Identifying user preferences for cycling infrastructure design using latent class discrete choice models

Abstract

During the past years, the use of bicycles to commute in Santiago has increased heavily along with new cyclingfacilities. Nevertheless, some of these facilities are underutilized due to the fact that they are poorly designed, notconsidering the needs and preferences of its potential users. This is why there is a need to understand these cyclists’preferences to guide future policy. This study analyzes a stated preferences survey carried out by Saud (2014) tocalibrate a multinomial logit model and one with latent classes. Both models showed similar results, with the latentclasses one providing richer information and a better fit to the observed results. Nevertheless, some respondentsampling errors were found, creating a bias in the results that do not fit with what has been described earlier inother studies. Considering the results obtained, some policy recommendations are given, along with future researchguidelines.