|
![](http://www.statmodel.com/discussion/icons/csc/1px.gif) |
Logistic model: bayes vs other estima... |
![Previous](http://www.statmodel.com/discussion/icons/csc/navgen_arrow_left.gif) ![](http://www.statmodel.com/discussion/icons/csc/1px.gif)
|
Message/Author |
|
|
Using version 7.4, I have run Example 3.5 in the User's Manual 3 ways: (a) as provided (U1 is specified categorical), (b) as mixture model, and (c) with U1 specified categorical but using Bayes estimator. Estimates for methods (a) & (b) match, but (c) has different estimates. Forgive what is probably a naïve question, but I have been unable to find information about why (c) would differ from (a) & (b), and how the (c) estimates could be interpreted (e.g.: OR for a & b are easily understood). Reference or brief explanation appreciated. Thanks... |
|
|
UG ex 3.5 uses ML and the default of logit link. Bayes uses probit. |
|
|
Thanks, Bengt - figured it was something simple I was missing... Michael |
|
Back to top |
|
|