Childhood Victimization and Subsequent Risk for Promiscuity, :. Prostitution,<.!.::~~~~ and Teenage Pregnancy: A Prospective Study

Cat/lix Spatz Widom, PhD. andJoseph B. Klhts, MA Introduction neglect. Finally, reliance on retrospective accounts of childhood victimization leaves Clinical and behavioral reports and thc data open to a host of potcntial research studies have linked childhood biases."> victimization to a varlety of negativc With some exceptions, dcsigns lack health and behavioral consequences. Pros- appropriate comparison or control groups. titution.'-'9 promiscuity,'"2' and teenage Since child abuse is often associatcd with pregnancv2-34 have been included among low-income familics,3A4" comparison or thcsc scquelac. From a public health control groups are important. Low- perspcctiv,c these consequences of abu- income families often have many prob- sive and neglectful early childhood experi- lems, child abusc being only one of them. ences are particularly important in terms Because other factors, such as poverty, of potential sexually transmitted diseases, unemployment, parental alcoholism, drug unwanted and early pregnancies, and the problems, or othcrwise inadequatc social potential to facilitate a youth's transition and family functioning. are often present into prostitution. In addition, anv relation- in such multiproblem homes, control ship between childhood victimization and groups matched on socioeconomic status teenage prcgnancv is important to un- and other relevant variables become vital cover because of the link between prema- components of this research.4' The child ture parcnting and inadequate child- sexual abuse literature has been fairly rearing practices, factors that perpetuatc vague in separating effects directly related the cycle of abusc from one generation to to sexual abuse from effects possibly due the next. to precxisting psychopathology in the Existing research on the connections child. family dysfunction, or the strcss bctwccn childhood victimization and pro- associated with disclosure.42 Such con- miscuity prostitution, and teenage preg- founding of corrclated characteristics nancv varics in terms of studv dcsigns and needs to be avoided so that the uniquc findings. One of the major limitations of contribution of childhood sexual abusc, this body of research is its reliancc on physical abuse, or neglect can bc discn- corrclational (cross-scctional) designs. tanglcd from other background charactcr- with data collccted at one point in time. istics. In sum, while there is corrclational Correlational studies do not permit exami- evidence that childhood victimization may nation of causal sequences. For example. bc associated with incrcased risk of one conccrn in interpreting this litcrature prostitution, promisculty, and tcenage is that a tcenager's pregnanc-y may have pregnancy, current knowledge of the resulted from forced sexual intercourse or extent and nature of thcsc relationships is from promiscuity or irresponsible contra- ceptive protection following a history of sCxual abusc.-3 Another possibility is that The authors are with the School of Criminal Justice. The University at Alban., Albany, NY. the pregnancy could bc planncd as an Requests for reprints should be sent to escape from an abusive homc environ- Cathy Spatz Widom. PhD, The Universitv at ment.- While the majoritv of this re- Albanv. School of Criminal Justice. 135 West- search is based on the belief that child- ern Ave. Albanv, NY 1222. This paper was accepted July 3. 1996. hood victimization leads to certain .\ote. The views expressed here are the outcomes, it is possible that some of these authors' and do not necessarily represent those problem bchaviors precede abuse or of the US Department of Justice.

