From a methodological and sociological perspective, analyzing social change is best done by using repeated cross-sectional data and by including individual level variables, time, and time-dependent macro variables. Furthermore, interest often focuses on whether the effects of explanatory variables change over time, e.g. the impact of education on one’s social position. We argue that multilevel models with time as context best meet these requirements. However, the methods for applying a multilevel approach to study time-dependent social processes are relatively poorly developed and therefore rarely used. Instead, most social change related analyses of repeated cross-sectional data, such as APC models or changing effect models, include time as an individual level variable and thereby neglect the contextual characteristics of time and changes at the macro level. In contrast, pooled time series analysis does not take account of individuals and micro level effects. Thus, the aim of this paper is to overcome these shortcomings by adopting a new approach to social change studies. We describe the application of multilevel analysis with years as contextual units on a step-by-step basis and emphasize the substantive advantages of each model for the study of social change.
展开▼
