Dear Colleagues,

I am using Stata 14.2 to estimate a multinomial probit model of vote choice in the 2016 U.S. presidential election. This model relies on individual-level data from the Cooperative Congressional Election Study. I am using multinomial probit to extend the research of other scholars who make a compelling case for the use of multinomial probit over multinomial logit.

I am having difficulty identifying a way to produce the proportional reduction in error and the percent of estimates correctly predicted following the mprobit model. When using multinomial logit, it seems that the "pre" command provides this information. However, I cannot obtain this same information using "pre" following the mprobit model. The command provides one line of information (usually a large negative percentage value which is obviously not accurate). I have searched for a solution online but have come up short.

Any recommendations as to how I can obtain the proportional reduction in error and the percentage correctly classified statistics following mprobit would be greatly appreciated.

Thank you in advance for your kind assistance.
-Kyle