The retrospective construction of metropolitan areas for longitudinal analysis: An application to racial occupational inequality

We analyze a theoretical model of racial occupational inequality using alternative boundary definitions for southern metropolitan areas. These include both "fixed" and "decade-specific" metropolitan boundary definitions. The findings show remarkable consistency among the statistical results across the two types of definitions. Our analysis suggests that defining Standard Metropolitan Statistical Areas according to a specified time point and moving back in time can yield important benefits to the researcher, including larger sample sizes with increased statistical power. Implications for future research are discussed.