Accounting for differences in modelled estimates of RP, SP and RP/SP direct petrol price elasticities for car mode choice: A warning

Abstract

Using a sample of 245 direct petrol price elasticities of car travel collected from 52 published mode choice studies, a random coefficient regression model is estimated to account for heterogeneity in the influence of the type of data used in the various studies (RP, SP and a combination of RP/SP). The focus on the type of data is designed to highlight a concern that has emerged in the way that an increasing number of researchers and consultants derive elasticities from stand-alone stated preference studies, and apply them. It is well known that this is not valid without model calibration (usually via the mode-specific constants), since the elasticity formula uses the probability of an alternative being chosen. To understand the extent of possible behavioural response bias when calibration is ignored is important, signalling a possible adjustment process to correct for the absence of calibration relative to calibrated RP and or RP/SP derived elasticities.