1

Most textbooks that I've read mention the nonsensical results that you might obtain with an OLS estimation of a probability of a variable being 0 or 1 -- as such, you use the transformations of the probit or the logit model. But are there any situations where, for instance, you're limited by the data and such results might come in handy?

John Doe
  • 1,275
  • 2
  • 15
  • 24
  • Could you explain what you mean by "limited by the data"? – whuber Mar 21 '16 at 22:56
  • 1
    I asked another question regarding a problem that I've been facing: http://stats.stackexchange.com/questions/202896/ols-and-probit-model-regarding-collinearity. I keep getting a dummy variable being dropped in a probit estimation, but the OLS estimation does not drop it. – John Doe Mar 21 '16 at 23:00

0 Answers0