ApolloChoiceModelling forum

Dear Prof. Stephane,

I have a question concerning the difference in the marginal rate of substitution values for the Multinomial Logit (MNL) and Mixed Logit models. Specifically, I am calculating 'Value of Risk Reduction,' which is the additional travel time food delivery riders are willing to undertake for a 10% reduction in the risk of a crash.

For the Mixed Logit model, I have assumed a log-normal distribution for both parameters (positive for risk reduction and negative for time). Here are the values I have obtained for both models:

Value of Risk Reduction (MNL_mean): 1.07 (90% CI: 0.28 - 1.85)
Value of Risk Reduction (Mixed_mean): 7.9 (90% CI: 1.64 - 14.1) - with a standard deviation of 53

While it is evident that the Mixed Logit model provides a better fit, I want to know what explains the difference in the values obtained (almost 7 times higher in the case of the Mixed Logit model). Could you please provide insight on this?

Thank you for your time and assistance.

Best regards,
K.

Hi

apologies for the slow reply

yes, this difference looks a bit too large to explain. Did you also look at the medians to see if they are reasonable (though we would want the means of course).

You might want to consider other distributions too with shorter tails

Stephane