A second limitation stems from our indirect measurement of exposure to crime-prone others. Drawing on respondent-generated measures of criminal associations (i.e. perceived peer delinquency) results in contaminated estimates of the delinquent peer effect. Due to false consensus and projection effects, respondents' perceptions of their associates' behaviors are systematically biased in the direction of their own behaviors, which leads to an inflation of the correlation between peer and personal delinquency.