A Test of Competing Hypotheses: Using Boost and Flag Variables to Predict Crime

• In Event: Strategic Analysis and Prediction of Spatial and Temporal Crime Patterns

Abstract

Two theoretical frames are used to predict future crime patterns. The first, commonly known as the flag hypothesis, suggests socio-structural factors of communities generate crime risk that is relatively stable over time. The second, known as the boost hypothesis, suggests recent crime events generate crime risk that changes from one day to the next. No one has examined whether or not these two theoretical frames carry equal weight in explaining crime patterns. It is possible that long-term indicators of crime carry more predictive weight compared to short-term indicators, or vice versa. Using data from Philadelphia, PA, crime risk is modeled using long-term and short-term crime indicators as suggested by the flag and boost hypotheses. Model comparisons will inform the relative importance of boost and flag variables. The results of this study will enhance our understanding of what drives crime patterns.

Authors

• Amber Perenzin, Temple University

• Jerry Ratcliffe, Temple University

• Ralph Brecken Taylor, Temple University