Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: Specifying categorical factors & GLMM parameter estimates


From   Richard Goldstein <richgold@ix.netcom.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: Specifying categorical factors & GLMM parameter estimates
Date   Wed, 14 Nov 2007 14:26:20 -0500

If you have more than 2 categories, then you must form the set of variables yourself, either prior to estimating the model (e.g., via tab or via xi) or by using xi at the time of the estimation

I am not clear about what you expect Stata or any other software to do that is special regarding a categorical variable that is ordinal; perhaps you could clarify what you are expecting.

You do not show any output, but from your description it appears that you have let Stata default to treating each right-hand-side variable as continuous.

Rich

Susan Lingle wrote:

Hi All,

I am new (1-day) to STATA and purchased it to run a specific test: a GLMM with a binary response variable (or logistic regression with mixed effects). I have a few specific questions.

(1) Categorical and ordinal variables: I re-named my categorical variables (which will be used as fixed factors in the GLMM) to numbers before pasting them into a STATA dta file. I now defined and assigned data labels to each number, and they appear okay.

But how can I make sure that STATA recognizes these as categorical variables? I see the format of ‘byte” and “%8.0g”, but does that tell me enough? And how can I tell STATA to distinguish between truly discrete variables (Europe, North America, Africa) and ordered variables (e.g., low, medium and high)? I assume there is no problem when there are only two categories (indicative), since there should be no confusion with a continuous or ordinal category. But I have a YEAR variable with 8 years (not to be ordered) and an AREA variable with four categories.

(2) Specific parameter estimates: The GLMM/binary response produces parameter estimates for each variable, but it gives no degrees of freedom (which would help me find out whether STATA is recognizing certain variables as categorical). It also does not produce specific parameter estimates, e.g., t-values with accompanying p-values for each level of a category. I am accustomed to seeing that output in most statistical programs. Advice on how to get this? Or does the absence of this output suggest that STATA ‘thinks’ my variables are continuous rather than categorical?

I have been using the menus (GUI) as much as possible.

Thanks!

Susan
*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   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