American Journal of Public Health 16f7 Widom and Kuhns ambiguous, at best, in the absence of cases of childhood physical and sexual individuals (20 years after their childhood longitudinal research. abuse and/or neglect processed during victimization) and control subjects to This paper examines the relationship the years 1967 through 1971 in the county document the long-term consequences of between early childhood victimization juvenile or adult criminal court (situated childhood victimization across a number and subsequent promiscuity, prostitution, in a metropolitan area in the Midwest). of domains of functioning (cognitive and and teenage pregnancy. The cohort de- These cases of early abuse and/or neglect intellectual, emotional, psychiatric, social sign and analytic techniques used permit- were restricted to children who were 11 and interpersonal, occupational, and gen- ted examination of the nature of these years of age or younger at the time of the eral health). Two-hour follow-up inter- relationships longitudinally while control- incident (n = 908). views, conducted between 1989 and 1995, ling for relevant demographic variables. Physical abuse cases included inju- consisted of a series of structured and On the basis of previous literature, the ries such as bruises, welts, burns, abra- semistructured questions and rating scales, overall hypothesis tested was that there sions, lacerations, wounds, cuts, bone and measures of IQ and reading ability, and a would be a direct relationship between skull fractures, and other evidence of psychiatric assessment. The interviewers early childhood victimization and subse- physical injury. Sexual abuse cases varied were unaware of the purpose of the study quent promiscuity, prostitution, and teen- from relatively nonspecific charges of and of the inclusion of an abused and age pregnancy. It was also hypothesized "assault and battery with intent to gratify neglected group. Similarly, the subjects that victims of childhood sexual abuse sexual desires" to more specific ones were unaware of the purpose of the study. would be at a particularly increased risk including "fondling or touching in an That is, subjects were told that they had for promiscuity, prostitution, and teenage obscene manner," sodomy, incest, and so been asked to participate as part of a large pregnancy. forth. Neglect cases reflected a judgment group of individuals who grew up in that that the parents' deficiencies in child care area during the late 1960s and early 1970s. Methods were beyond those found acceptable by Subjects who participated signed a con- community and professional standards at sent form acknowledging that they were In the current study, abused and the time. These cases represented ex- participating voluntarily. neglected children were matched with treme failure to provide adequate food, The findings described here are nonabused and nonneglected children clothing, shelter, and medical attention to based on completed interviews with 1196 and followed prospectively into young children. individuals. Of the original sample of adulthood.41'43 Characteristics of the pro- A critical component of this research 1575, 1291 subjects (82%) were located spective design included (1) an unambigu- involved the identification of a matched and 1196 were interviewed (76% overall, ous operationalization of abuse and ne- control group. Children who were under representing 74% of the abused and glect; (2) separate abused and neglected school age at the time of the abuse and/or neglected group and 78% of the control groups; (3) a large sample; (4) a control neglect incident were individually matched group). Of the 95 people not interviewed, group matched as closely as possible for with children ofthe same sex, race, date of 39 were deceased, 9 were incapable of age, sex, race, and approximate family birth (+1 week), and hospital of birth being interviewed, and 49 refused to social class background; and (5) assess- through the use of county birth record participate (a refusal rate of 3%). ment of the long-term consequences of information. For children of school age at Comparison of the current follow-up abuse and neglect beyond adolescence the time, elementary school records for sample with the original sample indicated and into adulthood (for complete details the same time period were used to find no significant differences in terms of of the study design and subject selection matches with children of the same sex, percentage male, White, or