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

Re: st: age-adjusted means

From   "Mona Mowafi" <[email protected]>
To   <[email protected]>
Subject   Re: st: age-adjusted means
Date   Mon, 25 Aug 2008 09:02:07 -0400

Thank you for your response, Maarten.  I forgot to mention, I do have the continuous BMI variable, so I would not have to assign values separately.  But I cannot do an ANOVA and then get the means b/c I violate ANOVA/regress assumptions of normality..  I am getting predicted probabilities by using multinomial for a separate table, but I thought a simple table of age-adjusted means would be helpful as well..

I will look up the commands you spoke of, but just wanted to throw that out there in case these commands don't work for this purpose..

Many thanks,

>>> Maarten buis <[email protected]> 8/25/2008 2:07 AM >>>
--- Mona Mowafi <[email protected]> wrote:
> I need to get the age-adjusted means of my outcome (3 categories of
> BMI: normal weight, overweight, obese) on each of my independent
> variables (indicators of SES), but my data is very non-normally
> distributed (over 50% of sample is obese) and I am conducting
> multinomial regression in my analyses.

If you want a mean than you will have to assign numeric values to each
category of your depedent variable, and those numeric values have to
mean something: For instance in case of a variable religion you can
give the value 1 to christian, 2 to muslim, and 3 to other, and than
compute the mean of the variable religion, but that does not mean
anything. Similarly, you can assign 1 to normal weight, 2 to
overweight, and 3 to obese, but what does that mean? Probabily a bit
more than the religion example (as this ordering corresponds with the
natural ordering of that variable), but not much: Why would the
distance between normal and overweight be exactly the same as the
distance between overweight and obese?

If you are using multinomial regression the natural thing to look at is
not the mean of the outcome, but the probabilities of ending up in each
category of the outcome. A very helpful tool is the -prgen- command
which is part of the -spost-, which you can download by typing 
-findit spost-. This command is discussed in the following Stata
Journal article:

Jun Xu and J. Scott Long (2005) "Confidence intervals for predicted
outcomes in regression models for categorical outcomes" The Stata
Journal, 5(4): 537--559. 

Hope this helps,

Maarten L. Buis
Department of Social Research Methodology
Vrije Universiteit Amsterdam
Boelelaan 1081
1081 HV Amsterdam
The Netherlands

visiting address:
Buitenveldertselaan 3 (Metropolitan), room Z434

+31 20 5986715 

Send instant messages to your online friends 
*   For searches and help try:
*   For searches and help try:

© Copyright 1996–2024 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index