Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy

DOCUMENT RESUME

ED 456 171 UD 034 268

AUTHOR Kirby, Douglas TITLE Emerging Answers: Research Findings on Programs To Reduce Teen Pregnancy. INSTITUTION National Campaign To Prevent Teen Pregnancy, Washington, DC. SPONS AGENCY David and Lucile Packard Foundation, Los Altos, CA.; Robert Wood Johnson Foundation, Princeton, NJ.; William and Flora Hewlett Foundation, Palo Alto, CA.; Target Stores/Dayton Hudson Corp.; Mott (C.S.) Foundation, Flint, MI. ISBN ISBN-1-58671-037-0 PUB DATE 2001-05-00 NOTE 242p.; Foreword by Sarah Brown. Also supported by the Summit and Turner Foundations. AVAILABLE FROM National Campaign To Prevent Teen Pregnancy, 1776 Massachusetts Avenue, NW, #200, Washington, DC 20036 ($15). Tel: 202-478-8500; Fax: 202-478-8588; e-mail: [email protected]; Web site: http://www.teenpregnancy.org). PUB TYPE Reports Descriptive (141) EDRS PRICE MF01/PC10 Plus Postage. DESCRIPTORS Acquired Immune Deficiency Syndrome; *Adolescents; *Comprehensive School Health Education; Contraception; Early Parenthood; *Pregnancy; Program Effectiveness; Program Evaluation; Secondary Education; Sex Education; *Sexuality; *Youth Programs IDENTIFIERS Abstinence; Condoms; Risk Reduction; *Risk Taking Behavior; Sexually Transmitted Diseases

ABSTRACT This report summarizes three bodies of research on teenagim pregnancy and programs to reduce the risk of teenage pregnancy. Studies included in this report were completed in.1980 or later, conducted in the United States or Canada, targeted adolescents, employed an experimental or quasi-experimental design, had a sample size of at least 100 in the combined treatment and control group, and measured the impact on sexual or contraceptive behavior, pregnancy, or childbearing. Six chapters focus on: (1) "Making the Case for Prevention Efforts: Adolescent Risk-Taking Behavior and Its Consequences"; (2) "Looking for Reasons Why: The Antecedents of Adolescent Sexual Behavior"; (3) "Assessing the Evidence: Factors Affecting the Strength of Research Results"; (4) "Emerging Answers: The Behavioral Impact of Programs To Reduce Adolescent Sexual Risk-Taking"; (5) "Looking Forward: Conclusions about the State of Research and the Effectiveness of Programs"; and (6)"Bringing It Home: Applying These Research Results in Communities." (Chapters contain references.) (SM)

Reproductions supplied by EDRS are the best that can be made from the original document. THE ,==3 NATIONAL

CAMPAIGN PREVENT -1" EN PREANCY Emerging Answers

RESEARCH FINDINGS ON PROGRAMS TO REDUCE TEEN PREGNANCY

Douglas Kirby, Ph.D.

MAY 2001

1 U.S. DEPARTMENT OF EDUCATION --\ Office of Educational Research and Improvement PERMISSION TO REPRODUCE AND EDUCATIONAL RESOURCES INFORMATION DISSEMINATE THIS MATERIAL HAS CENTER (ERIC) BEEN GRANTED BY /Thisdocument has been reproduced as c.i received from the person or organization originating it. triG6 0 Minor changes have been made to improve reproduction quality. 1,/teat4+ _gev.(41 Points of view or opinions stated in this TO THE ED AT ONRESOURCES document do not necessarily represent INFORMATION CENTER (ERIC) official OERI position or poky. BESTCOPYAVAILABLE THE IMATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY BOARD OF DIRECTORS

Chairman President Campaign Director THOMAS H. KEAN ISABEL V. SAWMILL SARAH BROWN Former Governor of New Jersey and Senior Fellow, President, Drew University The Brookings Institution

CAROL MENDEZ CASSELL BRUCE ROSENBLUM Director, Community Coalition Executive Vice President, Television, Partnership Programs for the Warner Brothers Prevention of Teen Pregnancy, CDC STEPHEN W. SANGER LINDA CHAVEZ Chairman and Chief Executive Officer President, General Mills, Inc. Center for Equal Opportunity VICTORIA P. SANT ANNETTE CUMMING President, The Summit Foundation Executive Director and Vice President, KURT L. SCHMOKE The Cumming Foundation Former Mayor of Baltimore FRANIUE SUE DEL PAPA and Partner, Wilmer, Cutler and Pickering Attorney General, VINCENT WEBER State of Nevada Partner, Clark & Weinstock WILLIAM GALSTON JUDY WOODRUFF School of Public Affairs, University of Maryland Prime Anchor and Senior Correspondent, CNN DAVID GERGEN ANDREW YOUNG Editor-at-Large, U.S. News & World Report Former Ambassador to the U.N. and WHOOPI GOLDBERG Co-Chair, GoodWorks International Actress KATHARINE GRAHAM Chairman of the Executive Committee, Trustees Emerki The Washington Post Company CHARLOTTE BEERS DAVID A. HAMBURG, M.D. Chairman, J. Walter Thompson President Emeritus, Carnegie Corporation of New York IRVINGB.HARRIS and Visiting Scholar, Department of Psychiatry Chairman, The Harris Foundation Weill Medical College, Cornell University BARBARA HUBERMAN ALEKINE CLEMENT JACKSON Director of Training, Advocates for Youth National President, YWCA of the USA SHEILA JOHNSON JUDITH E. JONES Executive Vice President of Clinical Professor, Corporate Affairs, BET, Inc. Columbia University School of Public Health LESLIE KANTOR NANCY KASSEBAUM BAKER Vice President of Education, Former U.S. Senator Planned Parenthood of New York City DOUGLAS KIRBY C. EVERETTKoop,M.D. Senior Research Scientist, ETR Associates Former U.S. Surgeon General JOHN D. MACOMBER JUDY MCGRATH Principal, President, MTV JDM Investment Group KRISTIN MOORE SISTER MARY ROSE MCGEADY President, Child Trends, Inc. President and Chief Executive Officer, HUGH PRICE Covenant House President, National Urban League, Inc. JODY GREENSTONE MILLER WARREN B. RUDMAN Venture Partner, MAVERON, LLC Former U.S. Senator and JOHN E. PEPPER Partner, Paul, Weiss, Rifkind, Wharton & Garrison Chairman, Board of Directors, ISABEL STEWART Procter & Gamble Company National Executive Director, Girls Inc. THE

NATIONAL

CAMPAIGN

PREVENT EN PRETANCY Emerging Answers

RESEARCH FINDINGS ON PROGRAMS TO REDUCE TEEN PREGNANCY

Douglas Kirby, Ph.D.

MAY 2001

4 Campaign Acknowledgments

The National Campaign gratefully acknowledge its many funders. Special thanks go to the David and Lucile Packard Foundation, the Robert Wood Johnson Foundation, the Summit Foundation, and the William and Flora Hewlett Foundation for generously supporting all of the Campaign's activities. Thanks also to the Turner Foundation and the Target Group of Stores for their support of Campaign publications and to the Charles Stewart Mott Foundation for its support of our research activities.

The National Campaign would also like to extend special appreciation to our Senior Editor, John Hutchins, for his unflagging dedication and tireless hard work on this project. Although this is an authored paper, its publication would simply not have been possible without John's guidance and editorial insight. John is a graceful and thoughtful editor and his talent is reflected throughout this manuscript. The Campaign also wishes to thank Bill Albert, our Director of Communications and Publications, and Ingrid Sanden, our Communications and Publications Assistant, for their many contributions to Emerging Answers and for helping shep- herd this publication carefully through to production. Lastly, the Campaign extends warm appreciation to Christine Flanigan for her careful attention to the demographic information provided in this and many other Campaign publications.

Author's Acknowledgments

The author would like to express a great deal of appreciation to John Hutchins and Sarah Brown for innumerable constructive suggestions for both the full volume and the summary, for their careful editing of both manuscripts, and for their continuous encouragement and support throughout the writing and editing of this volume. Without a doubt, this volume is more accurate, clear, and easy to read because of their efforts. Thanks also to Christine Flanigan for finding and checking statistics and to Ingrid Sanden for carefully editing the tables and references. And, finally, many thanks to the members of the National Campaign's Effective Programs and Research Task Force who read successive versions of this volume, offered many suggestions, and made sure that the reporting was accurate and balanced. In the end, however, the review's findings and conclusions are my own.

ISBN: 1-58671-037-0

Suggested citation: Kirby, D. (2001). Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy. Washington, DC: National Campaign to Prevent Teen Pregnancy.

Design: ampth-sand graphic design, inc. 1700 Connecticut Avenue, NW, Suite 401 Washington, DC 20009

5 EFFECTIVE PROGRAMS AND RESEARCH TASK FORCE

Chair SARA MCLANAHAN, PH.D. DOUGLAS KIRBY, PH.D. Professor Senior Research Scientist Office of Population Research ETR Associates Princeton University

BRENT MILLER, PH.D. Members Professor and Head CHIUSTINE BACHRACH, PH.D. Department of Family and Human Development Chief, Demographic and Behavioral Sciences Branch Utah State University National Institute of Child Health and Human Development KRISTIN MOORE, PH.D.* President ROBERT W. BLUM, M.D., PH.D. Child Trends, Inc. Professor and Director Division of General Pediatrics and Adolescent Health SUSAN PHILLIBER, PH.D. Department of Pediatrics Senior Partner University of Minnesota Philliber Research Associates

CLAIRE BRINDIS, DR.P.H. MICHAEL RESNICK, PH.D. Director Professor and Director Center for Reproductive Health Policy Research National Teen Pregnancy Prevention Research Center Institute for Health Policy Studies University of Minnesota University of California at San Francisco FREYA SONENSTEIN, PH.D.* SHERYLL D. CASHIN, J.D. Associate Professor of Law Director, Population Studies Center Georgetown University Law Center The Urban Institute

JACQUELINE E. DARROCH, PH.D. FELICIA H. STEWART, M.D. Senior Vice President and Vice President for Research Adjunct Professor, Department of Obstetrics, The Alan Guttmacher Institute Gynecology and Reproductive Sciences Co-Director, Center for Reproductive Health CYNTHIA A. GOmEz, PH.D., ED.M. Policy Research Assistant Professor, Department of Medicine University of California at San Francisco University of California at San Francisco Center for AIDS Prevention Studies STAN WEED, PH.D. Director SAUL D. HOFFMAN, PH.D. Institute for Research & Evaluation Professor Department of Economics BRIAN L. WILCOX, PH.D.* University of Delaware Director REBECCA A. MAYNARD, PH.D. Center on Children, Families, and the Law University Trustee Professor of Education University of Nebraska and Social Policy University of Pennsylvania BARBARA WOLFE, PH.D.* Director JOE S. MCILHANEY, JR., M.D. Institute for Research on Poverty President Department of Economics The Medical Institute for Sexual Health University of Wisconsin at Madison

* Former Task Force members who participated in the initial development and review of this report

6 Foreword

s Doug Kirby notes in his Author's tiveness of media campaigns. However, the Preface (p. v), much has changed for National Campaign's most requested the better in the four years since the research publication by far has been Dr. National Campaign published his first review Kirby's No Easy Answers, which is a testament of evaluation research on programs to pre- both to the quality of his work and to the vent teen pregnancy, No Easy Answers, in intense interest among program developers 1997. Teen pregnancy and birth rates have and political leaders alike in finding out been steadily declining, efforts to prevent "what works." I anticipate that this long- teen pregnancy at both the national and local awaited update of the research, with the levels have increased, and, as this report more hopeful title of Emerging Answers, will shows, the quality of evaluation research in prove to be as influential and popular as its this field has improved, bringing with it clear predecessor. evidence that several different kinds of programs can reduce teen sexual risk-taking and This report summarizes three bodies of pregnancy. This is good news for all of us research. First, it examines the current statis- who care about young people and about tics on teen pregnancy, childbearing, sexually the next generation of children who deserve transmitted diseases, and related problems to be raised by adult parents. (Chapter 1). Second, it identifies and summarizes the important risk and protective fac- From our founding in 1996, the tors associated with teen sexual behavior National Campaign has believed that "get- (Chapter 2). Third, it discusses how to assess ting the facts straight" is critically important evaluation research and then reviews numer- in our field a field that is subject to so ous evaluations of specific programs designed much controversy and conflict. Under the to reduce sexual risk-taking, teen pregnancy, guidance of the National Campaign's Task and HIV and other STDs (Chapters 3 and Force on Effective Programs and Research 4). It concludes with recommendations (EPR), which Dr. Kirby chairs, we have pub- about program implementation and evalua- lished a series of research reports on such tion (Chapters 5 and 6). The summary, topics as parental and family influence in ado- included in this volume on pages 1-11, is lescent sexual behavior, the role of peers in also available as a separately-published teens' sexual decision-making, and the effec- pamphlet.

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 7 I offer a few comments here to put this tion as a high-quality evaluation researcher report in context. Dr. Kirby's review focuses himself, and, as a consequence, a number of on organized programs. As such, the efforts his own studies of programs appear in this of parents, families, and other caring individ- review. In addition, in the interest of full dis- uals are not chronicled or assessed here. Nor closure, Dr. Kirby thought it was important does he review the efficacy of specific contra- to make it clear that ETR Associates, a non- ceptive methods or of programs designed profit organization that provides educational specifically to reduce pregnancies and births resources, training, and research in health among teens who are already mothers. In promotion, developed the Reducing the Risk addition, this review is narrowly focused on and Safer Choices curricula, two of the sex the effect these programs have on reducing and HIV education programs this review teen pregnancy and behavior that leads to concludes have the strongest evidence of teen pregnancy; it does not examine what effectiveness. ETR Associates continues to other results the programs might produce. market these curricula. In addition, several In particular, many of the programs dis- members of the National Campaign's Task cussed in this review may very well have had Force on Effective Programs and Research positive effects for their communities that were also involved in some of the studies have not been measured or may not be mea- reviewed in this report. surable for instance, enhancing teens' self- esteem or building a sense of common Although we believe that having accu- purpose in a community. Moreover, many rate, research-based information can only creative programmatic approaches to reduc- help communities make good decisions ing adolescent pregnancy have not yet been about preventing teen pregnancy, the evaluated and therefore could not be National Campaign recognizes that commu- included in this review. nities choose to develop particular prevention programs for many reasons other than On behalf of the National Campaign, I research including, for example, compati- would like to express our great appreciation bility with religious traditions, available to Doug Kirby for producing this excellent resources, community standards, and the research review. We Commend him for his personal values and beliefs of the leaders in diligence in searching high and low for relevant studies (published and unpublished), for charge. In this context, I would add that it is his unwavering commitment to being fair crucial for such leaders to understand that and evenhanded in his assessment of the community-based programs are only part of research, for his meticulous attention to the solution to the teen pregnancy challenge detail, and, most of all, for his great wisdom and that no single effort can be expected to and good humor throughout an extensive solve this problem by itself. Teen pregnancy process of review and editing. We also is, after all, a very complex problem, influ- extend our deep appreciation to the National enced by many factors, including individual Campaign's Task Force on Effective biology, parents and family, peers, schools Programs and Research (see the list after the and other social institutions, religion and title page), a distinguished and diverse group faith communities, the media, and the list of researchers and experts, under whose aus- goes on. In an ideal world, we would mount pices this review was developed. efforts to engage the help of all these forces, particularly popular culture, schools, faith It should be noted here that Doug communities, parents, and other adults. But Kirby, who is a Senior Research Scientist at we are a long way from doing so, and many ETR Associates, has a well-deserved reputa- communities mistakenly believe that modest

THE NATIONAL CAMPAIGN TO PREVENT TEEN ii community programs can do this single- on the complex problem of teen pregnancy. handedly. In many instances, these programs Instead, they should be encouraged by are fragile and poorly-funded; even appar- declining rates and new research showing ently "effective" programs often achieve only that some programs are making a difference. modest results; and not all teens at risk of They should continue to explore many ways pregnancy are enrolled in programs. The to address the various causes of teen preg-, simple point is that no single approach can nancy. They should replicate those programs solve this problem alone, whether it be a that have the best evidence for success, build national media campaign, a new move in their efforts around the common elements of faith communities to address this problem, successful programs, and continue to explore, or a well-designed community program. develop, and evaluate innovative and promis- Advocates of any single approach ing approaches. especially, in the context of this review, community programs should therefore be modest in both their promises and their expectations. Sarah Brown Director In the final analysis, professionals work- National Campaign to ing with youth should not adopt simplistic Prevent Teen Pregnancy solutions with little chance of making a dent May 2001

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy III 9 Author's Preface

n 1997, I wrote No Easy Answers: . recorded level ever. Second, larger, more rig- Research Findings on Programs to Reduce orous studies of some sex and HIV education Teen Pregnancy for the National programs have found sustained positive Campaign to Prevent Teen Pregnancy. At effects on behavior for as long as three years. that time, with only a few exceptions, most Third, there is now good evidence that one studies assessing the impact of programs to program that combines both sexuality educa- reduce teen sexual risk-taking failed either tion and youth development (i.e., the to measure or to find sustained long-term Children's Aid Society-Carrera Program) can impact on behavior. Among the few pro- reduce pregnancy for as long as three years. grams that appeared to have longer-term Fourth, both service learning programs (i.e., impact, none had been evaluated two or voluntary community service with group dis- more times by independent researchers and cussions and reflection) and sex and HIV found to be effective. Indeed, the two repli- education programs (i.e., Reducing the Risk) cations of programs that had previously have now been found to reduce sexual risk- shown positive effects on behavior failed to taking or pregnancy in several settings by corroborate those initial positive findings. In independent research teams. Fifth, there is general, the research evidence indicated that emerging evidence that some shorter, more there were "no easy answers" to markedly modest clinic interventions involving educa- reducing teen pregnancy in this country. tional materials coupled with one-on-one counseling may increase contraceptive use. Now, four years later, the research find- All of these findings are most encouraging. ings are definitely more positive, and there Of course, it is still very challenging to design are at least five important reasons to be more or operate programs,that actually reduce ado- optimistic that we can craft programs that lescent sexual risk-taking and pregnancy over help to reduce teen pregnancy. First, teen prolonged periods of time. However, we now pregnancy, abortion, and birth rates began to know it is possible, and we have clearer decrease about 1991 and have continued to guidelines for how to do it. decline every year since then. Not only have these rates maintained their downward trend, This report is, in many respects, a sec- but teen birth rates are now at their lowest ond edition of No Easy Answers. Much of the

Ernerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 1 0 content and organization remains the same. more consistent research findings demon- However, the methodological criteria for strating program effectiveness, we have enti- inclusion of studies has changed, more stud- tled it Emerging Answers. ies have been reviewed, and there are impor- Douglas Kirby, Ph.D. tant new findings. Given the stronger and May 2001

1 1

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY vi Table of Contents

Foreword

Author's Preface

Summary Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy

Chapter One 13 Making the Case for Prevention Efforts: Adolescent Risk-Taking Behavior and its Consequences

Chapter Two 25 Looking for Reasons Why: The Antecedents of Adolescent Sexual Behavior

Chapter Three 73 Assessing the Evidence: Factors Affecting the Strength of Research Results

Chapter Four 83 Emerging Answers: The Behavioral Impact of Programs to Reduce Adolescent Sexual Risk-Taking

Chapter Five 169 Looking Forward: Conclusions About the State of Research and the Effectiveness of Programs

Chapter Six 177 Bringing It Home: Applying These Research Results in Communities

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy vii

1 Summary Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy

Introduction However, what was true in 1997 is still true today: teen pregnancy and childbearing When the National Campaign to remain very serious problems in the United Prevent Teen Pregnancy released its first States. Even with recent declines, the United major report, No Easy Answers: Research States still has the highest teen pregnancy Findings on Programs *to Reduce Teen and birth rates among comparable industrial- Pregnancy (Kirby, 1997), it wasn't clear that ized nations, twice as high as Great Britain the recent modest reductions in rates of teen and ten times as high as the Netherlands, for pregnancy arid childbearing noted at the instance. In other words, this is no time to time were going to continue. Four years be complacent; there's still a long way to go. later, there is good news to report: teen pregnancy and childbearing rates have con- Not surprisingly, people from all over tinued their significant decline for several the country still come to the National years among all racial and ethnic groups and Campaign with one principal question: in all parts of the United States. The credit "What can I do in my community to prevent for this welcome trend goes, of course, to teen pregnancy what really works?" This teens themselves who have obviously new research review helps answer that ques- changed their behavior for the better. tion more definitively. However, it is impor- Evaluation research completed since No Easy tant from the outset to note some of its Answers was published offers additional good limitations. The full report, which is summa- news: more programs to prevent teen preg- rized here, discusses only those programs nancy are making a real difference in encour- that have been subjected to evaluation aging teens to remain abstinent or use research that meets certain methodological contraception when they have sex. As a result criteria (see below). It does not discuss what of these encouraging trends in the rates and parents can do; it does not evaluate the role in the research, this updated research review of broad cultural values and norms; and it is entitled Emerging Answers. does not review the relative efficacy of vari-

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy

13 ous methods of contraception. And the The Problem of Teen Pregnancy paper examines only primary prevention programs; it does not.review interventions to The recent and steady decline in teen prevent second pregnancies and births pregnancy and birth rates in the United among teen mothers, although some of the States should provide encouragement that continued progress is possible. However, conclusions would apply to these pregnancies there remain compelling reasons to increase and births as well. In addition, it is crucial prevention efforts: for leaders to understand that although effective programs can help reduce teen preg- oDespite the declining rates, more than nancy a few quite substantially it is four in ten teen girls still get pregnant at naive to think that they can completely solve least once before age 20, which translates the problem by themselves. Indeed, no sin- into nearly 900,000 teen pregnancies per gle approach to preventing teen pregnancy year. can provide a 100% solution. o Despite a leveling off of sexual activity among teens, about two-thirds of all stu- Nonetheless, prevention programs can dents have sex before graduating from be an important part of the answer, and it is high school potentially exposing encouraging that research is revealing more themselves to pregnancy and STDs. about what makes the successful ones work. oWhen teens give birth, their future The research reviewed here offers some prospects become more bleak. They important "emerging answers" about what become less likely to complete school effective programs look like. It summarizes and more likely to be single parents, for what has, and has not, worked in many com- instance. Their children's prospects are munities. Of course, local decisions about even worse they have less supportive programming are often affected by more and stimulating home environments, than research, including such important con- poorer health, lower cognitive develop- siderations as community values, available ment, worse educational outcomes, more resources, complementary services already behavior problems, and are more likely to become teen parents themselves. available, the preferences of teens and parents, and local politics. Fortunately, a num- o Despite indications of better use of con- ber of manuals to help communities put all traception by sexually active teens (par- these pieces together are available, including ticularly of condoms at first sex), many the National Campaign's Get Organized: A do not use contraceptives correctly and Guide to Preventing Teen Pregnancy. consistently every time they have sex. oAs a result of sexual risk-taking, about The following summary outlines some one in four sexually experienced teens facts that explain why communities must contract an STD each year some, of remain vigilant about teen pregnancy, child- which are incurable, including HIV, bearing, and STDs, outlines the criteria for which is, of course, life-threatening. including studies in this review, discusses oDespite recent encouraging trends in some of the antecedents of teen sexual risk- teen pregnancy, it is important to taking, and, finally, summarizes the findings remember that each year a new set of of the research review and their implications teens arrives on the scene, meaning that for communities. efforts to prevent teen pregnancy must

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 2 1 4 be constantly renewed. In addition, Antecedents to Sexual between 2000 and 2010, the population Risk-Taking, Pregnancy, of teen girls aged 15-19 is expected to and Childbearing increase by nearly 10 percent which means that even declining rates may not The reasons behind teen pregnancy are necessarily mean fewer numbers of teen complex, varied, and overlapping. In fact, from a review of at least 250 studies, pregnancies and births. Emerging Answers culls more than 100 pre- The Criteria for Inclusion cursors or "antecedents" to early teen sexual in this Review intercourse, poor contraceptive use, pregnancy, and childbearing. These risk factors Evaluation studies included in fall under such categories as community dis- Emerging Answers had to meet certain scien- advantage; family structure and economic disadvantage; family, peer, and partner atti- tific criteria. While No Easy Answers used tudes and behavior; and characteristics of publication in a peer-reviewed journal as the teens themselves, including biology, detach- primary qualification for including a study, ment from school, other behaviors that put this review relies on an expanded set of young people at risk, emotional distress, and methodological criteria. This change was sexual beliefs, attitudes, and skills. While all made for two reasons: (1) some studies teens are at some risk, some teens are at employed rigorous research methods but, for much higher risk than others. These a variety of reasons, were never published in antecedents can be used to identify those peer-reviewed journals, and (2) a few studies youth at higher risk of sexual risk-taking and published in peer-reviewed journals to guide the development of effective pro- employed very weak methods and provided grams. No single program could or misleading results. To be included in should try to address all of these Emerging Answers, a program evaluation had antecedents; yet, at the same time, effective programs are more likely to focus intention- to meet multiple criteria, the most important ally on several of them in a clear, purposeful of which were to have: way. obeen completed in 1980 or later, Because the reasons behind teen preg- obeen conducted in the United States nancy vary, so do the types of programs or Canada, adults design to combat the problem. When most people think of preventing teen preg- obeen targeted at adolescents of middle nancy, they probably conjure images of sex school or high school age (roughly or abstinence education classes or clinics that 12-18), offer contraceptive services. Although the most important antecedents of teen preg- oemployed an experimental or quasi- nancy and childbearing relate directly to sex- experimental design, ual attitudes, beliefs, and skills, many ohad a sample size of at least 100 in the influential family, community, cultural, and combined treatment and control group, individual factors closely associated with teen and pregnancy actually have little to do directly with sex (such as growing up in a poor com- o measured impact on sexual or contracep- munity, having little attachment to one's par- tive behavior, pregnancy, or childbearing. ents, failing at school, and being depressed).

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 3 1 5 In fact, one program with strong evidence Community-Wide Initiatives with Many for success in reducing teen pregnany con- Components centrates on the non-sexual antecedents of Programs That Focus on Non-Sexual teen pregnancy. Simply put, the antecedents Antecedents to teen pregnancy come in two categories: those that are sexual in nature (such as atti- Early Childhood Programs tudes toWard sex and contraception) and Youth Development Programs for Adolescents those that are not. o Service Learning Programs oVocational Education and Employment Findings on Programs Programs With these two categories of o Other Youth Development Programs antecedents in mind, one can divide pro- Programs That Focus on Both Sexual and grams to preYent teen pregnancy into three Non-Sexual Antecedents types: those that focus on sexual antecedents, those that focus on non-sexual antecedents, Programs with Both Sexuality and Youth and those that do both. Emerging Answers Development Components organizes its findings on programs into these Programs that Focus on Sexual three broad categories and then into sev- Antecedents ofTeen Pregnancy eral sub-categories and offers conclusions about the research in each. Of course, given Programs that focus on the sexual the great diversity of programs that exist, any antecedents of teen pregnancy are divided in typology will be inadequate to the task of this review into four groups: curricula-based capturing all the various ways that programs programs for young people (including absti- can be defined. nence education and sex education) that are typically offered in schools, sex and HIV Programs That Focus on the Sexual education programs for parents and families, Antecedents of Teen Pregnancy programs to improve access to condoms and other contraceptives, and multi-component, Curricula-Based Programs community-wide initiatives that have a oAbstinence-Only Programs strong emphasis on sex education or contraceptive services. o Sex and HIV Education Programs Sex and HIV Education Programs for Parents Curricula-Based Programs and Families According to recent national surveys, Clinic or School-Based Programs to Provide nearly every teenager in this country receives Reproductive Health Care or to Improve some form of sex or abstinence education, Access to Condoms or Other Contraceptives but the curricula vary widely in both focus o Family Planning Clinics and Services and intensity. This review places curricula into two groups: abstinence-only education o Protocols for Clinic Appointments and and sex or HIV education (sometimes also Supportive Activities called abstinence-plus or comprehensive sex o Other Clinic Characteristics and Programs education). There has been a great growth in the former category since the 1996 welfare o School-Based and School-Linked Clinics reform law made $85 million in federal and oSchool Condom-Availability Programs state funding available each year for absti-

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 4 1 6 nence-until-marriage interventions. programs included in this review do not However, in practice, curricula-based pro- increase sexual activity they do not hasten grams don't really divide neatly into these the onset of sex, increase the frequency of two groups; they actually exist along a dm- sex, nor increase the number of sexual part- tinuum. For instance, while all abstinence- ners. To the contrary, some sex and HIV only programs focus on abstinence as the education programs delay the onset of sex, only truly healthy and correct choice for young people, some also discuss condoms reduce the frequency of sex, or reduce the and other contraception, focusing primarily number of sexual partners. In fact, since the on their failure rates; others mention the publication of No Easy Answers, two inde- protective uses of condoms in a medically pendent studies have found that one particu- accurate manner, while still stressing absti- lar curriculum, Reducing the Risk, delayed nence. Similarly, many sexuality education the onset of intercourse. (Reducing the Risk programs describe abstinence as the safest, also increased the use of condoms or contra- and often the best, choice for teens but also ceptives among some groups of youth). This encourage the use of condoms and other contraception for sexually active teens. A few is the first time that research on replications particularly those for high-risk, sexually of a sex education program has confirmed active youth focus primarily on consistent initial findings of effectiveness. use of contraceptives, especially condoms. Other sex and HIV education programs

Abstinence-Only Programs including Safer Choices; Becoming a Responsible Teen; Making a Difference: An Very little rigorous evaluation of absti- Abstinence Approach to STD, Teen Pregnancy, nence-only programs has been completed; in and HIV/AIDS Prevention; and Making a fact, only three studies met the criteria for this review. The primary conclusion that can Difference: A Safer Sex Approach tolI), be drawn from these three studies is that the Teen Pregnancy, and HIV/AIDS Prevention evidence is not conclusive about abstinence- have also been shown to delay sex or only programs. None of the three evaluated increase condom or other contraceptive use programs showed an overall positive effect and thereby to decrease unprotected sex sub on sexual behavior, nor did they affect con- stantially. The studies of these four curricula traceptive use among sexually active partici- employed experimental designs and found pants. However, given the paucity of the positive behavioral effects for at least 12 to research and the great diversity of abstinence-only programs that is not reflected in 31 months. All five of these sex and HIV these three studies, one should be very care- education curricula have also been identified ful about drawing conclusions about absti- by the Centers for Disease COntrol and nence-only programs in general. Fortunately, Prevention (CDC) as having strong evidence results from a well-designed, federally- of success. sponsored evaluation of Title V-funded abstinence programs should be available The programs that have changed teens' within the next two years. sexual behavior share ten necessary characteristics (see sidebar on the next page). The Sex and HIV Education Programs absence of even one of these characteristics A large body of evaluation research appears to make a program appreciably less clearly shows that sex and HIV education likely to be effective.

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 5 17 I 0 Characteristics of Effective Sex and HIV Education Programs

The curricula of the most effective sex and HIV education programs share ten common characteristics.These programs:

I.Focus on reducing one or more sexual behaviors that lead to unintended pregnancy or HIV/STD infection.

Are based on theoretical approaches that have been demonstrated to influence other health-related behavior and identify specific important sexual antecedents to be targeted.

3. Deliver and consistently reinforce a clear message about abstaining from sexual activity and/or using condoms or other forms of contraception. This appears to be one of the most important characteristics that distinguishes effective from ineffective programs.

4.Provide basic, accurate information about the risks of teen sexual activity and about ways to avoid intercourse or use methods of protection against pregnancy and STDs.

5.- Includeactivities that address social pressures that influence sexual behavior.

6. Provide examples of and practice with communication, negotiation, and refusal skills.

7. Employ teaching methods designed to involve participants and have them personalize the information.

8. Incorporate behavioral goals, teaching methods, and materials that are appropriate to the age, sexual experience, and culture of the students.

9. Last a sufficient length of time (i.e., more than a few hours).

10. Select teachers or peer leaders who believe in the program and then provide them with adequate training.

Generally speaking, short-term curricula whether abstinence-only or sexuality education programs do not have measurable impact on the behavior of teens.

Sex and HIV Education Programs for intercourse found statistically significant Parents and Families effects, but the characteristics of the studies might have obscured possible program Most parents want to impart their val- impact. This does not mean that parental ues about sexuality to their children. But influence and parent/child communication because parents often have difficulty talking are not important. In fact, other research with their children about sexual topics, a confirms the importance of parent/child number of educational programs have been "connectedness," for instance, in reducing developed to improve parent/child commu- risky sexual behavior among teens. nication. Many studies have demonstrated short-term increases in parent/child commu- Programs Designed to Improve Access to nication, as well as increases in parent com- Condoms or Other Contraceptives fort with that communication, although the positive effects dissipate with time. Neither Many community family planning clin- of the two studies that measured whether ics, school-based health clinics, and school- these programs delayed the onset of sexual linked clinics offer services to teens,

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 6 1 8 including access to condoms and other con- clear messages about abstinence and traceptives. With regard to family planning contraception, and provided or prescribed clinics in particular, it is clear that they pro- contraceptives. vide many adolescents with contraceptive services, which presumably prevent pregnancies While studies of school condom-avail- among those teens. Nonetheless, because the ability programs consistently demonstrated long-term impact of family planning services that the programs did not increase sexual on the frequency of sexual behavior is not activity, they provided conflicting results known, the number of teen pregnancies pre- about their impact upon school-wide use of vented by family planning services is difficult condoms. These differences may reflect to estimate. methodological limitations, differences in the availability of condoms in the community, or However, there are clearer findings differences in the programs themselves. regarding particular clinic protocols or programs within health or family planning clin- Taken together, these studies suggest ics. These programs in which youth were that family planning clinic protocols or provided with information about abstinence, programs, school-based and school-linked condoms, and/or contraception; were clinics, and condom-availability programs in engaged in one-on-one discussions about schools that increased condom or other con- their own behavior; were given clear mes- traceptive use shared common characteristics. sages about sex and condom or contraceptive They focused primarily (or solely) upon use; and were provided condoms or contra- reproductive health and provided young ceptives consistently increased the use of people with a combination of educational condoms and contraception without increas- materials (however modest), the opportunity ing sexual activity. for one-on-one counseling or discussions, a clear message about abstinence and condom Many studies of schools with health or contraceptive use, and actual condoms or clinics and schools with condom-availability other contraceptives. programs have consistently shown that the provision of condoms or other contraceptives Community-Wide Initiatives with Many through schools does not increase sexual Components activity. Studies also show that substantial proportions of sexually experienced students In the past two decades, recognizing have obtained contraceptives from these pro- the complexity of the problem of teen preg- grams. However, given the relatively wide nancy, more communities have put in place availability of contraceptives in most commu- multi-component efforts to reduce rates of nities, most school-based clinics, especially teen pregnancy. These initiatives typically those that did not focus on pregnancy or combine such interventions as media cam- STD prevention, did not appear to markedly paigns, increased access to family planning increase the school-wide use of contracep- and contraception services, sex education tives that is, there appeared to be a "sub- classes for teens, and training in parent/child stitution effect," meaning that teens merely communication. The research evidence on switched from getting contraception from a these initiatives is mixed. Each of the studies source outside of school to getting it in reviewed in the report measured effects on school. By contrast, two studies suggested teens throughout the community, not just that school-based or school-linked clinics did on those teens directly served by programs. increase use of contraception when they The two most effective programs were the focused much more on contraception, gave most intensive ones, and, in fact, when the

Ernerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 7 1 9 interventions ceased, the use of condoms or Project, infants in low-income families were pregnancy rates returned to pre-program lev- randomly assigned to a full-time, year-round els, suggesting that such programs need to day care program focused on improving be maintained in order to have continuing intellectual and cognitive development or to effects. However, one of these two effective regular infant day care. In elementary school, programs did not show positive results when they were again randomly assigned to a it was tried again in a different community. three-year parent involvement program or to The bottom line seems to be that it is very a normal school environment. The children hard to change adolescent sexual or contra- were followed until age 21. The kids in the ceptive behavior throughout an entire preschool program delayed childbearing by community. When such change is accom- more than a year in comparison with the plished, it takes intense effort, which must be control group; they also performed higher sustained. on a number of intellectual and academic measures. While this is encouraging, it is Programs That Focus on Non-Sexual only one study with a small sample, albeit Antecedents with a strong scientific design.

Programs in this category focus on Youth Development Programs for broader reasons behind why teens get preg- Adolescents nant or cause a pregnancy, including disadvantaged families and communities, Service Learning Programs detachment from school, work, or other important social institutions, and lack of Service learning programs include two close relationships with parents and other parts: (1) voluntary service by teens in the caring adults. For instance, research suggests community (e.g., tutoring, working in nurs- that teens who are doing well in school and ing homes, and fixing up parks and recre- have educational and career plans for the ation areas), and (2) structured time for future are less likely to get pregnant or cause preparation and reflection before, during, a pregnancy. Increasingly, programs to pre- and after service (e.g., group discussions, vent teen pregnancy concentrate on helping journal writing, and papers). Sometimes the young people develop skills and confidence, service is part of an academic class. Service focus on education, and take advantage of learning programs may have the strongest job opportunities and mentoring relation- evidence of any intervention that they reduce ships with adults thereby helping them actual teen pregnancy rates while the youth create reasons to make responsible decisions are participating in the program. Among the about sex. These efforts include service programs with the best evidence of effective- learning, vocational education and employ- ness are the Teen Outreach Program and ment programs, and youth development pro- Reach for Health service learning program. grams, broadly defined. Early childhood Although the research does not clearly indi- programs also focus on non-sexual cate why service learning is so successful, sev- antecedents that may have an impact on the eral possibilities seem plausible: participants later sexual behavior of their participants. develop relationships with program facilitators, they gain a sense of autonomy and feel Early Childhood Programs more competent in their relationships with peers and adults, and they feel empowered Only one study evaluating an early by the knowledge that they can make a dif- childhood program met the criteria for this ference in the lives of others. All such factors, review. In the study of the Abecedarian in turn, may help increase teenagers' motiva-

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 8 tion to avoid pregnancy. In addition, partici- Programs with Both Sexuality and pating in supervised activities especially Youth Development Components after school may simply reduce the oppor- Three studies have examined programs tunities teens have to engage in risky behav- that address both reproductive health and ior, including unprotected sex. youth development simultaneously. The first study evaluated three programs in Washington Vocational Education Programs state that provided teens with small group and individualized education and skill-build- Vocational education programs provide ing sessions, as well as other individual ser- young people with remedial, academic, and vices. Results indicated that the programs did vocational education sometimes coupled not delay sex nor increase contraceptive use, with assistance in getting jobs and other but they did decrease the frequency of sex. health education and health services. Four The second study evaluated different pro- studies have evaluated the effect of such programs in 44 sites in California targeted to the grams on teen sexual risk-taking, pregnancy, sisters of teen girls who had become preg- and childbearing. A strong study of the nant an interesting strategy that is based Summer Training and Education Program on the well-known fact that having an older (STEP) revealed that the program did not sister become pregnant increases the chances have a consistent and significant impact on that younger sister will do the same. The either sexual activity or use of contraception. programs offered individual case manage- Similarly, evaluations of three programs, the ment and group activities and services. The Conservation and Youth Service Corps, the evaluation showed that the interventions Job Corps, and JOBSTART, revealed that delayed sex and decreased reported preg- they did not affect overall teen pregnancy or nancy nine months later. birth rates at 15- to 48-month follow-up. Finally, a recent and very rigorous study Thus, these studies provide rather strong evi- of the comprehensive Children's Aid Society- dence that programs like these four, which Carrera Program has demonstrated that, offer academic and vocation education and a among girls, it significantly delayed the onset few support services and are quite intensive, of sex, increased the use of condoms and will not decrease pregnancy or birth rates other effective methods of contraception, among disadvantaged teens. and reduced pregnancy and birth rates. The program did not reduce sexual risk-taking Other Youth Development Programs among boys. The CAS-Carrera Program, which is long-term, intensive, and expensive, Two other youth development pro- includes many components: (1) family life grams have been evaluated for their effect on and sex education, (2) individual academic teen pregnancy or birth rates. One of them, assessment, tutoring, help with homework, the Seattle Social Development Program, was preparation for standardized exams, and designed to increase grade schoolers' attach- assistance with college entrance, (3) work- ment to school and family by improving related activities, including a job club, teaching strategies and parenting skills. stipends, individual bank accounts, When these students were followed to age employment, and career awareness, (4) self- 18, those receiving the intervention were less expression through the arts, (5) sports activi- likely to report a pregnancy than the com- ties, and (6) comprehensive health care, parison group. This is encouraging, but the including mental health and reproductive evaluation design was not strong. health services and contraception. This is the

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 9 first and only study to date that includes ran- strated to substantially reduce teen preg- dom assignment, multiple sites, and a large nancy and birth rates among girls over a sample size and that found a positive impact long period of time. on sexual and contraceptive behavior, pregnancy, and births among girls for as long as These three categories of programs may three years. seem contradictory one focuses directly on issues of sex and contraception, one What Does All This Mean? addresses non-sexual factors, and the third targets both. But finding effective programs Just as in 1997, there are still no easy in each category is heartening news and answers to the problem of teen pregnancy. conforms with what the research says about However, recent research suggests that there the antecedents of teen pregnancy and child- are programs in each of the three main cate- bearing. If very different approaches prove to gories described above with evidence that be effective, then communities benefit they reduce sexual risk-taking, pregnancy, because they have more options from which and childbearing among teens (see to choose. "Programs with Strong Evidence of Success," Chapter 6, p. 179): Studies of a number of other types of interventions, including community-wide Programs That Focus on Sexual initiatives and collaboratives, school-based Antecedents: Several sex and HIV educa- clinics and school condom distribution pro- tion programs delay the onset of sex, grams, and some sex and HIV education reduce the frequency of sex, reduce the programs, offer mixed results of effective- number of sexual partners among teens, ness. In addition, the few rigorous studies of or increase the use of condoms and other abstinence-only curricula that have been forms of contraception. The most suc- completed to date do not show any overall cessful programs share ten specific char- effect on sexual behavior or contraceptive acteristics (see p. 91). In addition, several use. That said, one should not conclude that particular protocols and interventions in these various interventions have no value at clinic programs also increase the use of all or that they should necessarily be aban- condoms or other forms of contraception. doned as part of the overall mix of preven- oPrograms That Focus on Non-Sexual tion strategies. There may be a variety of Antecedents: Certain service learning pro- such interventions whose value has not yet grams, which do not focus on sexual been identified by rigorous evaluation. issues at all, have the strongest evidence In addition, the research indicates that that they actually reduce teen pregnancy encouraging abstinence and urging better rates. Other types of youth development use of contraception are compatible goals programs, especially vocational educa- for at least two reasons. First, the over- tion, have not reduced teen pregnancy or whelming weight of evidence shows that sex childbearing. education that discusses contraception does oPrograms That Focus on Both Sexual and not increase sexual activity. Second, those Non-Sexual Antecedents: A comprehen- programs that emphasize abstinence as the sive, intensive, and long-term safest and best approach, while also teaching intervention, the Children's Aid Society- about contraceptives for sexually active Carrera Program, which includes both youth, do not decrease contraceptive use. In youth development and reproductive fact, effective programs shared two common health components, has been demon- attributes: (1) being clearly focused on sexual

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 10 22 behavior and contraceptive use and (2) deliv- antecedents of these behaviors (i.e., the ering a clear message about abstaining from individual, family, social, and community sex as the safest choice for teens and using factors that predispose teens to risky protection against STDs and pregnancy if a behaviors), and (c) the particular pro- teen is sexually active. gram activities designed to change these antecedents. This way of thinking and So, what should communities do with planning usually results in programs that this information gleaned from the research literature? Emerging Answers suggests three have clear goals and orderly and plausible strategies for employing promising plans for reaching those goals. approaches: In the final analysis, professionals working with youth should not adopt simplistic 1.The best option is to replicate with solutions with little chance of making a dent fidelity (that is, carefully copy) programs that have been demonstrated to be effec- on the complex problem of teen pregnancy. tive with similar populations of teens. Instead, they should be encouraged by declining rates and new research showing 2. The next best option is to select or that some programs are making a difference. design programs with the common char- They should continue to explore many ways acteristics of programs that have been to address the various causes of teen preg- effective with similar populations. nancy. They should replicate those programs.. 3.If a community cannot do either #1 or that have the best evidence for success, build #2, it should use a careful, deliberate their efforts around the common elements of process to select or design new programs successful programs, and continue to explore, and not just rely on accustomed ways of develop, and evaluate innovative and promis- doing things. A useful strategy is to use a ing approaches. process adopted by many of the people who designed the effective programs Of course, all young people live in a reviewed above: develop logic models. A larger culture that is influenced by such dis- logic model (also called a causal or pro- parate forces as parents, peers, schools, the gram model) is a concise, causal descrip- economy, faith institutions, and the enter- tion of exactly how certain program tainmënt media. So even as professionals activities can be expected to affect partic- continue to develop, implement, and evalu- ular behaviors by teens. At a minimum, a logic model requires that one be specific ate better and more effective prevention pro- about what behavior one wants to grams, there is still enough work for other change. A logic model identifies in the sectors of society to help make adolescence following order: (a) the behaviors to be in America a time of education and growing changed, (b) the precursors or up, not pregnancy and parenting.

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy

II 23 Chapter 11 Making the Case for Prevention Efforts: Adolescent Risk-Taking Behavior and Its Consequences

Sexual activity by teens and teen preg- Notably, many of these trends among nancy are not new phenomena. adolescents paralleled similar trends among However, during the past century adults. During the last few decades, adults (and especially the last few decades), several have also engaged in a greater amount of sex trends changed the prevalence, nature, and outside of marriage and their out-of-wedlock outcomes of adolescent sexual activity in the birth rate rose (U.S. Department of Health United States. In the past hundred years, for and Human Services, 1995). Furthermore, example, the average age of menarche and when the birth rate for adults 20 and over spermarche decreased, while the average age rose or fell, the birth rate for adolescents typ- of marriage substantially increased. In addi-. ically did the same (Males, 1993). tion, societal values about sex and even pregnancy and childbearing before mar- The rest of this chapter details specific riage changed considerably. As a result of aspects of these changing trends among these and other factors, many young people teens: sexual activity itself, use of contracep- began having sexual intercourse at an tion, pregnancies and births, and STDs, increasingly early age during the 1970s and including HIV. the 1980s. Consequently, the widening gap between puberty and initiation of intercourse Adolescent Sexual Activity on the one hand, and marriage on the other hand, led to a larger percentage of youth In 1995, the median age of menarche who were sexually experienced at any given for females was 12.6 years and the median 'age, a greater number of acts of intercourse age of marriage was 25.1 (Alan Guttmacher before marriage, a greater number of sexual Institute, 2000). For males, the median ages partners before marriage, and higher rates of for spermarche and marriage in 1988 were unintended pregnancy, out-of-wedlock 14.0 and 26.5 respectively (Alan Guttmacher births, and sexually transmitted diseases Institute, 1994). Thus, for both male and (STDs), including HIV. females, 12.5 years elapse on average

Emerging Annniers: Research Findings on Programs to Reduce Teen Pregnancy

1 3 4 between the time when they become fertile Lindberg, et al., 1998). More recent data are and their sexual feelings intensify and the available from the Youth Risk Behavior time when they marry, creating a long Survey (YRBS), collected from national sam- period during which they need to avoid ples of high school students. According to unintended premarital pregnancy or STDs those data, among female students, the per- either through abstinence from sex or the centage who had ever had sex increased from use of contraception. 48 percent in 1990 to 52 percent in 1995 and then decreased to 48 percent again in The proportion of teens who have ever 1999, while among males, the percent who had sexual intercourse increases steadily with had ever had sex decreased continuously ag. e In 1995, among girls, the percentage from 61 percent in 1990 to 52 percent in wl-co had ever had sex varied from 25 percent 1999 (CDC, 1995; CDC, 1996; CDC, among 15-year-olds to 77 percent among 2000). Regardless of the data source, it is 19-year-olds, while among males, it varied clear that the percent of teen females and from 27 percent among 15-year-olds to 85 males who have ever had sex has stopped percent among 19-year-olds (Moore, increasing, and, especially among males, Driscoll, & Lindberg, 1998). Among stu- the percentage has actually decreased dents in grades 9-12 across the U.S. in 1999, (Santelli et al., 2000). 50 percent reported ever having had sexual intercourse. This ranged from 39 percent When teenage girls who had not had among 9th graders to 65 percent among 12 sex were asked to choose reasons for abstain- graders (CDC, 2000). ing from sex, the most common reason was that having sex would be against their reli- Rates of sexual activity also vary with gious or moral values. Other common race and ethnicity, although much of this choices were: they wanted to avoid preg- variation disappears if one controls for nancy; they wanted to avoid getting an STD; poverty and other forms of disadvantage. In and they hadn't met the right partner 1999, among high school students, 71 per- (Moore, Driscoll, & Lindberg, 1998). cent of African-Americans, 54 percent of Hispanics, and 43 percent of whites had ever When very young girls do have sex, had sex (CDC, 2000). many report that it was either involuntary or unwanted. For example, in 1995, among girls The percentage of all teenagers who who were 13- or 14-years-old when they first had ever had sex by any given age increased had sex, 8 percent reported their first sexual substantially during the 1970s and 1980s, experience was involuntary, and an additional but then stabilized during the 1990s. And, 31 percent indicated it was unwanted among males, the percentage who had ever (Moore, Driscoll, & Lindberg, 1998). had sex actually decreased during the 1990s. According to data from the National Survey Among male and female teenagers who of Family Growth, the percentage of 15- to have ever had sex, many only do so sporadi- 19-year-old females who had ever had sex cally. Sexually experienced unmarried adoles- increased from 29 percent in 1970 to 53 cents have sex during two-thirds of the percent in 1988 and then decreased non- months each year on average, but one- significantly to 51 percent in 1995 (Flanigan, quarter of these adolescents have sex during 2001; Moore, Driscoll, & Lindberg, 1998). fewer than half the months in a year (Alan Between 1988 and 1995, the percent of Guttmacher Institute, 1994). Among high males who had ever had sex decreased from school students in 1999, 50 percent had ever 60 percent to 55 percent (Sonenstein, Ku, had sex, but only. 36 percent had sex during

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 14 4. 5 the previous three months (CDC, 2000). or more years older (Abma & Sonenstein, Because of this sporadic sexual activity, 2001). Among 15- to 19-year-old sexually adolescents often do not plan to have sex on experienced females, 39 percent had their a particular occasion, but sometimes do so first sexual experience with a male three or anyway. more years older (Moore, Driscoll, & Lindberg, 1998). When girls have sex with Most sexually experienced teenagers do much older males, the chances are greater not have sexual intercourse with more than that their first sexual experiences are involun- one sexual partner during any given period tary or unwanted (Abma, Driscoll, & Moore, of time that is, they most commonly prac- 1998). tice premarital serial monogamy (Alan Guttmacher Institute, 1994). Furthermore, Use of Contraception about 70 percent of sexually experienced teen females and 54 percent of sexually expe- Most sexually experienced teenagers use rienced teen males have either zero or one contraception, at least some of the time. sexual partner each year. However, 13 per- About 71 percent of 15- to 19-year-old cent of sexually experienced females and 20 females used contraception the first time they percent of sexually experienced males have had sex, and 68 percent reported using con- three or more partners each year and, there- traception the last time they had sex fore, are at greater risk, especially of STDs (Flanigan, 2001). (Moore, Driscoll, & Lindberg, 1998). For most sexually active teenagers, their numbers The apparent trends over time in con- of partners accumulate over time, so if they traceptive use among teens depend upon the initiate intercourse at early ages, they have a measure of contraceptive use examined, the greater number of sexual partners before exact time period, and the data sets exam- marriage. ined. Data from the National Survey of Family Growth suggest that among female At least in part because they initiate teenagers there has been a large increase in intercourse earlier, African-American and contraceptive use at first sex from 48 percent Hispanic teenagers have more sexual partners in 1982 to 65 percent in 1988 and 71 per- than white teenagers by any given age. For cent in 1995. Among 15- to 19-year-old example, among all high school students in females who had sex in the previous three 1999, 16 percent had previously had sexual months, the proportion who used contracep- intercourse with four or more sexual part- tion at most recent sex decreased between ners. However, 34 percerit of non-Hispanic 1988 and 1995. Contraceptive use in the black, 17 percent of Hispanic, and 12 per- month of interview among teen girls at risk cent of non-Hispanic white students had of unintended pregnancy remained constant four or more partners (CDC, 2000). between 1988 and 1995 at 77 percent (Darroch & Singh, 1999; Flanigan, 2061; Some sexually active teens have sexual Terry & Manlove, 2000). partners close to their own ages, but not all do. For instance, among teen males with African-American and white female recent sexual partners, one-quarter have teens are equally likely to use contraception female partners who are the same age, 46 at their most recent act of sex, but Hispanic percent have partners 1-3 years younger, and female teens are much less likely to do so. 22 percent have partners 1-3 years older. According to the 1995 National Survey of Only 2 percent have partners 4 or more Family Growth, 71 percent of white teen years younger, and 5 percent have partners 4 females, 70 percent of African-American teen

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy I 5

rt_ C. (.) females, and 53 percent of Hispanic teen the less likely they are to use condoms (Ku, females used one or more methods of con- Sonenstein, & Pleck, 1994). traception during their last act of intercourse (Terry & Manlove, 2000). In 1999, among Like some adults, many teenagers do male and female high school students, 55 not consistently use contraceptives properly, percent of whites, 70 percent of African- thereby exposing themselves to risks of preg- Americans, and 55 percent of Hispanics used nancy or STDs. For example, among 15- to a condom the last time they had sex and 19-year-old girls relying upon oral contracep- thereby had some protection against some tives as their main contraceptive, only 70 per- STDs (CDC, 2000). cent took a pill every day (Abma, Chandra, et al., 1997). Among 15- to 19-year-old girls Condoms and oral contraceptives are relying on only coitus-dependent methods of the two most commonly used methods of contraception, only 62 percent used the contraception. Since the mid-1980s, the use method during every act of intercourse of condoms has increased (in part because of (Abma, Chandra, et al., 1997). Similarly, the HIV epidemic), while the use of oral among sexually experienced teen boys using contraceptives has declined (Darroch & condoms, only 45 percent used a condom Singh, 1999). However, small but increasing during every act of intercourse in the last year percentages of teens use hormonal contra- (Sonenstein, Ku, et al., 1998). ceptives such as Depo-Provera or Norplant. These long-acting methods were not avail- When adolescents were asked why they able before the 1990s, but, by 1995, they did not use contraception when they had accounted for 9 percent of contraceptive use sex, one of the most frequent responses is (Flanigan, 2001). that they did not expect or plan to have sex and, therefore, were not prepared (Kirby, Condoms are the most commonly used Brener, et al., 1999; Kirby, Waszak, & method of contraception at first sex. For Ziegler, 1989; Princeton Survey Research example, in 1995, 66 percent of 15- to 19- Associates, 1996). Adolescents say far less year-old sexually experienced females had frequently that they can't afford birth con- used a condom the first time they had sex trol, don't know where to get it, can't get it, (Moore, Driscoll, & Lindberg, 1998), and or don't know how to use it. 94 percent had ever used a condom (Abma, Chandra, et al., 1997). Pregnancies and Births

However, the use of condoms declines The U.S. teen pregnancy rate is very with age and sexual experience, and the use high. In 1996, among all females aged 15- of oral contraceptives increases. In 1995, 19, about 97 per 1,000 became pregnant condom use at last sex decreased from 78 (Henshaw, 1999). The rate is higher for 18- percent among 16-year-old boys to 63 per- to 19-year-olds (153 per 1,000) than for 15- cent among 19-year-old boys (Sonenstein, to 17-year-olds (62 per 1,000). For both age Ku, et al., 1998). Data from the 1999 YRBS groups combined, these pregnancy rates rep- reveals similar trends with age (CDC, 2000). resent about 880,170 teenage pregnancies Condoms are used disproportionately with annually. In addition, there were about casual partners and less commonly with close 24,830 pregnancies among adolescents 14 or romantic partners (Pleck, Sonenstein, & younger, for a grand total of about 905,000 Swain, 1988). Furthermore, the longer a pregnancies to youth under age 20 sexual relationship between two people lasts, (Henshaw, 1999).

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY I 6 0 The U.S. rate is much higher than The decrease in the 1990s in the preg- those in other western industrialized coun- nancy rate among all youth, both sexually tries with available data. For example, the experienced and inexperienced, probably U.S. rate of 97 pregnancies per 1,000 girls is reflects both a decrease in sexual behavior nearly twice as high as rates for Canada (52) among teens and an improvement in contra- and England and Wales (55), four to five ceptive use among sexually experienced times as high as rates in France (23) and teens. However, because of.a variety of Germany (19), and approximately seven methodological problems and limitations, it times the teen pregnancy rates in Italy, Spain, is difficult to know what proportion of the and the Netherlands, all of which have a rate decrease in pregnancy is due to less sexual of 14 per 1,000 (Flanigan, 2001; Singh & intercourse and what proportion is due to Darroch, 2000). improved contraceptive use (Darroch & Singh, 1999; Flanigan, 2001). The overall 1996 teen pregnancy rate has returned to about the same level as it Although roughly 9 percent of females was in the early 1970s, when it was first aged 15-19 become pregnant each year, the measured (Darroch & Singh, 1999). cumulative proportion of any cohort of teen However, it has fluctuated considerably dur- females who become pregnant increases during these last three decades. In 1972, the ing each year of their lives. Thus, more than rate was 95 per 1,000. It then increased 40 percent of girls in the United States gradually to 111 per 1,000 in 1980, become pregnant before they reach 20 years decreased slightly to 107 per 1,000 in 1987, of age, and many become pregnant a second increased rapidly to 117 per 1,000 in 1990, time before their twentieth birthday and then decreased to 97 per 1,000 in 1996 (National Campaign to Prevent Teen (Henshaw, 1999). Pregnancy, 2001).

These changes over time in U.S. preg- The.U.S. teen pregnancy rates vary nancy rates have roughly paralleled changes considerably by race and ethnicity. VVhereas in some other western countries (e.g., the pregnancy rate in 1996 for non-Hispanic England and Wales) (Singh & Darroch, white 15- to 19-year-old girls was 66 per 2000). Furthermore, the decrease in the 1,000, it was 179 per 1,000 among African- U.S. pregnancy rate since 1990 has roughly Americans and 165 per 1,000 among paralleled declines in pregnancy rates in Hispanics (Darroch & Singh, 1999). Once numerous other western countries, suggest- again, however, much of this variation in ing that similar forces may be at work. teen pregnancy rates reflects differences in levels of poverty. The increases in pregnancy rates during the 1970s and 1980s reflected the increases While the teenage pregnancy rate is, by in the proportion of 15- to 19-year-old definition, based upon female teenagers, this females who engaged in sexual intercourse. does not mean that the males involved in Among sexually experienced females aged 15- these pregnancies are teenagers. In 1994, 51 19, the pregnancy rate has decreased from percent of the male partners of teen girls 254 per 1,000 in 1972 to 197 per 1,000 in who became pregnant were within two years 1995 (Alan Guttmacher Institute, 1994; of their partners' ages, 29 percent were 3-5 Darroch & Singh, 1999). This decrease years older, and 19 percent were six or more reflects, in large part, greater use of contra- years older (Darroch, Landry, and Oslak, ception among sexually experienced youth 1999). Similarly, in 1994, while the preg- (Darroch & Singh, 1999). nancy rate for 15- to 19-year-old females was

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 17

4; 110 per 1,000, the corresponding rate for whites, it was 81 per 1,000 for African- teen males was only 52 per 1,000 (Darroch, Americans and 93 per 1,000 among Landry, & Oslak, 1999). This difference in Hispanics (Ventura et al., 2001). pregnancy rates between the two genders was large for both 15- to 17-year-old teens The overall teen birth rate in 1999 was (86 per 1,000 females versus 29 per 1,000 an all-time low (Ventura et al., 2001). males) and for 18- to 19-year-old teens (146 However, this obscures important trends. per 1,000 females versus 87 per 1,000 males). Birth rates declined quite dramatically between the mid-1950s and 1976, further In 1994, of all teenage pregnancies that decreased ever so slightly until 1987, did not end in miscarriages, about 78 per- increased rapidly until 1991, and then cent were unintended (Henshaw, 1998). decreased again. Recent decreases occurred That year, about 35 percent of teenpregnan- initially among second births to teens and cies were terminated by abortion, 43 percent then among first births as well. They also resulted in unintended births, and 22 percent resulted in intended births (Henshaw, occurred among all three major racial/ethnic 1998). The percent of pregnancies resulting groups, but the decreases between 1991 and in births decreased during most of the 1999 were greatest among blacks (30 per- 1970s, remained stable during most of the cent decline), second greatest among non- 1980s, and increased somewhat during the Hispanic whites (22 percent decline) and late 1980s and early 1990s (Henshaw, least among Hispanics (12 percent decline) 1999). Consistent with these trends, the per- (Ventura et al., 2001). cent of teen pregnancies ending in abortion also decreased between 1987 and 1994 Between 1992 and 1997, the rate of (Henshaw, 1998). second births among teen mothers declined even more than the rate of first births Consistent with the high pregnancy among teenagers. However, at 174 second rate, the teen birth rate is also very high. The births per 1,000 teen mothers, this rate is birth rate in the United States in 1999 was still very high (Moore et al., 1999). The 50 births per 1,000 15- to 19-year-old first birth rate for teens was 39 per 1,000 in females (Ventura et al., 2001). It is higher 1999, the lowest level since 1985 (Ventura for 18- to 19-years-olds (80 per 1,000) than et al., 2001). for 15- to 17-year-olds (29 per 1,000). Among unmarried adolescents aged 15 The U.S. teen birth rate is also much to,19, the birth rate rose from 22 per 1,000 higher than that in other western industrial- in 1970 to 40 per 1,000 in 1999 (Ventura et ized countries. In 1995, the U.S. birth rate al., 2001). This represents a steady increase was 57 per 1,000 in contrast to birth until 1994, and a slight decrease since then. rates of 24 per 1,000 in Canada, 28 per Similarly, among mothers under age 20, the 1,000 in England and Wales (the highest in percent of births that occur out-of-wedlock western Europe), and 6 per 1,000 in the Netherlands (the lowest) (Henshaw, 1999; has risen dramatically from 15 percent in Singh & Darroch, 2000). 1960 to 79 percent in 1999 (Ventura & Bachrach, 2000; Ventura et al., 2001). This Like the U.S. teen pregnancy rate, the large increase in non-marital childbearing has U.S. teen birth rate varies considerably by alarmed many people and motivated many race and ethnicity. Whereas the birth rate in efforts to reduce unintended pregnancy 1999 was 34 per 1,000 for non-Hispanic among teens.

