Targeted High-Risk Youth in Missouri PREP: Understanding Program Impacts on Youth Sexual Behavior Intentions

Abstract

Background:

Missouri Personal Responsibility Education Program (PREP) provides sexual health education programs to youth with goals of reducing unintended teen pregnancies. Theories of change provide that youth improve their sexual health knowledge, intentions, attitudes, and behaviors as a result of program implementation. Program evaluations are needed to assess the degree to which PREP programs are meeting their goals of improving youth outcomes.

Objective:

The purpose of this study is to examine youth sexual intentions to use a condom, engage in sexual behavior, and abstain from sex as a result of Missouri PREP program implementation. We evaluate the effectiveness of the Missouri program in modifying youth intentions toward healthier planned behaviors.

Methods:

All programs required youth to take pre- and post-program surveys. For this study, we evaluate a sample of 1,335 youth’s pre- and post-survey intentions related to condom use, sex, and abstention. We utilize t-tests as well as a lagged logistic regression approach to account for youth’s respective pre-intentions.

Results:

Youth’s scores on intentions, knowledge, and attitudes rise from pre- to post-survey. Knowledge gains are salient while attitudes remain relatively high and stable. Intentions to use condoms differ from those in intentions to have or abstain from sex. Program change in intentions to use a condom are highest among the three intention outcomes.

Conclusions:

Missouri PREP saw improvements in knowledge, attitudes, and intentions as a result of program implementation. Findings suggest that the Missouri PREP program is effective at positively influencing youth intentions to engage in risky or sexual behavior.

Introduction

The state of Missouri’s teen birth rate exceeds the national average (Martin et al. 2018)¹ and ranks 43^rd (where 50 is the worst) in the rate of decline (Power to Decide 2019). Since the early 1990s, U.S. teen birth rates have decreased substantially (Pew Research Center 2019); however, national rates surpass those of other developed countries such as Canada and the United Kingdom (Kearney and Levine 2012). In efforts to reduce teen pregnancy rates, Missouri, along with other states, has implemented the federally sponsored preventative program Personal Responsibility Education Program (PREP).

The foundation for PREP’s evidence-based programming² are theories of change that demonstrate how knowledge, attitudes, intentions, and self-efficacy are integral in influencing youth sexual health decisions and behaviors. Evaluations of teen pregnancy prevention programs suggest that improvements in knowledge, attitudes, and self-efficacy as a result of program implementation predict intentions to engage in sexual behavior (Futris et al. 2019) and later behaviors such as use of contraception (Guzzo and Hayford 2018). Intentions are important indicators of later health because they reveal the motivations behind behaviors (Bandura 1982; Ajzen 1991). Therefore, teen pregnancy interventions have typically focused on theories of change to guide their curriculum.

Evaluations of the PREP program, both in Missouri as well as nationally, generally focus on the fidelity of PREP program implementation as well as whether youth learn as a result of program participation. Aside from one study that focused on foster youth (Futris et el. 2019), to our knowledge, PREP program evaluations have not focused the role of PREP curriculum on youth intentions—specifically intentions to engage in sexual behavior or practice healthy sexual behaviors (i.e., use a condom if having sex) in the future.

In this study, we use pre- and post-program survey data collected by the Missouri Department of Health and Senior Services (DHSS) as a part of the national PREP program. We assess the post-program sexual intentions of youth participants in Missouri PREP including whether youth intend to use a condom, engage in sexual behavior, or refrain from sex within the six months that follow the survey. Understanding the influence of learned program components on youths’ post-program intentions can help improve program implementation and curriculum materials to educate youth about teen pregnancy prevention and sexual health behaviors.

Missouri PREP Education

Personal Responsibility Education

In response to teen pregnancies nationwide, the Patient Protection and Affordable Care Act of 2010 (ACA, P.L. 111–148) established the Personal Responsibility Education Program (PREP) to implement preventative health care programs for adolescents throughout the United States. The Missouri Department of Health and Senior Services (DHSS) was awarded state PREP funding from the Federal Youth Services Bureau (FYSB) to implement PREP. The Missouri PREP emphasizes the importance of both long-term (i.e., reducing teen pregnancy, STIs, and HIV rates) and short-term goals (i.e., increasing knowledge and positive behaviors) (Institute of Public Policy 2017). Following federal PREP program guidelines, Missouri chose from a list of pre-approved and rigorously evaluated curricular choices. The Missouri DHSS implemented three different curricula: Teen Outreach Program (TOP), Making Proud Choices (MPC), and Becoming a Responsible Teen (BART).

PREP programs follow a top-tier evidence approach, giving each grantee the option to: a) replicate evidence-based programs proven to be associated with measured outcomes or b) opt for other and more experimental programming approaches while incorporating desired elements of the evidence-based programs. Missouri DHSS selected option ‘a’ via these three PREP curricular programs. Three were chosen in order to accommodate the demographic, geographic, and socioeconomic diversity within the Missorui communities targeted for program implementation. Importantly, Missouri DHSS expected each program to produce similar results for youth sexual behavioral outcomes.

