Let's Talk About Sex Education: the Effect of State Mandated School

Home , Sex education curriculum

Let’s Talk About Sex Education: The Eﬀect of State Mandated School-Based Sex Education on Teenage Sexual Behaviors and Health

September 29, 2017

Abstract

This study is the first to examine the effect of US state mandated school-based sex education on teenage sexual behaviors and health. Using data from the Youth Risk Behavioral Surveys, National Vital Statistics, and the CDC’s Wonder statistics on sexually transmitted diseases, difference-in-difference results show the typical state sex education mandate increases teenage condom use by 3%, and decreases teenage chlamydia rates by 8%. No effect is found for sexual activity, contraception use, or teenage births. A back-of-the-envelope calculation indicates that the 8% decrease in chlamydia rates results in $41 million in savings for annual chlamydia care costs compared to the $5.4 million annually it would cost to educate teens about sex. Exploring the policy heterogeneity with respect to abstinence-based and comprehensive requirements within a states’ sex education law, the results generally point to no differential effect of abstinence-based or comprehensive teaching on teenage sexual behaviors and health. A causal interpretation of the results is supported by an event study exercise, and a falsification test on older young adults for whom sex education mandates do not bind.

Keywords: sex education; sexual health; sexually transmitted diseases; teenage births

JEL: I12, I18, J13, J18

1 1 Introduction

In the mid-1980s, once it was recognized that AIDS could be spread via sexual intercourse, Surgeon General Everett Koop called for increased sex education in schools begin- ning as early as the third grade (Cornblatt, 2009). By the late 1980s and 1990s, many states started implementing HIV and Sexually Transmitted Disease (STD) education and sex education due to concerns over HIV and teenage pregnancy. As the level of concern over HIV/STDs and teenage pregnancy steadily increased, states continued to implement and encourage sex education (Carter, 2001). Currently, 24 states and Washington DC mandate school-based sex education. There is, however, considerable heterogeneity in the timing and comprehensiveness of these mandates. For example, some states’ sex education programs are comprehensive in nature, meaning they require information on adolescent development, conception and pregnancy, abstinence and contraception effectiveness, while others are solely abstinence-based (Mullinax et al., 2017). Despite the potential benefits of sex education, these classes remain controversial. The debate over school-based sex education in the United States is centered on two major questions: do schools have a responsibility to teach students about issues related to sex, and if schools do teach sex education, what type of information should be presented? That is, what type of sex education is most effective at preventing teen pregnancy, and reducing disease transmission, for example. Using data from the Youth Risk Behavioral Surveys, National Vital Statistics, and the Center for Disease Control’s Wonder statistics on STDs, this study presents the first examination of the effect of state-level sex education mandates on teenage sexual behaviors, STDs, and birth rates. The primary goal of school-based sex education is to help young people build a foundation to mature into sexually healthy adults by assisting them in understanding a positive view of sexuality, providing them with information and skills for taking care of their sexual health, and promoting youth to make sound decisions now and in the future (Bridges & Hauser, 2014). As evidenced by the Surgeon General’s call, sex education programs are also viewed as an informational policy tool intended to reduce the future costs of STDs and teen pregnancy (Sabia, 2006). Sexually transmitted diseases are a severe public health problem in the United States. STDs cause harmful, often irreversible, and costly complications, especially among females (Weinstock et al., 2004). There are approximately 20 million new STD infections each year. Nearly half of these infections are among young people ages 15 to 24, who represent only twenty-five percent of the sexually active population (Weinstock et al., 2004). The estimated

2 cost to the US health care system from these new infections is $16 billion annually, including HIV and HPV (human papillomavirus) diagnoses (Workowski & Bolan, 2015). Among the non-viral STDs, chlamydia is the most common and costly infection, estimated at almost $517 million in annual health care costs (Owusu-Edusei Jr et al., 2013). Likewise, the economic costs of teenage childbearing may be sizable, especially for tax- payers and society as a whole. Several studies have found that teen childbearing is associated with declines in human capital attainment or future earnings for the teen mother (Angrist & Evans, 1999; Bronars & Grogger, 1994). However, the causal link remains unclear. Hotz et al. (1997) find little evidence that teenage childbearing harms teen mother’s socioeconomic outcomes when comparing teens who give birth with those who miscarry. Indeed, Hotz et al. (1997) cannot rule out the possibility that early childbearing adversely affects the fathers of children born to teen mothers, and the children themselves. Evidence suggests that men who father children of teen mothers would have had substantially higher incomes had they delayed childbearing, and the children of teenage mothers tend to fare poorly (i.e. lower cognitive development, academic achievement, health, behavior, etc.) compared to children born to older mothers (Hoffman & Maynard, 2008). The causal link between the effect of teen childbearing on the fathers and children is questioned due to these studies simply comparing children born to teenagers to children born to older individuals. By exploiting within-state variation in sex education mandates from 1991-2013 and addressing the potential endogeneity of state sex education mandates, the difference-in- difference results of the current study show that the typical state sex education mandate increases teenage condom use by 3%, and decreases teenage chlamydia rates by 8%. A back-of-the-envelope calculation implies that the 8% decrease in chlamydia rates results in $41 million in savings for annual STD care costs compared to the $5.4 million annually it would cost to educate teens about sex. Exploring the policy heterogeneity with respect to abstinence-only and comprehensive requirements within a states’ sex education law, the results generally point to no differential effect of abstinence-only and comprehensive teaching on teenage sexual behaviors or health.

2 Background

2.1 Brief History

Support for sex education began in the late 1800s when mass public campaigns promoted the “regulation of sexuality” and emphasized risk-reduction practices and health care pre-

3 vention in response to cholera and syphilis epidemics (Irvine, 2004). Momentum continued throughout the early and mid-1900s, until opposition towards sex education began to be or- ganized by groups such as the John Birch Society, Christian Crusade, and Parents Opposed to Sex and Sensitivity Education (Scales, 1981).1 By the early 1970s, twenty states had voted to restrict or abolish sex education. As mentioned above, the 1980s brought renewed interest in sex education as a result of concerns over teen pregnancy and HIV/AIDS that motivated widespread public support for sex education in schools (SIECUS, 2010). As a response, the Reagan administration began federal funding for abstinence-only-until-marriage programs, which gained momentum during the 1990s and early 2000s. Since 1997, Congress has provided over $1.5 billion for abstinence-only programs (SIECUS, 2010). Federal funds for such programs began easing after a report by Mathematica Policy Re- search was released in 2007 that found abstinence-only programs had no eﬀect on sexual behavior outcomes (Trenholm et al., 2007).2 In 2014, at the request of the Obama administration, Congress provided $185 million for medically accurate and age-appropriate sex education programs. In his proposed federal budget for 2017, former President Obama removed all funding for abstinence-only education.3Although the federal government has provided funding for public and private sex education programs, there is no federal law or policy that requires sex education to be taught in schools. Rather, the decision to mandate school-based sex education is left up to the state and local school districts.

