For more than 30 years, sex education research has struggled without success to produce compelling evidence that school-based comprehensive sex education (CSE) is an effective strategy for achieving its original purposes: reducing teen pregnancy, STDs, and sexual risk behavior. Yet, despite repeated claims that CSE programs have been “proven effective” at accomplishing such goals,1 when a credible scientific lens is used to examine studies designed to test the causal impact of CSE programs, the evidence disappears.
For example, a recently published systematic review of the research most often cited by CSE advocates found that only six out of 103 studies of school-based CSE worldwide showed a sustained protective effect (lasting one year post program) on either teen pregnancy, STDs, condom use, or abstinence, for the intended youth population, without also producing other negative effects on sexual risk behavior. Moreover, nearly three times as many studies (16) found harmful CSE effects (increases in teen sexual risk behavior).2 Two other landmark meta-analyses also found a dearth of positive CSE results.3 One, sponsored by the U.S. Centers for Disease Control and Prevention (CDC), found that school-based CSE programs did not significantly increase teen condom use or reduce teen pregnancy or STDs, key outcomes for which CSE was originally designed. The other, a meta-analysis of U.S. federally funded sex education programs (most were CSE), found no statistically significant results for any outcomes.
Perhaps being aware of CSE’s poor record at achieving these original goals, Goldfarb and Lieberman, the authors of a 2021 review of this body of research, looked for CSE success with other types of outcomes. The researchers conducted what they termed a “review of [the past] three decades of research on school-based programs to find evidence for the effectiveness of comprehensive sex education … [at producing outcomes] beyond pregnancy and sexually transmitted disease prevention,” including outcomes related to gender norms, gender identity, sexual orientation, and social justice. They detailed their findings in their published report, “Three Decades of Research: The Case for Comprehensive Sex Education.”4
Goldfarb and Lieberman state that their study found, “school-based CSE can lower homophobia and homophobic-related bullying, can increase understanding of gender and gender norms, can improve knowledge and skills that support healthy relationships, can build child sex abuse prevention skills, and can reduce dating and intimate partner violence.”5 They further assert that they have found “substantial evidence that sexuality education is most effective when begun early.”6 The authors give approximately 88 citations of purported evidence to support these claims for CSE effectiveness. Yet there are two major problems with the study that undermine its claims and call into question the overall usefulness of the analysis. However, before discussing these two problems, it is important to understand what is typically meant by the term “comprehensive sex education.”
What Is Comprehensive Sex Education?
Goldfarb and Lieberman make wide-ranging claims about “comprehensive sex education” (a.k.a., comprehensive sexuality education), without being very clear about what it is and what it is not. Of course, this is not a term that these authors coined nor that can be re-defined as they see fit. This term has been in use for over 30 years, and a common meaning has developed over that time, aided by definitional statements from organizations in this field. One of them is the (so-called) National Sexuality Education Standards (a set of recommendations that have not been endorsed by any federal agency but rather were self-labeled as “national standards” by their authors).7 These “Standards” are quoted by the study authors as stipulating that the “essential, minimum, core content … for sex education” should include instruction on “Anatomy and Physiology, Puberty & Adolescent Sexual Development … [and] Sexual Health” (italics added).8 In addition, the United Nations has defined CSE as “a curriculum-based process of teaching and learning about … sexual and reproductive anatomy and physiology; puberty and menstruation; reproduction, modern contraception, pregnancy and childbirth; and STIs, including HIV and AIDS.”9
Combining these two definitions, a CSE intervention will have at least two features; it will: a) be an articulated curriculum that can be taught to students, and b) contain content on sexual and reproductive health issues, including instruction on contraception and STI/HIV/AIDS prevention. These characteristics are consistent with the common usage of the term “comprehensive sex education” over the 30-year time period covered in the present study. Programs manifesting these characteristics are typically referred to as CSE while those that do not, are not. However, the Goldfarb and Lieberman study appears to consider any school-based activity addressing issues of student well-being as fitting into the same category as CSE programs. This is not scientifically appropriate; it is inaccurate and misleading to equate child sex abuse prevention or anti-bullying programs with CSE or to ascribe positive impacts from any type of prevention program in schools to CSE if the program does not exhibit the core elements of CSE (i.e., instruction on human sexuality, contraception, and STI/HIV/AIDS prevention).