Evaluations of teen pregnancy prevention programs suggest that regardless of curriculum, interventions are associated with positive teen pregnancy outcomes including delay of sex, reduction in the number of partners or frequency of sex, and increase in the use of condoms or contraceptives (Kirby 2012; Goesling et al. 2013). More specifically, the curriculuar programs chosen by Missouri DHSS have been nationally proven to positively impact youth in several ways. TOP participation is associated with reduced rates of pregnancy, school suspension, and class failure (Allen et al. 1990; Allen et al. 1997; Allen & Philliber 2001; Daley et al. 2015). MPC participation is associated with delays in initiation and reduced frequency of sex, decreases in occurrences of unprotected sex, and increased condom use (Jemmott et al. 1998). In Missouri specifically, MPC’s fidelity has been assessed cross-sectionally and found to significantly increase youth sexual knowledge and intent to use condoms in different settings and demographic groups from pre- to post-program surveys (Cronin et al. 2014). BART is associated with a reduction in male youths’ number of sexual partners and frequency of sex, and increase in condom use (St. Lawrence et al. 1993; St. Lawrence et al. 1995; Butts et al. 2002; Malow et al. 2009). The current study offers a Missouri-specific evaluation of youths intentions within TOP, MPC, and BART programs.

Missouri DHSS chose TOP, MPC, and BART because they were determined as the best fit for Missouri’s diverse population. High rates of teen pregnancy are not confined to a particular geographic region of the state or to a particular demographic group. Missouri PREP targets youth in locations where teen birth rates and teen pregnancy are especially high. Youth who are deemed high risk for teenage pregnancy such a foster youth for example, are especially vital to target for programming. Roughly 20 percent of Missouri PREP youth are in foster care (Ressel et al. 2017). For foster youth, sexual experiences are common and educational opportunities for sexual and reproductive health are rare (Rouse et al. 2020). Relative to the Missouri average, the counties selected for Missouri PREP had high rates of hardship based on measures of child poverty, teen birth rates, and unemployment (Missouri Kids Count 2018). Income inequality and poverty are linked to increased teen birth rates (Gold et al. 2002); economically disadvantaged youth are more likely to give birth than their less disadvantaged peers (Kearney and Levine 2014).

Each Missouri curricular program (TOP, MPC, and BART) vary in their delivery and implementation practices to best influence the youth they are purported to target. As shown in Table 1, TOP curricula is delivered over the course of the entire school year, either during or after school and is typically implemented in urban areas. MPC and BART are implemented over a shorter timeframe, sometimes on weekends or after school. The program curricula are taught in various settings yet are most often implemented within a school, typically during or after school hours (personal communications with the Institute of Public Policy).

Table 1:

Missouri PREP Curricular Types (BART, MPC, TOP)

Program	Program type	Number of Lessons	Intervention Length	Target Population	Age Group	Curricula Focus
BART	Comprehensive	8	8 weeks / 1 lesson per week	African American Youth	14–18 years	HIV/AIDS prevention, Communication, Negotiation, and Problem-Solving
MPC	Comprehensive	8	One day - 8 weeks / 1–4 modules a day (shorter) or 2 modules/week (longer)	At-Risk and Foster Youth	11–18 years	Abstinent choices, Healthy relationships, Sensitive to and addresses issues of concern to youth in care
TOP	Adult Preparation	25+	9 months / 1 lesson per week	At-Risk Youth	<13–19 years	Youth development, Community service learning, and Relationship building

Source: Missouri Department of Health and Senior Services. (2010). Personal Responsibility Education Program (PREP) Post Award State Plan.

Nationally, BART places focus on African American youth, but was commonly implemented in some of Missouri’s rural localities. Based on lesson offerings and adaptable program structures, MPC and TOP curricula offer the most amount of flexibility in course offerings. MPC commonly focuses on younger youth while TOP and BART target a wider age range. The MPC curricula is most utilized for foster and juvenile youth.

Each curricula is expected to be similar in producing positive life outcomes for youth. Thus, we do not focus particularly on differences by curricula youth are enrolled but do control for any differences due to contextual diversity in program site and youth characteristics. The three curricula primarily focus on comprehensive sexual education while stressing that abstinence is the safest way to prevent pregnancy. With three available choices, each contractor (the implementing organization) was granted flexibility in which curriculum they would offer, customizing programs to fit youth needs in terms of materials, timeline, and demographics. Standardized pre- and post-program surveys were used to assess PREP youths’ intentions. Each program integrates theories of change yet with fluctuating foci to educate their targeted population most effectively (see Table 1 below).

Theoretical Background of Pregnancy Prevention

Theories of Change

Theories of change that underlie many family planning and education programs (Weis 1995) guide the development of PREP curricular programs. Theories of change include a host of individual theories that describe how changes in knowledge, attitudes, intentions, and self-efficacy are associated with an individual’s behaviors. The theories central to PREP programming include the Theory of Planned Behavior (Ajzen 1991), Social Cognitive (Bandura 1986), Self-Efficacy (Bandura 1997), and Social Learning (Bandura 1977). As we explain below, theories of change are important to teen pregnancy prevention programs because behaviors shape life outcomes. Therefore, the theories provide a basis for evaluating the factors associated with behavioral change as they relate to pregnancy prevention programs.

Theories of change state that knowledge, attitudes, perceived behavior, and norms together shape intentions and behaviors (Ajzen 1991). This relationship is outlined in Figure 1, where knowledge, attitudes, efficacy, and behavioral norms inform intentions, which ultimately influence later life outcomes or behaviors. Attitudes about sexual health are found to predict family planning behavior (Guzzo and Hayford 2018); learned knowledge and behaviors in teen pregnancy prevention interventions are valid indicators of program effectiveness (Singh, Bankole, and Woog 2005). Learned knowledge and skills increase intended behavioral patterns toward making healthier decisions about sex and condom-use intentions (Futris et al. 2019). For example, healthy sexual behaviors and contraceptive use have been associated with a decrease in the rate of unintended youth pregnancy both in Missouri (Peipert et al. 2014) and nationally (Manlove et al. 2015; Santelli et al. 2007). Intentions are important to focus on because they foreshadow future behaviors (Ajzen 1991). An individual’s previous behaviors can foreshadow future actions. Based on these theories, behavior is influenced by an individual’s judgment of how well they are able to adapt to prospective situations (Bandura 1982).

