Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

st: Why no predicted probs but exponentiated linear prediction for -adjust- after -xtgee- or -xtmelogit-?


From   jverkuilen <jverkuilen@gc.cuny.edu>
To   <statalist@hsphsun2.harvard.edu>
Subject   st: Why no predicted probs but exponentiated linear prediction for -adjust- after -xtgee- or -xtmelogit-?
Date   Wed, 17 Dec 2008 22:05:05 -0500

Experimental psych generates reams of repeated measures binomial data (most bad analyzed but that's another story). 

Unsurprisingly I was running -xtmelogit- and -xtgee- today to analyze some repeated measures binomial data and was reminded of one of the oddities of the otherwise most excellent -adjust-, namely the fact that it will happily generate exponentiated linear predictions but not predicted probabilities. This seems odd because the predicted probs are just 

PredProb = ExpLinPred/(1 + ExpLinPred)

and we don't much care about ExpLinPreds but deeply care about PredProbs.  

Yes, I can drive -pred-, etc., but getting colleagues and students to drive Stata on their own can be task enough. -adjust- is a great selling point and students like it... except for this. 

Am I missing something?

JV 

(who wants to drag experimental psychologists away from the hell that is SPSS, and tilt at other windmills while he's at it)

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index