2.2 Theoretical Framework

According to the National Sexuality Education Standards (NSES), a representative school- based sex education mandate should include information on seven key components: anatomy and physiology, puberty and adolescent development, identity, pregnancy and reproduction, STDs and HIV, healthy relationships, and personal safety (Barr et al., 2014). However, the general requirements for school-based sex education courses vary signiﬁcantly across states.

1These groups argued that sex education was “smut”, “immoral”, and “a ﬁlthy communist plot” (Scales, 1981). 2The Mathematica study examines the eﬀect of four Title V, Section 510 grants: My Choice, My Future; ReCapturing the Vision; Families United to Prevent Teen Pregnancy; and Teens in Control, on teens’ sexual abstinence, their risks of pregnancy and sexually transmitted diseases, and other behavioral outcomes. 3Federal funding for abstinence-only programs included funding for the Community-Based Abstinence Edu- cation grant program and the abstinence-only funding granted as part of the Adolescent Family Life Act, among others. Funding under the Obama administration was slated for “competitive contracts and grants to public and private entities to fund medically accurate and age appropriate programs that reduce teen pregnancy”(111th Congress, 2009).

4 Some states require abstinence-based sex education, which stresses or covers the importance of abstinence until marriage and includes little to no information on contraceptives. Other states are comprehensive in nature, meaning they provide information that is closely aligned with NSES’s seven components (refer to Table 1). How might school-based sex education affect teenage sexual behavior? Rational individuals become sexually active at the age at which the perceived benefits from sexual intercourse surpass the perceived costs (Oettinger, 1999). If sex education teaches teenagers about the costs associated with pregnancy and sexually transmitted diseases, including mental, physical, and monetary costs, their expected costs and benefits of engaging in sexual behaviors may be altered. Sex education should only affect teens’ perceptions about sex and their sexual behaviors to the extent that it presents new information. This new information could be presented in multiple ways and in practice typically takes one of two forms: abstinence-based or comprehensive. Abstinence-based sex education typically stresses the following: students should abstain from sexual activity until after marriage; abstinence from sex is the only 100% effective way to avoid unwanted pregnancy, STDs and HIV; conceiving a child out of wedlock is likely to have harmful consequences for the child, the child’s parents and society; and failure rates associated with condom use (Alford, 2001). Abstinence-based sex education should have no effect on the sexual behavior of teens who, under any circumstances, would never have sex. That is, any new information they obtain will not alter their behavior given their preference for abstinence. For teens who prefer sexual activity but wish to avoid a pregnancy, abstinence-based education should decrease the level of sexual activity and reduce the risk of pregnancy, due to possible declines in the frequency of sexual activity. Comprehensive sex education courses tend to be age-appropriate, and include medically accurate information on topics like human development, relationships, decision making, abstinence, contraception, and disease prevention (Alford, 2001). Again, for teens who prefer to remain abstinent, comprehensive education should have no effect on their sexual behavior. For teens who prefer sexual activity but want to avoid a pregnancy, comprehensive sex education should decrease the risk of pregnancy or STDs for any level of sexual activity, by, for example, increasing contraception use. Recall that sex education should only affect teens’ sexual behaviors to the extent that it presents new information. Sex education should have a greater impact on teens who gain a lot of new information, such as teens without low-cost alternative sources of sexual information (Oettinger, 1999). Older teenagers, age 16 and up, are more likely to have a drivers license, a job, and potentially more access to the internet and media, all of which can provide more

5 opportunities to engage in sexual activity. Additionally, teens with older siblings and higher levels of peer interaction are likely to have more information about sex than those without such low-cost sources. Therefore, younger teenagers, those without older siblings, and those with less peer interactions are more likely to be affected by new information provided by sex education.4 Does this new information provided by sex education affect teenagers knowledge of sex? If so, does this increase in knowledge affect their sexual behaviors? Several studies have shown that sex education programs have increased teenagers’ knowledge about sexual health issues (Eisen & Zellman, 1986; Kim et al., 1997; Reichelt & Werley, 1975; Sanderson, 2000). Kim et al. (1997) and Dupas (2011) find that sex education programs not only increase teens’ knowledge about sex, but also affect their sexual behaviors.

2.3 Previous Literature

This study contributes to the existing literature in several ways. First, I exploit the within-state variation in state sex education mandates to estimate their effect on teenage sexual behaviors – an identification strategy superior to the cross-sectional identification employed in prior studies. Since the previous literature fails to address the potential endogeneity of sex education mandates, I assess the credibility of the common trends assumption of my identification strategy through an event-study exercise. Second, I am the first to examine the effect of sex education on teenage sexual behaviors using data drawn from repeated cross-sections of both the National and State Youth Risk Behavior Surveys (YRBS) from 1991 to 2013. The use of individual-level data allows me to estimate the effect of sex education on finer measures of teenage sexual behaviors, such as sexual activity, and condom use or contraception use at last sex. Finally, this study is one of the first to credibly explore whether the effect of school-based sex education extends to teenage STD rates, an outcome that has both private and public health consequences. The empirical work on sex education is extensive, and has attempted to provide a clearer understanding of the potential tradeoff in sex education. Cross-sectional studies examining the effect of sex education have found that abstinence-based sex education is not associated with a reduction in the likelihood of having sexual intercourse, and is associated with an increase in pregnancy rates among teens (Haffner, 1997; Kirby, 2008; Kohler et al., 2008;

4Since the YRBS does not include family composition information, such as if the survey participant has siblings, or information regarding friendships and peer interactions, I am unable to test the hypothesis that teens with older siblings and those with more peer interactions potentially respond diﬀerently to sex education than those who do not.

6 Stanger-Hall & Hall, 2011; Starkman & Rajani, 2002). Cross-sectional studies have also found that comprehensive sex education is associated with an increase in condom and contraception use, a decrease in teenage pregnancy rates, and a decrease in the number of teens entering into sexual intercourse (Haffner, 1997; Kirby, 2008; Kohler et al., 2008; Stanger-Hall & Hall, 2011; Starkman & Rajani, 2002). A causal interpretation of these studies is hindered by the fact that they all fail to address the potential endogeneity of sex education programs. Additionally, many studies have also implemented randomized control trials to assess the impact of sex education. Although a causal interpretation of the results is possible, most sex education randomized control trials were targeted toward at-risk populations, therefore lacking external validity.5 Studies using a quasi-experimental approach to assess the impact of sex education have generally found no evidence of a causal association between sex education and sexual behaviors and health. Sabia (2006) uses data from the National Longitudinal Study of Adolescent Health and finds that the causal link between sex education and adverse health outcomes is gone after controlling for unobserved heterogeneity via fixed effects and instrumental variables. Carr & Packham (2016) use a difference-in-difference design to estimate the effect of state-level abstinence education mandates on teen health outcomes. The authors find that the abstinence education mandates have no effect on teen birth rates or abortion rates, and may potentially affect teen STD rates in some states. Kearney & Levine (2015) also use a difference-in-difference design to determine the role of state-level demographic changes, economic conditions, and targeted policies on recent trends in the US teen birth rate. The authors control for state-level sex education curriculum and sex education curriculums requiring the teaching of contraception in their model, and find that these policies have no effect on the teen birth rate. The current study differs from Carr & Packham (2016) and Kearney & Levine (2015) by focusing generally on nationwide sex education mandates from 1991-2013, and by estimating the effect on individual-level teenage sexual behaviors to provide a clearer understanding of the mechanisms affecting teen birth, abortion, and STD rates. 5For a more comprehensive review of sex education RCTs, see Bennett & Assefi (2005); DiCenso et al. (2002); Kirby (2008).

