Covariance Function Regressions for Studying Culture: An Application to Racial Differences in Fertility

• In Event: 204 - Regular Session. Quantitative Methodology

Abstract

Renewed interest among sociologists and demographers in cultural explanations of inequality has re-opened long-standing debates about cultural essentialism and the connection between culture and action. In this paper we introduce a method that links culture to demographic outcomes while avoiding essentialist assumptions. We use this method to study fertility differences between African Americans and whites. We argue that between extreme characterizations of culture--as essential, coherent group properties versus particularistic individual perceptions or beliefs--lies a continuum that can be described quantitatively. We introduce a covariance function regression model to assess the extent to which groups' beliefs differ from each other in three ways: on average, in the amount of disagreement among group members, and in the tightness of connections between multiple beliefs and behaviors. Drawing on General Social Survey data from 1972 to 2010, we use this model to explore racial differences in fertility ideals and behavior. Our results suggest that African Americans' ideal and actual fertility are both higher than whites' on average but are also much more variable. The correlation between ideal and actual fertility is also much stronger among African Americans than it is among whites. Supporting neither essentialized between-group explanations based on fully cohesive notions of culture nor entirely particularistic within-group explanations based on individual traits, our results support a middle ground wherein differences between and within racial groups contribute to differences in fertility.

Authors

• Deirdre Bloome, University of Michigan

• Christopher Michael Muller, Columbia University

• Daniel Schrage, Harvard University