Two Major Methodological Problems
We find two major methodological problems with this study that negate its claims about CSE’s benefits.
1. Most of the Evidence Attributed to CSE is Not from CSE Studies
The first methodological problem arises when the consensus definition of CSE described above is applied to this review. Of the roughly 88 documents cited by the authors as the sources of their “evidence” for CSE, 72 of them, or four out of five (i.e., 80%, as stated earlier), are not studies of CSE programs at all. So, in a study subtitled “The Case for Comprehensive Sex Education,” only 16 of 88 evidentiary citations (fewer than one in five) are CSE studies. These 16 CSE studies are listed in Table 1. Throughout the report of study findings, strong assertions are made about the impact of CSE on many positive outcomes, but the “evidence” cited as proof of that impact is primarily not research about
comprehensive sex education, or any type of sex education.10 (See note 10 for more details.) The fact that this study attributes many positive findings from the 72 non-CSE sources (see Table 2) to the impact of CSE programs is a serious misrepresentation of the research on CSE and impugns the paper’s credibility.
2. Most of the Evidence Cited Does Not Meet Adequate Scientific Standards for Evidence of Effectiveness
The second fundamental problem with this study is that of those same 88 sources cited as the evidence upon which its claims of CSE effectiveness are based, the large majority (roughly 9 out of 10) do not meet recommended scientific standards for credible evidence of program effectiveness. That is, they do not produce reliable evidence of causal impact by a program on purported outcomes.
The scientific field of program effectiveness has recommended standards for evidence of effectiveness—scientific criteria that should be met to legitimate the assertion that a program is effective at producing claimed outcomes.11 First, the study of the program must be of adequate scientific quality. Recommended standards include: it must be designed to test cause and effect (i.e., an experimental or quasi-experimental study), have an adequate sample size, and measure long-term impacts (whether program effects endure beyond the program’s end). Second, in order to provide evidence of effectiveness, the program should produce results that have adequate effectual power. We evaluated program results according to criteria derived from the field of prevention research: positive effects should occur for the target population (not just a subgroup), they should endure well beyond the end of the program (for school-based programs, effects lasting from one school year to the next, or 12 months post program), and there should not be any countervailing negative program effects. Of the 88 sources cited as evidence by Goldfarb and Lieberman, only nine, or roughly one in 10,12 meet these scientific standards for evidence of effectiveness.13
The authors justify their lack of scientifically adequate studies by a claim that the “substantial number” of studies with inadequate research designs or inadequate sample sizes or qualitative (i.e., subjective) approaches “when taken together lead to strong evidence of outcomes” (see p. 4). This is a startling misrepresentation of scientific reality. In fact, the more error that is infused into a pool of evidence—by adding together studies of poor quality—the more, not less error will be found in the resulting evidence.
Combining these two fundamental problems—the inclusion of few actual CSE studies and the reliance on evidence of inadequate scientific quality—reveals the miniscule, shaky foundation upon which this paper and its sweeping claims stand: of the 16 citations that are studies of actual CSE programs, only three met scientific standards for studies of program effectiveness,14 and of these three, only one produced outcomes that met recommended standards for evidence of program effectiveness. This means that in a paper subtitled “The Case for Comprehensive Sex Education,” only one study provided reliable, credible evidence about CSE effectiveness (appropriate study design and sample size, with measurement of sustained effects) and produced outcomes that met recommended standards for evidence of effectiveness (positive effects for the target population, sustained at least 12 months, without other negative effects). Specifically, a program called Sexuality Education Initiative reported improvements in self-efficacy and attitudes about relationship rights. It also increased communication with partners, and the number of students using sexual health services and carrying a condom, 12 months after the program. However, it did not reduce teen sexual activity or number of sex partners, or increase condom or contraceptive use—major goals of the program and of CSE generally. Such a failure would disqualify it as an effective CSE program.15,16,17 (See notes 16 and 17 for details on the other two CSE studies.)