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 18 d The increase in non-marital births The birth rate for 15- to 19-year-old reflects, in large part, higher pregnancy rates males was 21 per 1,000 in 1999 (Ventura et among unmarried teens. It also reflects, in al., 2001). As with the rate for teen females, part, less legitimation by marriage that is, it also increased during the late 1980s, there has been a substantial decrease in mar- peaked in 1994, and is now declining riage among pregnant teenage girls. For (Ventura et al., 2001). example, in the early 1960s, 59 percent of first births that were conceived out-of- Consequences of Adolescent wedlock among 15- to 19-year-olds were . Childbearing legitimated by marriage, whereas only 16 percent were legitimated in the early 1990s Poverty and its related ills are clearly (Bachu, 1909). connected with teen pregnancy and childbearing. Poor teens are more likely to get Although 79 percent of teen births are pregnant and have children, and teens who out-of-wedlock, only 29 percent of all out- begin families are more likely to be poor. of-wedlock births are to teenagers (Ventura Therefore, poverty and other manifestations et al., 2001). In fact, the percentage of all of social disorganization can be both the out-of-wedlock births that are to teenagers consequences and the causes of teen preg- has actually declined over time, reflecting the nancy and childbearing. fact that the number of out-of-wedlock births among older women has increased Poverty and various manifestations of even more rapidly than the number among social disorganization are statistically associated with adolescent childbearing, but teenagers. It is also true that half of first out- assumptions about the direction of causality of-wedlock births are to teens. Thus, the pat- have biased estimates of the consequences of tern of giving birth out-of-wedlock often early childbearing. For many years, it was begins during the teen years (National widely believed that poverty and manifesta- Campaign to Prevent Teen Pregnancy, tions of social disorganization were primarily 2001). the consequences of early childbearing. This As is the case with pregnancies among belief overstated the actual impact of teenage teenage females, births among teenage childbearing because, in fact, poverty and females are disproportionately caused by social disorganization were among the causes somewhat older males. Indeed, in 1988, 19 of teenage childbearing as well. percent of the births to 15- to 19-year-old Nevertheless, when adolescents girls were fathered by males six or more years especially 15- to 17-year-old girls give older than their female partners (Alan birth, their future prospects decline in a Guttmacher Institute, 1994), and 50 percent number of ways (Maynard, 1997). They of births to teens aged 15-17 were fathered become less likely to complete school, more by males aged 20 or older (Landry & likely to have large families, and more likely Forrest, 1995). On the other hand, of all to be single parents. They work as much as fathers aged 22-29 years, only 2 percent women who delay childbearing for several fathered children borne by females aged 15- years, but their earnings must provide for a 17 (Lindberg et al., 1997). Because not all larger number of children (Maynard, 1997). 22- to 29-year-old males are fathers, the percentage of 22- to 29-year-old males who The children of teenaged mothers may father children borne by very young females bear the greatest brunt of their mothers' is even smaller. young age. In particular, children born to

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 19 30 mothers aged 15-17 in comparison with one-third of all sexually active young people those born to mothers aged 20 or 21 have become infected by an STD by age 24 less supportive and stimulating home envi- (American School Health Association, 1998). ronments, poorer health, lower cognitive Of course, many have been treated and development, worse educational outcomes, cured, but others have not. Furthermore, higher rates of behavior problems, and approximately one-quarter of all reported higher rates of adolescent childbearing them- cases of STDs occur among adolescents, and selves (Maynard, 1997). about another one-third occur among young Finally, adolescent childbearing leads to adults aged 20-24 (Eng & Butler, 1997). considerable cost to taxpayers and society Rates of STDs are typically much higher more generally. After adjustment for other for African-American and Hispanic teens factors related to teen parenthood, the esti- . mated annual cost to taxpayers of births to than white teens. For example, in 1997, the young women who became mothers when rate of gonorrhea among African-American they were 15-17, instead of 20-21, was at 15- to 19-year-olds was about 24 times least $6.9 billion in 1996. This estimate higher than the rate among white teens includes only five categories of costs: lost tax (Division of STD Prevention, 1998). In part, revenues, increased spending on public assis- these higher rates reflect greater poverty, less tance, health care for the children, foster access to health services, larger numbers of care, and the criminal justice system sexual partners, and possibly differences in (Maynard, 1997). And there are additional reporting by clinics serving low-income costs associated with young women who minority youth (Santelli et al., 1999). gave birth when they were 18- to 19-years- old instead of when they are older. Although Adolescents have the highest age-spe- the cost per child is lower for the older teen cific rates for some STDs, such as chlamydia mothers, there are many more 18- to 19- and gonorrhea (Division of STD Prevention, year-old mothers than 15- to 17-year-old 1998). When data were analyzed by gender mothers. Thus, the aggregate costs to tax- in 1997, 15- to 19-year-old girls had the payers of 18- or 19-year-olds giving birth highest rates of chlamydia and gonorrhea, may be substantial. while the rates for 15- to 19-year-old boys were second only to those for 20- to 24- Sexually Transmitted Diseases year-old young men (Division of STD Prevention, 1998). Teen sexual activity also leads to high rates of sexually transmitted diseases (STDs). Such high rates of STDs among About three million teenagers acquire an teenagers are caused, in part, by the fact that STD every year (Office of National AIDS they are less likely to be married than older Policy, 1996). This means that roughly one sexually active people and therefore have in eight young people between the ages of 13 and 19 and about one in four of those more sexual partners. In addition, they may who have ever had sexual intercourse con- have sex with other partners at higher risk tract an STD every year (Alan Guttmacher and may be less likely to receive health care Institute, 1994). In some geographic areas, for curable STDs. For some STDs, such as rates are much higher. For example, in one chlamydia, adolescent women may also be community, 40 percent of 14- to 19-year-old more physiologically susceptible to infection girls who came to a teen clinic had an STD than older women (Division of STD (Bunnell et al., 1999). In addition, about Prevention, 1998).

THE.NATIONAL CAMpAIGN TO PREVENT TEEN PREGNANCY 20 31 Rates of some curable STDs that have The human and monetary costs of been targeted by STD prevention programs STDs are very high. STDs other than HIV have been reduced among adolescents, just can lead to infertility, ectopic pregnancy, can- as they have been reduced among adults. For cer, and numerous other health problems. example, chlamydia rates among teen women They can also increase the chances of HIV and both gonorrhea and syphilis rates among transmission. The Institute of Medicine esti- both genders have declined. These declines mated that the 1994 monetary costs of have occurred among all three major STDs, other than HIV, among all people, racial/ethnic groups (Division of STD not only adolescents, exceed $10 billion per Prevention, 1998). These data demonstrate year (Eng & Butler, 1997). Sadly, because that it is possible to reduce STDs among HIV can still lead to death and because teens. On the other hand, the prevalence of treatment for HIV and AIDS is so expensive, some incurable STDs has increased among the human and monetary costs of HIV and teens. For example, since the 1970s, the AIDS are extremely high. prevalence of herpes simplex virus type 2 has increased substantially (Fleming et al., 1997). Implications for Pregnancy Prevention Programs In 1996, the Office of National AIDS These patterns of sexual activity and Policy estimated that one-quarter of all new risk-taking among teenagers, as well as their HIV infections occur among young people consequences, point to some important ideas between the ages of 13 and 21. By the end that should shape programs designed to pre- of 1999, 3,725 13- to 19-year-old teenagers vent teen pregnancy: had been reported with AIDS (CDC, 1999). In addition, in the 32 areas with confidential 0Despite the recent declines, teen preg- HIV infection reporting, an additional 4,797 nancy rates remain far too high, both in teenagers were reported to be HIV-positive. comparison with other developed coun- Because of the long and variable time tries and in terms of the human costs to between HIV infection and AIDS, rates of the teens and their children. Thus, we HIV infection provide a more accurate pic- should not become complacenfabout ture of current trends in the epidemic than recent progress, but instead should be rates of AIDS. Among teenagers, a majority encouraged, should try harder, and of these infections occur among girls (56 should build upon our success. percent), and many.occur through hetero- 0 Postponing the initiation of sex or sexual contact (CDC, 1999). Among the returning to abstinence should be an teenagers who were infected with HW but important goal of comprehensive preg- did not yet have AIDS and for whom expo- nancy prevention initiatives. Many youth sure risk was reported, 87 percent of the have sex at an early age, thereby making females and 10 percent of the males con- high pregnancy rates possible. Further- tracted HIV from heterosexual contact. more, the apparent decline in the percent Rates of HIV infection are also higher of youth who had ever had sex in the among African-Americans than among 1990s simply demonstrates that the pre- whites. Through December 1999, among vious trend toward ever earlier initiation young people ages 13 to 19 with HIV, 67 of sex can be halted and possibly percent were African-American youth, while reversed. In addition, abstinence is the only 26 percent were non-Hispanic whites most effective way to avoid teen preg- (CDC, 1999). nancy and represents the only option

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 21 32 for unmarried teens that some people that do not require planning just before can support. or during sex. 0 Comprehensive pregnancy prevention 0 Whenever appropriate, efforts to prevent initiatives should also promote consistent pregnancy should also address STDs and correct use of effective methods of because (1) the costs of STDs, especially contraception for those youth who do HIV, are high, (2) the desire to avoid have sex. Because a majority of youth ini- STDs, like the concern about pregnancy, tiate sex before they complete their teen causes some adolescents to avoid sex or years but many of them do not use con- to use condoms, and (3) abstinence and traception consistently and correctly, pre- to a lesser extent condoms protect vention strategies need to improve against both pregnancy and STDs. In contraceptive use by teens. During the addition, the increase in condom use last few decades, contraceptive' use over time demonstrates that condom use, among sexually experienced youth has like abstinence, can increase. Programs increased, and their pregnancy rates have addressing STDs should emphasize the declined accordingly, demonstrating that risks of sex especially with multiple efforts to improve eontraceptive use can partners over time and the conse- decrease pregnancy rates. In addition, quences of STDs, the effectiveness of contraception is an additional preventive abstinence as prevention, the ability of approach that many people support in condoms to provide considerable protec- combination with an emphasis upon tion against some but not all STDs, and abstinence. the importance of screening and treatment fOr those teens who might possibly oPrograms addressing contraceptive use be infected. should recognize that sexual activity among teens is often sporadic, that youth oBecause the human costs of teen preg- often have sex without planning to do so nancy and STDs are large, reducing ahead of time, and that they therefore do teen pregnancy and STDs should be a not always use contraception. Among national priority. Because the financial other things, youth need to identify the costs of unintended pregnancy, STDs, situations in which they are most likely and HIV are high, especially among to have unplanned and unprotected sex, high-risk youth, even programs that have to learn skills to avoid those situations, a relatively high cost per youth may be and to be knowledgeable about and have cost-effective. access to long-lasting methods of contraception and emergency contraception

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 22 33 References Centers for Disease Control and Prevention. (1998). Trends in sexual risk behavior among Abma, J., Chandra, A., Mosher, W., Peterson, high school students United States, L., & Piccinino, L. (1997). Fertility, family plan- 1991-1997. Morbidity and Mortality Weekly ning, and women's health: New data from the Report, 47(36), 749-752. 1995 National Survey of Family Growth. Vital and Health Statistics, 23(19). Centers for Disease Control and Prevention. (1999). HIV/AIDS Surveillance Report, 11(2), Abma, J., Driscoll, A., & Moore, K. (1998). 1-44. Young women's degree of control over first intercourse: An exploratory analysis. Centers for Disease Control and Prevention. Family Planning Perspectives, 30(1), 12-18. (2000). Youth risk behavior surveillance United States, 1999. Morbidity and Mortality Abma, J., & Sonenstein, F. (2001). Sexual activ- Weekly Report, 49(SS-5). ity and contraceptive practices among teenagers Darroch, J.E., Landry, D.J., & Oslak, S. (1999). in the United States, 1988 and 1995. Vital and Age differences between sexual partners in the Health Statistics Series, 23(21). United States. Family Planning Perspectives, Alan Guttmacher Institute. (1994). Sex and 31(4), 160-167. America's teenagers. New York: Author. Darroch, J.E., & Singh, S. (1999). Why is teenage pregnancy declining? The roles of Alan Guttmacher Institute. (2000). Fulfilling the abstinence, sexual activity and contraceptive use. promise: Public policy and US. family planning Occasional Report, No. 1. New York: Alan clinics. New York: Author. Guttmacher Institute. American School Health Association. (1998). Division of STD Prevention. (1998). Sexually Sexually transmitted diseases in America: How transmitted disease surveillance, 1997. Atlanta: many cases.and at what cost? Menlo Park, CA: Centers for Disease Control and Prevention. Kaiser Family Foundation. Eng, T.R., & Butler, W.T. (eds). (1997). Bachu, A. (1999). Trends in premarital child- The hidden epidemic: Confronting sexually bearing: 1930-1994. Current Population transmitted diseases. Washington DC: National Reports, P23-197. Academy Press.

Bunnell, R.E., Dahlberg, L., Rolfs, R., Ransom, Flanigan, C. (2001). What's behind the good R., Gershman, K., Fashy, C., Newall, W., news: The decline in teen pregnancy rates during Schmid, S., & Stone, K. (1999). High preva- the 1990s. Washington, DC: National Campaign lence and incidence of sexually transmitted dis- to Prevent Teen Pregnancy. eases in urban adolescent females despite moderate risk behaviors. Journarof Infectious Fleming, D.T., McQuillan, G.M., Johnson, RE., Diseases, 180(5), 1624-1631. Nahmias, A.J., Aral, S.O., Lee, F.K., & St. Louis, M.E. (1997). Herpes simplex virus type 2 Centers for Disease Control and Prevention. in the United States: 1976-1994. New England (1995, February 24). Trends in sexual risk Journal of Medicine, 337, 1105-1111. behavior among high school students United States, 1990, 1991, and 1993. Morbidity and Henshaw, S.K. (1998). Unintended pregnancy in Mortality Weekly Report, 44(7), 124-125, 131- the United States. Family Planning Perspectives, 132. 30(1), 24-29, 46.

Centers for Disease Control and Prevention. Henshaw, S.K. (1999). US. teenage pregnancy (1996, September 27). Youth risk behavior statistics with comparative statistics for women surveillance United States, 1995. Morbidity aged 20-24. New York: Alan Guttmacher and Mortality Weekly Report, 45(SS-4). Institute.

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 23 3 Kirby, D., Brener, N.D., Brown, N.L., Pleck, J.H., Sonenstein, F.L., & Swain, S.O. Peterfreund, N., Hillard, P., & Harrist, (1988). Adolescent males' sexual behavior and (1999). The impact of condom distribution in contraceptive use: Implications for male respon- Seattle schools on sexual behavior and condom sibility. Journal of Adolescent Research, 3, 275- use. American Journal of Public Health, 89(2), 284. 182-187. Princeton Survey Research Associates. (1996). Kirby, D., Waszak, C., & Ziegler, J. (1989). An Kaiser Family Foundation survey on teens and assessment of six school-based clinics: Services, sex: What they say teens today need to know and impact, and potential. Washington, DC: Center who they listen to. Menlo Park, CA: Kaiser Family for Population Options. Foundation.

Ku, L., Sonenstein, F.L., & Pleck, J.H. (1994). Santelli, J.S., DiClemente, R.J., Miller, K.S., & The dynamics of young men's condom use dur- Kirby, D. (1999). Sexually transmitted diseases, ing and across relationships. Family Planning unintended pregnancy, and adolescent health Perspectives, .26(6), 246-251. promotion. Adolescent State of the Art Reviews, 10(1), 87-180. Landry, D., & Forrest, J.D. (1995). How old are U.S. fathers? Family Planning Perspectives, Santelli, J.S., Lindberg, L.D., Abma, J., 27(4), 159-165. McNeeley, C.S., & Resnick, M. (2000). Adolescent sexual behavior: Estimates and trends Lindberg, L.D., Sonenstein, F., Ku, L., & from four nationally representative surveys. Martinez, G. (1997). Age differences between Family Planning Perspectives, 32(4), 156-165, minors who give birth and their adult partners. 194. Family Planning Perspectives, 29(2), 61-66. Singh, S., & Darroch, J. (2000). Adolescent pregnancy and.childbearing: Levels and trends in Males, M. (1993). School-age pregnancy: Why developed countries. Family Planning hasn't prevention worked? Journal of School Perspectives, 32(1), 14-23. Health, 63(10), 429-432. Sonenstein, F.L., Ku, L., Lindberg, L.D., Maynard, R.A. (Ed.) (1997). Kids having kids: Turner, C.F., & Pleck, J.H. (1998). Changes in Economic costs and social consequences of teen sexual behavior and condom use among pregnancy. Washington, DC: The Urban teenaged males: 1988 to 1995. American Institute. Journal of Public Health, 88(6), 956-959.

Moore, K.A., Driscoll, A.K., & Lindberg, L.D. Terry, E., & Manlove, J. (2000). Trends in sex- (1998). A statistical portrait of adolescent sex, ual activity and contraceptive use among teens. contraception, and childbearing. Washington, Washington, DC: National Campaign to Prevent DC: National Campaign to Prevent Teen Teen Pregnancy. Pregnancy. U.S. Department of Health and Human Moore, K.A., Papillo, A.R., Williams, S., Jager, Services. (1995). Report to Congress on out-of- J., & Jones, F. (1999). Facts at a glance. wedlock childbearing. Hyattsville, MD: Author. Washington DC: Child Trends. Ventura, S.J., & Bachrach, C.A. (2000). National Campaign to Prevent Teen Pregnancy. Nonmarital childbearing in the United States, (1997). Whatever happened to childhood? The 1940-99. National Vital Statistics Reports, problem of teen pregnancy in the United States. 48(16). Washington, DC: Author. Ventura, S.J., Martin. J.A., Curtin, S.C., Office of National AIDS Policy. (1996). Youth Menacker, F., & Hamilton, B.E. (2001). Births: and HIV/AIDS: An American agenda. Final data for 1999. National Vital Statistics Washington, DC: Author. Reports, 49(1).

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 24 35 Chaipter Looking for Reasons Why: The Antecedents of Adolescent Sexual Behavior

ogic and experience suggest that the but not always, they causally affect that more we know about the causes of behavior. In the field of adolescent sexual 'risky sexual behavior, the more success behavior, common sense sometimes tells us we'll have in designing sound programs to that particular antecedents are, in fact, reduce such behavior. Fortunately, for many causally related to certain sexual behaviors by years, researchers have tried to define the fac- teens. For example, community poverty is an tors in the lives of young people that influ- antecedent of pregnancy as well as a causal ence whether they will have sex, use factor, because community poverty reduces contraception, or become pregnant (or cause employment and career opportunity, which a pregnancy). These factors are often called in turn reduce youths' motivation to pursue "antecedents." Antecedents that increase the education and jobs and to avoid early child- chances of sexual risk-taking and pregnancy bearirig. Conversely, community wealth and are called "risk factors." Those that reduce opportunity increase motivation to pursue the chances are called "protective factors." education and careers and avoid early child- Some antecedents relate directly to sexuality bearing. Similarly, hormonal changes and (e.g., onset of puberty and teens' beliefs puberty are antecedents of sexual initiation, about premarital sex), and others do not and common sense tells us these phenomena (e.g., levels of poverty and parents' attitudes increase young people's sexual desire, about education). increase their sexual attractiveness to others, By definition, all antecedents must be and increase their chances of having sex. correlated with (that is, be associated with) However, sometimes causality (as opposed to the outcome behavior in question (e.g., initi- mere association) is not well-established by ation of sex, use of contraception, or preg- either research or common sense. For exam- nancy), and they must occur before that ple, smoking cigarettes is associated with behavior. Because the antecedents of a early initiation of intercourse, but it is behavior must be correlated with that behav- unlikely to cause early intercourse. Rather, ior and must occur prior to it, then often, youth who engage in a variety of risk-taking

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 25 behaviors may be more likely both to smoke most important antecedents becomes critical. cigarettes and to engage in early sex. If people who develop programs focus on unimportant antecedents, then they are likely Why should we care about the to target youth at lower risk of unprotected antecedents of adolescent sexual and contra- sex and/or their programs will not concen- ceptive behavior, pregnancy, and childbear- trate on those factors known to affect sexual ing? There are two simple and important risk-taking. Such programs will probably be reasons. First, the antecedents can be used to much less effective than if they had been identify particular groups of young people built upon the most important antecedents. who are especially likely to initiate sex early, fail to use contraception effectively, become Clusters of Antecedents pregnant, or bear children. These youth can then be targeted with more intensive inter- The chart on the next page presents ventions to help reduce their sexual risk- some of the most important clusters of taking. It is important to add here that antecedents of adolescent sexual activity, use when identifying the highest risk youth, of contraception, pregnancy, and childbear- antecedents should be reasonably highly ing. It summarizes and simplifies the much correlated with sexual risk-taking behavior, more detailed picture that is presented in the but they need not be causally related. For tables at the end of the chapter(2.1-2.7). example, programs might target youth who (For more information about the methods engage in other unhealthy or delinquent and criteria used to identify and select behavior, such as smoking, drinking and studies, see p. 35 or Kirby, 1999.) The driving, or other illegal activity, because these chart opposite and Tables2.1-2.7reveal that youth may be more likely to engage in sexual a large number of antecedents have been risk-taking. linked to one or more sexual or contraceptive behaviors, pregnancy, and childbearing. Second, knowledge about antecedents In fact, depending on how precisely the should be used to design more effective antecedents are defined, more than 100 dif- interventions. Neither parents nor anyone ferent ones are identified among numerous else can constantly monitor children to research studies and across the seven tables. make sure that they do not engage in unpro- They describe not only individual teens tected sex. In the final analysis, young people themselves (both their biological and psy- make their own decisions about having sex chosocial attributes) but also important peo- and using contraception. However, parents ple and institutions within their environment and other adults need to take steps to influ- their partners, peers, families, schools, ence those factors that are both highly and religious affiliations, communities, and even causally related with adolescent sexual and states as well as the teens' relationships to contraceptive behavior that is, they need them. Together, the antecedents in these to focus upon those factors that make a dif- tables paint a rich, detailed, and complex ference. If they don't, then they are unlikely portrait. to be effective in helping teens delay sex, increase contraceptive use, or reduce their In part because so many antecedents are chances of pregnancy. related to teens' sexual behavior, few antecedents are very highly related to behav- Because antecedents can be used both ior that is, in general, the more factors to identify young people at greater risk of linked to sexual behavior, the weaker the unprotected sex and pregnancy and to design effect of any single one. Rather, most of the more effective programs, understanding the antecedents are weakly or, in some instances,

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 26 3 7 Important Antecedents of Adolescent Sexual Behavior, Use of Contraception, Pregnancy, and Childbearing! Community Teen Community disadvantage and Biological antecedents disorganization Older age and greater physical maturity + High level of education Higher hormone levels High unemployment rate + High income level Ethnicity High crime rate + Being white (vs. black or Hispanic) Family Attachment to and success in school + Good school performance Structure and economic advantage of + Educational aspirations and plans for the future the teenagers' families + Two (vs. one) parents Attachment to religious institutions Changes in parental marital status + Frequent religious attendance + High level of parents' education + High parental income level Problem or risk-taking behaviors Tobacco, alcohol, or drug use Positive family dynamics and attachment Problem behaviors or delinquency + Parental support and family connectedness Other risk behaviors + Sufficient parental supervision and monitoring Emotional distress Higher level of stress Family attitudes about and modeling of sexual Depression risktaking and early childbearing Suicide ideation Mother's early age at first sex and first birth Single mother's dating and cohabitation behav- Characteristics of relationship iors with partners + Conservative parental attitudes about Early and frequent dating premarital sex or teen sex Going steady, having a close relationship + Positive parental attitudes about Greater number of romantic partners contraception Having a partner 3 or more years older Older sibling's early sexual.behavior and age of first birth Sexual abuse History of prior sexual coercion or abuse [Peer Sexual beliefs, attitudes, and skills Peer attitudes and behavior + Conservative attitudes toward + High grades among friends premarital sex Peers' substance use and delinquent + Greater perceived susceptibility to and non-normative behavior pregnancy, STDs/HIV Sexually active peers (or perception thereof) + Importance of avoiding pregnancy, + Positive peer norms or support for childbearing, & STDs condom or contraceptive use + Greater knowledge about contraception + More positive attitudes about contraception Partner + Greater perceived self-efficacy in using Partner attitudes condoms or contraception + Partner support for contraception

) 1 "+" denotes a protective factor; "" denotes a risk factor.

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 27 3 3 moderately related to behavior. And some of cal theories (Miller et al., 1998), theories these antecedents are undoubtedly causally suggesting that sexual risking-taking is part related to each other. In particular, some of of a larger syndrome of risk-taking or deviant the more distal antecedents (i.e., theoretically behavior (Costa et aL, 1995), and sociopsy- "distant" from the teen) affect the more chological theories of rational behavior proximal sexual antecedents (i.e., "closer in" (Fishbein et al., 1991). However, the to the individual and his or her behavior). antecedents summarized in these tables For example, community levels of education, clearly demonstrate that no single theoretical employment, and opportunity (distal perspective is sufficient; the total picture is antecedents) may affect individual teens' much more complex. educational goals, perceived costs of early childbearing, and motivation to avoid early The communities teens live in influence childbearing (proximal antecedents). their sexual behavior. When teens live in Similarly, the sexual beliefs and behaviors of poor communities with less advantage and family members, peers, and parmers opportunity and more disorganization, they undoubtedly affect teens' sexual beliefs and are more likely to engage in sex at an earlier behavior. In other words, many of the more age and to become pregnant. More specifi- distal non-sexual antecedents first affect the cally, when the adults in their communities more proximal sexual antecedents, which in have lower levels of education, are less likely turn affect sexual behavior. to be employed, have lower incomes, and engage in higher rates of crime, then these While nearly all youth are at some point teens are more likely to engage in sex and in their lives at some risk of engaging in sex, become pregnant than teens in communities failing to use contraception effectively, and with higher levels of education and income becoming pregnant, the risk and protective and lower rates of unemployment and crime. factors identified in the tables and discussed In communities with higher levels of educa- below substantially increase or decrease the tion, employment, and income, adults may risks of these events. As noted above, few of place greater emphasis upon obtaining the antecedents are critically important. higher education, pursuing careers, and Some teens with few risk factors and many avoiding early childbearing; they may also protective factors nevertheless engage in provide role models for these behaviors, and unprotected sex; and some teens with many the communities more generally may provide risk factors and few protective factors avoid opportunities for education and careers. This unprotected sex. Nevertheless, as the num- does not mean that teens in communities ber of risk factors in a given teen's life with high levels of education, high employ- increases and as the number of protective ment, high income, and little crime never factors diminishes, that teen's chances of engage in unprotected sex. It simplY means engaging in unprotected sex and becoming they are less likely to do so, or do so less pregnant (or creating a pregnancy) increase. frequently.

These antecedents support a wide vari- Similarly, the characteristics of a teen's ety of theories about adolescent sexual risk- family are important as well in determining taking behavior for instance, theories risk. When teens have parents with low levels involving economic disadvantage and oppor- of education and income, they are more tunity (Billy, Brewster, & Grady, 1994), the- likely to engage in sex at an earlier age, to ories involving parent child-rearing practices fail to use contraception consistently, and to and parent values about adolescent sexuality become pregnant than teens in families with (Jaccard, Dittus, & Litardo, 1999), biologi- high levels of education and income. This

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 28 39 may be due, in part, to the greater emphasis also be related to teen behavior. However, as that more educated and wealthier parents Tables 2.1-2.7 suggest, studies about the place upon obtaining an education, pursuing impact of parent-child communication about a career, and avoiding early childbearing, in these topics are far more mixed some part to the greater resources they may have found it to be a protective factor, while oth- to support the teens' education and career ers did not. A couple of studies even found it planning, and in part to other reasons. to be a risk factor, perhaps because parents may anticipate that their teens are going to When teens live with both parents and initiate sex and then decide to talk to them have close relationships with them, they are about sex and contraception. When their less likely to engage in unprotected sex and teens do initiate sex, it appears that greater become pregnant. More specifically, when communication about sexuality may have teens live with both parents (instead of only contributed to teen sexual behavior, when, in one parent or neither parent), they are less fact, the teens' impending sexual debut led likely to engage in sex, more likely to use to both the parent-child discussions about contraception if they have sex, and less likely sex followed soon by actual initiation of sex. to become pregnant or cause a pregnancy. To complicate matters further, some studies Furthermore, if teens believe they have con- indicate that the strength (or even existence) siderable parental support, feel connected to of the relationship between parent-child their parents, and are appropriately super- communication and sexual risk-taking vised or monitored by their parents, they depends upon the gender of the teen, the become less likely to have unprotected sex gender of the parent, the closeness of the and become pregnant. parent-child relationship, the parents' values, If family members, especially parents, the characteristics of the communication pro- either express values or model behavior con- cess, and other factors (Miller et al., 1998). sistent with sexual risk-taking or early child- So while parents do have an impact upon bearing, then teens are more likely to engage their teens' sexual behavior, the actual causal in unprotected sex and become pregnant. impact of parent-child communication has Parents can do this in a variety of ways, not been fully delineated. including having permissive attitudes about Not surprisingly, peers also influence premarital sex or teen sex, having negative the sexual behavior of teens. When a teen's attitudes about contraception, having sex peers obtain poor grades, are unattached to outside of marriage themselves, cohabitating, giving birth outside of marriage, or having school, and engage in a variety of negative given birth at an early age. Similarly, if the behaviors, then the teen is more likely to , siblings of teens model early childbearing by engage in sex. When peers get good grades, giving birth at an early age, then the teens, are more attached to school, and engage in themselves, are more likely to engage in fewer negative behaviors, a teen is less likely unprotected sex and become pregnant. to become pregnant. Furthermore, when teens believe that their peers are having sex, If parental values about sex and parental they become more likely to have sex them- sexual behavior and childbearing experience selves, and when they believe their peers sup- influence teen behavior, then it would seem port condom or contraceptive use, they plausible that the amount of parent-child become more likely to use condoms or other communication about these topics should contraceptives.' 1 For research on the impact of peers and how to influence that impact, see Peer Potential: Making the Most of How Teens Influence Each Other (Bearman et al., 1999).

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 29 4 0 Turning to the teens themselves,,as ment is important, get good grades, do not everyone knows, the older teens become, the drop out of school, or have plans for higher more likely they are to have sex and to education beyond high school, they initiate become pregnant, reflecting many important sex later, use contraception more effectively, changes that come with increasing age. Some and are less likely to become pregnant or are biological, including physical maturity bear children. In short, such positive feelings and higher hormone levels, which may lead increase motivation to avoid risky sexual to a greater desire for love, intimacy, and sex behavior. Attachment to faith communities or to greater sexual attractiveness. Others are may also reduce the chances that teens will social for instance, more pressure from engage in sex, although the evidence sup- others to have sex; changes in perceived porting this relationship is less strong. Teens norms about sexual and contraceptive behav- who describe themselves as more religious, ior; and the increased opportunity to have who attend religious services more frequently, sex that comes with greater freedom and and have a stronger religious affiliation are independence more generally. While these less likely to initiate sex by any giyen age. biological antecedents associated with older However, the direction of causality is not age cannot be changed by programmatic entirely clear, for just as attachment to faith interventions, they can be used to identify communities may affect sexual behavior, sex- higher risk youth. Furthermore, some of the ual behavior may also affect attachment to social factors associated with being older faith communities. For example, teens who (e.g., greater pressure to have sex) may be have had sex may feel less comfortable in amenable to change. their churches, synagogues, or mosques and be less likely to attend services. As the previous chapter noted, both African-Americans and Hispanics have sex at Using alcohol and drugs, engaging in an earlier age than whites and are also more other problem or risk-taking behaviors, and likely to become pregnant and bear children suffering from emotional distress (including at an earlier age. This reflects, in large part, depression) all increase teens' chances of differences in community levels of education, engaging in unprotected sex and becoming poverty, and opportunity, rather than race or pregnant. These antecedents are significantly ethnicity per se. When family and community related to one another and may represent characteristics associated with race and ethnic- either more general personality traits or ity are held constant, the impact of race and exposure to higher risk environments. There ethnicity diminishes considerably but not are at least two common interpretations for entirely in these studies. Differences in the relationship between substance use and cultural values help explain their residual dif- risky sexual behavior: (1) they are all part of ference. For example, Hispanic families may a general inclination to take risks and of an give greater importance to family and may be environment that supports such behavior, more accepting of early childbearing than and (2) drug and alcohol use diminishes non-Hispanic white families. both inhibitions and rational decision-making, thereby actually increasing the incidence Schools and the relationships that of unprotected sex. Both interpretations are teens have with schools also influence probably valid. teens' sexual behavior. Attachment to school and success in school reduce the chances that Not surprisingly, early romantic involve- teens will engage in unprotected sex and ment increases the chance of early sexual become pregnant. When teens feel close to activity. When teens begin dating at an early their schools, believe that academic achieve- age, date frequently, have a greater number

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 30 of romantic partners, and go steady at an and childbearing, are less motivated to use early age, they become more likely to initiate contraception, and do not always use contra- sex at an earlier age. In part, these early ception, then they are more likely to become romantic relationships may provide both pregnant. These sexual beliefs, attitudes, and greater opportunity and greater pressure to skills can together be called the "sexual psy- have sex. Furthermore, sex within a romantic chosocial" antecedents of teen sexual risk- relationship may be more consistent with taking behavior. teens' values and perceived norms than sex in casual relationships. When the romantic part- These sexual psychosocial antecedents ner of a teen is three or more years older, are particularly important for four reasons. then the teen is especially likely to engage in First, they are well-supported by sociopsy- sex. The impact of this antecedent is quite chological theory (Fishbein et al., 1991). large, especially among middle school youth. Second, they are very proximal to sexual and contraceptive behavior that is, they are Prior sexual abuse is also an important more directly linked conceptually with sexual risk factor for early initiation of sex, poor behavior than other non-sexual antecedents, contraceptive use, pregnancy, and childbear- such as community level of poverty. Third, as ing. Youth who have been sexually abused noted above, they are more highly related to undoubtedly have been exposed to a variety some of the sexual and contraceptive behav- of risk factors. In addition, the past sexual iors than most of the other antecedents. abuse may warp what teens understand as And, finally, as will be demonstrated later in appropriate sexual and contraceptive behav- this report, some of these sexual psychosocial ior and may reduce their ability to refrain antecedents have formed the theoretical basis from sex or to use contraception. for most of the sex and HIV education pro- Finally, teens' own sexual beliefs, atti- grams that have succeeded in reducing sexual tudes, and skills affect their sexual behavior, risk-taking. For all these reasons, sexual psy- of course. In.fact, these antecedents in gen- chosocial antecedents should be an impor- eral are the most strongly related to sexual tant focus of certain types of pregnancy behavior. When teens have permissive atti- prevention and STD/HIV prevention pro- tudes toward premarital sex, perceive per- grams, particularly sex and HIV education sonal and social benefits and few costs to programs. having sex, do not care if their friends know they have had sex, lack the confidence in Summary Themes their ability to avoid sex, have less concern about pregnancy or STDs, and intend to At least three overarching observations have sex, then, not surprisingly, they are can be made about all these clusters of more likely to engage in sex. When sexually antecedents. First, teens, like all people, are active teens do not accept the fact that they strongly influenced by their physical and are having sex, do not perceive they are at social environments by their families risk of pregnancy, do not perceive that preg- (especially their parents), by their peers and nancy would have a negative effect on their friends, by their romantic partners (if they lives, are less knowledgeable about contra- have any), by their schools and faith commu- cePtion, and have more negative attitudes nities, and by their communities more gener- toward contraception, then they are less ally. All of these groups influence teens by likely to use contraception. Similarly, when emphasizing particular beliefs and norms, teens initiate sex earlier, have more sexual modeling certain behaviors, providing partners, are ambivalent about pregnancy opportunities for particular behaviors, and

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 3 4r' sometimes directly applying pressure to Limitations of These Research engage or not engage in specific behaviors. Findings

Second, a substantial proportion of all It would be ideal if one could estimate the risk factors involve some form of disad- accurately the relative causal effects of each of vantage, disorganization, or dysfunction the antecedents on behavior. Unfortunately, disadvantaged communities with high unem- this is not possible for at least three method- ployment, low income, and high crime rates; ological reasons: (1) most studies focus on poorly educated, low-income families with only a small subset of the antecedents and are only one parent; parents who provide insuffi- therefore limited in their ability to assess the cient support for and monitoring of their relative importance of all antecedents; (2) teens; friends who do poorly in school, use when studies measure only one aspect of drugs, and engage in unprotected sex; and poverty or social disorganization, the esti- teens themselves who are not attached to mated effect is increased by the other mani- family, school, or church and are emotionally festations of poverty or social disorganization distressed, use drugs, and have been sexually with which the measured aspect is correlated; coerced or abused. and (3) few studies can demonstrate causality very well. Therefore, accurate estimates of the Third, attachment to people or groups relative causal effect of each antecedent are who express protective values and model not available. positive behaviors reduces sexual risk-taking. When youth are more strongly attached to It is also true that the magnitude of the their parents, to their schools, or to their effect of particular antecedents may vary at faith communities, they are less likely to different times. Our society, is changing engage in unprotected sex. In general, par- rapidly. In recent years, for example, there ents, schools, and faith communities discour- has been considerable immigration from age youth from engaging in sex in the first other countries, greater employment oppor- place and then also discourage unprotected tunity during an economic expansion, wel- sex. Consequently, when youth are more fare reform, more widespread use of the strongly attached to their parents, schools, or internet, greater knowledge about the spread and risks of HIV/AIDS, a more conservative faith communities, they are more likely to climate regarding sexuality, and greater behave consistently with these values. For knowledge about and access to emergency example, teens who belong to either contraceptives and long-acting contracep- Catholic or fundamentalist Protestant tives. These changes and others affect the churches, which have strong norms against relative importance of these antecedents at premarital sex, are less likely to engage in sex different points in time. than youth who belong to other Protestant churches that have less strict norms. Furthermore, different antecedents may Nonetheless, attachment is not always a be important for different groups of youth. good thing. Being very popular with peers For example, among middle-class youth with (who often have less conservative attitudes ample educational and employment opportu- about sex than parents, schools, or faith nities, the most important antecedents may communities) and having high-risk peers are be skills to resist peer pressure to have sex or both associated with earlier initiation of sex. use drugs or alcohol. Among youth in disad- Therefore, it is attachment to individuals or vantaged communities, the most important groups of people who express and model risk factors may be lack of educational and low-risk norms that is protective. economic opportunity, little belief in the

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 32 4 'D future, and little motivation to avoid early ethnicity, income level, school perfor- childbearing. Among girls, the sexual norms mance, and engagement in other risk- and values of peers may have a greater taking behavior), they can be used to impact than among boys. identify teens most in need of help.

These important caveats do not mean O In addition, these antecedents should that people designing programs should not provide the basis for developing inter- focus on antecedents. What they do mean is ventions. Programs should focus upon that people designing programs should be those antecedents that (1) are the most aware of these limitations and to the extent highly and causally related to sexual risk- feasible should identify through their own taking behavior and (2) are amenable to research the important antecedents of sexual change by the program. After all, pro- risk-taking for the group of youth they are grams are less likely to change behavior if targeting with their particular programs. the selected antecedents have no impact upon behavior, or if the programs cannot Implications of Theories and change the important antecedents Antecedents for Pregnancy selected. Prevention Programs oGiven that there are a large number of antecedents linked to teen pregnancy, Taken together, these theories and each one with only a modest impact on antecedents related to adolescent sexual sexual behavior, creating powerful pro- behavior have strong implications for the grams to reduce pregnancy is very chal- design and targeting of pregnancy prevention lenging. The fact that many important programs: antecedents are manifestations of social 0 On the one hand, most youth are at risk disadvantage and disorganization (which of unprotected sex and pregnancy. Given are difficult to change) makes the chal- that about four-fifths of young people lenge even greater. Consequently, it is have sex while still in their teens and not likely that there are any simple, easy- given that many do not always use con- to-implement prevention programs doms and other forms of contraception "magic bullets" that will substantially consistently or effectively, pregnancy and change adolescent sexual behavior and STDs are real risks in the lives of most pregnancy rates. Few programs, after all, teens. Thus, all teens need appropriate can modify more than a few risk or pro- education about the value of delaying sex tective factors at a time. as well as accurate information about O To reduce pregnancy markedly, commu- contraception. And all teens who become nities may have to address many sexually active need access to reproduc- antecedents among different groups tive health services. (e.g., teens, their families, schools, and 0 On the other hand, some teens are at communities), and they may have to much greater risk than others, and address both the sexual and non-sexual understanding the antecedents outlined antecedents. In practice, at the commu- here can help programs target high-risk nity level this may mean that a patchwork teens with more intensive and effec- of programs can be effective, if each tive interventions. Because some of addresses a specific set of antecedents the more important antecedents are that in the aggregate improve most of readily measurable (e.g., gender, age, the most important antecedents.

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 33 4 oBecause teens' sexual beliefs, attitudes, more responsible models of sexual perceived norms, confidence in their abil- behavior by parents, teens, and siblings. ities, intentions, and actual skills are more strongly related to their sexual and Because some non-sexual antecedents are contraceptive behavior than most other modestly related to pregnancy, and non-sexual antecedents, and because because some of them can also be modi- these sexual psychosocial antecedents can fied by programs, they should also be be modified, pregnancy prevention addressed by some interventions. initiatives should include sex and HIV O Given that both sexual and non-sexual education programs and other sexuality- antecedents need to be addressed, adults focused interventions that address these antecedents as effectively as possible. and community organizations that both Given that teens' sexual beliefs, attitudes, do and do not address sexuality directly and behavior are affected by the model- can make an important difference in ing of sexual and contraceptive behavior helping teens avoid pregnancy. In other and pregnancy by parents, siblings, and words, while focusing on sexual peers, pregnancy prevention initiatives antecedents may be the most direct and should also encourage these groups to effective way to reduce the risk of teen model appropriate sexual behavior, as pregnancy, it is not required of every well as encourage the media to present program. There are other ways to help.

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 34 4 5 C lapter Motes Criteria for the Inclusions of Studies on Antecedents

The following seven tables present reviews (e.g., Miller, 1995; Moore et al., antecedents that selected research studies 1995; Santelli & Beilenson, 1992). have found to be related to different adolescent sexual and contraceptive behaviors, Space considerations and the large pregnancy, and childbearing. Those number of studies reviewed prevented the antecedents were identified in more than inclusion of every reference for every signifi- 250 studies that met the following criteria: cant antecedent for every study in the tables. Four criteria guided the inclusion of refer- flMet the scientific standards requisite ences in the tables: (1) if a study included for inclusion in professional journals or multivariate analyses, the variables found sig- publications. nificant in the multivariate analyses were

O Published in 1980 or later. included, but those found significant in only bivariate analyses were not; (2) if a data set O Analyzed data collected from U.S. had a large and representative sample, all the adolescents, most of whom were 19 or significant antecedents in these multivariate younger. analyses were included; (3) if a given O Used a sample size of at least 100. antecedent had only a few references supporting it, all were included, even if the sam- O Measured the relationship between the ple was not nationally representative; and (4) antecedents and one or more of the fol- if more than a few studies found an lowing sexual behaviors: initiation of sex, antecedent significant, then more than a few frequency of sexual intercourse, number (but not all) references for that antecedent of sexual partners, use of condoms, use were included. of any type of contraception, pregnancy, or childbearing. (Studies that measured These tables were designed to provide only out-of-wedlock pregnancy or child- an overall picture of the antecedents of sex- bearing were not included.) ual and contraceptive behaviors, pregnancy, The procedures employed to identify and childbearing, not to be a thorough anal- these antecedents are described more fully ysis of any particular antecedent. Thus, a elsewhere (Kirby, 1999). These tables also variable was included as an antecedent if one benefitted considerably from previous or more studies found it to be significantly

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 35 46 related to any of the outcome behaviors, a particular point in time, or might be signif- even if other studies failed to do so. By icant only when the antecedent is measured reporting only the relationships that were in a particular way or when other variables significant, a bias is, of course, introduced were (or were not) controlled in the study. and this bias only compounds the natural And, of course, it is probable that a few of tendency of authors to report relationships the antecedents were found to be significant that are significant and not to report those only because of chance. On the other hand, that are not. Consequently, some of the vari- many of the antecedents were found to be ables included in the tables might be signifi- significant in multiple studies, thereby repli- cant antecedents for only particular groups cating their importance across multiple of youth, might have been significant at only groups of adolescents at different times.

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 36 4 7 Table 2.1: Antecedents of Initiation of Sex Environment/Context

Community: +Rural area (Lammers et al., 2000) +/Higher percent black or Hispanic vs. white (Billy, Brewster; & Grady, 1994; Brewster; Billy, & Grady, 1993; Gibbs, 1986) Higher percent foreign born (Billy, Brewster; & Grady, 1994) Higher percent with college education (Brewster; Billy, & Grady, 1993) Higher divorce rates (Brewster; Billy, & Grady, 1993) Higher rate of residential turnover (Brewster; Billy, & Grady, 1993; Gibbs, 1986) Higher percent of working females working full time (Billy, Brewster; & Grady, 1994) Higher unemployment rate (Brewster; 1994) Higher family income (Brewster; Billy, & Grady, 1993) Higher percent religious adherents (Billy, Brewster; & Grady, 1994) Higher crime rate (Billy, Brewster; & Grady, 1994) Greater neighborhood monitoring by adults in community (Small & Luster; 1994) Greater number of clinics (Brewster, Billy, & Grady, 1993) Higher teen non-marital birth rate (Brewster, Billy, & Grady, 1993)

School:

. +Parochial school (Resnick et al., 1997)

Religious Institution: +Catholic or fundamentalist Protestant vs. Protestant and other (Brewster et al., 1998)

Family: Higher parental education (Billy, Brewster; & Grady, 1994; Brewster; 1994; Brewster et al., 1998; Forste & Heaton, 1988; Grady, Hayward, & Bill, 1989; Hamard, Grady, & Billy, 1992; Lauritsen, 1994; Miller et al., 1997; Moore, Morrison, & Glei, 1995; Sante Ili et al., 2000; Zelnik, Kantner; & Ford, 1981) + Two (vs. one) parents (Afxentiou & Hawley, 1997; Bearman & BrOckner; 1999; Billy, Brewster; & Grady, 1994; Blum et al., 2000; Brewster; 1994; Brewster et al., 1998; Day, 1992; Dorius & Barber; 1998; Flewelling & Bauman, 1990; Forge & Heaton, 1988; Hogan & Kitagawa, 1985; Inazu & Fox, 1980; Ku, Sonenstein, & Pleck,I 993a; Lammers et al., 2000; Utile & Rankin, forthcoming; Lock &Vincent, 1995; Meschke et al., 2000; Miller & Bingham, 1989; Miller et al., 1997; Moore, Morrison, & Glei, 1995; Murry, 1992; Newcomer & Udry, 1987; Raine et al., 1999; Rodgers, 1983; Sante Ili et al., 2000; Thornton & Camburn, 1987; Udry & Billy, 1987; Upchurch et al., 1998; Whitbeck, Simons, & Kao, 1994) Divorce or change to single-parent household (Capaldi, Crosby, & Stoolmiller 1996; Devine, Long, & Forehand, 1993; Dorius, Heaton, & Steffen, 1993; Miller et al., 1997; Newcomer & Udry, 1987; Wu & Martinson, 1993) Working mother during ages 5-15 (Billy,.Brewster; & Grady, 1994; Mott et al., 1996) Higher income level (Afxentiou & Hawley, 1997; Blum et al., 2000; Brewster; Billy, & Grady, 1993; Hayward, Grady, & Billy, 1992; Hogan & Kitagawa, 1985; Inazu & Fox, 1980; Lammers et al., 2000; Little & Rankin, forthcoming) Intergenerational receipt of welfare (Moore, Morrison, & Glei, 1995) continued + = a protective factor; = a risk factor; +/ = a protective factor in some studies and a risk factor in others

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 37 4 3 Table 2.1: Antecedents of Initiation of Sex continued Family: continued Health insurance (Bearman & BrUckner, 1999) Greater number of siblings (Benson &Torpy, 1995; Hogan & Kitagawa, 1985) Being a younger sibling (Haurin & Mott, 1990; Rodgers, 1983; Rodgers & Rowe, 1988; Widmer, 1997) Greater family religiosity (Billy, Brewster, & Grady, 1994; Inazu & Fox, 1980) Recent family suicide attempts (Resnick et al., 1997) Older mother's age at first sex (Mott et al., 1996; Newcomer & Udry, 1984). Older mother's age at first birth (Ku, Sonenstein, & Pleck, 1993a; Newcomer & Udry, 1984; Widmer, 1997) Single mother's dating behavior (Whitbeck, Simons, & Kao, 1994) Single mother's cohabitation (Inazu & Fox, 1980) An older sibling who had sex (East & Felice, 1992; East, Felice, & Morgan, 1993; Haurin & Mott, 1990; Widmer, 1997) An older sister who gave birth as an adolescent (East, Felice, & Morgan, 1993; East,I 996a; East,I 996b; Hogan & Kitagawa, 1985) Conservative parental attitudes about teen or premarital sex (Baker,Thalberg, & Morrison, 1988; Blum, Buehring, & Rinehart, 2000; CvetkOvich & Grote, 1981; Dittus & Jaccard, 2000; Gibson & Kempf, 1990; Halpern et al., 2000; Hove II et al., 1994; Jaccard & Dittus, 1991; Jaccard, Dittus, & Gordon, 1996; Jaccard, Dittus, & Litardo, 1999; Little & Rankin, forthcoming; Resnick et al., 1997; Shah & Zelnik, 1981;Thomson, 1982; Weinstein &Thornton, 1989; Widmer, 1997) ConserVative parental attitudes about contraception (Resnick et al., 1997)

Peer: Older age of peer group and close friends (Bearman & Brikkner, 1999) Peers with poor grades and high non-normative behavior (Bearman & Bruckner, 1999) Friends with good grades and little non-normative behavior (Bearman & Brikkner, 1999) Peers with lower achievement orientation (Meschke et al., 2000) Close friends' closeness to parents (Bearman & Bruckner, 1999) Deviant life trajectories of peers (Bearman & Brikkner, 1999; Brewster, 1994) Peers with positive attitudes about preventive health (Boyer,Tschann, & Shafer, 1999) Peers who drink alcohol (Blum, Buehring, & Rinehart, 2000; Kinsman et al., 1998) Peers with permissive attitudes toward premarital sex (Carvajal et al., 1999; Gibson & Kempf, 1990; Little & Rankin, forthcoming) Sexually active peers (Alexander & Hickner, 1997; Benda & DiBlasio, 1991; East, 1994; East & Felice, 1992; East, Felice, & Morgan, 1993; Jaccard, Dittus, & Litardo, 1999; Kinsman et al., 1998; Little & Rankin, forthcoming; Lock & Vincent, 1995; Mahn et al., 2000; Miller et al., 1997; Rodgers & Rowe, 1990; Stanton, Li, Black et al., 1994; Whitaker & Miller, 2000) Individual

Biological: Being male (AGI, 1994; Benson &Torpy, 1995; Boyer,Tschann, & Shafer, 1999; Ford & Norris, 1993; Kinsman et al., 1998; Raine et al., 1999)

continued

+ = a protective factor; = a risk factor; +/ = a protective factor in some studies and a risk factor in others

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 38 4 9 Table 2.1: Antecedents of Initiation of Sex continued

Biological: continued Older age (AG1, 1994; Bearman & BrUckner; 1999; Brewster et al., 1998; Dorius, Heaton, & Steffen, 1993; lnazu & Fox, 1980; Jaccard, Dittus, & Litardo, 1999; Lock &Vincent, 1995; Newcomer & Baldwin, 1992; Raine et al., 1999; Sante Ili et al., 2000; Stanton, Li, Black, et al., 1994) Having specific dopamine receptor genes (Miller, Pasta, et al., 1998) Higher testosterone levels in both genders (Halpern et al., 1993; Halpem, Udry, & Suchindran, 1997; Lauritsen & Swicegood, 1997; Udry, 1988; Udry & Billy, 1987) Older pubertal development and timing (Benson &Torpy, 1995; Capaldi, Crosby, & Stoolmiller, 1996; Dittus & Jaccard, 2000; Flannery, Rowe, & Gulley, 1993; Halpern et al., 1993; Phinney et al., 1990; Rowe & Rodgers, 1994; Udry & Cliquet, 1982; Zabin et al., 1986) Older age of menarche (Billy, Brewster; & Grady, 1994; Brewster et al., 1998; Gibbs, 1986; Miller et al., 1997) Greater physical maturity (appears older than most) (Capaldi, Crosby, & Stoolmiller; 1996; Resnick et al., 1997)

Race/Ethnicity: Race (black vs. white) (Blum et al., 2000; Braverman & Strasburger; 1993; lnazu & Fox, 1980; Miller et al., 1997; Mott et al., 1996; Sante Ili et al., 2000; Upchurch et al., 1998; Warren et al., 1998) Ethnicity (Hispanic vs. white) (Aneshensel et al., 1990; Halpern et al., 2000) +Ethnicity (Asian Pacific Islander vs. white) (Hou & Basen-Engquist, 1997)

Relationship with Family: Being a younger (rather than older) sibling (Rodgers & Rowe, 1990) +Higher quality of family interactions, childrearing practices, support of parents, connectedness (Bearman & BrUckner; 1999; Dittus & Jaccard, 2000; lnazu & Fox, 1980; Jaccard, Drttus, & Gordon, 1996; Jensen, DeGaston, & Wee, 1994; Jessor et al., 1983; Karofsky, Zeng, & Kosorok, 2000; Luster & Small, 1994; Miller et al., 1997; Resnick et al., 1997; Smith, 1997;Turner et al., 1993) +More appropriate parental supervision and monitoring (Dorius & Barber; 1998; Ensminger; 1990; Hogan & Krtagawa, 1985; Miller et al., 1986; Romer et al., 1999; Small & Luster; 1994;Whitbeck et al., 1992) +/Greater parent/child communication about sex and birth control (East,I 996b, Fox & Inazu, 1980; Jaccard & Dittus, 1991; Jaccard, Dittus, & Gordon, 1996; Leland & Barth, 1993; Miller; Levin, et al., 1998; Moore, Peterson, & Furstenberg, 1986; Mueller & Powers, 1990; Murry, 1992; Nagy, DiClemente, & Adcock, 1995; Newcomer & Udry, I985;Whitaker & Miller; 2000; Widmer; 1997)

Attachment to and Success in School: Enrolled in school (Brewster et al., 1998) Greater school attendance (Resnick et al., 1997)

Better educational performance (Billy, Brewster, & Grady, 1994; Gibbs, 1986; Gibson & Kempf, 1990; . Halpern et al., 2000;tammers et al., 2000; Miller & Sneesby, 1988; Ohannessian & Crockett, 1993; Raine et al., 1999; Resnick et al., 1997) Either very high or very low intelligence scores (Halpern et al., 2000) Greater connectedness to school (Resnick et al., 1997) Greater participation in school clubs (Halpern et al., 2000) Greater importance of academic achievement (Jessor et al., 1983) Plans to attend college (Blum, Buehring, & Rinehart, 2000; Halpern et al., 2000) continued

+ = a protective factor; = a risk factor; +/ = a protective factor in.some studies and a risk factor in others

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 39 Table 2.1: Antecedents of Initiation of Sex continued

Attachment to and Success in School: continued Fighting at school (Miller et al., 1997) +Received AIDS education (Ku, Sonenstein, & Pleck,I 993a) +/ Received sex education (Billy, Brewster; & Grady, 1994; Dawson, 1986; Furstenberg, Moore, & Peterson, 1985; Ku, Sonenstein, & Pleck, 1993a; Marsiglio & Mott, 1986; Zelnik & Kim, 1982)

Attachment to Faith Communities: Greater religiosity (Billy, Brewster; & Grady, 1994; Ku et al., 1998; Lammers et al., 2000; Resnick et al, 1997) Having a religious affiliation (Bearman & BrUckner; 1999) Having a conservative religious affiliation (Beck, Cole, & Hammond, 1991; Miller & Olson, 1988; Weinbender & Sossignol, 1996) More frequent attendance (Billy, Brewster; & Grady, 1994; Halpern et al., 1994; Halpern et al., 2000; Miller et al., 1997; Miller & Olson, 1988; Mott et al., 1996)

Relationships with Peers: Not being part of a peer group (Bearman & BrOckner; 1999) Being popular with peers (Bearman & Bruckner; 1999) Greater importance of popularity (Meschke et al., 2000) More social activities with peers (Boyer,Tschann, & Shafer, 1999) Engaging in physical fights (Harvey & Spigner; 1995; Miller et al., 1997) More social bonding (McBride et.al., 1995)

Relationships with Romantic Partners: Dating alone (Meschke et al., 2000) Having a romantic relationship, going steady with a boy/girlfriend, closeness of relationship with partner (Blum, Buehring, & Rinehart, 2000; Halpern et al., 2000; Little & Rankin, forthcoming; Lock & Vincent, 1995; Rosenthal et al.,I 997;Thornton, 1990) Having a relationship with an older romantic partner (Leitenberg & Sattzman, 2000; MaHn et al., 2000) Greater number of romantic partners (Bearman & BrUckner; 1999)

Healthy Behaviors: +Greater participation in sports (Miller, Sabo, et al., 1998) +Greater involvement in other healthy behaviors (Crosby,Yarber, & Kanu, 1998; Harvey & Spigner; 1995)

Problem or Risk-Taking Behaviors: Greater impulsivity (Halpern et al., 2000; White & Johnson, 1988) Tobacco use (Dorius, Heaton, & Steffen, 1993; Harvey & Spigner; 1995; Raine et al., 1999) Substance use (Boyer;Tschann, & Shafer; 1999; Capaldi, Crosby, & Stoolmiller 1996; Crosby,Yarber; & Kanu, 1998; Dorius, Heaton, & Steffen, 1993; Harvey & Spigner; 1995; Kinsman et al., 1998; Kowaleski-Jones & Mott, 1998; Little & Rankin, forthcoming; Lowry et al., 1994; Mott et al., 1996; The National Center on Addiction and Substance Abuse at Columbia Universfty, 1999; Srrah, 1997;Weinbender & Rossignol, 1996)

continued

+ = a protective factor; = a risk factor; +/ = a protective factor in some studies and a risk factor in others

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 40 Table 2.1: Antecedents of Initiation of Sex continued Problem or Risk-Taking Behaviors: continued Greater involement in delinquent behaviors (Capaldi, Crosby, & Stoolmiller, 1996; Devine, Long, & Forehand, 1993; Gibbs, 1986; Harvey & Spignen 1995; Ketterlinus et al., 1992; Miller et al., 1997; Mott et al., 1996; Rosenbaum & Kande!, 1990) Running away from home (Kowaleski-Jones & Mott, 1998) Greater involvement in general unconventional behavior (Costa et al., 1995; Rosenbaum & Kandel, 1990; Rowe, Rodgers, & Meseck-Bushey, 1989)

Other Behaviors: Paid work more than 20 hours/week (Resnick et al., 1997) Watching a larger proportion of TV with sexual content (Brown & Newcomer, 1991).