7 3 Data and Measures

3.1 Data

My primary analysis will use data drawn from three sources. First, I use repeated cross- sections of both the National and State Youth Risk Behavior Surveys (YRBS) from 1991 to 2013. The National and State YRBS surveys are conducted biennially by the Center for Disease Control and Prevention (CDC) and administered to high school students. While the state surveys are coordinated by the CDC, they are usually conducted by state education and health agencies.6 The YRBS is well suited for this study because it contains data on several measures of student sexual behaviors, including sexual activity, and condom and contraception use. Second, I use state-level chlamydia and birth rates for 15 to 19 year olds to supplement the YRBS data. These data were obtained from the CDC’s Wonder statistics on STDs, and the National Vital Statistics, respectively, to estimate the eﬀect of sex education mandates on teenage STD and birth rates.

3.2 YRBS Survey Outcomes

Using the YRBS data, I identify three key measures of teenage sexual activity. First, I create an indicator equal to 1 if the student indicated they had ever had sex, and zero otherwise. I ﬁnd that 48 percent of the sample indicated they have had sex (see Table 2). Next, respondents were asked about condom and contraception use the last time they had sex. I operationalize this by creating binary variables to measure condom and other contraception use at last sex. Thus, both condom and other contraception use are conditional on sexual activity. Other contraception use captures whether a respondent reported using any kind of FDA-approved contraception method, excluding condoms. This includes birth control pills, an IUD or shot, patch, or birth control ring. According to the sample, 52 percent of students indicated using a condom at last sex, while 18 percent used an alternative FDA- approved contraception method, and 30 percent used no medically approved method at last sex. 6The augmentation of national with state YRBS data has been employed in a number of recent studies examining the eﬀects of many state-level public policies, including cigarette taxes (Hansen et al., 2013), medical marijuana laws (Anderson et al., 2015), anti-bullying laws (Sabia & Bass, 2016), and parental involvement laws for abortion (Sabia & Anderson, 2014), on risky behaviors.

8 3.3 Health Outcomes

Though the YRBS is rich with individual level sexual behavior data, the surveys fail to ask students questions regarding sexually transmitted diseases, pregnancy, and the outcome of a pregnancy if one occurred. Additionally, each of the above measures of sexual behavior is self-reported. If sex education mandates prompt more students to be willing to report their sexual behavior, then estimated effects of sex education mandates may be biased upward. Furthermore, STDs, pregnancy, and pregnancy outcomes are likely relatively low frequency events in the YRBS, so measures of incidence might be inaccurate. To supplement the self- reported measures of student sexual behaviors, I augment my YRBS analysis with objective STD and birth rate data. According to the CDC’s Division of STD Prevention, young people ages 15 to 24 are at a higher risk of acquiring an STD, especially chlamydia. Since many chlamydia infections go unnoticed and undiagnosed, the result of such a disease can be severe, especially for a woman’s reproductive health (Cornblatt, 2009). State-level chlamydia rates for 15-19 year olds, per 1,000, for 1996-2014 were obtained from the CDC’s STD Surveillance Data.7 These data are derived from information from the official statistics for the reported8 occurrence of nationally notifiable STDs in the United States, test positivity and prevalence data from numerous prevalence monitoring initiatives, sentinel surveillance, and national health care services surveys (CDC, 2015). Additionally, data on pregnancies and the outcome of such pregnancies is not available in the YRBS. Thus, I obtain state-level birth rates for 15 to 19 year old females for the years 1991-2013 from the National Vital Statistics. In the United States, state laws require birth certificates to be completed for all births, and Federal law mandates national collection and publication of births and other vital statistics data (CDC, 2015).

3.4 Sex Education Mandates

I begin by generating a binary sex education variable that measures whether a state had enacted and was enforcing a sex education mandate. Information about sex education eﬀective dates was obtained from multiple sources. Policy information from 1991 to 2000 for each state was retrieved from numerous volumes of the SIECUS (Sexuality Information and Education Council of the United States) Report. The SIECUS Report was published from 1972 to 2005, and includes scholarly articles, opinion pieces, policy information, and

7Prior to 1996, STD rates were not available by age brackets. 8The CDC’s STD Surveillance Data only accounts for reported STDs. If sex education induces changes in STD testing rates, which in turn accounts for changes in STD rates, then the estimated results will be biased (Carr & Packham, 2016).

9 other works regarding sexuality information and education. Sex education mandates from 2001 to 2013 were collected from the Guttmacher Institute State Policies in Brief: Sex and HIV Education publications, which contain detailed sex education policy information at the state level. Throughout the sample period, 20 states mandated sex education (see Table 1). Given the substantial heterogeneity in the type of sex education mandate enacted by each state, SIECUS Reports and the Guttmacher Institute (GI) categorize these mandates by their comprehensiveness and requirements. Given the debate about what type of sex education should be oﬀered in schools among policymakers, identifying the most eﬀective type of sex education, whether it be abstinence-based or more comprehensive in nature, is critical. I use the SIECUS and GI reports from 1995-20139 to indicate whether a state mandating sex education requires abstinence-based, or comprehensive sex education. I generate binary indicators for abstinence-based and comprehensive sex education that measure whether a state already enacting a sex education mandate requires the teaching of abstinence-based or comprehensive education. Ten states that mandate sex education include abstinence-based education, while 11 states mandating sex education require a comprehensive curriculum.10

4 Empirical Approach

My econometric approach will estimate a reduced form diﬀerence-in-diﬀerence model that takes the form: 0 0 Yist = α + βSexEdst + δ Xist + γ Zst + θs + τt + ist (1) where i indexes the individual, s indexes the respondent’s state, and t indexes the survey year.

Yist is a measure of individual teenage sexual behavior, including ever had sex, condom use 11 during last sex, and contraceptive use during last sex. SexEdst is an indicator for whether a sex education mandate was in effect in state s at year t. The vector X includes individual level controls, such as age, race, gender, and grade level; the vector Z includes state level economic and policy controls, such as the unemployment rate, per capita income, beer taxes, blood alcohol content laws, zero tolerance drunk driving laws, parental involvement laws for abortion, and mandatory wait laws for abortion services. Finally, θs and τt are state and year fixed effects, and standard errors are clustered at the state level.

