Statalist


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

RE: 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   RE: st: Why no predicted probs but exponentiated linear prediction for -adjust- after -xtgee- or -xtmelogit-?
Date   Thu, 18 Dec 2008 09:51:54 -0500

"Maarten buis" wrote:

>Yes, we can/do care about ExpLinPreds, these are the odds. Odds and
probabilities are just alternative measures of the same thing. The one
represents the likelihood of an event as the expected number of
successes per 100 (the probability*100), while the other as the
expected number of successes per failure. >


Sloppy wording on my part: *I* am aware of odds but it isn't an outcome measure most of my colleagues or students care about.  

APA publication guidelines push CIs. One of Stata's real hit-you-between-the-eyes features is its ability to make tables of CIs for a model like logistic regression quickly and easily based on arbitrary combinations of variables that don't even need to be in the model. But when you have just got done convincing someone that the program does what they need... except this one last little step where the payoff comes, it is a bit of a letdown. 

It would be awesome if it plotted the table of predictions and CIs too. People like the graphs I show them, but are less impressed by the work they would have to do to get them (they don't know about all the underlying numerics needed to fit, say, a mixed logit model).  

-adjust- is close to being a killer app is all I'm saying, as a large chunk of labor is spent grinding thru making graphs with CIs. 

(To answer the obvious followup: Yes I could work on something but it wouldn't be official with snazzy dialog box. More crucially, I don't believe my employer would care like they do about the three Psychometrika articles I am working on.) 



*
*   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