Emotional Well-Being and Distress: Higher self-esteem (Millen Christensen, & Olson, 1987; Orr et al., 1989) Higher decision-making autonomy (Halpern et al., 2000) Greater perceived risk of untimely death (Halpern et al., 2000; Resnick et al., 1997) Greater level of stress (Harvey & Spignen 1995) Higher level of depression (Kowaleski-Jones & Mott, 1998) Suicidal ideation (Benson &Torpy, 1995) Receipt of help for emotional problems (Kowaleski-Jones & Mott, 1998)

Sexual Beliefs,Attitudes, Skills, and Behaviors: Viewing of TV shows with sexual content (Brown & Newcomen 199 I) More stereotypical gender roles (Foshee & Bauman, 1992; Lock &Vincent, 1995) More permissive attitudes toward premarital sex and abstinence (Carvajal et al., 1999; Christopher; Johnson, & Roosa, 1993; Gibson & Kempf, 1990; Lock & Vincent, 1995; Miller et al., 1986; Miller; Norton, et al., I998;Thomson, 1982) More perceived personal and social benefits to sex (Blum, Buehring, & Rinehart, 2000) More perceived personal and social costs to sex (Blum, Buehring, & Rinehart, 2000) Belief that boys gain respect if have sex (Kinsman et al., 1998) +Greater desire to have friends believe youth is virgin (Stanton, Li, Black et al., 1994) Greater feelings of guilt if were sexually active (Stanton, Li, Black et al., 1994) Greater perceived costs of pregnancy (Blum, Buehring, & Rinehart, 2000) Greater embarrassment if pregnant (Halpern et al., 2000) Greater perceived risk or concern about STDs or AIDS (BoyenTschann, & Shafer, 1999; Harvey & Spigner, 1995; Stanton, Li, Black, et al., 1994) Greater self-efficacy to refrain from sex (Robinson,Telljohann, & Price, 1999) Same-sex attraction or behavior (Resnick et al., 1997) Dating at an early age or frequent dating (Dorius, Heaton, & Steffen, 1993; Miller et al., 1997; Murry, 1992; Thornton, 1990) Ever kissed or necked (Blum, Buehring, & Rinehart, 2000) Greater intention to have sex (Kinsman, et al., 1998; Miller & Norton, 1998)

continued

+ = a protective factor; = a risk factor; +/ = a protective factor in some studies and a risk factor in others

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 41 Table 2.1: Antecedents of Initiation of Sex continued 11

Sexual Beliefs, Attitudes, Skills, and Behaviors: continued +Pledge of virginity (Bearman & Bruckner, 2001; Blum, Buehring, & Rinehart, 2000; Resnick et al., 1997)

Sexual Abuse: Sexual pressure, coercion, and abuse (Boyer & Fine, 1992; Browning & Laumann, 1997; Gershenson et al., 1989; MaHn et al., 2000; Miller, Monson, & Norton, 1995; Moore, Nord, & Peterson, 1989; Silverman & Amaro, 2000; Small & Luster, 1994; Stock et al., 1997)

+ = a protective factor; = a risk factor; +1 = a protective factor in some studies and a risk factor in others

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 42 Table 2.2: Antecedents of Frequency, of Sexual Intercourse Environment/Context

Community: Higher ratio of unmarried men to women (Billy, Brewster, & Grady, 1994) Higher high school dropout rate (Billy, Brewster, & Grady, 1994) Higher level of unemployment (Ku, Sonenstein, & Pleck,I 993b) Higher residential turnover (Billy, Brewster, & Grady, 1994) Higher percent of family planning patients under 20 (Billy, Brewster, & Grady, 1994)

School: +Receipt of AIDS education in school (Ku et al., 1998)

Family: Higher maternal education (Ku et al., 1998) + Two (vs. one) parents (Billy, Brewster, & Grady, 1994; Ku et al., 1998) Parental divorce (Devine, Long, & Forehand, 1993) More conservative parental attitudes about teen or premarital sex (Benda & DiBlasio, 1994; larttus & Jaccard, 2000; Jaccard, Dittus, & Gordon, 1996; Miller, Forehand, & Kotchick, 1999) Having an older sister who gave birth as an adolescent (Cox, Emans, & Bithoney, 1993; East, 1996a; East, Felice, & Morgan, 1993; Friede et al., 1986; Hogan & Kitagawa, 1985)

Peer: Sexually active peers (Benda & DiBlasio, 1991; Benda & DiBlasio, 1994) Individual

Biological: Being male (Benda & Corwyn, 1996; DiBlasio & Benda, 1990) Older age of menarche (Billy, Brewster, & Grady, 1994) Older age (Benda & Colwyn, 1996; Ku et al., 1998; Miller, Forehand, & Kotchick, 1999)

Race/Ethnicity: +/ Race (black vs. white) (Ku et al., I998;Warren et al., 1998)

Relationship with Family: Better quality of family interactions and connectedness (Benda & Corwyn, 1996; DiBlasio & Benda, 1990; Dittus & Jaccard, 2000; Jaccard, Dittus, & Gordon, 1996; Miller, Forehand, & Kotchick 1999) More appropriate parental supervision and monitoring (Benda & Conwyn, 1996; Benda & DiBlasio, 1994; Miller, Forehand, & Kotchick, 1999)

Relationship with School: +Greater educational investment (Ohannessian & Crockett, 1993) +Higher academic performance and lower school failure (DiBlasio & Benda, 1990; Ku et al.,I 998)

continued

+ = a protective factor; = a risk factor; +1 = a protective factor in some studies and a risk factor in others

Emerging Answers: Research Findings on Programs to Reduce.Teen Pregnancy 43 5 4 Table 2.2: Antecedents of Frequency of Sexual Intercourse continued Relationship with School: continued +Received AIDS education (Ku, Sonenstein, & Pleck 1992) +Participation in sports (Sabo et al., 1998)

Attachment to Faith Communities: +Greater religiosity (DiBlasio & Benda, 1990; Du Rant & Sanders, 1989; Ku, Sonenstein, & Fleck, 1992)

Relationships with Partners: +Greater female power in the relationship (Jorgensen, King, &Torrey, 1980)

Problem or Risk-Taking Behaviors: Alcohol use (Bailey, Cam lin, & Ennett, 1998) Drug use (Bailey, Cam lin, & Ennett, 1998; DiBlasio & Benda, 1990) Greater involvement in other problem behaviors (DiBlasio & Benda, 1990)

Emotional Well-Being and Stress: Suicide attempts (Burge et al., 1995)

Sexual Beliefs, Attitudes, Skills, and Behaviors: Older age of first sex (Du Rant & Sanders,I 989;Thornton, 1990) Greater worry about AIDS (Ku, Sonenstein, & Pleck, 1992) More permissive attitudes toward premarital sex (Ku et al., 1998) More permissive attitudes about abortion (Ku et al., 1998) Greater number of years being sexually active (Du Rant & Sanders, 1989)

+ = a protective factor; = a risk factor; +1 = a protective factor in some studies and a risk factor in others

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 44 Table 2.3: Antecedents of Number of Sexual Partners Environment/Context

Community: Living in an urban area (Sante lli et al., 1998)

School: +Taught AIDS education (Ku, Sonenstein, & Pleck, 1992)

Family: Parental divorce (Devine, Long, & Forehand, 1993) Higher parental income level (Ku, Sonenstein, & Pleck 1992) Greater religiosity (Ku, Sonenstein, & Pleck, 1992) More conservative parental attitudes about teen or premarital sex (Luster & Small, 1997; Miller, Forehand, & Kotchick, 1999; Shah & Zelnik, 1981; Thornton & Camburn, 1989)

Peer: More communication about HIV (Holtzman & Rubinson, 1995) Sexually active peers (Whitaker & Miller, 2000)

Partner: Greater difference in age of first partner eber et al., 1992) Individual

Biological: Being male (Durbin et al., 1993; Ford & Norris, 1993; Holtzman & Rubinson, 1995) Older age (Holtzman & Rubinson, 1995; Miller, Forehand, & Kotchick, 1999; Richter et al., 1993;Sanielli et al., 1998)

Race/Ethnicity: Race (black vs. white) (Durbin et al., 1993; Holtzman & Rubinson, 1995; Santelli et at, 1998; Shrier et al., 1997;Warren et al., 1998) Ethnicity (Hispanic vs. white) (Santelli et al., 1998) Greater acculturation by Hispanics (Ford & Norris, 1993)

Relationship with Family: +Greater general communication (Miller, Forehand, & Kotchick, 1999) + More appropriate parental supervision and monitoring (Luster & Small, 1997; Miller, Forehand, & Kotchick, 1999) +/Greater parent/child communication about sex, condoms, or AIDS (Holtzman & Rubinson, 1995;Leland & Barth, 1993;Whitaker & Miller, 2000) Physical abuse (Luster & Sma11,1997)

continued

+ = a protective factor; = a risk factor; +/ = a protective factor in some studies and a risk factor in others

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 45 5 6 Table 2.3: Antecedents of Number of Sexual Partners continued Relationship with Faith Communities: +Greater religious attendance (Seidman, Mosher, & Aral, I994;Thornton & Cambum, 1989) +Having a conservative religious affiliation (Ku, Sonenstein, & Pleck, 1992)

Relationship with Community: +Having a mentor (Beier, Rosenfeld, et al., 2000) +Greater level of social support (St Lawrence et al., 1994)

Teen Relationships with Partners: +Being married (Sante Ili et al., 1998) Dating violence (Valois, Oeltmann, Waller, & Hussey, 1999)

Healthy Behaviors: Greater involvement in healthy behaviors (Crosby,Yarber, & Kanu, 1998) Greater participation in sports (Sabo et al., 1998)

Problem or Risk-Taking Behaviors: More traditional attitudes toward masculinity (Heck, Sonenstein, & Ku,I 993b) Substance use (Crosby,Yarber, & Kanu, 1998; Lowry et al., 1994; Middleman et al., 1995; Millstein & Moscicki, I 995;The National Center on Addiction and Substance Abuse at Columbia Universfty, 1999; Richter et al., 1993; Santelli et al., 1998; Shrier et al., 1997; Valois, Oeltmann,Waller, & Hussey, 1999) Physical fighting (Valois, Oeltmann,Waller, & Hussey, 1999) Carrying weapons (Valois, Oeltmann,Waller, & Hussey, 1999) Greater involvement in other problem behaviors (Richter et al., 1993)

Sexual Beliefs,Attitudes, Skills, and Behaviors: +Older age of first sex (or years since first sex) (Durbin et al., 1993; Shrier et al., 1997; Santelli et al., 1998; Seidman, Mosher, & Aral, 1994;Thomton, 1990;Weber et al., 1992) +/Greater knowledge about AIDS (Holtzman & Rubinson, 1995; Weinman, Smith, & Mumford, 1992) +Greater perceived susceptibility to STDs/AIDS (Catania et al., 1989; Ku, Sonenstein, & Pleck 1992) + More conservative attitudes and norms toward number of sexual partners (Basen-Engquist & Parcel, 1992, Catania et al., 1989) +Sexual communication skills (Catania et al., 1989)

Sexual Abuse: Sexual abuse (Browning & Laumann, 1997; Luster & Small, 1994; Luster & Small, 1997; Silverman & Amaro, 2000; Stock et al., 1997)

+ = a protective factor; = a risk factor; +/ = a protective factor in some studies and a risk factor in others

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 46 5 7 Table 2.4: Antecedents of Use of Condoms Environment/Context

Family: +Higher parental education (Brewster et al., 1998; Murphy & Boggess, 1998) +Higher income level (Wilson et al., 1994)

Peers: +Greater peer norms and support for condom use (Boyer,Tschann, & Shafer; 1999; Di Clemente et al., 1996; Whitaker & Miller; 2000) +Greater peer use of condoms (Di Clemente, 1990; Stanton, Li, Black, et al, 1994; Whitaker & Miller; 2000)

Partner: Partner 3 or more years older (Miller; Clark, & Moore, 1997) Greater partner support for condom use (Murphy & Boggess, 1998; Pendergrast, Du Rant, & Gaillard, 1992; Pleck, Sonenstein, & Ku, 1991) Greater partner sexual experience (Ku, Sonenstein, & Pleck, 1994) Higher risk status of partner (Ku, Sonenstein, & Pleck, 1994) Individual

Biological: Older age (Di Clemente et al., 1996; Goodwin, 1990; Hingson et al., 1990; Ku, Sonenstein, & Pleck, 1994; Murphy & Boggess, 1998; Pendergrast, Du Rant, & Gaillard, 1992; Pleck, 1989; Pleck, Sonenstein, & Swain, 1988; Reitman et al., 1996; Richter et al., 1993; Rickrnan et al., 1994; Shrier et al., 1997) Being male (Brown, Di Clemente, & Park, 1992; Di Clemente et al., 1996; Shrier et al., 1997)

Race/Ethnicity: +/ Race (black vs. white) (CDC, 1996; Hingson et al., 1990; Marsiglio, 1993; Murphy & Boggess, 1998; Warren et al., 1998) Ethnicity (Hispanic vs. white) (Ku, Sonenstein, & Pleck, 1994; Murphy & Boggess, 1998)

Relationship with Family: +/ Greater parent-child communication about sex, condoms, or birth control (Leland & Barth, 1993; Miller; Levin, et al., 1998; Romer et al., 1999;WKrtaker & Miller; 2000;Whitaker et al., I999;Wilson et al., 1994) + More appropriate parental supervision and monitoring (Miller; Forehand, & Kotchick, 1999)

Attachment to and Success in School: +Higher educational performance (Fleck Sonenstein, & Swain, 1988; St Lawrence, 1993) +Plans to attend college (Heck, Sonenstein, & Swain, 1988)

Religiosity: Roman Catholic vs. Protestant & other (Brewster et al., 1998)

continued

+ = a protective factor; = a risk factor; +/ = a protective factor in some studies and a risk factor in others

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 47 tr Table 2.4: Antecedents of Use of Condoms continued Relationships with Partners: +More discussions about sexual risk (Whitaker et al., 1999) Steady close romantic relationship (Pleck, Sonenstein, & Swain: 1988)

Relationships with Partners: continued Longer relationship (Ku, Sonenstein, & Pleck, 1994) Being married (Ku, Sonenstein, & Pleck, 1994)

Healthy Behaviors: +Greater participation in sports and exercise (Crosby,Yarben & Kanu, 1998; Sabo et al., 1998)

Problem or Risk-Taking Behaviors: Alcohol use (Bailey, Cam lin, & Ennett 1998; Goodwin, 1990; Hingson et al., 1990) Drug use (Bailey, Cam lin, & Ennett, 1998; Crosby,Yarber, & Kanu, 1998; Goodwin, 1990; Hingson et al., 1990; Lowry et al., 1994; Middleman et al., 1995; Richter et al., 1993; Shrier et al., 1997; Sonenstein, Pleck, & Ku, 1989) Greater involvement in general risk-taking (Brown, Di Clemente, & Park, 1992; Richter et al., 1993)

Emotional Well-Being and Distress: Greater internal locus of control (St Lawrence, 1993) Greater impulse control (Di Clemente et al., 1996) Greater self-control (St. Lawrence, 1993) More social support (St. Lawrence et al., 1994)

Sexual Beliefs,Attitudes, Skills, and Behaviors: Greater acceptance of non-traditional gender roles by men and women (Pleck, Sonenstein, & Ku,I 993b) Greater desire to have friends believe one is virgin (Stanton, Li, Black, et al, 1994) Older age of first s,ex (Ku, Sonenstein, & Fleck, 1994; Murphy & Boggess, 1998; Shrier et al., 1997) Higher frequency of sex (DiClemente et al., 1996; Ku, Sonenstein, & Pleck, 1994) Greater number of sex partners (Goodwin, 1990; Sonenstein, Pleck, & Ku, 1989; Wilson et al., 1994) Sex with a prostitute (Sonenstein, Pleck, & Ku, 1989) Use of alcohol or drugs before sex (Hou & Basen-Engquist, 1997) Received STD education (Pendergrast, DuRant, & Gaillard, 1992) Discussed AIDS with their physician (Goodwin, 1990) Discussed AIDS with others (Hingson et al., 1990) Greater knowledge about AIDS (Ku, Sonenstein, & Pleck, 1992) Knowing someone who was HIV-positive (Hingson et al., 1990; Rickman et al., 1994) Greater knowledge about condoms (Reitman et al., I996;Wilson et al., 1994) Greater perceived.male responsibility for pregnancy prevention (Murphy & Boggess, 1998; Fleck, Sonenstein, & Ku, 1991; Pleck, Sonenstein, & Swain, 1988) Stronger belief that causing a pregnancy was a sign of manhood (Murphy & Boggess, 1998)

continued + = a protective factor; = a risk factor; +/ = a protective factor in some studies and a risk factor in others

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 48 Table 2.4: Antecedents of Use of Condoms continued Sexual Beliefs,Attitudes, Skills, and Behaviors: continued Greater perceived susceptibility to pregnancy/STDs/HIV (Hingson et al., 1990; Ku, Sonenstein, & Pleck, 1992; Murphy & Boggess, 1998; Heck Sonenstein, & Ku; 1991; Fleck Sonenstein, & Ku, 1993a; Reitman et al., 1996) Greater motivation to avoid AIDS (Ku, Sonenstein, & Fleck 1992) Greater importance of avoiding STDs (Wilson et al., 1994) Stronger belief that condoms are effective in reducing STDs/HIV (Hingson et al., 1990) More positive attitudes toward condoms and other forms of contraception (Hingson et al., 1990; Reitman et al., 1996) Greater embarrassment and barriers to getting condoms (Murphy & Boggess, 1998; Sieving et al., 1997) Higher perceived barriers or costs of using condoms (e.g., reduce pleasure) (Catania et al., 1989; Di Clemente et al., 1992; Hingson et al., 1990; Murphy & Boggess, 1998; Pendergrast, Du Rant, & Gaillard, 1992; Fleck, Sonenstein, & Ku, 199 I ) Greater perceived self-efficacy in using condoms (BoyerTschann, & Shafer 1999; Di Clemente et al., 1992; Pendergrast, Du Rant, & Gaillard, 1992; Sieving et al., 1997; Reitman et al., 1996 ) Greater self-efficacy to demand condom use (Di Clemente et al.,.1996) Stronger sexual communication skills (Catania et al., 1989) Greater motivation to use condoms (Bailey, Cam lin, & Ennett, 1998) Greater intent to use condoms (Brown, Di Clemente, & Parl<, 1992) Carry condoms (Hingson et al., 1990) Previous condom use (St Lawrence, 1993)

Sexual Abuse: Sexual abuse (Silverman & Amaro, 2000)

+ = a protective factor; = a risk factor; +/ = a protective factor in some studies and a risk factor in others

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 49 60 Table 2.5: Antecedents of Use of Contraception Environment/Context

Community: +Better neighborhood quality (Hogan, &tone, & Krtagawa, 1985)

Family: Higher parental education (Brewster et al., 1998; Hayward, Grady, & Billy, 1992; Manning, Longmore, & Giordano, 2000) Two (vs. one) parents (Forste & Heaton, 1988; Gispert et al., 1984; Manning, Longmore, & Giordano, 2000; Moore, Morrison, & Glei, 1995) Higher income level (AGI, 1994; Hogan, Astone, & Kitagawa, 1985; Galavotti & Lovick 1989) Mother's receipt of welfare (Moore, Morrison, & Glei, 1995) More positive parental values about contraception (Baker,Thalberg, & Morrison, 1988; Jorgensen & Sonstegard, 1984;Thomson, 1982)

Peers: Substance use (Kowaleskidones & Mott, 1998)

Partner: Much older male (Darroch, Landry, & Os lak,I 999b) Greater partner support for contraceptive use (Kastnen 1984; Weisman et al., 1991; Whitley & Schofield, 1985- 86) Agreement with partner about contraceptive method (Brindis et al., 1998) Individual

Biological: +Older age (AGI, 1994; Glei, 1999; Ku, Sonenstein, & Pleck, 1994; Santelli et al., 2000; Scher, Emans, & Grace, 1982)

Race/Ethnicity: +/ Race (black vs. white) (Manning, Longmore, & Giordano, 2000; Marsiglio, 1993) Ethnicity (Hispanic vs. white) (AGI, 1994; Manning, Longmore, & Giordano, 2000)

Relationship with Family: Appropriate family strictness and discipline (Cvetkovich & Grote, 1981) Greater parental monitoring (Luster & Small, 1994) Greater parental connectedness and support (Drttus & Jaccard, 2000; Gispert et al., 1984; Jaccard, Dittus, & Gordon, 1996; Luster & Small, 1994) +/Greater parent-child communication about sex, condoms, or other forms of birth control (Handelsman, Cabral, & Weisfeld, 1987; Inazu & Fox, 1980; Jaccard, Dittus, & Gordon, 1996; Kastner, 1984; Newcomer & Udry, I 985;Thomson, 1982)

Attachment to and Success in School: Dropped out of school (Darroch, Landry, & Oslak 1999)

continued

+ = a protective factor; = a risk factor; +/ = a protective factor in some studies and a risk factor in others

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 50 61 Table 2.5: Antecedents of Use of Contraception continued Attachment to and Success in School: continued +Plans to attend college (Scher, Emans, & Grace, 1982) +Received sex education (Manning, Longmore, & Giordano, 2000)

Attachment to Faith Communities: More frequent attendance (Collins & Robinson, 1986)

Relationships with Partners: Going steady (Manning, Longmore, & Giordano, 2000) Monogamous relationships (Whitley & Schofield, 1985-86) Greater female power in the relationship (Jorgensen, King, &Torrey, 1980) Discussed contraception with partner (Collins & Robinson, 1986)

General Skills and Personality Traits: Higher level of cognitive development (Holmbeck et al., 1994) Greater problem-solving skills (Flaherty et al., 1983) More future orientation (Whitley & Schofield, 1985-86) Greater egocentrism (Holmbeck et al., 1994)

Healthy Behaviors: +Greater participation in sports (Sabo et al., 1998) +Greater involvement in other healthy behaviors (Costa et al., 1995; Fortenberry, Costa, & Jessor, 1997)

Problem or Risk-Taking Behaviors: Greater general risk-taking (Costa et al., 1996) Greater sensation-seeking (Arnett, 1990) +Greater general psychosocial conventionality (Costa et al., 1996)

Emotional Well-Being and Distress: Stronger self-image and self-esteem (Holmbeck et al., 1994) Depression (Kowaleski-Jones & Mott, 1998)

Sexual Beliefs,Attitudes, Skills, and Behaviors: Received sex education (Dawson, 1986; Mauldon & Luker 1996; Marsiglio & Mott, 1986; Zelnik & Kim, 1982) Greater acceptance of non-traditional gender roles for women (Kowaleski-Jones & Mott, 1998) Greater importance of avoiding pregnancy (Philliber et al., I986;Weisman et al., 1991; Zabin,Astone, & Emerson, 1993) Perception of positive side effects of oral contraceptives (Weisman et al., 1991) More permissive attitude toward premarital sex (Collins & Robinson, 1986; Morrison, 1989) Pledge of virginity (Bearman & Bruckner 2001) Older age of first sex (Galavotti & Lovick, 1989; Manning, Longmore, & Giordano, 2000; Melchert & Burnett, 1990; Scott-Jones &Turner 1988) continued

+ = a protective factor; = a risk factor; +1 = a protective factor in some studies and a risk factor in others

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 5 I 6 t) Table 2.5: Antecedents of Use of Contraception continued Sexual Beliefs, Attitudes, Skills, and Behaviors: continued Greater wantedness of first sex (Abma, Driscoll, & Moore, 1998) Greater frequency of sex (Collins & Robinson, 1986; Morrison, 1989;Whitley & Schofield, 1985-86) Greater acceptance of own sexual behavior (Whitley & Schofield, 1985-86) Greater knowledge about contraception (Morrison, 1989; Scott-Jones &Turner, 1988) More positive attitudes toward contraception (Jorgensen & Sonstegard, 1984; Kastner, 1984; Morrison, 1989; Philliber et al., 1986; Weisman, 1991; Whitley & Schofield, 1985-86) Greater comfort and satisfaction with contraceptive method (Brindis et al., 1998; Shea, Herceg-Baron, & Furstenberg, 1984) Greater perceived susceptibility to pregnancy/STDs/HIV (Arnett, 1990; Philliber et al., 1986) Previous contraceptive history and experience (Gorosh, 1982; Marsiglio, I993;Weisman et al., 1991) Greater number of visits to a family planning clinic (Brindis et al., 1994) Greater. satisfaction with family planning clinic visit (Shea, Herceg-Baron, & Furstenberg, 1984; Scher, Emans, & Grace, 1982)

Sexual Abuse: Sexual abuse (Boyer & Fine, 1992; Luster & Small, 1994; Silverman & AFriao, 2000; Stock et al., 1997)

)+ = a protective factor; = a risk factor; +/ = a protective factor in some studies and a risk factor in others

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 52 63 Table 2.6: Antecedents of Pregnancy' Environment/Context

State: Restrictive laws regarding contraceptive licensing, advertising, or selling (Lundberg, 1990; Lundberg & Plotniclk 1995) +Coordinated programs and policies for addressing teen pregnancy (Moore et al., 1994)

Community: Higher community socio-economic status (Hogan & Kitagawa, 1985; Mayer & Jencks, 1989) Greater residential mobility (Singh, 1986) Higher level of unemployment (Ku, Sonenstein, & Pleck,I 993b) Higher violent crime rate (Moore et al., 1994) Higher teenage suicide rate (Singh, 1986) Higher levels of community stress (Singh, 1986)

School: Catholic (vs. public) school (Manlove, 1998) Higher percent of minority students (Manlove, 1998) Higher percent of students receiving free lunch (Manlove, 1998) Learning-focused school setting (Kasen, Cohen, & Brook 1998) Higher school dropout rates (Singh, 1986) Higher rates of school vandalism (Chandy et al., 1994)

Family: Foreign language spoken at home (Lundberg & Plotnick 1995) Higher parental education (Chandy et al., 1994; Hayward, Grady, & Billy, 1992; Lundberg & Plotnick, 1995; Plotnick, 1992; Roosa et al., 1997) Larger family size (Hogan & Kitagawa, 1985; Lundberg & Plotnick, 1995) + Two (vs. one) parents (Barnett, Papini, & Gbur 1991; Chandy et al., 1994; Hogan & Kitagawa, 1985; Lundberg & Plotnick, 1995; Manlove, 1998; Murry, 1992) Higher income level (AGI, 1994; Barnett, Papini, & Gbur, 1991; Hogan & Krtagawa, 1985; Ireson, 1984; Manlove, 1998; Murry, 1992) Partial coverage with public health insurance (Bearman & Bnkkner, 1999) Older sister who was an adolescent parent (East, 1996a; Hogan & Kitagawa, 1985) Greater parental disapproval of teen sex or use of contraception (Dittus & Jaccard, 2000; Resnick et al., 1997)

Peers: +Friends with good grades (Kasen, Cohen, & Brook, 1998) Friends and peers with good grades and little non-normative behavior (Bearman & BrUckner 1999) Friends who are teen mothers (Holden et al., 1993)

continued

+ = a protective factor; = a risk factor; +1 = a protective factor in some studies and a risk factor in others

1 Includes antecedents of males who impregnate females.

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy '53 6 4 Table 2.6: Antecedents of Pregnancy' continued Partner: Male partner 3 or more years older (Darroch, Landry, & Oslak, 1999; Miller, Clark, & Moore, 1997) Individual

Biological: Older age (AG1, 1994; Bearman & Bruckner, 1999; Darroch, Landry, & Oslak, 1999; Guagliardo, Huang, & D'Angelo, 1999; Pierre et al, 1998; Spingarn & Du Rant, 1996) Greater physical development (Dttus & Jaccard, 2000)

Race/Ethnicity: Race (black vs. white) (AGI, 1994; Bearman & BrUckner, 1999) Ethnicity (Hispanic vs. white) (AGI, 1994)

Relationship with Family: +Greater teen/family connectedness (Barnett, Papini, & Gbur, 1991; Bearman & Bruckner, 1999; Dittus & Jaccard, 2000; Resnick et al., 1997; Scott, 1993) +Appropriate parental supervision and monitoring (Hogan & Kitagawa, 1985) +/Greater parent/child communication about sex (Adolph et al., 1995; Barnett, Papini, & Gbur, 1991; Baumeister, Flores, & MaHn, 1995; Leland & Barth, 1993; Murry, 1992) Physical abuse (Chandy et al., 1994) General maltreatment by family (Smith, 1996)

Attachment to and Success in School: Positive attitude toward school (Plotnick, 1992) Higher school performance (Ireson, 1984; Holden et al., 1993; Manlove, 1998; Robbins, Kaplan, & Martin, 1985) School dropout (Manlove, 1998; Murry, 1992) Higher education plans (Manlove, 1998; Plotnick, 1992)

Religiosity: Mainstream Protestant versus other (Bearman & Bruckner, 1999; Lundberg & Plotnick, 1995) Protestant, Catholic, or Jewish versus none (Plotnick 1992)

Relationships with Peers: +Being a member of the leading crowd (Bearman & Brakner, 1999)

Relationships with Partners: Dating or dating at an early age (Hogan & Kitagawa, 1985) Being married (AGI, 1994; Barnett, Papini, & Gbur, 1991; Darroch, Landry, & Oslak,I 999a) Greater number of sexual partners (Pierre et al., 1998)

Healthy Behaviors: +Greater participation in sports (Sabo et al., 1998) continued )+ = a protective factor; = a risk factor; +/ = a protective factor in some studies and a risk factor in others

1 Includes antecedents of males who impregnate females.

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 54 6 Table 2.6: Antecedents of Pregnancy' continued Problem or Risk-Taking Behaviors: Alcohol use (Pierre et al., 1998; Spingarn & Du Rant, 1996) Substance use (Guagliardo, Huang, & D'Angelo, 1999; Pierre et al., 1998; Spingarn & Du Rant, 1996) Greater involvement in other risk behaviors (Pierre et al., 1998; Resnick, Chambliss, & Blum; 1993; Spingarn & Du Rant, 1996) Greater involvement in delinquent behaviors (Kasen, Cohen, & Brook, 1998; Pierre et al., 1998)

Emotional Well-Being and Distress: +Greater positive self-concept (Pete-McGadney, 1995) Attempted suicide (Pierre et al., 1998)

Sexual Beliefs,Attitudes, Skills, and Behaviors: Sex education (Zelnik & Kim, 1982) More egalitarian gender and family roles (Plotnick, 1992) Greater sex role competencies (Ireson, 1984) Stronger belief that causink a pregnancy was a sign of manhood (Resnick Chambliss, & Blum, 1993) +Greater perceived negative consequences of pregnancy (Resnick et al., 1997) Greater perceived ease of childbearing and parenting (Holden et al., 1993) Same-sex sexual behavior (Vermont Department of Health, 1997) Older age of first sex (Bearman & Bruckner, 1999; Melchert & Burnett, 1990; Murry, 1992; Spingarn & Du Rant, 1996) Greater number of years since first sex (Resnick et al., 1997) Greater number of sexual partners (Bearman & BrUckner, 1999; Guagliardo, Huang, & D'Angelo, 1999; Spingarn & Du Rant, 1996) Greater motivation to use contraception (Landry et al., 1986) Greater use of condoms (Guagliardo, Huang, & D'Angelo, 1999) Greater use of contraception (Barnett, Papini, & Gbur, 1991; Bearman & BrUckner, 1999; Glei, 1999; Holden et al., 1993; Murry, 1992; Pierre et al., 1998; Resnick et al., 1997) History of STDs (Guagliardo, Huang, & D'Angelo, 1999)

Sexual Abuse: Sexual abuse (Boyer & Fine, 1992; Kenney, ReinhoFtz, & Angelini, 1997; Nagy Di Clemente, & Adcock, 1995; Pierre et al., 1998; Silverman & Amaro, 2000; Stock et al., 1997)

+ = a protective factor; = a risk factor; +1 = a protective factor in some studies and a risk factor in others

1 Includes antecedents of males who impregnate females.

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 55 6 6 Table 2.7: Antecedents of Childbearing' Environment/Context

State: Higher education level (Moore et al, 1994) Higher levels of female labor force participation (Moore et al., 1994) Higher crime rate (Moore et al., 1994) Higher level of state funding for family planning (Anderson & Cope, 1987; Forrest, Hermalin, & Henshaw, 1981; Moore et al., 1994; Singh, 1986) Higher level of state funding for abortion (Moore et al., 1994)

Community: Higher ratio of males to females (Ku, Sonenstein, & Pleck,I 993b) Higher level of unemployment (Ku, Sonenstein, & Pleck,I 993b) Higher community income (Brooks-Gunn et al., 1993) More high-status workers (Crane, 1991) More community opportunities (Bickel et al., 1997) Higher levels of community stress (Singh, 1986) Greater community social disogranization (Billy, Brewster, & Grady, 1994)

School: Private religious school (Zill, Nord, & Loomis, 1995) High minority enrollment (Zill, Nord, & Loomis, 1995) Higher levels of safety (Moore et al., 1998) Higher sch.00l dropout rates (Singh, 1986) College prep program (vs. vocational or general) (Zill, Nord, & Loomis, 1995)

Family: Parent education (Zill, Nord, & Loomis, 1995) Greater family emphasis upon responsibility (Hanson, Myers, & Ginsburg, 1987) Having an older sister who gave birth as an adolescent (Cox, Emans, & Bithoney, 1993; East et al., 1996a; East, Felice, & Morgan, 1993; Friede et al., 1986; Hogan & Kitagawa, 1985) Higher maternal educational level (An, Haveman, & Wolfe, 1993; Brooks-Gunn et al., 1993; Kahn & Anderson, 1992; Manlove et al., 2000;Thornberry, Smith, & Howard, 1997) + Two (vs. one) parents (Manlove et al., 2000; Moore et al., 1998; Stouthamer-Loeber & Wei, 1998; Zill, Nord, & Loomis, 1995) Presence of an adequate father figure (Gohel, Diamond, & Chambers, 1997) Presence of a grandparent in the home (Astone & Washington, 1994) Marital disruptions or remarriages (An, Haveman, & Wolfe, 1993; Manlove et al., 2000;Wu, 1994) Higher family income level (Brooks-Gunn et al, 1993, Zill, Nord, & Loomis, 1995) Mother was a teen mother (Gohel, Diamond, & Chambers, 1997; Kahn & Anderson, 1992; Manlove et al., 2000; Thornberry, Smith, & Howard, 1997) More negative parental view of early parenthood (Gohel, Diamond, & Chambers, 1997) continued )+ = a protective factor; = a risk factor; +1 = a protective faCtor in some studies and a risk factor in others 2 Includes antecedents of males who father children.

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 56 Table 2.7: Antecedents of Childbearing2 contin'ued

Peers: A best friend who has been pregnant (Landry et al., 1986)

Partner: Older age of male partner (Ku, Sonenstein, & Pleck, 1993a; Landry & Forrest, 1995; Males & Chew, 1996) Individual

Biological: Older age (Darroch, Landry, & Oslak, 1999; Ventura, Curtin, & Mathews, 1998) +Older age of menarche (Kahn & Anderson, 1992)

Race/Ethnicity: Race (black vs. white) (An, Haveman, &Wolfe, 1993; Brooks-Gunn et al., 1993; Manlove et al., 2000; Moore, et al.,I 998;Thornberry, Smith, & Howard, 1997; Zill, Nord, & Loomis, 1995) Ethnicity (Hispanic vs. white) (Aneshensel et al., 1990; Manlove et al., 2000; Moore et al., 1998; Thornberry, Smith, & Howard, 1997)

Relationship with Family: Greater family social support (Thornberry, Smith, & Howard, 1997) Greater parent involvement in adolescent's education (Manlove, 1995) Living away from home (Landry et al., 1986) Having run away from home (Kowaleski-Jones & Mott, 1998)

Attachment to and Success in School: Changed schools multiple times (Moore et al., 1998) Greater school involvement (Moore et al., 1998; Zill, Nord, & Loomis, 1995) Higher school performance (Moore et al., 1998; Plotnick & Butler, 1991; Stouthamer-Loeber & Wei, I998;Thornberry, Smith, & Howard, 1997; Zill, Nord, & Loomis, 1995) School dropout (Manlove et al., 2000) Higher educational aspirations (Moore et al., 1998; Plotnick & Butler, 1991) Higher parental college expectations for teen (Thornberry, Smith, & Howard, 1997)

Religiosity: +Having a religious affiliation (An, Haveman, &Wolfe, 1993; Manlove et al., 2000)

Relationship to Community: +Youth participation in a stable community (Bickel et al., 1997)

Relationships with Peers: Perceived by peers as controversial or aggressive in elementary school (Underwood, Kupersmidt, & Coie, 1996) + More popular in elementary school (Underwood, Kupersmidt, & Coie, 1996) Membership in a gang (Thornberry, Sm'rth, & Howard, 1997) continued

+ = a protective factor; = a risk factor; +/ = a protective factor in some studies and a risk factor in others

2 Includes antecedents of males who father children.

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 57 63 Table 2.7: Antecedents of Childbearing2 continued Teen Relationships with Partners: Being married .(AGl, 1994; Darroch, Landry, & Oslak, 1999)

Healthy Behaviors: +/Participation in sports (Sabo et al., 1998)

General Skills and Personality Traits: +Greater internal locus of control (Kowaleski-Jones & Mott, 1998)

Emotional Well-Being and Distress Higher self-esteem (Kowaleskidones & Mott, 1998; Plotnick & Butler, 1991; Robinson & Frank, 1994) Depression (Kowaleski-Jones & Mott, 1998) Receipt of help for emotional problems (Kowaleski-Jones & Mott, 1998)

Sexual Beliefs,Attitudes, Skills, and Behaviors: More non-traditional attitudes toward family and gender roles (Plotnick & Butler, 1991) Older age of first sex (Manlove et al., 2000; Smith, I 997;Thornberry, Smith, & Howard, 1997) Greater desire to have a child or ambivalence about having one (Gohel, Diamond, & Chambers, 1997; Hanson, Morrison, & Ginsberg, 1989; Nesmith et al., 1997; Zabin, 1994a; Zabin,Astone, & Emerson, 1993)

Sexual Abuse: Sexual abuse (Browning & Laumann, 1997)

Problem or Risk-Taking Behaviors More permissive attitudes toward risk-taking (Kowaleski-Jones & Mott, 1998) Substance use (Kowaleskidones & Mott, I 998;Thornberry, Smith, & Howard, 1997) Greater involvement in delinquent behavior (Stou-thamer-Loeber & Wei, 1998) Greater involvement in non-sexual risk-taking behaviors (Serbin et al., 1991)

+ = a protective factor; = a risk factor; +/ = a protective factor in some studies and a risk factor in others

2 Includes antecedents of males who father children.

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 58 69 References (1998). Substance use and risky sexual behavior among homeless and runaway youth. Journal of Abma, J., Driscoll, A., & Moore, K. (1998). Adolescent Health, 23(6), 378-388. Young women's degree of control over first intercourse: An exploratory analysis. Family Baker, S.A., Thalberg, S.P., & Morrison, D.M. Planning Perspectives, 30(1), 12-18. (1988). Parents' behavioral norms as predictors of adolescent sexual activity and contraceptive Adolph, C., Remos, D.E., Linton, K.L., & use. Adolescence, 23(90), 265-282. Grimes, D.A. (1995). Pregnancy among Hispanic teenagers: Is good parental Barnett, J.K., Papini, D.R., & Gbur, E. (1991). communication a deterrent? Contraception, Familial correlates of sexually active pregnant 51(5), 303-306. and nonpregnant adolescents. Adolescence, 26(102), 457-472. Afxentiou, D., & Hawley, C.B. (1997). Explaining female teenagers' sexual behavior and Basen-Engquist, K., & Parcel, G.S. (1992). outcomes: A bivariate probit analysis with selec- Attitudes, norms, and self-efficacy: A model of tivity correction. Journal of Family and Economic adolescents' HIV-related sexual risk behavior. Issues, 18, 91-106. Health Education Quarterly, 19(2), 263-277.

Alan Guttmacher Institute. (1994). Sex and Baumeister, L.M., Flores, E., & Marin, B.V. America's teenagers. New York, NY: Author. (1995). Sex information given to Latina adolescents by parents. Health Education Research, Alexander, E., & Hickner, J. (1997). First coitus 10(2), 233-239. for adolescents: Understanding why and when. Journal of the American Board of Family Bearman, P.S., & Bruckner, H. (1999). Power in Practice, 10(2), 96-103. numbers: Peer effects on adolescent girls' sexual debut and pregnancy. Washington, DC: The An, C.B., Haveman, R., & Wolfe, B. (1993). National Campaign to Prevent Teen Pregnancy. Teen out-of-wedlock births and welfare receipt: The role of childhood events and economic cir- Bearman, P.S., Bruckner, H., Brown, B.B., cumstances. The Review of Economics and Theobald, W., & Philliber, S. (1999). Peer poten- Statistics, 75(2), 195-208. tial: Making the most of how teens influence each other. Washington, DC: The National Campaign Anderson, LE., & Cope, L.G. (1987). The to Prevent Teen Pregnancy. impact of family planning program activity on fertility. Family Planning Perspectives, 19(4), Bearman, P.S., & Briickner, H. (2001). 152-157. Promising the future: Virginity pledges and the transition to first intercourse. American Journal Aneshensel, C.S., Becerra, R.M., Fielder, E.P., & of Sociology, 106(4), 859-912. Schuler, R.H. (1990). Onset of fertility-related events during adolescence: A prospective com- Beck, S.H., Cole, B.S., & Hammond, J.A. (1991). parison of Mexican-American and non-Hispanic Religious heritage and premarital sex: Evidence white females. American Journal of Public from a national sample of young adults. Journal Health, 80(8), 959-963. for the Scientific Study of Religion, 30, 173-180.

Arnett, J. (1990). Contraceptive use, sensation Beier, S.R., Rosenfeld, W.D., Spitalny, K.C., seeking, and adolescent egocentrism. Journal of Zansky, S.H., & Bontempo, A.N. (2000). The Youth and Adolescence, 19(2), 171-180. potential role of an adult mentor in influencing high-risk behaviors in adolescents. Archives of Astone, N.M., & Washington, M.L. (1994). The Pediatrics and Adolescent Medicine, 154, 327-331. association between grandparental coresidence and adolescent childbearing. Journal of Family Benda, B.B., & Corwyn, R.F. (1996). Testing a Issues, 15(4), 574-589. theoretical model of adolescent sexual behavior among rural families in poverty. Child and Bailey, S.L., Camlin, C.S., & Ennett, S.T. Adolescent Social Work Journal, 13(6), 469-494.

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 59 70 Benda, B.B., & DiBlasio, F.A. (1991). Brewster, K.L., Cooksey, E.C., Guilkey, D.K., & Comparison of four theories of adolescent sexual Rindfuss, R.R. (1998). The changing impact of exploration. Deviant Behavior, 12(3), 235-257. religion on the sexual and contraceptive behavior of adolescent women in the United States. Benda, B.B., & DiBlasio, F.A. (1994). An Journal of Marriage and the Family, 60, 493- integration of theory: Adolescent sexual contacts. 504. Journal of Youth and Adolescence, 23(3), 403- 420. Brindis, C., Starbuck-Morales, S., Wolfe, A.L., & McCarter, V. (1994). Characteristics associated Benson, M.D., & Torpy, E.F. (1995). Sexual with contraceptive use among adolescent females behavior in junior high school students. in school-based family planning programs. Obstetrics and Gynecology, 85(2), 279-284. Family Planning Perspectives, 26(4), 160-164.

Bickel, R., Weaver, S., Williams, T., & Lange, L. Brindis, C., Boggess, J., Katsuranis, F., Mantell, (1997.) Opportunity, community, and teen M., McCarter, V., & Wolfe, A. (1998). A profile pregnancy in an Appalachian state. Journal of of the adolescent male family planning client. Educational Research, 90(3), 175-181. Family Planning Perspectives, 30(2), 63-66, 88.

Billy, J.O.G., Brewster, K.L., & Grady, W.R. Brooks-Gunn, J., Duncan, G.J., Klebanov, P.K., (1994). Contextual effects on the sexual behav- & Sealand, N. (1993). Do neighborhoods influ- ior of adolescent women. Journal of Marriage ence child and adolescent development? and the Family, 56(2), 387-404. American Journal of Sociology, 99(2), 353-395.

Blum, R.W., Beuhring, T., & Rinehart, P.M. Brown, J.D., & Newcomer, S.F. (1991). (2000). Protecting teens: Beyond race, income and Television viewing and adolescents' sexual behav- family structure. Minneapolis, MN: Center for ior. Journal of Homosexuality, 21(1-2), 77-91. Adolescent Health, University of Minnesota. Brown, L.K., DiClemente, R.J., & Park, T. Blum, R.W., Beuhring, T., Shew, M.L., (1992). Predictors of condom use in sexually Bearinger, L.H., Sieving, R.E., & Resnick, M.D. active adolescents. Journal of Adolescent Health, (2000). The effects of race, income, and family 13(8), 651-657. structure On adolescent risk behavior. American Journal of Public Health, 90(12), 1879-84. Browning, C.R., & Laumann, E.O. (1997). Sexual contact between children and adults: A Boyer, C.B., Tschann, J.M., & Shafer, M. (1991). life course perspective. American Sociological Predictors of risk for sexually transmitted diseases Review, 62(4), 540-560. in ninth grade urban high school students. Journal of Adolescent Research, 14(4), 448-465. Burge, V., Felts, M., Chenier, T., & Parrillo, A.V. (1995). Drug use, sexual activity, and suici- Boyer, D., & Fine, D. (1992). Sexual abuse as a fac- dal behavior in U.S. high school students. tor in adolescent pregnancy and child maltreatment. Journal of School Health, 65(6), 222-227. Family Planning Perspectives, 24(1), 4-11, 19. Capaldi, D.M., Crosby, L., & Stoolmiller, M. Braverman, P., & Strasburger, V.C. (1993). (1996). Predicting the timing of first sexual Adolescent sexual activity. Clinical Pediatrics, intercourse for at-risk adolescent males. Child 32(11), 658-668. Development, 67(2), 344-359.

Brewster, K.L. (1994). Race differences in sexual Carvajal, S., Parcel, G.S., Basen-Engquist, K., activity among adolescent women: The role of Banspach, S.W., Coyle, K.K., & Kirby, D.B. neighborhood characteristics. American (1999). Psychosocial predictors of the delay of Sociological Review, 59( 3):408-424. sexual intercourse by adolescents. Health Psychology, 18(5), 443-452. Brewster, K.L., Billy, J.O.G., & Grady, W.R. (1993). Social context and adolescent behavior: Catania, J., Coates, T., Kegeles, S., Peterson, J., The impact of community on the transition to Marin, B., & Fullilove, M. (1989). Predicting sexual activity. Social Forces, 71(3), 713-740. risk behavior in a random household probability

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 60 7 1 sample: The 'Amen" Study. San Francisco, CA: Darroch, J.E., Landry, D.J., & Oslak, S. (1999). University of California, Center for AIDS Age differences between sexual partners in the Prevention Studies. United States. Family Planning Perspectives, 31(4), 160-167. Centers for Disease Control and Prevention. (1996). CDC surveillance summaries. Morbidity Dawson, D.A. (1986). The effects of sex and Mortality Weekly Report, 45(No. SS-4), 64-65. education on adolescent behavior. Family Planning Perspectives, 18(4), 162-170. Chandy, J.M., Harris, L., Blum, R.W., & Resnick, M.D. (1994, December). Female adolescents of Day, RD. (1992). The transition to first inter- alcohol misusers: Sexual behaviors. Journal of course among racially and culturally diverse Youth and Adolescence, 23(6), 695-709. youth. Journal of Marriage and the Family, 54(4), 749-762. Christopher, F.S., Johnson, D.C., & Roosa, M. (1993). Family, individual, and social correlates Devine, D., Long, P., & Forehand, R. (1993). A of early Hispanic adolescent sexual expression. prospective study of adolescent sexual activity: The Journal of Sex Research, 30(1), 54-61. Description, correlates, and predictors. Advances in Behaviour Research and Therapy, 15(3), 185-209. Collins, J.K., & Robinson, L. (1986). The contraceptive knowledge, attitudes and practice of DiBlasio, F.A., & Benda, B.B. (1990). unmarried adolescents. Australian Journal of Adolescent sexual behavior: Multivariate analysis Sex, Marriage and Family, 7(3), 132-152. of a social learning model. Journal of Adolescent Research, 5(4), 449-466. Costa, F.M., Jessor, R., Donovan, J.E., & Fortenberry, J.D. (1995). Early initiation of DiClemente, R.J. (1990). The emergence of sexual intercourse: The influence of psychosocial adolescents as a risk group for human immuno- unconventionality. Journal of Research on deficiency virus infection. Journal of Adolescent Adolescence, 5(1), 93-121. Research, 5(1), 7 -17.

Costa, F.M., Jessor, R., Fortenberry, J.D., & DiClemente, R.J., Durbin, M., Siegel, D., Donovan, J.E. (1996). Psychosocial conventionality, Krasnovsky, F., Lazarus, N., & Comacho, T. health orientation, and contraceptive use in adoles- (1992). Determinants of condom use among cence. Journal of Adolescent Health, 18(6), 404-16. junior high school students in a minority, inner- city school district. Pediatrics, 89(2), 197-202. Cox, J., Emans, S.J., & Bithoney, W. (1993). Sisters of teen mothers: Increased risk for adoles- DiClemente, R.J., Lodico, M., Grinstead, O.A., cent parenthood. Journal of Adolescent and Harper, G., Rickman, RL., Evans, P.E., & Pediatric Gynecology, 6, 138-142. , Coates, T.J. (1996). African-American adolescents residing in high-risk urban environments do use Crane, J. (1991). Effects of neighborhoods on condoms: Correlates and predictors of condom dropping out of school and teenage childbear- use among adolescents in public housing developing. In C. Jencks, & P.E. Peterson (Eds.), The ments. Pediatrics, 98(2, part 1), 269-278. urban underclass. Washington DC: The Brookings Institute. Dittus, P.J., & Jaccard, J. (2000). Adolescents' perceptions of maternal disapproval of sex: Crosby, R.A., Yarber, W.L., & Kanu, A.J. Relationship to sexual outcomes. Journal of (1998). Relationship of HIV/STD sexual risk Adolescent Health, 26(4), 268-278. behaviors to other health behaviors among a sample of Indiana rural youth. The Health Dorius, G., & Barber, B. (1998). Parental Education Monograph, 16(2), 51-59. support and control and the onset of sexual intercourse. Unpublished manuscript. Cvetkovich, G., & Grote, B. (1981). Psychosocial maturity and teenage contraceptive Dorius, G.L., Heaton, T.B., & Steffen, P. use: An investigation of decision-making and (1993). Adolescent life events and their associa- communication skills. Population and tion with the onset of sexual intercourse. Youth Environment, 4(4), 211-226. and Society, 25(1), 3-23.

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 61 72 Durbin, M., Diclemente, R.J., Siegel, D., Ford, K., & Norris, A.E. (1993). Urban Krasnovsky, F., Lazarus, N., & Camacho, T. Hispanic adolescents and young adults: (1993). Factors associated with multiple sex Relationship of acculturation to sexual behavior. partners among junior high school students. Journal of Sex Research, 30(4), 316-323. Journal of Adolescent Health, 14(3), 202-207. Forste, R.T., & Heaton, T.B. (1988). Initiation DuRant, R.H., & Sanders, J.M. (1989). Sexual of sexual activity among female adolescents. behavior and contraceptive risk taking among Youth and Society, 19(3), 250-268. sexually active adolescent females. Journal of Adolescent Health Care, 10(1), 1-9. Forrest, J.D., Hermalin, A.I., & Henshaw, S.K. (1981). The impact of family planning clinic East, P.L. (1996a). The younger sisters of child- programs on adolescent pregnancy. Family bearing adolescents: Their attitudes, expectations, Planning Perspectives, 13(3), 109-116. and behaviors. Child Development, 67(2), 267-282. Fortenberry, J.D., Costa, F.M., Jessor, R., & East, P.L. (19966). Do adolescent pregnancy Donovan, J.E. (1997). Contraceptive behavior and childbearing affect younger siblings? Family and adolescent lifesytles: A structural modeling Planning Perspectives, 28(4), 148-153. approach. Journal of Research on Adolescence, 7(3), 307-329. East, P.L., & Felice, M.E. (1992). Pregnancy risk among the younger sisters of pregnant and child- Foshee, V., & Bauman, K. (1992). Gender bearing adolescents. Journal of Developmental stereotyping and adolescent sexual behavior: A and Behavioral Pediatrics, 13(2), 128-136. test of temporal order. Journal of Applied Social Psychology, 22(20), 1561-1579. East, P.L., Felice, M.E., & Morgan, M. (1993). Sisters' and girlfriends' sexual and childbearing Fox, G.L., & Inazu, J.K. (1980). Patterns and behavior: Effects on early adolescent girls' sexual outcomes of mother-daughter communication outcomes. Journal of Marriage and the Family, about sexuality. Journal of Social Issues, 36(1), 55(4), 953-963. 7-29. Friede, A., Hogue, C.J., Doyle, L., Hammerslough, Ensminger, M.E. (1990). Sexual activity and C.R., Sniezek, J.E., & Arrighi, H. (1986). Do problem behaviors among black, urban adoles- sisters of childbearing teenagers have increased cents. Child Development, 61, 2032-2046. rates of childbearing? American Journal of Public Health, 76(10), 1221-1224. Fishbein, M., Bandura, A., Triandis, H.C., Kanfer, F.H., Becker, M.H., & Middlestadt, S.E. Furstenberg, F.F., Moore, K.A., & Peterson, J.L. (1991, October). Factors influencing behavior (1985). Sex education and sexual experience and behavior change. Final report to the among adolescents. American Journal of Public Theorist's Workshoii, Washington, DC. Health, 75(11), 1131-1132.

Flaherty, E.W, Marecek, J., Olsen, K., & Wilcove, Galavotti, C., & Lovick, S.R. (1989). School- G. (1983). Preventing adolescent pregnancy: An based clinic use and other factors affecting ado- interpersonal problem-solving approach. lescent contraceptive behavior. Journal of Prevention in Human Services, 2(3), 49-64. Adolescent Health Care, 10(6), 506-512.

Flannery, D, Rowe, D., & Gulley, B. (1993). Gershenson, H.P., Musick, J.S., Ruch-Ross, Impact of pubertal status, timing, and age on H.S., Magee, V., Rubino, K.K., & Rosenberg, adolescent sexual experience and delinquency. D. (1989). The prevalence of coercive sexual Journal of Adolescent Research, 8(1), 21-40. experience among teenage mothers. Journal of Interpersonal Violence, 4(2), 204-219. Flewelling, R.L., & Bauman, K.E. (1990). Family structure as a predictor of initial Gibbs, J.T. (1986). Psychosocial correlates of substance use and sexual intercourse in eakly sexual attitudes and behaviors in urban early ado- adolescence. Journal of Marriage and the Family, lescent females: Implications for intervention. 52(1), 171-181. Journal of Social Work and Human Sexuality

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 62 713 (Special Issue Adolescent sexualities: Overviews Halpern, C.T., Udry, J.R., & Suchindran, C. and principles of intervention); 5(1), 81-97. (1997). Testosterone predicts initiation of coitus in adolescent females. Psychosomatic Medicine, Gibson, J.W., & Kempf, J. (1990). Attitudinal 59(2), 161-171. predictors of sexual activity in Hispanic adolescent females. Journal of Adolescent Research, Handelsman, C.D., Cabral, R.J., & Weisfeld, 5(4), 414-430. G.E. (1987). Sources of information and adolescent sexual knowledge and behavior. Journal of Gispert, M., Brinich, P., Wheeler, K., & Krieger, Adolescent Research, 2(4), 455-463. L. (1984). Predictors of repeat pregnancies among low-income adolescents. Hospital and Hanson, S.L., Morrison, D.R., & Ginsberg, A.L. Community Psychiatry, 35(7), 719-723. (1989). The antecedents of teenage fatherhood. Demography, 26, 579-596. Gohel, M., Diamond, J.J., & Chambers, C.V. (1997). Attitudes toward sexual responsibility Hanson, S.L., Myers, D.E., & Ginsburg, A.L. and parenting: An exploratory study of young (1987). The role of responsibility and knowledge urban males. Family Planning Perspectives, 29(6), in reducing teenage out-of-wedlock childbearing. 280-283. Journal of Marriage and the Family, 49(2), 241-256. Goodwin, F.K. (1990). Which kids use condoms or don't. Journal of the American Medical Harvey, S.M., & Spigner, C. (1995). Factors Association, 264(11), 1389. associated with sexual behavior among adolescents: A multivariate analysis. Adolescence, Gorosh, M. (1982). Patterns of contraceptive 30(18), 153-264. use among female adolescenis. Journal of Adolescent Health Care, 3(2), 96-102. Haurin, R.J., & Mott, F.L. (1990). Adolescent sexual activity in the family context: The impact Grady, W.R., Hayward, M.D., & Billy, J.O.G. of older siblings. Demography, 27, 537-557. (1989). A dynamic model of premarital pregnancy in the U.S. Final report to the Office of Hayward, M.D., Grady, W.R., & Billy, J.O.G. Population Affairs. Washington, DC: U.S. (1992). The influence of socioeconomic status Department of Health and Human Services. on adolescent pregnancy. Social Science Quarterly, 73(4), 750-772. Guagliardo, M.F., Huang, Z., & D'Angelo, L.J. (1999). Fathering pregnancies: Marking health- Hingson, RW., Strunin, L., Berlin, B.M., & risk behaviors in urban adolescents. Journal of Heeren, T. (1990). Beliefs about AIDS, use of Adolescent Health, 24(1), 10-15. alcohol, drugs and unprotected sex among Massachusetts adolescents. American Journal of Halpern, C.T., Joyner, K., Udry, R., & Public Health, 80(3), 295-299. Suchindran, C. (2000). Smart teens don't have sex (or kiss much either). Journal of Adolescent Hogan, D.P., Astone, N.M., & Kitagawa, E.M. Health, 26(3), 213-225. (1985). Social and environmental factors influencing contraceptive use among Black adolescents. Halpern, C.T., Udry, J.R., Campbell, B., & Family Planning Perspectives, 17(4), 165-169. Suchindran, C. (1993). Testosterone and pubertal development as predictors of sexual activity: A Hogan, D.P., & Kitagawa, E.M. (1985). The panel analysis of adolescent males. Psychosomatic impact of social status, family structure, and neigh- Medicine, 55(5), 436-447. borhood on the fertility of Black adolescents. American Journal of Sociology, 90(4), 825-855. Halpern, C.T., Udry, J.R., Campbell, B., Suchindran, C., & Mason, G.A. (1994). Holden, G.W., Nelson, P.B., Velasquez, J., & Testosterone and religiosity as predictors of sex- Ritchie, K.L. (1993). Cognitive, psychosocial, ual attitudes and activity among adolescent and reported sexual behavior differences between males: A biosocial model. Journal of Biosocial pregnant and nonpregnant adolescents. Science, 26(2), 217-234. Adolescence, 28(111), 557-572.

Emerging Answers: Research Findings on PrograrnpoiReduce Teen Pregnancy 63 e Holmbeck, G.N., Crossman, R.E., Wandrei, Jessor, R., Costa, F., Jessor, L., & Donovan, J.E. M.L., & Gasiewski, E. (1994). Cognitive devel- (1983). Time of first intercourse: A prospective opment, egocentrism, self-esteem, and adoles- study. Journal of Personality and Social cent contraceptive knowledge, attitudes, and Psychology, 44(3), 608-626. behavior. Journal of Youth and Adolescence, 23(2), 169-193. Jorgensen, S.R., King, S.L., & Torrey, B.A. (1980). Dyadic and social network influences on Holtzman, D., & Rubinson, R. (1995). Parent adolescent exposure to pregnancy risk. Journal ,and peer communication effects on AIDS-related of Marriage and the Family, 42(1), 141-155. behavior among U.S. high school students. Family Planning Perspectives, 27(6), 235-240. Jorgensen, S.R., & Sonstegard, J.S. (1984). Predicting adolescent sexual and contraceptive Hou, S.I., & Basen-Engquist, K. (1997). behavior: An application and test of the Fishbein Human immunodeficiency virus risk behavior model. Journal of Marriage and the Family, among white and Asian/Pacific Islander high 46(1), 43-55. school students in the United States: Does culture make a difference? Journal of Adolescent Kahn, J.R., & Anderson, K.E. (1992). Health, 20(1), 68-74. Intergenerational patterns of teenage fertility. Demography, 29(1), 39-57. Hovell, M., Sipan, C., Blumberg, E., Atkins, C., Hofstetter, C.R., & Kreitner, S. (1994). Family Karofsky, P.S., Zeng, L., & Kosorok, M.R. influences on Latino and Anglo adolescents' sex- (2000). Relationship between adolescent- ual behavior. Journal of Marriage and the parental communication and initiation of first Family, 56(4), 973-986. intercourse by adolescents. Journal of Adolescent Health, 28(1), 41-45. Inazu, J.K., & Fox, G.L. (1980). Maternal influ- Kasen, S., Cohen, P., & Brook, J.S. (1998). ence on the sexual behavior of teen-age daugh- Adolescent school experiences and dropout, ado- ters: Direct and indirect sources. Journal of lescent pregnancy, and young adult deviant behav- Family Issues, 1(1), 81-99. ior. Journal of Adolescent Research, 13(1), 49-72. Ireson, D.J. (1984). Adolescent pregnancy and Kastner, L.S. (1984). Ecological factors predict- sex roles. Sex Roles, 11(3-4), 189-201. ing adolescent contraceptive use: Implications for intervention. Journal of Adolescent Health Care, Jaccard. J., & Dittus, P. (1991). Parent-teen 5(2), 79-86. communication: Thward the prevention of unintended pregnancies. New York, NY: Kenney, J.W., Reinholtz, C., & Angelini, P.J. Springer-Verlag. (1997). Ethnic differences in childhood and adolescent sexual abuse and teenage pregnancy. Jaccard, J., Dittus, P.J., & Gordon, V.V. (1996). Journal of Adolescent Health, 21(1), 3-10. Maternal correlates of adolescent sexual and contraceptive behavior. Family Planning Perspectives, Kinsman, S.B., Romer, D., Furstenberg, F.F., & 28(4), 159-165, 185. Schwarz, D.F. (1998). Early sexual initiation: The role of peer norms. Pediatrics, 102(5), Jaccard, J., Ditius, P.J., & Litardo, H.A. (1999). 1185-1192. Parent-adolescent communication about sex and birth control: Implications for parent-based Kirby, D. (1999). Looking for reasons why: The interventions to reduce unintended adolescent antecedents of adolescent sexual risk-taking, preg- pregnancy. In W. Miller & L. Severy (Eds.), nancy, and childbearing. Washington, DC: Advances in population: Psychological perspectives. National Campaign to Prevent Teen Pregnancy. London: Kingsley Publishers. Kowaleski-Jones, L., & Mott, F. (1998). Sex, Jensen, L.C., DeGaston, J.F., & Wee, S.E. contraception and childbearing among high-risk (1994). Societal and parental influences on youth: Do different factors influence males and adolescent sexual behavior. Psychological Reports, females? Family Planning Perspectives, 30(4), 75(2), 928-930. 163-169.

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 64 Ku, L., Sonenstein, F.L., & Pleck, J.H. (1992). Leitenberg, H., & Saltzman, H. (2000). A The association of AIDS education and sex edu- statewide survey of age at first intercourse for cation with sexual behavior and condom use adolescent females and age of their male part- among teenage men. Family Planning ners: Relation to other risk behaviors and statu- Perspectives, 24(3), 100-106. tory rape implications. Archives of Sexual Behavior, 29(3), 203-15. Ku, L., Sonenstein, F.L., & Pleck, J.H. (1993a). Factors influencing first intercourse for teenage Leland, N.L., & Barth, R.P. (1993). men. Public Health Reports, 108(6), 680-694. Characteristics of adolescents who have attempted to avoid EIW and who have commu- Ku, L., Sonenstein, F.L., & Pleck, J.H. (1993b). nicated with parents about sex. Journal of Neighborhood, family, and work: Influences on Adolescent Research, 8(1), 58-76. the premarital behaviors of adolescent males. Social Forces, 72(2), 479-503. Little, C.B1, & Rankin, A. (forthcoming). Why do they start? Explaining reported early-teen sex- Ku, L., Sonenstein, F.L., & Pleck, J.H. (1993c). ual activity. Sociological Forum. Young men's risk behaviors for HIV infection and sexually transmitted diseases, 1988 through Lock, S.E., & Vincent, M.L. (1995). Sexual 1991. American Journal of Public Health, decision-making among rural adolescent females. 83(11), 1609-1615. Health Values: The Journal of Health Behavior, Education and Promotion, 19(1), 47-58. Ku, L., Sonenstein, F.L., & Pleck, J.H. (1994). The dynamics of young men's condom use dur- Lowry, R., Holtzman, D., Truman, B.I., Kann, ing and across relationships. Family Planning L., Collins, J.L., & Kolbe, L.J. (1994). Perspectives, 26(6), 246-251. Substance use and HW-related sexual behaviors among U.S. high school students: Are they Ku, L., Sonenstein, F.L., Lindberg, L.D., related? American Journal of Public Health, Bradner, C.H., Boggess, S., & Pleck, J.H. 84(7), 1116-1120. (1998). Understanding changes in sexual activity among young metropolitan men: 1979-1995. Lundberg, S. (1990). Effects of state welfare, Family Planning Perspectives, 30(6), 256-262. abortion, and family planning policies on premarital childbearing among white adolescents. Lammers, C., Ireland, M., Resnick, M., & Blum, Family Planning Perspectives, 22(6), 246-251. R. (2000). Influences on adolescents' decisions to postpone onset of sexual intercourse: A survival Lundberg, S., & Plotnick, R.D. (1995). analysis of virginity among youths aged 13 to 18 Adolescent premarital childbearing: Do eco- years. Journal of Adolescent Health, 26(1), 42-48. nomic incentives matter? Journal of Labor Economics, 13(2), 177-200. Landry, D.J., & Forrest, J.D. (1995).' How old are U.S. fathers? Family Planning Perspectives, Luster, T., & Small, S.A. (1994). Factors associ- 27(4), 159-161. ated with sexual risk-taking behaviors among adolescents. Journal of Marriage and the Family, Landry, E., Bertrand, J.T., Cherry, F., & Rice, J. 56(3), 622-632. (1986). Teen pregnancy in New Orleans: Factors that differentiate teens who deliver, abort, and Luster, T., & Small, S.A. (1997). Sexual abuse successfully contracept. Journal of Youth and history and number of sex partners among Adolescence, 15(3), 259-274. female adolescents. Family Planning Perspectives, 29(5), 204-211. Lauritsen, J.L. (1994). Explaining race and gender differences in adolescent sexual behavior. Males, M., & Chew, K.S. (1996). The ages of Social Forces, 72(3), 859-883. fathers in California adolescent births. American Journal of Public Health, 86(4), 565-568. Lauritsen, J.L., & Swicegood, C.G. (1997). The consistency of self-reported initiation of sexual Manlove, J. (1995). Early motherhood on an activity. Family Planning Perspectives, 29(5), intergenerational perspective: The experiences of a 215-221. British cohort. Unpublished manuscript.