It should be noted that three of the 16 CSE studies cited in this paper are cross-sectional or correlational studies that produced conflicting findings on the association between LGBTQ-sensitive sex education and various outcomes (see Section 4, below).18,19,20 It is well known that correlational analyses cannot test causal impact nor thereby provide reliable evidence about program effectiveness. As authors of two of the studies put it, “The data we collected were … not designed to demonstrate causal relationships between independent and dependent variables,”21 and, “…we should interpret these correlational findings with some caution—causal mechanisms cannot be inferred…”22 The findings of these studies may raise questions meriting further experimental research, but they cannot be considered conclusive evidence of CSE impact.
Thus, only one CSE study in this review produced reliable evidence of positive CSE impact, but it improved only minor outcomes and failed to achieve the major goal—to reduce teen sexual risk behavior. Such results clearly do not make “the case for comprehensive sex education.”
Even allowing for a broader category of interventions that includes storybook reading programs for 3rd graders, musical performances, personal testimonials, etc., would add to the evidence base only eight studies meeting standards for evidence of effectiveness: five studies finding evidence of effectiveness for child sex abuse prevention programs and three for dating violence prevention programs. It would still leave eight of the ten outcomes claimed as benefits of this so-called “CSE” without credible supporting evidence.
Below we describe the actual evidence, or lack thereof, for some of the major claims in this paper.
1. No Evidence that Sex Education Should Begin in Early Elementary School
The study claims, “This review offers substantial evidence that sexuality education is most effective when begun early” (p.10),23 meaning, “beginning in the earliest grades” (see “University News” article at: https://www.montclair.edu/newscenter/2020/12/14/experts-sex-education-should-begin-in-kindergarten/). However, of the nine studies cited as evidence for this claim,24 only one was a study of sex education in the early grades (kindergarten through 3rd grade) and this study did not meet recommended standards of scientific quality for studies of effectiveness—it was a non-experimental design and did not measure effects beyond the program’s end.25 Two of the other nine studies were of child sex abuse prevention programs in the early grades that did meet scientific standards of evidence, and they did find positive program effects beyond the programs’ end.26 However, child sex abuse prevention is not sexuality education, and evidence that these programs have been beneficial for young children is not evidence that CSE or any sex education program will be. In other words, the authors produced no credible evidence supporting their claim that “substantial evidence” shows “sex education is most effective when begun early.”
2. No Evidence that CSE Can Help Prevent Child Sex Abuse
The authors claim their review found that “school-based CSE…can build child sex abuse prevention skills.” However, none of the 12 studies cited as evidence for this claim were studies of CSE.27 All 12 were studies of child sex abuse prevention programs for young children, not sex education. At least five of these met scientific standards for studies of effectiveness and these found positive effects sustained beyond the program’s end.28 However, child sex abuse prevention programs do not contain the sex education content that defines CSE, and attributing their positive results to CSE program impact is a misrepresentation of the research on CSE and on child sex abuse. Preventing child sex abuse is a worthy goal for classroom-based interventions, but giving CSE credit for building child sex abuse prevention skills, when there is no evidence for it, is factually erroneous and misleading.
3. Inadequate Evidence that CSE Reduces Dating/Intimate Partner Violence
The authors claim their review found that “school-based CSE…can reduce dating and intimate partner violence.” Yet out of the 32 studies29 cited to support this claim, only four were studies of CSE or any type of sexuality education program.30 Of the four CSE studies, only two met scientific standards for studies of effectiveness, but these programs’ outcomes did not show evidence of effectiveness. One was a study of It’s Your Game (IYG) by the program’s authors, which found it reduced dating violence after one year. However, multiple other studies have found IYG has also increased teen sexual risk behavior.31 These negative effects rule out IYG’s designation as an effective violence prevention program. The other CSE study found significantly differing results for males and females (i.e., subgroup effects). The program reduced violence for boys but not girls and increased condom use for boys while girls were reported to have reduced condom use (OR=.76), although it did not appear to be significant at the p<.05 level.32 (A third study found short-term impact but did not measure effects beyond six months, as recommended for studies testing effectiveness.33) These results do not provide adequate evidence that school-based CSE is an effective strategy for reducing dating violence.
