Key points are not available for this paper at this time.
Accelerated longitudinal designs enable researchers to study individual development over a long interval of the life course by gathering data during a comparatively short interval of time. Such designs also create possibilities not available in standard panel designs for separating developmental effects from cohort and period effects. However, these designs confront the investigator with a special set of inferential challenges and introduce complexity into statistical analysis. In this article the authors employ a hierarchical linear model to illustrate the application of growth curve analysis to data from an accelerated longitudinal design. The goal is to construct a picture of the development of attitudes toward deviance from ages 11 to 18 by linking data from two cohorts of the National Youth Survey, each observed for only 5 years. The example illustrates how the analyst may control for time-varying and time-invariant covariates and test for cohort effects and cohort-by-age interactions. Interesting features of growth include an inflection point (age at which the rate of increase in prodeviant attitude begins to slow down) and a peak age (age of maximally prodeviant attitude).
Raudenbush et al. (Sun,) studied this question.