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 65 Manlove, J. (1998). The influence of high Middleman, A.B., Faulkner, A.H., Woods, E.R., school dropout and school disengagement on Emans, S.J., & DuRant, R_H. (1995). High-risk the risk of school-age pregnancy. Journal of behaviors among high school students in Research on Adolescence, 8(2), 187-220. Massachusetts who use anabolic steroids. Pediatrics, 96(2, Part 1), 268-272. Manlove, J., Terry, E., Gitelson, L., Papillo, A.R., & Russel, S. (2000). Explaining demo- Miller, B.C. (1995). Risk factors for adolescent graphic trends in teenage fertility, 1980-1995. nonmarital childbearing. In Report to Congress on Family Planning Perspectives, 32(4), 166-175. out-of-wedlock childbearing (pp. 217-227). Washington, DC: U.S. Department of Health Manning, W.D., Longmore, M.A., & Giordano, and Human Services. P.C. (2000). The relationship context of contraceptive use at first intercourse. Family Planning Miller, B.C., & Bingham, C.R. (1989). Family Perspectives, 32(3), 104-110. configuration in relation to the sexual behavior of female adolescents. Journal of Marriage and Marin, B.V., Coyle, K.K., Gomez, C.A., Carvajal, the Family, 51(2), 499-506. S.C., & Kirby, D.B. (2000). Older boyfriends and girlfriends increase risk of sexual initiation in Miller, B.C., Christensen, R.B., & Olson, T.D. young adolescents. Journal of Adolescent Health, (1987). Adolescent self-esteem in relation to sex- 27(6), 409-418. ual attitudes and behavior. Youth and Society, 19(1), 93-111. Marsiglio, W. (1993). Adolescent males' orientation toward paternity and Contraception. Family Miller, B.C., Monson, B.H., & Norton, M.C. Planning Perspectives, 25(1), 22-31. (1995). The effects of forced sexual intercourse Marsiglio, W, & Mott, F. (1986). The impact of on white female adolescents. Child Abuse and sex education on sexual activity, contraceptive use, Neglect, 19(10), 1289-1301. and premarital pregnancy among American teenagers. Family Planning Perspectives, 18(4), 151-162. Miller, B.C., McCoy, J.K., Olson, T.D., & Wallace, C.M. (1986). Parental discipline and Mauldon, J., & Luker, K. (1996). The effects of control attempts in relation to adolescent sexual contraceptive education on method use at first attitudes and behavior. Journal of Marriage and intercourse: Family Planning Perspectives, 28(1), the Family, 48, 503-512. 19-24, 41. Miller, B.C., Norton, M.C., Curtis, T., Hill, Mayer, S.E., & Jencks, C. (1989). Growing up E.J., Schvaneveldt, P., & Young, M.H. (1997). in poor neighborhoods: How much does it mat- The timing of sexual intercourse among adoles- ter? Science, 243(4897), 1441-1445. cents: Family, peer, and other antecedents. Youth and Society, 29(1),' 54-83.

McBride, C.M., Curry, S.J., Cheadle, A., . Anderman, C., Wagner, E.H., Diehr, P., & Psaty, Miller, B.C., Norton, M.C., Fan, I., & B. (1995). School-level application of a social Christopherson, C.R. (1998). Pubertal bonding model to adolescent risk-taking behav- development, parental communication, and ior. Journal of School Health, 65(2), 63-68. sexual values in relation to adolescent sexual behaviors. Journal of Early Adolescence, 18(1), Melchert, T., & Burnett, K.F. (1990). Attitudes, 27-52. knowledge, and sexual behavior of high-risk adolescents: Implications for counseling and sexual- Miller, B.C., & Olson, T.D. (1988). Sexual atti- ity education. Journal of Counseling and tudes and behavior of high school students in Development, 68(3), 293-298. relation to background and contextual factors. Journal of Sex Research, 24, 194-200. Meschke, L.L., Zweig, J.M., Barber, B.L., & Eccles, J.S. (2000). Demographic, biological, Miller, B.C., & Sneesby, K.R. (1988). psychological, and social predictors of the timing Educational correlates of adolescents' sexual atti- of first intercourse. Journal of Research on tudes and behavior. Journal of Youth and Adolescence, 10(3), 315-338. Adolescence, 17(6), 521-530.

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 66 Miller, K.E., Sabo, D:F., Farrell, M.P., Barnes, Moore, K.A., Peterson, J.L., & Furstenberg, F.F. G.M., & Melnick, M,J. (1998). Athletic partici- (1986). Parental attitudes and the occurrence of pation and sexual behavior in adolescents: The early sexual activity. Journal of Marriage and the different world of boys and girls. Journal of Family, 48, 777-782, Health and Social Behavior, 39(2), 108-123. Morrison, D.M., (1989). Predicting contracep- Miller, K.S., Clark, L.F., & Moore, J.S. (1997). tive efficacy: A discriminant analysis of three Sexual initiation with older male partners and groups of adolescent women. Journal of Applied subsequent HIV risk behavior among female Social Psychology, 19(17 , part 1), 1431- 1452. adolescents. Family Planning Perspectives, 29(5), 212-214. Mott, F.L., Fondell, M.M., Hu, P.N., Kowaleski- Jdnes, L., & Menaghan, E.G. (1996). The deter- Miller, K.S., Forehand, R., & Kotchick, B.A. minants of first sex by age 14 in a high-risk (1999). Adolescent sexual behavior in two ethnic adolescent population. Family Planning minority samples: The role of family variables. Perspectives, 28(1), 13-18. Journal of Marriage and the Family, 61(1), 85-90. Mueller, K.E., & Powers, W.G. (1990). Parent- Miller, K.S., Levin, M.L., Whitaker, D.J., & Xu, child sexual discussions: Perceived communicator X. (1998). Patterns of condom use among ado- style and subsequent behavior. Adolescence, lescents: The impact of mother-adolescent com- 25(98), 469-482. munication. American Journal of Public Health, 88(10), 1542-1544. Murphy, J.J., & Boggess, S. (1998). Increased condom use among teenage males, 1988-1995: Miller, W.B., Pasta, D.J., MacMurray, J., Chiu, The role of attitudes. Family Planning C., Wu, H., & Comings, D.E. (1998). Dopamine Perspectives, 30(6), 276-280, 303. receptor genes are associated with age at first sexual intercourse. Journal of Biosocial Science, Murry, V.M. (1992). Incidence of first preg- 31(part 1), 17-28. nancy among Black adolescent females over three decades. Youth and Society, 23(4), 478-506. Millstein, S.G., & Moscicki, A.B. (1995). Sexually-transmitted disease in female adoles- Nagy, S., DiClemente, R.J., & Adcock, A.G. cents: Effects of psychosocial factors and high- (1995). Adverse factors associated with forced risk behaviors. Journal of Adolescent Health, sex among southern adolescent girls. Pediatrics, 17(2), 83-90. 96, 944-946. Moore, K.A, Blumenthal, C., Sugland, B.W., Hyatt, B., Snyder, N.O., & Morrison, D.R. National Center on Addiction and Substance (1994). State variation in rates of adolescent Abuse. (1999). Dangerous liaisons: Substance pregnancy and childbearing. Washington, DC: abuse and sex. Washington, DC: National Center Child Trends, Inc. on Addiction and Substance Abuse.

Moore, K.A., Morrison, D.R., & Glei, D.A. Nesmith, J.D., Klerman, L.V., Oh, M.K., & (1995). Welfare and adolescent sex: The effects Feinstein, R.A. (1997). Procreative experiences of family history, benefit levels, and community and orientations toward paternity held by incar- context. Journal of Family and Economic Issues cerated adolescent males. Journal of Adolescent (Theme issue Family economic policy: The inter- Health, 20(3), 198-203. generational effects of welfare), 16(2-3), 207-237. Newcomer, S.F., & Baldwin, W. (1992). Moore, K.A., Manlove, J., Glei, D.A., & Morrison, Demographics of adolescent sexual behavior, D.R. (1998). Nonmarital school-age motherhood: contraception, pregnancy, and STDs. journal of Family, individual, and school characteristics. School Health, 62(7), 265-270. Journal of Adolescent Research, 13(4), 433-457. Newcomer, S.F., & Udry, J.R. (1984). Mothers' Moore, K.A., Nord, C.W., & Peterson, J.L. (1989). influence on sexual behavior of their teenage Nonvoluntary sexual activity among adolescents. children. Journal of Marriage and the Family, 46, Family Planning Perspectives, 21(3), 110-114. 477-485.

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 67 ( 6 Newcomer, S.F., & Udry, J.R. (1985). Parent- between perceived cost-benefits and consistency. child communication and adolescent sexual Journal of Marriage and the Family, 53(3), 733- behavior. Family Planning Perspectives, 17(4), 745. 169-174. Pleck, J.H., Sonenstein, F.L., & Ku, L. (1993a). Newcomer, S.F., & Udry, J.R. (1987). Parental Changes in adolescent males' use of and attitudes marital status effects on adolescent sexual behav- toward condoms, 1988-1991. Family Planning ior. Journal of Marriage and the Family, 49(2), Perspectives, 25(3), 106-110, 117. 235-240. Pleck, J.H., Sonenstein, F.L., & Ku, L.C. Ohannessian, C., & Crockett, L. (1993). A lon- (1993b). Masculine ideology: Its impact on ado- gitudinal investigation of the relationship lescent males' heterosexual relationships. Journal between educational investment and adolescent of Social Issues, 49(3), 11-29. sexual activity. Journal of Adolescent Research, 8(2), 167-182. Pleck, J.H., Sonenstein, F.L., & Swain, S.O. (1988). Adolescent males' sexual behavior and Orr, D.P., Wilbrandt, M.L., Brack, C.J., Rauch, contraceptive use: Implications for male responsi- S.P., & Ingersoll, G.M. (1989). Reported sexual bility. Journal of Adolescent Research, 3(3-4), 275- behaviors and self-esteem among young adoles- 284. cents. American Journal of Diseases of Children, 143, 86-90. Plotnick, RD., & Butler, S.S. (1991). Attitudes and adolescent nonmarital childbearing: Evidence Pendergrast, R.A., DuRant, RH., & Gaillard, from the National Longitudinal Survey of Youth. G.I. (1992). Attitudinal and behavioral correlates Journal of Adolescent Research, 6(4), 470-492. of condom use in urban adolescent males. Journal of Adolescent Health, 13(2), 133-139. Plomick, RD. (1992). The effects of attitudes on teenage premarital pregnancy and its resolution. Pete-McGadney, J. (1995). Differences in adolescent self-concept as a fimction of race, geo- American Sociological Review, 57, 800-811. graphic location, and pregnancy. Adolescence, 30(117), 95-105. Raine, T.R., Jenkins, R, Aarons, S.J., Woodward, K., Fairfax, J.L., El-Khorazaty, M.N., & Herman, Philliber, S., Namerow, P.B., Kaye, J.W., & A. (1999). Sociodemographic correlates of virgin- Kunkes, C.H. (1986). Pregnancy risk-taking ity in seventh-grade black and Latino students. among adolescents. Journal of Adolescent Journal of Adolescent Health, 24(5), 304-312. Research, 1(4), 463-481. Reitman, D., St. Lawrence, J.S., Jefferson, K.W., Phinney, V.G., Jensen, L.C., Olsen, J.A., & Alleyne, E., Brasfield, T.L., & Shirley, A. (1996). Cundick, B. (1990). The relationship between Predictors of African American adolescents' con- early development and psychosexual behaviors dom use and HIV risk behavior. AIDS Education in adolescent females. Adolescence, 25(98), and Prevention, 8(6), 499-515. 321-332. Resnick, M.D., Bearman, P.S., Blum, R.W., Pierre, N., Shrier, L.A., Emans, S.J., & DuRant, Bauman, K.E., Harris, K.M., Jones, J., Tabor, J., R.H. (1998). Adolescent Males involved in preg- Beuhring, T., Sieving, RE., Shew, M., Ireland, nancy: Associations of forced sexual contact and M., Bearinger, L.H., & Udry, J.R (1997). risk behaviors. Journal of Adolescent Health, Protecting adolescents from harm: Findings from 23(6), 364-369. the National Longitudinal Study on Adolescent Health. Journal of American Medical Association, Neck, J.H. (1989). Correlates of black adoles- 278(10), 823-832. cent males' condom use. Journal of Adolescent Research (Special issue: Black adolescents), 4(2), Resnick, M.D., Chambliss, S., & Blum, R.W. 247-253. (1993). Health risk behaviors of urban adolescent males involved in pregnancy. Families in Society: Pleck, J.H., Sonenstein, F.L., & Ku, L. (1991). Journal of Contemporary Human Services, 74, Adolescent males' condom use: Relationships 366-374.

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 68 (9 Richter, D.L., Valois, F.G., McKeown, R.E., & romantic relationships and sexual behaviors of Vincent, M.L. (1993). Correlates of condom use young adolescent girls. Journal of Adolescent and number of sexual partners among high school Health, 21(4), 238-243. adolescents. Journal of School Health, 63(2), 91-96. Rowe, D.C., & Rodgers, J.L. (1994). A social Rickman, R.L., Lodico, M., DiClemente, KJ., contagion model of adolescent sexual behavior: Morris, R., Baker, C., & Huscroft, S. (1994). Explaining race differences. Social Biology, 41(2), Sexual communication is associated with condom 1-18. use by sexually active incarcerated adolescents. Journal of Adolescent Health, 15(5), 383-388. Rowe, D.C., Rodgers, J.L., & Meseck-Bushey, S. (1989). An 'epidemic' model of sexual intercourse Robbins, C., Kaplan, H.B., & Martin, S.S. prevalences for black and white adolescents. Social (1985). Antecedents of pregnancy among unmar- Biology, 36, 127-145. ried adolescents. Journal of Marriage and the Family, 47(3), 567-583. Sabo, D., Miller, K., Farrell, M., Barnes, G., & Melnick, M. (1998). The Women's Sports Robinson, K.L., Telljohann, S.K., & Price, J.H. Foundation report: Sport and teen pregnancy. East (1999). Predictors of sixth graders engaging in Meadow, NY: Women's Sports Foundation. sexual intercourse. Journal of School Health, 69(9), 369-375. Santelli, J.S., & Beilenson, P. (1992). Risk factors for adolescent sexual behavior, fertility, and sexu- Robinson, KB., & Frank, D.I. (1994). The rela- ally transmitted diseases. Journal of School Health, tion between self-esteem, sexual activity, and 62(7), 271-279. pregnancy. Adolescence, 29(113), 27-35. Santa, J.S., Brener, N.D., Lowry, R., Bhatt, A., Rodgers, J.L. (1983). Family configuration and & Zabin, L.S. (1998). Multiple sexual partners adolescent sexual behavior. Population and among U.S. adolescents and young adults. Family Environment, 6, 73-83. Planning Perspectives, 30(6), 271-275.

Rodgers, J.L., & Rowe, D.C. (1988). Influence of Santelli, J.S., Lowry, K, Brener, N.D., & Robin, L. siblings on adolescent sexual behavior. (2000). The association of sexual behaviors with Developmental Psychology, 24(5), 722-728. socioeconomic status, family structure, and race/ethnicity among U.S. adolescents. American Rodgers, J.L., & Rowe, D.C. (1990). Adolescent Journal of Public Health, 90(10), 1582-1588. sexual activity and mildly deviant behavior: Sibling and friendship effects. Journal of Family Issues, Scher, P.W., Emans, S.J., & Grace, E.M. (1982). 11(3), 274-293. Factors associated with compliance to oral contraceptive use in an adolescent population. Journal of Romer, D., Stanton, B., Galbraith, J., Feigelman, Adolescent Health Care, 3(2), 120-123. S., Black, M.M., & Li, X. (1999). Parental influence on adolescent sexual behavior in high-poverty Scott, J.W. (1993). African American daughter- settings. Archives of Pediatrics & Adolescent mother relations and teenage pregnancy: Two Medicine, 153(10), 1055-1062. faces of premarital teenage pregnancy. Western Journal of Black Studies, 17(2), 73-81. Roosa, M.W., Tein, J.Y., Reinholtz, C., & Angelini, P.J. (1997). The relationship of child- Scott-Jones, D., & Turner, S.L. (1988). Sex edu- hood sexual abuse to teenage pregnancy. Journal cation, contraceptive and reproductive knowledge, of Marriage and the Family, 59(1), 119-130. and contraceptive use among Black adolescent females. Journal of Adolescent Research, 3(2), 171- Rosenbaum, E., & Kandel, D.B. (1990). Early 187. onset of adolescent sexual behavior and drug involvement. Journal of Marriage and the Family, Seidman, S.N., Mosher, W.D., & Aral, S.O. 52(3), 783-798. (1994). Predictors of high-risk behavior in unmarried American women: Adolescent environment as Rosenthal, S.L., Burklow, KA., Lewis, L.M., risk factor. Journal of Adolescent Health, 15(2), Succop, P.A., & Biro, F.M. (1997). Heterosexual 126-132.

Emerging Answers:.ResearchFindings on Programs to Reduce Teen Pregnancy 69 so Serbin, L.A., Peters, P.L., McAffer, F.J., & awareness among adolescent males. Family Schwartzman, A.E. (1991). Childhood aggres- Planning Perspectives, 21(4), 152-158. sion and withdrawal as predictors of adolescent pregnancy, early parenthood, and environmental Spingarn, R.W., & DuRant, R.H. (1996). Male risk for the next generation. Canadian Journal adolescents involved in pregnancy: Associated

of Behavioral Science, 23(3), 318-331. . health risk and problem behaviors. Pediatrics, 98(2, Part I), 262-268. Shah, F., & Zelnik, M. (1981). Parent and peer influence on sexual behavior, contraceptive use, St. Lawrence, J.S. (1993). African-American and pregnancy experience of young women. adolescents' knowledge, health-related attitudes, Journal of Marriage and the Family, 43, 339-348. sexual behavior, and contraceptive decisions: Implications for the prevention of adolescent Shea, J.A., Hercog-Baron, R., & Furstenberg, HW infection. Journal of Consulting and F.F. (1984). Factors associated with adolescent Clinical Psychology, 61(1), 104-112. use of family planning clinics. American Journal

of Public Health, 74(11), 1227-1230. . St. Lawrence, J.S., Brasfield, T.L., Jefferson, K.W., Allyene, E., & Shirley, A. (1994). Social Shrier, L.A., Emans, S.J., Woods, E.R., & support as a factor in African-American adoles- DuRant, R.H. (1997). The association of sexual cents' sexual risk behavior. Journal of Adolescent risk behaviors and problem drug behaviors in Research, 9(3), 292-310. high school students. Journal of Adolescent Health, 20(5), 377-383. Stanton, B., Li, X., Black, M., Ricardo, I., Galbraith, J., Kaljee, L., & Feigelman, S. (1994). Sieving, R., Resnick, M.D., Bearinger, L., Sexual practices and intentions among pre-ado- Remafedi, G., Taylor, B.A., & Harmon, B. lescent and early adolescent low-income, urban (1997). Cognitive and behavioral predictors of African-Americans. Pediatrics, 93(6), 966-973. sexually transmitted disease risk behavior among sexually active adolescents. Archives of Pediatrics Stock, J.L., Bell, M.A., Boyer, D.K., & Connell, and Adolescent Medicine, 151(3), 243- 251. F.A. (1997). Adolescent pregnancy and sexual risk-taking among sexually abused girls. Family Silverman, J.G., & Amaro, H. (2000). The rela- Planning Perspectives, 29(5), 200-203, 227. tionship between sexual abuse and sexual risk 'among high school students: Findings from the Stouthamer-Loeber, M., & Wei, B.S. (1998). 1997 Massachusetts Youth Risk Behavior Survey. The precursors of young fatherhood and its Maternal and Child Health Journal, 4(2), effect on delinquency of teenage males. Journal 125-134. of Adolescent Health, 22(1), 56-65.

Singh, S. (1986). Adolescent pregnancy in the Thomson, E. (1982). Socialization for sexual United States: An interstate analysis. Family and contraceptive behavior: Moral absolutes Planning Perspectives, 18(5), 210-220. verses relative consequences. Youth and Society, 14(1), 103-128. Small, S.A., & Luster, T. (1994). Adolescent sexual activity: An ecological risk-factor Thornberry, T.P., Smith, C.A., & Howard, G.J. approach. Journal of Marriage and the Family, (1997). Risk factors for teenage fatherhood. 56(1), 181-192. Journal of Marriage and the Family, 59(3), 505- 522. Smith, C.A. (1996). The link between childhood maltreatment and teenage pregnancy. Social Thornberry, T.P., Wei, E.H., Stouthamer- Work Research, 20(3), 131-141. Loeber, M., & Van Dyke, J. (undated). Teenage fatherhood and delinquent behavior. Washington, Smith, C.A. (1997). Factors associated with early DC: U.S. Department of Justice, Office of sexual activity among urban adolescents. Social Juvenile Justice and Delinquency Prevention. Work, 42(4), 334-346. Thornton, A. (1990). The courtship process and Sonenstein, F.L., Pleck, J.H., & Ku, L.C. adolescent sexuality. Journal of Family Issues, (1989). Sexual activity, condom use, and AIDS 11(3), 239-273.

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 70 Thornton, A., & Camburn, D. (1987). The influ- Warren, C.W., Santelli, J.S., Everett, S.A., Kann, ence of the family on premarital sexual attitudes L., Collins, J.L., Cassell, C., Morris, L., & and behavior. Demography, 24(3), 323-340. Kolbe, L.J. (1998). Sexual behavior among U.S. high school students, 1990-1995. Family Thornton, A., & Camburn, D. (1989). Religious Planning Perspectives, 30(4), 170-172, 200. participation and adolescent sexual behavior and attitudes. Journal of Marriage and the Family, Weber, F.T., Gearing, J., Davis, A., & Conlon, 51(3), 641-653. M. (1992). Prepubertal initiation of sexual experiences and older first partner predict promiscu- Turner, R.A., Irwin, C.E., Jr., Tschann, & ous sexual behavior of delinquent adolescent Millstein, S.G. (1993). Autonomy, relatedness, males Unrecognized child abuse? Journal of and the initiation of health risk behaviors in early Adolescent Health, 13(7), 600-605. adolescence. Health Psychology, 12(3), 200-208. Weinbender, M.L., & Rossignol, A.M. (1996). Udry, J.R. (1988). Biological predispositions and Lifestyle and risk of premature sexual activity in a social control in adolescent sexual behavior. high school population of Seventh Day American Sociological Review, 53, 709-722. Adventists: Valuegenesis 1989. Adolescence, 31(122), 265-281. Udry, J.R., & Billy, J.O.G. (1987). Initiation of coitus in early adolescence. American Weinman, M.L., Smith, P.B., & Mumford, D.M. Sociological Review, 52, 841-855. (1992). A comparison between a 1986 and 1989 cohort of inner-city adolescent females on Udry, J.R., & Cliquet, R.L. (1982). A cross-cul- knowledge, beliefs, and risk factors for AIDS. tural examination of the relationship between Journal of Adolescence, 15(1), 19-28. ages at menarche, marriage, and first birth. Demography, 19, 53-56. Weinstein, M., & Thornton, A. (1989). Mother- child relations and adolescent sexual attitudes Underwood, M.K., Kupersmidt, & Coie, and behaviors. Demography, 26(4), 563-577. J.D. (1996). Childhood peer sociometric status and aggression as predictors of adolescent child- Weisman, C.S., Plichta, S., Nathanson, C.A., bearing. Journal of Research on Adolescence, 6(2), Chase, G.A., Ensminger, M.E., & Robinson, 201-223. J.C. (1991). Adolescent women's contraceptive decision making. Journal of Health and Social Upchurch, D.M., Levy-Storms, L., Sucoff, C.A., Behavior, 32(2), 130-144. & Aneshensel, C.S. (1998). Gender and ethnic differences in the timing of first sexual intercourse. Whitaker, D.J., & Miller, KS. (2000). Parent- Family Planning Perspectives, 30(3), 121-127. adolescent discussions about sex and condoms: Impact of peer influences of sexual risk behavior. Valois, R.F., Oeltmann, J.E., Waller, J., & Journal of Adolescent Research, 15(2), 251-273. Hussey, J.R. (1999). Relationship between number of sexual intercourse partners and selected VVhitaker, D.J., Miller, K.S., May, D.C., & health risk behaviors among public high school Levin, M.L. (1999). Teenage partners' commu- adolescents. Journal of Adolescent Health, 25(5), nication about sexual risk and condom use: The 328-335. importance of parent-teenager discussions. Family Planning Perspectives, 31(3), 117-121. Ventura, S.J., Curtin, S.C., & Mathews, T.J. (1998). Teenage births in the United States: Whitbeck, L., Hoyt, D., Miller, M., & Kao, M. National and state trends, 1990-1996. National (1992). Parental support, depressed affect, and Vital Statistics System. Hyattsville, MD: National sexual experiences among adolescents. Youth Center for Health Statistics. and Society, 24(2), 166-177.

Vermont Department of Public Health. (1997, Whitbeck, L., Simons, R.M., & Kao, M. (1994). April 9). Gay, lesbian, and bisexual.students at The effects of divorced mothers' dating behaviors risk. In Nuggets: News, notes and findings from and sexual attitudes on the sexual attitudes and the VDH Office of Alcohol and Drug Abuse behaviors of their adolescent children. Journal of Programs. Montpelier, VT: Author. Marriage and the Family, 56(3), 615-621.

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 71 82 White, H.R., & Johnson, V. (1988). Risk-taking Zabin, L.S. (1994). Adolescent sexual behavior as a predictor of adolescent sexual activity and use and childbearing. Women's Health Issues, 4(2), of contraception. Journal of Adolescent Research, 73-75. 3(3-4), 317-331. Zabin, L.S., Astone, N.M., & Emerson, M,R. Whitley, B.E., & Schofield, J.W. (1985-1986). A (1993). Do adolescents want babies? The rela- meta-analysis of research on adolescent contra- tionship between attitudes and behavior. Journal ceptive use. Population and Environment, 8(3-4), of Research on Adolescence, 3(1), 67-86. 173-203. Zabin, L.S., Hardy, J.B., Smith, E.A., & Hirsch, Widmer, E.D. (1997). Influence of older siblings M. (1986). Substance use and its relation to sex- on initiation of sexual intercourse. Journal of ual activity among inner-city adolescents. Marriage and the Family, 59(4), 928-938. Journal of Adolescent Health Care, 7, 320-331.

Wilson, M.D., Kastrinakis, M., D'Angelo, L.J., & Zelnik, M., Kantner, J.F., & Ford, K. (1981). Getson, P. (1994). Attitudes, knowledge, and Sex and pregnancy in adolescence. Beverly Hills, behavior regarding condom use in urban black CA: Sage Publications. adolescent males. Adolescence,. 29(113), 13-26. Zelnik, M., & Kim, Y. (1982). Sex education Wu, L.L. (1994). Effects of family structure and and its association with teenage sexual activity, income on the risk of a premarital birth (Center pregnancy, and contraceptive use. Family for Demography and Ecology Working Paper No. Planning Perspectives, 14(3), 117-118, 123-126. 94-07). Madison, WI: University of Wisconsin. Zill, N., Nord, C.W., & Loomis, L.S. (1995). Wu, L.L., & Martinson, B.B.-(1993). Family Adolescent time use, risky behavior, and outcomes: structure and the risk of a premarital birth. An analysis of national data. Rockville, MD: American Sociological Review, 58, 210-232. Westat, Inc.

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 72 8 3 Chalpter 3 Assessing the Evidence: Factors Affecting the Strength of Research Results --Ihe research presented in Chapters 1 All the studies included in this review and 2 on adolescent sexual activity have used experimental or quasi-experimental and the antecedents of sexual behav- designs to measure the impact of specific ior can help people design programs to programs. By definition, experimental reduce teen pregnancy by helping define designs randomly assign study participants to teens at high risk and by suggesting the risk intervention and control groups and then and protective factors that programs should compare the two groups, while quasi-experi- strive to change. However, only good evalua- mental designs do not randomly assign study tion of programs themselves can provide evi- participants to either group but do compare dence for whether or not those programs the intervention group with a comparison actually have the desired effect on behavior. group of purportedly similar youth. Both types of studies include youth in both pro- Strengths and Limitations of gram and either control or comparison Evaluation Methods groups, use some method of making the program and comparison groups similar This chapter discusses some of the crite- before program implementation, track indi- ria for judging the strength of evidence from vidual youths over time, and then link their evaluation studies. Many studies have exam- pretest and posttest scores. ined the effects of programs on sexual behavior and pregnancy, but both the quality of Studies employing experimental or the research methods employed and the quasi-experimental designs are very different strength of the resulting evidence have varied from those based on national surveys of enormously. Because of that variation, the youth or on census tract-, county-, or state- quality of the research methods and the level data. Studies based on national surveys strength of the evidence should be thought- generally depend upon the respondents' fully considered when assessing the results. recall of whether they ever participated in a Results that come from strong, well-designed particular type of program. They also have studies are generally taken more seriously very poor measures of the quality of any pro- than results from poorly designed ones. grams in which the participants were

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 73 84 involved and may have considerable difficulty I. Sampling of Programs. controlling statistically for other factors that might produce spurious statistical relation- In order to be confident that the find- ships or obscure actual relationships. ings from a group of somewhat similar evalu- Similarly, studies based on data aggregated at ated programs apply to other programs of the census tract, county, or state levels also the same type, the evaluated programs must have poor measures of participation in vari- be a representative sample of the larger ous programs, have difficulty controlling for group of programs. Otherwise, conclusions spurious relationships, and sometimes have based on the sample cannot be generalized very small sample sizes (e.g., studies of the to the larger group of programs. For exam- 50 states). Thus, neither of these two groups ple, evaluations of several sex education programs employing particular strategies for of studies is reviewed in this paper. changing behavior probably do not help us Studies using experimental or quasi- understand sex education programs that take experimental designs have at least three very different approaches. important strengths not typically found in There are so many different approaches studies using national survey data or aggre- to reducing adolescent sexual risk-taking that gated data. First, they evaluate specific pro- even the many evaluations in this review can- grams with known program characteristics. not possibly represent all of the various types Second, they can clearly distinguish between of programs adequately. There are undoubt- who did and did not receive the interven- edly some approaches to reducing teen preg- tion. Third, such studies typically control for nancy and STDs that have never been other factors that may have affected the evaluated, and there are other types of pro- results. Thus, this group of studies has grams for which there are only a few evalua- greater potential ability to assess the actual tions (e.g., abstinence-only programs or causal impact of specific programs than stud- school condom-availability programs). The ies based upon national surveys or aggre- key point here is that one must be very care- gated data. ful when drawing inferences about types of programs with only a few existing studies. However, even those studies employing The problem with generalizing the results experimental or quasi-experimental designs from a small group of studies is heightened vary greatly in their adherence to the follow- by the fact that some programs may work ing important principles of research method- only in some settings with some youth but ology. Some incorporate most of these not in others. Additional studies are often principles and, therefore, yield high quality needed to learn about the effects of pro- results; those that do not are more likely to grams across many different types of youth. produce biased results. This chapter identifies ten principles of good research methods, Furthermore, the programs that are estimates the extent to which each principle evaluated, especially those in large well- is met by the research studies reviewed in funded projects, are more likely than other this volume, and estimates how violation of programs to 11..ve well-trained staff, to have each principle would bias the overall results been well and carefully run, and to have a of these studies. For specific information real focus on specific outcomes. Thus, these about how particular studies in this review evaluations are likely to overstate the positive adhere to these principles, see the detailed results that might emerge when the same study descriptions in Tables 4.1-4.14 in programs are adopted and replicated in other Chapter 4. settings by other organizations. This bias

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 74 83 may be offset slightly by the fact that some Of the studies in Tables 4.1-4.14 mea- programs may be evaluated prematurely suring impact upon behavior, about one- (before formative evaluation has led to pro- third employed random assignment while the gram improvements) and by the fact that rest used a quasi-experimental design with- some programs may not be implemented out random assignment. This means that with fidelity when they are evaluated. self-selection biases or other kinds of biases may affect the results of two-thirds of the studies. These biases could either obscure 2.Random Assignment. actual positive program effects or erro- The evaluation design that demon- neously suggest positive program effects strates causality most conclusively is one when they did not actually occur. On the using random assignment of individual other hand, in the studies reviewed in this youth, groups of youth (e.g., classrooms), or volume, possible biases produced by lack of entire schools or communities. Random random assignment should not be great for assignment is especially important because two reasons: (1) the studies with the weakest adolescents who voluntarily participate in quasi-experimental designs were excluded from this review, and (2) most important interventions are typically different in either conclusions are confirmed by one or more subtle or not-so-subtle ways from those who studies with experimental designs. do not. These differences are what is meant by "selection bias." For example, youth who It should also be noted that the control voluntarily participate in an abstinence-only groups in nearly all these studies typically program may be more inclined to be absti- received other instruction or services, rather nent than their same-aged peers, and youth than nothing at all. For example, in some who visit a health center for contraception studies, the control groups received the stan- may already be more motivated to use con-' dard sex education course for that institu- traception than other sexually experienced tion, instead of the new or experimental young people. Therefore, youth who do not intervention. Or, youth who did not receive voluntarily participate in an abstinence-only subsidized family planning services from .a programs may be a biased comparison group clinic received such services from a private for those youth who do volunteer, and sexu- physician or obtained non-prescription con- ally active youth who don't attend a family traceptives from a drug store. Thus, these planning clinic may be a biased comparison studies typically measure only the incremen- group for those who do attend. tal effect of the intervention and not the cumulative effect of that intervention plus Furthermore, youth who receive per- whatever sexuality education or reproductive mission from their parents to participate in health services the youth might have "programs are likely to be different in many received. If the cumulative effect of all edu- ways from those who do not receive such cation and services were measured, results would undoubtedly be stronger. permission. Even when participation is not determined primarily by either the teens or 3. Sample Size. their parents, selection biases occur. For example, school-wide and community-wide Rigorous studies include a sufficiently programs are disproportionately likely to be large sample size. The sample size needed implemented in schools or communities varies according to several factors, such as where there is the greatest need and where the variability in the outcome measure, the there is greater sexual risk-taking. magnitude of the effect that needs to be

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 75 8 6 detected, and the chosen level of statistical desired length of the follow-up varies significance (a measure of how confident one depending on different characteristics of the can be in the results). However, a growing study (e.g., the outcomes being measured, body of experience suggests that a combined the length of time before they become sample of at least 500 or even 1,000 is com- apparent, and the length of time they are monly needed. Much larger sample sizes may likely to endure). Experience in this field be needed if (1) groups of youth (e.g., class- suggests that follow-up for at least one year rooms or entire schools) rather than individ- is important, but follow-up of less duration ual youth are assigned to intervention or can provide information about short-term comparison groups, (2) the intervention has effects. a modest effect, (3) some of the sample's subgroups need to be examined separately There are at least two important rea- for instance, those who are sexually experi- sons why long-term follow-up is needed. enced versus those who are not or males ver- First, it can detect effects that are not appar- sus females, or (4) highly skewed variables ent in short periods of time. This is particu- with very asymmetrical distributions (e.g., larly true for studies measuring the impact of pregnancy or birth rates) are being mea- programs trying to delay the initiation of sured. intercourse. If young people are tracked for much less than 12 to 18 months, then too Large samples are needed for at least few youth in Most control groups will initiate three reasons. Most important, if the sample intercourse to allow for the treatment group size is too small, then a program may pro- to emerge as doing measurably better than duce programmatically significant results, but the control group that is, the program those results may not be statistically signifi- may actually delay intercourse, but that delay cant because of insufficient statistical power. would not be observed because the follow- Second, when sample sizes are too small and up period was too short. Second, some pro- when researchers search for significant find- gram effects (e.g., increase in participants' ings, then anomalous results are more likely knowledge) are likely to diminish substan- to occur. Third, when sample sizes are too tially with time, and, thus, we cannot assume small, the magnitude of effect is unknown that positive short-term results will endure. because much of the apparent effect could And, of course, if short-term results do not have been caused by chance. endure, then they may have only a very limited impact upon the goals of reducing ado- Many published evaluations of pro- lescent pregnancy and STD rates over a grams to reduce adolescent risk-taking have longer period of time. been based on small sample sizes. Thus, some of these programs may have had pro- Relatively few studies have measured grammatically significant results that were long-term effects, although their number is not found to be statistically significant, some increasing. A few published studies attempt- may have produced anomalous results that ing to measure impact upon initiation of sex are reported as statistically significant pro- clearly did not measure behavior for a long gram effects, and some may have had signifi- enough period of time and, therefore, may cant effects whose magnitude was unknown. have incorrectly concluded that programs were not effective. For this reason, these 4. Long-Term Follow-Up. studies are not included in this review, even though they were published. Many studies Studies need to conduct long-term fol- did meet the criteria for inclusion in this vol- low-up with program participants. The ume, but measured only relatively short-term

THE NATIONAL CAMPMGN TO PREVENT TEEN PREGNANCY 76 87 effects on behavior. These studies are some- Studies can also examine the program what informative, although they tell us little impact on birth rates. However, if a pro- about the programs' long-term impact on gram's goal is to reduce pregnancy, measur- rates of teen pregnancy. ing the impact on birth rates as a proxy for pregnancy rates does not take into account 5. Measurement of Pregnancy Versus the program's impact on abortion rates. As Sexual and Contraceptive Behaviors. such, results may be misleading. For example, if a program's only effect is to decrease Given the programmatic goal of reduc- the rate of abortion, the pregnancy rate ing pregnancy, the single best outcome mea- would remain the same while the birth rate sure for a program would logically be a would increase. If only birth rates were measure of pregnancy rates. Several studies examined, one could reach the erroneous have, in fact, measured the impact of pro- conclusion that the pregnancy rate had grams on pregnancy rates. However, there increased due to the program when, in fact, are at least two methodological problems it had not. A few studies have employed very with using pregnancy rates as an outcome large sample sizes and have properly analyzed self-reports of pregnancies or births or have measure. First, regardless of how they are estimated birth rates from vital records. A measured, pregnancy rates are a very insensi- few other studies have failed to recognize the tive measure of program impact. For exam- issues of statistical power and have con- ple, if a program reduces the pregnancy rate ducted misleading analyses of pregnancy or per year from 100 per 1,000 to 80 per birth rates. 1,000, that is a 20 percent reduction, which would suggest a very successful program. Because of the limitations of the preg- However, that decrease represents a differ- nancy and birth rate measures, other mea- ence of only 20 pregnancies per 1,000 (or 2 sures of behavioral change are often used. percentage points), and a very large sample Because the single most important size (e.g., more than 6,000) would be antecedent of pregnancy is the frequency of required to find that change to be statistically unprotected sexual activity, this measure and significant. To even a greater extent, this its constituent components are often used to same problem applies to birth rates. Second, measure the impact of programs. Frequently pregnancy rates are often hard to determine measured components include age of initia- and may, therefore, be an inaccurate mea- tion of intercourse, frequency of intercourse, sure. If pregnancy rates are estimated from use of contraception (at first sex, at last sex, self-reported data, pregnancies are likely to or consistency of use), and actual frequency be under-reported (Fu et al., 1998; Jones & of sex without contraception. Typically, these behaviors can only be estimated from data Forrest, 1992). This is especially true for that individuals report about themselves. males who do not know of pregnancies they Although some under- and over-reporting of have caused and for females who have ended these behaviors undoubtedly exists, these their pregnancies by abortion and are reluc- data are generally believed to be reasonably tant to report it. Moreover, it is commonly reliable and valid. Furthermore, when under- impossible to estimate pregnancy rates for or over-reporting occurs, these biases are study participants from vital records. While believed to cancel each other out in good birth rates can sometimes be determined experimental designs (Sonenstein, 1996). from birth certificates, public records rarely yield abortion rates (and therefore pregnancy On the other han.d, under-reporting of rates). sexual risk-taking in both intervention and

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 77 0 0 U comparison groups may not always cancel hypotheses, regardless of whether they are each other out. The resulting bias is most positive or negative. likely to occur when participants become close to intervention staff, when intervention Many studies in this field have failed staff administer the surveys, and when large to conduct proper statistical analyses. For incentives are provided to study participants example, some apparently did not specify to complete the surveys. To the extent that hypotheses before testing them, a few appear this bias occurs, under-reporting would to have searched for and reported only desir- probably be higher in the intervention able results without reporting negative groups and favor more desirable program results, some failed to adjust for group clus- results. tering, and most failed to account for multiple tests of significance. All of these statistical 6. Measurement of Behavior. problems would tend to bias studies in favor of more desirable results. This may be quite Given the goal of reducing adolescent a large bias. pregnancy, studies should measure actual sexual and contraceptive behavior or preg- 8. Publication of Results. nancy, rather than proximal antecedents of those behaviors such as intent to have sex or If evaluation studies are conducted well, intent to use contraception. This is particu- researchers sliould attempt to publish all the larly important because adolescents' sexual important results, regardless of whether beliefs, attitudes, and even intentions are those results are positive or negative. only moderately (not highly) related to their Otherwise, the published literature may be actual behavior, which makes them weak very biased. Consider the following example. proxies for actual behaviors. For example, Given that the federal government, many many adolescents intend to use contracep- states, and several foundations have funded tion every time they have sex, but many do the implementation and evaluation of preg- not actually do so. nancy prevention, HIV prevention, and youth development programs in many com- Many studies have examined program munities throughout the country, many eval- impact on proximal antecedents of behavior uations are underway at any point in time. rather than on behavior itself. Although such For the sake of argument, let's say that as studies can inform the development of pro- many as 100 studies with experimental or grams, they are not included in this review. quasi-experimental designs are completed during any five-year period of time. If each 7. Proper Statistical Analyses. of these 100 studies examines 10 different behavioral outcomes, then as many as 1,000 Studies should conduct proper statisti- behavioral outcomes could be examined cal analyses of the collected data. This across the studies. If these outcomes are includes, for example, stating precisely the examined for two different sub-groups (e.g., hypotheses to be tested before conducting males and females), then 2,000 outcomes the tests of significance, conducting the will be examined. By chance alone, 5 percent proper statistical tests, correcting for cluster- (100) of these outcomes will be statistically ing if respondents are randomly assigned by significant at the .05 level of significance, group instead of individually, accounting for and, of these, again by chance alone, half multiple tests of significance, especially when (50, or 10 per year) will be in the desired numerous tests are conducted, and then direction. If only these studies are published, reporting all the results of the tests of then the literature will obviously be very

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 78 89 biased in favor of the programs even 10. Independent External Evaluators. though those positive results may only be the Evaluations of impact should ideally be result of chance. conducted by independent third parties who In sum, the large number of studies do not have a direct stake in the results of being conducted at any point in time, the the evaluation. People who evaluate their own programs are not dishonest, but they large number of outcomes examined, chance, sometimes strongly believe in the efficacy of and the natural tendency to publish only their programs and spend greater effort positive results may markedly bias published searching for elusive positive results rather conclusions about program effects in a posi- than elusive negative results. Thus, their tive direction. To partially counter this bias, reported results may be biased. This caution- greater emphasis should be given to studies ary perspective is supported by a respected that (1) are large, well-funded, and well- literature about "experimenter effects" that designed, (2) have advisory boards that underscores the need for independent (or include experts in the field, and (3) are blinded) evaluations. Many of the studies widely known while in progress, because reviewed here were conducted by people such studies are more likely to publish results who helped design the programs evaluated regardless of whether they are positive or and undoubtedly had personal interests in negative. the outcomes. This may have biased the results in favor of more positive results.

9. Replication of Studies. Conducting studies that meet even A program achieving positive results in most of these principles requires considerable one study should be replicated by others time and resources. In fact, doing so may take five or more years: one or more years to elsewhere and re-evaluated in order to learn design the program, train staff to implement more about the true impact of the model. it, obtain the approval of schools or other This is important because many factors organizations to put it in place, and recruit unique to the first evaluation may be respon- youth; one or more years to actually run the sible for producing the positive results. For program with a sufficient number of youth; instance, the teacher(s) may have been par- two years to collect long-term follow-up ticularly charismatic, the program may have data; and one or more years to clean the met the particular needs only of that specific data, conduct statistical analyses, and write group of teens, the program might have reports. In addition, conducting studies with been reinforced by other programs in the these methodological principles requires very community, or the results might have knowledgeable and competent staff and occurred by chance. When tried in another sometimes hundreds of thousands, if not setting without these factors present, the millions, of dollars to cover all costs. results may not be so pdsitive. So far, a few Consequently, this discussion of programs have been replicated and evaluated methodological principles is not meant to in other sites. The earlier positive findings suggest that it is easy, possible, or even desir- were confirmed in some cases and were not able to incorporate every principle into each confirmed in others. Whenever possible, study. For some studies, it may be ethically effective programs shOuld be replicated and unacceptable or administratively impossible studied before being adopted in statewide or to randomly assign participants to program nationwide initiatives. and control groups or to have large sample

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 79; sizes, for instance. In other studies, it may be cations for this review, for selecting programs ethically and theoretically possible to incor- to be replicated, and for future research: porate these principles, but insufficient resources may make it impossible to do so. a The limitations of the evaluations, cou- Moreover, the fields of pregnancy and STD pled with possible biases both big and prevention may advance more rapidly if small that operate in both directions (but many innovative programs are developed and probably disproportionately favor desired evaluated less rigorously and less expensively results), reduce our ability to make at the outset, and then only those programs definitive statements about either specific with encouraging results are subsequently programs or types of programs. Generali- evaluated more rigorously. After all, with zations should be made cautiously. limited resources, it is not possible to evalu- The quality of the evaluation design and ate every creative program rigorously at the of the resulting research evidence should outset. be given considerable weight when assessing reported program results and Despite the very real limitations of the when selecting programs for replication. research in this field, we should be very encouraged by the incredible progress that aDespite great progress, there still exists a has been made recently in research methods real need for more evaluations with rig- employed. For example, twenty years ago orous designs, proper statistical analyses, there were only a few studies of sex educa- and accurate reporting of results. Many tion programs. Of those studies, only a few examples of rigorous research and used experimental designs with random progress in research methods over the assignment; most had small sample sizes; decades demonstrate that such studies most measured only short-term effects; few can be conducted. Public and private measured actual behavior; some used inap- funders need to fully recognize the time propriate statistical tests; some reported only and resources necessary to conduct these positive results; and none replicated previous rigorous studies. studies. By contrast, many studies now use a Although few studies meet all ten random assignment, have large sample sizes, research criteria discussed above, much measure long-term effects of a year or more, can be learned from the many less-than- measure a variety of sexual and contraceptive perfect studies that have been conducted, behaviors, use proper statistical tests, and particularly if the limitations of each report both positive and negative results study are kept in mind and the patterns (c.f., Kirby, 1999). A few have even repli- of results across studies are observed. cated previous studies and reported similar The review of studies in Chapter 4 takes positive results. Thus, we should be very into account these limitations and pat- encouraged by this progress and can now terns; a study-by-study summary of the have much greater confidence in the characteristics of each evaluation can be reported results of research findings. found in the chapter's tables. Implications of Methodological oWhen positive results are achieved in an Strengths and Limitations evaluation, the program should be put in place and evaluated with different staff These ten criteria and the actual and different research teams and with methodological strengths and limitations of different types of youth in different loca- the studies in this field have important impli- tions all in an effort to confirm the

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 80 91 original findings. When the positive semination of that program. Even then, results of the first study are found again attention should be given to the particu- by one or more subsequent studies, then lar groups and conditions in which the it is wise to consider the widespread dis- program is found to be effective.

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 81 9'2 References Kirby, D. (1999). Reflections on two decades of research on teen sexual behavior and pregnancy. Fu, H., Darroch, J.E., Henshaw, S.K., & Kolb, Journal of School Health, 69(3), 89-94. E. (1998). Measuring the extent of abortion underreporting in the 1995 National Survey of Sonenstein, F.L. (1996). Measuring sexual risk Family Growth. Family Planning Perspectives, behaviors. Paper presented a meeting of the 30(3), 128-133, 138. American Enterprise Institute, Washington, DC.

Jones, E.J., & Forrest, J.D. (1992). Underreporting of abortion in surveys of U.S. women: 1976 to 1988. Demography, 29(1), 113-126.

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 82 93 Chaipter 4 Emerging Answers: The Behavioral Impact of Programs to Reduce Adolescent Sexual Risk-Taking

aving set the stage in the previous Criteria for Selection of Studies chapters by describing the scope of for This Review L the problem of teenpregnancy, the antecedents of teen sexual and contraceptive Except where noted, this review focuses behavior, and the criteria for assessing evalua- only on those studies meeting certain metho- tion research, this chapter turns to the main dological criteria.' The six most important business of this report: reviewing and sum- criteria are presented below. A fuller explana- , marizing evaluations of programs to prevent tion of these criteria,- as well as of other less teen pregnancy and sexual risk-taking. This important criteria are presented at the end of chapter first explains the criteria that studies this chapter (p. 115). The six primary criteria had to meet to be included in the review. It are: then presents a typology for grouping the various prevention programs that have been EI The study was conducted in 1980 or later. evaluated. And, finally, the findings of the [3 The study was conducted in the United evaluations themselves are presented. General States or Canada. conclusions about particular types of pro- o grams can be found at the end of each sec- The study targeted adolescents of middle school or high school age (roughly tion; detailed descriptions of each study can 12-18). be found in the tables. Overall conclusions about both the findings and future directions Li The study used an appropriate experi- for research are in Chapter 5. mental or quasi-experimental design.

1 The first edition of the National Campaign's research review, No Easy Answers: Research Findings on Programs to Reduce Teen Pregnancy (1997), used publication in a peer-reviewed journal as the principal criterion for inclusion of a study. However, this volume uses methodological criteria for inclusion for two reasons: (1) some studies have employed rigorous research methods, but for a variety of reasons were never published in peer-reviewed journals, and (2) a few studies published in peer-reviewed journals employed very weak methods and provided misleading results that could bias interpretation of findings. Using methodological criteria for inclusion, instead of a publication criterion, improved both the number and quality of studies reviewed.

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 83 CI A 0The sample size was at least 100 in the STDs, particularly HIV. Some target adoles- combined treatment and control group. cents, while others target parents of adoles- This minimum number applies to the cents. Some focus on sexually inexperienced actual statistical analyses measuring youth, others on the sexually experienced. impact on behavior. Some use structured curricula with groups of The study measured impact on sexual or youth, others provide one-on-one instruction contraceptive behavior or pregnancy or or counseling, and still others use commu- childbearing. nity-wide media approaches. Some programs are implemented by adults, others are led by A Typology of Programs that the peers of adolescents. Given all these May Reduce Teen Pregnancy important differences and dimensions, it is and STDs very difficult to create any single typology to organize programs to prevent teen pregnancy. Innumerable programs have been developed to reduce adolescent pregnancy and However, one useful way to categorize STDs, including HIV. They are found in prevention programs is,by the behavioral schools; in health, family planning, and STD antecedents that they target (for more on clinics; and in community organizations antecedents, see Chapter 2). Accordingly, working with youth (including faith commu- this review first divides programs and their nities). Some are designed to support and encourage abstinence; others to improve respective evaluations into those that focus knowledge, attitudes, and skills about contra- primarily on sexual antecedents of adolescent ception; some to improve access to contra- sexual and contraceptive behavior, those that ception; and still others to improve focus primarily upon non-sexual antecedents, education, life skills, and life opportunities and those that address both. Within these more generally. Some focus on preventing three broad categories, programs are divided first pregnancies; some target subsequent into seven major groups and several sub- pregnancies; and others concentrate on groups (see box below).

Programs.Focusing Primarily on Sexual Antecedents Curriculum-based sexuality education programs in any setting o Abstinence-only education programs o Sexuality and HIV education programs Sex and HIV education programs for parents and their families Clinic or school-based programs designed to provide reproductive health care or to improve access to condoms or other contraceptives o Family planning clinics and services o School-based and school-linked clinics o Protocols for clinic appointments and o School condom-availability programs supportive activities o Other clinic characteristics and programs Community-wide pregnancy or HIV prevention initiatives with multiple components Programs Focusing Primarily on Non-Sexual Antecedents Early childhood programs Youth development programs for adolescents o Service learning programs o Other youth development programs o Vocational education and employment programs Programs Focusing on BOth Sexual and Non-Sexual Antecedents Multi-component programs with both sexuality and youth development components

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 84 9 5 Programs that Focus Primarily (appropriately called HIV education pro- on Sexual Antecedents grams).

Curriculum-Based Sexuality In regard to their emphasis on only abstinence versus abstinence and the use of Education Programs protection against pregancy and STDs, pro- Most adolescents in this country know grams actually fall along a continuum and no a lot about the risks of unprotected sexual longer clump into distinct groups. Whereas intercourse and about methods of reducing all abstinence-only programs emphasize that those risks. For example, nearly all youth abstinence from sexual intercourse is the only know that unprotected sexual intercourse truly healthy and safe choice for young peo- can lead to pregnancy or STDs, and most ple, some talk only about abstinence, some know that condoms provide some protection mention condoms and contraception but against pregnancy and STDs and can be only emphasize their fallibility, and some obtained at stores. They learn this and other offer medically accurate information about information from a variety of sources, includ- the benefits and limitations of condoms and ing school sex and HIV education programs, contraceptives while still emphasizing absti- the media, their parents and other adults, nence. This last group of abstinence-only their friends, and others. Presumably, this programs is harder to distinguish from sexu- information does reduce the amount of ality education and HIV education pro- unprotected sex among teenagers. For exam- grams, most of which also emphasize abstinence as the safest choice for young ple, after knowledge about HIV was dissemi- people but encourage use of condoms and nated throughout the country in the 1980s, contraceptives if youth do have sex, and a adolescents markedly increased their use of few of which especially those for high- condoms (Sonenstein, Pleck, & Ku, 1989). risk, sexually active youth give primary However, there remain the following impor- emphasis to consistent condom use. tant questions: Given some adolescent knowledge about these topics, does addi- Abstinence-Only Programs tional instruction about these topics or other aspects of sexuality affect teenagers' sexual Despite their common emphasis on behavior in positive or negative ways? If yes, abstinence, abstinence-only programs are what are the characteristics of programs that very diverse in many ways. For example, have positive effects? some faith-based programs may begin with a prayer for God's guidance and teach that it is This section will examine the effects of immoral to have sex before marriage, while programs that stress abstinence as the only most programs especially with the acceptable choice for preventing pregnancy increased availability of federal funding for (often called abstinence-only programs) and abstinence programs since 1996 are programs that discuss both abstinence and entirely secular. Some programs emphasize methods of protection against pregnancy and that abstinence until marriage is the only STDs (broader programs often called sexual- truly safe choice, while others encourage ity education or abstinence-plus programs). youth to postpone sex until an unspecified This latter group can be further divided into later age. Some abstinence-only programs are those that address both pregnancy and entirely curriculum-based, while others STDs/HIV (sometimes called sexuality edu- include a wide variety of youth development cation programs) and those that focus pri- activities. Some last only one or two sessions, marily on STD/HIV prevention while others last for 15 to 20 sessions. Some

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 85 9 6 are very didactic, while others engage the One of these studies actually evaluated participants in group activities, using role- the impact of three different curricula, Sex playing and other active-learning strategies to Respect, Teen-Aid, and Values and Choices, change group norms and teach assertiveness which were implemented in junior and skills. senior high schools (Weed et al., 1992). (Notably, Values and Choices was modified During most of the 1980s and 1990s, to make it more consistent with abstinence- relatively few resources were devoted to eval- only and Utah state guidelines). Results indi- uating abstinence-only programs. For cated that after one year, these curricula did instance, the federal Title XX Adolescent not significantly affect the initiation of sex Family Life Act (AFLA) of 1981 provided among either junior high school or senior funding for the development and implemen- high school students. tation of abstinence-only programs and for short-term, low-cost evaluations of these Another study measured the impact of programs; however, it did not provide ade- Stay SMART, a health education unit quate time or financial support for more rig- designed to delay sex and prevent alcohol, orous evaluation to measure longer term cigarette, and marijuana use (St. Pierre et al., impact of such programs on sexual behavior. 1995). It did not have a consistent and sig- And few other sources of funding were avail- nificant impact on the frequency of sexual able. Fortunately, this has changed in recent activity. years. The U.S. Department of Health and Human Services' Office of the Assistant The third study was the largest and Secretary for Planning and Evaluation most rigorous of the three studies. It evalu- (ASPE) is currently funding a rigorous evaluated the impact of Postponing Sexual ation of a number of abstinence-only pro- Involvement (PSI), a relatively short, five- grams that are receiving federal funding session program that was taught either by (mainly from the 1996 welfare reform act adults or peers. PSI was the primary compo- and administered through Title V). The first nent in a statewide California initiative to results of that evaluation are expected in late delay sex and prevent teen pregnancy. 2002. Designed to detect even small changes in behavior, the study included random assign- To date, only three impact studies of ment, very large sample sizes, and measure- abstinence-only programs have been identi- ment of behavior at both three months and fied that meet the criteria for this review.2 Of 17 months (Kirby et al., 1995). Results these three, two have been published in peer- demonstrated that the program had no sig- reviewed journals (Kirby et al., 1995; St. nificant impact on the initiation of sex, the Pierre et al., 1995). All three studies mea- frequency of sex among those students who sured program impact on the initiation of sex had ever had sex, or the number of sexual or frequency of sex (see the box on page 87 partners among those who had had sex. The for a summary of the effects and Table 4.1 study examined the impact of PSI on various for descriptions of each study). sub-groups of youth (e.g., determined by

2 Some discerning readers may recognize that there are fewer abstinence-only studies included in this review than in its predecessor, No Easy Answers. The reason is that the criteria for inclusion are different in the two reviews. Whereas No Easy Answers included all published studies meeting weaker criteria, this review includes all studies meeting stronger, fairer criteria, regardless of whether the studies were published. Four abstinence-only studies that were included in No Easy Answers were not included in this volume, primarily because their sample sizes were too small or because they failed to measure initiation of sex for a sufficiently long period of time following completion of the program, and consequently they did not give the programs a fair chance at demonstrating success in delaying initiation of sex.

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 86 97 The Number of Programs with Effects on Sexual and Contraceptive Behaviors by Type of Program Sexuality HIV Sum of Sexuality Abstinence-Only Education Education and HIV Education Programs Programs Programs Programs Initiation of Sex Delayed initiation 6 3 9 Had no significant impact 3 12 6 18

Hastened initiation 0 1 0 1 Total Number of Programs 3 19 9 28 Frequency of Sex Decreased frequency 0 2 3 Had no significant impact 2 8 5 13

Increased frequency 0 1 0 1 Total Number of Programs 2 I I 8 19 Number of Sexual Partners Decreased number 0 3 3 Had no significant impact 4 3 7 Increased number 0 0 Total Number of Programs 4 6 10 Use of Condoms

Increased use 2 ' 8 10 Had no significant impact 5 3 8 Decreased use 0 0 0 Total Number of Programs 7 I I I 8 Use of Contraception Increased use 4 4 Had no significant impact 7 7 Decreased use 0 Total Number of Programs I I 11

gender, ethnicity, and grade level) and found a significantimpact one way or the other on no effects on sexual behavior within any of contraceptive use (Kirby et al., 1995). these sub-groups. Similarly, the study of PSI was the only Although abstinence-only programs are study among these three studies of absti- not designed to affect contraceptive use, some nence-only programs to measure impact on people have expressed concern that if absti- pregnancy. Students who received adult-led PSI were neither more likely nor less likely to nence-only programs discuss contraception report pregnancy during the following 17 primarily in a negative light, then participants months. However, unexpectedly, students will be less likely to use contraception when who received PSI taught by peers were more they do have sex. Accordingly, some people likely to report being pregnant or causing a believe that the impact of abstinence-only pro- pregnancy than their control group (Kirby et grams on contraceptive use should also be A., 1995). The authors could not explain studied. So far, only the study of PSI has mea- this anomalous finding but did attribute it, sured that impact. PSI did not discuss contra- in part, to survey responses from a small ception in a negative light, and it did not have group of seventh grade males in one school.

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 87 9 3 The primary conclusion to reach from behavior and some may not, and particular these three studies is that the evidence is not programmatic characteristics may distinguish conclusive about the impact of abstinence- effective programs from ineffective ones. If only programs. There are only three studies this is true, then communities wishing to put of abstinence-only programs meeting the cri- abstinence-only programs in place may teria for inclusion in this review, and two of increase their chances of selecting effective them had important methodological limita- ones if they choose programs with the com- tions. Given this small number of studies, mon characteristics of effective abstinence- their methodological limitations, and the based sex and HIV education programs great diversity of abstinence-only programs described below (see page 91). that have not been evaluated, one should be careful about making any generalizations Sexuality and HIV Education Programs about all abstinence-only programs. In par- Covering Both Abstinence and ticular, one should not conclude that all Contraception or Condoms abstinence-only programs either do or do not delay sex (or do or do not affect contra- Sexuality and HIV education programs ceptive use). tend to differ from abstinence-only programs in two ways: first, although abstinence is However, it is also true that these few commonly advocated as the best strategy for early results are not encouraging. None of school-aged youth, it is not stressed as the these three studies evaluating the impact of only acceptable behavior; and second, the five different curricula found any overall benefits of contraception are described, effects on sexual behavior. Furthermore, along with their failure rates and/or side there do not currently exist any abstinence- effects. That is, they usually emphasize that only programs with reasonably strong evi- abstinence is the safest method for prevent- dence that they actually delay the initiation ing STDs and pregnancy and that using con- of sex or reduce its frequency. This may doms and other methods of contraception change in the future as rigorous studies cur- provide some protection against STDs and rently underway produce their findings (e.g., pregnancy and, accordingly, are safer than the federal Title V Abstinence Education unprotected sex. Program Evaluation being conducted by Mathematica Policy Research, Inc., and the As noted above, in this review, "sex" or University of Pennsylvania).3 "sexuality" education refers to programs that cover protection against both pregnancy and It may be the case that, in some STDs (and possibly other broader sexuality respects, abstinence-only programs are like topics), while "HIV" education programs sexuality and HIV education programs (dis- refers to programs that focus primarily on cussed below). that is, some abstinence- HIV (and some other STDs). Both groups only programs may be effective at changing of programs include a wide variety of

3Recent publicity has focused attention on a study of the effect of virginity pledges, which suggested that, in some circumstances among certain populations, virginity pledges were associated with later initiation of sex (as well as less use of contraception at first sex for those pledgers who did have sex) (Bearman tic Brlickner, 2001). Virginity pledges are a component of some abstinence-only programs. This study is not discussed in this review because it was not a program evaluation with an experimental or quasi-experimental design and, therefore, did not meet the criteria for inclusion. Instead, it was based on survey data from the nationally representative Add Health data set and provides weaker evidence for a causal impact (see Chapter 3, pages 73-74, for a discussion of the differences between analyses of survey data and program evaluation and why studies based on survey data yield weaker evidence for causal effects).

