Evaluation of endogeneity due to the omission of psychosocial variables in mode choice

Endogeneity is a frequent anomaly in discrete choice models, leading to inconsistent parameter estimates that are used to obtain economic indicators, such as the subjective value of travel time (SVT) and elasticities in mode choice models. While the omission of variables has been identified as one of the most relevant sources of endogeneity for mode choice, the omission of multiple psychosocial variables has received limited attention thus far. This paper proposes two methods to evaluate endogeneity arising from omitted psychosocial constructs: the Multiple Indicator Solution (MIS) and the Integrated Choice Latent Variable method (ICLV). The analysis uses travel survey data from the cities of Concepción and Temuco, Chile, where three psychosocial constructs—affection, attitude, and habit—were measured and incorporated into different utility function specifications. The results indicate that omitting psychosocial variables from transport choice models leads to lower SVT estimates compared to models that explicitly include these factors. Travel time elasticities decrease for bicycle and car when applying the proposed methodologies but increase for bus. Cost elasticity decreases for car and increases for bus once psychological factors are accounted for. These findings suggest that infrastructure project appraisals may undervalue SVT and misestimate elasticities if psychosocial variables are omitted, with important implications for transport policy and investment decisions.