ABSTRACT
Numerous studies have examined the measurement problems associated with the use of ratio variables in aggregate level research, but have seldom considered the substantive implications of using a particular deflator in lieu of another. The present study questions the conventional wisdom of employing the total population size of social units to deflate raw counts of crime. The OLS regression analyses of a sample of 105 U.S. cities reveal that the effects of structural predictors vary dramatically across alternative rate measures of burglary and motor vehicle theft. The implications of these findings for macro social criminological research are discussed.