UCLA

Over the past few decades, college student survey response rates have been declining. This is a problematic trend because student survey data are used extensively in endeavors such as accreditation, institutional improvement, and scholarly research. While low survey response rates are not necessarily a problem, they will be if they impact the representativeness of survey samples. Unfortunately, the limited literature on student survey nonresponse suggests that nonresponse is usually not random, though for college students little is known about the type of student, institutional, or administrative characteristics that promote student survey response.

The purpose of this study was to examine predictors of college student survey response, in a comprehensive model that takes into account both student and institutional factors. Drawing on sociological, organizational, and psychological theories, a conceptual model of student- and institution-level influences on survey response was developed and tested using national longitudinal surveys administered by the Higher Education Research Institute (HERI) to first-time, full-time students enrolling at four-year institutions in the falls of 2003, 2004 and 2005. The study utilized hierarchical generalized linear modeling (HGLM) to examine predictors of longitudinal survey nonresponse one and four years after matriculation, for all students as well as for groups disaggregated by gender and self-identified race/ethnicity (White, Black/African American, Latino/a, and Asian American).

Results revealed that a key group of response predictors was consistent across aggregated and disaggregated groups of students, one and four years after college entry. For virtually all students, a small set of student-level characteristics (most notably high school achievement, gender, personality, and self-rated likelihood of transfer) strongly predicted response propensities, indicating that students' entering characteristics have an enduring impact on their survey response likelihoods over the entire course of college. Institution-level results revealed that students were far less likely to respond to web surveys and mail surveys than they were to paper surveys handed out in person; survey incentives showed mixed effects. Institutional size was a consistent predictor across all students and surveys, while institutional survey climate significantly impacted response propensities for seniors only. Findings are discussed in terms of their implications for both researchers and practitioners.