Figure 1:

Hypothesized Relationship of Missouri PREP Factors on Youth Sexual Behavior Intentions

Based on theories of change and as discussed in the text above, Figure 1 outlines our hypotheses. Our study is informed by a meta-analysis conducted by Sheeran et al. (2016) who found that change in attitudes, norms, and self-efficacy affect intentions and behaviors; studies that elicit change in these cognitions promote healthy behaviors. We predict the following:

1. Program-change in attitudes and knowledge of sex and health behaviors following PREP participation will be related to sexual intentions at the end of a youth’s PREP program experience.

2. Efficacy and classroom behavioral norms will be related to post-program intentions.

Within our stated hypothesis and as a part of our analysis, attitudes and knowledge are assessed as measures over time, whereas self-efficacy and classroom norms are not due to measurement differences in self-efficacy from pre-post program surveys and minimal change in behavioral outcomes over time. We include pre-program, or baseline measures of efficacy and classroom behavioral norms to better understand youth post-program intentions. Classroom behaviors are used as a proxy measure for overall behavioral norms.

Data and Methods

For this study, we utilize secondary pre- and post-survey data of Missouri PREP youth that were obtained from the Institute of Public Policy (IPP) at the University of Missouri who oversee PREP evaluation for the Missouri DHSS, the grantee leading the program. Survey questions were designed by the FYSB to assess the overarching PREP initiatives, and the Missouri DHSS along with the IPP added questions to reflect the three curricula being utilized. We utilize results from the surveys to assess Missouri youth and their intentions post-PREP implementation.

In accordance with TREND guidelines for nonrandom interventions (https://www.cdc.gov/trendstatement/), we provide details about the Missouri PREP implementation. Futher details about Missouri PREP can be found online (https://motpp.missouri.edu). In order to implement PREP, the Missouri DHSS performed readiness assessments to determine a community’s PREP eligibility and a needs assessment of communties with the highest rates of teen pregnancy. Once deemed eligible based on overall readiness and receptiveness to delivering PREP programming, site-coordinators identifed organizations within the community to oversee PREP programming. The selected community organizations included schools, health and/or community organizations. Organizations then selected faciliators to undergo training in order to become certified to teach PREP curriculum to youth. Students were recruited by coordinators to participate in PREP classes within the communities that they lived. High risk youth were especially targeted for programming based on the program that was being implemented (e.g., TOP targets youth in schools whereas MPC is focused on youth within community-based or foster care organizations). Classes were administered in settings including schools, churches, and community organizations and varied in duration and format depending on the curriculum (e.g., BART, MPC, or TOP). The communities in which PREP classes were implemented were select in that they were particulary ready and able to provide teen pregnancy intervention. The students who participated in their community PREP program were not-randomly selected, a fact that should be considered when discussing the findings and the external validity of the findings to the broader population of PREP participants nationally or other similar aged youth who have not participated in PREP. To help alleviate this issue, we review other studies that have included both longitudinal data and information on effect sizes from which to benchmark our findings, however we do not directly compare results due to differences in methods and sample design. We also calculate Cohen’s D to approximate the effect sizes of our results.

Upon entrance into the program, parental consent forms were sent home with students to obtain permission to participate in the class and for the release of the student’s survey data (N=3,047). We restricted our study sample to those who completed the program (69 did not complete) and those who completed both pre and post survey questionnaires (N=2,628). A total of 388 participants were dropped because they participated in the program for multiple years or in more than one curricular program in a given year. Youth who participated in the program more than once may yield different results than those who participated only once. Respondents with missing information on key analytic variables are excluded from the sample using casewise-deletion yielding a final analytic sample of 1,335 youth. Using similar methodology to obtain their analytic sample, Futris and colleagues (2019) were able to retain about 41.9 percent of participants by excluding youth who did not complete the program, did not complete both pre- and post-surveys, and those with 30 percent or more missing data. This retention rate is similar to our rate of 43.8 percent. T-test comparisons between our analytic sample (N=1,335) and the full sample without duplicate entries (N=2,659) are shown in Appendix A.

At the first PREP class, students completed a pre-program survey asking basic demographic information along with questions about current knowledge, attitudes, intentions, norms, self-efficacy, and behaviors associated with pregnancy and parenting, STDs, HIV, life skills, and related components. Upon completion of the program’s final class, students completed a similar post-program survey. Student survey data were compiled from five completed program years beginning with 2012–2013 (year two) and ending with program year 2016–2017 (year six). Year one was Missouri PREP’s pilot year and is not included in this study.³ IRB exemption was obtained as no identifying or sensitive data were utilized.

In addition to the survey data, we merge on publicly available county-level contextual data obtained from Missouri Kids Count collected by University of Missouri Office of Social and Economic Data Analysis (OSEDA) and University of Missouri Extension in collaboration with Family and Community Trust (FACT) and Children’s Trust Fund.⁴ These data include county-level contextual information about where the PREP was implemented. We include measures of the percent of children in poverty, births to teens aged 15–19, and adult unemployment.