9Information regarding the components of a state’s sex education mandate was not available prior to 1995. 103 states (Washington DC, New Jersey, and Georgia) already had sex education mandates implemented before the 1991 period, but the type of sex education wasn’t enacted until after 1991. See Table 1. 11 Chlamydia and birth rates are obtained at the state level, thus Equation 1 can be modiﬁed as: Yst = 0 α + βSexEdst + γ Zst + θs + τt + st. This modiﬁcation can be carried over to all subsequent equations where chlamydia and birth rates are the main outcomes of interest.

10 Identification of the variable of interest, β, comes from within-state variation in sex education mandates during the 1991-2013 sample period. To produce unbiased estimates of β in the equation above, the parallel trends assumption of difference-in-difference models must be satisfied. This assumption may be violated if, for example, states enact sex education mandates in response to unfavorable teenage sexual behaviors, STD or birth rate trends, or if there are time-varying state characteristics not captured in state-specific time-varying economic and policy controls that are associated with both the adoption of sex education mandates and with the outcomes under study. To address the potential policy endogeneity as best I can, I test whether sex education mandates were implemented in response to pre-existing teenage sexual behavior and health trends with the following event study specification:

≥7 X 0 0 Yist = µ + στ I(t − t0 = τ)st + δ Xist + γ Zst + θs + τt + ist, (2) τ=≤−5

where I(t − t0 = τ) is an indicator equal to 1 if the observation is τ years away from the implementation of the sex education mandate, and zero otherwise.12 To capture heterogeneity across states’ sex education mandates, I re-estimate Equation 1 above by including the policy variable Abstinence, or Comprehensive which indicates whether a states’ sex education mandate requires the teaching of abstinence-based education or comprehensive education. Identiﬁcation of φ comes from within-state variation in states’ mandates that require abstinence-based or comprehensive sex education from 1995-2013.13

0 0 Yist = α + βSexEdst + φEducRequirementst + δ Xist + γ Zst + θs + τt + ist (3) where EducRequirement indicates either abstinence-based or comprehensive education.14 It may be the case that states begin requiring schools to teach abstinence education instead of comprehensive education, or vice versa, in response to trends in teenage STD rates, birth rates, and other risky sexual behaviors. If this is so, the estimated eﬀect of the education

12Due to the substantial variation in the sex education mandates (i.e. states adopting a mandate for a few years at a time (see Table 1)), Equation 2 is estimated on a more “balanced” sample including only states that implemented and retained a sex education mandate one time throughout the entire sample period, and all control states. These states also have a sufficient number of pre and post implementation observations. Estimates for the main specification for this subset of the sample are presented in the Appendix. 13See Table 1. 14EducRequirement is an interaction term. It is identical to being multiplied by SexEd. There is no main effect since Comprehensive education, for example, is only available if a sex education mandate was already in place.

11 requirements on teenage sexual behaviors will be biased. Therefore, the type of sex education provided should be uncorrelated with pre-treatment trends in the outcomes. This assumption can be tested using a similar event study speciﬁcation as discussed above.

5 Results

5.1 YRBS Survey Outcomes

Table 3 presents the results from Equation 1 for the effect of any state sex education mandate on measures of teenage sexual behavior. Difference-in-difference estimates indicate that the typical state sex education mandate increases the probability of condom use by about 3 percent (0.015/0.525). Sex education mandates appear to have no effect on FDA-approved contraception methods, suggesting that teenagers’ preferred method of contraception is condoms. Additionally, sex education mandates have no statistically significant effect on the probability of engaging in sexual intercourse. Based on the 95% confidence interval, I can rule out the probability of sexual activity decreasing by more than 0.9 percent and increasing by more than 2.1 percent. Although teens are not engaging in more sexual activity, they are increasing their use of condoms. A causal interpretation of the effect of state mandated sex education on teenage sexual behaviors can be supported by the event study figures presented in Figure 1. Panels (a) and (b) show the estimates of α from Equation 2 on teenage sexual activity and condom use.15 The points in the plots give the estimate of α, with year -1 excluded such that all estimates are relative to this year, while the lines extending from them represent 95% confidence intervals that are calculated using standard errors clustered at the state level. If the adoption of sex education mandates has an effect on teenage sexual behaviors before their implementation, then the results may be driven by unobserved trends (Meer & West, 2015). This is not the case, and there exists no evidence of pre-treatment trends in the sexual activity or condom use outcome measures.16 These figures suggest that sex education implementation is exogenous to teenage sexual behaviors.17 Table 4 presents the results for the effect of sex education by individual age to test which

15The event study figure for contraception use is presented in Appendix Figure A-1, panel (a). 16Given the absence of pre-treatment trends, state-specific linear time trends are not included as a control in any of the specifications since such an inclusion may attenuate the results (Meer & West, 2015). 17Recall the event study figures are estimated on a more “balanced” panel, as described in footnote 12 of the text. Estimates from equation 1 on the more balanced panel are presented in Appendix Table 2. The results are quantitatively similar for the YRBS outcomes.

12 teenagers, younger or older, respond more to school-based sex education. The estimates indicate that older teens, those 15 to 18, respond most to sex education. Specifically, sex education increases the probability that 15 year olds engage in sexual activity by 1.5 percentage points. The positive effect of sex education on sexual activity is only present for 15 year olds, suggesting that younger teens, who may be most affected by the mandate, begin engaging in sexual behaviors as a result of new information about sex being obtained. Addi- tionally, condom use marginally increases for sexually active 14 year olds by 2.3 percentage points. However, according to a national sex study done by researchers at Indiana Univer- sity, younger teens engage in oral sex significantly more than vaginal intercourse, and the average age of first intercourse is 17 (Herbenick et al., 2010). Thus, it may be the case that the significant increase in sexual activity by 15 year olds and condom use by 14 year olds is due to an increase in oral sex, and protected oral sex, rather than sexual penetration.18 Condom use also significantly increases for 16 and 17 year olds, who are more likely to be sexually active, by 4.4 percent and 3.6 percent, respectively.19 It may be the case that length of exposure to school-based sex education affects teenage sexual behaviors. Those who have been exposed to sexual health information for longer periods of time may be affected more than those most recently exposed. This effect may only exist if the education being provided is new, and teens are updating their sexual health information. Since younger teens could have only been in high school at most one or two years and may have never been exposed to a sex education law, or only been exposed for a maximum of one or two years, this analysis is focused on teens ages 16 to 18. The results for length of exposure to sex education are presented in Table 5.20 Older teens exposed to sex education for more than 1 year significantly increase their condom use. The safer sexual behavior effects for older teens may take time to unfold due to teens potentially gaining new information, updating their knowledge, and eventually changing their behavior.

5.2 Health Outcomes

Given the significant effect of a sex education mandate on teenage condom use at last intercourse, it is useful to explore whether this effect translates to teenage STD and birth

18The YRBS does not distinguish between oral sex or sexual intercourse when asking respondents if they have ever had sex. Therefore, I am unable to test this prediction that younger teens are engaging in oral sex rather than sexual penetration. 19The analysis was also performed by gender and race. There was no statistical diﬀerence between males and females, or whites, blacks, or other race students, for the outcomes under study. 20This analysis augments equation (1) by including a separate dummy variable equal to 1 if the student was exposed to a sex education law for 1 year, 2 years, or 3+ years, and 0 otherwise.

