another shortcoming of typical transport modeling i just realized while reading a paper by anable (2005). the four step model uses basic demographics (income, household size, vehicle ownership etc) during trip generation and mode choice. the assumption is that individuals equivalent on these metrics will have identical trip rates and mode choices. this is completely absurd. why do the models not account for differences in personalities (as assessed by attitudes). what would be the difficulty with sticking an additional 25-50 question battery on a travel survey to complexify the model. further, what are the cost tradeoffs with this approach vs. moving to more advanced models. to what extent can the 4sm be "saved" by marginal increases in complexity?