The Modifiable Areal Unit Problem in Person–Context Research

Abstract

Objectives: To examine whether the modifiable areal unit problem (MAUP) observed in studies of aggregate crime rates poses a threat to studies examining the moderating effect of neighborhood processes on the association between individual risk factors and self-reported violence. Methods: A series of negative binomial regression models were estimated on a sample of 11,288 youth participating in the Add Health study to assess the moderating influence of neighborhood disadvantage on the association between impulsivity and violence aggregated to the census block group and the census tract. Results: One-unit increases in neighborhood disadvantage and impulsivity were associated with 5.1 percent and 4.9 percent increases in the expected count of self-reported violent behaviors, respectively. Neighborhood disadvantage exacerbated the association between impulsivity and self-reported violence when measured at the tract level but had no moderating effect when measured at the block group. Conclusions: The findings reported here suggest that the MAUP poses a unique concern for person–context research. The geographic aggregation of neighborhood effects warrants careful consideration in theoretical and empirical models linking individual behavior with broader social ecologies.