Addressing Ethnic Differences in the Validity of Self-reported Criminal Behaviour Through a Social Desirability Measure

Abstract Objectives Self-reported criminal behaviour has the potential to provide clearer insights into patterns of criminality compared to using police statistics. The risk of inaccurate responses however forms a major obstacle to its validity. This study therefore examines underreporting bias of self-reported criminal behaviour among five ethnic groups and compares different methods to facilitate the creation of valid intergroup comparisons. Methods This study includes data from the Monitor on Youthful Delinquency (N = 6,218) which was connected to police suspect registrations. To identify patterns of underreporting, we compared self-reported and police recorded crime with a social desirability measure, which was adjusted to be invariant across ethnic groups. Three different methods to correct for underreporting bias were subsequently compared; partialling out the effect of social desirability, listwise deletion, and a novel technique which we named Social Desirability based Score Replacement (SDSR). Results The study reveals that police suspects with a high social desirability score display a low likelihood to self-report crime when they have an ethnic minority background, but not when they have a native Dutch background or when they have a moderate to low social desirability score. This finding points towards systematic differences in underreporting bias. Model outcomes are shown to be significantly impacted depending on the method that is used to address this issue. Conclusion Neglecting to correct underreporting-bias hinders the validity of intergroup comparisons of self-reported criminal behaviour. The inclusion of a social desirability measure is therefore recommended to help identify and correct underreporting bias, particularly through the use of SDSR.