Measures

Students responded to demographic questions including age, grade, male/female sex, and race. They were then asked about their sexual history, knowledge, attitudes, self-efficacy, and classroom behavioral components in relation to their thoughts, feelings, and attitudes on sexual health and their own individual health and behaviors. The study’s pre-post design is utilized to measure change in program outcomes resulting from PREP participation. Both pre- and post-surveys measured youth intentions identically allowing us to evaluate these outcomes by curricula over time. Knowledge and attitudes were also identically measured from pre- to post-survey, yet self-efficacy was changed in the post-survey to reflect how the program influenced thoughts and feelings.

Sexual Intention Outcomes

Three separate intentions were measured by asking youth whether within the next six months they intended to: use a condom if intending to have sex, engage in sexual behavior, or abstain from sex. The four-point Likert scale ranged from “no, definitely not” (1) to “yes, definitely” (4). We dichotomized each outcome measure to equal one if the respondent answered either of the two “yes” responses and 0 if “no”. Sensitivity analyses confirmed minimal loss of information by collapsing response categories (available on request). We include the respondent’s pre-intentions responses as independent variables so that the post-program responses can be interpreted relative to the initial, pre-survey responses.

Independent and Control Variables

Our analytic approach is guided by a meta-analysis conducted by Sheeran et al. (2016) who find that attitudes, behavioral norms, and self-efficacy are associated with intentions. Teen pregnancy prevention programs were positively related with factors such as attitudes, norms, and self-efficacy in increasing healthy behaviors (Sheeran et al. 2016). To follow and expand on the list of measures that may be informative of later intentions, our primary independent variables include change in knowledge, change in attitudes, pre-program efficacy, and classroom behavior questions.

The same knowledge and attitude questions were asked of all youth in Missouri’s three PREP curricula⁵. Ten knowledge questions captured information relating to pregnancy and HIV/STDs. Respondents were prompted to answer true/false/don’t know. Some examples include, “Most people who have HIV know they have it”, and “A women cannot get pregnant the first time she has sex”. The “don’t know” responses were recoded as missing values since these responses may reflect other processes. The ten knowledge questions were summed to create a score for number of correct or positive responses at the pre- and post-survey time periods. The final knowledge variable is the difference in pre- and post-program survey responses, yielding change. This change score measure includes a possible range of −10 to 10 correct responses.

Youth were prompted to respond to a list of seven attitudes using a five-point Likert scale on whether they agreed or disagreed with a given statement. These questions relate to sexual intercourse and condom usage on a personal level and include statements such as, “I could say no to the person going out with me if I don’t want to have sex.” Attitude questions were also summed to create a single scored variable in both pre- and post-surveys. The final attitudes variable consists of program change from pre- to post-program surveys with a possible range of −35 to 35.

For efficacy, pre-survey responses were utilized. Post-survey questions were asked differently from those in the pre-survey and were therefore omitted⁶. Eight pre-survey self-efficacy questions were asked regarding how often the respondent said that they felt a certain way about a particular topic within the last three months. The four response options ranged from “all” to “none of the time” with higher values corresponding to higher self-efficacy. Efficacy topics included being able to manage stress, caring about doing well in school, and managing friendships and conflict, for example. This measure was summed with a range from 0 to 32. Higher values indicate more positive feelings toward topics.

Behavioral questions included classroom norms such as failing a grade or class and cutting class, and were coded dichotomously (1=yes; 0=no). These three measures were summed to create a classroom behavior score for the number of reported occurrences. The score ranged from 0 to 3 and higher values indicate greater levels of deviant behavior.

We also control for youth demographic characteristics (i.e., age, male/female sex, and race) and a contextual county-level variable measuring the economic wellbeing of the locality in which the respondent participated in PREP⁷. The contextual county-level measure is for each county in which a PREP program was held. In our analysis, contextual measures include the percent of children in poverty, births to teens aged 15–19, per 1,000, and the percent of adult unemployed. The three measures were transformed into comparable variables due to their measurement differences through attention to values in their range, mean, and standard deviation. For example, a new variable was created for percent of children in poverty by allocating values that were two or more standard deviations below the mean variable value to equal one. The new variable would equal 2 if percent poverty was one standard deviation below the mean up to the mean value, 3 if a value was equal to the variable mean up to a standard deviation above the mean, and 4 if two or more standard deviations above the mean. The three measures were then summed to create an index score ranging from 3 to 12 with higher scores relating to worse overall county wellbeing⁸.

Analytic Strategy

Our goal is to evaluate the correlates of post-program sexual behavior intentions of Missouri PREP participants. We utilize a two-step approach to first determine whether youth benefit from the program, and second to investigate the predictive components of sexual behavior intentions. We begin with descriptive analysis to evaluate mean differences in our intention outcomes, youth knowledge, and their attitudes from pre- to post-program implementation using t-tests. Second, we estimate lagged logistic regression models to predict post-program intentions. We include respective pre-program intention measures to control for beginning program characteristics. For example, when accounting for post-program intentions to use a condom, we control for pre-program condom-use intentions to account for pre-program responses. Other pre-program measures are self-efficacy and classroom behaviors. We also include contextual-level measures, demographic control variables, and program-change in knowledge and attitudes to determine PREP’s impacts on youths within TOP, MPC, and BART PREP curricular programs. The lagged logistic regression models allow us to test the relationship between PREP program components and end of program intentions, while also accounting for youth beginning intentions. We expect that results should be consistent in that gained positive attributes (i.e., knowledge and attitudes) from the Missouri PREP program will be associated with improvements in sexual behavior intentions (i.e., decrease in intentions to engage in sexual behavior and/or increase in intentions to use a condom if deciding to have sex).

