An Empirical Investigation of Biased Survey Data and an Attempted Cure

In this paper I investigate response bias in survey data on annual driving mileage and evaluate the performance of a proposed remedy, Orbit. Individuals systematically exaggerate their deviation from the sample average, and using the self-reported data leads to misleading estimates of the income elasticity of travel mileage. I extend the Orbit procedure, which is designed to correct for reporting bias, to allow misreporting at the lower censoring point. Orbit fails to detect the nature of the bias and distorts the income elasticity estimate even further. The message for practitioners using biased data is therefore a cautionary one.