Mode Effects in Self-Administered Surveys Among Pregnant and Parenting Adolescents
Derecho, A. A., Head, B. F., Kan, M. L., Ashley, O. S., Williams, C., & Jones, S. B. (2011, May). Mode Effects in Self-Administered Surveys Among Pregnant and Parenting Adolescents. Presented at AAPOR 2011, Phoenix, AZ.
Asking survey respondents questions about sensitive topics may elicit socially desirable answers. Mode of administration has been shown to affect the magnitude of social desirability bias. Comparisons of computer-assisted self-interviewing (CASI) and self-administered questionnaires (SAQs) among general adolescent samples have produced modest evidence of mode effects, with more accurate reports on sensitive items (e.g., alcohol or other drug use) when the mode of administration is CASI. Few studies have tested for mode effects among pregnant and parenting adolescents. We collected baseline data from a convenience sample of 1,036 pregnant and parenting adolescents enrolled in a cross-site evaluation of the Adolescent Family Life Program, which involved projects in 10 states. The purpose of this study is to examine mode effects among adolescents who completed the survey using CASI (32%) and those who completed using SAQs (68%) on responses to questions about contraceptive use, sexually transmitted disease (STD) prevention, number of pregnancies, desire to have another baby before finishing high school, marriage attitudes, educational aspirations, communication with their parents, and avoiding pressure from others. We explored associations between data collection mode and responses to these questions using multiple linear and logistic regressions, controlling for project site and other potential confounders (e.g., demographic and socioeconomic variables). Results showed that data collection mode was not associated with any of the items examined. These findings suggest that among pregnant and parenting adolescents, responses to SAQs and CASI are generally comparable. When planning a survey within this population, mode may be less important to consider than other factors, such as capacity or budget. Future research should examine other potentially sensitive items relevant to this population to determine whether mode effects exist.