abused and/or criteria, see Widom41). race, date ofbirth (+6 months), and same neglected; poverty in childhood census It was important to match for social class in the same elementary school tract; or mean current age (abuse/neglect class in this study because it is theoreti- during the years 1967 through 1971. group = 28.6 years, control group = 28.9 cally plausible that any relation between Overall, there were matches for 74% of years). The interviewed group (follow-up child abuse or neglect and the outcomes the abused and neglected children. sample) was significantly more likely to under investigation would be confounded Nonmatches occurred for a number have had an official criminal history than or explained by social class differences. It of reasons. In terms of birth records, the original sample of 1575 (42% vs 36%). is difficult to match exactly for social class nonmatches occurred in situations in However, this is not surprising since because higher income families may live which the abused or neglected child was people with a criminal history are gener- in lower social class neighborhoods, and born outside the county or state or ally easier to find, in part because they vice versa. However, the matching proce- date-of-birth information was missing. In have more "institutional footprints" to dure used here was based on a broad terms of school records, nonmatches assist in locating them. definition of social class that included occurred because class registers were Variables neighborhoods in which children were unavailable as a result of the closing of the reared, schools they attended, and hospi- elementary school over the past 20 years Teenage pregnancy was defined as tals in which they were born. Busing was or lack of adequate identifying informa- having had a child before 18 years of age. not operational at this time, and students tion for the abused and neglected chil- This was based on a calculation of the age in elementary schools in the study county dren. of the oldest child in comparison with the were from small, socioeconomically ho- Initially, official criminal histories of age of the person at the time of the mogenous neighborhoods. the abused and neglected children were interview. Subjects who indicated that examined and compared with those of the they were less than 18 years old at the Subjects control group.41 A second phase of the birth of their first child were coded as The abused and neglected group research involved the locating and inter- having had a teenage pregnancy (n = 159). comprised individuals with substantiated viewing of these abused and/or neglected Subjects were defined as promiscuous if

1608 American Journal of Public Health November 1996, Vol. 86, No. 11 Child Abuse and Sexual Behaviors they reported having had "sex with 10 or more people within any single year" TABLE 1 -Prevalence of Promiscuity, Prostitution, and Teenage Pregnancy In a (n = 207). Prostitution was defined as a Cohort of 676 Young Adults Abused as Children and In 520 Control positive response to a question about Subjects whether the person had "ever been paid for having sex with someone" (n = 101). Teenage % These outcome variables were dichoto- No. Promiscuity, % Prostitution, % Pregnancy, mous (coded 1 or 0). Overall Independent variables included child- Abuse/neglect group 676 17.79 10.73** 14.35 hood victimization as well as control Neglect 543 18.92 11.13*** 15.29 variables of age, sex, and race. Group was Physical abuse 110 19.63 12.04* 11.82 a dichotomous variable based on official Sexual abuse 96 13.68 9.38 12.50 520 16.99 5.60 11.92 reports of early childhood abuse and/or Control group neglect (0 = control, 1 = abuse/neglect). Female subjects Abuse/neglect group 338 7.14 8.93** 23.96 In analyses focusing on consequences Neglect 257 7.84 9.02** 26.46 associated with different types of abuse or Physical abuse 47 12.77 12.77** 25.53 neglect, three categories (separate dummy Sexual abuse 76 6.58 10.53** 15.79 variables) based on official reports of Control group 244 5.74 2.87 20.90 childhood victimization were used: any Male subjects physical abuse, any sexual abuse, and any Abuse/neglect group 338 28.53 12.54 4.73 Neglect 286 28.87 13.03 5.24 neglect. It should be noted that respon- Physical abuse 63 25.00 11.48 1.59 dents may have experienced more than Sexual abuse 20 42.11 5.00 0.00 one type of abuse or neglect. Control group 276 27.01 8.03 3.99 Multivariate analyses controlled for sex, race/ethnicity, age, and