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 88 9 9 approaches, ranging from programs taught tion programs on the initiation of inter- during regular school classes, to programs course: taught on school campuses after school, to programs taught in homeless shelters and flNine of them (or about one-third) found detention centers. They reflect the consider- that the programs delayed the initiation able creativity and differing perspectives of of sex (Aarons et al., 2000; Blake et aL, the organizations implementing them. 2000; Coyle et al., 2000; Ekstrand et al., 1996; Howard & McCabe, 1990; During debates about sex and HIV Hubbard, Giese, & Rainey, 1998; education programs throughout the country, Jemmott, Jemmott, & Fong, 1998; people have expressed the concern that if Kirby, Barth, Ldand, & Fetro, 1991; programs talk about condoms and contra- Klaus et al., 1987; St. Lawrence et al., ception, they may put sexually engaging 1995). ideas into the heads of youth, may make it OEighteen found no significant impact seem safe and easy for youth to have sex, and (Coyle et al., 1999; Eisen, Zellman, & therefore may actually increase sexual activity. McAlister, 1990; Gottsegen & Philliber, This concern is plausible. Because it is an 2000; Jemmott, Jemmott, & Fong, important issue, many studies have examined 1992; Jemmott, Jemmott, & Fong, this possibility and have consistently found 1998; Kirby, 1985; Kirby et al., 1997; evidence that sex or HIV education does not Levy et al., 1995; Lieberman et al., increase sexual activity. 2000; Little & Rankin, 2001; Main et Other people have argued that if sex al., 1994; Nicholson & Postrado, 1991; and HIV education programs discuss adoles- Thomas et al., 1992; Walter & Vaughn, cent sexual behavior, its consequences, and 1993; Warren & King, 1994). methods of protection (including abstinence) O Only one study out of 28 found that a franldy and realistically, then youth will be sex or HIV education program hastened less likely to engage in sex (or, at the very the initiation of sex (Moberg & Piper, least, will not engage in more sex) and will 1998). be more likely to use contraception effectively. This view is supported by multiple Overall, this is very strong evidence that studies. More specifically, evaluations of these programs do not hasten sex and that many sex and HIV education programs some of them actually delay sex. strongly support the conclusion that these Nineteen studies examined the impact curricula do not increase sexual intercourse, of sexuality and HIV education programs on either by hastening the onset of intercourse, the frequency of intercourse: increasing the frequency of intercourse, or increasing the number of sexual partners, oFive studies found that they reduced the and that some, but not all, programs can frequency of sex (Coyle et al., 2001; delay and reduce sexual activity (see the box Howard & McCabe, 1990; Jemmott, on page 87 for a summary of the effects and Jemmott, & Fong, 1992; Jemmott, Tables 4.2 and 4.3 for summaries of each Jemmott, & Fong, 1998; St. Lawrence study). et al., 1995).

Twenty-eight studies meeting the crite- OThirteen found no significant impact ria discussed above have examined the (Blake, Ledsky, Lohrmann, et al., 2000; impact of middle school, high school, or Coyle et al., 1999; Jemmott, Jemmott & community-based sexuality or HIV educa- Fong, 1998; Kirby, 1985; Kirby et al.,

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 89 100 1991; Kirby, Korpi, Adivi et al., 1997; Borus et al., 1991; St. Lawrence et al., 1995; Levy et at, 1995; Little & Rankin, 2001; Walter & Vaughn, 1993). Similarly, four of Main et al., 1994; Moberg & Piper, eleven programs that measured contraceptive 1990; Rotheram-Borus et at, 1991). use more generally significantly increased its use (Aarons et al., 2000; Coyle et al., 1999; 0Only one of nineteen studies found a sig- Gottsegen & Philliber, 2000; Kirby, Barth, nificant increase in frequency (Moberg & Leland, & Fetro, 1991). None of the pro- Piper, 1998). grams reduced either condom or contracep- Again, this is strong evidence that these pro- tive use. Taken together, these results are grams do not increase the frequency of sex quite positive. and that some of them reduce the frequency. A disproportionate number of the programs that significantly increased either con- Finally, of the ten studies that examined dom or contraceptive use more generally impact on number of sexual partners, three were HW education programs that increased found a significant decrease in partners condom use. Eight out of eleven HIV edu- (Jemmott, Jemmott, & Fong, 1992; Main et cation programs found significant effects on al., 1994; St. Lawrence et al., 1995), seven condom use, while two out of seven sex edu- found no impact (Coyle et al., 1999; cation programs found significant effects on Gillmore et al., 1997; Kirby et al., 1991; condom use and four out of eleven found Kirby, Korpi, Adivi, et al., 1997; Levy et al., significant effects on contraceptive use more 1995; Little & Rankin, 2001; Magura, Kang, generally. It cannot yet be determined & Shapiro, 1994), and none found a signifi- whether HIV education programs are inher- cant increase. Once more, this is strong evi- ently more effective than more general sex dence that these programs do not increase education programs that cover pregnancy, the number of sexual partners. STDs, HIV, and other topics, or whether HIV education programs have simply been In sum, these data strongly indicate that better funded, provided better training, had sex and HIV education programs do not sig- studies with larger sample sizes, or had some nificantly increase any measure of sexual other advantage that might improve effec- activity, as some people have feared. These tiveness. The special effectiveness of HW results are also consistent with reviews of education programs may also reflect the fact programs evaluated in other countries that that AIDS is undoubtedly a more salient have also found that sex and HIV education threat than pregnancy for high-risk males in programs do not increase any measure of some communities. sexual activity (Grunseit et al., 1997). The data also suggest that these sex and These studies also demonstrate that HIV education programs may be more effec- some programs increased condom use or tive with higher risk youth than with lower contraceptive use more generally. Of the risk youth. This may be partly due to the eighteen programs for which impact on con- behavioral characteristics of high-risk youth dom use was evaluated, ten programs (or that is, when youth engage in a large more than half) significantly increased some amount of unprotected sex, there is greater measure of condom use (Coyle et al., 1999; room for improvement than if they engage Hubbard, Giese, & Rainey, 1998; Jemmott, in little unprotected sex to begin with. In Jemmott, & Fong, 1992; Jemmott, addition, these findings may be due, in part, Jemmott, & Fong, 1998; Magura, Kang, & to methodological and statistical factors. A Shapiro, 1994; Main et al., 1994; Rotheram- program that reduces the proportion of

THE NATIONAL cAMPAIGN TO PREVENT TEEN PREGNANCY 90 IA/11 ri lower risk youth who initiate sex from 6 per- five behavioral effects that the first study cent to 4 percent is more difficult to measure found. For example, the initial positive than a program that reduces the proportion results for Postponing Sexual Involvement in of higher risk youth who initiate sex from 12 Atlanta, Georgia, were not replicated in percent to 8 percent, even though the pro- California (Howard & McCabe, 1990; Kirby, portional reductions are the same. Korpi, Barth, et al., 1997). However, there is now evidence of one successful replication. The strength of the evidence for the Two separate research teams in California effectiveness of some sex and HIV education and Arkansas trained people to implement programs has improved considerably during Reducing the Risk, implemented it in multi- recent years. In 1997, No Easy Answers ple schools in each study, and evaluated the raised serious concerns about the method- impact of the curriculum on adolescent sex- ological rigor of some of the studies evaluat- ual behavior. Both found that it delayed the ing these programs. It noted that (1) many onset of sexual intercourse and increased use studies did not include random assignment, of condoms or contraception more generally large sample sizes, long-term follow-up, among some groups of youth (Hubbard, measurement of behavior, and proper statisti- Giese, & Rainey, 1998; Kirby et al., 1991). cal analyses, and (2) the few studies that did Such confirmation of positive behavioral include all these methodological strengths findings is most encouraging, providing failed to find positive and significant effects greater evidence that Reducing the Risk can on behavior (Kirby, Korpi, Barth, et al., delay the onset of intercourse in different 1997; Kirby, Korpi, Adivi, et al., 1997; communities throughout the country. "Thomas et al., 1992). However, there are now three studies with random assignment, Only one study has estimated the cost- large sample sizes, long-term follow-up, effectiveness and cost-benefit of a sex educa- measurement of behavior, and proper statisti- tion program; that study found that for every cal analyses that have shown statistically sig- dollar invested in the Safer Choices program, nificant and programmatically important $2.65 in total medical and social costs were reductions in adolescent sexual risk-taking saved (Wang et al., 2000). The savings were (Coyle et al., 1999; Jemmott, Jemmott, & produced by preventing pregnancy and Fong, 1998; St. Lawrence, 1995). These STDs, including HIV. three studies clearly indicate that certain school-based and community-based sex and Common Characteristics of Effective Curricula HIV education programs can delay sex, decrease the frequency of sex, increase con- Those curricula with evidence that they dom or cOntraceptive use, or decrease unpro- reduce sexual risk-taking share ten particular tected sex. In previous years, few studies characteristics, noted below. Some of these measured or found long-lerm effects. characteristics have also been identified in However, that too has changed. Several other reviews of impact studies (Frost & recent studies have found lasting effects for Forrest, 1995; Miller & Paikoff, 1992; one year, some have found effects for about Moore et al., 1995). These characteristics 18 months, and one study found effects that reflect different aspects of effective teaching lasted more than 31 months after the inter- and are similar to the characteristics of edu- vention (Coyle et al., 2001). cational programs found to reduce substance abuse (Dusenbury & Falco, 1995). In years past, there were also few replications of studies. When they did occur, the The ten characteristics appear to be second study usually failed to find the posi- necessary characteristics that is, when

Emerging Answers: Research Findings on Programs to,. Reduce Teen Pregnancy 9 I 1 2 evaluated programs lacked one or more of schools, and others, and indirectly through these characteristics, they.were typically observing the behavior of others and the found to be ineffective at changing behavior. consequences that befall them. In addition, However, there is little evidence specifying social influence theories address societal pres- which of these factors or combinations of sures on youth and the importance of help- factors contributes most to the overall suc- ing young people understand those pressures cess of the programs. and resist the negative ones. Thus, these programs strive to go far beyond the cognitive These ten characteristics of effective level; they focus on recognizing social influ- programs are: ences, changing individual values, changing group norms and perceptions of those 1. Effective programs focused on norms, and building.social skills. reducing one or more sexual behaviors that lead to unintended pregnancy or These theories help to specify which HIV/STD infection. These programs particular antecedents the interventions are focused narrowly on a small number of spe- trying to change (e.g., the beliefs, attitudes, cific behavioral goals, such as delaying the norms, confidence, and skills related to sex- initiation of intercourse or using condoms or ual behavior), so that changes in these other forms of contraception; relatively little antecedents would lead to voluntary change time was spent addressing other sexuality in sexual or contraceptive behavior. Thus, issues, such as gender roles, dating, or par- each activity was designed to change one or enthood. Nearly every activity was directed more antecedents specified by the particular toward the behavioral goals. theoretical model for the curriculum, and each important antecedent in the theoretical Few studies evaluated the impact of a model was addressed by one or more activi- focused and potentially effective curriculum ties. While all of the effective curricula unit that was embedded in a larger more focused on antecedents specified by their comprehensive sexuality education program. adopted theories, some program developers Such units may or may not effectively change actually surveyed students and empirically behavior, but only additional research will determined which possible antecedents best answer this question. predicted desired behavior. Activities in their programs then focused on those particular 2. Effective programs were based on antecedents. theoretical approaches that have been demonstrated to be effective in influenc- By focusing on specific behavior (char- ing other health-related risky behaviors acteristic #1), by identifying particular such as social cognitive theory (Bandura, antecedents causally related to that behavior, 1986), social influence theory (McGuire, and by designing activities to change each of 1972), social inoculation theory (Homans, those important antecedents, the developers 1965), cognitive behavioral theory (Bandura, of these programs were, in fact, designing 1986; Schinke et al., 1981), theory of rea- "logic models" and basing their interven- soned action (Fishbein & Ajzen, 1975) and tions on those models (Kirby, 2000). Logic theory of planned behavior (Ajzen, 1985). models are discussed in Chapter 6. These theories together address many of the individual sexuality-related antecedents iden- 3. Effective programs gave a clear tified in Chapter 2. They recognize the fact message about sexual activity and condom that the beliefs and values of youth are influ- or contraceptive use and continually rein- enced directly through education by parents, forced that message. This particular charac-

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 92 103 teristic appeared to be one of the most doms). Some of them also addressed peer important criteria that distinguished effective norms about having sex or using condoms. from ineffective curricula. The effective pro- For example, some curricula provided data grams did not simply lay out the pros and showing that many youth do not have sex or cons of different sexual choices and implicitly do use condoms, or they had students let the students decide which was right for engage in activities in which they concluded them; rather, most of the curriculum activi- that students should abstain from sex or use ties were directed toward convincing the stu- condoms and then expressed those beliefs to dents that abstaining from sex, using other students. At least one curriculum condoms consistently, or using other forms addressed media influences (e.g., how sex is of contraception consistently was the right used to sell products and how television choice, and that unprotected sex was clearly often depicts characters having unprotected an undesirable choice. To the extent possi- intercourse but rarely experiencing negative ble, they tried to use group activities to consequences). change group norms about what was the expected behavior. 6. Effective programs provided modeling of and practice with communication, 4. Effective programs provided basic, negotiation, and refusal skills. Typically, accurate information about the risks of the programs provided information about teen sexual activity and about methods of skills, demonstrated the effective use of those avoiding intercourse or using protection skills, and then provided some type of skill against pregnancy and STDs. Effective rehearsal and practice (e.g., verbal role-play- programs provided basic information that ing and written practice). Some curricula students needed to assess risks and avoid taught different ways to say "no" to sex or unprotected sex. Typically, this information unprotected sex, how to insist on the use of was not detailed or comprehensive. For condoms or other methods of contraception, example, the curricula did not provide how to use body language that reinforced detailed information about all methods of the verbal message, how to repeatedly refuse contraception or different types of STDs. sex or insist on condom use, how to suggest Instead, they provided a foundation: they alternative activities, and how to help build emphasized the basic facts needed to per- the relationship while refusing unprotected suade youth to avoid unprotected sex, and sex or refusing to have sex at all. Some cur- they provided information that would lead to ricula started with easier scenarios in role changes in beliefs, attitudes, and perceptions playing and then moved to more challenging of peer norms. Some curricula also provided ones. Some started with fully scripted role more detailed information about how to use plays and moved to more improvisational condoms correctly. ones, in which the youth resisting unprotected sex had to use their own words. 5. Effective programs included activi- Although all effective curricula gave some ties- that address social pressures that attention to skills, there were significant vari- influence sexual behavior. These activities ations in the quality of activities designed to took a variety of forms. For example, several teach skills and also in the time devoted to curricula discussed situations that might lead practicing the skills. to sex. Most of the curricula discussed "lines" that are typically used to get someone 7. Effective programs employed a to have sex, and some discussed how to variety of teaching methods designed to overcome social barriers to using condoms involve the participants and have them (e.g., embarrassment about buying con- personalize the information. Instructors

Emerging Answers: Research Findings on Programs4o1Ruce Teen Pregnancy 93 iUt reached students by engaging them in the tant activities adequately. In general, it learning process, not through didactic requires considerable time and multiple instruction. Students were involved in activities to. change the most important numerous experiential classroom and home- antecedents of sexual risk-taking and to work activities, such as small group discus- thereby have a real influence on behavior. sions, games or simulations, brainstorming, Thus, short programs that lasted only a cou- role-playing, written exercises, verbal feed- ple of hours did not appear to be effective, back and coaching, interviewing parents, while longer programs that had many activi- locating contraception in local drugstores, ties had a greater effect. More specifically, and visiting or telephoning family planning clinics. In addition to these experiential activ- effective programs tended to fall into two ities, a few effective curricula used peer edu- categories: (1) those that lasted 14 or more cators or videos with characters (either real hours and (2) those that lasted a smaller or acted) who resembled the students and number of hours but recruited youth who with whom the students could identify. All of voluntarily participated and then worked these activities kept the students more with these youth in small group settings with involved in the program, got them to think a leader for each group. (When youth volun- about the issues, and helped them personal- teer to participate, they may be more open ize the information in their own lives. to instruction than if they are required to sit in a school class. And when they work in 8. Effective programs incorporated small groups, instructors may be able to behavioral goals, teaching methods, and materials that were appropriate to the involve the youth more completely, to tailor age, sexual experience, and culture of the the material to each group, and to cover students. For example, programs for more material and more concerns more younger youth in junior high school, few of quickly.) whom had engaged in intercourse, focused on delaying the onset of intercourse. 10. Effective programs selected Programs designed for high school students, teachers or peer leaders who believed in some of whom had engaged in intercourse the program they were implementing and and some of whom had not, emphasized that then provided them with training. Given students should avoid unprotected inter- the challenges of implementing programs course; that abstinence was the best method that focused on a sensitive topic and incor- of avoiding unprotected sex; and that con- porated a variety of interactive activities, the doms or contraception should always be used effective programs carefully selected teachers if they did have sex. And programs for and provided them with training. The train- higher-risk youth, most of whom were ing ranged from approximately six hours to already sexually active, emphasized the three days. In general, the training was importance of always using condoms and designed to give teachers and peer leaders avoiding high-risk situations. Some of the information on the program as well as prac- curricula, such as Becoming a Responsible tice using the teaching strategies included in Teen and Making a Difference, were designed the curricula (e.g., conducting role-playing for specific racial or ethnic groups and exercises and leading group discussions). emphasized statistics, values, and approaches that were tailored to those groups. Some of the teachers in these effective programs also received coaching and/or follow- 9. Effective programs lasted a suffi- up training to improve the quality of their cient length of time to complete impor- teaching.

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 94 1 5 Conclusions about Abstinence-Only, Sex substantial period of time that is, for Education, and HIV Education Programs as long as 31 months. Given both the magnitude and duration of their impact, Taken as a whole, the studies reviewed they may actually reduce teen pregnancy above suggest the following conclusions and STDs. about abstinence-only, sex education, and HIV education programs: 0 The programs that effectively changed risk-taking behavior focused on sexual oGiven the great diversity of abstinence- and contraceptive behavior, gave a clear only programs, the few studies that have message about what behavior was desired been completed, and the methodological or appropriate, and, more generally, limitations of those studies, few conclu- shared all ten program characteristics sions can be reached about abstinence- identified above. Notably, short pro- only programs. Clearly, some programs grams, regardless of whether they were are not effective at significantly delaying abstinence-only programs or sex and the onset of intercourse, but others HIV education programs, did not have a might turn out to be effective. Currently, measurable impact on behavior. there are no specific abstinence-only pro- 0 These educational programs are not a grams with strong evidence for their suc- complete solution to reducing unpro- cess, but this may change when research tected sexual intercourse some pro- in progress is completed. gram participants continued to engage in oSexuality and HIV education programs unprotected intercourse even after com- that include discussion of condoms and pleting the most effective programs; only contraception do not increase sexual a few studies measured and demon- intercourse; they do not hasten the onset strated long-term effects; few studies of intercourse, do not increase the fre- even measured impact on pregnancy quency of intercourse, and do not rates and those that did found none; and increase the number of sexual partners. none of the studies measured impact on To the contrary, some of the programs STD rates. Nevertheless, some of these actually delay sex, reduce the frequency programs did show important long-term of sex, or reduce the number of sexual behavioral effects (e.g., reduction of partners. And, notably, at least two inde- unprotected sex), which suggests that pendent studies found that one particular they make effective components of sex education curriculum delays the larger, more comprehensive initiatives to onset of sexual intercourse. prevent pregnancy. ci Studies of some, but not all, of these Sex and HIV/AIDS Education same sex and HIV education programs Programs for Parents and Their provided strong evidence that they sub- Families stantially increased condom or contracep- Most parents and their children have tive use. Thus, some sexuality or HIV remarkably few conversations about sexual education programs reduced unprotected topics, often because both parents and their sex both by reducing sexual intercourse teens feel so uncomfortable doing so. To and by increasing the use of protection help alleviate this problem, many brief edu- against STDs and pregnancy. cational programs have been designed to oSome sex and HIV education programs increase parent/child communication. These had a positive impact on behavior for a include programs for parents only, programs

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 95 106 for parents and their children together, grams designed to increase parent/child homework assignments in school sex educa- communication actually reduce teen sexual tion classes requiring communication with risk-taking. parents, and video programs with written materials to be completed at home. Clinic or School-Based Programs Designed to Provide Reproductive Many studies indicate that few parents Health Care or to Improve Access to are willing or able to participate in such pro- Condoms or Other Contraceptives grams and that getting them to enroll in these programs is very challenging. With In this section, studies of five groups of remarkable consistency, however, many stud- interventions are summarized: family plan- ies also indicate that, when parents do partic- ning services, protocols for clinic appoint- ipate, they do increase their communication ments and supportive activities, other clinic with their children about sexuality in the characteristics and programs, school-based short term, as well as their comfort with that and school-linked clinics, and school communication (Kirby, 2001). However, condom-availability programs. those positive effects on communication seem to dissipate with time. Family Planning Services

Because there does not exist a simple The primary objective of family plan- relationship between parent/child communi- ning clinics or family planning services cation and adolescent sexual and contracep- offered within other health settings is to pro- tive behavior, it is not sufficient to simply vide clients with contraception and other examine the impact of programs on commu- reproductive health services and with the nication; studies must also examine their knowledge and skills to use them. According effects on sexual behavior. Only two studies to a 1992 national survey of family planning have done so (see Table 4.4). The first of clinics, many have special facilities for these, implemented by Girls Inc. (formerly teenagers, three-fourths encourage their Girls Clubs), included five, two-hour sessions counselors to spend extra time with clients for mothers and their daughters (Nicholson under age 18, four-fifths have outreach pro- & Postrado, 1991). Although the program grams for teenagers, many have programs for group appeared less likely to initiate inter- teenage parents as well as for the parents of course, the result was not statistically signifi- these teenagers, and many have sex educa- cant, and the program and comparison tion training programs for adults and teens groups were probably not equivalent before (Henshaw & Torres, 1994). the program. The second program included a well-designed video and written materials Adolescents do not immediately begin to be used at home (Miller et al., 1993). It using organized family planning services increased parent/child communication but when they first have sex. As noted above, failed to delay the onset of intercourse signif- they are far more likely to use a condom the icantly, in part because very few youth in first time they have sex (66 percent) than either the intervention or control group ini- oral contraceptives (which require the help tiated sex in the conservative community of health care providers), and 60 percent of where it was implemented. Thus, despite the teen girls wait a year or more after initiating widespread belief that parent/child commu- intercourse before visiting a doctor or clinic nication about sexuality will delay sex and for contraception (Alan Guttmacher reduce adolescent sexual risk-taking, there is Institute, 1994). However, many sexually little evidence available indicating that pro- active female teenagers do receive family

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 96 17 planning services. According to the 1995 Lundberg & Plotnick, 1990; Olsen & Weed, National Survey of Family Growth, an esti- 1986; Singh, 1986; Weed & Olsen, 1986), mated 2.6 million 15- to 19-year-old females the strength of their conclusions is greatly (or nearly one-third of all females in that age weakened by conflicting results among stud- group) made one or more visits to a clinic, ies and by several severe methodological private medical source, or counselor for fam- limitations. Thus, the actual impact on ado- ily planning in a single year of those, lescent pregnancy rates of either family plan- almost two-thirds visited a clinic (Abma et ning services generally or subsidized family al., 1997). Many of these young women planning clinics specifically has not been received oral contraceptives and other con- accurately estimated. traceptives that are more effective at preventing pregnancy than condoms or other However, there are other important, non-prescription methods that can be pur- though narrower, questions that have been chased without the requirement of a clinic addressed by studies employing experimental visit or prescription. Accordingly, these fam- or quasi-experimental designs: How can family planning services presumably prevented ily planning services be improved so that many adolescent pregnancies that would adolescents who use them engage in less have occurred if these services did not exist unprotected sex? More specifically, how can or were significantly curtailed. the protocols for clinic appointments be improved and supported by other activities? The estimated number of adolescent And how can broader clinic policies and pregnancies averted by family planning ser- other characteristics be improved? vices depends greatly on what assumptions one makes about what the prevalence of teen Protocols for Clinic Appointments and sexual and contraceptive activity would be if Supportive Activities prescription contraceptives were unavailable. That is, it is not known how teenagers would Six studies have examined what happens change their behavior if highly effective con- during a clinic visit the counseling and traceptive methods were not available; some instruction that takes place between a medi- might refrain from sex altogether, some cal provider and a teen patient and the other might have sex and use less effective meth- materials and activities that can support and ods, and some might have sex and not use reinforce that counseling. Four found positive any method at all. If one assumes that youth effects on condom or contraceptive behavior would use over-the-counter methods of con- (see Table 4.5). All six interventions were traception if prescription methods were not part of longer medical appointments. available, then the estimated number of additional teen pregnancies per year that would During the most modest of the six result ranges from 40,000 to 160,000 interventions, African-American teen males nationally, depending upon changes in sexual attending an STD clinic received either a activity (Kalm, Brindis, & Glei, 1999). 14-minute video, or a one-on-one session with a health educator, or standard care Unfortunately, there is remarkably little (DeLameter, Wagstaff, & Havens, 2000). All research about the impact of family planning males received the results of their STD tests services generally. While there have been sev- and appropriate treatment. The experimental eral studies of the effects of family planning design was a rather strong one, but the study clinics on pregnancy or birth rates (Anderson failed to find any significant differences & Cope, 1987; Brewster et al., 1993; among the behavioral effects of these three Forrest, Hermalin, & Henshaw, 1981; treatment models. Notably, condom use, as

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 97 108 well as frequency of sex and number of sex- patient's physician then reviewed the risk ual partners, increased among all three assessment with the patient and discussed groups over time. concerns and methods of avoiding unprotected sex. An experimental design indicated The second study evaluated a very that the program increased use of condoms modest intervention for female patients with during the three months after the interven- chlamydia (Orr et al., 1996). A nurse spent tion. It might also have increased the chances about 10 to 20 minutes discussing chlamydia of having sex but not the frequency of sex with the aid of a pamphlet, demonstrated during the following three months, but the how to put a condom on a banana (and got results were mixed depending upon the type the patient to practice), and engaged the of analysis. However, both results ceased patient in a brief role-play involving a woman being significant by nine months after the getting her partner to use a condom. An intervention. experimental design was used to measure the impact at six months and found that those In the fifth study, a family planning youth who received the special instruction clinic substantially improved its clinic proto- were substantially more likely to use con- col for adolescents by placing greater focus doms than those youth who received the upon non-medical problems, providing more standard intervention. information and more counseling, delaying the medical examination until the second The third study evaluated a program for visit, and giving more attention to partner males that included two parts: (1) a slide- and parent involvement (Winter & tape program that focused on anatomy, Breckenmaker, 1991). It also designated one STDs, contraception, couple communica- staff person as a teen counselor. The study tion, and access to health services and (2) a did not have a strong evaluation design, but visit with a health care practitioner who its results indicate that it did increase contra- focused on contraception, reproductive ceptive use. health goals, health risks, and the patient's related interests. Both parts emphasized Finally, the last study examined two dif- abstinence and the use of contraception if ferent interventions, both of which were very sexually active. A strong experimental design modest (Hercog-Baron et al., 1986)..During and questionnaire data collected a year later a clinic visit, patients either (1) were invited indicated that the program did not signifi- to bring their parents to six subsequent visits cantly affect sexual activity but did increase or (2) received two to six telephone calls use of contraception, especially by the males' from the clinic staff regarding their use of partners and by program participants who the contraceptives method(s) they had cho- were not sexually experienced at baseline sen. Only 36 percent of the teens in the first (Danielson et al., 1990). group attended any subsequent visits with their parents, but most youth in the second The fourth program focused on group (84 percent) did receive the phone HIV/STD prevention and served equal per- calls. Results indicated that the program centages of males and females (Boekeloo et failed to have a significant impact on use of al., 1999). It included a 15-minute audio- contraception or pregnancy. taped risk assessment and education program, a discussion ice-breaker, two brochures on The fact that four of these six studie skills and ways to avoid unprotected sex, a found positive effects on behavior with such brochure on community resources, and par- brief, modest interventions is quite encour- ent brochures. On a one-to-one basis, the aging. It should be noted that all four of the

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 98 109 effective interventions focused on sexual and School-Based Health Centers and contraceptive behavior, gave clear messages School-Linked Reproductive Health about appropriate sexual and contraceptive Clinics behavior, and included one-on-one consulta- School-based health centers are clinics tion about the client's own behavior. At the located on school grounds that offer services very least, these studies suggest that such to students in their schools. The purpose of approaches should be further developed and these clinics is to provide primary health care rigorously evaluated. These results should services that are affordable and accessible to also encourage medical providers to review students who otherwise might not have ongo- their instructional protocols with youth and ing access to such services. Consequently, all to spend more time talking with individual of these clinics provide basic primary health adolescent patients about their sexual and care services. In 1999, there were at least contraceptive activity. 1,135 school-based health Centers, and 70 percent of these served students in grades 7- Other Clinic Characteristics 12 (Making the Grade, 2000). and Programs While 90 percent of clinics in secondary Another study evaluated the effect of a schools provide at least one reproductive city-wide effort to improve family planning health service (such as gynecologic exams, services for young people (Hughes, birth control counseling, pregnancy testing, Furstenberg, & Teitler, 1995) (see Table and STD diagnosis and treatment), only 29 4.6). That initiative, called RESPECT percent of all clinics write prescriptions for (Responsible Education on Sexuality and birth control pills that can be filled else- Pregnancy for Every Community's Teens), where, and only 18 percent actually dispense involved nine existing clinics that initiated or birth control pills (Fothergill & Feijoo, expanded after-school or evening hours, 2000). About 28 percent dispense condoms. began teenage walk-in hours, decreased the average waiting time for appointments, and When school-based clinics are well- increased the hours reserved for teenagers staffed and well-run and dispense contracep- only. The program also trained staff to work tives, they potentially have many qualities with teens. In addition, the initiative con- that might be seen as those of ideal adoles- ducted school and community activities, as cent reproductive health programs that is, well as a media campaign. This study differed their location is convenient to the students, from most others in this review in that it they reach both females and males, they pro- measured the impact of this city-wide effort vide comprehensive health services, they are on all the youth in the targeted geographical confidential, their staff are selected and areas, not just on those youth who actually trained to work with adolescents, they can used the clinic. This is a far more demanding easily conduct follow-up, their services are challenge, but it was an appropriate one free, and they can integrate education, coun- given the initiative's goal to expand services seling, and medical services. On the other and its community-wide efforts, including hand, school-based clinics typically do not the media campaign. A comparison of reach out-of-school teenagers (e.g., those changes over time between the catchment who have graduated from or dropped out of area and a comparison area revealed no sig- high school), nor do they reach older males, nificant changes in contraceptive use or preg- the ones who most often father children nancy or childbearing rates. born to adolescent girls.

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 99 When school-based clinics make contra- in this review and is not summarized.) The ception available, many sexually-experienced first two studies (Kirby, 1991; Kirby, Waszak, students obtain it from the clinics. For exam- & Ziegler, 1991) found that the presence of ple, in a study of four clinics that provided the clinic did not affect the onset of sexual prescriptions or actually dispensed contracep- intercourse either positively or negatively. tives, the proportion of sexually-experienced Kisker, Brown, and Hill (1994) found some females who obtained contraceptives through data indicating that the clinic and its educa- the clinic varied from 23 percent to 40 per- tional programs may have delayed the onset cent (Kirby, Waszak, & Ziegler, 1991). of intercourse. The first two studies also found that clinic presence was not associated Six studies have examined the effects on with greater frequency of intercourse. In behavior of schOol-based health centers (see combination, these studies indicate that pre- Table 4.7) (Edwards et al., 1980; Kirby, scribing or dispensing contraceptives on high 1991; Kirby, Waszak, & Ziegler, 1991; Kirby school campuses does not hasten the onset et al., 1993; Kisker, Brown, & Hill, 1994; of intercourse nor increase its frequency, as Newcomer, Duggan, & Toczek, 1996). (An some people have feared. additional study discussed. below examined the impact of a school-linked reproductive These same three studies measured health clinic.) Five of the six studies exam- clinic impact on contraceptive use, but their ined school-based clinics in three or more results were mixed. Kisker, Brown, and Hill schools. Although the quasi-experimental (1994) produced the most negative results designs used in the studies varied consider- clinic presence was associated with lower ably, they were not strong in general. For rates of contraceptive use. It is not clear what example, some of them did not include both produced this counter-intuitive result. The intervention and comparison groups and also first Kirby study (1991) examined the impact pretest and posttest data in the same design. of a clinic that did not focus on reproductive Consequently, inferences should be drawn health and that prescribed but did not dis- cautiously from these studies. In addition, pense contraception. That clinic did not these studies measured population effects increase contraceptive use in that school. The that is, they typically measured the effects on results of the Kirby, Waszak, and Ziegler the entire school population and not just on study (1991) varied with the site. At one site those students who actually used the clinics for family planning services. Measuring that was run by Planned Parenthood, impact on all students is the appropriate cri- focused on high-risk youths, emphasized terion for judging the effectiveness of such pregnancy prevention, and dispensed oral an intervention, but it is a more demanding contraceptives, there was a significantly one. greater use of oral contraceptives among female students than among female students Three of these studies measured the in the comparison school many miles away; effect of school-based clinics on sexual however, there was no significant difference behaVior, and they found consistent evidence in condom use. At two other sites that dis- that the clinics did not increase sexual activ- pensed both condoms and oral contracep- ity (Kirby, 1991; Kirby, Waszak, & Ziegler, tives but did not have a strong educational 1991; Kisker, Brown, & Hill, 1994). The component, no significant differences were three studies evaluated the effects of one, six, found between the clinic and comparison and 19 clinics respectively. (However, one of schools in use of condoms by male students the six sites in the Kirby, Waszak, and Ziegler or use of oral contraceptives by female stu- study did not meet the criteria for inclusion dents. At these latter two schools, there

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 100 lii clearly were substitution effects even schools with clinics and 19 schools without though many sexually experienced students clinics in a mid-Atlantic city. They also found obtained contraception from the clinics, no evidence that opening clinics in the four most of those students would have obtained high schools reduced birth rates. contraception elsewhere if the clinics had not been there. In contrast to these school-based health centers that were located on the school Three studies of school-based clinics grounds and that provided primary health examined their impact on pregnancy rates, care services, the Self Center in Baltimore and all failed to find an impact. Kirby (1991) was located across the street from a high found that opening a clinic did not signifi- school and four blocks away from a junior cantly reduce the pregnancy rate in the one high school and provided only reproductive site in that study. Similarly, Kirby, Waszak, health services (Zabin et al., 1986). More and Ziegler (1991) found that clinic pres- specifically, it provided educational, counsel- ence was not significantly related to preg- ing, and reproductive health services in the nancy rates in any of the five school-based clinic, as well as educational and counseling clinic sites, after background characteristics services in the two schools. In both schools, of the students were statistically controlled. the staff implemented a peer education pro- Finally, Kisker, Brown, and Hill (1994) also gram and afterschool group discussions, found that clinic presence was not related to while in the clinic the staff provided individ- overall pregnancy rates in the multiple sites ual counseling, group counseling, and con- they examined. traceptive services. According to survey data collected from the two program schools and Three studies examined clinic impact on two matched comparison schools, there was birth rates, and they found mixed results. a delay in the intervention schools in the Edwards and colleagues (1980) gave a great onset of sexual intercourse among those impetus to school-based clinics with a report youth who had not yet initiated sex and an that birth rates declined in three different increase in the use of contraception among schools in St. Paul after clinics providing those who had ever had sex. In addition, reproductive health care (including prescrip- there was an apparent decrease in pregnancy tions for contraception) were opened. rates in the program schools two years after However, these conclusions were based on the Self Center opened. only one baseline year for each school and on the clinic staff's knowledge of births These studies of comprehensive school- among students. Subsequently, Kirby and based health centers and a school-linked colleagues (1993) studied five St. Paul reproductive health clinic provide inconsis- school-based clinics (including the three tent results. While all the studies provided studied by Edwards and colleagues), over- evidence that the clinics did not increase sex- coming the previous study's limitations by ual activity, their evidence regarding contra- generating birth rates from school and public ceptive use and pregnancy rates was mixed. records in St. Paul for two to five baseline There are at least two possible explanations years and for multiple post-clinic years. That for this. The first is methodological: few of study found large year-to-year variations in these studies had strong quasi-experimental school-wide birth rates but no evidence indi- designs, and all had important limitations cating that the clinics significantly reduced that could have affected results either one birth rates. Similarly, Newcomer, Duggan, way or the other. A second explanation is and Toczek (1999) used public records to programmatic: it is notable that only two of generate birth rates for ten years for four the clinics appeared to increase contraceptive

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 10 112 use one was a school-based clinic that was Thus far, only four studies meeting the run by Planned Parenthood, focused consid- criteria for this review have presented results erably on reproductive health, and gave a on the behavioral effects of condom-avail- clear message about contraceptives; the sec- ability programs in schools (Table 4.8). The ond was a school-linked clinic that focused study with the strongest evaluation design solely on reproductive health and also had assessed the effects of making condoms avail- charismatic staff who gave a clear message able through vending machines in five about abstinence and contraceptives. It is Seattle schools without school-based clinics possible that, in increasing school-wide rates and through vending machines and baskets of contraceptive use, location of the clinics is in five additional Seattle schools with pre- not as important as a clear focus on repro- existing school-based clinics (Kirby, Brener, ductive health. Further research is needed to Brown, et al., 1999). School-wide data were determine which of these (or other) explana- collected both before condoms were made tions best explains the apparent differential available and then again two years later. In success of the clinics. neither group of schools was there an increase in sexual activity. In the schools with School Condom-Availability Programs only vending machines and no clinics, there were no significant changes in condom use Given the threat of AIDS, other STDs, (or in use of oral contraceptives). By con- and pregnancY, more than 300 schools with- trast, in the schools with clinks and baskets out school-based clinics have made condoms of condoms, there was, surprisingly, a signifi- available through school counselors, nurses, cant decrease in condom use and a significant teachers, vending machines, or baskets increase in oral contraceptive use, suggesting (Kirby & Brown, 1996). These schools are that the clinics may have begun encouraging in addition to about 250 schools that make oral contraceptive use in addition to provid- condoms available to students through ing condoms. Notably, these schools also had school-based clinics. a strong HIV education intervention, but, because it existed before the baseline data When available in school, the number collection, the effect of the educational com- of condoms obtained per student from ponent was not measured by this study. schools varies greatly from program to program (Kirby & Brown, 1996). In general, A second study measured the impact of students in smaller alternative schools (prob- making condoms available in baskets in nine ably with more high-risk students) obtained Philadelphia schools (Furstenberg et al., many more condoms per student than stu- 1997). Students in those schools could dents in larger schools or students in main- receive reproductive health information, con- stream schools. In addition, when schools doms, and general health referrals. Both made multiple brands of condoms available before and after the centers were opened, in baskets in convenient and private locations young people were randomly selected for and without any restrictions, students personal interviews from census tracts sur- obtained many more condoms than when rounding these nine schools and other com- there were restrictions (e.g.,when students parison schools. Results revealed that in the could only obtain a small number of con- schools with centers, changes over time in doms from school personnel at specified four measures of sexual behavior or condom times after brief counseling). Finally, students use were not significantly different from the obtained many more condoms in schools changes over time in the schools without that had clinics. centers. However, the authors noted that

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 102

1 I "T? there were consistent non-significant differ- last time they had sex, and were less likely to ences in the trends over time: students in use a method of contraception other than schools with centers reduced their sexual condoms the last time they.had sex (Blake, activity and increased their condom use more Ledsky, Goodenow, et al., 2000) than students in schools without centers. Relatively small samples may have limited the What conclusions can one reach from ability to detect programmatically meaning- these four studies about the impact of con- ful results. dom availability on condom use? There are three logical possibilities. First, the differ- The third study evaluated a comprehen- ences in results could be caused by differ- sive AIDS prevention program in the New ences in the research methods. If this is true, York City high schools. It included instruc- then this group of studies provides weak tion about AIDS, school-wide activities, and overall evidence that school condom avail- condom availability. Analyses compared stu- ability increases condom use because the dents in New York schools after the program strongest study found a negative effect, the was put in place with a matched sample from Chicago schools; that is, no baseline data second study found a non-significant trend were collected or analyzed. Results revealed in the desired direction, and the third and that students in the New York schools were fourth studies found a significant positive not more likely to have initiated intercourse effect on condom use. Second, the differ- but were more likely to have used a condom ences in results could be caused by differ- the last time they had sex (Guttmacher et al., ences in the communities and in student 1997). needs. If youth already have ample access to condoms in their communities, as focus Finally, the fourth study measured the group data suggest they did in Seattle, then impact of making condoms available in nine making condoms available in schools may randomly selected Massachusetts schools. not increase condom use. By contrast, if Atilyses compared students in these schools communities do not provide condoms in with students in 50 randomly selected convenient and confidential or private loca- Massachusetts schools not making condoms tions, then making them available in schools available. Schools that made condoms avail- may increase student access to condoms and able apparently were also more likely to pro- subsequently increase use of condoms. (This vide their students with instruction about would suggest that when schools consider avoiding HIV. No baseline data were col- making condoms available, they should first lected that measured student sexual and con- conduct a student and community assess- dom behavior before the condoms were ment to determine whether condoms are made available in the schools. Thus, there may have been a variety of known and readily available in convenient and comfort- unknown differences between the two able locations for youth and whether making groups of schools before condoms were them available in schools would meet a real made available. Nevertheless, results indi- need.) Third, the differences in study results cated that students in schools with condom could be due, in part, to the addition of other programmatic components (e.g., edu- availability were less likely to have ever had. sex, were less likely to have had sex in the cational components and the availability of last three months, had about the same num- small group discussions or one-on-one ber of sexual partners in the last three counseling) in three of the studies. This is months but had fewer lifetime sexual part- consistent with the data above showing that ners, were more likely to use a condom the some sex and HIV education programs and

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 103 1I4-1 some brief interventions providing individ- those with few restrictions on condoms ual counseling increase condom use. and comfortable, easy access to condoms substantial proportions of sexually Conclusions about the Effects of experienced students did obtain contra- Programs Designed to Provide ceptives and condoms. Reproductive Health Care or to Improve However, given the relatively wide avail- Access to Condoms or Other ability of contraceptives in most commu- Contraceptives nities, school-based clinics that did not Studies of programs designed to pro- focus on pregnancy or STD prevention vide reproductive health services and to did not appear to markedly increase the increase access to contraceptives have pro- school-wide use of contraceptives that duced both consistent and inconsistent is, there appeared to be a substitution results. Nevertheless, several patterns and effect. Consistent with this finding, the conclusions emerge, including the following: provision of contraceptives through school-based clinics did not decrease the fiLarge numbers of young people do school-wide pregnancy or birth rates in obtain contraceptives from family plan- these sites. ning clinics and other providers, and O By contrast, when school-based or these contraceptives presumably prevent school-linked clinics not only provided many adolescent pregnancies each year. contraceptives but also focused more However, because the long-term impact intensely on contraception and gave a of family planning services on sexual clear message about abstinence and behavior is not known, the net effect of oral contraceptives, results suggested family planning services on pregnancy is that the programs did increase use of difficult to estimate. contraception. Brief clinic protocols and programs that Studies of school condom-availability provided youth with more information programs provide conflicting results that about abstinence, condoms, and/or may reflect methodological limitations, other forms of contraception; engaged differences in the overall availability of youth in one-on-one discussions about condoms in the communities, or differ- their own behavior; gave a clear message; ences in the programs themselves. and provided condoms or contraceptives typically did not increase sexual activity In sum, multiple studies of community but did consistently increase the use of clinics, school-based health clinics, and condoms and contraception. school-linked clinics that were devoted primarily (or solely) to reproductive According to many studies, providing health and that combined educational contraceptives in school-based health material (however modest), the opportu- centers and providing condoms in school nity for one-on-one counseling or discus- condom-availability programs consis- sions, a clear message about abstinence tently did not hasten the onset of sexual and condom or contraceptive use, and intercourse nor increase its frequency. actual condoms or contraceptives did O In schools with health clinics that pro- rather consistently find significant vided prescriptions or actually dispensed increases in condom or contraceptive use contraception and in schools with con- either among individual participants or dom-availability programs especially among students school-wide. Despite the

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 104 1 CI' methodological limitations of these stud- pamphlets describing how to use them. In ies and the variation among these pro- addition to these peer-led activities, there grams, the number of studies and the were radio and television public service consistency of their findings suggest that announcements and posters in local busi- this may be an effective combination of nesses and other public locations. In compar- components and that even brief interac- ison with youth in another city, males in the tions can have an beneficial impact. intervention community were less likely to initiate intercourse, and females in the inter- Community-Wide Pregnancy vention group were less likely to have multi- or HIV Prevention Initiatives ple partners. However, the program did not with Multiple Components appear to significantly affect other measures During the last two decades, there has of sexual activity or condom use. been a growing recognition that it might take more than just single programs focusing A second initiative, the RESPECT initia- on discrete populations of teens to change tive in Philadelphia discussed above (see p. teen pregnancy rates markedly. Thus, many 99), improved, clinic services and conducted communities have developed community- community-wide activities (Hughes, wide collaboratives or initiatives with the Furstenberg, & Teitler, 1995). The clinics goal of reducing teen pregnancy. conducted hundreds of school and community progams over a three-year period and Seven different stlidies have examined organized a two-year media campaign with the impact of community-wide programs posters and public transit cards. It was the (see Table 4.9). All of them measured impact subject of many radio programs and several on community-wide measures of sexual or newspaper articles. As noted above, the analy- contraceptive behavior or pregnancy or birth ses revealed no significant changes in contra- rates; they did not measure the impact on ceptive use or pregnancy or childbearing rates. those directly served. As noted above, this is the proper goal for a community-wide initia- A third study examined the impact of a tive, but it is a much more challenging and large, comprehensive social marketing cam- demanding one methodologically. These paign called Project Action (Polen & studies are discussed roughly in order of the Freeborn, 1995). Three public service intensity of the programs' interventions, announcements were aired multiple times on beginning with the least intensive. television, condom vending machines were installed in locations recommended by youth, The first study measured the effects of a comprehensive community-wide HIV educa- and teenagers were trained to facilitate small- tion program for adolescents a multi- group workshops that focused on decision- faceted intervention in New England making and assertiveness skills. Results designed to increase the use of condoms and indicated that the campaign did not increase reduce HW transmission (Sellers, McGraw, the proportion of higher-risk youth who had & McKinlay, 1994). Trained peer leaders ran ever had intercourse, nor did it increase their workshops in schools, community organiza- acquisition of condoms or their use of con- tions, and health centers; organized group doms with their main partners. However, discussions in the homes of youth; gave pre- after the campaign began, there was a signifi- sentations at large community events; con- cant increase in their use of condoms with ducted street corner and door-to-door casual sexual partners; after the campaign canvassing; and passed out condoms and ended, this use returned to baseline levels.

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy

1°51 ; 6 The fourth study (Alstead et al., 1999) The sixth community initiative the examined the impact of a social marketing most intensive was designed to reduce campaign that was very similar to the one teen pregnancy in a small, rural South described in the third study. A mass media Carolina community. The program included campaign targeted sexually active teens with the following components: teachers, adminis- pamphlets, posters, t-shirts, radio spots, bus trators, and community leaders were given signs, and billboards with a message to use training in sexuality education; sex education condoms. The campaign also arranged with was integrated into all grades in the schools; multiple agencies, organizations, and busi- peer counselors were trained; the school nesses to put 22 bins of condoms and 25 nurse counseled students, provided male students with condoms, and took female stu- condom vending machines in restrooms and dents to a nearby family planning clinic; and other locations recommended by youth. local media, churches, and other community Results indicated that it did not increase sex- organizations highlighted special events and ual activity nor did it increase condom use. reinforced the messages of avoiding unintended pregnancy (Koo et al., 1994; The fifth initiative was the multi-year Vincent, Clearie, & Schluchter, 1987). After Plain Talk program (Grossman & Pepper, the program was put in place, the pregnancy 1999). It was implemented in five communi- rate for 14- to 17-year-olds declined signifi- ties, three of which participated in the candy for several years. After parts of the impact evaluation. The initiative focused on program ended for example, some of the sexually active youth and strove to increase community efforts declined in intensity, the adult-youth communication about sex and school nurse resigned and her links to family contraception and to increase access to con- planning clinics and her distribution of contraceptive services. To do this, it launched a doms ended, some teachers left the school, variety of community activities to create a and more generally program momentum consensus among adults about the need to declined the pregnancy rate returned to protect sexually active youth by encouraging pre-program levels. From the existing data, it contraceptive use, and it provided parents is not clear which of the program compo- and other community adults with the knowl- nents or other unknown factors produced edge and skills to communicate more effec- the changes over time in pregnancy rates. tively with teens about sexual behavior and contraception. In one of the three communi- The final community initiative did not appear to be as intensive as the South ties, a clinic serving adults and teens was Carolina initiative, but it was a partial repli- opened; in the second community, an adoles- cation of that initiative in three communities cent clinic opened; and in the third, a clinic in Kansas (Paine-Andrews et al., 1999). The increased its hours for adolescents. Repro- initiative got the community to become ductive health information was also given to more involved in preventing teen pregnancy, youth at several community events. Survey enhanced sexuality education for teachers results revealed no significant changes in use and parents, improved age-appropriate K-12 of contraception at first or last sex. They also sex education, increased access to health ser- suggested no significant overall effects on vices, increased collaboration with school reported pregnancy.4 administrators, implemented mass media

4 The results between the baseline survey and the four-year post survey did indicate a significant decrease,in pregnancy rates among girls, but there was a non-significant increase reported by boys, and neither finding controlled for the decrease in nationwide teen pregnancy rates during this time period.

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 106 1 1 events, increased alternative activities for Programs that Focus Primarily youth, and got faith communities involved. on Non-Sexual Antecedents Despite the many activities, it is not clear how much the initiative really focused on Research clearly suggests that improv- sexual behavior and contraceptive use. ing young women's education, employment, Although there were a few favorable trends, and other life options reduces their preg- there were no consistent and significant nancy and birth rates. In many countries changes in sexual behavior, condom use, throughout the world, as young women's pregnancy rates, or birth rates. educational levels and employment opportunities increased, theinfertility rates declined. Conclusions about Multi-Component In this country, between the mid-1950s and Community-Wide Initiatives the mid-1970s, increasingly large percent- Although these community initiatives ages of young women pursued higher educa- and their impact differed considerably, their tion and more challenging professional results suggest several conclusions, some of careers and postponed marriage and child- them more tentative than others: bearing. During these years, the teen birth rate declined markedly (Alan Guttmacher O These results confirm that community initiatives focusing on pregnancy or con- Institute, 1994). And as noted in Chapter 2, dom use do not hasten or increase sexual there remains an important relationship activity, even when they focus primarily between educational and career plans and on condom or contraceptive use. adolescent pregnancy among today's adolescents. Observing these trends, some profes- O With regard to contraceptive use or sionals working with youth believe that one pregnancy or birth rates in general, these of the most promising approaches to reduc- studies are not particularly encouraging. The failure of most of these initiatives to ing teen pregnancy is to improve educational significantly improve adolescent sexual or and career opportunities through youth contraceptive behavior or to decrease development programs. Whereas the pro- pregnancy and birth rates may reflect the grams summarized in the first four groups fact that it is very challenging to improve above focused primarily on changing the sex- community-wide adolescent sexual or ual antecedents of adolescent sexual behavior contraceptive behavior. (e.g., the knowledge, attitudes, norms, and skills involving sexual and contraceptive O The most effective program was clearly the most intensive. However, when oth- behavior), the youth development programs ers attempted to replicate it, they did not included in this review focused primarily on achieve consistent positive effects. the non-sexual antecedents of adolescent sexual behavior (e.g., involvement with other O The effects of both Project Action and adults, attachment to school, educational the South Carolina program also suggest goals, and community eniployment opportu- that programs must be maintained if they are to continue to have an effect. In both ,nity). In other words, they were designed to cases, after the programs ended, use of improve the participants' education, life condoms or pregnancy rates returned to skills, and employment options. The first of pre-program levels. This finding is con- these groups of youth development pro- sistent with studies of marketing efforts grams was designed for very young children. more generally. The second was designed for adolescents.

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 107 1 1 8 Early Childhood Programs meeting the criteria of this review prevent any more definitive conclusions.5 Only one study meeting the criteria for this review evaluated an early childhood pro- Youth Development Programs gram in relation to teen pregnancy and child- for Adolescents bearing, the Abecedarian Project (Campbell, 1999) (Table 4.10). Infants in low-income Service Learning Programs families were randomly assigned either to participate in a full-time, year-around day care By definition, service learning programs program that focused on improving the chil- include (1) voluntary or unpaid service in the dren's intellectual and cognitive development community (e.g., tutoring, working as a or to receive whatever day care the families teacher's aide, working in nursing homes, or and communities could provide. Then, in ele- helping fix up parks and recreation areas) and mentary school, all the study participants (2) structured time for preparation and reflec- were again randomly assigned to receive tion before, during, and after service (e.g., either the normal school environment or a group discussions, journal writing, or papers). three-year program to involve parents and Often the service is voluntary, but sometimes improve parent-school communication about it is prearranged as part of a class. And often, the child. All youth were tracked until age but not always, the service is linked to aca- 21. Although the sample size was small demic instruction in the classroom. (n=104), the children who received the Service learning programs may have preschool program delayed childbearing by stronger evidence that they reduce actual more than a year in comparison with the con- teen pregnancy rates while youth are in the trol group. Notably, they also performed programs than any other type of interven- higher on a number of intellectual and aca- tion. Four different studies, three of which demic measures and received more years of evaluated programs in multiple locations, education than the control group. The chil- have consistently indicated that service learn- dren who participated in the elementary ing reduces either sexual activity or teen school program (as opposed to the pre- pregnancy (Allen,,Philliber, Herrling, & school day care program) also performed sig- Kupermanic, 1997; Melchior, 1998; nificantly better on all of these outcomes, but O'Donnell et al., 1999; O'Donnell et al., the results were not quite as strong. 2000; Philliber & Allen, 1992) (Table 4.11).

These results are encouraging; the The fifst study of a service learning pro- Abecedarian program's impact on educa- gram evaluated multiple sites using the Teen tional attainment may partially explain why Outreach Program (TOP) (Philliber & Allen, the program participants delayed childbear- 1992). It found that youth were less likely to ing. The results of this study also suggest report becoming pregnant during the school that other studies of early childhood pro- year in which they participated in TOP. grams should measure long-term impact on Because the comparison group consisted of teen pregnancy and childbearing. However, youth identified by participants as similar to the small sample size of the Abecedarian themselves, there was the potential for self- study and the fact that it is the only study selection effects. Consequently, a second

5 It should be noted that the results of the Abecedarian study are consistent with the pregnancy results for females of the High/Scope Perry Preschool Study, which measured the long-term impact of a high quality, active-learning preschool program. However, the High/Scope study is not sununarized in this review because of its small sample size for pregnancy results (n=49) (Schweinhart, Barnes, & Weikart, 1993).

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 108 r) study was completed, this time with an exper- but the addition of service learning did sig- imental design, including random assignment nificantly reduce sexual activity. In the short of youth to participate in TOP or not to par- term, it delayed the onset of sex while, in the ticipate (Allen, Philliber, Healing, & long term (more than three years later), it Kupermanic, 1997). Again, this study evalu- both delayed the onset of sex and reduced ated the impact of TOP in multiple sites the percentage of students who had sex the around the country. On the average, these previous month. These studies suggest that TOP participants sOent about 46 hours doing service learning may reduce teen pregnancy service. TOP participants again reported lower pregnancy rates during the school year rates in part by reducing sexual activity. in which they participated in TOP than did It is not known for sure why service the control group. It should also be noted that TOP participants also had lower rates of learning has positive effects on pregnancy, school failure than the control group. but several explanations have been suggested: participants developed on-going rela- A third study measured the impact of tionships with caring program facilitators; exemplary Learn and Serve programs some may have developed greater autonomy throughout the country (Melchior, 1998). and felt more competent in their relation- Students in these programs spent an average ships with peers and adults; some may have of 77 hours providing service. This study did been heartened by the realization that they not employ an experimental design with ran- could make a difference in the lives of others dom assignment, but it did identify similar students in other sChool classes or other all of which might have increased motiva- schools as a comparison group. Its results tion to avoid pregnancy. The volunteer expe- tended to confirm the TOP results in that riences also encouraged youths to think participants in the Learn and Serve programs more about their futures. It may also be that reported lower pregnancy rates during the both supervision and alternative activities school year in which they participated. simply reduced the opportunity for partici- However, the result was not quite significant pants to engage in problem behaviors, (p=.10). Notably, this study also evaluated including unprotected sex. After all, these the longer term impact of participation in programs were time intensive the average Learn and Serve and found that the impact number of hours that youth spent in TOP on pregnancy (and also on most other out- and Learn and Serve programs during the comes) did not last through the school year following the year of participation. This sug- academic year were 46 hours and 77 hours gests that participation in service learning respectively. The study of TOP found that programs may reduce teen pregnancy rates the kinds of volunteer service varied consid- only during the semesters in which youth erably from site to site, but TOP appeared to actually participate. be most effective when young people had some control over where they volunteered Finally, a pair of studies measured the (Allen, Philliber, Healing, & Kupermanic, impact of a health education curriculum 1997). The effectiveness of TOP was not alone and the combined impact of the same dependent upon the fidelity of the imple- health education curriculum and service learning (O'Donnell et al., 1999; O'Donnell mentation of the TOP curriculum (Allen, et al., 2000). Results indicated that the Philliber, & Hoggson, 1990), which suggests health education curriculum alone did not that the service itself is the most important significantly decrease recent sexual activity, component of the programs.

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 109 I Vocational Education and Employment vantaged youth, only JOBSTART was Programs restricted to school dropouts. The Job Corps was mostly residential; JOBSTART and the Vocational education and employment Conservation and Youth Service Corps were programs typically include academic instruc- for the most part not residential. All three tion (or an educational requirement) and studies incorporated strong experimental either vocational education or actual jobs. designs with random assignment of individ- Four different studies evaluated such pro- ual youth, large sample sizes, long-term mea- grams, all in multiple sites (Table 4.12). surement (15-48 months), and measurement The first study evaluated the impact of of either pregnancy or childbearing (but not the Summer Training and Education sexual behavior). Results from these three Program (STEP) (Grossman & Sipe, 1992; studies demonstrate that the programs did not affect overall pregnancy or birth rates. Walker & Vilella-Velez, 1992). Youth in the There was no significant impact of the treatment group received 90 hours of aca- Conservation and Youth Service Corps on demic remediation and half-time summer unmarried pregnancy rates measured 15 employment. They also received other sup- months later; no impact ofJob Corps on port services, including 36 sessions of life l;ir-th rates measured 30 months later; andno skills education and 5-15 hours of other sup- impact ofJOBSTART on pregnancy or birth port during the school years. By contrast, the rates measured 48 months later among control group received full-time summer women not residing with any of their own employment. In other words, in comparison children at baseline. There were, however, with the control group, the intervention two small exceptions to these findings that group received less summer employment but showed no effects. Among African-American did receive the academic remediation, life women in the Conservation and Youth skills education, and personal support. A very Service Corps study, participants were signifi- strong experimental design revealed that the cantly less likely to experience an unmarried program did not have a consistent and signif- pregnancy than the control group, and, in icant impact on either sexual activity or use the JOBSTART study, among teen mothers of contraception. residing with their children, participants were more likely to experience a pregnancy or The remaining three programs, the birth than non-participants. Overall, these Conservation and Youth Service Corps, the studies provide rather strOng evidence that Job Corps, and JOBSTART, were all imple- these programs did not have significant mented in the late 1980s or 1990s and were positive overall effects on pregnancy of targeted toward somewhat older youth. In childbearing. the Job Corps and JOBSTART programs, about 73 percent of the participants were 16- to 19-years-old; the other 27 percent Other Youth Development Programs were older. These programs combined reme- Two other studies have examined the dial, academic, and vocational education impact of youth development programs for (Cave et al., 1993; Jastrab et al., 1997; adolescents. The programs differed consider- Schochet, Burghardt, & Glazerman, 2000). ably from each other (see Table 4.13). The To varying degrees, these programs also pro- first study evaluated the impact of a very vided other support services, including life comprehensive program called the Quantum skills education, health education, health Opportunities Program (Hahn, 1994), which care, child care, and job placement assistance. was implemented among high school stu- While all three programs focused on disad- dents from families receiving public assis-

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 110 I tance. The program included educational Conclusions about the Effects ofYouth activities (e.g., tutoring arid computer-based Development Programs instruction), community service activities, All these studies of different types of and development activities (e.g., arts and youth development programs, for either career and college planning). Thus, it focused on academic achievement but gave small children or adolescents, support several considerable attention to social competence conclusions: as well. Participants received small stipends Cl Service learning, which combines and bonus payments for participation and community service with reflection on completion of activities and matching funds those experiences, appears to reduce teen for approved activities after the school day. pregnancy during the academic year in Although the evaluation had a strong which youth complete the service. It research design, its sample size was small appears to be effective even without (n=156 for analyses of birth rates), and there are many questions about the quality of the addressing sexuality directly. actual program implementation. Results sug- Vocational education programs that gested that the program participants had a include academic remediation, vocational lower birth rate than the control group, but education, and a few support services do this result was not quite significant (p=.10). not significantly reduce teen pregnancy or birth rates in the long run. The second study evaluated the impact of the Seattle Social Development Program oOther youth development programs, (Hawkins et al., 1999), Which was designed such as strong preschool childcare pro- to increase children's attachment to school grams, programs to improve the quality and family by improving teaching strategies of teaching in elementary school and (e.g., cooperative learning) and parenting student attachment to school, and very skills. To improve the quality of teaching in comprehensive and intensive youth schools, the program provided five days of development programs, have produced in-service training for teachers of grades 1-6 a few encouraging results, but there are each year. Many teachers participated in this. too few studies and too many important To improve parenting skills, the program study limitations to reach any offered parenting classes for parents of chil- conclusions. dren in grades 1-3 and 5-6; relatively few parents attended. The program also imple- O At this point, it is simply not clear why mented a curriculum unit in schools to some youth development programs (i.e., increase students' social skills (e.g., decision- service learning) reduce teen pregnancy making and refusal skills). When these grade and others with some similar characteris- school students were followed to age 18, tics (i.e., vocational education programs) those receiving the intervention were less do not. This is an important area for likely to report a pregnancy than the com- further research. parison group. They were also more attached to school, got higher grades, and engaged in fewer delinquent acts.

