773990_en-US

liliconcom

Nominal Logistic vs GenReg w/ Binomial Distribution

When you fit a logistic model in Fit Model, you do not get confidence intervals on the modeled probability.  When you fit a logistic model with GenReg, you do get confidence intervals.

So this question is why does the Logistic Model option in Fit Model not have confidence levels but GenReg does?  And is it possible to get confidence levels of logistic models in JMP without having access to JMP Pro?

Logistic Model

GenReg - Logistic Model

 

Re: Nominal Logistic vs GenReg w/ Binomial Distribution

No, unfortunately, you cannot get the confidence limits for predicted probability in the standard JMP. To calculate CLs with a formular you’d need the estimated standard error of the linear predictor, or at least the estimated covariance matrix of the parameter estimates, neither of which is reported.

As for why?  Back in the old days, logistic regression, like the standard least squares, were primarily used for explanatory modeling focusing on the coefficient estimates.

Re: Nominal Logistic vs GenReg w/ Binomial Distribution

Thank you for confirming Jian.  It would be nice to have the CI's in Logistic Regression, but at least I know I'm not missing something here.