approximate Note. AJl statistical comparisons (chi-square tests) were made with the control group. Respondents could be coded 1 for more than one type of abuse (thus, the sum of the abuse/neglect group family social class. Race/ethnicity was sizes does not equal the total number of abused/neglected subjects). based on respondents' self-identification. *P < .05; **P < .01; ***P < .001 (two-tailed tests). White non-Hispanic was coded 1; all other race/ethnicity categories (Hispanic and non-Hispanic Black, Hispanic White, American Indian, and Pacific Islander) regression. However, since all subjects, 9% of the abused and neglected female were coded 0. Age was an interval-level not only matched pairs, were used in the subjects, as compared with 3% of the variable indicating the person's age at the analysis, conditional analyses proved infea- female control subjects, reported having time ofthe interview. The mean age of the sible. The results of these analyses were engaged in prostitution. Physically abused sample at the time of the interview was essentially the same as those presented female subjects had the highest rates 29.2 years (SD = 3.8). As an indicator of here. (12.8%) of prostitution, followed by sexu- the family's approximate social class dur- ally abused (10.5%) and neglected (9.0%) ing the person's childhood, a variable Results female subjects. labeled "welfare as a child" was included. Multivariate results are presented A positive score (1) on this dichotomous The first hypothesis examined separately for promiscuity (Table 2), variable reflected the respondent's report whether experiencing childhoodvictimiza- prostitution (Table 3), and teenage preg- that his or her family had received welfare tion was associated with increased risk for nancy (Table 4). Group (abuse/neglect vs when the respondent was a child. promiscuity, prostitution, or teenage preg- control) was not a significant predictor of Initial results are presented in terms nancy. Table 1 presents the prevalence of promiscuity for the sample overall or for of bivariate statistics. Multivariate analy- promiscuity, prostitution, and teenage male and female subjects separately (Table ses using logistic regression were con- pregnancy by group, gender, and type of 2). Furthermore, none of the specific ducted to determine whether the findings abuse or neglect. The most striking find- types ofabuse or neglect predicted promis- at the bivariate level would be supported ing was the lack of difference between the cuity (Table 2). in adjusted analyses after demographic abused/neglected group and the control Group (i.e., childhood abuse or ne- characteristics had been controlled. Logis- group (overall and for male and female glect) remained a significant predictor of tic regression equations were estimated subjects separately) in terms of promiscu- prostitution, after control for demo- with age, race, sex, and welfare as a child ity and teenage pregnancy (see Table 1). graphic characteristics and welfare as a as control variables predicting the dichoto- Only the likelihood of engaging in child, for the sample overall and for mous dependent variables of prostitution, prostitution differentiated childhood abuse female subjects (Table 3). For male promiscuity, and teenage pregnancy. A and neglect victims from individuals who subjects, the effect of group did not reach second set of equations was estimated to did not have official records of abuse or the customary level of significance examine the contributions of different neglect, and a significant increase in risk (P = .13), suggesting that the effect was interac- types of abuse or neglect. Analyses were for prostitution was confined to female confined to female subjects. The performed for the overall sample and for subjects. The difference in prevalence of tion between gender and group was not male and female subjects separately. prostitution for abused and neglected significant (P < .14). Childhood sexual of These analyses were also replicated with male subjects vs male control subjects abuse was a significant predictor prosti- Ne- matched pairs only and with a version of approached, but did not reach, the conven- tution in female subjects (Table 3). conditional maximum likelihood logistic tional level of significance. Overall, about glect also predicted prostitution for the

November 1996, Vol. 86, No. 11 American Journal of Public Health 1609 Widom and Kuhns