Six of the non-CSE studies were dating violence prevention programs that met scientific standards for studies of effectiveness,34 and two found reductions in dating violence for the target population after one year—evidence of program effectiveness.35 However, to attribute the positive impact of these dating violence programs to CSE, when their goals and content are very different from that of CSE, is erroneous and misleading.
4. Inadequate Evidence that CSE Reduces Homophobic Bullying
Three of the five studies cited as evidence for this claim were CSE studies, but none met scientific standards for studies of effectiveness, that is, all three were correlational designs, rather than an experimental design that could test CSE’s causal impact. Moreover, the findings across the three studies were inconsistent. A study by Baams, et al., conducted in high schools in the Netherlands, found that neither teaching about “STI prevention” or “sexual orientation and gender” (common topics in current CSE programs) was associated with a reduction in LGBTQ+ name-calling in schools. Goldfarb and Lieberman actually misrepresent this study’s “lack of findings for the topic of sexual orientation and gender” (as stated in Baams, et al., 2017, pp. 937, 938; for males, p<.539, for females, p<.332,),36 and assert that its inclusion in curriculum reduced homophobic name-calling (see Goldfarb and Lieberman, p. 4), a claim that contradicts the data and conclusions reported in the study. Unfortunately, Baams, et al., actually misrepresent their own reported findings in the study Abstract. To be clear, the only factor they found to be associated with homophobic name-calling was the amount of sexuality education received (without respect to the specific content of the program), but this was a subgroup effect, that is, observed by females only.
Blake, et al., 2001, did not find that LGBTQ+ youth in schools with LBGTQ+ sensitive sex ed curriculum (identified by teachers’ subjective ratings) were less likely to be suicidal or feel unsafe at school than those in schools without it.37 But Proulx, et al., 2019, reported that LGBTQ+ sex ed content (again, not objectively measured but subjectively reported by school personnel) was associated with lower levels of bullying, depression, and suicidality.38
The non-experimental study design and conflicting pattern of results across these three CSE studies does not substantiate the claim that CSE reduces homophobic bullying.
5. A Pervasive Lack of Scientific Evidence for Any Claimed CSE Benefits
A similar lack of scientific evidence can be seen for all of the other positive outcomes that Goldfarb and Lieberman report to be the result of school-based CSE. To cite another example, for the claim that CSE produces “expanded understanding of gender/gender norms,” none of the nine supporting citations were studies of CSE.39
A New Sex Education Agenda
We wonder about finding these erroneous representations of research evidence published in a scholarly paper in a mainline academic journal. Having examined it thoroughly, we can only conclude that, rather than providing scientific evidence about a set of potential CSE benefits, the Goldfarb and Lieberman paper seems to provide the appearance or veneer of scientific support to a new sex education agenda, an agenda the authors articulate and endorse.
This agenda comes into focus around the claim that research shows sex education should be delivered to children as young as kindergarten age. The authors acknowledge the lack of research data on this topic: “there are only limited examples of sexuality-related content in the K-5 curriculum” (p. 10). Then they claim that this (nonexistent) “data strongly indicate that such topics are developmentally appropriate and produce positive outcomes, … that not only are younger children able to discuss sexuality-related issues but that the early grades may, in fact, be the best time to introduce topics related to sexual orientation, gender identity and expression, gender equality, and social justice related to the LGBTQ community before hetero- and cis-normative values and assumptions become more deeply ingrained and less mutable” (p.10). The authors provide no research studies that show scientific evidence for these claims, despite their efforts to give the appearance that they have done so. A couple of classrooms in which LBGTQ+ storybooks are read to 4-year-old preschoolers or 3rd graders, with positive reports by their teachers, do not constitute scientific evidence.