13 rates. Table 6 presents the results for any sex education mandate on chlamydia and birth rates, per 1,000, for teenagers ages 15-19 for 1996-2013 and 1991-2013, respectively. The estimates suggest that school-based sex education mandates decrease chlamydia rates by 7.9 percent. There is no significant effect of sex education on the teenage birth rate. Since sexual activity is not affected, and condom use increases, it is not surprising that there is no effect on the teenage birth rate. A causal interpretation of the teenage chlamydia and birth rate result is supported by the event study figures in Figure 1, panel (c), and Appendix Figure A-1, panel (b), respectively. There exists no evidence of pre-treatment trends for teenage chlamydia rates. After implementation of the sex education mandate, chlamydia rates decline substantially and persist for up to six years.21 Additionally, a causal interpretation of the decline in teenage chlamydia rates can be supported by a falsification test on chlamydia rates for older young adults ages 20-24 for the years 1996-2013.22 The result is presented in Table 7. There is no evidence that school-based sex education mandates effect chlamydia rates for young adults for whom sex education mandates do not bind. The significant decrease in chlamydia rates is substantial, given that the estimated direct medical costs for treating young people with sexually transmitted diseases is $16 billion annually, excluding costs associated with HIV/AIDS (CDC, 2013). Furthermore, among the non-viral STDs, chlamydia is the most common and costly infection, estimated at almost $517 million in annual health care costs (Owusu-Edusei Jr et al., 2013). A back-of-the- envelope calculation estimates that the annual savings associated with a 7.9 percent decrease in chlamydia rates is roughly $41 million.23

5.3 Exploring Heterogeneity

While the average school-based state sex education mandate appears to have a significant positive effect on teenage condom use and chlamydia rates, I next explore whether there may be heterogeneity in the effect of sex education by type of law implemented. Tables 8 and 9 present the results from Equation 3 for abstinence-based and comprehensive sex education on teenage sexual behaviors, and teen STD and birth rates, respectively. The results imply that there is no significant differential effect of abstinence-based or comprehensive sex ed-

21Recall the event study ﬁgures are estimated on a more “balanced” panel, as described in footnote 9 of the text. Estimates from equation 1 on the more balanced panel are presented in Appendix Table 2. The results are qualitatively similar for the sexual health outcomes. 22The falsiﬁcation test is only performed on chlamydia rates for older teens since data on sexual behavior outcomes, like those in the YRBS, are not available for older teens. 23This calculation is based on the 395,612 reported cases of chlamydia in 2013 among 15-19 year olds (CDC, 2013) and the $517 million in annual chlamydia health care costs (Owusu-Edusei Jr et al., 2013).

14 ucation to the typical school-based sex education mandate.24 Appendix Table 1 highlights the different components of each type of sex education curriculum. A general sex education curriculum includes information on abstinence, the risks associated with sexual activities, and information, whether medically accurate or not, about contraception methods, including condoms. The common elements of the type of sex education may be enough to alter teen’s behaviors, and the results may suggest that the type of school-based sex education does not differentially affect teenage sexual behaviors, but simply being exposed to any information regarding sexual health alters teens’ behaviors. The null effect of sex education components is surprising, especially from a policymakers’ standpoint given the debate about what type of school-based sex education should be taught. As discussed in the background section, abstinence-based and comprehensive sex education should change the behavior of teens who wish to engage in sexual activity, so long as it is new information. However, this result could suggest that the information they are receiving in abstinence-based and comprehensive sex education is not new and they are not changing their sexual behavior, or the type of specific sex education is not effective in changing their behavior. For example, comprehensive sex education programs that distribute condoms to the students or demonstrate the proper use of contraception may be more effective than a program that simply talks about how to use a condom. A shortcoming of my analysis presented here is that I am unable to identify the intensity (e.g. distributing condoms to the students versus talking about their proper use) of the sex education mandate enforced in each state.

6 Conclusion

Given the policy debate over school-based sex education, credible estimates of the causal link between sex education and teenage sexual behaviors are needed. Despite the exhaustive literature examining the effect of sex education on adolescent sexual behaviors, the effect of state-level sex education mandates on sexual behavior has not been studied in a way that allows for a causal interpretation of the results. The current study presents the first examination of the relationship between school-based sex education and teenage sexual behavior using within-state variation in sex education mandates from 1991-2013. Difference-in-difference results suggest that the typical state sex education mandate increases the probability of condom use by 3 percent and decreases the chlamydia rate among teens ages 15 to 19 by 8

24Event study ﬁgures for comprehensive and abstinence-based sex education are presented in Appendix Figures A-2 and A-3, respectively.

15 percent. A causal interpretation of the results is supported by the event study exercise that indicates no pre-trends were present before the implementation of a sex education mandate, and the falsification test on chlamydia rates of older young adults for whom school-based sex education mandates do not bind. When heterogenous types of sex education are analyzed, I find that abstinence-based and comprehensive sex education have no differential effect on teenage sexual behaviors, a noteworthy result that contrasts the previous literature. There are some limitations of this study worthy of note. Although this study contributes to estimating the effect of school-based sex education on teenage sexual behaviors and health using a more credible research design than previous studies, the analysis could benefit from a more comprehensive examination of the sex education laws. Specifically, school-based sex education is mandated at the state-level, but the curriculum is largely left up to the local school districts. Obtaining information about the specific components of the state sex education mandate at the school district level would provide a clearer understanding of what type of information is driving the results presented in this study. Despite these limitations, the results here suggest that there are clear benefits to sex education. Are these benefits worth the cost? The common cost for a school-based sex education curriculum is roughly $200, which typically only includes the cost of teaching materials (NARAL, 2009).25 In 2012, there were approximately 27,000 US public secondary schools (NCES, 2012). If each one of these schools implemented a sex education curriculum, it would cost around $5.4 million annually to educate teens about sex. Given the $41 million in annual STD savings, clearly the benefits of implementing sex education outweigh the costs.

25Most sex education teachers are health or physical education teachers already employed by the school (FoSE, 2014). Additionally, the total average time dedicated to teaching sex education in high school is about 4 hours, and is commonly taught during the health or physical education class period (CDC, 2014).

16 References

111th Congress. (2009). Consolidated appropriations act, 2010 (Tech. Rep.). United States Government.

Alford, S. (2001). Sex education programs: Deﬁnitions and point-by-point comparison (Tech. Rep.). Advocates for Youth.

Anderson, D. M., Hansen, B., & Rees, D. I. (2015). Medical marijuana laws and teen marijuana use. American Law and Economics Review, 17 (2), 495–528.

Angrist, J. D., & Evans, W. N. (1999). Schooling and labor market consequences of the 1970 state abortion reforms. Research in labor economics, 18 , 75–113.

