Importance of Areal Aggregation: The Case of Violence Against the Police Research

• In Event: Advances in Statistical Methods

Abstract

Some debate exists about the appropriate level of aggregation for analyses. On the one hand, smaller levels of aggregation (e.g., cities v. states) allow for a better understanding of an issue and for specific policy recommendations. However, data availability may become more limited as units become smaller, posing a problem for researchers. With rare outcomes, smaller aggregations lead to more cases with zero counts, adding a layer of analytical hurdles. The question remains, for criminal justice research, how much does level of aggregation matter?

We explore this question using the case of violence against the police as the outcome measure, specifically chosen because of the rarity of this type of violence. The number of incidents where Baltimore police officers were injured or assaulted with firearms in the line of duty was compared with theoretical indicators drawn from census data: concentrated disadvantage, residential mobility, immigration concentration and racial diversity. We compare the results at three levels of analysis: the census tract, the block group, and the neighborhood level. Findings and implications for research are discussed.

Authors

• Jennifer Gibbs, Penn State Harrisburg

• Jonathan Lee, Penn State Harrisburg