What the authors have done, however, is to articulate an ideological rationale for this new agenda, as follows: “Children learn gender role attitudes at an early age from observing the people in their families … it is important to introduce concepts that would disrupt stereotypical and harmful biases related to gender and sexual orientation, during this formative time” (pp. 10 & 11). Revealing Goldfarb and Lieberman’s lack of evidence for this agenda is timely because today, many proponents of this new approach want it to begin with children in kindergarten or younger, and are looking to studies like this one for evidence that supports such early inculcation of a debatable ideology.
We want to be clear. The Goldfarb and Lieberman paper does not contain any scientifically reliable evidence justifying sex education for young children, nor for teaching gender ideology or social justice theory in school classrooms.
Finally, one thing that was not made clear in their paper is the extent to which Goldfarb and Lieberman may have a vested interest in their own findings. Their final conclusion is that these “findings strengthen justification for the widespread adoption of the National Sex Education Standards.” Not disclosed is that both Goldfarb and Lieberman are “contributors and reviewers” on these same National Sex Education Standards.40 Further, they conclude that their study “provides strong support for comprehensive sex education.” Not disclosed is that Goldfarb is the author of one CSE curriculum, Our Whole Lives,41 and co-author of another, The 3Rs.42 Thus, there is potential financial benefit in finding “strong support” for CSE programs.
Perhaps more important, The 3Rs curriculum appears to reflect the sex education agenda that Goldfarb and Lieberman endorse in their research report. Included in the content of The 3Rs: in kindergarten a graphic lesson on identifying the genitals that “most boys have” or “most girls have;” for 6-yearolds, a lesson on gender non-conformity based on the book, “My Princess Boy;” for 7-year-olds a graphic lesson identifying male and female genitals and their functions; a lesson for 10-year-olds on homosexuality; lessons for 11-year-olds on gender roles and identity, on “gender-neutral” language (e.g., to say “someone with a vulva” instead of “girl” or “female”), a lesson that includes gay and transgender romantic relationships in which sexual activity is discussed, as well as a lesson directing 11-year-olds to a website called “sexetc.org” with stories on the homepage about masturbation and transgender men getting pregnant; for 12-year-olds, a lesson that describes “bathing together” and “mutual masturbation” with a boyfriend/girlfriend as “important because they can help … build connection between people without any risk of STDs [or pregnancy],” four lessons about transgenderism and homosexuality, and a lesson on how to decide “whether [you] want to be in a sexual relationship.”43
Thus, it appears legitimate to ask whether Goldfarb and Lieberman may have overlooked the weakness of their evidence in order to support an agenda in which they appear to have multiple vested interests.
The Goldfarb and Lieberman research review does not provide “strong support” that CSE is an effective strategy for producing the benefits it claims; it provides almost no support at all. Less than one in five of the studies cited as evidence are actually studies of CSE, and only one of those provides scientifically credible evidence of CSE’s effectiveness for any positive outcomes. Even there, the evidence is weak: there was no reduction in teen sexual risk behavior. Two of the 9 CSE studies that were designed to measure cause and effect show short-term positive impacts and one shows effects on a demographic subgroup. But these few minor indicators of program potential are not evidence of CSE effectiveness. Instead of “strong support” for CSE, this review gives the appearance of scientific backing to a new and arguable CSE agenda, for readers who do not examine the purported evidence closely.
The title of Goldfarb and Lieberman’s paper declares that three decades of research have made “the case for CSE.” Yet, when the lack of evidence in their review is combined with the poor results of school-based CSE at achieving its original purposes (reducing teen pregnancy and STDs),44 the case for CSE falls apart. If three decades of research have made any “case,” it is that it’s time to rethink CSE as the go-to strategy for sex education in school classrooms worldwide. A strategy with such a consistently poor track record for 30 years does not merit the expenditure of public funds nor the trust of parents and school districts who look to it to protect young people from harm. Rather than being expanded to encompass new, unproven, and debatable purposes, CSE should be retired and the search for more effective strategies should be intensified. Newer evidence supporting sexual risk avoidance paradigms should be a part of that search.45