Question

Suppose that the options for 18-year-old students are (Go to college, Fisherman, Hang out at their parents' place). Suppose you estimate the Conditional Logit model to understand individual preferences. Now, suppose that a 2 year ban on fishing makes the choice to become a fisherman no longer feasible. The model implies that you can still predict the ratio of the probability of going to college over the probability of hanging out at their parents' place. What would be this prediction? Do you find the prediction of this model credible? Carefully motivate your answer.

Answer 1

Fishing is banned. In this circumstances, previous conditional logit model may be used to predict the ratio of the probability of going to college over the probability of hanging out at their parents' place. But this is erronomous as the true ratio may differ in general.

The prediction model is not at all credible. Chance of selecting a preferred choice is clearly dependent on availability of options. As the set of available options differs here, use of previous model cannot be credible. However, it would be acceptable only when the students who would like to choose fishing choose other two in the same ratio as that of going to college and hanging out at their parents' place earlier.