prostitution. However, after these factors 3), the physical TABLE 2-Logistic Regressions Predicting Promiscuity in Young Adults Abused were controlled for (Table as Children abuse effect became more borderline and not significant. Overall Female Subjects Male Subjects Table 4 presents the results of logis- (n = 1187) (n = 580) (n = 607) tic regression equations predicting teen- age pregnancy by group (abuse/neglect vs OR 95% Cl OR 95% Cl OR 95% Cl control) and by type of abuse or neglect. Abuse/neglect 1.06 0.77, 1.48 1.07 0.53, 2.16 1.06 0.73,1.54 Surprisingly, neither group nor type of abuse or neglect was a significant predic- Type of abuse Sexual abuse 1.57 0.81, 3.06 1.17 0.42, 3.22 2.37 0.91, 6.16 tor of teenage pregnancy for either the Physical abuse 1.10 0.65, 1.86 1.89 0.73, 4.87 0.90 0.48, 1.67 overall sample or the females. This find- Neglect 1.14 0.83,1.59 1.22 0.61, 2.46 1.10 0.76,1.60 ing contrasts dramatically with previous literature, although differences between Note. Sample sizes given are both case and control subjects. All statistical comparisons were made the present design and others may help with the control group. Age, race, gender, and an indicator of socioeconomic status (receiving welfare as a child) were controlled in the regression equations. OR = odds ratio; Cl = confidence explain the discrepancy. interval. Discussion Despite widespread belief that child- hood victimization is associated with TABLE 3-Logistic Regressions Predicting Prostitution in Young Adults Abused as Children promiscuity, prostitution, and teenage pregnancy, our findings, based on a Overall Female Subjects Male Subjects prospective cohort design, indicate that (n = 1189) (n = 580) (n = 609) childhood victimization is not a signifi- icant risk factor for promiscuity or teen- OR 95% Cl OR 95% Cl OR 95% Cl age pregnancy. That is, we found no Abuse/neglect 1.88** 1.19, 2.99 2.96* 1.26, 6.96 1.54 0.88, 2.71 significant relationship between early childhood abuse/neglect and promiscuity Type of abuse in bivariate Sexual abuse 1.73 0.80, 3.78 2.54* 1.02, 6.32 0.63 0.08, 5.00 and teenage pregnancy, either Physical abuse 1.79 0.94, 3.40 2.63 0.99, 6.96 1.35 0.58, 3.18 or in multivariate analyses that controlled Neglect 1.78* 1.15, 2.77 2.58* 1.20, 5.55 1.48 0.86, 2.55 for age, race, sex, and welfare status as a child. Thus, our results do not conform to Note. Sample sizes given are both case and control subjects. All statistical comparisons were made general expectations or to previous re- with the control group. Age, race, gender, and an indicator of socioeconomic status (receiving welfare as a child) were controlled in the regression equations. OR = odds ratio; Cl = confidence ports about these two outcomes. How- interval. ever, this study differs from most past *P < .05; **P < .01 (two-tailed tests). research in the prospective nature of the design and the inclusion of a control group. The prospective nature of this study allows some issues of causality to be and the effects of TABLE 4-Logistic Regressions Predicting Teenage Pregnancy in Young Adults examined disentangles Abused as Children childhood victimization from other poten- tial confounding effects. Overall Female Subjects Male Subjects The outcomes that other researchers (n = 1196) (n = 582) (n = 614) have found and attributed to early child- hood victimization experiences (particu- OR 95% Cl OR 95% Cl OR 95% Cl larly sexual abuse) may not be specifically Abuse/neglect 1.10 0.76, 1.59 1.11 0.73,1.67 1.07 0.47,2.40 associated with childhood victimization or sexual abuse in particular. Although a Type of abuse high percentage of young women become Sexualabuse 0.62 0.32,1.20 0.65 0.33,1.28 ...... Physical abuse 1.05 0.55,1.98 1.31 0.65, 2.64 ... pregnant as teenagers and many report Neglect 1.30 0.90,1.87 1.31 0.87,1.98 ...... having experienced earlier sexual abuse, there may be other factors that explain Note. Sample sizes given are both case and control subjects. All statistical comparisons were made the apparent relationship. For example, with the control group. Age, race, gender, and an indicator of socioeconomic status (receiving welfare as a child) were controlled in the regression equations. OR = odds ratio; Cl = confidence Leibowitz et al.44 adopted a primarily interval. economic perspective to explain teenage pregnancy, whereas Cooksey45 examined the effects of family background factors, including family structure and size, paren- sample overall and for female subjects, with no control for demographic character- tal education and employment status, and whereas the effect for male subjects did istics or approximate family social class religious affiliation. Cooksey46 showed not reach the customary level of signifi- (Table 1), physically abused female sub- that many background factors predicted there was a cance (P = .16). It should be noted that, jects had significantly higher rates of premarital pregnancy (e.g.,

