Many questions in social research must be evaluated over time. For example, in studies of intragenerational mobility, measuring opportunity for economic advancement requires longitudinal data. The authors develop and use a class of hybrid functional models to demonstrate how different models can lead to extremely different substantive conclusions. They provide guidelines for longitudinal data analyses in which variance partitions are central to the inquiry. In their analysis of the National Longitudinal Survey of Youth, the authors conclude that in a period of rising wagedispersion, the bulk of inequality is persistent over the life course. Their models provide support for the scenario in which wage inequality rises steadily while instability slowly diminishes over time. They obtain mild evidence of increased wage instability for somewhat older workers in the early 1990s, matching a recessionary trend. These findings contribute significantly to understandingwage inequality in United States over the past 25 years.
Key Words: covariance structure • variance components • functional data analysis • wage inequality • National Longitudinal Survey