An examination of the influence of institutional context on persistence at four-year colleges and universities: A multilevel approach.

다국어 초록 (Multilingual Abstract)

This study contributes to our understanding of the most appropriate theoretical and methodological approaches to examining differences between institutions in college student persistence. Using constructs from Bean's (1990) student attrition model and the Berger-Milem (2000) college impact model, national data from the Beginning Postsecondary Students (BPS) and the Integrated Postsecondary Education Data System (IPEDS) surveys, and hierarchical generalized linear modeling (HGLM) techniques, this study addresses the limitations of prior research by exploring the extent to which persistence is influenced by institution-level variables after taking student-level variables into account. The analytical sample is limited to first-time, full-time, degree-seeking undergraduates attending four-year colleges and universities nationwide.
Five conclusions may be drawn from this study. First, several student-level constructs from Bean's (1990) student attrition model help to explain persistence within a four-year college or university. This study supports Bean's (1990) claims that student academic background and such measures of experiences as college academic performance, involvement, satisfaction, and institutional commitment positively influence persistence. Aspects of environmental pull, specifically a student's plan to transfer and working off campus, are validated as negative predictors of persistence.
Second, even after controlling for student-level variables, differences in persistence between four-year colleges and universities exist. Third, this study finds that the two student-level constructs from the student attrition model, satisfaction and plan to transfer, that are aggregated to the institutional level reflect student-level rather than institutional-level effects. Fourth, this study finds that selectivity, as measured by the average student academic ability at an institution, has a contextual effect on college student persistence that reflects a positive increment to the chance of persistence that accrues to a student beyond a student's individual academic ability as a result of attending a more, rather than a less, selective institution. Fifth, this study demonstrates that multilevel statistical techniques can be used to examine persistence utilizing the student and the institution as units of analysis and simultaneously take into account the effects of variables that operate within an institution and between institutions to influence persistence.