1610 American Journal of Public Health November 1996, Vol. 86, No. 1 1 Child Abuse and Sexual Behaviors higher likelihood for Blacks than for prostitution reduced the contribution of underestimate of the extent of teenage Whites or Hispanics and a lower likeli- childhood victimization as a predictor of pregnancies experienced. Our measure hood for individuals with an intact family prostitution overall. However, childhood did not include women who were preg- structure and few siblings). victimization remained a significant predic- nant but whose pregnancies ended in Thus, high rates of teenage preg- tor (data not shown). abortions or miscarriages. It is possible nancy may be found in young women who One issue in considering these prosti- that women whose children died might come from economically deprived back- tution findings, however, is the extent to have failed to mention these children. We grounds (i.e., from families who may have which the relationship between early hope that other researchers engaged in multiple problems, both including and childhood victimization and prostitution prospective research with different excluding child abuse or neglect). In is part of a larger syndrome of problem samples from different geographic areas previous studies not involving a demo- behaviors for which abused and neglected and time periods will examine these issues graphically matched control group, this children are at risk.48 Previous research and attempt to replicate our findings. O pattern would not have been evident. has indicated that abused and neglected Also, because studies often involve preg- children are at an increased risk of being nant teenagers who are asked about their arrested as a juvenile, as an adult, or for a Acknowledgments earlier backgrounds, abused and ne- violent crime.49 Future analyses will at- This research was supported by grants from the glected children who did not become tempt to disentangle whether the in- National Institute of Mental Health (MH49467) pregnant would not be included, thus creased risk of prostitution described here and the National Institute ofJustice (86-IJ-CX- an incomplete picture of the 0033 and 89-IJ-CX-0007). providing is specific (and should be interpreted in We express great appreciation to Patricia long-term consequences of early child- the victimization context) or whether it is Glynn for her help with data management and hood victimization. part of a lifestyle that places abused and statistical analyses and to Deborah Coppola for On the other hand, we did find that neglected children at increased risk in her help in preparing the manuscript. early childhood victimization was associ- general for criminal behavior. ated with increased risk for prostitution, Another issue concerns the role References although this was true only for female substance abuse plays in relation to 1. Allen D. Young male prostitutes: a psycho- subjects. These findings reinforce the prostitution. Among Bagley and Young's3 social study. Arch Sex Behav. 1980;9:399- earlier literature calling attention to the sample of prostitutes, the primary reason 426. role of childhood victimization in the 2. Bagley R. Sexual Offenses against Children. for entrance into prostitution reportedly New York, NY: Putnam, 1984. backgrounds of prostitutes. Early child- was the need to support a drug habit. 3. Bagley C, Young L. Juvenile prostitution hood abuse and neglect appear to place Seng's9 sample of prostitution-involved and child sexual abuse: a controlled study. children at increased risk of becoming rates Can J Community Ment Health. 1987;6:5- children reported significantly higher 26. prostitutes, which reinforces the impor- of alcohol and drug abuse than did the tance ofviewing prostitution in a victimiza- 4. Bracey D. The juvenile prostitute: victim sexually abused children who were not and offender. Victimology. 