Barr, E. M., Goldfarb, E. S., Russell, S., Seabert, D., Wallen, M., & Wilson, K. L. (2014). Improving sexuality education: the development of teacher-preparation standards. Journal of school health, 84 (6), 396–415.

Bennett, S. E., & Asseﬁ, N. P. (2005). School-based teenage pregnancy prevention programs: a systematic review of randomized controlled trials. Journal of Adolescent Health, 36 (1), 72–81.

Bridges, E., & Hauser, D. (2014, May). Youth health and rights in sex education. Retrieved August 25, 2016, from http://futureofsexed.org/youthhealthrights.html

Bronars, S. G., & Grogger, J. (1994). The economic consequences of unwed motherhood: Using twin births as a natural experiment. The American Economic Review, 1141–1156.

Carr, J. B., & Packham, A. (2016). The eﬀects of state-mandated abstinence-based sex education on teen health outcomes. Health economics.

Carter, J. B. (2001). Birds, bees, and venereal disease: Toward an intellectual history of sex education. Journal of the History of Sexuality, 10 (2), 213–249.

CDC. (2013, February). Incidence, prevalence, and cost of sexually transmitted infections in the united states. Online.

CDC. (2014). Results from the school health policies and practices study 2014 (Tech. Rep.). US Department of Health and Human Serviced.

CDC. (2015, March). Interpreting std surveillance data. Retrieved October 26, 2016, from http://www.cdc.gov/std/stats15/appendixa.htm

Cornblatt, J. (2009). A brief history of sex ed in america (Tech. Rep.). Newsroom.

DiCenso, A., Guyatt, G., Willan, A., & Griﬃth, L. (2002). Interventions to reduce unintended pregnancies among adolescents: systematic review of randomised controlled trials. Bmj , 324 (7351), 1426.

17 Dupas, P. (2011). Do teenagers respond to hiv risk information? evidence from a field experiment in kenya. American Economic Journal: Applied Economics, 3 (1), 1–34. Eisen, M., & Zellman, G. L. (1986). The role of health belief attitudes, sex education, and demographics in predicting adolescents’ sexuality knowledge. Health Education & Behavior, 13 (1), 9–22. FoSE. (2014). Future of sex education - current barriers to sex education implementation. Retrieved June 5, 2017, from http://www.futureofsexed.org/currentbarriers.html Haffner, D. W. (1997). Whats wrong with abstinence-only sexuality education programs? SIECUS report, 25 (4), 9–13. Hansen, B., Rees, D. I., & Sabia, J. J. (2013). Cigarette taxes and how youth obtain cigarettes. National Tax Journal, 66 (2), 371–394. Herbenick, D., Reece, M., Schick, V., Sanders, S. A., Dodge, B., & Fortenberry, J. D. (2010). Sexual behavior in the united states: results from a national probability sample of men and women ages 14–94. The journal of sexual medicine, 7 (s5), 255–265. Hoffman, S. D., & Maynard, R. A. (2008). Kids having kids: Economic costs & social consequences of teen pregnancy. The Urban Insitute. Hotz, V. J., McElroy, S. W., & Sanders, S. G. (1997). The impacts of teenage childbearing on the mothers and the consequences of those impacts for government. Kids having kids: Economic costs and social consequences of teen pregnancy, 55–94. Irvine, J. M. (2004). Talk about sex: The battles over sex education in the united states. Univ of California Press. Kearney, M. S., & Levine, P. B. (2015). Investigating recent trends in the us teen birth rate. Journal of health economics, 41 , 15–29. Kim, N., Stanton, B., Li, X., Dickersin, K., & Galbraith, J. (1997). Effectiveness of the 40 adolescent aids-risk reduction interventions: a quantitative review. Journal of Adolescent Health, 20 (3), 204–215. Kirby, D. B. (2008). The impact of abstinence and comprehensive sex and std/hiv education programs on adolescent sexual behavior. Sexuality Research & Social Policy, 5 (3), 18–27. Kohler, P. K., Manhart, L. E., & Lafferty, W. E. (2008). Abstinence-only and comprehensive sex education and the initiation of sexual activity and teen pregnancy. Journal of Adolescent Health, 42 (4), 344–351. Meer, J., & West, J. (2015). Effects of the minimum wage on employment dynamics. Journal of Human Resources. Mullinax, M., Mathur, S., & Santelli, J. (2017). Adolescent sexual health and sexuality education. In International handbook on adolescent health and development (pp. 143–167). Springer.

18 NARAL. (2009). The healthy youth act - comprehensive sex education for wisconsin’s youth. Retrieved October 27, 2016, from http://www.prochoicewisconsin.org/what -is-choice/sexuality-education/healthy-youth-act.shtml

NCES. (2012). Number of educational institutions, by level and control of institution: Selected years, 1980-81 through 2011-12 (Tech. Rep.). Institute of Education Sciences: National Center for Education Statistics.

Oettinger, G. S. (1999). The eﬀects of sex education on teen sexual activity and teen pregnancy. Journal of Political Economy, 107 (3), 606–644.

Owusu-Edusei Jr, K., Chesson, H. W., Gift, T. L., Tao, G., Mahajan, R., Ocfemia, M. C. B., & Kent, C. K. (2013). The estimated direct medical cost of selected sexually transmitted infections in the united states, 2008. Sexually transmitted diseases, 40 (3), 197–201.

Reichelt, P. A., & Werley, H. H. (1975). Contraception, abortion and venereal disease: Teenagers’ knowledge and the eﬀect of education. Family Planning Perspectives, 83–88.

Sabia, J. J. (2006). Does sex education aﬀect adolescent sexual behaviors and health? Journal of Policy Analysis and Management, 25 (4), 783–802.

Sabia, J. J., & Anderson, D. M. (2014). Parental involvement laws, birth control, and mental health: New evidence from the yrbs (Tech. Rep.). Working Paper, San Diego State University.

Sabia, J. J., & Bass, B. (2016). Do anti-bullying laws work? new evidence on school safety and youth violence. Journal of Population Economics, 1–30.

Sanderson, C. A. (2000). The eﬀectiveness of a sexuality education newsletter in inﬂuencing teenagers’ knowledge and attitudes about sexual involvement and drug use. Journal of Adolescent Research, 15 (6), 674–681.

Scales, P. (1981). The new opposition to sex education. Journal of School Health, 51 (4), 300–304.

SIECUS. (2010). A brief history of federal funding for sex education and related programs (Tech. Rep.). Sexuality Information and Education Council of the United States.

Stanger-Hall, K. F., & Hall, D. W. (2011). Abstinence-only education and teen pregnancy rates: why we need comprehensive sex education in the us. PLoS One, 6 (10), e24658.

Starkman, N., & Rajani, N. (2002). The case for comprehensive sex education. AIDS patient care and STDs, 16 (7), 313–318.

Trenholm, C., Devaney, B., Fortson, K., Quay, K., Wheeler, J., & Clark, M. (2007). Impacts of four title v, section 510 abstinence education programs. ﬁnal report. Mathematica Policy Research, Inc..

