a logit model with extreme value error

[Patrick Fang]

Hi listers,
recent I find a paper which fits a logit model by assuming an i.i.d. extreme value distribution of the error term. How can I implement such model in Stata?

[Phil Schumm]

Assuming the errors (or an underlying tolerance distribution) are distributed according to the extreme value distribution leads to the complementary log-log link. Such models can be fit with cloglog or with other commands that permit specification of this link (e.g., glm).