Estimating Macro-Relationships Using Micro-Data

Abstract

Widespread recognition that many sociological processes occur at multiple levels has led researchers to search for valid forms of cross-level modeling such as contextual analysis. In this article, we propose an alternative cross-level method for examining the effects of macro-factors on macro-processes while controlling for processes at a lower-level of aggregation. This method applies existing statistical techniques but in a new fashion. It has the advantages of more precise control for lower-order effects, reduction of problems with degrees of freedom, and more error-free prediction of the higher-level relationships than using alternative estimation strategies. As a result, parameter estimates of the macro-level effects are more efficient. An empirical example studying macro-determinants of unemployment rates in states in the United States demonstrates that estimation of this model leads to more valid conclusions than more commonly used methods.