ABSTRACT
Repeated measures data generated from longitudinal designs are often used when studying correlates of individual change. Such studies pose several challenges: (1) The measurement scale must be invariant over time; (2) covariates of interest are often multilevel (e.g., measured at the person and neighborhood level); (3) some item-level missing data can be expected. To cope with these challenges, we propose a repeated measures, multilevel Rasch model with random effects. Under assumptions of conditional independence, additivity, and measurement invariance over time, the approach enables the investigator to calibrate the items and persons on an interval scale, incorporate covariates at each level, and accommodate data missing at random. Using data on eight items tapping violent crime from 2,842 adolescents ages 9 to 18 nested within 196 census tracts in Chicago, we illustrate how to test key assumptions, how to adjust the model in light of diagnostic analyses, and how to interpret parameter estimates.
