Using self-reported data to assess the validity of driving simulation data.

In this article, we use self-reported driving behaviors from a written questionnaire to assess the measurement validity of data derived from a driving simulation. The issue of validity concerns the extent to which measures from the experimental context map onto constructs of interest. Following a description of the experimental methods and setting, an argument for the face validity of the data is advanced. Convergent validity was assessed by regressing behaviors observed in thedriving simulatoron self-reported measures of driving behaviors. Significant relationships were found across six measures: accidents, speeding, velocity, passing, weaving between traffic, and behavior at stop signs. Concurrent validity was evaluated with an analysis of simulator accident involvement and attention deficit hyperactivity disorder status. Discriminant validity was assessed using a multitrait-multimethod matrix of simulator and questionnaire data. We concluded that although the relationship between self-reported behaviors and observed responses in the simulator falls short of perfect correspondence, the data collected from the driving simulator are valid measures of the behaviors of interest.