19 Weinstock, H., Berman, S., & Cates, W. (2004). Sexually transmitted diseases among american youth: incidence and prevalence estimates, 2000. Perspectives on sexual and reproductive health, 36 (1), 6–10.

Workowski, K. A., & Bolan, G. A. (2015). Sexually transmitted diseases treatment guidelines (2015). Reproductive Endocrinology(24), 51–56.

20 Figure 1: Event Study Estimates of the Eﬀect of Sex Education Mandates on Teenage Sexual Behaviors and Health

(a)

(b) (c)

Notes: Estimates of Equation 2 described in the text on the “balanced panel”. All estimates include individual controls (when applicable), state-level controls, and state and year fixed effects. 95 percent confidence intervals are shown extending from each point. All estimates are relative to year -1.

21 Table 1: State-level Sex Education Mandates, 1991-2013 Sex Education Abstinence-Based Comprehensive Alabama 1993-2000 Florida 2001-2010 Washington DC 2004-2013 Alaska 2001-2006 Georgia 2004-2013 Hawaii 1997-2010 Arizona 1994-1996 Kentucky 1999-2013 Illinois 2004-2005 Florida 1992-1996, 2001-2010 Minnesota 2011-2013 Maine 2003-2013 Hawaii 1995-2010 Montana 2007-2013 Maryland 2001-2013 Illinois 1994-2005 New Jersey 2004-2006 New Jersey 1997-2003, 2011-2013 Kentucky 1999-2013 North Carolina 2004-2009 New Mexico 2009-2013 Maine 2001-2013 North Dakota 2012-2013 North Carolina 1995-2003 Minnesota 1995-2013 Ohio 2011-2013 Oregon 2008-2013 Montana 2007-2013 Tennessee 2001-2013 Tennessee 1995-2000 Mississippi 2012-2013 West Virginia 1997-2006 New Mexico 2009-2013 North Carolina 1995-2013 North Dakota 2012-2013 Ohio 2011-2013 Oregon 2008-2013 Tennessee 1995-2013 Texas 1994-1996 West Virginia 1995-2013 Wyoming 2001-2013 Notes: The states contributing to the identifying variation are listed above. States that passed sex education mandates during the 1991-2013 period, and states that implemented abstinence-based versus comprehensive sex education from 1995-2013. New Jersey and North Carolina switched from comprehensive to abstinence-based in 2004, and Tennessee switched from comprehensive to abstinence- based in 2001. Washington DC, New Jersey, and Georgia already had sex education mandates implemented before the 1991 period, but the type of sex education wasn’t enacted until the dates listed above.

22 Table 2: Descriptive Statistics of Analysis Variables N Mean SD Min Max Demographics Grade 1,178,736 10.378 1.102 9 12 Age 1,183,001 15.99 1.237 12 18 Male 1,186,880 0.491 0.500 0 1 White 1,193,759 0.548 0.498 0 1 Black 1,193,759 0.151 0.358 0 1 Hispanic 1,193,759 0.151 0.358 0 1

Outcome Measures Sexual Activity 919,775 0.481 0.499 0 1 Condom Use 411,315 0.523 0.499 0 1 FDA Birth Control Method 411,315 0.183 0.386 0 1 Chlamydia Rate 865 15.918 7.696 2.92 66.25 Birth Rate 1,122 42.893 15.541 2.2 112.8

Treatment Measures Sex Education 1,189,637 0.468 0.499 0 1 Abstinence-Based 1,074,319 0.130 0.336 0 1 Comprehensive 1,074,319 0.278 0.448 0 1 Notes: Data on treatment measures come from the SIECUS Reports from 1991-2000, and the Guttmacher Institute State Policies in Brief: State Sex and HIV Education, 2001-2013. Data on individual-level outcome measures and demographics come from the State and National YRBS from 1991-2013. Data on state-level STD rates for 15-19 year olds come from CDC WONDER Online Database for the years 1996-2014, and data on female birth rates for 15 to 19 year olds come from the National Vital Statistics from 1991-2013. Condom use, and FDA birth control methods are all conditional on having had sex.

Table 3: The Eﬀect of Sex Education on Teenage Sexual Behaviors Had Sex Condom Use FDA Method SexEd 0.006 0.015*** 0.001 (0.008) (0.006) (0.004)

Observations 854,596 384,010 384,010 Outcome Means 0.484 0.525 0.181 State-level controls Yes Yes Yes Individual controls Yes Yes Yes State FE Yes Yes Yes Year FE Yes Yes Yes Notes: Linear probability model estimates are obtained using data from the 1991-2013 YRBS. Individual controls include age, grade, gender, and race. State-level controls include the unemployment rate, real income per capita, state- level beer taxes, blood alcohol content laws, zero tolerance drunk driving laws, parental involvement laws for abortion, and mandatory wait laws for abortion services. Standard errors clustered at the state level are in parentheses. *** p<0.01, ** p<0.05, * p<0.1

23 Table 4: The Eﬀect of Sex Education on Teenage Sexual Behaviors, by Age Age 14 Age 15 Age 16 Age 17 Age 18 Panel I: Had Sex SexEd 0.009 0.015** 0.005 0.005 -0.006 (0.007) (0.060) (0.011) (0.010) (0.011)

Observations 90,103 216,629 227,366 202,993 115,027 Outcome Means 0.250 0.359 0.489 0.601 0.684 Panel II: Condom Use SexEd 0.023* 0.008 0.024*** 0.018** 0.002 (0.012) (0.010) (0.009) (0.007) (0.010)

Observations 20,985 72,065 103,142 113,732 73,188 Outcome Means 0.602 0.587 0.545 0.499 0.457 Panel III: Other Contraception Use SexEd 0.002 0.001 -0.004 0.002 0.006 (0.008) (0.006) (0.006) (0.006) (0.007)

Observations 20,985 72,065 103,142 113,732 73,188 Outcome Means 0.085 0.112 0.163 0.218 0.246 Notes: Linear probability model estimates from equation 1 are obtained using data from the 1991-2013 YRBS. Regressions include individual-level and state-level controls, and state and year ﬁxed eﬀects. Standard errors clustered at the state level are in parentheses. *** p<0.01, ** p<0.05, * p<0.1

24 Table 5: Years Exposed to a Sex Education Mandate for 16 to 18 year olds Had Sex Condom Use FDA Method 1 year exposed 0.026 -0.003 -0.0001 (0.018) (0.011) (0.009) 2 years exposed -0.018 0.014** -0.004 (0.011) (0.007) (0.005) 3+ years exposed -0.002 0.012* -0.0002 (0.012) (0.007) (0.005)

Observations 570,763 302,390 302,390 Notes: Linear probability model estimates are obtained using data from the 1991-2013 YRBS. Regressions include individual-level and state-level controls, and state and year ﬁxed eﬀects. Standard errors clustered at the state level are in parentheses. *** p<0.01, ** p<0.05, * p<0.1

