Community views on crime and policing: Survey mode effects on bias in community surveys*

Abstract

This study compared two modes of administration, mail and random-digit dialing (RDD), in surveys targeting four diverse community areas within a large metropolitan area. The modes' differential susceptibility to nonresponse bias and response bias is revealed by directly comparing the two modes and procedures. The authors conclude that RDD phone community surveys on attitudes toward the police are likely to overestimate the population's true level of approval, whereas mail surveys are likely to underestimate it. Researchers should consider the mode of data collection when examining relationships and building models that explain community reactions to policing. Policy makers are cautioned to be aware that the survey mode may have a strong impact on estimates of prevalence, especially in neighborhoods with higher crime and lower income.