| Abstract: | Different operational definitions of neighborhood boundaries could generate differences in empirical relationships between ecological dynamics and crime. As a call for research on whether aggregation bias is an issue for related studies, this note compares bi‐level models predicting re‐arrest for domestic violence with census tracts versus officially defined city neighborhoods as macro units. Results indicate that both types of units generate the same substantive findings at the aggregate and individual levels of analysis. Similarities also exist between these results and those from previous multilevel studies of crime and victimization despite differences in the aggregate units across studies (including block groups). |