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy

III 4" 4., Programs that Focus on Both The second study evaluated different Sexual and Non-Sexual programs in 44 different sites in California Antecedents (East & Kiernan, 2000). Their unusual and common quality was that they all targeted Multi-Component Programs with girls at high risk of becoming pregnant Both Sexuality and Youth because their sisters had become pregnant as Development Components teenagers. The primary goals of these programs were to delay sex, increase contracep- A number of studies have examined tive use, and decrease risky behaviors programs that focused on both sexual and associated with teen pregnancy (e.g., drink- non-sexual antecedents of teen pregnancy ing and drug use). To achieve these goals, that is, they had components addressing programs were designed to help youth both sexuality and youth development more remain in school or return to school, broadly. The first two studies evaluated sev- increase self-esteem, improve knowledge and eral different types of programs, all of which skills to make healthy decisions, improve were relatively modest in intensity and dura- access to health and reproductive health tion, and the third study evaluated a very services, and increase communication with intensive, long-term program. parents and adults. The sites used multiple strategies, including both individual case The first study evaluated the impact of management and group activities and ser- three programs in Washington state vices, but the activity that occupied the (McBride & Gienapp, 2000) (Table 4.14). greatest amount of time was recreation. All three programs started with the premise Youth also spent a large proportion of their that many adolescents, especially high-risk time in small group activities focusing on adolescents, have a variety of emotional various topics. In general, these were not needs and problems that affect their sexual intensive youth development programs; the behavior. These three programs implemented mean number of hours of participation in a "client-centered" approach that was based these programs was less than 19, although on the service providers' understanding of they varied from less than one hour to more why the teens they were working with were than 95 hours. While the evaluation was lim- involved in sexual risk-taking. The service ited by lack of random assignment and possi- providers generally believed that their teen ble selection effects, the results indicated that clients lacked one or more of the following: the interventions both delayed sex and decreased reported pregnancy rates nine (1) information about sex, (2) a variety of months later, but they did not significantly coping skills, (3) emotional support, (4) pos- increase contraceptive use. itive guidance, and (5) adults they could trust and talk to about sensitive issues. The The third program that addressed both programs tried to address these problems sexual and non-sexual antecedents was the and others by providing small group and Children's Aid Society-Carrera Program individualized education and skill-building (CAS-Carrera Program) (Philliber et al., sessions, as well as several other individual- 2000). It was a long-term and intensive pro- ized services, such as counseling, mentoring, gram that recruited youth when they were referrals, and advocacy, that were tailored for about 13- to 15-years-old and encouraged each teen. Results indicated that the program them to participate throughout high school. did not delay sex nor increase contraceptive During those school years, it operated 5 days use, but did decrease the frequency of sex. a week. Some programs had regularly sched-

THE NATIONAL, CAMPAIGN TO PREVENT TEEN PREGNANCY 112 123 uled special events, education programs, and sex than boys in the control group. This was entrepreneurial activities. During the summer found among males who had initiated sexual months, paid employment, including activity prior to the onset of the program. entrepreneurial activities, were emphasized, Notably, these findings are reported for all along with evening maintenance programs. the members of the treatment and control Participants spent an average of 16 hours per groups, even though some members of the month in the program during the first three treatment group (especially the boys) did not years; many spent more time in the program. participate extensively in the program, and The CAS-Carrera Program used a holistic some members of the control group received approach, providing multiple services: (1) a few services from the same organization or family life and sex education, (2) an educa- other organizations in the community. tion component that included individual academic assessment, tutoring, help with This study of the CAS-Carrera homework, preparation for standardized Program is the first and only evaluation to exams, and assistance with college entrance, date using random assignment, multiple (3) a work-related intervention that included sites, and a large sample size that found a a job club, stipends, individual bank positive impact on sexual and contraceptive accounts, employment, and career awareness, behavior and pregnancy and birth rates (4) self-expression through the arts, and (5) among girls for as long as three years. In individual sports. In addition, the program fact, the pregnancy rate among girls in the provided mental health care and comprehen- intervention group was less than halfthe rate sive medical care, including reproductive among the control group (10 percent v. 22 health and contraception when needed. In all percent). These are strong, very important these areas, staff tried to create close caring results. It should also be recognized that this relationships with the participants. Although is a complex program to implement, requir- the program focused on youth, it also pro- ing significant financial and staff resources, vided services for the participants' parents and sites that do not implement all the com- and other adults in the community. ponents or that do not fully engage young people over time cannot expect to achieve The evaluation study of the CAS- these positive results. Carrera Program was a very rigorous one. It included multiple sites, random assignment, Why was the CAS-Carrera Program a large sample size, long-term measurement, successful where other programs with some measurement of behavior, and proper statisti- similar components (JOBSTART, Job Corps, cal analyses. The study found that, after three and the Conservation and Youth Service years and among girls, the program signifi- Corps) were not? There are several possible cantly delayed the onset of sexual inter- explanations: (1) the vocational education course, increased the use of condoms as a programs targeted somewhat older youth secondary method with another highly effec- and tracked them into their early twenties tive methods of contraception, reduced preg- when childbearing is more normative and nancy rates, and reduced birth rates. Among less costly; (2) the CAS-Carrera Program males, the program did not have significant included a strong sexuality education compo- positive behavioral effects, but the study did nent in combination with improved access to have one unexpected finding males in the reproductive health and other health ser- programs were significantly less likely to vices, as well as the intensive youth develop- report using both condoms and another ment components; and (3) staff in the highly effective contraception method at last CAS-Carrera Program may have developed

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 113

4.4 closer relationships with the young people pregnancy. The ability of some programs they served. to have success in reducing both sexual risk-taking and even pregnancy suggests Conclusions about the Effects of that further research is needed to better Programs that Address Both Sexual and define the qualities of effective programs. Non-sexual Antecedents 0Intensive long-term programs, like the Although there are only three studies CAS-Carrera Program, that have multi- in this group, they support the following ple components addressing both the conclusions: reproductive health needs and the other 0Modest interventions that address both emotional and social needs of youth can sexual and non-sexual antecedents may have a substantial long-term impact upon reduce adolescent sexual risk-taking and pregnancy among girls.

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 114 1 C lapter Motes List of Criteria for Selection of Studies

The study was completed during or after intervention schools or communities and 1980. well-matched comparison schools or com- The study was conducted in the United munities with statistical adjustments for States or Canada. baseline differences, or another appropriate and equally valid quasi-experimental The study primarily targeted adolescents design. of high school age or younger (roughly 18 or younger) who were not yet O Whether each study participant was in the parents. intervention or comparison group must be known independently by the researchers; O The intervention was implemented with reasonable integrity or fidelity. it cannot be based solely on the recall of the respondents one or two years after O The sample size was at least 100 in the participation. combined treatment and control group. This minimum number applies to the O The study must have measured impact on actual statistical analysis measuring impact actual sexual behavior, i.e., measures of on behavior. sexual intercourse such as the timing of initiation of intercourse, condom O The same method of parental consent was or contraceptive use, pregnancy, or child- used with both the intervention and comparison/baseline groups. bearing. The study used an appropriate and valid O When the study measured impact on the experimental or quasi-experimental design. initiation of sex, then it must have fol- For example, if the intervention targeted lowed youth for at least six months after either individual youth or classrooms of the intervention in order to allow suffi- youth, then at a minimum the study must cient time for fewer youth in the interven- have used a quasi-experimental design tion group than in the comparison group with both intervention and comparison to initiate sex. When the study measured groups and baseline and follow-up data. impact on the frequency of sex or use of If the intervention targeted entire schools condoms or contraception, then the study or entire communities, then at a minimum must have measured impact for at least 2 the study must have used a quasi- months after the intervention or 4 months experimental design with school-wide or after baseline, whichever was shorter. community-wide time-series data (at least two points in time), or alternatively a O The study employed proper statistical quasi-experimental design with numerous analyses.

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 1 1541 T ble 4. hStudy Studies Information of Abstinence Sample Description 11) gr rns Study Results Publication DateProgram(s)Author(s) / / I SampleSESLocation (N) / Post / GenderEthnicity Grade/ / Age / Program Description Design MethodsAnalytic Change in Outcome' CommentsAdditional involvement/ENABLPostponing Sexual CaliforniathroughoutDispersed 7th 12-13and 8th years design.communitySetting: Classrooms organizations in most in one designs; RandomExperimental. assignment of entire comparisoninterventiont-tests between and FrequencyInitiation of ofintercourse: sex in previous 0 assignment,rigorous;The evaluation it had large wasrandom Sample very CagamPangKirby,I 995 Korpi, Barth, VariedN=7,753 SES F=58%M=42%graders ContentSessions: Designed 5 I-hour sessions both to help adult-taughtdentsyouths.schools, were In classrooms, part randomlyPSI, of peer-taughtthe or study,assigned individual stu- PSI, to or changegroups usingscores. 12Frequency3 months:0 months:0 of sex in previous examinedstatisticalup,sizes, and long-term appropriate analyses. the impact follow- It also of B1=9%His=31%Wh=38% uponandpressuresyouth apply postponingunderstand resistanceto have sexual socialsex skills; and andinvolvement emphasis to peerdevelop collectedMatcheda control questionnaireatgroup. baseline, 3 anddata 17 were UseNumber of condoms:0 of sexual partners: 0 rooms,tings,eitherPSI implemented individual community or entire class- schools. in set- basedMethods:Taught on social influence by adults theory. or teens. ComparisonmonthsInterVention post-intervention. post-test post-test N=4,056 N=3,697 Pregnancy:Use of birth control pills:6 AdultTeen led:led: Teen-Aid,Sex Respect, Utah .Mean=15.5years Setting: School classrooms Quasi-experimental. suresRepeated analysis mea- of HighInitiation school: of sex: designThe strength was weakened of this by PrigmoreWeed,Values Olsen, and Choices DeGaston, N=1,963Mixed SES schoolseniorMiddle highand uponContenttwoSessions:Not to abstinence three All threeweeks. reported; (Values curricula appearand focusedChoices to be 3grams;highThree matched schools high 2 matched juniorschools implemented high high and schools schools 5 junior the pro-and relatedablestrollingcovariance, that forto were inter-vari- con- Overall:MiddleMost 0permissive permissive values values group: + bilityshowassignment,the lack betweenbaseline of random failure compara-interven- to 1992 Wh=90%F=51%M=49% uations.taughtonlywas andedited skills ItUtah focused to to guidelines).conform avoid on sex self-control, to in abstinence-Sex difficult Respect self- sit- latercollectedMatchedserved (at as the questionnaireat comparison baseline,end of each 3 groups. datato program), 4 weekswere Loglinearmeasure.andvention outcome group mod- HighLeast futurepermissive orientation values group: 0 analyses.sizesgroups,tion and for and comparison someBehavior small sub-group sample data Oth=7%His=3% thatteachesTeen-Aidrespect, abstinence decision-makingand provides respect is the information for best others.it emphasizeschoice. and It ComparisonandIntervention one year group:later. group: N=756 N=1,207 sex.transitionels used for to Low future orientation group: 0+ mostAmongcohortwere permissivebased data youth only.upon with values, first the abstinencesexualityprovidesIt also anda covers broaderfrom covers drugs, other understanding dating healthalcohol, standards. issues, and of e.g., unior high school: Overall:Most 0permissive values group: 0 threetionthe differencesrates groups among were in theinitia- not nencemaking.informationtobacco.Values as It thegives and only lessand promotes correct Choicesemphasis choice. decision- provides to absti- LeastMiddle permissive permissive values group:values 0group: 0 samplebeensignificantThis due size. to insufficient may have ) 1 Change in Note:For all studies, the significance level was set at p < 0.05.1f authors reported significant findings at the p < 0.10 level, they wereoutcome treated for group receiving intervention: no significant change = 0; significant desirable change = +; significant undesirable change = as insignificant in this table. Table 4.1: StudiesStudy of Information Abstinence Programs continued Sample Description Study Results I 1 Age / Publication DateProgram(s)Author(s) / / 1I Sample (N) SESLocation / Post / GenderEthnicityGrade / / Program Description Design MethodsAnalytic Change in Outcome' CommentsAdditional Stay SMART i Mostly in . Mean=13.6 Setting: Boys and Girls Clubs of I Quasi-experimental. Repeated mea- Virgins: I For non-virgins, results St. Pierre, Mark, Kaltreider, I throughouturban areas years I America I Fourteen clubs were assigned to 3 usedsures toANCOVA control At 3 months: Recency of last I SMART withoutwere the inconsistent. Stay Aikin the U.S. Not reported Sessions: 12 I groups: comparison group, which for the pre-test' intercourse: 0 I booster appeared to 1995 Low SES F=25%M=75% delayContent sex Multi-focus: and prevent Designedalcohol, to 1 tionreceived group, nothing; which receivedthe first interven-Stay outcomemeasure ofvari- the Frequency of intercourse:0 I intercoursereduce frequency at 27 of I N=273 Wh=45%131=42% personalcigarette, andand socialmarijuana competence use. Based on I SMART without thewhichthe booster; second received and intervention Stay SMART group, and andable, ethnicity. gender, age, At 15 months: Combined measure: 0 I tobooster,months.With reduce it frequency. did thenot appear His=14% ofmodel social of influence prevention theory). (broader Included version 9 the boosters. There were few 1 Recency of last I These inconsistent sessions on life skills training (general If youths did not participate in encessignificant at base- differ- Frequency of intercourse: intercourse: 0 I of randomresults, assignment, coupled with lack tivecoping peer skills influences) and skills and to resist3 on post-nega- I droppedmost of the from sessions, the intervention they were line; none on 0 I small sample sizes, very behavior out- Combined measure:0 I high attrition rates, and sexponing in media, sexual lines involvement to have sex,(discussed and groups. comes. At 27 months: I failure to adjust for clus- Matched questionnaire data were I tering effect at the club consequencesplaying). of sex and did role I collected at baseline, 3 months sesSeparate for virgins analy- Recency of last intercourse: 0 I level render these results andMethods:A a 4.5-hour 5-session 2-year boosterI-year booster were I 3-monthmonthslater, 15 later. months post-test later, and 27 pre-test.asand measured non-virgins at CombinedFrequency measure:0of intercourse: 0, + inconclusive. Taughtpositiveknowledgedesigned by role to staff and reinforcemodels. members.Youth to help the older skills youth and volun- be Comparison:Stay SMART:SMART N=83+N=I09 booster: N=8I AtNon-virgins: 3 months: Combined measure:0 teered to participate. At 15At months:27 months: Combined measure:measure:0 0 ) 1 Change in Note: Fo'r all studies, the significance level was set at p < 0.05. If authors reported significant findings at the p < 0.10 level, they129outcome for group receiving intervention: no significant change = 0; significant desirable change = +; significant undesirable change = were treated as insignificant in this table. 130 Tabie,11 4.2:StudiesStudy Information of Sex Education Progr ms Sample Description Study Results Publication DateProgram(s)Author(s) / / SampleSESLocation (N) / Post / GenderEthnicity Grade/ / Age / Program Description Design MethodsAnalytic Change in Outcome' CommentsAdditional Urban area Mean=11.5 1, Experimental. Repeated mea- In general, this was a RespectDraw the the Line/ Line I Californiain northern years Sessions:Setting: Middle 20 sessions: schools 5 in the 6th I Nineteen middle schools randomly modelssures logistic for At endInitiation of 7th grade:of intercourse: Boys: + veryrandom rigorous assignment study with of Coyle, Kirby, Marin, 1 6th grade grade, 8 in the 7th grade, and 7 in the : assigned to intervention and con- males and Girls: 0 schools, large sample 2000Gomez Gregorich N=2,829Mixed SES F=50%M=50% quencesContent8th grade. ofTopics unplanned included sex (pregnancy,conse- 1 trol nancyreceivedconditions. prevention usual Control STD, education. HIV, schools and preg- variablelingratelyfemales for control- the sepa-that one At end of 8th grade: Girls:Boys: +0 analysis.andsurementsizes, proper However, long-termof statistical behavior, the mea- B1=5%Asn=16%Wh=17%His=59% ulationsocialsetting,STDs, learningand andtheoryActivities HIV), refusal theory personal skills. and wereBased sexualsocial highly on inoc-limit gradecollectedinMatched thebefore springin thequestionnaire the springof intervention the of7th the dataand 6thand 8thwere twobetweenwas groupsdifferent the and At endHad of sex 7th ingrade: last 12 months: Girls:Boys: +0 monthssignificanceupon depended sex of in the the impactupon last 12 Oth=3 interactive. grades. . tionrelated of sex. to initia- At end of 8th grade: Girls:Boys: 0+ analysisthe type used. of statistical Health for Life Project I Wisconsin 1 6th grade Setting: Middle schools Experimental. sionLogistic was regres- used Initiation of intercourse: At grade 8: 0 i" very rigorous studyIn general, with this was a Moberg, Piper 1 Mixed and I M=48% Sessions: 58 total; 16 on sexuality Twenty-one middle schools ran- 1998 middle SES F=52% Content Comprehensive health cur- Ii controldomly assigned conditions, to interventionstratifying for and suresbaselineto control of sub-mea- for At grade 10:9: sizes,schools,random long-term large assignment sample of N=1,98I 1 W=96% Oth=4% Baseduse,alcoholriculum nutrition, on use,designed social tobacco and influences tosexual positivelyuse, behavior.marijuana model affect and chooseinterventionsubstance between use. condition an Schools age-appropri- asSignedcould to involvementnon-intercoursestance use, Sex last month: At grade 9:8: 0 analysis.ior,measurement and However, proper of itstatistical behav-was Skill-based.principles of Peer, adolescent parent, development.and commu- 1 ate intensiveversion tiught version in gradestaught in 6-8, grade or 7. sexwith andopposite to Consistent condom use: At grade 10: whethernot possible students to had measure parentsvalues,socialnity components.Topics: situations, media,and the communication oppositerefusal skills, understandingsex, parental respon-with Matchedprevention-orientedControl schools questionnaire received curricula. data other were Adjustmentsdesign.ANCOVAmimic an for At gradeAt grade I 0: 09:08:0 ever had sex at baseline. 3 1 controlsexualsibility foractivity, information. health sexuality behaviors, facts, risks and of birth 8th,collected 9th, and in the10th fall grades. of the 6th, 7th, I clustering. 13 2 ) 1 Change in Note: For all studies, the significance level was set at p < 0.05. If authors reported significant findings at the p < 0.10 level, they outcome for group receiving intervention: no significant change = 0; significant desirable change = +; significant undesirable cl.lrige = were treated as insignificant in this table. Table 4.2:Studies of Sex Education Programs continued 1 Study Information Program(s) / SESLocation / PostSample / Description Grade / Age / Study Analytic Results Additional McMaster Teen Publication Date Author(s) / SampleOntario, (N) Mean=12.7 GenderEthnicity / Setting Health classes in junior high Program Description Experimental. Design Logistic regres- Methods Initiation of intercourse:Change in Outcome' The content of the treat- Comments Program Canada years schools. Students were divided into 1 sion was used At 3 months: 0 ment intervention did rn small groups of 6-8 students each. 1 Twenty-one schools randomly to control for At 1 year:0 not appear much -4CD Goldsmith,Thomas, Mitchell, Singer,Watters Devlin, reportedNot graders7th and 8th Sessions: 10 conditions.assigned to treatment and control size,school, gender, school and At 4 years: 0 1 controlstronger intervention. than that of the 1 any other vari- Contraceptive use: 1 992 N=2,570 M=48% Content Adolescent development, 1 Ten control schools received the ables related to Consistent use The evaluation was rig- Mitchel-DiCenso,Thomas, F=52% bilitypeer pressure,in relationships, gender stages roles, responsi-of inti- 1 riculum.conventional sex education cur- the outcome At 3 months:0 assignmentorous; it had of random schools, z Devlin, Goldsmith,Willan, Not reported 1 variables. At 1 year:0 macy, adolescent pregnancy and Males: + 1 large sample sizes, short- Singer, Marks,Watters, childbearing, and decision-making. 1 Matched questionnaire data were Treatment English as Female: 0 1 and long-term follow-up, Hewson primary 1 collected at baseline, 3-month group waS At 4 years: 0 1 low dropout rates, and ri)(I) 1997 language=75% films,sion,Methods:Tutors question-and-answerand role-plays. used group periods, discus- 1 periodspost-intervention, at I-year intervals. and 4 follow-up 'test itsadded effect last to Pregnancy rates: analyses.appropriate statistical At 3 months: 0 1 D- 1 Intervention post-test N=1,593 Both overall and 1 However, despite the 1 Comparison post-test separate analy- At I year: 0 -n N= 977 ses by gender. At 4 years: 0 1 random assignment of cro 5 1 baselineinterventionschools, maleswere group morein the at likely .43 cr() 70 controlcourseto have groupthan initiated males at baseline. inter- in the a 1 1 Postponing Sexual Atlanta, GA 13-14 years Setting Regular classrooms Quasi-experimental. t-tests between 1 Initiation of intercourse: A limitation of the study 3 Involvement (PSI) and intervention and I. End 8th grade: + was the fact that inter- 0 Human Sexuality Low SES 8th graders Sessions:PSI=5 hr. 10 1 receivedIntervention:one PSI and Humanschool Sexuality.district groupscomparison at pre- End 12th9th grade: grade: + 0 groupsvention livedand comparison in different Howard,1990 McCabe N=536 F=66%M=34% ContentHuman PSI:Designed Sexuality=5 hr. both to help 1 trictsComparison:3 received existingsmaller programs.school dis- andInitial post-test. equiva- InexperiencedFrequency of at intercourse: pre-test backgroundgeographic areas. characteris- Some (pH B1=99% youth understand social and peer 1 lence of inter- but initiating by follow-up: tics were controlled CD 1 Telephone interviews were con- through sampling proce- 7:17 uponandpressures apply postponing resistanceto have sexual sex skills; and involvement emphasis to develop ; grades.ducted in the 86,9th, and 12th .ventioncomparison / End 9th12th grade: grade:0 + impactdures to of reduce this difference. the CrCI a)0 basedHuman on social Sexuality:5 influence basic theory. sessions on 1 wereSample patients = children at a publicof parents hospital. who t-tests.established with ContraceptiveExperienced use:at pre-test 0 tiesOther that events might or have activi- 7 human sexual4 decision-making, and 1 Inexperienced at pre-test affected either the inter- contraceptives. 1 Matched questionnaire data were but initiating by follow-up: vention group or com- Methods:Taught by peer leaders 1 intervention.collected at baseline and post- End 9th126 grade: grade: + 0 wereparison not group controlled. differently (11th and 12th graders). 1 Experienced at pre-test 1 ComparisonIntervention post-testpost-test N=395N=I41 End 1269th grade:0grade: 0 ) 1 Change in Note: For all studies, the significance level was set at p < 0.05. If authors reported significant findings at the p < 0.10 level, theyoutcome for1 group 3 3receiving intervention: no significant change = 0; significant desirable change = +; significant undesirable change = were treated as insignificant in this table. 134 Table 4.2:Studies of Sex Education Programs continued Study Information Sample Description , Study Results 1: Age / Publication DateProgram(s)Author(s) / / SESLocation / Post / 1 Gender Grade/ / Program Description Design MethodsAnalytic Change in Outcome' CommentsAdditional Sample (N) 1 Ethnicity ! I InvolvementPostponing Sexual (PSI) and NYCortland, I YearsMean=14.0 Setting: Regular classrooms I Quasi-experimental. interventiont-tests between and Ever had consenting sex:0 evaluationThis was not design. a strong Pre- 1 Sessions:4 Intervention: 1 school received PSI. comparison Frequency of intercourse in and post-questionnaires Little,(Adapted)Human Rankin Sexuality N=271reportedNot Not8th graders reported youthpressuresContent: understand PSI:Designed to have social sex and and both peer to develop to help I PSIreceivedComparison:3 or other existing peer smaller programs, programs. schools but not andgroups post-tests. at pre- theFrequency last 12month: months:of intercourse 0 0 in tiontherewere among notwas matched greater the interven- andattri- 2001 Wh=88% and apply resistance skills; emphasis 1 lenceInitial equiva-of tion group. Oth=12% 1 Unmatched quesionnaire data Number of sexual partners: 0 abuseIncludedbasedupon information. postponing on media social analysis influence sexual andinvolvement; theory substance I weremonths, collected and at 6 baseline,months. 2 establishedcomparisonintervention/ behavioralprogrammightModest have sample impact, resultsobscured sizes but were no (10thMethods:Taught graders). by peer leaders 1 ComparisonIntervention 6-month:6-month: N=166N=105 with t-tests. close to significance. Mean=12.8 Experimental. Logistic regres- Initiation of sex: I Although 6 schools were HealthHumanInvolvementPostponing Screening Sexuality, Sexual (PSI), and LowDCWashington, SES 7thyears graders healthonContent:HumanSetting:Schools. reproductive professionals. health sexuality:3 PSI: taught 5 sessions sessions by ventionthenSix schools andrandomly control were assigned conditions. paired andto inter- the groupsbetweenforsion differences controlling of the 2 Girls: EndBeginning of 8th:7th: of+ 8th:0 thestantiallyofrandomized, this failure design weakened to the wascollect strength sub- by Williams,Aarons,Khorazaty,Woodward, Jenkins, Clar,Wingrove Raine, El- N=2,099 I1 M=48% B1=83%F=52% standDesignedtaught social by 10thboth and andto peer help Ilth pressures youth graders. under- to wereUnmatched collected duringquestionnaire the middle data ratelyconductedschools, for and each sepa- Boys: End of 7th:0 ingtionnairematched and relocation data,pre-post the ques-clos-of one 2000 Oth=2%His=13% socialponingresistancehave sex influence sexual and skills; toinvolvement; theory. developemphasis Included and uponbased apply mediapost- on 8ththe ofgrade.8th the grade, 7th gradegrade, and the (baseline), the end beginning of the the endof gender. sex:Contraceptive use at last EndBeginning of 8th: of0 8th:0 clusteringthecompositionschool, failure changes in to schools,of adjust schools, in ethnic for and viewsassessment:Studentstion.analysishealth Health were and risk screening conductedsubstance assessment. completedand abusewith individualIndividual youth informa- a with inter- AnalBaseline 8th grade7th grade followup: survey:N=522 N=422 Girls: EndBeginning of 8th:7th: of + 8th: + behavior.ferencescontrolapparently for in baselinethe the failure outcome dif- to groupities.healthproblems sessions:8 topics. or risk Othersessions behaviors. school-wide on different Small activ- Boys: EndBeginning of 7th:8th:0 of 0 8th: 0 136 135 ) 1 Change in outcome for group receivingNote:For intervention: all studies, no thesignificant significance change level = 0;was significant set at p

Wise Guys Greensboro, 12 or Setting: Schools. Quasi-experimental. Not reported. Initiation of sex:0 I This study had a very Gottsegen, Philliber NC I 3=74% Sessions:8 class sessions, 1 per week earlyPrincipals sexual identified involvement males in at both risk of Contraceptive use at last sex: I groupsweak design.The did not appear two Unpublished (2000) youthHigh-risk M=I00%7th=76% reproduction,Content Provides contraception information and about STDs; the intervention and comparison Frequency of contraceptive I verybaseline similar atand baseline, follow-up N=335 Wh=63% sexuality,seeks to develop responsibility, healthy respect values for about schools.Unmatched questionnaire data col- use: + I notquestionnaire matched, sample data were Oth=37% Methods:cationwomen, skills. resistance A yariety skills, of active and communi-learning Baseline6leted months at baseline, intervention: later. 8 weeks N=94 later, and i resultsweakstatisticalsizes were(only were analysis small,significant presented andwas the methods, including role playing. Baseline comparison: N=55 !I didin the not tables appear and to results con- I groups).betweentrol for differences the two ) 1 Change in outcome for group receiving intervention: no significant change = 0; significant desirable change = +; significant undesirable change = Note: For all studies, the significance level was set at p < 0.05. If authors reported significant findings at the p < 0.1 0 level, 149 they were treated as insignificant in this table. 1 Table 4.3:Study Studies Information of HOWAIDS Education Programs Sample Description I Study Results Publication Date(s) Program(s)Author(s) / / SampleSES (N)Location / Post / GenderEthnicity Grade/ / Age / Program Description Design MethodsAnalytic Change in Outcome' CommentsAdditional AdolescentsAIDS Prevention in School for NYNew York, 9thNot and reported I I th I Sessions:Setting: General 6 education classes. FourQuasi-experimental. high schools were divided interventiont-tests between and I Abstinence:0Intercourse with high-risk 'IdidIThe not 3-month allow measure- post-test triX Walter,Vaughan Low SES graders 1 Content Health belief model, social eachinto two pair, matched one school pairs. provided Within 9th groupscomparison using Ipartner: + I effects.ment of Changes long-term in 1993 N=867 F=58%M=42% ence:focuscognitive theory, upon modelcorrecting of social facts influ- grade comparisonprogram classes classes, and while11 th Multiplechange scores. regres- IIMonogamous relationships: + behavior were modest 0 Wh His=35%BI=37%=13% resources,ingappraiseabout knowledge AIDS, risk changing teachingof of transmission, A1DS-prevention perceptions cognitive increas- skills of to gradethe programcomparison other school classes. classes provided and 9th Ilth characteristicsbackgroundto controlsion was for used Consistent condom use: + c-) Asn=11%Othr=4% standingiors,frequency clarifying external of peer personal influences, risk-taking values, and behav- under- AIDSComparison prevention classes program. received no . scores.and baseline r-,Co 0 H and/orteaching consistently skills to delay use intercoursecondoms. .Interventionpost-intervention.collectedMatched post-test at baseline questionnaire N=477 and 3-month data were ResponsibleBecoming A Teen Jackson, MS yearsMean=15.3 center.Setting: Conference room in a health Experimental.Comparison post-test N=390 suresRepeated MANOVA mea- Initiation of intercourse: + evaluationThis was a design very strong with tri Jefferson,Alleyne,O'Bannon,St. Lawrence, Shirley Brasfield N=225Low SES grade=9.7Mean Content:meetings.Sessions: Based 8 90- toupon 120-minute social learning weekly hourinterventionassignedIndividual educational to youth receiveor an were intervention. alternative the randomly study 2- gender.groupsureused and impactto mea- of NumberviousSexual two intercourse of months: sex partners: +during + pre- sophisticatedoutcometermrandom follow-up, assignment, measures, statistical multiple and long- c-) 1995 BI=100%F=72%M=28% petencies,riskyandtheory. emotional behavior, Designed and meaningsmodelprovide to affect behavioral practice, attached cognitive feed-corn- to monthslater,collectedMatched 6 monthslater. questionnaireat baseline,2 later, and monthsdata 12 were pre-test.differencesNo significant at vaginalFrequency intercourse: of unprotected Females:Males: + 0 analysis.reportedOn some risksoutcomes, fluctuated doms,"lines,"decisionsCoveredback, and AIDS andeffectivereinforce pressures, information, social new skills. skills,use sexual of and con- Frequencytected vaginal of unprotectedcondom-pro-intercourse: oral+ timeconsiderably period to from another. one 1 handle.situations that would be difficult to sex:Frequency + of unprotected anal I andwithMethods: female5-15 Smallyouths co-facilitators. group were discussions led by male Frequencysex: + of condom- I SessionsConsiderable with role-playing HIV+ youth. and practice. protectedPercent of byanal acts condoms: sex:0of intercourse + 151 ) 1 Change Note:For all studies, the significance level was set at p < 0.05. If authors reported significant findings at thein p outcome< 0.10 level, for group they werereceiving intervention: no significant change = 0; significant desirable change = +; significant undesirable change = treated as insignificant in this table. Table 4.3: Studies of HIV/AIDS Education Programs continued , Study Information Program(s) / 1 LocationSample / Description Age / Study Results Publication Date Author(s) / SampleSES (N) / Post GenderEthnicity Grade/ / i Program Description Design MethodsAnalytic Change in Outcome' CommentsAdditional Philadelphia, Mean= I 4.6 I Setting: At a school on a Saturday I: Experimental. Arialysis of . Had sex: 0 In general, this was a ResponsibleBe Proud, Be PA I. years . Session= 5 continuous hours. 1 Youths were randomly assigned to covariance was I NUmber of days had sex: + strong evaluation design. Jemmott III, Jemmott, Fong Low SES I Not reported i. Content Based upon theory of rea- I the treatment group and a control forused the to gender control 1I Number of partners: + wasHowever, relatively thesmall, sample and size 1992 N=I50 I M=100% ; soned action. Included information , , - iI opportunitiesgroup, which received intervention. a career, andof I. the respec- facilitator I Rated frequency of condom the did3-month not allow post-test measure- BI=100% I cises,about role-playing,risks videotape% and other games active exer I Matched questionnaire data were interventiontive pre" I use: + effects.ment of long-term 1 learning activities. All were culturally ! collected at baseline, at the end of 1 Number of days of sex with- I and developmentally appropriate. i later.the interventions, and 3 months . outcomemeasure ofbeing the I out condoms: + I1 facilitatorsMethods: Implementedwith backgrounds by 27 in black human iI measured. . Had heterosexual anal sex: + 1 asexuality mean of Taught about 6in youths. small groups with I1 Number of days hadsexual hetero- anal sex:0 I partners:Number of 0 female anal sex

) 1 Change in Note: For all studies, the significance level was set at p < 0.05. If authors reported significant findings at the p < 0.10 level, theyoutcome153 for group receiving intervention: no significant change = 0; significant desirable change = +; significant undesirable change = -. were treated as insignificant in this table. Table 4.3: Studies of HIV/AIDS Education Programs continued Results Study Information Sample Description : Age / Study Publication DateProgram(s)Author(s) / / SampleSESLocation (N) / Post / I GenderEthnicity Grade/ / Program Description Design MethodsAnalytic Change in Outcome' CommentsAdditional Responsible!Be Proud! Be A Sexual PAPhiladelphia, yearsMean=11.8 SaturdaySetting: Recruitedprogram on from school schools campuses. for a trOltoExperimental. 2 grouptreatment that Random groupsreceived andassignment different I con- testsChi-squared or f-tests. I Abstinence-based:Initiation of intercourse: At 3 months: + I wasstudy.This strongwas Both a andverythe designthestrong CurriculumAbstinence Low income 6-7th graders overSessions: two Saturdays. 8 I-hour modules delivered intervention. I Frequency of sex: I Effectsresults werein mediating positive. Responsible!Be Proud! Be A Safer Total N=659 F=53%M=47% Content 2 curricula, 1 abstinence- . collectedMatched questionnaireat baseline, 3 months,data were 6 At 63 months:0 I behavioralvariables supported effects. Non- ("BeSex Proud!Curriculum Be BI=100% cognitive-behaviorbased, 1 safer sex-based. theories Based and elicita-on months, and I 2 months. At 12 months:0 I effectssignificant were behavioral typically in asResponsible!" "Making a Difference") now known videos,tion research. games, Small brainstorming, group discussions, experi- Condom use: At 3 months: 0 I Thethe desired safer sex direction. curriculum Jemmott, Jemmott, Fong exercises.ential exercises, The safer and skill-buildingsex curriculum At 126 months: months: 0 + I uponhad significant frequency effects of 1998 Trainedcondomalso addressed adultuse. or hedonistic peer facilitators. beliefs about I Frequency of unprotected sex: At 63 months:0 I notencedyouthsunprotected all atyouth. sexually baseline, sex experi- among but 1 At 12 months:0 I Results did nOt differ by I SaferInitiation sex-based: of intercourse: At 3 months:0 I andmatching staff onparticipants gender, nor tr, I Frequency of sex: At 1263 months:months:0 months: + + I facilitators.by adult versus peer I Condom use: At 1263 months: months: + + Frequency of unprotected sex: At 1236 months: months: + 0 1 5 6 1 5 5 ) 1 Change in outcome for group receivingNote:For intervention: all studies, no significant the significance change =level 0; significant was set at desirable p < 0.05.If change authors reported significant findings at the p < 0.10 level, they were treated = +; significant undesirable change = as insignificant in this table. Table 4.3: Studies of HIV/AIDS Education Programs continued Study Information Program(s) / LocationSample / Description Age / Study Results Publication Date Author(s) / Sample (N)SES / Post I Grade / Program Description Design MethodsAnalytic Change in Outcome' CommentsAdditional Ethnicity I Get Real About AIDS Colorado yearsMean=15 Setting: Classrooms. Quasi-experimental. logisticRandom-effects and lin- Ever had sex:0 1 lowedThe students for a relatively were foi- KolbeBanspach,Main, Iverson, Collins, McGloin, Rugg, reportedNot 9th=60%1 Oth=13% ContentSessions: Based 15. upon social cognitive I comparisontoSeventeen the intervention schools group, werewhichgroup assigned receivedand the usedmodelsear regression to analyzewere NumberEver purchased of sexual condoms: encounters + mentalGivenshort periodthe design, quasi-experi- of sophisti- time. 1994 N=979 M=5I%12th=5%Ilth=22% avoidance,functionalEmphasizedtheory and knowledge impacttheory skills. Topicsof norms,reasoned about covered: HIV-riskcondom action. I sonthe program. school'sschools usualoffered Four HIVof no the education program. 6 compari- fordata,student-level clustering controlling of duringNumber last of 2sexual months: partners +0 correctedmethodscated and were forproper the used analyticclus- that Wh=65%F=49% avoid,use, and and skills leave to identify,risky situations. manage, I aftercollectedMatched the intervention,at baseline, questionnaire 2and months 6data were Analysesschools.students withincon- months:Condom + use during last 2 matchingTheretering effectwere pre-test difficultiesof students. and in BI=6%His=21% PostersTeachers were received distributed 40 hours throughout of training. I months after the intervention. denttrolled pre-test for students.post-test All scoresthe results for stu- Asn=3% 1 Intervention post-test N=560 I response (out- were just barely signifi- Othr=5% HIVthe school information and students cards. were given Comparison post-test N=4I9 I baseline),measuredcome variable grade, at cant at the .05 level. curriculum.Methods: 25 teachers taught I gender.ethnicity, and Healthy Oakland Teens Oakland, CA I 12-16 years Setting: Social science classes at a I Quasi-experimental. I1 The 2 groups I Initiation of sex: + The validity of these Ekstrand, Siegel, Nido, Low SES Mean=13.1 middle school. 1 were compared . results was reduced by Krasnovsky,Faigeles, Cummings, Chiment, Battle, N=250 years Sessions: 5 adult-led / 8 peer-led. iI A cohort of studentsvention in school the inter- were compared I regressionwith logistic assignment,the lack of random some differ- Coates 7th graders Content 5 adult-led sessions : withnearby cohorts schools. of students in similar 1 baselinecontrolling for ventionences between and comparison the inter- M=48% included basic information on I' I differences. 1996 F=52% andanatomy, preventive substance behaviors. abuse, EightHIV/STDs, peer- I1 collectedBaseline in thequestionnaire 7th grade anddata 8- were samplegroups, sizerelatively for analyses small His=B1=78% I I% andled sessions included wereperception more ofinteractive risk, values 11 months later in the 8th grade. I (N=190),of initiation and of failuresex to Wh=7% ventiveclarification, behaviors, costs andinfluence benefits of alcohol of pre- ControlIntervention post-test post-test N=I43 N=I07 I effects.adjust for In clusteringaddition, par- Othr=5% and drugs,condom peer use. norms, refusal skills, I; mentsent consent changed, require- but the I thosestudy wasrespondents restricted who to I waspassivecompleted still parentalin effect.surveys consent when ) 1 Change in Note: For all studies, the significance level was set at p < 0.05. If authors reported significant findings at the p < 0.10 level, they outcome for group receiving intervention: no significant change =157 0; significant desirable change = +; significant undesirable change = were treated as insignificant in this table. 158 Table 4.3: StudiesStudy ofInformation HIV/AIDS Education Programs continued Sample Description Age / Study Results Publication Date Program(s)Author(s) / / SampleSESLocation (N) / PostI / I GenderEthnicity Grade/ / Program Description Design MethodsAnalytic Change in Outcome' CommentsAdditional Gillmore,Untitled Morrison, NotSeattle,WA I Mean=9th14-1.9 years I healthSetting: STD Juvenile clinics, detention, and other public clinics. YouthExperimental. were randomly assigned to covariancesuresRepeated analysis was mea- of CondomNumber of use:0 sexual partners: 0 werebehavioralIn fewaddition significant impact,to the lack there of Gutierrrez,Richey, Balassone, Fanis reported I grade bookContent with Group basic information1 received anda comic Matched3 groups. questionnaire data were forused baseline to control condom:Refusing 0sex without variables.changes in mediating N=3I4 F=46%M=54% 1 examples of communication skills. collected at baseline,3 months differences. 1997 1 Group 2 received the comic book and later, and 6 months later. I In general this was a Wh=48%BI=52% I skillsa videotape for negotiating with teen condom actors modelinguse. i butstrong the research validity ofdesign, the Group 3 participated in 2 4-hour skill- 1 results were limited by training sessions with role playing, 1 the small sample sizes I thesmall comic group book exercises, and the games, videotape. and both for some analyses. Magura,Untitled Kang, Shapiro NYNew York, yearsMean=17.8 center.adolescentSetting: Department reception andof Correction's detention program;InterventionQuasi-experimental. the group comparison received group the sionlogisticMultiple or multiple regres- regres- High-riskMultiple sexual sexual partners:0 partners:0 smallassignmentThe lack sample of randomand size relatively 1994 N=157reportedNot M=100%NotB1=65% reported problem-solvingContentSessions:4 Based I-hour upontherapy. sessions. techniques Included of wereprogram.members released were before wait-listed receiving but the tosion measureinterventionimpact were used of the and sex:CondomAnal + sex:0 use during vaginal theHowever,reduced studyfindings thedesign. the onstrength veracity condom of of Wh=2%His=33% aboutabuse.healthsmall-group drugs,issues, Also discussionsincluded sex, especially and decision-making HIV AIDSfocusing and and role- upon drug jail).laterlectedMatched (S at months baselineinterview after and data release 10 were months from col- theto control outcomemeasurebaseline for of the Condomsex: + use induring general: oral/anal + groups,tionsimilarityuse areandthe increased controllingcomparison of the interven- by the playing.Participants volunteered for program. ComparisonIntervention post-test:N=58post-test N=99 variable.backgrounddifferenceNo significant on condomconsistentandfor pre-test the combinationuse findings measures, and lack on of of iorsincludingcharacteristics, at pre-test. behav- ! drugtakingfindings use). behaviors for other (e.g., risk- used.ofOne-tailed significance tests r n ) 1 Change in Note:For all studies, the significance level was set at p < 0.05.1f authors reported significant findings at the p < 0.10 level, they wereoutcome treated for group receiving intervention: no significant change = 0; significant desirable change = +; significant undesirable change = as insignificant in this table. Table 4.3: StudiesStudy of Information C-OlVIAIDS Education Programs continued Sample Description Study Results Program(s) / I Location / 1 Age / 1 SES / Post 1 Grade / i Design Analytic Change in Outcome' Additional Publication Date Author(s) / Sample (N) i GenderEthnicity / 1 Program Description . Methods Comments New York, 11-18 years Setting: Shelter for runaway youth 1 Quasi-experimental. Outcomes were At 3 months: Several things reduced Rotheram-Borus,Untitled NY I Mean=15.5 Sessions:Designed as 20, but was 3 I1 One shelter for runaway youth theregressed number onto of AbstainedConsistent from condom sex:0 use: + the validitylack of randomof this design: Koopman, Haigners, Low SES . . years : to 30. Mean = 13 sessions. 1 offered the program, while a sec- sessions that Avoidance of high-risk assignment the use of Davies I 1 1 ond similar shelter in the same city mnaways panic- situations: + only two groups; the rel- (runaway Not reported I Content Included general knowledge 1 serving similar youth did not, ipated in and atively small sample size; 1991 youths) 1 about HIV/AIDS, training in coping N=I45 F=64%M=36% Ii inskills high-risk (including situations), unrealistic access expectation to I atMatched baseline, interview 3 months data later, collected and 6 variables.demographic At 6 months:Abstinence:Consistent 0 condom use: + quatelyand the failurecontrol to for ade- other i B1=63% 1i methodshealth care of andsurmounting other resources, individual and I1i months later. There were no Avoidance of high-risk *differencesthose between youth who His=22% i 1 significant differ- situations: + remained in the shelter 1 Intervention post-test N=78 Wh=8%othr=7% I ing).barriers Activities (covered were in interactiveprivate counsel- (e.g., Comparison post-test N=67 theences 2 groups between at timefor longer and those periods who of I1 ioraldramatizationsdeveloped coping raps responses). andand practicedsoap opera behav- baseline. periods.remained for shorter 1 Methods:Taught in small groups thereOn the were other no hand, significant 1 female(N=I0) 4team days of per trained week leaders. by male and graphicsexualdifferences characteristics risk behaviorsin demo- or at twoappearedmagnitudebaseline groups, betweenand large. of the effectsthe

) 1 Change in Note: For all studies, the significance level was set at p < 0.05.1f authors reported significant findings at the p < 0.10 level, they wereoutcome for group receiving intervention: no significant change = 0; significant desirable161 change = +; significant undesirable change = . treated as insignificant in this table. 162 Table 4.3: StudiesStudy of Information DIM/AIDS Education Programs continued Sample Description Study Results Publication DateProgram(s)Author(s) / / Sample (N)SESLocation / Post / GenderEthnicityGrade / / Age / Program Description Design MethodsAnalytic Change in Outcome' CommentsAdditional Ozawa,Slonim-Nevo,Auslander,Untitled Jung St. Louis,reportedNot MO yearsNotMean=14.7 reported wholems.neglectedSetting: had been Residential or delinquent, had mental centers abused, health for orprob- youth domlyExperimental.15 residentialassigned to centers skills training were ran- ofvarianceAnalysis outcomebaseline ofwith co-scores 0Sex with an unknown partner: follow-up,assignment,tionThis designwas a multiple strong with long-term random evalua- 1996 N=218 I M=56% Sessions: 9 1.5-2 hour sessions deliv- group,control discussion-only group. group, or covariates.variables as Vaginal sex without condom:0 sophisticatedoutcome measures, statistical and I Wh=54% BI=46%F=44% contentbehavioraleredContent over the atheory. same3-weekBased forSubstantive onperiod. both. cognitive- The HIV skills- 9-12collectedMatched months. questionnaireat baseline, post-test, dam were and cometinuousmodelsGeneral variables forout- con-linear Anal sex without condom:0 Theanalysis.modest, sample sizebut resultswas were groupandmodeling,based practice, wasgroup demonstrations, taught waswhile taughtusing the discussion problemskills role using plays, solv- dichotomous.modelsand logistic for closedesirednot to consistently significance. direction in nor the itatorsing and10)Methods:Taught per ofdiscussions. same group. age and in smallgender. groups Two facil- (8- YouthPrevention AIDS Project Chicago, IL Not reported 1 Setting: School classrooms. Experimental. testsSignificance showed no Initiation of intercourse: 0 wereLarge randomly groups assigned of students (YAPP)Levy, Perhats,Weeks, N=1,669Low SES F=51%M=49%7th graders 1 gradeContentSessions: booster. Based 10 in on7th social grade, learning 5 in 8th mentdomlyFifteen group, assigned school which districts to includedthe firstwere treat- ran- at pre-testbetweendifferences groups NumberFrequency of ofsexual sexual partners:0 activity 0 atanalyses(school the individual districts), were conducted level. but 1995Handler, Thu, Flay Wh=24%BI=59% tion,theory,STD decision-making, topics prevention, included pregnancy and HIV/AIDS/ resistance/ preven- group,activities,classroom which a instruction second received treatment plusclassroom parent ANOVAsThree-way con- CondomEver used use condoms:0 last sec 0 I (56%)Attrition by rates 9th grade. were high Weeks,Handler, Levy, Perhats, Gordon, Flay Othr=4%His=13% lectures,negotiation class skills. discussions, Activities small included group curriculum.whichinstruction, received and thethe standard control group, AIDS andtrollinglogistic gender, for regres-race and I theoccurredPost-test 8th grade mayshortly haveinterven- after 1997 Methods:box.petitions,exercises, and roleInstruction an plays, anonymous educational was imple- question com- ventionincollectedMatched the 7th booster questionnaire beforegrade, in theafter the intervention the 8thdata inter-grade, were gender.forsion racerandom-effects controlling Alsoand for ainitiationdeclinetion, possible not in allowingof thedelay intercourse impact in for aor tors.mented by professional health educa- ControlInterventionand in the post-test 9th post-test grade. N=668 N=1,00I missinghandleregression some data. to to be measured. 164 163 ) 1 Change in Note: For all studies, the significance level was set at p < 0.05. If authors reported significant findings at the p < 0.10 level, they outcome for group receiving intervention: no significant change = 0; significant desirable change = +; significant undesirable change = were treated as insignificant in this table. Table 4.4:StudiesStudy Information of Sex and ='IHV/AHS Education Programs for Parents and Their Families Sample Description Study Results Program(s)Author(s) / / SampleSESLocation (N) / Post / GenderGrade / / Age / Program Description Design MethodsAnalytic Change in Outcome' CommentsAdditional Facts and Feelings Publication Date Northern 12-14 yearsEthnicity Setting: Home-based video. Experimental. I Repeated mea- Initiation of intercourse:0 This was a strong design. Lee,Miller, Christopherson, Norton, Jenson, King middleUpper-Utah SES graders7th and 8th rials.toSessions: 20-minute 6 units, video each and including written mate- a 15- group,toFamilies 3 which groups:the were received randomly first the intervention videos assigned i variance.sures analysis of initiatedtheHowever, youths intercourse only in any 3-5% group of 1993 Not reported 1I and mailed newSletters; the second Thus,during itthe was elapsed difficult year. to N=503 Wh=95% mation,Content modeled The video parent/child provided commu- infor- 1 theintervention videos onlr group, and which the control received produce an impact on 86%Mormon= consistentnication, and with emphasized abstinence. sexual The values writ- I group, which received nothing. canttherethis outcome. impact Was not upon aHowever, signifi- youths' coveredtopicsten materials for changes discussion. suggested in puberty, Together questions facts they and I atcollectedMatched baseline, questionnairefrom after teens 3 months, and data parents and were followingcourseintent to during have year theinter-or until sexualitymunication,about reproduction, in thevalues media, and parent/teen decision-makingsexual behavior, com- I Interventionafter I year. group 1 post-test marriage. skills, and communication skills. 1 InterventionN=120 group 2 post-test encouragePhone calls use were of themade materials. bi-weekly to 1 N=I22 Control group post-test N=261 i Nicholson,Growing Together Postrado TN;Memphis,Dallas,TX; Omaha, yearsMean=12.4 Sessions:Setting: Girls 5 2-hour Clubs. classes. IiQuasi-experimental.toGirls participate and parents in programs who volunteered within ; linganalysis,Multiple-logistic for back- control- Initiate intercourse:0 thedesignThe lack strength was of randomreduced of this by 1991 DEWilmington,NE; F=100%Not reported IncludedcommunicateContent factsDesigned withabout their to adolescent help daughters. parents I the programfour Girls group, Clubs while constituted those :1 ground and characteristics.baseline smallassignment sample and sizes. relatively Low SES Wh=BI=74% 1 5% tionships.sexuality, values about dating and rela- 1 who chose not to participateconstituted the comparison group. I There were few wasThe comparisonabout 2.5 times group N=20IBaseline: Othr= 1 1% daughtersticeMethods: communicating Interactiveand others. withexercises; their own prac- I .at baselineQuestionnaire and 1 and data 2 years were later. collected atbetweendifferences baseline. groups almostmore(p=.054),intercourse. likely significant to but initiate This not quite.was

) 1 Change in Note: For all studies, the significance level was set at p < 0.05. If authors reported significant findings at the p < 0.10 level, they outcome for group receiving intervention: no significant change = 0;significant desirable change = +; significant undesirable change = were treated as insignificant in this table. Table 4$:Study Studies Information if Protocols for Clink App Sample Description Age / Studyintments and Supportive Activities Results Publication DateProgram(s)Author(s) / / SampleSESLocation (N) / PostI / I GenderEthnicity Grade/ / Program Description Design MethodsAnalytic Change in Outcome' CommentsAdditional Simmens,Boekeloo,ASSESS Cheng,Schamus, LowD.C.Washington, SES Not14-15=47%12-13=54% reported 1 Sessions:Setting: Managed 1 care clinic sites. treatmentYouthExperimental. were and randomly control assigned groups. to AdjustmentstestChi-squaredFisher's exact and for . Vaginal,months:VaginalAt 3 months: intercourseanal, or oral in sex last in 3last large.sampleresearchThis was sizes design,a strong were although not O'Connor, D'Angelo 1 Content Major message: unpro- 3 months:0 1999 N=197 F=50%M=50% 1 safer,tected abstinence sex was unsafe, was safest. condoms Focus were Control group received usual providers.clustering within Frequency of sex:0 Bivariate analyses 1 health examination. revealed no significant 1 upon STDs and HIV. Materials pro- Multiple logistic 1 Number of partners:0 impact upon sex in last BI=64% Matched interview surveys were or linear regres- 1 Use of condoms at last sex: + W=19% 1 minutevided to audiotaped each youth risk included:a assessment 15- conducted at end of clinic appoint- sion models. Unprotected sex: + Multivariatethree months. analyses His=4% 1 and education program, a discussion ment, 3 months later, and 9 1 STD diagnoses:0 found it to be just barely Oth=13% 1 ice-breaker, two brochures on skills months later. At 9 months: significant 1 and self-efficacy to avoid unprotected Intervention follow-up:N=94 1 Intercourse in last 3 months:0 1 sex, a brochure on community Control follow-up: N=I03 1 Vaginal, anal, or oral sex in last 1 Physiciansresources, andreviewed parent the brochures. risk assess- Frequency3 months: 0of sex: 0 1 ment with each youth and discussed Number of partners:0 .1 unprotectedconcerns and sex. methods of avoiding STDUnprotectedUse ofdiagnoses:0 condoms sex:0 at last sex:0 Untitled OR,Portland, and 9th=15-18 I 8% years I organization.Setting: Health maintenance Experimental. sionLogistic was regres- used Frequency of sexual activity 0 imentalThis was design. a strong exper- Wiest,Danielson, Greenlick Marcy, Plunkett, WAVancouver, Ilth=24%lOth=34% 1 appointmentSessions: Part of a one-hour medical treatmentYouth were and randomly control assigned groups. to potentiallyvarietyto control of for a inContraceptive previous year:0 methods used catedThe results greater mostly pill use indi- and 1990 N=971Middle SES Wh=9I%M=I00%12th=24% anatomy,slide-tapeContent STDs, program Included contraception, that two focused parts: couple upon (I) a onecollectedMatched year later.questionnaireat baseline and data about were variables.confounding ousPrimaryuse:Frequency year:0 method + of contraceptive used in previ- afterthoseoccurred baseline. who mostly initiated among sex B1=5% 1 communication, and access to health Asn=4% 1 services, and (2) a visit with a health Method used at last inter- 1 care practitioner who focused upon course: + 1 contraception, reproductive goals, I1 contraception.healthEmphasized risks, and abstinence the patient's and interests. use of 167 ) 1 Change in outcome for group receiving Note:Forintervention: all studies, no significant the significance change = level0; significant was set desirableat p < 0.05. change If authors = +; reportedsignificant significant undesirable findings change at the= p < 0.10 level, they were treated as insignificant in this table. 16o Table 4.5: StudiesStudy Information of Protocols for Clinic Appointments and Supportive Activities continued Sample Description Age / Study Results Publication DateProgram(s)Author(s) / / SampleSESLocation (N) / Post / 1 GenderEthnicityGrade / / Program Description Design MethodsAnalytic Change in Outcome' CommentsAdditional rn DeLameter,VVagstaff,HavensUntitled WIMilwaukee, yearsMean=18.315-19 years Sessions:Video:1Setting: Urban STD 14-minute clinic. video. aYouthExperimental. video were intervention, randomly health assigned educa- to pre-testpost-testcovarianceAnalysis differ-of with 1 Number of partners:Frequency 0 of sex: modestventionsorousAlthough design, interventions, werethis wasthe extremely inter- a rig- 074 ro 2000 N=562reportedNot M=I00%Not reported STD/AIDS.ContentlastingEducator/patient about Video:focused Designed .14 minutes.session: to affectI onsession per- AlltreatmentControltor groupsintervention, classes andreceived education receiyed or controlresults standardprogram. ofgroup. STD variable.dependentence as the CasualSteady partner:0 andSTD,withespecially receivingbeing learning in tested comparison the stan-results,for an CDus BI=100% wasresponseceived culturally risk, efficacy, motivational appropriate, and self-efficacy. factors, presented It Matchedtests. questionniare data were adjustedBonferroni-Used t-tests. Condom use: Steady partner:0 Thedard standard education care and care. CD(D couldby people identify, with evoked whom socialthe viewers responsi- later.post.collected 30 days at baseline, later, and immediate 6 months Casual partner:0 described.intervention was not : Americanbility,included and emphasizedyouth personal are at stories, that risk. African- It involved ref- reportedthreeOver timegroups, increases among there all in were vs7 offdemonstratederence a condom, group andnorms how presented to for put condom on informa-and use, take condoms.offrequency partners, of and sex, use number of 4...11 0C) allowedHealthtion in aneducator:face-to-face tailoring unauthoritative of message, manner. sessionpersonal- CYO a3s condomquestions.ization of on risk, Included plastic and penis.answers practice to putting 0(D On;Untitled Langefeld, Katz Caine LowINIndianapolis, SES Mean=17.914-19 years clinics.Setting: Family planning and STD wereTwoExperimental. family randomly planning assigned clinic tosites treat- withwereChange Wilcoxoncompared scores Condom use: + Thelinepoorly two on groups matched ethnicity:the were at base- CD(IS 1996 N=112 F=I00%years Methods:Sessions: One-on-one I session. 10-20 minute session. Individualsment and control at the STDconditions. clinic were pariedsigned t-tests.rank and controlmostlyintervention blacksite mostly siteand was the (D Oth=45%BI=55% I chlamydiamodel.Content Included based (with upon use (I) discussionof the a pamphlet),health ofbelief Matchedrandomly questionnaireassigned. data col- analysisgencyAlso contin- table control- white.chlamydia, However, and all the had C/c) 7s 1 chlanyaiaAll hadpositive tested for getting(3)put(2) briefmodelinga condom her role partner playand on (over involvingpractice to use a banana), a ofa condom. womanhow toand traditionalControllow-up.lected at group baseline discussion got andsimilar with 6-month lengthnurse fol- HaenszelMantel-(Cochran-ling for baseline test). groupsattritionTherematched were was (46%). onwell- also behaviors. Analyseshigh treatment.aboutComparisonIntervention condom use post-test and partner N=50N=55 planningteringdid not at control clinics.the two for family clus- ) 1 Change in Note: For all studies-, the significance level was set at p < 0.05.1f authors reported significant findings at the p < 0.10 level, they wereoutcome for group receiving intervention: no significant change = 0; significant desirable change = +; significant undesirable change = 169 treated as insignificant in this table. 1_70 Table 4.5: Studies of Protocols for Clinic Appointments and Supportive Activities continued Study Information 1 Sample Description Study Results Publication Date Program(s)Author(s) / / I SampleSES (N)Location / Post / ! ! GenderEthnicityGrade / / Age / Program Description Design MethodsAnalytic Change in Outcome' CommentsAdditional Untitled 1 Pennsylvania 15=16% Setting: Family planning clinic. Quasi-experimental. 1 Analysis of Among all patients: The validity of the design 1 16=34% variance. Use of contraception at 6 was reduced by the Winter, Breckenmaker 17=>40% 1 Sessions: 2 sessions, then more fre- Six clinics assigned to receive the 1 months: + 1 quent visits. improved patient protocols or the 1 Small pre-test inability to compare 1991 1 Not 1 Not reported standard protocols. differences in contraceptive use at 1 Use of contraception at 12 reported 1 Content: Greater focus upon the satisfaction. months: 0 1 baseline, the failure to F=100% 1 non-medical needs of adolescents:- Some baseline data collected. 1 demonstrate equivalence 1 more information, more concrete Patient records for the subsequent 1 between the two groups : N=672 I1 Wh=98% BI=1% 1 comfortinstruction, during moreexam, counseling, more involve- more 9-20 months examined. encingAmong problems: patients experi- 1 to controlin other for ways, clustering and failure His=1% : 1 1 ment of male partners, more recogni- Intervention post-test N=228 Use of contraception at 6 effect. 1 tion of peer pressure, and more Comparison post-test N=444 months: + parental involvement. 1 Other data supported Methods: One-on-one sessions. months:Use of+ contraception at 12 1 the expandedpositive impact program. of I Furstenberg,Herceg-Baron,Untitled Shea, Harris PAPhiladelphia, <16=31%17=36%16=33% clinics.hoodSetting: health Hospital clinics, clinics, and family neighbor- planning PatientsExperimental. were randomly assigned tionamongSignificance and interven- control UseClinic of attendance:0 contraception:0 imentalThis was design. a strong exper- 1986 reportedNot Not reported support.Sessions: Patients First intervention:Family were asked to 1 3trolto and the groups I2 interviews intervention (which respectively).participated and 2 con- in suredgroups at mea- post- Pregnancy: 0 supportpatientsThe proportion groupin the familythat of actu- N=358 BI=53%F=100% withattend a family 6 weekly member. counseling sessions I1 Interviews conducteri at baseline, 6 Notest significant periods. memberally brought was a veryfamily low Wh=47% Second intervention: Periodic sup- 1 months later, and 15 months later. differencespre-test. at (30%). Participants weekscallsport. from following Patients clinic staffthe received initialduring clinic 2-6 the telephone 4-6visit. 1 ControlSecondFirst intervention:N=93 intervention:groups: N=I98 N=61 sionsreceived each. only 2.3 ses-

) 1 Change Note: For all studies, the significance level was set at p < 0.05. If authors reported significant findings at the pin