Results

Descriptive Statistics

Table 2 presents descriptive statistics of youth demographic information for the analytic sample. The average participant is roughly 15 years of age. The sample is 57 percent female with 44 percent of respondents identifying as white, 37 percent African American, and 20 percent of another race. Twenty-one percent of youth were enrolled in the BART curricula, 56 percent in MPC, and 23 percent in TOP. According to the Missouri Census Data Center (2020), 49 percent of youth ages 0–17 in Missouri are female, 75 percent identify as non-Hispanic white, and 15 percent are non-Hispanic black.

Table 2:

Descriptive Statistics of Missouri PREP Youth Demographics and Select Survey Components

Demographics	BART	MPC	TOP	Overall	Range
Age	15.96	15.38	13.13	15.11	[10,21]
Female	0.50	0.55	0.69	0.57	[0,1]
White	0.25	0.50	0.44	0.44	[0,1]
African American	0.59	0.27	0.40	0.37	[0,1]
Other	0.16	0.22	0.15	0.20	[0,1]
Other Survey Components
Program	0.21	0.56	0.23	-	[0,1]
County Wellbeing Index	7.06	6.31	8.86	7.06	[3,11]
Pre-efficacy	23.95	23.06	23.87	23.43	[7,32]
Classroom Behaviors	0.92	0.95	0.45	0.83	[0,3]
N	275	748	312	1,335

Out of a possible score of 12, the average county wellbeing score is 7.06, with higher scores indicating poorer county wellbeing. Turning to the program components, for classroom behaviors the average negative behavior was 0.83 on a scale of three negative behaviors (fail class/grade/cut class). Average pre-program efficacy for the utilized sample was 23.43 of a possible score of 35 positive thought/feeling points.

Pre-Post Learning

Table 3A presents pre- and post-survey measures for knowledge, attitudes, and the three sexual intention measures. We descriptively test for evidence of change in sexual intentions, knowledge, and attitudes from pre- to post-survey overall and within each curricular program. Although results vary across curricula, change is apparent overall for each outcome. Table 3A suggests significant differences in intentions to use a condom, engage in sexual behavior, abstain from sex, youth knowledge, and youth attitudes overall. In Table 3B, we report Cohen’s d for each of our pre to post measures to estimate effect sizes. While differences for the three intentions are modest, change in knowledge is large. The reason for reporting Cohen’s d is to assess the likelihood that differences in our sample are due to program implementation and not random chance. Given the presence of change and by benchmarking our results against prior studies that find similar program change in attribues (see Kim et al. 1997; Malow et al. 2009; Boustani et al. 2017), results suggest evidence of change. Effect size comparisons are difficult to account for across studies because not all studies include effect sizes in their analyses; if so, they are not always comparable due to differences in study design, methods, or metric used to measure the effect (Goesling et al. 2014).

Table 3A:

Missouri PREP Youth Pre- to Post-Program Change in Intentions, Knowledge, and Attitudes by Curricular Program

	BART	MPC	TOP	Overall
Intentions
Pre-Condom	0.87	0.85	0.82	0.85
Post-Condom	0.93	0.91	0.90	0.91
Difference	0.05	0.06	0.07	0.06
Test Statistic	2.25**	4.16***	3.03**	5.61***
Pre-Sex	0.39	0.38	0.11	0.32
Post-Sex	0.44	0.39	0.13	0.34
Difference	0.05	0.02	0.02	0.02
Test Statistic	1.67*	1.19	1.40	2.34**
Pre-Abstain	0.55	0.56	0.75	0.60
Post-Abstain	0.60	0.58	0.79	0.63
Difference	0.04	0.03	0.04	0.03
Test Statistic	1.30	1.42	1.47	2.38**
Pre-Knowledge	7.03	6.66	5.25	6.40
Post-Knowledge	8.05	8.37	6.76	7.93
Difference	1.03	1.72	1.51	1.53
Test Statistic	7.77***	20.83***	11.69***	24.66***
Pre-Attitudes	29.26	28.59	29.20	28.87
Post-Attitudes	29.27	29.74	28.88	29.44
Difference	0.01	1.15	−0.31	0.57
Test Statistic	0.04	6.58***	−1.03	4.33***
N	275	748	312	1,335

Note: Test statistics are shown for paired t-tests

^*p<.10;

^**p<.05;

^***p<.01

Table 3B:

Effect Sizes for Pre- to Post-Program Measures

	Cohen’s D
Intentions to:
Use a condom	0.19
Engage in sexual behavior	0.05
Abstain	0.07
Knowlegde	0.71
Attitudes	0.14

N = 2,670

On average, about 91 percent of youth state that they intend to use a condom if they have sex within the next six months on the post-survey yielding a six percent increase from pre-survey responses. Youth in each of the three curricular programs increase their condom-use intentions by about 5–7 percentage points. For intentions to engage in sexual behavior, youth increase 2 percentage points overall: post-PREP, 34 percent plan to engage in sexual intercourse within the next six months. Statistically significant changes for sex intentions occur both overall and for BART youth. Simultaneously, intentions to abstain from sex increase overall by three percentage points: 63 percent of youth planning to abstain from sex within the next six months.