1983;8:151. tion context. involved in prostitution. Other studies 5. Burgess AW, Hartman CR, McCormack Despite an emphasis on childhood have reported correlations between pros- A. Abused to abuser: antecedents of sexual abuse as an antecedent to prostitu- socially deviant behaviors. Am J Psychiatry. titution and drug use.5056 Although such tion, we found that childhood neglect is 1987;144:1431-1436. was the of the also associated with increased risk for an analysis beyond scope 6. Earls CM, David. Early family and sexual we to examine the experiences of male and female prosti- prostitution. By definition, childhood ne- present paper, plan between substance use and tutes. Canada's Ment Health. December glect involves inadequate supervision and relationship 1990:7-11. inadequate caretaking by a parent or abuse and prostitution (particularly look- 7. Greenwald H. The Elegant Prostitute. New described pros- ing at temporal sequences) in future York, NY: Ballantine Books; 1970. guardian. Bracey4 juvenile 8. James J, Meyerding J. Early sexual experi- titutes as being neglected at home, quot- research. While this study design has clear ence as a factor in prostitution. Arch Sex ing one young woman who reported that Behav. 1977;7:31-42. her "parents didn't tell me to get out, but strengths, certain caveats should also be 9. Seng M. Child sexual abuse and adolescent they didn't come looking for me when I kept in mind. First, these cases of child- prostitution: a comparative analysis.Adoles- did." Neglected young children, on the hood abuse and neglect came to the cence. 1989;24:665-675. attention of the authorities and were 10. Silbert M, Pines A. Sexual child abuse as streets alone, are at risk of being victim- an antecedent of prostitution. Child Abuse ized or enticed into prostitution. For skewed toward the lower end of the Negl. 1981;5:407-411. example, runaways may come under the socioeconomic spectrum.39 Members of 11. Silbert M, Pines A. Early sexual exploita- control of pornographers and pimps and our matched control group came from the tion as an influence in prostitution. Soc same neighborhoods as the abused and Work ResAbstracts. 1983;28:285-290. become susceptible to subsequent physi- 12. Simons RL, Whitbeck LB. Sexual abuse as cal and sexual victimization by pimps and neglected children and were also from a precursor to prostitution and victimiza- customers.47 lower socioeconomic classes. This means tion among adolescent and adult homeless We have also found that abused and that the sample is not representative of women. JFam Issues. 1991;12:361-379. the as a whole and that these 13. Schaffer B, DeBlassie R. Adolescent pros- neglected children are at increased risk, population titution.Adolescence. 1984;19:689-696. relative to nonabused and nonneglected findings cannot be generalized to all cases 14. Newton-Ruddy L, Hundelsman M. Jung- control children, of being picked up by the ofabuse and neglect. For example, middle- ian feminine psychology and adolescent police for running away48 and of reporting and upper-class women who might have prostitutes. Adolescence. 1986;21:21 1. experienced childhood sexual abuse were 15. Widom CS, Ames MA. Criminal conse- having run away overnight. The introduc- quences of childhood sexual victimization. tion of "running away" into our regression less likely to be included here. ChildAbuse Negl. 1994;18:303-318. equations to estimate its role as a media- Second, the teenage pregnancy mea- 16. Conte JR. The effect of sexual abuse on tor between childhood victimization and sure used here may have represented an children: a critique and suggestions for

November 1996, Vol. 86, No. 11 American Journal of Public Health 1611 Widom and Kuhns

further research. Victimology. 1985;10:110- 31. O'Brien JD. The effects of incest on female factors for child abuse. Child Abuse Negl. 130. adolescent behavior. J Am Acad Psycho- 1982;6: 113-123. 17. Browne A, Finkelhor D. Impact of child anal. 1987;15:83-92. 44. Leibowitz A, Eisen M, Chow WK. An sexual abuse: a review of the research. 32. Oz S, Fine M. A comparison of childhood economic model of teenage pregnancy Psychological Bull. 1986;99:66-77. backgrounds of teenage mothers and their decision-making. Demography. 1986;23:67- 18. Burgess AW. The Sexual Victimization of non-mother peers: a new formulation. J 77. Adolescents. Washington, DC: National Adolescence. 1988;11:251-261. 45. Cooksey EC. Factors in the resolution of Institute of Mental Health, 1985. 33. Gershenson HP, Musick JS, Ruch-Ross adolescent premarital pregnancies. Demog- 19. Courtois CA. The incest experience and its HS, Magee V, Rubino KK, Rosenberg D. raphy. 1990;27:207-218. aftermath. Victimology. 