Table 6: The Eﬀect of Sex Education on Teenage STD and Birth Rates Chlamydia Birth Rate SexEd -1.243** 0.023 (0.577) (0.015)

Observations 865 1,122 Outcome Means 15.640 3.730 State-level controls Yes Yes State FE Yes Yes Year FE Yes Yes Notes: Linear probability model estimates are obtained using data on STD rates for 15-19 year olds from CDC WONDER Online Database for the years 1996-2013. Birth rates for females ages 15 to 19 were obtained from the National Vital Statistics for the years 1991-2013. Standard errors clustered at the state level are in parentheses. *** p<0.01, ** p<0.05, * p<0.1

Table 7: Falsiﬁcation Tests on Older Young Adult Chlamydia Rates Chlamydia SexEd -0.361 (0.632)

Observations 865 State-level controls Yes State FE Yes Year FE Yes Notes: Linear probability model estimates are obtained using data on STD rates for 20-24 year olds from CDC WONDER Online Database for the years 1996-2013. Standard errors clustered at the state level are in parentheses. *** p<0.01, ** p<0.05, * p<0.1

25 Table 8: The Eﬀect of Sex Education Components on Teenage Sexual Behaviors Had Sex Condom Use FDA Method Panel I: Abstinence-Based Sex Education SexEd 0.007 0.020*** -0.004 (0.008) (0.007) (0.005) Abstinence 0.002 -0.007 0.008 (0.010) (0.011) (0.006)

Observations 777,490 342,454 342,454 Panel II: Comprehensive Sex Education SexEd 0.011 0.016** 0.0001 (0.008) (0.008) (0.005) Comprehensive -0.013 0.006 -0.001 (0.011) (0.008) (0.006)

Observations 777,490 342,454 342,454 Outcome Means 0.475 0.531 0.184 Notes: Linear probability model estimates are obtained using data from the 1995-2013 YRBS. Regressions include individual-level and state-level controls, and state and year ﬁxed eﬀects. Standard errors clustered at the state level are in parentheses. *** p<0.01, ** p<0.05, * p<0.1

26 Table 9: The Eﬀect of Sex Education Components on Teenage STD and Birth Rates Chlamydia Birth Rate Panel I: Abstinence-Based Sex Education SexEd -1.169* 0.013 (0.676) (0.011) Abstinence -0.177 0.018 (0.531) (0.014)

Observations 865 918 Panel II: Comprehensive Sex Education SexEd -1.677** 0.023 (0.670) (0.014) Comprehensive 1.514 -0.010 (0.916) (0.019)

Observations 865 918 Outcome Means 15.640 3.670 Notes: Linear probability model estimates are obtained using data on STD and birth rates for 15 to 19 year olds. All regressions are for the years 1995-2013, since information on Abstinence-based versus Comprehensive sex education was only available for those years. All regressions include state-level controls, and state and year ﬁxed eﬀects. Standard errors clustered at the state level are in parentheses. *** p<0.01, ** p<0.05, * p<0.1

27 7 Appendix

Figure A-1: Event Study Estimates of the Eﬀect of Sex Education Mandates on Teenage Contraception Use and Birth Rates

(a)

(b)

28 Figure A-2: Event Study Estimates of the Eﬀect of Comprehensive Sex Education Mandates on Teenage Sexual Behaviors

(a) (b)

(e)

29 Figure A-3: Event Study Estimates of the Eﬀect of Abstinence-Based Sex Education Man- dates on Teenage Sexual Behaviors

(a) (b)

(e)

30 Table A-1: Types of Sex Education Curriculum, Descriptions General Sex Education Abstinence-Based Comprehensive • Abstinence

• Risks associated with • Only 100% eﬀective • Adolescent development sex way to avoid pregnancy/STDs – Puberty – Pregnancy • Abstain from sex until • Abstinence – STDs after marriage • Contraception methods • Information about con- • Consequences of sex be- traception methods – Medically accurate fore marriage information – Pregnancy – Proper use and ef- – STDs fectiveness

• Failure rates of contra- • Conception and Preg- ception methods nancy

• Adoption as an option – Options for unin- for unintended pregnan- tended pregnancy cies • STDs

– Symptoms and treatments – How to access local testing and treatment services

Table A-2: The Eﬀect of Sex Education on Teenage Sexual Behaviors, “Balanced” Panel Had Sex Condom Use FDA Chlamydia Birth Rate Method Rate SexEd 0.009 0.022*** -0.006 -0.879 0.015 (0.010) (0.008) (0.005) (0.787) (0.021)

Observations 586,556 249,194 249,194 561 726 Outcome Means 0.466 0.522 0.191 14.176 3.704 State-level controls Yes Yes Yes Yes Yes Individual controls Yes Yes Yes Yes Yes State FE Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Notes: Linear probability model estimates are obtained using data from the 1991-2013 YRBS, CDC WONDER STD statistics, and National Vital Statistics. Standard errors clustered at the state level are in parentheses. *** p<0.01, ** p<0.05, * p<0.1

31 Table A-3: The Eﬀect of Sex Education Type on Teenage Sexual Behaviors, “Balanced” Panel Had Sex Condom Use FDA Method Panel I: Abstinence-Based Sex Education SexEd 0.006 0.018*** -0.005 (0.008) (0.006) (0.006) Abstinence 0.004 0.010 0.002 (0.010) (0.009) (0.010)

Observations 702,557 311,217 311,217 Outcome Means 0.474 0.527 0.188 Panel II: Comprehensive Sex Education SexEd 0.018** 0.018** -0.0003 (0.008) (0.009) (0.006) Comprehensive -0.030 0.010 -0.005 (0.023) (0.010) (0.014)

Observations 678,778 300,085 300,085 Outcome Means 0.473 0.533 0.185 Notes: Linear probability model estimates are obtained using data from the 1995-2013 YRBS. All regressions are for the years 1995-2013, since information on Abstinence-based versus Comprehensive sex education was only available for those years. Regressions include individual-level and state-level controls, and state and year ﬁxed eﬀects. Standard errors clustered at the state level are in parentheses. *** p<0.01, ** p<0.05, * p<0.1

Table A-4: The Eﬀect of Sex Education Type on Teenage STD and Birth Rates, “Balanced Panel” Chlamydia Birth Rate Panel I: Abstinence-Based Sex Education SexEd -1.233* 0.011 (0.726) (0.012) Abstinence -0.455 0.027 (0.731) (0.017)

Observations 814 864 Outcome Means 15.679 3.672 Panel II: Comprehensive Sex Education SexEd -2.085** 0.029* (0.841) (0.016) Comprehensive 2.678 -0.034 (1.415) (0.027)

Observations 763 810 Outcome Means 15.700 3.662 Notes: Linear probability model estimates are obtained using data on STD and birth rates for 15 to 19 year olds. All regressions are for the years 1995-2013, since information on Abstinence-based versus Comprehensive sex education was only available for those years. All regressions include state-level controls, and state and year ﬁxed eﬀects. Standard errors clustered at the state level are in parentheses. *** p<0.01, ** p<0.05, * p<0.1