Regarding knowledge, youth answered on average 7.9 out of 10 answers correctly on the post-program survey, a gain of 1.5 additional questions answered correctly. Youth from all three curricular programs experienced a statistically significant increase in knowledge over the course of PREP. Attitudes also increased, yet the increase is smaller with an additional 0.6 of a point increase in score on average; this increase appears to be improved mainly via the MPC program. Out of a possible score of 35, post-program intentions average 29 points, indicating relatively positive attitudes on agreeance with ideas toward sexual intercourse and condom usage prior to PREP.

Lagged Regression Analyses

Table 4 reports logistic regression results in terms of odds ratios while accounting for respective pre-intention survey responses. Generally, intentions to use a condom have disparate predictors than those intentions to engage in sexual activity or to abstain from sex. As expected, the results for intentions to engage in sexual behavior work in the opposite direction of intentions to abstain from sex.

Table 4:

Logistic Regression Results of Missouri PREP Youth Post-Program Intentions

	Condom		Sex		Abstain
	b	SE	b	SE	b	SE
Pre-intentions
Intentions to use condom	7.20***	1.54
Intentions to engage in sexual behavior			22.89***	3.93
Intentions to abstain					5.94***	0.81
Age	0.95	0.06	1.16**	0.06	0.85***	0.03
Female	1.36	0.28	0.54***	0.09	1.75***	0.24
Race (White=reference)
African American	0.54**	0.13	0.77	0.15	1.06	0.17
Other	0.76	0.22	0.87	0.20	0.77	0.14
County Wellbeing Index	1.02	0.05	1.03	0.04	1.02	0.03
Program (BART=reference)
MPC	0.59	0.18	0.87	0.19	0.79	0.14
TOP	0.52	0.19	0.51*	0.15	1.09	0.26
Knowledge change	1.07	0.05	0.84***	0.03	1.16***	0.04
Attitude change	1.06**	0.02	1.00	0.02	1.00	0.01
Pre-Efficacy	1.02	0.02	0.97	0.02	1.03	0.02
Classroom behaviors	0.91	0.10	1.312***	0.11	0.82**	0.06
Constant	4.39	5.45	0.05	0.05	2.94	2.44
R2	0.12		0.42		0.21

Missouri PREP Pre- and Post- Survey Data (Program years 2011–2017); Missouri Kids Count Data (2017)

^*p<.05;

^**p<.01;

^***p<.001

N= 1,335

Results reported in odds ratios

Overall, compared to whites, African American youth are statistically and significantly (46 percent) less likely to report intentions to use a condom in the post-program survey. The gain in attitudes during the program is predictive of higher condom use intentions. However, several of these factors are statistically significant predictors of planned sex or abstention intentions. For example, age, female sex, knowledge change, and classroom behaviors are consistently significant predictors of intentions to engage in sexual behavior or abstain from sex within the next six months. For intentions to engage in sexual behavior, older youths and males are more likely to state that they are planning to engage in sexual intercourse. Knowledge is negatively related to sex intentions: each additional point of knowledge gain is associated with a 16 percent decrease in intentions to engage in sexual behavior. For behaviors, each additional negative classroom behavior is associated with a 31 percent increase in the intention to engage in sexual behavior. Youth in the TOP program are also less likely than youth in BART to intend to engage in sexual behavior.

For intentions to abstain, the same characteristics are significant, yet operate in the opposite direction of intentions to engage in sexual behavior. Females are overwhelmingly more likely than males to intend to abstain from sex (75 percent more likely). Younger youth, as opposed to older, are more likely to state that they plan to abstain from sex within the next six months. Also, larger gains in knowledge and fewer occasions of negative classroom behaviors are associated with intentions to abstain. For both intentions to engage in sexual behavior and intentions to abstain, neither race nor attitudes are significant predictors of post-program responses. County wellbeing and pre- measures of self-efficacy are not significant predictors of any of the three intention behaviors though sensitivity analyses suggest that county wellbeing is statistically different across the three curricular programs (shown below in Table 5).

Table 5:

Multinomial Logistic Regression Results for Curricular Program Differences (TOP = Reference group)

	BART		MPC		BART		MPC
	b	SE	b	SE	b	SE	b	SE
Pre-intentions
Intentions to use condom	0.36	0.26	0.24	0.20	0.19	0.30	0.10	0.24
Intentions to engage in sexual behavior	0.93***	0.30	1.03***	0.26	0.44	0.35	0.56	0.31
Intentions to abstain	−0.04	0.23	−0.04	0.19	0.06	0.27	0.00	0.23
Post-intentions
Intentions to use condom	0.29	0.33	0.19	0.25	0.48	0.37	0.23	0.30
Intentions to engage in sexual behavior	0.98***	0.28	0.63**	0.25	0.79*	0.33	0.54	0.30
Intentions to abstain	−0.18	0.24	−0.39*	0.20	0.23	0.27	0.02	0.23
Age					0.77***	0.06	0.45***	0.05
Female					−0.53**	0.21	−0.40*	0.18
Race (White=reference)
African American					1.51***	0.23	−0.16	0.19
Other					0.65*	0.30	0.04	0.24
County Wellbeing Index					−0.30***	0.05	−0.48***	0.04
Constant	−1.01***	0.37	0.42	0.28	−10.78***	1.16	−2.20*	0.93

^*p<.05;

^**p<.01;

^***p<.001

N = 1,335

Supplementary analysis suggests that knowledge and attitudes are weakly correlated (r=.107, p=0.0001). We tested whether they moderated one another but found no evidence of significant interactions in predicting intentions (results available from authors). While significant, the correlations between pre-program self-efficacy and intentions also were all weak (r= <.10). In supplementary models, post-program self-efficacy replaced pre-program self-efficacy and significantly and positively predicted intentions to abstain from sex, but not the other intentions. The correlations between post-program self-efficacy and intentions are modest (between .04 and .13); the magnitude is greater than pre-program self-efficacy and intentions.

