ABSTRACT
This chapter investigates the validity of some of self-reported variables, notably crashes. All self-reported variables therefore need to be validated, and the best method is comparison to other, more reliable sources, if such exist, thereby at least achieving a measure of the error involved in the data thus gathered. When self-reported citations have been compared with state records, the ensuing moderate correlations have, if commented upon, been explained by the same kind of 'logic' as presented for accidents; the error is assumed to lie with the records, because more violations are reported by the subjects. Use of self-reported accidents raises the possibility that a 'dishonesty' bias might underlie associations found with attitude or personality measures. The technique of placing the responsibility elsewhere seems to indicate that even some researchers using self-reported data, mainly accidents, feel a certain awkwardness about this method.