) 1 Change in outcome for group receiving intervention:noNote: For all studies,significant the changesignificance = 0; significantlevel was set desirable at p < 0.05. change If authors = +; significant reported significantundesirable findings change at =the p < 0.10 level, they 173 were treated as insignificant in this table. 174 Table 4.7:Study Studies Information of School- ased and School-Link d Heakh Centers Sample Description Study Results 1 Program(s) / i SESLocation / Post / 1 Grade / Age / Design Analytic Change in Outcome' Additional Publication Date Author(s) / ,1 Sample (N) 1 GenderEthnicity / Program Description Methods Comments 1 1 1 St Paul school-based i St. Paul MN Not reported Setting: One junior/senior high Quasi-experimental. I1 No tests of sit ' Birth rates: + 1 The validity of these clinics (see Kirby et aL I school and two high schools. 1 nificance were 1 results was limited by below) 1 SESLow-middle 9-12 graders 1 Pregnancy and birth rates for the Ii conducted. . : the small number of - . I 1 first year each clinic was open F=100% Content School-based clinics pro- i 1 baseline year% incom- Edwards, Steinman, vided comprehensive health service% 1 were compared with subsequent trI Arnol& Hakanson i N'apProx.1,000 1 Not reported reproductive health care provided in I years; pregnancy and birth rates I :I nanciesplete knowledge and birth% of lack pret 1980 i the clinic; contraceptives picked up at 1 were estimated by clinic staff from 1 of comparison schools, 1 their personal knowledge of Su, - special hospital clinic; day care for 1 and lack of tests of 0 infants provided. . dent pregnancies and births. . significance. 1 I St. Paul school-based :i St. Paul, MN Not reported Setting: High schools. Quasi-experimental. Several methods Birth rates: 0 1 There were no compari- clinics (see Edwards et I 1 were used to 1 son schools in the study al. above) i: Not 9-12 graders Content Provided comprehensive 1 In four school% birth rates for 5 compare pre- : There were large year- 1 reported F=100% health services; reproductive health 1 years before and 6 years after din- clinic and post- to-year fluctuations in Kirby Resnick, Downes, :! care provided in the clinic; contracep- 1 ics opened were compare& in one clinic birth birth rates that may have Kocher, Gunderson, N=1 838 to ' tives picked up at special hospital 1 school birth rates 2 years before rate% Tests of obscured a small clinic 1i Not reported 0 Pothoff, Zelterman, Blum I depending2,988 clinic; day care for infants provide& 1 wereand 3 compared;years after birththe clinic rates opened were significance impact. St. Paul students 1993 upon year 1) estimated by matching school conducted goodhad access health tocare relatively and had 1 enrollment records and county tri birth certificates Onlyopened.weightsrelatively school-wide before low birth the effects clinics -d contraception.usedthoughwere measured,the not clinics all students foreven 1 P 175 ) 1 Change in outcome for Note: For all studies, the significance level was set at p < 0.05. If authors reported significant findings at the p < 0.10 level, they group receiving intervention: no significant change = 0; significant desirable change = +; significant undesirable change = . were treated as insignificant in this table. Table 4.7:StudiesStudy Information of School-Based and School-Linked Health Centers continued Sample Description Study Results Publication DateProgram(s)Author(s) / / SampleSESLocation (N) / Post / GenderEthnicity Grade/ /Age / Program Description Design MethodsAnalytic Change in Outcome' CommentsAdditional Self Center MDBaltimore, I Not reported clinic.Setting:Classrooms and storefront 1 Quasi-experimental. sisLife used table to analy- Initiation of intercourse: Females: + I resultsThe validity were oflimited the by Streett,Zabin, Hirsch, Hardy Smith, Low SES 7-12 graders 1 Content Clinic staff made presenta- 1 S.choolOne junior were and selected 1 senior for high the pro- relatedadjust for differ- age- Males:Not provided 1 higherseveral risk factors: students some 1986 F=55%M=45% tions to each home room at least jf gram group; another junior and ences in some 1 Contraceptive1 use: Ij dropped out between N=6,596 once per year, provided individual con- 1 senior high school were selected tables. All students: the first fall survey and I B1=99% duringsultation the in school the school day, healthand provided suite a for the comparison group. Multiple tests of 1 intercourseTime from first to clinic schoolthree subsequent surveys; the spring pro- full range of reproductive health ser- I significance attendance: + 1 gram schools (one of counseling,vices (group contraceptive discussions, individual services, 1 from students school-wideQuestionnaire in the data were collected comparewere used inter- to Females: j I magnetwhich was, school) in part, may a have I i. Use at last pregnancy testing, refeiral) at the clinic 1 fall before the program was imple- vention and intercourse: + i differed in unknown located across the street from the mented and the three following comparison 1 ways from the compari- juniorhighcontacts school high school. andwere a throughfew Most blocks students the fromsmall- the I analyzedsprings. Pregnancy for only 28 rates months. were groups. Pregnancy rates: High school females: + sonnearby;middle schools; school some an additional sampleopened groupMethods: discussions. 2 self care teams (social I forstudy clusteringsizes failed were to effectssmall; control and of the workerspokespeopleservices. and nurse) Peer forleaders delivered program. acted bulk as of . students in schools. Untitled Norfolk,VA 13=2% I Setting:High school. Quasi-experimental. j Changes over j Ever had sex:0 1 The comparison school 1 14=15% 1 1 time were 1 was a good match for Kirby Low SES 15=23% 1 Content The clinic provided com- The intervention school was 1 determined by 1 Age of 1st sex:0 I the clinic school. 1991 N=6,260 j 16=26%17=23% I ingprehensive prescriptions primary for healthcare, contraception. includ- It sonmatched school. with a nearby compari- testschi-squared and t- Frequency of sex: Males: 0 sampleHowever, at thethe selectedcompari- 19=2%18=9%, did not dispense contraception. fromQuestionnaire either the entire data school were collected or tests. I Females; 0 I veryson schoolrepresentative was not sam-a 9th=30% clinicsamples opened, of students 2 years just later, before and 3the Use of contraception: 0 1 I duringple of the the entire2-year school data I 10th=24%1 1 th=23% years later. Pregnancy rate: Males: 0 I datacollection. included The data survey from M=47%12th=24% Birth rate: Females: 0 I the clinic.little9th graders exposure Only school-wide who to hadthe F=53% Males: 0 1I effects were measured, B1=58% Females: 0 1 even though not all stu- I Asn=5%Wh=34% I contraceptives.dents used the Also,clinics the for i Oth=2%His=1% 1 simplistic.statistical analyses were ) 1 Change in Note: For all studies, the significance level was set at p < 0.05. If authors reported significant findings at the p < 0.10 level, theyoutcome for group receiving intervention: no significant change = 0; significant desirable change = +; significant undesirable177 change = . were treated as insignificant in this table. 17° Table 4.7:Studies of School-Based and School-Linked Health Centers continued Study Information ' Sample Description Age / Study Results Publication DateProgram(s)Author(s) / / I Sample (N) SESLocation / Post / I EthnicityGenderGrade / / Program Description Design MethodsAnalytic I Change in Outcome' CommentsAdditional Kirby,Waszak,Untitled Ziegler MI;Muskegon,Gary, IN; 9-12Not reportedgraders I Content:Setting: High All clinics schools. provided com- SchoolsQuasi-experimental. were matched to the ductedsesSeparate were for analy-con- each Initiation of intercourse: Males: 4 sites:I 0 site: + anddesignThis communitypost-test-only was weak differ- school 1991 Jackson,Quincy, MS; FL; F=55%M=45% 1 prescribedprehensiveI referred contraception,health for contraception, services. and In addition, 3 dis-I withwideextent clinicspost-clinic feasible. were Indata compared4 sites, from school-schools with race;gender regression and Females: 4 sites:I 0 site: + theences clinic may school/compar- have affected LowDallas,TX SES B1=100% pensed contraception. wereschools;data fromcollected in similar before comparison and 2 years I site, school-wide data backgroundothercontrolwas used student for to 6 activity:Frequency of sexual Males: 5 sites:0 andunknownison post-testschool ways. analyses design The inpre- 3=8242=1,3601Site: =1,295 N widewereafter surveys.thecollected clinics through opened. school- Data characteristics. Use of contraceptives: Condoms:Females: 5 sites: 0 wereOnlystrongestQuincy measured,school-wide was design. probablyeven effects the 5=1,2744=1,185 All students: 4 sites:1 0 sites:+ usedthough the not clinics all students for Pill: Females: 4 sites:0I Thecontraceptives. clustering effect of I Pregnancy: 5 sites: 0 site: + I notstudents controlled in schools statistically. was I 1 Untitled Clinics in 18=46%17=35% Setting: School health centers. I Quasi-experimental. Numerous Initiation of intercourse: + The basic design was Kisker, Brown, Hill urbanschools areas in 19=19% I dentsSessions: attended By their the seniorclinics anyear, average stu- I schoolsCohort data with fromclinics students were com- in statisticalmulti-variate analy- I 0Had intercourse in last month: wereweak notBaseline collected data from 1994 thethroughout country. I Not reported I of 1.9 times per year. I frompared a national with cohort sample data of students collected ductedses wereto con- I Used a condom at last inter- schoolsthe students before in the stu-clinic Low SES I Not reported I rangeContent: of primary Clinic providedhealth care a wideservices. I in urban schools. impactmeasure the course: 0 receiveddents could services; have the N=3,909 BI=32%His=44% I forSome contraception; clinics only referredothers provided students Comparison:Intervention: N=3,050N=859 !I birthUsed control an effectiveat last inter- method of notcomparison always equivalent; group was I Wh=15% Othr=2%Asn=7% contraception. I Used a method consistentlycourse: I ofcontroland students the fordesign thein schools. clusteringdid not. I Pregnancy:during last 0month:0 theThe impactevaluation upon measured all stu- healthjustclinicdents upon services. forinthose thereproductive schools,using the not ) 1 Change in Note: For all studies, the significance level was set at p < 0.05.1f authors reported significant findings at the p < 0.10 level, they outcomewere for group receiving intervention: no significant change = 0; significant desirable change = +; significant undesirable change = treated as insignificant in this table. Table 4.7: Studies of School-Based and School-Linked Health Centers continued Study Information Program(s) / LocationSample / Description Age / Study Results Publication Date Author(s) / SampleSES (N) / Post I GenderEthnicity Grade/ / Program Description Design MethodsAnalytic Change in Outcome' CommentsAdditional Mid-AtJantic 1 Not reported Setting: High schools. Quasi-experimental. Birth rates were Birth rates:0 This quasi-experimental Untitled 1 city 1 adjusted for stu- design was strengthened Newcomer, Duggan, 9-12 graders Content School-based clinics. 1 Birth rates for 4 clinic schools and dent age and by the relatively large a Toczek1999 Low SES F--= 100% vices,Provided contraceptive comprehensive cciunseling, health and ser- ! lated19 by non-clinic matching schools school were enroll- calcu- ethnicity. comparisonnumber ofschools, clinic and the thanN=more 200,000 B1=80% severalprescriptions years, forcontraception contraception. was Afterdis- 1 Birthment ratesdata andwere birth calculated certificates. for 1 matching of these eq 1 Wh=18% schools, the time series pensed, not just prescribed. 1 yearsbefore the clinics opened and Othr=2% 1 9 years after they opened. Only birthdata, ratesthe generation from public of 1 births that were conceived while !1 1 the student was in school or calrecords, control and of theage statisti- and (.13CD 1 inwithin school 6 months were counted. of being enrolled bystrengthethnicity. the single was However, baselineweakened its. -11D- clinicandAfteryear ethnicity, ofadjustmentschools data. girlswho for inused theage Go oUi thethosely to clinic.clinic give who werebirth did lessthannot use like- 00. 3in00.- ) CD0t-I- CD cra 0.) ) 1 Change in Note: For all studies, the significance level was set at p < 0.05.1f authors reported significant findings at the p < 0.10 level, they wereoutcome for group receiving intervention; no significant change = 0; significant desirable change = +; significant undesirable181 change = treated as insignificant in this table. Table 4.8:Study Studies Information of School Condom-Availability Programs Sample Description Study Results Publication Date Program(s)Author(s) / / Sample.Location (N)SES // Post GenderEthnicityGrade / / Age / Program Description Design MethodsAnalytic Change in Outcome' CommentsAdditional Blake,Massachusetts Ledsky, Goodenow, MixedMassachusetts SES yearsMean=16.I ContentSetting: HighSchools schools. made condoms I1 Post-programQuasi-experimental. questionnaire data logisticcovarianceAnalysis regres- of and AgeEver at first had sex:0 sex: + condomsuredAlthough the availability this impact study of inmea- WindsorSawyer, Lohrmann, N=4,I66 9rh=28%I Oth=26% available to students. I selectedwere collected Massachusetts from 59 highrandomly forsion, community controlling Had sex in last 3 months: + strengthmultiple schools,of its design the was Unpublished (2000) M=51%12th=22%Ilth=24% i able,schools; 50 did9 made not. condoms avail- characteristics.and student lastNumber 3 months:0 of sexual partners in werethesubstantially fact not that collected baselineweakened and data by B1=6%Wh=75%F=49% severalweightedData were factors. by Condomlifetime:Number +ofuse sexual at last partners sex: + in unknowndifferedmadethe fact condoms thatin ways known schools availabilefrom and that Oth=6%Asn=4%His=8% Pregnancy.condomContraceptive at 0 last use sex: other than (e.g.,makeschools theycondoms that were did available morenot 1 aboutstudentslikely condom to provide with use). instruction their Guttmacher,New York Lieberman, NYNew York. 15=23%14=IRX Setting: High schools. Post-programQuasi-experimental. questionnaire data weightedData were to orEver anal engaged sex:0 in oral, vaginal, I designThis was because a very baseline weak Radosh,Ward, Freudenberg, DesJarlais N=12,857Mixed SES 18=10%17=23%16=25% lessonsAIDSContent team in Components each in each grade, school, a included:an resource 6 HIV/AIDS room selectedwere collected New York from schools 12 randomly and IR Cityallmatch New high the York school over- Condom use at last sex: + ; encesanddata pre-existing were between not collected interven- differ- 1997 I Not reported condomdomswhere were informational resource available, volunteers materialsstaff serving inand the as con- Interventionmatched Chicago (New schools. York): N=7,119 Multi-variatesystem. bonschooli and comparision could only be 1 F=54%M=46% mationresource session room, forand parents. an HIV/AIDS infor- N=5,738Comparison (Chicago): comparedlogistic models the 2 controlled.statistically. 1 His=28%BI=47% forsites, numerous controlling As=7%Wh=17%Oth=1% use.iortovariables sexual and condom relatedbehav- ) 1 Change in Note: For all studies, the significance level was set at p < 0.05. If authors reported significant findings at the p < 0.10 level, they outcome for group receiving intervendon: no significant change = 0; significant desirable change =.+; significant undesirable change = were treated as insignificant in this table. 184 Table 4.8:StudiesStudy of Information School Condom-Availability Programs continued Sample Description I Study Results Program(s)Author(s) / / SESLocation / Post / I Gender Grade/ / Age / Program Description Design MethodsAnalytic Change in Outcome' CommentsAdditional CentersHealth Resource Publication Date SamplePAPhiladelphia, (N) I yearsMean=16 Ethnicity Setting: Public high schools. Quasi-experimental. toData match weighted school Ever had sex:0 differedIntervention some from schools com- Furstenberg,Geitz,Teitler, Mixed SES 1 9th-12th . wereContent: opened. Health They provided resource counsel- centers InterviewMultiple designs data were were collected used. characteristics. Had sex in past 4 weeks:0 parison schools. Weiss 1 grade ing on the importance of abstinence, from a random sample of youth in Logistic regres- Used condom at last sex:0 About 32 percent of the 1997 N=1,435 1 F=62%M=38% nearbydoms,reproductive and health general health facilities. health information, referrals con- to sometractsvention,the communities ofafter athe different the pre-intervention intervention, before group the of census inter-and variables.for backgroundsion to control lastHad 4 sex weeks:0 without condom in Severalhealthstudents resource trends had used were center. the in I Chh=1VVh=23%N=66% I% Comparison:Intervention:respondents wereN=519N=916 re-interviewed. notthe significant.desired direction, but Kirby,Seattle Brener, Brown, MixedSeattle,WA SES 16=26%15=24%14=9% ContentSetting: Public In 5 high high schools schools. without QuestionnaireQuasi-experimental. data were collected matchweightedData were Seatde to SexEver in had past sex:0 3 months: + I1 permeanStudents year. 4.6 obtainedcondoms a Peterfreund, Hillard, N=40,572 17=225% health centers, condoms were made school-wide in 10 Seattle schools schools. Had 4 lifetime partners:0 1 Harrist1999 9th-12th18=16% domshighavailable schools were in madevending with available,health machines. centers, unre- In con-5 countryplesand independentlyof schools before thethroughout inprogram random the was sam- sisTwo-way of variance, analy- months:Had + 4 partners in last 3 1 werecomparisonThe well-matched. baseline samples Seattle and grade 1 M=49% informationcenters.stricted, inNear baskets on each abstinence, in basket the health wascondom free differentTheimplemented 2 national samples and samples of2 yearsschools. included later. forwhich clustering. adjusted Mean age of first sex:0 I clinicsThe Seattle might schoolshave with Wh=40%His=5%B1=19%F=51% use, and HIV prevention. Comparison:Intervention: N=27,200 N=I5,072 At last sex: Pill:Condom + use: 1 dureschanged regarding their proce- pill use. 1 Oth=5%As=32% ) 1 Change in Note:For all studies, the significance level was set at p < 0.05. If authors reported significant findings at the p < 0.10 level, they wereoutcome for group receiving interventi.on: no significant change = 0; significant desirable change1._85 = +; significant undesirable change = treated as insignificant in this table. 186 Table 4.9:Study Studes Information of Community-Wide Pregnancy or HIV Prevention Initiatives with Multiple Components Sample Description Study Results Publication DateProgram(s)Author(s) / / SampleSESLocation (N) / Post / GenderEthnicity Grade/ / Age / Program Description Design .AnalyticMethods Change in Outcome' CommentsAdditional Condom Campaign County3 King 16=35%15=34% Setting 3 communities 1 Quasi-experimental. Fisher'sChi-square exact or Ever had sex:0 1 design,This was because a challenging it Alstead, Campsmith, (Seattle) 17=31% 1 Content: 3 primary strategies: (1) Cross-sectional interview survey test were used Sex in last 90 days: 0 attempted to measure Goldbaum,WoodHalley,1999 Hartfield, reportedNotcommunities M=49%Not reported sexually(2)communitymobilization a mass active mediato of supportteens multiple campaign using the levels campaign;pamphlets, targeting of the ! andfirstpaign,data wave7 collected months2 months of the laterbefore medialater (after the(after campaign), thecam- the sections.testpre-to compare propor-and post- I Use of condoms:0 the munities,activepaignimpact upon youth of not the all injust sexuallycam- the upon com- N=1,425 BI=26%Wh=40%F=51% businessesmultipleandposters, billboards; agencies,t-shirts, to distribute and radio organizations, (3) spots, recruitmentcondoms bus and signs, of I: ond wave). Samplescongregated.convenience selected from sites where youth thethose campaign.study who was observed reducedThe validity the of Asn=18%Oth=16% from 22 bins or 25 vending machines 2-month:Baseline: N=341N=478 ! assignmentby the lack and of no random tom- in restrooms and other locations. I 7-month: N=606 Thereicantparison were differences group.also no signif-in con- ! dom use (or in intentandthoseuseexposure condoms)those whoto whoreported to between thedid campaignnot I 0 u ) 1 Change in Note: For all studies, the significance level was set at p < 0.05.1f authors reported significant findings at the p < 0.10 level, they wereoutcome for group receiving intervention: no significant change = 0; significant desirable change = +; significant undesirable change = treated as insignificant in this table. Table 4.9:StudiesStudy ofInformation Community-Wide Pregnancy or HIV Prevention Initiatives with Multiple Components continued Sample Description Study Results Publication Date Program(s)Author(s) / / SampleSES (N)Location / Post / GenderEthnicity Grade/ / Age / Program Description Design MethodsAnalytic Change in Outcome' CommentsAdditional 1 1 I Grossman,Plain Talk Pepper Atlanta,New GA; 14-16=41%12-13=32% I1 Setting: 3 communities Cross-sectionalQuasi-experimental. survey data were regression.Logistic 1 sex:Use of birth control at first I design,This was because a challenging it 1999 SanOrleans, Diego, LA; 17-18=27% I needconsensusContent to protect Designed among sexually adults to: (I) activeabout create theyouth a andcollected again before 3 years the later intervention (I site) or 4 Girls:All: 0 0 1 theattempted impact toof measurethe cam- CA Not reported 1 Boys: 0 paign upon all sexually Low SES M=50% I1 (2)by encouragingprovide parents contraceptive and other use; commu- years later (2 sites). ! Use of birth control at last active youth in the corn- N=1,268 B1.5F=50% skillsnity adults to communicate with the knowledge more effectively and sex: All: 0 I thosecampaign.munities, who observed not just uponthe His=40% =8% with teen about sexual behavior and Girls: 0 1 Oth=2% 1 contraception; and (3) improve access Boys: 0 1 The strength of the evi- contraception.to reproductive Ahealth clinic care,was openedincluding in I Pregnancy: All: 0 1 betweenthedence long was periodpre reduced and of post time by mentedthird.aopenedone clinic community; Community increasedinand the reproductive evening an itsevents adolescent hours in a werehealthsecond, in the clinicimple- infor- and Girls:Boys: + 0 tookmanycontrolsurveys place other welland both changes inabilityfor thenation- tothat mation provided. I thatally andtime locally (e.g., changesduring in ! cationlocal school programs, health state- edu- 1 welfarepreventionwide pregnancy reform, initiatives, national I rates).nationalteen sexual Pre teen andbehavior, pregnancy post sur-and ! veys were also con- 1 ofducted the year.pregnancy at different The analysis rates times I1 pregnancydecreasesdid not adjust inrates. national for ) 1 Change in Note: For all studies, the significance level was set at p < 0.05.1f authors reported significant findings at the p < 0.10 level, they wereoutcome for group receiving intervention: no significant change = 0; significant desirable change = +; significant S'undesirable 0 change = treated as insignificant in this table. Table 4.9:Studies of Community-Wide Pregnancy or HIV Prevention Initiatives with Multiple Components continued Study Results Study Information I Sample Description Age / Publication DateProgram(s)Author(s) / / I Sample (N) .LocationSES // Post I GenderEthnicity / Program Description Design MethodsAnalytic Change in Outcome' CommentsAdditional Poder Latino Boston, MA, Mean=16.9 Setting School and community 1 Quasi-experimental. Multivariate Initiation of intercourse: This was not a strong and years 1 regression and Males: + design, because basically 1 Areal probability samples selected I Sellers,1994 McGraw, McKinlay Hartford, CT Not reported Sessions: I8-month program. in the inner-city areas of Boston sionlogistic were regres- used Females: 0 1 beingtwo different compared cities and were N=536Low SES His=100%Not reported nizations,educatorsContent Workshops andin schools, health centers; community run by group peer orga- 1 intervention).. Hartford(with the (withoutintervention) the and betweendifferencesto control the for ingHaving the previousmultiple partners6 months: dur- Males: 0 I outcomescontributedother factors (or to could lackthese ofhave PuertoRican= 94% andlargediscussions street community corner in homes; canvassing;events; presentations door-to-door provision at I afterInterviews the intervention conducted (18 beforemonths and Theretwo samples. were Frequency of sex in previous 6 Females: + weremanyoutcomes). controlledbaseline However, differences statisti- posters;of condoms; quarterly radio andnewsletters. TV PSAs; Ilater). intheencespre-test whether two between groupsdiffer- they Usemonths: of condom during last Females:Males: 0 0 i ofSamplestimetally, youth wasand (sexually sizeschangemeasured. for oversubsets inex- AIDS.someoneSTDhad ever or knew had with an intercourse: Females:Males: 0 0 '- inexperiencedwereperienced small at(e.g., pre-test) males). 89- separatelyanalyzedfemalesMales were and Proportionused: of time a condom Females:Males: 0 0 1 The percentage of programsAIDSwhoyouth had workshops participated was similar and in 1 condomsyouthboth cities, had in received butBoston. more free

1 1 ) 1 Change in Note:For all studies, the significance level was set at p < 0.05.1f authors reported significant findings at the p < 0.10 level, they wereoutcome treated for group receiving intervention: no significant change = 0; significant desirable change = +; significant undesirable change = as insignificant in this table. Table 4.9:StudiesStudy ofInformation Community-Wide Pregnancy or HIV Prevention Initiatives with Multiple Components continued Sample Description Age / Study Results Program(s) / 1 Publication Date Author(s) / SampleSESLocation (N) / Post / Gender Grade/ / I Program Description Design MethodsAnalytic Change in Outcome' CommentsAdditional Ethnicity I Portland, OR 12-21 years Quasi-experimental. Used regression 1 This was a challenging 1 Polen,Project Freeborn ACTION I Median=15- Setting: Community. and logistic Initiation of intercourse:0 I design, because it rrl reportedNot 16 years (I)Content a community The intervention mobilization included: to localCross-sections agencies serving of teen youth clients at of assessregression change to Acquiredmonth: a 0 condom in the last 1 attempted to measure 074 co 1995 N=2,212 Not reported effortsincrease to communityincrease teen acceptance condom use;of viewedhigher risk before of STDs the intervention were inter- over time, con- : Used a condom at last inter- I: youth-servingpaignthe impact upon ofall agencies,the youth cam- in 04 )1.z 1 F=42%M=58% airing(2) a media on TV; campaign (3) the installation involving 3 ofPSAs 240 forbegan about.2 and continuouslyyears. thereafter trollingder, forage, gen- race, course with main partner:0 I not just upon those who !"! B1=35%Wh=42% recommendedcondom vending by machines youth; and in (4)locations a PostBaseline:N=508 intervention:N=1,704 andsexual number partners. of ! coursepartner:Used with a condom +other or at casual last inter- I wasTheobserved validityreduced the of campaign.by the the study lack CD I Am Ind =4% His=5% teen skills-building workshop program. someof random differences assignment, among 5.)(1" )' I Asn=2%Othr=12% andtionalthe differentthe samples relatively cross-sec- of smallyouth, cro 0_.213- partners.whosample had sizes sex It was withfor youthnot casual 3o I condomshort-termpossible touse increaselink with the the in Cra 30-3-s-c) PSATeenscampaign. or familiar logo were with not the CD077 notdommore familiar. than likely those to Attitudes use teens a con- CD0_ condomself-efficacytoward condoms use regarding improved. and SQ (D ) 1 Change in Note: For all studies, the significance level was set at p < 0.05. If authors reported significant findings at the p < 0.10 level, theyoutcome for group receiving intervention: no significant change = 0; significant desirable change = +; significlant undesirable change = were treated as insignificant in this table. 1 9 4 Table 4.9: Studies of Community-Wide Pregnancy or HIV Prevention Initiatives with Multiple Components continued Study Results Study Information Sample Description .. Age / Program(s)Author(s) / / SESLocation / Post / 1 GenderGrade / / Program Description Design MethodsAnalytic Change in Outcome' CommentsAdditional (ResponsibleRESPECT Publication Date SamplePAPhiladelphia, (N) 14-18 yearsEthnicity theirSetting: communities. 9 family planning clinics and Quasi-experimental. sionLogistic was regres- used Initiation of intercourse:0 impactThis study upon measured the entire the Education on Sexuality Cross- Not reported 1 Sessions:Not reported. entireRandom catchment samples areas of ofteens the from the age,to control race, and for Frequency of sex: 0 justtargeted upon population, those actually not Teens)Everyand Pregnancy Community's for citysection of Not reported hours,expandedContent began after-school Clinics teenage initiated walk-in or evening hours,or I citysamplesclinic excluding were of comparedteens the catchmentfrom with the entirerandom areas. overmeasuregender, time and change in tothe PregnancyUse of contraception:0 rates:0 outcomeareceiving much more services;criterion. demanding this is TeitlerHughes, Furstenberg, N=1,96I appointments,reserveddecreased for theand teenagers average increased only.waiting the They hours time also for I andbeforeTelephone about the 2.5 programinterviews years waslater. with initiated teens theversuscatchment remainder that area in Birch rates:0 component)nentsSome (e.g., program the may mediacompo- have appearsNote:This1995 in Table study 4.6 also wellschooltrainedIn as addition, and athe media communitystaff thecampaign. to initiativework activities, with conducted teens. as Wave 21I pre-testpost-test: N=907N=680 N=I17 of the city. groupreached as thewell. comparison School/CommunityProgram for Sexual in SouthRural county 14-17 years Setting: K-I2 classrooms. AnnualWaveQuasi-experimental. 2 pregnancy post-test:N=257 rates for 14- to proportionalZ statistic for Pregnancy rates: + Thisdesign, was a becausechallenging it TeensRisk Reduction Among CarolinaLow SES I F=100%Not reported instruction.Sessions: Integrated into other mated17-year-old for the years females 1977-1988 were esti- waspregnancydifferences used torates in gramtheattempted impact upon toofall measure theyouth pro- in Vincent,Schluchter Clearie, 4,430,N=3,800 to I Wh=42% 131=58% 1 designedtraining.schoolContent staff ClassroomAboutto wereincrease two-thirds given instructionknowledge, sex education of thewas deci- community;countyfor: (I) thesurrounding western (2) the easternthepart program of thepart of ventioncomparecomparison and inter- the receiveduponcommunity, those interventions. who not just Green,VincentKm,1987 Dunteman, George, dependingupon year I; skills,sion-making and self-esteem, skills, communication and to align val- tlie tiessoncounty servinggroup; serving andas comparison as(3) a3 compari-similar groups. coun- groups. Afterratesimplemented, the declined;program pregnancy afterwas parts 1994 IIues with those of theanduponfocus community assuming problem-solving,was not personalalways The onrisk responsibility. sexuality assessment, but Comparison:N=rangeIntervention: from 319 to 333 topregnancyof thetheir program previous rates ended, returnedlevel. 1 Peer ;education was included.nurse School provided consultation, condoms, N=range from 391 to 1,630 yearsThe small and thenumber small of num- ; churchesclinic.and transportation Community implemented groupsto a classes family and planningand validityfemalesber of 14- of limit these to Ithe 7-year-old findings. internal I; thepapers,special radio. events.and announcements Articles appeared were inon isolationtionThe andvery theofsmall the geographical popula-county mostofItlimit isthe notits important. program externalclear what were validity parts ) 1 Change in outcome for group receiving Note:intervention: For all studies, no significant the significance change = level0; significant was set atdesirable p < 0.05. change If authors = +; reported significant significant undesirable findings change at the = p < 0.10 level, they were treated as insignificant in this table. ou Table 4.9:Studies of Community-Wide Pregnancy or HIV Prevention Initiatives with Multiple Components continued i Study Information Ii . Results 1 Sample Description i Study :1 1 Age / Program(s) / 1 Location / Author(s) / SES / Post .I GenderGrade / / . Program Description Design Analrtic i Change in Outcome' Additional Publication Date I Sample (N) I Etnnicity . Methods . Comments 1 3 Kansas I I I Ever had sex: I SexualSchool/Community Risk Reduction 'I communities: iI Not reported clinicsSetting:Communities and organizations and within schools, them. I The intervention zipQuasi-experimental. codes were forChi-square behavioral tests I Geary: + I design,This was because a challenging it Not reported 1 Paine-Andrews,Replication Initiative Harris, Franklin,Geary, I Not reported include&Content:Community-wide enhanced sexuality initiatives education I matched with other zip codes to z-statisticdata; an adjusted for I Age at first sex: Franklin:- I theattempted impact to of measure the pro- : I form comparison groups 1 Fawcett,VincentFisher, Lewis,Williams, , WichitaMixed SES Geary: priatefor teachers sex education and parents, (K-12), age-appro- increased I In Geary and Franldin, question- birthpregnancy rate data. and Franklin:Geary: 0 0 I thegram community, upon all youth not justin 1999 I Wh=66% B1=23% access to health services, collaboration I schoolsmire indata the were intervention collected areas in I Condom use: I receivedupon those interventions. who I As=4%His=6% masswith school media, administrators, greater involvement use of of I before and after the intervention. - Geary:0 . I1 It employed a th I Franklin: community in teen pregnancy preven- I In all these communities, birth rate I . Franl din: + I change evaluation strat- eory of don, peer support and education, 1 I Wh=97% His=2% alternative activities for youth, and I yearsdata were before collected the intervention for up to and5 Pregnancy: Geary: 0 Ii categy Pcess data indi- ed numerous school ro I Bl= I% involvement of the faith communities. I vention3 years and during, comparison in both zip the codes. inter- I, , Franklin: 0 I Conclusionsand community are changes. limited I Wichita:Not Geary: Pre-test N=1,004 Births: Wichita: 0 I tionsby various and results design that limits- dif- . reported Post-test N=952 - i fered across the sites. Franklin: Pre-test N=7I 0 I1 andChanges birth inrates pregnancy were not Post-test N=817 I andconsistently never statistically favorable, sig- I codes.favornificant, the but intervention tended to zip

, ) 1 Change in outcome for group receivingNote: intervention: For all studies, no significant the significance change level = 0; was significant set at p desirable < 0.05. If changeautho.rs = reported +; significant significant undesirable findings change at the p = < -. 0.10 level, they 197 were treated as insignificant in this table. 100 Table 44Study 0: Information study of tn E nly Childhood Pr gram 1 Sample Description 1 . Study Results 1 i Program(s) / Location / 1 Age I i 1 Grade / i Analytic Additional SES / Post 1 Design ' Change in Outcome' Publication Date Author(s) / Sample (N) i i GenderEthnicity / 1 Program Description Methods Comments 1 Infancy I Setting: Pre-schools and elementary Experimental. General linear Birth rates: + 1 The preschool treatment Abecedarian Project reportedNot I through 21 i schools. 1 model for con- I group delayed childbear- Campbell Low SES years I1 Duration: i1 toChildren 4 groups: were preschool randomly plus assigned ele- cometinuous variables. out- I Bothing by treatment more than groups a year. 1999 N=104 Not relevant 1 PreschookBoth: 8 years 5 years i treatmentmentary treatment, only elementary preschool treat- Chi-square I also performed higher 1 I I analyses. on intellecutal and aca- F=49%M=51% Elementary:3 years I ment only, and control. I educationaldemic measures level and 1 Children were followed to age 21. I Oth=2%B1=98% Preschool:Full-dayContent child care all year I1 Questionnaire and survey data col- I attained. Elementary:from infancy Ato home-school kindergarten. resource lected near 2Ist birthday. I elementaryBoth the preschool interventions and involvementteacher strove in child'sto increase learning parent and to i I preschoolcant,were but,statistically in treatment general, signifi- the givenparentsconvey for issuesand parents school; of concernand curriculum their between children. packets I strongerappeared impact. to have the

. r 199 ) 1 Change in outcome for group receivingNote: intervention: For all studies, no significant the significance change = level 0; significant was set at desirable p < 0.05.1f change authors = +; reported significant significant undesirable findings change at the = p. < 0.10 level, they were treated as insignificant in this table. Tabie 4.0Study Information : ti ides of ServiceSample Description Learning Programs Age / 1 Study Results Program(s) / Location / 1 Publication Date Author(s) / SampleSES (N) / Post GenderEthnicity Grade/ / I Program Description Design MethodsAnalytic Change in Outcome' CommentsAdditional 1 Seventeen i 1 Learn and Serve I Not reported Setting: Middle and high schools. Quasi-experimenal. ANCOVA, con- Pregnancy: 0 1 This strength of this MelchiorAmerica 1 incommunities U.S. 9th-12th=70%7th-8th=30% included:Content Service(I) "meaningful" learning service projects in similarComparison classes groups in the weresame otherschool ences,baselinetrolling and for differ- 1 effects.possibledesign was self-selection weakened by 1998 M=40% the community (e.g., tutoring, working or classes in other schools. ANCOVA using 1 F=60% as a teacher's aide, working in a nurs- Questionnaire data were collected change scores. 1 There was an impact Wh=58% meaning home, of 77 or hours; homeless and (2)shelter) structured for a at baseline, after program partici- I1 p=.10upon pregnancy level at post-test, at the B1=17%His=19% writing,time for papers,reflection and (discussions, group presenta- journal onepation year (end after of schoolparticipation year), (endand 1 follow-upbut not the test. one-year Oth=6% tions). Both parts are linked to the of following school year). 1 academic curriculum. Intervention group: N=608 1 gramAt post-test, also had the a small pro- programsOnly well-designed were selected. and high quality Comparison group: N=444 1 educationalpositive impact attitudes upon and 1 some measures of 1 school performance, but 1 these did not last I year.through the following CommunityReach for Health Youth and New York 1 age=12.7Mean Setting: Middle schools. 1 experimental.Part experimental: part quasi- controlMIXOR forused to 1 CYS & RFH vs. Control Had sex in last 3 months: + designThe strength was reduced of the by Service Learning Low SES years Sessions: 40/year for Reach for 1 baseline scores. lack of random assign- N=1,061 1 Health, plus 3 hours/week for commu- I Thirty-five classrooms of students Adjusted for RFH vs. Control ment to the control Duran,O'Donnell, Haber,Atnafou, Stueve, Doval, 8th=48%7th=52% nity service. I wereReach randomly for Health assigned plus to community receive clustering. Had sex in last 3 months:0 modestgroup, small long-term sample sizes, 1 Content: Community Youth Service Johnson, Grant, Murray, 1 service or only Reach for Health. follow-up, and an 1 M=47% (CYS) included service in nursing Juhn,Tang, Piessens I F=53% homes, neighborhood clinics, child day 1 Anothercomparison school school.was matched as a icanceabsence for of some tests ofout- signif- (See1999 next entry) B1=79% ter. locations.careEach centers, student Students andwas a assigned seniordebriefed citizen to experi- 2 cen- 1 lectedMatched in fallquestionnaire (baseline) and data spring col- comes. ences. 1 (follow-up). opposeduponThe greatest 8th gradersto 7th effect graders). (as was Reach for Health (RFH) focused on 1 RFH & CYS: N=255 behavior.substance use, violence, and sexual 1 Comparison:RFH: N=222 N=584 cialalsotheTrend greatesteducation upon data youth indicated impact classes. in spe- was that ) 1 Change in outcome for group receiving intervention:Note: For noall studies,significant the change significance = 0; significant level was setdesirable at p < 0.05.1fchange authors = +; significant reported undesirable significant findings change at= the p < 0.10 level, they were 201 treated as insignificant in this table. 202 Table 4.11:StudiesStudy ofInformation Service Learning Programs continued Sample Description Study Results Program(s) / I SESLocation / Post / Grade / Age / Analytic Additional Reach for Health and Publication Date Author(s) / SampleNew (N) York Not Ethnicity Setting: Middle schools. Program Description Experimental. Design Logistic regres- Methods Initiation of sex: +Change in Outcome' 1 A few students trans- Comments IGender1 / ServiceCommunity Learning Youth Low SES 7th=reported 100% communitySessions: 3 service hours/week or 90 forhours 30 weeksalto- wereEighteen randomly classrooms assigned ofto studentsreceive forsion, the adjusted cluster- Had sex in last month: + 1 ferredprogram into or because out of theof O'Donnell, Stueve,Duran, Doval, N=I95 M=41% gether. serviceReach foror onlyHealth Reach plus forcommunity Health. ining classrooms. of students 1 schedulingexpanded resourcesconflicts. or PleckWilson, Haber, Perry, B1=71%F=59% i zen(CYS)Content homes, included nursing CommunityYouth servicehomes, inhealth senior Service cen- citi- ' collectedMatched in fall questionnaire of the 7th grade data were 0 (SeeUnpublished previous (2000) entry) Oth=3%His=26% I Studentsstudentters, and was debriefedchild assigned day careexperiences. to centers. 2 locations. Each grade(baseline) (follow-up). and spring of the 10th behavior.substanceReach for use,Health violence, (RFH) andfocused sexual on Ln (TOP)ProgramTeen Outreach Study150 #1:sites Stud);1 1-19 #1: years manySetting: sites Schools throughout and communities the country. in Quasi-experimental.Study #1: controllingwereTwo groups compared, for Study #1: Pregnancy rates: + design.strongThe second experimental It confirmed study had the a tri Philliber,Allen1992 (#1) StudyMixednationwide SES #2: NothighMiddle reportedschools and values,smallContentSessions: group decision-making, 2 Weekly. major classroom components: discussions communica- (1) of staff,TOPComparison participants,TOPor random group assignment. selected or school by andlemencesbaseline behaviorsother in differ-prob- char- Study #2: Pregnancy rates: + studyweakerresults measuredwhich founddesign. had in Neitherthelong-a first Allen,Kuperminc Philliber,1997 (#2)Herrlin, Mixednationwide25 sites SES StudyNot reported #2: e.g.,teervolunteertion skills, serviceserved experiences;parenting, asin aidesschool lifein orhospitalsand options, community, (2) volun- andand schoolprogramcollectedMatched year). (beginning questionnaireat intake and and dataexit end from wereof acteristics. andreducedTheterm school programresults. course suspensions. also failure 9th=36%MeanIlth=20%10th=33%15.8 age: years service,activities.walkathons,nursing homes,but A meanminimumpeer participeted tutoring,was of46 20 hours. and inhours other of StudyComparisonIntervention #2: post-testpost-test N=2,624N=3,032 F=85%M=15%12=11% ofcollectedMatchedExperimental. school questionnaireduringyear and first end several dataof school wereweeks Oth=3%His=11%Wh=BI=67% 1 9% ComparisonInterventionyear. pest-testpost-test N=342N=353 263 Note:For1 Change all in studies,outcome the for significancegroup receiving level intervention: was set at pno < significant0.05. If authors change reported = 0; significant significant desiMble findings change at the p= <+; 0.10 significant level, they undesirable were change = treated as insignificant in this table. Tabie 40Study 2: Information Studies ofVocationaiProgram(s)Author(s) Education / / and Empioyment Programs SESLocation / SamplePost / Description Gender Grade/ / Age / Program Description Study Design MethodsAnalytic Change in Outcome' Results CommentsAdditional Publication Date Sample (N) Ethnicity I Conservation and Four cities 18-25 years Setting Communities. Experimental. Not reported. Unmarried pregnancr 1 The evaluation had a CD Jastrab,Youth Blomquist, Service Masker, Corps thethroughout country Out of school average.Duration: About 4-5 months on assignedIndividual toyouth the corpswere randomlyor to BlackOverall: females: 0 + 1 wasimpactstrong not in design,uponthe same pregnancy but the 014 a Orr1997 reportedNotLow SES M=57%F=43% tionalContent and Provides academic remedial, education, voca- and baseline;Interviewnothing. telephonedata were interview collected data at WhiteHispanic females:0 females: 0 groups.direction for all ethnic B1=50% work experience within the context 1 Intervention group had Wh=14%His=25% andcommunityof community other activities). service service and Corps (about 20% memberseducation 80% Samplecollected size not reported. I 5-months later. I suchother aspositive lower outcomes,arrest Oth=6%As=S% Communitywork full time projects in small last teams. 2-12 weeks, i. paidrates employment and higher rates of projects.housing,homes,e.g., provide help or complete assistancerenovate environmental low-income to nursing

) 1 Change in Note:For all studies, the significance level was set at p < 0.05. If authors reported significant findings at the p < 0.10 level, they wereoutcome for group receiving intervention: no significant change = 0; significant desirable change = +; significant undesirable change = 205 treated as insignificant in this table. 206 Table 4.12:StudiesStudy InformationofVocational Education and Employment Programs continued Sample Description Study Results Program(s) / SESLocation / Post, / I Grade / Age / Program Description Design Analytic Change in Outcome' Additional Publication Date Author(s) / Sample (N) 1 Gender / Methods Comments SchochetJob Corps nitiesI 19 in commu-the 18-19=30%16-17=44% Ethnicity non-residentialSetting Mostly Jobresidential, Corps settings. but a few A randomExperimental. sample of youth eligible t-tests.Chi-square and Birth rates: 0 strongThis was experimental an extremely GlazermanBurghardt LowU.S. income Not20-24=26% reported siveContent: program An intensive, whose major comjarehen- service : wideto participate were selected in Job for Corps the study.nation- sampleAppropriate weights. randomU.S.design: eligible randomassignment participants, sample to of 2000 N= I 1,787 B1=49%F=43%M=57% counseling,don,living,components vocational health and training, jobincludecare, placement health academicresidential education, assis- educa- participateControlJobThey Corps were group andrandomlyin other members control programs assigned groups. could to births.up,group, and long-term measurement follow- of His=17%Wh=26% I participationtance. Mean wasnumber 8 months. of months Mean of (mean is 637 hours of programs). membersMany control partiCipated group in Oth=7% I vocationalnumber of instructionhours of academic was greater and months.baseline,Interview 12 months, data were and collected 30 at programs.other similar types of than 1,000. : ControlIntervention group: N=4,476. group:N=7,311 Jobled Corps to more participation GED com- employmentCorpsafterpletion, 2 andyears, higher higher duringafter lowerearnings 2 Job tionlowerpublicyears, rates, assistance,arrest lower but andreceipt most andconvic- of attendancecantThereeffects changes were were and novery in signifi-collegesub- small. monthmuchNotestance the olderuse.follow-up sample at the survey.was 30- 208 267 ) 1 Change in Note:For all studies, the significance level was set at p < 0.05. If authors reported significant findings at the p < 0.10 level, they wereoutcome for group receiving intervention: no significant change = 0; significant desirable change = +; significant undesirable change = treated as insignificant in this table. Table 4.12:StudiesStudy Information of Vocational Education and Employment Programs continued Sample Description Study Results Program(s) / Location / Age / I Publication Date Author(s) / Sample (N)SES / Post GenderEthnicityGrade / / Program Description Design MethodsAnalytic 1 Change in Outcome' CommentsAdditional 1 " JOBSTART 13 cities in 16-19=73% 1 Setting: Community organizatiohs, Experimental. t-tests. Non-custodial women: .This was a strong experi- Cave, Bos, Doolittle, U.S. 20-21=27% I schools, and Job Corp settings. Youth who applied for the pro- BirthPregnancy rates:0 rates:0 mental design. Toussaint Low income School Content: Included individualized gram were randomly assigned to 1 JOBSTART led to 1993 N=1,941 F=54%M=46%dropouts seling,(i.e.,occupationalinstruction transportation, life skills),in basictraining, and academic child job support placementcare, skills, services coun- Surveyvicesgroupparticipate from could data other or andwere not programs.did collectedThe receive control at ser- ' Custodial mothers: BirthPregnancy rates: rates: 1 STARTincomepletionincreases andparticipation.during in lower GED JOB- earnedcom- Oth=3%Wh=9%His=44%B1=44% activities.assistance. Mean of 400 hours of ControlInterventionbaseline, group:N=496 12, group:24, and N=533 48 months. custodialamongincreaseThere was women in mothers achildbearing significant who when Were todialthosebutthey no entered mothers whodifference were (about when notamong half),cus- follow-upmuchNotethey entered.the survey.older sample at the was 4-year (STEP)EducationSummer Training Program and I 5 urbanU.S.areas in the 15=43%14=57% jobs.Setting Classroom and part-time RandomExperimental. assignment to the STEP Not reported. IUse1 of contraceptives:Sexual 0 activity: 0 assignment,Therigorous; evaluation large it had was sample random very Walker,Vilella-Velez1992 behind)(academicallyLow SES, F=52%M=48%Not reported Content:summers.Sessions: Life 36 sessionsskills education over 2 in such Matchedduringprogram the questionnaireor summer. to a guaranteed data werejob 1 Births: 0 follow-up.recognizedsizes, and It should long-term that thebe con- Grossman,1992 Sipe 1 N=4,800 His=18%Asn=19%BI=49% makingIncareers,areas sexuality, as and sexualand theit community focused behavior,importance on involvementdrug decision- of use, respon- Control:N=2,400Treatmentcollected annually N=2,400 for 5 years. half-timetimetrol summer.group jobs jobs) (ratherreceived during than full- the Othr=14%Wh/ of(half-time)sible academicsupport behavior. atduring recitation,minimum Ninety the school wage,andhours 5-15 years. 90of workhours ) 1 Change in Note:For all studies, the significance level was set at p < 0.05.1f authors reported significant findings at the p < 0.10 level, they wereoutcome treated for group receiving intervention: no significant change = 0; significant desirable change = +; significant undesirable change = 209 as insignificant in this table. 210 TabDe 400 3: Studes of Otherlf uth PeveDopment Progra Sample Description Study Results Study InformationPublication DateProgram(s)Author(s) / / Sample (N)SESLocation / Post / GenderGrade / / Age / Program Description Design MethodsAnalytic Change in Outcome' CommentsAdditional San Antonio, Not reported Ethnicity Setting: Community youth-serving Experimental. Means for 2 I Birth rates: 0 Aside from the relatively OpportunitiesQuantumHahn,Program Leavitt, Aaron PA;Philadelphia,TX;Milwaukee, 9th Notgraders reported tiesagenciesContent:Included (e.g., tutoring and educationalcomputer-based activi- mentwereIn each randomlyand of controlthe 5 assignedsites, conditions. 50 studentsto treat- groupscompared. were smalldesign.was sample a strong size, evaluation this 1994 OldahomaandSaginaw,WI; MI; Not reported ties,Participantscareer,instruction), and development and college receivedcommunity planning).activities small service stipends (arts, activi- collectedFollow-up 6 questionnaire times over 4 datayears. were levelbirthsignificantWhile ratesfor there the effectsat firstwere the 3p=.05upon noyears, City,N=149Low OK SES 'andand completion bonus payments of activities. for participation ratespositiveafter 4at years theeffect p=.09 there on level.birth was a SeattleKosterman,Hawkins,Development Social Catalano, Abbott, Project Hill Seatde,WASES=57%Low M=50%Not reported improvingattachmentContentSetting: ElementaryDesigned teaching to school strategiesto schools.and increase family and by par- andrandomlyEighteenQuasi-experimental. control; schoolsassigned classrooms were to treatment non- within chi-squareuousanceAnalysis variables;for ofcontin- testvari- Pregnancy:NumberEver had ofsex: +sex + partners: + assignmentthedesignThe lack strenith was of randomweakened of this by 1999 N=598 I Oth=56% Wh=44%F=50% 6:makingchildren'senting 5 days skills. and socialof in-servicerefusalAlso skills designed skills). (e.g., training decision-Grades to increasefor I- Interviewtosome treatment schools data and randomlycollected control. assignedat age 18. earLogisticmousfor dichoto-regression measures. and lin- acts,suchotherIntervention less aspositive fewer drinking, group outcomes,delinquent less had ingandforteachers parentsfor 5-6, children eachand of social year,children in grades parentingcompetence in grades1-6. classes train-1-3 also used. achievementandattachmentschool higher misbehavior, academic to school, more 1 Change in outcome for group receiving intervention: no significant change = 0; significant desirable change = +; significant undesirable change = ' 2 Note: For all studies, the significance level was set at p < 0.05. If authors reported significant findings at the p < 0.10 level, they were treated as insignificant in this table. Table 4. ti Study4: Studies Information of VI ki-Cornponent Programs with Both Sexuality and Sample Description Study Youth Development Components Results Publication Date Program(s)Author(s) / / SampleSES (N)Location / Post / I Gender Grade/ / Age / Program Description Design MethodsAnalytid Change in Outcome' CommentsAdditional I Ethnicity California's Adolescent 16 locations Mean=13.7 1 Settings: Multiple settings, including Quasi-experimental. Analyses statis- Initiation of sex: + The strength of this PreventionSibling Pregnancy Program Notin California . I MeanYears serviceschools, agencies. health departments, and social certainAll study criteria. participants Comparison had to meet differencestrolledtically con- for in Frequency of sex:0 mentlackdesign of and randomwas possible weakened assign- self- by 2000East, Kiernan N=1,270reported F=60%M=40%grade=8.2 MultipleContentSessions:An strategies All 44 average sites were wereof used, 18.4 different e.g.,hours. found(wereselectedgroup deemedthroughmembers from the outreachat were lowerwaiting often risk) efforts. list and pationandcharacteristicsbackground for in partici- other use:ConsistencyNumber 0 of sexual of contraceptive partners:0 tionselectionhigherHowever, youth effects. riskappeared the on interven-some at I Wh=11%B1=11%H is=68% remainincreaseservices.individual in self-esteem, Programs caseschool management or werereturn help designedyouth to and school, group to later.collectedMatched questionnaireat baseline and data 9 months were services.non-ASPPP Youthindices. who received Oth=10% increasehealthhealthyimprove and decisions, knowledgecommunication reproductive improve and services, skillswith access parentsto makeand to ComparisonIntervention post-test N=82IN=450 moregreaterlesshours likely numberslikely of toservice touse initiate of con- were sex, I use,toand delay adults.and sex,decrease They increase were risk contraceptivealsobehaviors designed asso- possibletotraception, becomedifferences self-selection andpregnant, lesswere likely not but I recreation.ingciated and with drug teen use). pregnancy Programs (e.g., included drink- statistically controlled.

) 1 Change in Note: For all studies, the significance level was set at p < 0.05.1f authors reported significant findings at the p < 0.10 level, they wereoutcome for group receiving intervention: no significant change = 0; significant desirable change = +; significant undesirable change0 01 = 10 0 treated as insignificant in this table. 214 Table 4.14:StudiesStudy Informationof Multi-Component Programs with Both Sexuality and Youth Development Components continued Sample Description Study Results Publication Date Program(s)Author(s) / / i SampleSESLocation (N) / 1Post / 1 EthnicityGrade / Age / / Program Description Design MethodsAnalytic Change in Outcome' CommentsAdditional 1 1 Children's Aid Society 1 New York, 13=36% 1 Setting: Community organizations Experimental. Lotstic regres- Ever had sec This was a very rigorous Carrera Program 1 NY 14=37% serving youth. 1 sion, separate All: + study with multiple sites, IGender 1 In each of the 6 sites, youth were 1 15=26% for each gender, Girls: + random assignment, a Philliber, Kaye, Herr ling, Low SES 1 Sessions:5 days a week during the 1 randomly assigned to treatment controlling for a Boys: 0 large sample size, long- West 1 N=484 Not reported 1 school year, and special sessions in the 1 and control conditions. Some few baseline term measurement, 2000 M=45% 1 summer. Mean number of hours of 1 control group members received a characteristics. 1 Condom use at last sex: measurement of behav- F=55% 1 hoursparticipation per month. for first 3 years was 16 1 recreation.few after-school services, typically Girls:All: 0 0 positiveior, and, amongfindings females, for sex- BI=56% 1 1 Boys: 0 ual behavior and preg- His=36% 1 Content This was an intensive pro- 1 collectedMatched questionnaireat baseline and data annually were 1 Use of condom plus an effec- nancy (as opposed to 1 gram lasting through high school. It 1 Oth=2%BI&His=7% 1 toused develop a holistic close approach, relationships and staff with tried reported.for 3 years. Three-year data I tive method of contraception: All: 0 pregnancy).only behaviors affecting i youth. It included 5 components: 1 Girls: + Notably, the findings are 1 (I) family life and sex education; (2) an Intervention 3-yr: N=242 Boys: reported for all the Control 3-yr: N=242 1 1 individualeducation academic component assessment, that included tutor- Pregnancy: All: + mentmembers and controlof the treat- 1 ing, help with homework, preparation Girls: + groups, even though :1 withfor standardized college entrance; exams, (3) and a assistancework- Boys: 0 treatmentsome members group of did the not 1 related intervention that included a Births: participate much and 1 job club, stipends, individual bank All: 0 some members of the 1 awareness;accounts, employment, (4) self-expression and career through Boys:Girls: 0+: patedcontrol in group some partici-services. 1 addition,the arts; andthe program(5) individual provided sports. mental In Positive effects upon sex- 1 health care and comprehensive medi- comesual behaviors were limited and out- to cal care, including contraception. tivetheretogirls boys. effectsand were Indid otherafor notfew both extend areas;posi- girls someaccounts,experience,and boys health (e.g.,and bank services). receipt work of 215 ) 1 Change in outcome for group receivingNote: intervention: For all studies, no significant the significance change = level 0; significant was set at desirable p< 0.05. change If authors reported significant findings at the p < 0.10 level, they were = +; significant undesirable change = treated as insignificant in this table. Table 4.14: Studies of Multi-Component Programs with Both Sexuality and Youth Development Components continued Study Information Sample Description ,1 Study Results Age / Ii Program(s) / Location / 1 I Grade / 1 Anal)ftic Additional. Author(s) / SES / Post 1 Program Description Design Change in Outcome' Publication Date Sample (N) GenderEthnicity / Methods . Comments Untitled I Washington Mean=15.4 I: Setting:There were 3 programs iI Experimental. t-tests for base- i Ever had sex: 0 The study evaluated 7 McBride, Gienapp stateLow SES years I planningwhich were clinic, administered a middle and in higha family Ii Individual youth were randomly Covarianceline differences, Sex in past month + 3programs/sites, measured impact but upononly 2000 N=690 M=10%Not reported i:I t.ing.school, One and was a community-based run by the health set- I onlyassigned. 2 hours Control of program group services.received model.adjustment month:Contraceptive 0 use during last Thisbehavior. was a strong exper- F=90% i department and two were run by fam- i Matched questionnaire data were . ily planning clinics. I:I collected at baseline and 6 to 9 Contraceptive use during last strengthimental design, was weakened but its i Oth=26%Wh=74% ' Sessions: Mean=27 hours, 1 months later at end of sex 0 by small sample sizes for Content: Employed a "client-cen- :Ii Interventionintervention. group: N=191 Always use contraceptives:0 some analyses. sexualwhythetered" service teens approachbehaviors, become providers' thate.g., involved is theybeliefs based lacic in about upon risky(1) I Control group: N=166 guidance.(4)whominformation emotional they The canabout suppom approach trust; sex; (3) and(2) addressescoping adults(5) positive skills; Ii theseindividualizedcationing and small andother group skill-building problemsservices and individualized such by with combin- as other coun- edu- i , advocacyseling, mentoring, tailored referrals,for each teen.and

) Note:1 ChangeFor all studies, in outcome the significancefor group receiving level was intervention: set at p < 0.05.no significant If authors change reported = 0;significant significant findings desirable at the change p < 0.10 = +; level, significant they undesirable change = . were treated as insignificant in this table. 9 18 References Bearman, P.S., & Bruckner, H. (2001). Promising the future: Virginity pledges and the Aarons, S.J., Jenkins, R.R., Raine, T.R., El- transition to first intercourse. American Journal Khorazaty, M.N., Woodward, K.M., Williams, of Sociology, 106(4), 859-912. R.L., Clark, M.C., & Wingrove, B.K. (2000). Postponing sexual intercourse amorig urban Blake, S.M., Ledsky, R., Goodenow, C., Sawyer, R., Lohrmann, D., & Windsor, R. (2000). junior high school students A randomized Condom availability programs in Massachusetts controlled evaluation. Journal of Adolescent high schools: Relationship to condom use and sex- Health, 27(4), 236-247. ual behavior. Unpublished.

Abma, J., Chandra, A., Mosher, W., Peterson, Blake, S.M., Ledsky, R., Lohrmann, D., L., & Piccinino, L. (1997). Fertility, family plan- Bechhofer, L., Nichols, P., Windsor, R., ning, and women's health: New data from the Banspach, S., & Jones, S. (2000). Overall and 1995 National Survey of Family Growth. Vital differential impact of an HIV/STD prevention and Health Statistics, 23(19). curriculum for adolescents. Washington, D.C.: Academy for Educational Development. Ajzen, I. (1985). From intention to actions: A Unpublished manuscript. theory of planned behavior. In J. Kuhl & J. Beckman (Eds.), Action control from cognition to Boekeloo, B.O., Schamus, L.A., Simmens, S.J., behavior (pp. 11-39). New York: Springer- Cheng, T.L., O'Connor, K., & D'Angelo, L.J. Verlag. (1999). A STD/HIV prevention trial among adolescents in managed care. Pediatrics, 103(1), Alan Guttmacher Institute. (1994). Sex and 107-115. America's teenagers. New York: Author. Brewster, K.L., Billy, J.O.G., & Grady, W.R. Allen, J.P., Philliber, S., & Hoggson, N. (1990). (1993). Social context and adolescent behavior: The impact of community on the transition to School-based prevention of teen-age pregnancy sexual activity. Social Forces, 71, 713-740. and school dropout: Process evaluation of the national replication of the Teen Outreach Campbell, F.A. (1999, April 16). Long-term Program. American Journal of Community outcomes from the Abecedarian study. In F.A. Psychology, 18(4), 505-524. Campbell (Chair), How high-quality early childhood programs enhance long-term development: A Allen, J.P., Philliber, S., Herrling, S., & comparison of finding's and models. Symposium Kuperminc, G.P. (1997). Preventing teen preg- presented at the Biennial Meeting of the Society nancy and academic failure: Experimental evalua- for Research in Child Development, tion of a developmentally-based approach. Child Albuquerque, NM. Development, 64(4), 729-742. Cave, G., Bos, H., Doolittle, F., & Toussaint, C. Alstead, M., Campsmith, M., Halley, C.S., (1993). JOBSTART: Final report on a program Hartfield, K., Goldbaum, G., & Wood, R.W. for school dropouts. New York: Manpower (1999). Developing, implementing, and evaluat- Demonstration Research Corporation. ing a condom promotion program targeting sex- Chamie, M., Eisman, S., Forrest, J.D., Orr, ually active adolescents. AIDS Education and M.T., & Torres, A. (1982). Factors affecting Prevention, 11(6), 497-512. adolescents' use of family planning clinics. Family Planning Perspectives, 14(3), 126-139. Anderson, J.E., & Cope, L.G. (1987). The impact of family planning program activity on Coyle, K.K., Basen-Enquist, K.M., Kirby, D.B., fertility. Family Planning Perspectives, 19(4), Parcel, G.S., Banspach, S.W., Collins, J.L., 152-157. Baumler, E.R., Caravajal, S., & Harrist, R.B. (forthcoming). Safer Choices: Long-term impact Bandura, A. (1986). Social foundations of of a multi-component school-based HW, STD, thought and action. Englewood Cliffs, NJ: and pregnancy prevention program. Public Prentice Hall. Health Reports.

THE NATIONAL CAMPMGN TO PREVENT TEEN PREGNANCY 162 219 Coyle, K.K., Basen-Enquist, K.M., Kirby, D.B., Fishbein, M., & Ajzen, I. (1975). Belief, atti- Parcel, G.S., Banspach, SW., Harrist, R.B., tude, intention, and behavior. Reading, MA: Baumler, E.R., & Weil, M.L. (1999). Short-term Addison-Wesley. impact of Safer Choices: A multi-component school-based HIV, other STD, and pregnancy Forrest, J.D., Hermalin, A.I., & Henshaw, S.K. prevention program. Journal of School Health, (1981). The impact of family planning clinic 69(5), 181-188. programs on adolescent pregnancy. Family Planning Perspectiljes, 13(3), 109-116. Coyle, K.K., Kirby, D., Marin, B., Gomez, C., & Gregorich, S. (2000). Effect of Draw the Fothergill, K., & Feijoo, A. (2000). Family plan- Line/Respect the Line on sexual behavior in mid- ning services at school-based health centers: dle schools. Santa Cruz: ETR Associates. Findings from a national survey. Journal of Unpublished manuscript. Adolescent Health, 27(3),166-169.