To aid in interpreting results, we produced predicted probabilities of each sexual intention by curricula. As shown in Figure 2, TOP youth appear to have low predicted probabilities of intending to engage in sexual behavior and higher predicted probabilities of intending to abstain than those of the BART or MPC curricula. Age and male/female sex differences likely contribute to differences given that TOP participants are on average the youngest and disproportionately female compared to other programs.

Figure 2:

Youth Sexual Behavior Intentions by PREP Curricular Program

In sensitivity analysis, we tested multinomial logistic regression models (Table 5) to identify whether our six key intention measures (pre/post intentions to use a condom, engage in sexual behavior, or abstain from sex) and key demographic information (age, male/female sex, race, and county wellbeing) differed among programs. We find significant differences in intentions to engage in sexual behavior among TOP versus MPC and BART youth (in support of descriptive findings in Table 3A) and differences in TOP versus MPC youth intentions to abstain. MPC and BART youth intentions were not significantly different from each other (results not shown). When controls for demographic characteristics are introduced (right side of Table 5), age, male/female sex, and county wellbeing accounted for the program differences among all three curricular programs, rendering all but one intention (post-sex intentions in TOP compared to BART) non-significant. Race was also significantly different among programs except for when comparing TOP and MPC (TOP vs. BART and MPC vs. BART were significantly different) (results not shown).

Due to program implementation differences, we ran logistic regression models (Table 6) by curricular program to determine whether key demographic characteristics are related to program differences in intentions to engage in sexual behavior. We show that age and female sex significantly predict post-intentions to engage in sexual behavior when run separately by each curricular program in all cases except for sex in TOP. Older youth and males are more likely than younger participants and females to report intentions to engage in sexual behavior post-program implementation. For BART participants, race is a significant predictor of sex intentions; African American youths are less likely than white youth to report intentions to engage in sexual behavior post-PREP program.

Table 6:

Logistic Regression of Descriptive Youth Characteristics on Post-Intentions to Engage in Sexual Behavior

	BART		MPC		TOP
	b	SE	b	SE	b	SE
Age	0.26**	0.09	0.22***	0.04	0.49***	0.10
Female	−1.21***	0.27	−0.83***	0.16	−0.72	0.38
Race (White=reference)
African American	−0.63*	0.31	−0.11	0.19	0.01	0.39
Other	−0.68	0.43	−0.07	0.20	−0.34	0.61
County Wellbeing Index	−0.02	0.07	−0.03	0.04	0.08	0.09
Constant	−3.32*	1.58	−3.08***	0.80	−9.06***	1.78
N	275		748		312

^*p<.05;

^**p<.01;

^***p<.001

Discussion

Missouri participated in the nationally implemented federal PREP program as an initiative to combat high teen pregnancy rates. Grant funding was obtained from the Federal Youth Services Bureau for the state to implement pregnancy prevention programming to high-risk youths. Five years of data were collected from youth pre- and post-participation. We observe program characteristics related to three separate youth sexual intentions: condom use, sexual activity, and abstention. This study adds to prior PREP literature by providing insight to the factors that influence youths’ intentions as a result of PREP program implementation. The Missouri-specific focus helps to further understand how youth in Midwestern localities are faring post-PREP. This work is an extension of prior program evaluations that have proven program effectiveness and that youths benefit from program implementation.

Overall, youth gain knowledge and report improvements in attitudes. They also report increases in their intentions to use condoms, engage in sexual behavior, and abstain from sex. The predictors of post-program intentions vary by outcome. Consistent with past research (see Jemmott and Jemmott 2010; Kim et al. 1997), the increase in intentions to use a condom is larger than that of the change in intentions to engage in sexual behavior or abstain from sex. For condom use intentions, race (African American youths) and attitude change are predictive of post-program survey responses. For sex and abstention intentions, age, female sex, knowledge change, and classroom behaviors are significant predictors of post-program intentions. These results are consistent with Sheeran et al. (2016) who find that learned program components positively influence youths’ later life outcomes. Our results for program change also compare to findings from prior studies (see Kim et al. 1997, Malow et al. 2009, and Boustani 2017), however we caution the comparison of effect sizes due to differences in sample design between studies.

Supporting prior literature (i.e., Bandura 1977; Ajzen 1991), learned knowledge and attitudes resulting from teen pregnancy prevention programming (i.e., change in knowledge and attitudes, respectively) are associated with increases in positive sexual behavior intentions. However, our findings show that while knowledge and attitudes are associated with intentions, they are not both significant predictors of each outcome. It is possible that although youth may know the protective benefits of condom use, their attitudes may ultimately influence their decisions to use them. Alternatively, knowledge may contribute to youths’ intentions to engage in sexual behavior or abstain from sex because of an increased awareness of the implications of intercourse as well as methods to prevent pregnancy. Motives for intended condom use likely differ from the reasons that individuals choose to have (or not have) sex. For example, motives for having sex could include intimacy or desire, or could be for other reasons such as it “just happened” (Dawson et al. 2008). Decisions to use a condom can be mediated by levels of intimacy and the relationship context (Cooper et al. 1998). These findings suggest that comprehensive teen pregnancy prevention programs should continue focusing on the myriad of factors that influence sexual intentions. Additionally, focusing on more than one type of intention will continue the program’s comprehensive nature of learning both for youth and for evaluators.