1979;4:337-347. The prevalence of coercive sexual experi- 46. Cooksey EC. Outcomes ofAdolescent First 20. DeYoung M. The Sexual Victimization of ence among teenage mothers. J Interpers Premarital Pregnancies: The Influence of Children. Jefferson, NC: McFarland Press; iol. 1989;4:204-219. Family Background. Providence, RI: Brown 1982. 34. Boyer D, Fine D. Sexual abuse as a factor University; 1988. Dissertation. 21. Finkelhor D. Sexually Victimized Children. in adolescent pregnancy and child maltreat- 47. Fisher B, Weisberg DK, Marotta T. Report New York, NY: Free Press; 1979. ment. Fam Plan Perspect. 1992;24:4-11. on Adolescent Male Prostitution. San Fran- 22. Finkelhor D, Hotalling G, Lewis I, Smith 35. Briere J. Methodological issues in the cisco, Calif: Urban Rural System Associ- C. Sexual abuse and its relationship to later study of sexual abuse effects. J Consult Clin ates; 1982. sexual satisfaction, martial status, religion, Psychol. 1992;60:196-203. 48. Widom CS. Childhood victimization and and attitudes. J Interpers Viol. 1989;4:379- 36. Cicchetti D, Rizley R. Developmental risk for adolescent problem behaviors. In: 399. perspectives on the etiology, intergenera- Lamb ME, Ketterlinus R, eds. Adolescent 23. Herman J, Hirschman K. Father-daughter tional transmission, and sequelae of child Problem Behaviors. New York, NY: incest: a feminist theoretical perspective. J maltreatment. New Dir Child Dev. 1981;1 1: Erlbaum, 1994. Women Cult. Soc. 1977;2:735-756. 31-55. 49. Widom CS. The cycle of violence. Science. 37. Yarrow MR, Campbell JD, Burton RV. 1989;244:160-166. 24. Lukianowicz N. Incest I: parental incest. Br Recollections of childhood: a study of the 50. Inciardi JA. Trading sex for crack among J Psychiatry. 1972; 120: 130-133. retrospective method. Monogr Soc Res juvenile drug users: a research note. Con- 25. Cavaiola A, Schiff M. Behavioral sequelae Child Dev 1970;135(138, 5). temp Drug Probl. Winter 1989:689-700. of physical and/or sexual abuse in adoles- 38. Widom CS. Sampling biases and implica- 51. Inciardi JA, Pottieger AE, Fomey MA, cents. ChildAbuse Negl. 1988;12:181-188. tions for child abuse research. Am J Chitwood DD, McBride DC. Prostitution, 26. Laumann EO, Gagnon JH, Michael RT, Orthopsychiatry. 1988;58:260-270. IV drug use, and sex-for-crack exchanges Michaels S. The Social Organization of 39. Newberger EH, Reed RB, Daniel JH, among serious delinquents: risks for HIV Sexuality: Sexual Practices in the United Hyde JN, Kotelchuck M. Pediatric social infection. Criminology. 1991;29:221-235. States. Chicago. Ill: University of Chicago illness: toward an etiological classification. 52. Kuhns JB, Heide KM, Silverman I. Sub- Press; 1994. Pediatncs. 1977;50:178-185. stance use/misuse among female prosti- 27. Butler JR, Burton LM. Rethinking teenage 40. Straus MA, Gelles RJ, eds. Physical Vio- tutes and female arrestees. Int J Addict. childbearing: is sexual abuse a missing lence inAmerican Families: Risk Factors and 1992;27:1283-1292. link? Fam Relations. 1990;39:73-80. Adaptations to Violence in 8145 Families. 53. Philpot CR, Harcourt CL, Edwards JM. 28. DeFrancis V. Protecting the Child Victim of New Brunswick, NJ: Transaction Pulish- Drug use by prostitutes in Sydney. Br J Sex Crimes Committed by Adults. Denver, ers, 1990. Addict. 1989;84:499-505. Colo: American Humane Association, 1969. 41. Widom CS. Child abuse, neglect and adult 54. Marshall N, Hendtlass J. Drugs and prosti- 29. Donaldson PE, Whalen MH, Anastas JW. behavior: research design and findings on tution. J Drug Issues. Spring 1986:227-248. Teen pregnancy and sexual abuse: explor- criminality, violence, and child abuse. Am J 55. Goldstein PJ. Prostitution and Drugs. ing the connection. Smith Coll Stud Soc Orthopsychiatry. 1989;59:355-367. Toronto, Ontario, Canada: DC Heath & Work 1989;59:289-300. 42. Beitchman JH, Zucker KJ, Hood JE, Co; 1979. 30. Landry E, Bertrand J, Cherry J, Rice J. daCosta GA, Akman D. A review of the 56. Cottler LB, Helzer JE, Tipp JE. Lifetime Teen pregnancy in New Orleans: factors short-term effects of childhood sexual patterns of substance use among general that differentiate teens who deliver, abort abuse. Child Abuse Negl. 1991;15:537-556. population subjects engaging in high risk or successfully contracept. J Youth Adoles- 43. Leventhal JM. Research strategies and sexual behaviors: implications for HIV risk. cence. 1986;15:259-274. methodologic standards in studies of risk AmJDrugAlcoholAbuse. 1990:16:207-222.

1612 American Journal of Public Health November 1996, Vol. 86, No. 11