Population size, change, and crime in US cities

The sometimes noted contradiction between cross-sectional and longitudinal relationships concerning city population size and crime rates is reexamined using more complex analytic procedures, controlling for extraneous variables, and allowing for non-monotonic relationships. Instead of a simple cross-sectional relationship between population size and crime rates, the more sophisticated analysis reveals either no association or a quadratic relationship. Similarly, instead of a simple lack of longitudinal relationship or a negative one, the more complicated analysis shows a non-monotonic pattern for three of six offenses. However, we contend that these divergent patterns for cross-sectional relative to longitudinal data are not necessarily indicative of an "anomaly." Instead, they represent different aspects of a dynamic process in need of more extensive theorizing. Finally, the cross-sectional results showing that city size and crime rates are either not linked or when linked are in a non-monotonic pattern call into question one of the accepted relationships in criminology that have long guided thinking about crime.