Improving choice model parameter estimates by jointly modelling the SP choices with corresponding elicited certainty ratings

The paper proposes a closed-form econometric model of joint SP choices and corresponding elicited certainty ratings. The connection between the SP choices and corresponding elicited certainty ratings is modelled through an entropy-based measure of SP choice task complexity. Empirical applications of the proposed model are presented by using two SP survey datasets collected in Vancouver and Toronto. Empirical models reveal that the SP choice task complexity, measured through SP choice entropy, directly influences SP certainty ratings. Such direct relationship proves to be important through establishing an endogenous relationship between them. It is clear that capturing such endogeneity improves the efficiency of parameter estimates of the SP choice model. However, the level and extent of such benefit gains vary by the nature and complexity of the SP survey. Empirical investigation presented in the paper proves that both efficiency gain and higher goodness-of-fit are probable with the choice contexts with higher number of choice alternatives. However, even for the smaller number of choice alternative case, efficiency in parameter estimates can be increased by proposed joint model formulation.