Chapter 10: Modeling Nonlinear Longitudinal Change with Mixed Effects Models

Investigations of cognitive and noncognitive developmental processes in education and in the social and behavioral sciences necessitate that individuals be measured repeatedly over time or other analytic conditions. Such longitudinal data are typically comprised of continuous or categorical response scales collected from direct observation, experimental manipulation, self-reports, survey instruments and assessment batteries. The scores from these data sources reflect individual differences in performance (e.g., reading comprehension) or gradations of individual attributes or attitudes (e.g., motivation, self-efficacy). These differences can be marked by dissimilar patterns of change, such as differences in performance levels at different ages, as well as different rates of change as individuals refine their skills, abilities and attitudes. Notably, change is not uniform but rather faster during some periods and slower in others.