Danielson, R., Marcy, S., Plunkett, A., Wiest, W., Frost, J.J., & Forrest, J.D. (1995). & Greenlick, M.R. (1990). Reproductive health Understanding the impact of effective teenage counseling for young men: What does it do? pregnancy prevention programs. Family Family Planning Perspectives, 22(3), 115- 121. Planning Perspectives, 27(5), 188-195.

DeLamater, J., Wagstaff, D.A., & Havens, K.K. Furstenberg, F.F., Geitz, L.M., Teitler, J.O., & (2000). The impact of a culturally appropriate Weiss, C.C. (1997). Does condom availability STD/AIDS education intervention on black make a difference? An evaluation of male adolescents' sexual and condom use behav- Philadelphia's health resource centers. Family ior. Health Education & Behavior, 27(4), 453- Planning Perspectives, 29(3), 123-127. 469. Gillmore, M.R., Morrison, D.M., Richey, C.A., Dusenbury, L., & Falco, M. (1995). Eleven Balassone, M.L., Gutierrez, L., & Farris, M. components of effective drug abuse prevention (1997). Effects of a skill-based intervention to curricula. Journal of School Health, 65(10), 420- encourage condom use among high-risk, hetero- 425. sexually active adolescents. AIDS Prevention and East, P., & Kiernan, E. (2000). California's Education, 9(Supplement A), 44-67. Adolescent Sibling Pregnancy Prevention Gottsegen, E., & Philliber, W.W. (2000). Impact Program: Evaluating the impact of pregnancy prevention services to the siblings of pregnant and of a sexual responsibility program on young males. parenting teens. San Diego: University of Unpublished manuscript. California, San Diego Medical Center. Grossman, J., & Pepper, S. (1999). Plain Talk Edwards, L., Steinman, M., Arnold, K., & and adolescent sexual behavior. Philadelphia: Hakanson, E. (1980). Adolescent pregnancy pre- Public/Private Ventures. vention services in high school clinics. Family Planning Perspectives, 12(1), 6-14. Grossman, J.B., & Sipe, C.L. (1992). Summer Training and Education Program (STEP): Eisen, M., Zellman, G.L., & McAlister, A.L. Report on long-term impacts. Philadelphia: (1990). Evaluating the impact of a theory-based Public/Private Ventures. sexuality and contraceptive education program. Family Planning Perspectives, 22(6), 261- 271. Grunseit, A., Kippax, S., Aggleton, P., Baldo, M., & Slutkin, G. (1997). Sexuality education Ekstrand, M.L., Siegel, D.S., Nido, V., Faigeles, and young people's sexual behavior: A review of B., Cummings, G.A., Battle, R., Krasnovsky, F., studies. Journal of Adolescent Research, 12(4), Chiment, E., & Coates, T.J. (1996, July). Peer- 421-453. led AIDS prevention delays onset of sexual activity and changes peer norms among urban junior high Guttmacher, S., Lieberman, L., Ward, D., school students. Paper presented at the XI Freudenberg, N., Radosh, A., & DesJarlais, D. International Conference on MDS, Vancouver, (1997). Condom availability in New York City Canada. public high schools: Relationships to condom

Emerging Answers: Research Findings oh Programs to Reduce Teen Pregnancy 1 63 220 use and sexual behaviors. American Journal of Jemmott, J.B., III, Jemmott, L.S., & Fong, G.T. Public Health, 87(9), 1427-1433. (1998). Abstinence and safer sex: A randomized trial of HIV sexual risk-reduction interventions Hahn, A., Leavitt, T., & Aaron, P. (1994). for young African-American adolescents. Journal Evaluation of the Quantum Opportunities of the American Medical Association, 279(19), Program (QOP): Did the program work? 1529-1536. Waltham, MA: Brandeis University, Center for Human Resources. Kahn, J.G., Brindis, C.D., & Glei, D.A. (1999). Pregnancies averted among U.S. teenagers by Hawkins, J.D., Catalano, R.F., Kosterman, R., the use of contraceptives. Family Planning Abbott, R., & Hill, K.G. (1999). Preventing Perspectives, 31(1), 29-34. adolescent health-risk behaviors by strengthening protection during childhood. Archives Pediatrics Kirby, D. (1984). Sexuality education: An evalu- Adolescent Medicine, 153(3), 226-234. ation of programs and their effects. Santa Cruz, CA: Network Publications. Hercog-Baron, R., Furstenberg, F.F., Shea, J., Jr., & Harris, K.M. (1986). Supporting Kirby, D. (1985). The effects of selected sexual- teenagers' use of contraceptives: A comparison of ity education programs: Toward a more realistic clinic services. Family Planning Perspectives, view. Journal of Sex Education and Therapy, 18(2), 61-66. 11(1), 28-37.

Henshaw, S.K., & Torres, A. (1994). Family Kirby, D. (1991). An evaluation of the Lake planning agencies: Services, policies, and fund- Taylor High School Health Center. Scotts Valley, ing. Family Planning Perspectives, 26, 52-59, 82. CA: ETR Associates.

Homans, G.C. (1965). Group factors in worker Kirby, D. (1997). No easy answers: Research find- productivity. In H. Proshansky & L. Seidenbert ings on programs to reduce teen pregnancy. Washington, DC: National Campaign to Prevent (Eds.), Basic studies in social psychology. New Teen Pregnancy. York: Holt, Reinhart, & Winston.

Howard, M., & McCabe, J. (1990). Helping Kirby, D. (2000). Logic models: A useful tool for designing, strengthening, and evaluating pro- teenagers postpone sexual involvement. Family grams to reduce teen pregnancy. Santa Cruz, CA: Planning Perspectives, 22(1), 21-26. ETR Associates.

Hubbard, B.M., Giese M.L., & Rainey, J. Kirby, D., Barth, R., Leland, N., & Fetro, J. (1998). A replication of Reducing the Risk, a (1991). Reducing the Risk: A new curriculum to theory-based sexuality curriculum for adoles- prevent sexual risk-taking. Family Planning cents. Journal of School Health, 68(6), 243-247. Perspectives, 23(6), 253-263.

Hughes, M.E., Furstenberg, F.F., Jr., & Teider, Kirby, D., Brener, N.D., Brown, N.L., J.O. (1995). The impact of an increase in family Peterfreund, N., Hillard, P., & Harrist, R. planning services on the teenage population of (1999). The impact of condom distribution in Philadelphia. Family Planning Perspectives, 27(2), Seattle schools on sexual behavior and condom 60-65, 78. use. American Journal of Public Health, 89(2), 182-187. Jastrab, J., Masker, J., Blomquist, J., & Orr, L. (1997). Youth Corps: Promising strategies for Kirby, D., & Brown, N. (1996). School condom young people and their communities. Cambridge, availability programs in the United States. MA: Abt Associates, Inc. Family Planning Perspectives, 28(5), 196-202.

Jemmott, J.B., III, Jemmott, L.S., & Fong, G.T. Kirby, D., Korpi, M., Adivi, C., & Weissman, J. (1992). Reductions in HIV risk-associated sexual (1997). An impact evaluation of Project SNAPP: behaviors among black male adolescents: Effects An AIDS and pregnancy prevention middle of an AIDS prevention intervention. American school program. AIDS Education and Journal of Public Health, 82(3), 372-377. Prevention, 9(Supplement A), 44-61.

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 164 Kirby, D., Korpi, M., Barth, R.P., & group pregnancy prevention program in New Cagampang, H.H. (1997). The impact of the York City schools. Family Planning Perspectives, Postponing Sexual Involvement curriculum 32(5), 237-245. among youths in California. Family Planning Perspectives, 29(3), 100-108. Little, C.B., & Rankin, A. (2001). An evaluation of the Postponing Sexual Involvement curricu- Kirby, D., & Miller, B. (forthcoming). lum among upstate New York eighth graders. Intervention designed to promote parent-teen Sociological Forum, 15(4). communication about sexuality. New Directions in Child and Adolescent Development. Lundberg, S., & Plotnick, R.D. (1990). Effects of state welfare, abortion, and family planning Kirby, D., Resnik, M.D., Downes, B., Kocher, policies on premarital childbearing among ado- T., Gunderson, P., Pothoff, S., Zelterman, D., & lescents. Family Planning Perspectives, 22, 246- Blum, R.W. (1993). The effects of school-based 251. health clinics in St. Paul upon school-wide birth rates. Family Planning Perspectives, 25(12), 12- Magura, S., Kang, S., & Shapiro, J.L. (1994). 16. Outcomes of intensive AIDS education for male adolescent drug users in jail. Journal of Kirby, D., Waszak, C., & Ziegler, J. (1991). Six Adolescent Health, 15(6), 457-463. school-based clinics: Their reproductive health services and impact on sexual behavior. Family Making the Grade. (2000, Summer). National Planning Perspectives, 23(1), 6-16. census survey identifies 1,135 school-based health clinics nationwide. Access. Kisker, E.E. (1984). The effectiveness of family planning clinics in serving adolescents. Family Main, D.S., Iverson, D.C., McGloin, J., Planning Perspectives, 16(5), 212-218. Banspach, S.W., Collins, K., Rugg, D., & Kolbe, L.J. (1994). Preventing HIV infection among Kisker, E.E., Brown, R.S., & Hill, J. (1994). adolescents: Evaluation of a school-based educa- Health caring: Outcomes of the Robert Wood tion program. Preventive Medicine, 23(4), 409- Johnson Foundation's school-based adolescent 417. health care program. Princeton, NJ: Robert Wood Johnson Foundation. McBride, D., & Gienapp, A. (2000). Using ran- Klaus, H., Bryan, L.M., Bryant, M.L., et al. domized designs to evaluate a client-centered (1987). Fertility awareness/natural family plan- program to prevent adolescent pregnancy. ning for adolescents and their families: Report of Family Planning Perspectives, 32(5), 227-235. a multisite pilot project. International Journal of Adolescent Medicine and Health, 3(2), 101-119. McGuire, W. (1972). Social psychology. In P.C. Dodwell (Ed.), New horizons in psychology. Koo, H.P., Dunteman, G.H., George, C., Green, Middlesex, England: Penguin Books. Y., & Vincent, M. (1994). Reducing adolescent pregnancy through school and community-based Melchior, A. (1998). National evaluation of education: Denmark, South Carolina, revised Learn and Serve America school and community- 1991. Family Planning Perspectives, 26(5), 206- based programs. Waltham, MA: Center for 217. Human Resources, Brandeis University.

Levy, S.R., Perhats, C., Weeks, K., Handler, A., Miller, B.C., Norton, M.C., Jenson, G.O., Lee, Zhu, C., & Flay, B.R. (1995). Impact of a T.R., Christopherson, C., & King, P.K. (1993). school-based AIDS prevention program on risk Impact evaluation of FACTS and feelings: A and protective behavior for newly sexually active home-based video. Family Relations, 42(4), 392- students. Journal of School Health, 65(4), 145,- 400. 151. Miller, B.C., & Paikoff, R.L. (1992). Comparing Lieberman, L.D., Gray, H., Wier, M., adolescent pregnancy prevention programs: Fiorentino, R., & Maloney, P. (2000). Long- Methods and results. In B.C. Miller, J.J. Card, term outcomes of an abstinence-based, small- R.L. Paikoff, & J.L. Peterson (Eds.), Preventing

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy I 613.1 e)2 adolescent pregnancy (pp. 265-284). Newbury Reach for Health community youth service learn- Park, CA: Sage Publications. ing HIV prevention program. Unpublished manuscript. Mitchel-DiCenso, A., Thomas, B.H., Devlin, M.C., Goldsmith, C.H., Willan, A., Singer, J., Olsen, J.A., & Weed, S.E. (1986). Effects of Marks, S., Watters, D., & Hewson, S. (1997). family-planning programs for teenagers on ado- Evaluation of an educational program to prevent lescent birth and pregnancy rates. Family adolescent pregnancy. Health Education & Perspective, 20(3), 153-170. Behavior, 24(3), 300-3412. Orr, D.P., Langefeld, C.D., Katz, B.P., & Caine, Moberg, D.P., & Piper, D.L. (1990). An out- V.A. (1996). Behavioral intervention to increase come evaluation of project model health: A mid- condom use among high-risk female adolescents. dle school health promotion program. Health Journal of Pediatrics, 128(2), 288-295. Education Quarterly, 17(1), 37-51. Paine-Andrews, A., Harris, K.J., Fisher, J.L., Moberg, D.P., & Piper, D.L. (1998). The Lewis, R., Williams, E.L., Fawcett, S.B., & Healthy for Life Project: Sexual risk behavior Vincent, M.L. (1999). Effects of a replication of outcomes. AIDS Education and Prevention, a multi-component model for preventing adoles- 10(2), 128-148. cent pregnancy in three Kansas communities. Family Planning Perspectives, 31(4), 182-189. Moore, K.A., Blumenthal, C., Sugland, B., Hyatt, B., Snyder, N., & Morrison, D.R. Philliber, S., & Allen, J.P. (1992). Life options (1994). State variation in rates of adolescent and community service: Teen Outreach pregnancy and childbearing. Washington, DC: Program. In B.C. Miller, J.J. Card, R.L. Paikoff, Child Trends. & J.L. Peterson (Eds.), Preventing adolescent pregnancy (pp. 139-155). Newbury Park, CA: Moore, K.A., Sugland, B., Blumenthal, C., Glei, Sage Publications. D., & Snyder, N. (1995). Adolescent pregnancy prevention programs: Interventions and evalua- Philliber, S., Kaye, J.W., Herring, S., & West, E. tions. Washington, DC: Child Trends. (2000). Preventing teen pregnancy: An evaluation of the Children's Aid Society Carrera pro- Newcomer, S., Duggan, A., & Toczek, M. gram. Accord, NY: Philliber Research Associates. (1999). Do school-based clinics influence adolescent birth rates? In A. Mundigo (Ed.), Polen, M.R., & Freeborn, D.K. (1995). Reproductive health: Programme and policy Outcome evaluation of Project ACTION. changes post-Cairo. Belgium: IUSSP. Portland, OR: Kaiser Permanente Center for Health Research. Nicholson, H.J., & Postrado, L.T. (1991). Girls Incorporated preventing adolescent pregnancy: A Rotheram-Borus, M.J., Koopman, C., Haigners, program development and research project. New C., & Davies, M. (1991). Reducing HIV sexual York: Girls Incorporated. risk behaviors among runaway adolescents. Journal of the American Medical Association, O'Donnell, L., Stueve, A., Doval, A.S., Duran, 266(9), 1237-1241. R., Haber, D., Atnafou, R., Johnson, N., Grant, U., Murray, H., Juhn, G., Tang, J., & Piessens, Schinke, S.P., Blythe, B.J., Gilchrist, L.D., & P. (1999). The effectiveness of the Reach for Burt, G.A. (1981). Primary prevention of ado- Health community youth service learning pro- lescent pregnancy. Social Work with Groups, 4(1- gram in reducing early and unprotected sex 2), 121-135. among urban middle school students. American Journal of Public Health, 89(2), 176-181. Schochet, P.Z., Burghardt, J., & Glazerman, S. (2000, February 9). National Job Corps study: O'Donnell, L., Stueve, A., O'Donnell, C., The short-term impacts of Job Corps on partici- Duran, R., Doval, A.S., Wilson, R.F., Haber, D., pants' employment and related outcomes. Report Perry, E., & Pleck, J.H. (2000). Long-term and Evaluation Report Series 00-A. Washington, reduction in sexual initation and sexual activity DC: U.S. Department of Labor, Employment among urban middle school participants in the and Training Administration.

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 166 0 r; 4. j Schweinhart, L.J., Barnes, H.V., & Weikart, D.P. Walker, G., & Vilella-Velez, F. (1992). Anatomy (1993). Significant benefits: The High/Scope of a demonstration. Philadelphia, PA: Perry Preschool study through age 27. Ypsilanti, Public/Private Ventures. MI: The High/Scope Press. Walter, H.J., & Vaughn, RD. (1993). AIDS risk Sellers, D.E., McGraw, S.A., & McKinlay; J.B. reduction among a multi-ethnic sample of urban (1994). Does the promotion and distribution of high school students. Journal of the American condoms increase teen sexual activity? Evidence Medical Association, 270(6), 725-730. from an HW prevention program for Latino youth. American Journal of Public Health, Wang, L.Y., Davis, M., Robin, L., Collins, J., 84(12), 1952-1958. Coyle, K., & Baumler, E. (2000). Economic evaluation of Safer Choices. Archives of Pediatric Singh, S. (1986). Adolescent pregnancy in the Adolescent Medicine, 154(10), 1017-1024. United States: An interstate analysis. Family Planning Perspectives, 18(5), 210-220. Warren, W.K., & King, A.J.C. (1994). Development and evaluation of an Slonim-Nevo, V., Auslander, W.F., Ozawa, AIDS/STD/sexualiPy program for grade 9 stu- M.N., & Jung, K.G. (1996). The long-term dents. Kingston, Ontario: Social Program impact of AIDS-preventive interventions for Evaluation Group. delinquent and abused adolescents. Adolescence, 31(122), 409-421. Weed, S.E., & Olsen, J.A. (1986). Effects of family planning programs for teenage pregnancy

Sonenstein, F.L., Pleck, & Ku, L.C. replication and extension. Family Perspectives, . (1989). Sexual activity, condom use and AIDS 20(3), 173-195. awareness among adolescent males. Family Planning Perspectives, 21, 152-158. Weed, S.E., Olsen, TA., DeGaston, J., & Prigmore, J. (1992, December). Predicting and St. Lawrence, TS., Jefferson, K.W., Alleyne, E., changing teen sexual activity rates: A comparison Brasfield, T.L., O'Bannon, R.E., III, & Shirley, of three Title XX programs. Washington DC: A. (1995). Cognitive-behavioral intervention to Office of Adolescent Pregnancy Programs. reduce African American adolescents' risk for HIV infection. Journal of Consulting and Weeks, K., Levy, S.R., Gordon, A.K., Handler, Clinical Psychology, 63(2), 221-237. A., Perhats, C., & Flay, B.R. (1997). Does parental involvement make a difference? The St. Pierre, T.L., Mark, M.M., Kaltreider, D.L., & impact of parent interactive activities on students Aikin, K.J. (1995). A 27-month evaluation of a in a school-based AIDS prevention program. sexual activity prevention program in Boys & AIDS Education and Prevention, 9(Supplement Girls Clubs across the nation. Family Relations, A), 90-106. 44(1), 69-77. Winter, L., & Breckenmaker, L.C. (1991). Thomas, B., Mitchell, A., Devlin, M., Tailoring family planning services to the special Goldsmith, C., Singer, J., & Watters, D. (1992). needs of adolescents. Family Planning Small group sex education at school: The Perspectives, 23(1), 24-30. McMaster Teen Program. In B.C. Miller, J.J. Card, R.L. Paikoff, & J.L. Peterson (Eds.), Zabin, L.S., Hirsch, M.B., Smith, E.A., Streett, Preventing adolescent pregnancy (pp. 28-52). R., & Hardy, J.B. (1986). Evaluation of a Newbury Park, CA: Sage Publications. pregnancy prevention program for urban teenagers. Family Planning Perspectives, 18(3), Vincent, M., Clearie, A., & Schluchter, M. 119-126. (1987). Reducing adolescent pregnancy through school and community-based education. Journal of the American Medical Association, 257(24), 3382-3386.

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy. 1 67 224 ampter Looking Forward: Conclusions About the State of Research and the Effectiveness of Programs

Ihe previous chapter discussed in group doesn't mean it will be equally so in a detail a multitude of studies on the different environment with a different target effects of programs to prevent teen population. pregnancy. This chapter offers summary observations about the state of research in Second, there have been very few repli- this field and then about what the research cations of most evaluations of specific pro- tells us about the effectiveness of programs. grams. When programs have been replicated and evaluated, sonietimes the initial study's Conclusions About the Research positive results were confirmed; sometimes they were not. This calls into question our At least four factors limit the conclu- ability to make general statements about the sions that can be drawn from the many stud- effectiveness of types of some specific pro- ies reviewed in this report. First, although grams based on results from non-replicated more than 70 studies have examined the studies. impact of programs designed to reduce teen pregnancy or HIV/STDs, there are many Third, many of these studies were lim- different approaches to reducing sexual risk- ited by methodological problems or con- taking that have not been evaluated suffi- straints. Too often, studies have not used ciently if at all. Consequently, it is not experimental designs (and therefore suffered appropriatc to rcach conclusions about an from selection biascs); have had sample sizes entire approach to pregnancy prevention on that were too small (and therefore failed to the basis of one or even a small handful of detect programmatically important outcomes, studies. This is especially true when programs produced anomalous results, or had very are implemented in different settings because large confidence intervals, meaning that the their success may vary by the setting and findings were more likely to be partly the group targeted. That is, just because a pro- result of chance); have used exploratory ana- gram is effective in one setting or with one lytic techniques instead of confirmatory tech-

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 169 225 niques (i.e., have searched for positive development programs, and mass media results, rather than stating hypotheses first programs for youth. and then reporting results); have failed to control for the clustering of youth in schools O Conduct basic research on other topics or agencies (and thereby have exaggerated that may help to develop new approaches the statistical significance of results); or have to reducing teen pregnancy (such topics failed to report and publish negative results. include, for example, approaches to pre- Accordingly, the results that are published venting and dealing with sexual abuse, are undoubtedly biased in multiple unknown methods for discouraging romantic ways but are especially likely to be biased in relationships with much older partners, favor of positive results. methods for addressing other antecedents of early voluntary sexual activity, Fourth, some studies have produced approaches to increasing contraceptive inconsistent results. Some programs continuation, ways to involve males in appeared to affect behavior, while other programs, important characteristics of seemingly similar programs did not. When effective youth development programs, these inconsistencies occur, it is difficult to and approaches to improving the life know whether the results varied because of options that young people perceive they methodological differences, whether the have, especially those who are poor or programs were implemented differently, or otherwise disadvantaged). whether the programs were effective with only specific groups of youth in selected O Identify rigorous intervention studies in settings. other fields (e.g., substance abuse prevention, violence prevention, and youth For these four reasons, it is difficult to development programs) that may.have reach many conclusions that are strongly data on fertility outcomes, and encour- supported by a body of evidence. As a result, age new studies in these fields to add some programs in this country are based on measures of fertility to the outcomes somewhat naive assumptions about what will they are tracking in order to expand the change adolescent sexual behavior, and range of program types evaluated that resources are undoubtedly not directed may help to reduce teen pregnancy. toward the most effective approaches. In addition, in a field as contentious as teen Improve the quality and rigor of research pregnancy prevention, it is important to be by using experimental designs with ran- clear about the extent to which one's conclu- dom assignment, obtaining large sample sions about programs are supported by sizes, tracking youth for at least a year research evidence. and preferably longer, measuring actual sexual and contraceptive behavior, con- To produce more definitive conclusions ducting proper statistical analyses, and about what works, the research community reporting results regardless of whether needs to: they are positive or negative. Whenever possible, these evaluations should be Continue to identify and evaluate those completed by independent parties. programs that appear promising but that have not been well-evaluated, including OIdentify programs that yield positive abstinence-only programs, clinic proto- behavioral results in one study and re- cols and outreach efforts, pre-school evaluate them in different communities interventions, additional types of youth and with different program and research

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 170 0 4- 6 teams to determine their success under childbearing, and they fall into three dif- different conditions. ferent groups. The first group includes those programs that address the sexual O Conduct evaluations of effective pro- antecedents of sexual risk-taking (e.g., grams specifically to identify which com- school- or community-based sex and AIDS ponents or characteristics of these education programs and some health clinic programs are the critical to their success. programs). The second group primarily O Conduct evaluations of selected pro- addresses non-sexual antecedents (e.g., ser- grams (e.g., service learning) to under- vice learning programs). And the third stand more fully how they successfully group addresses both groups of antecedents affect the antecedents of teen pregnancy. (e.g., the Children's Aid Society-Carrera Program). O Conduct research to improve our understanding of how to replicate with fidelity 0Abstinence and use of contraception are those programs that are found to be compatible goals and topics. There are two effective in two or more studies. ideas behind this simple statement. First, the overwhelming weight of the evidence Conclusions About Programs demonstrates that programs that focus on sexuality and discuss contraception, Despite the caveats expressed above, the including sex and HIV education patterns of results among existing studies do programs, school-based clinics, and warrant several general conclusions about the condom-availability programs, do not effectiveness of particular types of programs: increase sexual activity. Many studies

O Both the studies of antecedents and the have reached this conclusion, and the evaluations of programs indicate that rare study finding an increase in any there are no single, simple approaches that measure of sexual activity is about what will dramatically reduce adolescent preg- would be expected by chance. Further- nancy across the country. While there are more, a number of programs that dis- a number of effective programs, there are cussed condoms or other forms of no "magic bullets." If pregnancy preven- contraception and encouraged their use tion programs and initiatives are to among sexually active youth also delayed reduce pregnancy markedly, they may or reduced the frequency of sexual inter- need to combine several effective compo- course. Second, programs that empha- nents that address both the more "proxi- sized abstinence, that gave it clear mal" sexual antecedents of adolescent prominence, and that presented it as the sexual behavior (such as values about safest and best approach, while also abstinence and attitudes about contra- emphasizing condoms or contraceptives ceptive use) as well as the more "distal" for sexually active youth, did not non-sexual antecedents (such as commu- decrease contraceptive use. Thus, giving nity poverty, lack of opportunity, family appropriate emphasis to both abstinence disorganization, peer behaviors, attach- and condoms or other contraceptives ment to school, substance abuse, and does not have the negative effects that risk-taking more generally). This is par- peopld sometimes fear, and it can, in fact, ticularly true for more disadvantaged have many positive effects. adolescents. 0Relatively little is known about the impact O A number of programs now appear to of programs that stress abstinence as the reduce sexual risk-taking, pregnancy, or only acceptable behavior for unmarried

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 171n , 44 teens. Until now, only a limited effort has females attend family planning clinics or been made to evaluate these programs, visit private physicians where they obtain and very few well-designed studies have contraceptives with higher effectiveness been completed. The rare study that was ratings than over-the-counter contracep- well designed did not find a delay in sex- tives. All else being equal, these contra- ual intercourse, and this is not encourag- ceptives would logically reduce the ing. However, because abstinence-only pregnancy rates among those youth who programs are such a diverse group, those use these prescribed methods. Non- few evaluations make it impossible to experimental studies have estimated that reach any overall conclusions about all tens or hundreds of thousands of preg- abstinence-only programs. It seems plau- nancies are prevented, the exact estimate sible that some abstinence-only programs depending on assumptions about how especially those that incorporate the teens would behave sexually if these ten characteristics of effective sex and prescription contraceptives were not HIV education programs or those that available. However, few studies have include an effective youth, development examined the impact of subsidized program with an abstinence message family planning services on pregnancy may delay sexual intercourse. On the or birth rates, and none met the criteria other hand, currently there does not for this review. exist any research with reasonably strong 0Several studies have consistently indicated evidence demonstrating that any particu- that when clinics provide improved educa- lar abstinence-only program actually tional materials, discuss the adolescent delayed the onset of sexual intercourse or patient's sexual and condom or contracep- reduced any other measure of sexual tive behavior, give a clear message about activity. When studies currently under that behavior, and incorporate other way are completed, this may change. components into the clinic visit, clinics can Studies of some sex and HIV education increase condom or contraceptive use, programs have produced credible evidence although not always for a prolonged period that they reduce sexual risk-taking either of time. by delaying the onset of sex, reducing the 0School-based and school-linked clinics and frequency of sex, reducing the number of school condom-availability programs do sexual partners, or increasing the use of not increase sexual activity, but it is not condoms or other forms of contraception. clear whether they increase condom and Some studies found these positive effects contraceptive use. Substantial percentages to endure for as long as 31 months. of sexually experienced female students Studies of other sex and HIV education in schools with school-based clinics programs have failed to find positive obtain contraceptives from those clinics. effects on behavior. Effective sex and Similarly, students obtain large numbers HIV education programs share ten char- of condoms from schools, when schools acteristics that are described in Chapter 4 provide those condoms in private loca- on pages 91-94. tions and with few restrictions. Studies 0Family planning clinics probably prevent a have consistently shown that making large number of teen pregnancies condoms or other contraceptives avail- although there is remarkably little evidence able in schools does not hasten or to support this common-sense view. Each increase sexual behavior. However, stud- year many sexually active adolescent ies measuring the impact of school-based

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 172 9 0 (-) clinics and school condom-availability on programs have been evaluated many contraceptive use have produced mixed times and in many communities, and dif- results; a few found increases in condom ferent studies by different investigators or contraceptive use, while most did not. have found positive effects on sexual risk- Notably, school-based or school-linked taking or pregnancy. This is very encour- clinics that focused on contraceptive use aging. However, it is unclear, whether and gave a clear message about absti- these programs reduce teen pregnancy nence and contraceptive use may have beyond the academic year in which the delayed sex and increased contraceptive students are involved in the programs. use, while other clinics did not. Notably, these programs did not focus on the.sexual antecedents of unprotected O The small number of studies measuring the sex, and some were not even designed to impact of community-wide initiatives and reduce teen pregnancy. However, they collaboratives to reduce teen pregnancy were intensive in that youth participated found mixed results. One study reported in them for many hours and often for evidence of a reduction in teen preg- prolonged periods of time. In addition to nancy rates, while others did not. It their success in reducing teen pregnancy, should be fully recognized, however, that some of these programs had other these studies were population-based, positive results, such as reducing school which presented their respective pro- failure. grams with a very challenging goal: significantly changing the behavior of all Some other youth development programs the youth in their targeted communities, (e.g., vocational education programs) that not just the behavior of a much smaller addresied non-sexual antecedents did not number of youth who participated reduce teen pregnancy or childbearing directly in their programs. even though they addressed seemingly important antecedents (such as basic read- O Effective programs that addressed sexual ing, writing, and math skills and prepa- antecedents shared two common attributes: ration for employment), were relatively they focused clearly on sexual behavior and intensive, and were long-lasting. condom or contraceptive use, and they gave clear messages about abstaining from sex 0Other youth development programs, such as or using protection against STDs and strong preschool child care programs, pro- pregnancy. These qualities helped define grams to improve the quality of teaching the effective school- and community- in elementary school and student attach- based sex and HW education proirams, ment to school, and very comprehensive the effective school-based clinic pro- and intensive youth development programs, the effective health clinic programs, have produced a few encouraging grams, and even the effective community results, but there are too few studies and collaboratives. Of those programs too many important study limitations to designed to address sexual antecedents, reach any conclusions. those programs that did not do so 0The only evaluated program that inten- directly and that did not give clear mes- sively addressed both the sexual and non- sages about abstinence and the use of sexual antecedents of disadvantaged youth protection appeared to be less effective. (the CAS-Carrera Program) substantially oService learning programs have the reduced teen pregnancy reported by girls strongest evidence that they actually reduce over a long period of time. It also had teen pregnancy rates. Service learning other positive effects.

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 173 2 9 These conclusions indicate that pro- STDs, although the evidence for that is dra- grams may be effective both when they focus matically weaker than it is for reducing teen clearly on sexual behavior and when they pregnancy. This is an important area for fur- don't address sexuality at all. Moreover, it is ther investigation. apparent that if programs target sexual antecedents, they must do so directly and must give a clear message. For example, they may discuss realistic situations that might 0 .0 0 lead to unprotected sex and methods for avoiding those situations, for remaining abstinent, and for using protection. If pro- In conclusion, considerable progress has grams target non-sexual antecedents, then been made to reduce teen pregnancy and they need to intervene sufficiently intensively STDs in this country. Our understanding of in the lives of youth so that youth become the antecedents of adolescent sexual risk-tak- more motivated to avoid pregnancy and ing and pregnancy continues to grow and to childbearing or simply have less opportunity broaden. The number of evaluations of spe- to engage in unprotected sex. And, of cific programs continues to accumulate, and course, the CAS-Carrera Program demon- the quality of these studies continues to strates that addressing both sexual and non- improve markedly. More important, the sexual antecedents can be very effective. number of programs with evidence for All in all, these findings bring good behavioral change continues to increase and news because they mean that quite different the strength of that evidence is improving. approaches to reducing sexual risk-taking are In just the last four years, a number of rigor- emerging: some programs that address ous studies with random assignment that sexuality directly and a few programs that show long-term behavioral impact have been address youth development more broadly published, as have the first replications of can reduce sexual risk-taking or pregnancy. studies with positive results. Most important, This increases the choices that communities pregnancy and childbearing rates have have for addressing teen pregnancy with declined every year since 1991. Certainly, effective programs. there remains much to learn, but there are now some "emerging answers" to the ques- While this review has focused principally tion, "What works?" Adults working with on whether programs reduce teen pregnancy youth should feel proud of their contribution or the behaviors that lead to teen pregnancy, to this.success. it is also true that much can be learned from this research about whether programs effec- On the other hand, we should not be tively reduce STDs, including HW. After all, lulled into complacency. The evidence for every program that delayed sex or increased the success of these programs, while improv- condom use should logically reduce STD ing markedly, is still not as strong as it transmission. Most of these programs should be. Teen pregnancy (and STD) rates focused on sexual antecedents and were cur- in the United States are still much too high, riculum-based sex or HW education pro- both in terms of international standards and grams or clinic-based programs. However, in terms of the well-being of the youth and service learning programs, which do not infants involved. Currently, relatively few focus on sexual antecedents, may delay the youth in this country participate in the kinds onset of sex and clearly can decrease teen of programs that have evidence for success, pregnancy. Thus, they might also reduce and past experience suggests that it is very

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 174 un difficult to replicate successful programs with effective programs and services rapidly in fidelity. This desperately needs to be done. order to meet their needs. The final chapter And, finally, the number of teenagers in this in this report offers three strategies for using country is expected to grow rapidly during these research findings to select and design programs to prevent teen pregnancy at the the coming years, placing a much greater community level. demand on all of our programs that serve youth. Clearly, we need to expand our

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 175 9 31 ChEpter Bringing It Home: Applying These Research Results in Communities 7 ow can one use this report's find- oThe availability of condoms and rings on the antecedents of adoles- reproductive health services in the Lcent sexual risk-taking and on the community; effects of various programs to select or design programs to reduce sexual risk-taking O The extent to which most health care and pregnancy? This is not an easy question providers address adolescent sexual to answer, and, indeed, several excellent and behavior and how they do so; detailed guides are available to help people O The availability of employment opportu- design and implement programs (see nities in the community, the quality of "Excellent Guides" sidebar on the next schools, the stability and closeness of page). Many issues need to be considered families, the monitoring and supervision when choosing what type of programs to put of youth, and the existence of commu- in place. Several of the most important are: nity programs for youth;

O The actual sexual and contraceptive O The need for and interest in youth behavior of teens in the. community; development programs for adolescents; oThe values of the community about ado- and lescent abstinence, sexual behavior, and O The resources both staff resources and contraceptive use; monetary resources available to imple- O The community's concerns about teen ment new programs. pregnancy only versus concerns about With such considerations in mind, there teen pregnancy, STDs, and HIV; are three strategies that communities should O The quality of existing abstinence and sex follow to increase the chances that the pro- and HIV/STD education programs in grams they select or design on their own will the schools and in other organizations actually reduce sexual risk-taking or that work with youth in the community; pregnancy:

Emerging Answers: Research Rndings on Programs to Reduce Teen Pregnancy 1.73 2 1.Implement with fidelity programs programs with fidelity that have been demonstrated to be effective with demonstrated to be effective in other places' similar populations. with similar youth. If a program has strong evidence that it reduced sexual risk-taking or 2.Select or design programs that teen pregnancy, if your target population is incorporate the key characteristics of similar, and if you implement the program in programs that have been effective the same way, then the chances are rather with similar populations. good that you will achieve similar results. If 3.Use logic models to select or design the program was found to be effective in sev- new programs. eral sites when implemented by different, independent groups, then your chances are I. Implement with fidelity programs even greater. demonstrated to be effective with Of course, the question still remains: similar populations. what are the most effective programs to The single most promising strategy for reduce teen pregnancy? Unfortunately, that reducing teen pregnancy is to implement question cannot be answered definitively because there are thousands of programs and most have never been evaluated. Further- more, the programs that have been evaluated Excellent Guides to Designing and have not been directly compared with one Implementing Teen Pregnancy another in an experimental design. Typically, Prevention Programs they have served different groups at different times, and it is simply not possible to deter- Kreinin,T, Kuhn, S., Rodgers, A.B., & Hutchins, J. mine which are the most effective. (Eds.). (1999). Get Organized:A Guide to Preventing Teen Pregnancy. Washington, DC: On the other hand, it is possible to National Campaign to PreventTeen Pregnancy answer a related question: Which programs (wwwteenpregnancy.org) have the strongest evidence indicating that Brindis, C., & Davis, L (1998). Communities they delay sex, increase condom or contra- Responding to the Challenge of Adolescent ceptive use, or reduce actual pregnancy or Pregnancy Prevention. Washington, DC: Advocates childbearing? This review has identified eight for Youth. (www.advocatesforyouth.org) programs with strong evidence of effectiveness within the categories of sex and HIV Card, J.J., Brindis, C., Peterson, J.L., & Niego, S. education, service learning, and multiple- (2001). Guidebook: Evaluating Teen Pregnancy component interventions (see "Programs Prevention Programs (2nd edition). Los Altos, CA: Sociometrics. (wvAmsocio.com) with Strong Evidence" sidebar on the next page). In order to be included in this list, a Moore, K.A., & Sugland, B.W. (2001). Next Steps program must meet either of two sets of and Best Bets:Approaches to Preventing Adolescent conditions. The first set of conditions Childbearing (2nd edition).Washington, DC: includes an evaluation with an experimental Child Trends. (www.childtrends.org) design with random assignment, a large sam-

Philliber Research Associates. (Ed.). (2001). ple size, strong statistical analyses, and statis- Creating and Evaluating Successful Teen Pregnancy tically significant and programmatically Programs. Accord, NY: Author. important behavioral effects for at least one (www.philliberresearch.com) year. The second set of conditions includes two or more studies conducted by independent research teams, each with at least a

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 178 quasi-experimental design with both inter- Programs with Strong vention and comparison groups and baseline Evidence if Success and follow-up data, acceptable sample size, acceptable statistical analyses, and statistically I.Programs that Focus Primarily on significant and programmatically important Sexual Antecedents behavioral effects for at least one year. Sex education programs covering both pregnancy and STDs/HIV"

Programs that focus primarily on sex- oReducing the Risk ual antecedents. Two studies found that o Safer Choices Reducing the Risk delayed the initiation of .sexual intercourse and increased condom or HIV education programs' oBecoming a Responsible Teen: An HIV Risk contraceptive use among some groups of Reduction Intervention for African- youth (see Chapter 4, p. 91). These results American Adolescents lasted for 18 months. The other four sex and o Making a Difference: An Abstinence HIV education programs on the list all had Approach to STD, Teen Pregnancy, and HIV/AIDS Prevention very strong experimental designs and positive results. Safer Choices increased condom and o Making a Difference: A Safer Sex Approach to STD, Teen Pregnancy, and HIV/AIDS contraceptive use and decreased the fre- Prevention quency of unprotected sex over a 31-month II. Programs that Focus Primarily on period. Making a Difference: An Abstinence Non-Sexual Antecedents Approach to STD, Teen Pregnancy, and Service learning2 HIV/AIDS Prevention delayed the initiation of sex at three months and increased the oTeen Outreach Program (TOP) use of condoms. Making a Difference: A o Reach for Health Community Youth Service Learning Safer Sex Approach to STD, Teen Pregnancy, and HITT/AIDS Prevention reduced the Ill. Programs that Focus Upon Both frequency of sex, increased the use of con- Sexual and Non-Sexual Antecedents doms, and decreased unprotected sex during Multi-component programs with a 12-month period, and Becoming A intensive sexuality and youth development component Responsible Teen delayed the initiation of sex, reduced the incidence of sex, reduced the E3 Children's Aid Society-Carrera Program3 number of sexual partners, increased the use While the sex and HIV education programs identi- of condoms, and reduced the frequency of fied in this table demonstrated a positive impact upon sexual behavior and condom and contraceptive use, unprotected sex during a 12-month period. some other sex and HIV education programs did not For all five programs, some of their behav- have positive effects. Studies indicated that the sex and HIV education programs in this table reduced ioral effects may have lasted longer than the sexual risk-taking, but they did not provide evidence time periods specified above, but the studies they reduced teen pregnancy. did not measure longer term effects. 2 All the service learning programs that have been evaluated, including the Learn and Serve programs, Although these results are very encouraging, have found results suggesting a positive impact upon none of these programs provided evidence either sexual behavior or pregnancy. The Learn and Serve study is not included on this list because it did for reduction in actual teen pregnancy. not meet the criteria for being on this list, but it did Although reduced sexual risk-taking should confirm the efficacy of service learning. According to the analysis of TOP, the particular curriculum used lead to reductions in pregnancy, the impact in the small group component did not appear to be of these programs on pregnancy was either critical to the success of service learning. not measured or the findings were not statis- 3 This program has provided the strongest evidence for a three-year impact upon pregnancy. tically significant.

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 179 3 4 It should also be noted that all five of Teen Outreach Program, and the Children's these curricula and their respective studies Aid Society-Carrera Program have all been have been independently reviewed by the implemented in five or more communities and Centers for Disease Control and Prevention been found to be effective. This suggests that (CDC) and selected for their "Programs that their success is not limited to a single type of Work" initiative. Two additional programs to community or to a particular program team. reduce sexual risk-taking were selected by CDC for the "Programs that Work" initia- A very important question remains, tive, but those two studies did not meet the however: what is the magnitude of the criteria specified above. impact of these programs? How much of an improvement in behavior do they make? Programs that focus primarily on non- This is another question that is surprisingly sexual antecedents. In contrast to the sex and difficult to answer. On the one hand, each of HIV education curricula, several studies sug- these programs had meaningful behavioral . gest that service learning programs actually results. For example, Reducing the Risk reduce teen pregnancy, at least during the decreased the percent of sexually inexperi- academic year in which students participate enced youth who subsequently engaged in (see Chapter 4, p. 108). The single service unprotected sex by about half, from 16 per- learning program with the strongest evidence cent to 9 percent (Kirby et al., 1991). Safer for a reduction in teen pregnancy is the Teen Choices reduced the mean number of acts of Outreach Program. However, the Reach for sexual intercourse not protected by a con- Health service learning program delayed the dom from 3.82 times in a three-month onset of sexual intercourse and reduced sex- period to 2.44 times, a 37 percent reduction ual activity for as long as three years. Given (Coyle et al., forthcoming). Making a that no studies have measured the impact of Difference: An Abstinence Approach to SID, service learning programs and found that Teen Pregnancy, and HIV/AIDS Prevention they failed to reduce teen pregnancy, it may reduced the percent of youth who had sex be the case that many intensive service learn- during the previous 3 months from 22 per- ing programs are effective. As has been cent to 13 percent, a 42 percent reduction. noted before, all the service learning pro- Making a Difference: A Safer Sex Approach to grams that have been evaluated were very SID, Teen Pregnancy, and HIV/AIDS intensive and lasted many months with many Prevention reduced the percentage of youth hours of structured time for participants. who reported unprotected sex during the A program that focuses on both sexual previous month from 10.8 percent to 5.4 and non-sexual antecedents. The Children's percent, a 50 percent reduction, and the Aid Society-Carrera Program is the only pro- mean number of acts of unprotected inter- gram with strong evidence that (among girls) course during the previous month from .51 it delayed sex, increased simultaneous use of to .17, a 67 percent reduction (Jemmott et condoms and other more effective contra- al., 1998). Becoming A Responsible Teen ceptives, and decreased both teen pregnancy reduced the percent of sexually inexperienced and teen birth rates. Furthermore, the study youth who initiated sex during a 12-month indicated that the program had very long- period frorri 31 percent to 12 percent, a 63 term effects. effects that lasted three years percent reduction (St. Lawrence et al., after youth joined the program. 1995). The Reach for Health service learning program reduced the percent of youth who Of the eight programs with strong evi- initiated sex from 27 percent to 19 percent, dence, Reducing the Risk, Safer Choices, the a 29 percent reduction (O'Donnell et al.,

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 180 2000). Finally, the Children's Aid Society- evidence is simply not as strong. And, as Carrera Program reported a reduction in noted above, there are undoubtedly many pregnancy rate among girls after three years effective programs that simply have not been in the program from 22 percent to 10 per- evaluated. In the future, other programs will cent, a 55 percent reduction (Philliber et al., meet the specified criteria and can be added 2000). All of these reductions are very to this list. encouraging. When selecting programs to put in However, for several reasons, these place, should community leaders confine reductions are somewhat misleading, and themselves to these eight programs with the people replicating these programs should not strongest evidence of effectiveness? On the expect to obtain such positive results. First, one hand, community leaders should obvi- while all of these results measure important ously give these programs serious considera- aspects of sexual risk-taldng or pregnancy, tion because of both the magnitude of their they are also among the largest effects in each positive effects and the strength of the evi- study. Typically, these studies did not have dence for their success. On the other hand, such large effects on other behavioral out- community leaders will probably want to comes. For example, Reducing the Risk did consider other programs. After all, these pro- not have as large or significant an impact on grams are not appropriate for youth of all frequency of sex as it did upon the frequency ages, they may not match the needs of par- of unprotected sex, and Making a Difference: ticular groups of youth, and they may not A Safer Sex Approach to SID, Teen Pregnancy, match the values or resources of some com- and HIV/AIDS Prevention did not have as munities. And, of course, none of these pro- large an impact on delaying the initiation of grams eliminated sexual risk-taking or sex as it did on reducing unprotected sex. pregnancy. It is probably the case that to dra- Second, while all of these results were statisti- matically reduce teen pregnancy, broad-scale, cally significant, all of these estimates had community-wide initiatives will require a rather large confidence intervals meaning number of programmatic components to that chance probably affected the size of address the wide variety of sexual and non- some of these results. In theory, chance could sexual antecedents of adolescent sexual either enhance or diminish the apparent behavior described in Chapter 2. effects of these programs. However, given that these were among the largest effects, it is 2. Select or design programs that likely that they enhanced some of them, and incorporate the key characteristics replications would be more likely to report of programs that have been smaller effects. effective with similar populations.

Although these eight programs have the When it is not possible to implement strongest evidence for success and the esti- programs that have been demonstrated to be mated magnitude of their success is impres- effective, then a second promising strategy is sive, it is important to emphasize that these to select or design programs that incorporate are not the only programs in this country the common characteristics of effective pro- with evidence that they have a positive grams, especially programs that have impact on adolescent sexual behavior. Both involved similar target populations. In doing the text and tables in Chapter 4 identify so, the chances are increased that positive numerous other programs that provide some behavioral results will be obtained. Of evidence that they changed behavior in course, whether or not the program is actu- desired directions, but the strength of that ally effective may depend upon (1) properly

Emerging Answers: Research Findings on Programs.to Reduce Teen Pregnancy 181 ) 3 6 identifying the critical characteristics of effec- 3. Use logic models to select or tive programs, and then (2) designing or design new programs. implementing a program that actually incor- porates those characteristics. If it is not appropriate or possible to put in place an existing program that Chapter 4 identified ten common char- research has shown to be effective or one acteristics of effective sex and HIV education with the common characteristics of effective programs (see p. 91-94). These characteris- programs, then a final strategy can be tics appear to distinguish effective programs adopted. Communities can employ a process from ineffective ones, and they are similar to used by the many developers of effective pro- identified common characteristics of effective grams they can create a logic model to drug prevention programs. Thus, those ten design the program. Many of the people characteristics may be a helpful guide for who developed effective programs explicitly selecting programs that have not yet been used a logic model; others used the principles evaluated or for designing new ones. inherent in logic models.

The research on programs other than This third strategy for reducing teen sex and HIV education programs (i.e., clinic- pregnancy is clearly much more challenging based programs and others noted in Chapter and more problematic than either of the first 4) does not allow for the identification of a two because it involves actually developing a similar list of key characteristics. However, completely new intervention, rather than given that important caveat, these other implementing an existing one. However, this groups of effective programs do appear to strategy of using a logic model is far more share a few common characteristics. First, flexible than trying to build on the specific nearly all of the effective programs that characteristics of other programs and can addressed sexual antecedents focused clearly. be combined with either of the first two on changing particular sexual behaviors and strategies. gave a clear message about those behaviors Logic models are concise, causal (i.e., to remain or become abstinent or to descriptions of the mechanisms through always use protection against pregnancy and which specific program activities can affect STDs). A particularly good example of this behavior. Thus, at a minimum, they must was the group of clinic-based programs with include a specification of (1) the behaviors to modest clinician-patient interventions giving be changed, (2) the antecedents of each of a clear message to clinic patients (see those behaviors, and (3) the particular pro- Chapter 4, p. 97). Second, effective pick- gram activities designed to change each grams that incorporated youth development selected antecedent. They also include a components (e.g., service learning and the specification of the expected causal links CAS-Carrera Program) were all very inten- among these three basic model components. sive and involved youth for many months (if Logic models are sometimes called causal not years). However, it is not known which models or program models. characteristics or components of these programs (i.e., type of staff, duration of special A full discussion of logic models would classes, etc.) are the most critical. be too detailed to include in this report.'

1 A recent guide, BDI Logic Models: A Useful Tool for Designing, Strengthening, and Evaluating Programsto Reduce Adolescent Pregnancy (Kirby, 2000), describes how to develop logic models for pregnancy prevention initiatives. For a much more extensive discussion of this topic and processes for designing programs,see Intervention Mapping: Designing Theory and Evidence-Based Health Promotion Programs (Bartholomew et al., 2001).

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 182 4n However, it is worthwhile here to describe selected behavior. However, when this is not the three important, basic steps in develop- possible, focus groups with targeted popula- ing one: tions can be a less costly substitute method for verifying potentially important antece- Step #1: Identify specific sexual behav- dents. When identifying these antecedents, it ior goals for the different groups of teens is particularly important to identify those being targeted. Three logical behavioral that have a causal impact on sexual behavior goals for reducing teen pregnancy are to (1) and not merely a non-causal association. delay the onset of sexual intercourse, (2) Determining causality can be difficult. reduce the frequency of sexual intercourse (including returning to abstinence), and (3) Once the most important antecedents increase the use of effective contraception for the target population have been identi- among those having sex. Ideally, these fied, the program designer then needs to behavioral goals should be determined by select those antecedents that can be changed current measured levels of sexual risk-taking substantially with the programmatic and should also reflect the values and other resources available. Thus, the program must goals of the community. This step is also an be able to change the selected antecedents, important activity to complete even when and, in turn, the changes in the antecedents using either of the first two strategies for must have an impact upon behavior. For selecting programs. example, if one's goal is to delay the initiation of sex among teens in a particular Step #2: For each targeted group of school, there are many potentially important youth, identify (to the extent feasible) the antecedents that are highly relevant (see antecedents that are most highly related to Chapter 2, Table 2.1). However, in a partic- the sexual behaviors you want to change, ular community, program designers might and then select those antecedents that can be find that there are only two particular changed. The chart on page 27 of Chapter 2 antecedents that they believe they can summarizes some of the most important change and that are highly related to the ini- antecedents for teen sexual risk-taking and tiation of sex: (1) the belief that peers have pregnancy, while Tables 2.1-2.7 identify permissive attitudes toward adolescent sex many additional antecedents for each partic- and (2) the belief that peers are having sex. ular behavior. Although these lists should be These are just examples; when designing informative and helpful, the antecedents in effective programs, program designers typi- those tables are based on studies of teens in cally need to select more than two many different communities. To the extent antecedents. possible, communities should identify the most important antecedents for the behav- Step #3: Identify the particular activi- ioral goals of the targeted groups in their ties or programmatic components that will own communities. If a community is some- affect each selected antecedent. That is, hav- what heterogeneous, then the most impor- ing chosen the antecedents in Step #2, then tant antecedents of each behavioral goal may spccify the particular activities or program vary from one targeted group to another. components that will change those antecedents. Careful consideration should be Ideally, program designers would use given to whether or not the program activi- surveys or existing records to measure possi- ties being considered are sufficiently strong ble antecedents and sexual behaviors and to markedly change the antecedents. Ideally, then observe the strength of the relationships each targeted antecedent will have several between each possible antecedent and each activities or program components designed

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 183 238 to affect it. However, some activities or components can also affect more than one antecedent. Although not reviewed in this report, there is a large amount of experience, theory, and research about what types of Thus, in the coming years, communities program activities affect different antecedents should not naively focus on simplistic solu- (for example, there is research describing tions that have little chance for markedly what types of activities change perceptions of reducing teen pregnancy. Instead, they peer norms and increase self-efficacy). should (1) replicate much more broadly and By way of illustration, take the with fidelity those programs with the greatest antecedents related to peers selected in the evidence for success, (2) replicate more example under Step #2. The program devel- broadly programs incorporating the common opers in this hypothetical community could qualities of effective programs, and (3) conduct a survey of the students in the design and implement programs that effec- school to measure their beliefs about adoles- tively address the important antecedents of cent sex and their actual sexual behavior. sexual risk-taking. In addition, we should all Such surveys often reveal that most teens continue to explore, develop, and rigorously believe that it is a good idea to wait until evaluate promising approaches. they are older to have sex and that fewer teens have had sex than most students would We live in a hopeful time. Declining think. The program intervention that logi- teen pregnancy and birth rates, combined cally comes next is for these data to be pre- with a stronger body of research on the sented to the students and discussed in small groups in an attempt to counter the beliefs antecedents of teen sexual risk-taking and on that (1) peers favor early sex and (2) every- the impact of programs, should increase our one is having sex. (Of course, if most teens confidence to continue building on current actually do favor sex and most are having sex, success. The challenge will be to integrate then this program intervention would not be what we learn from experience with what we appropriate.) learn from research and to have that pooled knowledge guide our development of more When the logic model is completed, it provides a detailed guide specifying what key effective programs for youth. Such programs program activities will be implemented, will help young people avoid pregnancy and which antecedents they will affect, and which STDs, make a successful transition to adult- sexual behaviors will be modified by the hood, become educated and self-sufficient, changes in these antecedents. Consequently, and then become ready to be parents for the it can guide the development and implemen- next generation. tation of the program, as well as any possible evaluation of it.

THE NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 184 239 References Kirby, D., Barth, R., Leland, N., & Fetro, J. (1991). Reducing the Risk: A new curriculum to Bartholomew, L.K., Parcel, G.S., Kok G., & prevent sexual risk-taking. Family Planning Gottlieb, N.H. (2001). Intervention mapping: Perspectives, 23(6), 253-263. Designing theory and evidence-based health promotion programs. Mountainview, CA: Mayfield O'Donnell, L., Stueve, A., O'Donnell, C., Publishing CompanyA McGraw-Hill Duran, R., Doval, A.S., Wilson, R.F., Haber, D., Company. Perry, E., & Pleck, J.H. (2000). Long-term Coyle, KK., Basen-Enquist, K.M., Kirby, D.B., reduction in sexual initation and sexual activity Parcel, G.S., Banspach, S.W., Collins, J.L., among urban middle school participants in the Baum ler, E.R., Caravajal, S., & Harrist, R.B. Reach for Health community youth service learn- (forthcoming). Safer Choices: Long-term impact ing HIV prevention program. Unpublished of a multi-component school-based HIV, STD, manuscript. and pregnancy prevention program. Public Health Reports. Philliber, -S., Kaye, J.W., Herring, S., & West, E. Jemmott, J.B., III, Jemmott, L.S., & Fong, G.T. (2000). Preventing teen pregnancy: An evalua- (1998). Abstinence and safer sex: A randomized tion of the Children's Aid Society Carrera pro- trial of HW sexual risk-reduction interventions gram. Accord, NY: Philliber Research Associates. for young African-American adolescents. Journal of the American Medical Association, 279(19), St. Lawrence, J.S., Jefferson, K.W., Alleyne, E., 1529-1536. Brasfield, T.L., O'Bannon, RE., III, & Shirley, A. (1995). Cognitive-behavioral intervention to Kirby, D. (2001). BDI logic models: A useful tool for designing, strengthening, and evaluating pro- reduce African American adolescents' risk for grams to reduce adolescent pregnancy. Santa Cruz, HIV infection. Journal of Consulting and CA: ETR Associates. Clinical Psychology, 63(2), 221-237.

Emerging Answers: Research Findings on Programs to Reduce Teen Pregnancy 185240 About the Author

ouglas a more comprehensive and detailed picture Kirby, of the risk and protective factors associated Ph.D., is with adolescent sexual behavior, contracep- a Senior Research tive use, and pregnancy, and has identified Scientist at ETR important common characteristics of effec- Associates in tive sexuality education and HIV education Scotts Valley, programs. Over the years, he has also California. For authored or co-authored more than 100 vol- more than 22 umes, articles, and chapters on adolescent years, he has sexual behavior and programs designed to directed state- change that behavior. These have included wide and nation- A wide studies of reviews of the field for the Centers for adolescent sexual behavior, abstinence-only Disease Control and Prevention, the programs, sexuality and HIV education pro- National Institutes of Health, the former grams, school-bascd clinics, school condom- Office of Technology Assessment, and vari- availability programs, and youth develop- ous foundations. Dr. Kirby serves on the ment programs. He co-authored research on Board of the National Campaign to Prevent the Reducing the Risk, Safer Choices, and Teen Pregnancy and chairs the National Draw the Line curricula, all of which signifi- Campaign's Task Force on Effective candy reduced unprotected sex, either by Programs and Research. He is the author of delaying sex, increasing condom use, or No Easy Answers: Research Findings on increasing contraceptive use. He has painted Programs to Reduce Teen Pregnancy (1997). THE

NATIONAL

CAMPAIGN T PREVENT JEEN PREANCY

The National Campaign to Prevent Teen Pregnancy is a nonprofit, nonpartisan initiative supported almost entirely by private donations. The Campaign's mission is to improve the well-being of children, youth, and families by reducing teen pregnancy. Our goal is to reduce the rate of teen pregnancy by one-third between 1996 and 2005.

NATIONAL CAMPAIGN TO PREVENT TEEN PREGNANCY 1776 MASSACHUSETTS AVENUE, NW SUITE 200 WASHINGTON, DC 20036 (202) 478-8500 (202) 478-8588 FAX [email protected] WWW.TEENPREGNANCY.ORG

2 US: Defiant:sent of EsicatiOn office ar Educational Reseaucli and Improvement MEW. .I . IklalfonaI binary of Ede:Winn E IC..1 EducatIonal Rascal= Infonnadon Canter (ERIC)._ abc;34 REPRODUCTION RELEASE (Spec& Document) DOCUMENT IDENTIFICAMON:

TdbxEfrini r\s AJex-5.

Carp:rate Sam= Pudic:alba Date: AICPTP nay, ant IL REPRODUCrION RELEASE: In atter ta diesenthints as widely an possible timely and significant matedale afinteneat to the educational community, documents annourmad In I monthly abnlinct *neat at the MC systemRennumas in EducOm(FUE) aro comily ntede amiable tra unarm In nacreifiche, reproducedpaper cop and electronic made, and odd through the ERIC Document Repmcluctfon Service GEOM. Coedit la given to the acnuca Of end' document. and, neneductIon release le sided, we a/ the It:diming notices. la alibied to ifte document.

. Itpaindealou la started to owed= and dfasenticatelut Identilliatt drunecent, please CHECK ONE ante folloarinattnee options and sign atthe lustto

. . al*, PIP& . . . . Thaasamb addrar alaviaa Imimmill he ilma lempb ladarabara Wawa ba Ms damps Mbar ahem !aim data sand to alLaral2Adatuanta . aglaad ft MLA/al 1 dicusais Maxi to al Lauda dammanta . . PERMS= TO REPROCUCEAND PERMISSION TO REPRODUCE AND DISSEMINATE THM MATERIAL IN PERMS= TO REPROCUCEAND DISSENDIATE THIS MATERIAL HAS MICROFICHE ANC IN EUECTRONIC MEDIA DISSEMINATE THIS MATERIAL IN BEEN GRANTED BY FOR ERIC COLLECTION SUBSCRIBERS ONLY, MICRORCHEONLY HAS MEN GRANTED 81 HAS BEEN GRANTED BY

ACti sso Sgr TO INE EDUCATIONAL RESOURCE TO THE EDUCATIONAL RESOURCES TO TtE EDUCATIONAL RESOURCES INFORMATION COSTER (ERIC) INFORMAITON CENTER (ERIC) INFORMATION CENTER (ERIC) 2A 2B

Laval 2A Laval 2B

Cbastiterallettand 1 adorn. paemillas Chick Ma tar Laval Mar% panalllai Cloak lam tar Laval= ralam-pmace810 tepadutedon =MI daaraloellasea/malleas or alba mazaamllaa dbomaullosalmalleba hi ragradueloa dkaaulallim ta ERIC =Mal midi almlaale) mamma almlanlammla tirfaNC/theksIcallealbs =PA ainalanaalf

Ilacumsdasdil paisammed prouldad mmoduallaagy Istes paamilaa 1a madam samed. Mao lamla chadoribtaxamolavallimoomal *Laud I.

I hereby veleta the Ezetz: zekend assetacentaltsmeticrn.Center (ERIC; nantacinnatedenzinstoneitemtAce end daseneanta this=WIWI aglinciceted above. Reonzaction low the ERIC mianiche r ebabenic media *persona other Sea ERICengidayeesend 12 system nnattactoratequintensanktelamtememcepselghtheicha aceptinniensedettenamenzetzepandixthes Wired= andathersenem z agenda ia ma* !Ozanne= nen= of edungots n sewn=to Maui=Odes SIP here, 1'0 wint% ;deans SMILAddimw. AlCP TP/1776 .massitvq10,260 . Oosh 087-(r), DC, 2003 IL Doctmerr AVAILABILITY INFORMATION(FROM NON-ERICSOURC4. 11permission to rept:deals not *granted to thic, or, liyaawleh MCto distils aSalhabillly Ofthe provide the framing infonnetionregarding-the avalablity af the downed. documentiomemtlietscance, available, and a dependable VCRIC elf notannounce a doomed anima *sit dome can be spelled. Contributorashosidalso beware thenERIC it is stringent for documents thatcannot be made available through EDRS.) selection miteriamesitoicandymote

Publistrenebbibutor

Rice

IV. REFERRALOF-ERIC TO COPYRIGHT1REPRODUCTIONRIGHTS HOLDER: ithe right togrant this reemductdon .addretar Meas. is held by someone otherthenthe addressee, *lige pulite the appropdateawe. and

V. WHERETO SEND MISFORM:

Send this farm to thelaming ERIC Clearinghouse

However, it saleSedby the 811C-Fscifiy, ant conlehoesd) to: maidng atruseitliclett conttendhirto ERIC.' rehab thin faint (eM dm damns*hying =lc Owing/homeon Urban. ?escheats Colifige,ColmenheaUniversity- Box 40 525-w..12401 Street Piew Yak NY 10027

ToB ire= (BOO)601-4863 EFF-ime 6lee. mom C71Z47P4en Rim*. eric4no@oolumibtadedn