This brief gives evaluators of teen pregnancy prevention programs survey items and outcome measures to consider when assessing the impact of their programs.
The outcomes included are drawn from Child Trends’ evaluation of Re:MIX, a comprehensive sexual health education program that aims to reduce rates of unintended pregnancy and sexually transmitted infections (STIs) among primarily Latinx adolescents (grades 8-10) in Texas. Child Trends included measures from four mediating outcome domains (intentions, knowledge, self-efficacy, and attitudes about sexual activity and contraceptive use) that are linked to reproductive health behaviors.
The sections below outline the outcomes, related survey questions (full survey questionnaire available here), outcome response format, and survey question source within each outcome domain. In sections located below the evaluation outcomes, we give more detail about the development of the survey items and the development of the evaluation outcome measures.
a – Alphas were determined based on Re:MIX baseline sample.
The survey, administered to students participating in the Re:MIX program and to those in the control condition, assessed students’ demographic characteristics, sexual activity, and contraceptive use. It also included questions about intentions, knowledge, attitudes, and self-efficacy related to sexual activity and contraceptive use—factors that are associated with early and unprotected sex and are targeted by the Re:MIX curriculum.
The study team included survey items that reflected the goals of Re:MIX and were appropriate for the age and background of the adolescents in the study sample. Questions were based on established measures from various sources, including those mandated by the Office of Population Affairs (OPA) (formerly known as the Office of Adolescent Health, or OAH), surveys used to evaluate programs that were part of the federal Adolescent Pregnancy Prevention Approaches (PPA) Study (specifically, Teen PEP and Gender Matters), and existing national surveys that have tested measures with adolescent populations, such as the National Survey of Adolescent Males and the Survey of Young Adults 2009 (also known as The Fog Zone).
Before the study team distributed the survey to Re:MIX evaluation participants, they conducted cognitive interviews over the phone with White, Black, and Latinx students in grades 8 through 10, located in three Texas counties. The goal of these interviews was to ensure that participants easily understood and accurately interpreted the survey items, and to confirm that respondents had the information needed to answer the questions. Survey questions were adapted based on the findings.
To develop the evaluation outcome measures, the study team recoded survey items and often combined several survey items into measures that would capture the underlying constructs of interest. For outcomes that utilized three or more question items, we formed unique constructs using principal component analysis with varimax rotation. We kept items with a factor loading of 0.5 or higher. Internal consistency for created scales (three or more items) was assessed with Cronbach’s alpha. All scales had an alpha of at least 0.7, considered the threshold for internal consistency. For several constructs, we created binary measures based on theory and/or the distribution of our sample. The response distributions and item loadings that informed the study team’s decisions around the final measures were dependent on the Re:MIX evaluation sample—specifically, the full sample with baseline survey data. Future evaluators will want to test these measures against their own samples. All outcomes, apart from intentions to have sex and the behavioral outcomes, were coded such that a higher value is negatively associated with early and unprotected sex.
© Copyright 2023 ChildTrendsPrivacy StatementNewsletter SignupLinkedInYouTube