Faking it: the influence of insurance-related contingencies on the accuracy of self-reported driving behaviors

Abstract

The validity of self-reported questionnaires has been debated extensively, particularly when respondents are aware of contingencies associated with their responses. In this study, self-report for potential auto insurance purposes was addressed to determine (a) if individuals altered their responses when aware that contingencies such as higher insurance rates were associated with their self-report, and (b) if tests developed under anonymous conditions were still valid when used in motivated contexts. This study involved self-report of driving anger and driving anger expression, as previous research showed that high anger drivers tend to be more aggressive and risky drivers. Insurance companies could be interested in this information to assess risk. Three hundred students in an introductory psychology class completed questionnaires on driving anger, driving anger expression, trait anger, impulsivity, and speech anxiety under three conditions. The first was an anonymous self-assessment of these variables. The second asked students to complete the questionnaires as if submitting them to their automobile insurance company, assuming the information would be used to set premium rates. The third resembled the second, but students were informed of the insurance company instructions after describing their driving behaviors, but before describing their anger, anger expression, and impulsivity, (i.e., psychological variables which cannot be externally validated). These conditions are designed to assess the potential impact of insurance contingencies on both criterion and psychological variables. Generally, no condition effect was found, as only three of the analyses involved conditions. The insurance later respondents reported less driving anger and fewer major accidents than those in other conditions. Also, condition interacted with gender to predict differing patterns of self-report of speaking anxiety. While men's anxiety increased over insurance conditions, women's decreased. Potential differences in correlations as a function of condition were tested by comparing the strengths of independent correlations. No meaningful differences in correlations between variables were found as a function of condition. Data were collapsed across conditions. Correlations among driving anger, driving anger expression and trait anger reflected previous research. Potential reasons for lack of finding include actual lack of condition impact, possible lack of understanding of the impact of increased insurance rates, and lack of understanding of instructions.