Interestingly, we do not find evidence of a relationship between pre-program self-efficacy and intentions as hypothesized. Given differences in how the efficacy questions were asked in the pre- and post-program surveys, we were unable to construct an efficacy change score. On the one hand, the absence of significant results may suggest that changes in sexual intentions are not dependent on starting levels of efficacy. Though we are unable to measure this directly, it is possible that participation in the programs made youths feel more efficacious in ways that spanned beyond the measures included in the survey. Self-efficacy as it relates to sexual knowledge, attitudes, and intentions is a promising area for future research. An individual’s self-efficacy, or judgment of how well they are able to deal with prospective situations determines later behaviors (Bandura 1982). Future studies should ask questions to allow for assessments of changes in efficacy over the course of program participation as well as covering additional domains of efficacy related to sexual behaviors.

We find age to be positively related to intentions to engage in sexual behavior, and negatively related to abstention (meaning younger youth are more likely to abstain). This finding is supported by Poobalan et al. (2009) who emphasize the importance of attention to timing and intention-related risk factors in youth pregnancy prevention programs. Educating youth before they become sexually active can play an important role in influencing intention behaviors (Poobalan et al. 2009). Once youth are already sexually active, they are less likely to change their sexual behaviors (Dickson et al. 1997). In addition, special populations, such as foster youth, report fewer resources, support, and educational opportunities despite reports of heightened sexual frequency (Rouse et al. 2020). These youth are especially important to continue to target for pregnancy prevention and health resources especially at early ages.

In addition to age, male-female sex variation is evident in prevention programming. Kirby (2012) found that females were more receptive to programming than males, and also more likely to show change in knowledge, attitudes, and intentions (Applegate 1998). Similarly, our study shows that females are more likely than males to state abstention intentions, and also less likely to state intentions to engage in sexual behavior. These results could be explained through male-female sex variation in perceived risk factors or emotions. Houck et al. (2014) find that females display more negative emotions (e.g. sad, nervous, scared, etc.) before having sex than males; youths’ negative emotions are associated with a decrease in intentions to engage in sexual activity in the future. In addition, females may weigh the risks of teen pregnancy more heavily than males because they would more directly experience an unintended pregnancy. These are plausible avenues for future research but beyond the scope of the current Missouri PREP data.

Future research, both qualitative and longitudinal in nature, is necessary to disentangle the precursors of each sexual intention. A closer examination of gender variation in program outcomes as well as youth in other curriculum outside of Missouri could verify the generalizability of our findings. Follow up with PREP youth as well as data on actual behaviors could enhance knowledge surrounding the relationship between youth intentions and health behaviors. These results could also be compared with abstinence-only based classes to isolate intention factors related to behavior outcomes. Furthermore, enhancements to Missouri PREP programming could be made through conducting a follow-up study of previous PREP participants to understand the program’s long-term impacts.

Appendix A:

Descriptive Statistics of Missouri PREP Youth Demographics and Survey Components (Full vs. Analytic Samples)

	Full Sample (N=2,659)		Analytic Sample (N=1,335)
	Mean / %	Range	Mean / %	Range	Test statistic
Age	14.56	[10,21]	15.11	[10,21]	−14.31***
Female	0.56	[0,1]	0.57	[0,1]	−0.62
Race					0.84
White	0.42	[0,1]	0.44	[0,1]	−1.72*
African American	0.39	[0,1]	0.37	[0,1]	2.19*
Other	0.19	[0,1]	0.20	[0,1]	−0.55
County Wellbeing Index	7.66	[3,11]	7.06	[3,11]	1.37***
Program					16.29***
BART	0.19	[0,1]	0.21	[0,1]	−2.59***
MPC	0.37	[0,1]	0.56	[0,1]	−21.58***
TOP	0.44	[0,1]	0.23	[0,1]	23.57***
Knowledge
Pre-program	6.08	[0,10]	6.40	[0,10]	−8.05***
Post-program	7.64	[0,10]	7.93	[0,10]	−9.92***
Attitudes
Pre-program	27.77	[2,35]	28.87	[5,35]	−11.24***
Post-program	28.71	[5,35]	29.44	[11,35]	−12.15***
Pre-efficacy	22.96	[1,32]	23.43	[7,32]	−5.26***
Classroom behaviors	0.78	[0,3]	0.83	[0,3]	−2.88**
Intentions to use condom
Pre-program	0.80	[0,1]	0.85	[0,1]	−7.59***
Post-program	0.88	[0,1]	0.91	[0,1]	−6.32***
Intentions to engage in sexual behavior
Pre-program	0.29	[0,1]	0.32	[0,1]	−3.07**
Post-program	0.31	[0,1]	0.34	[0,1]	−4.61***
Intentions to abstain
Pre-program	0.60	[0,1]	0.60	[0,1]	0.09
Post-program	0.65	[0,1]	0.63	[0,1]	2.34*

Note: Full sample includes non-duplicate information on all survey participants with consent to participate in the survey. Casewise survey eliminates those who dropped out of the program or those with missing data. Test statistics are shown for paired t-tests in difference between full and casewise samples.

^*p<.05;

^**p<.01;

^***p<.001