A Dynamic Model of School Effects on Students' Academic Achievement

Andrew Halpern-Manners, University of Minnesota

Prior research on contextual effects nearly always measures “context” at a single point in time (e.g., children's school conditions in grade 10 or their family socioeconomic status at age 16). In this paper, I argue that this measurement strategy fails to account for (1) age-based variation in children's sensitivity to their surroundings; (2) differential effects stemming from differences in the length of children's exposures; and (3) moves between contexts and changes over time within them. To evaluate the implications of this argument, I specify and test a more dynamic model of school effects on young people's academic performance. Drawing on nationally-representative longitudinal data and recent advances in growth mixture modeling, I identify a series of qualitatively distinct trajectories of school exposure that extend across a substantial portion of respondents' childhood and adolescent years. I then use these trajectories as predictors in models of math and reading achievement.