Statalist


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

st: basic repeated measures question


From   "Airey, David C" <david.airey@Vanderbilt.Edu>
To   "statalist@hsphsun2.harvard.edu" <statalist@hsphsun2.harvard.edu>
Subject   st: basic repeated measures question
Date   Tue, 17 Nov 2009 21:43:13 -0600

> I have 3 questions regarding performing an ANOVA and 1 regarding doing
> a repeated measures logistic regression. I have tried to find examples
> for my questions but been unsuccessful.

You have lots of resources online that could help you:

http://www.ats.ucla.edu/stat/stata/faq/xtmixed.htm

http://www.stata.com/support/faqs/stat/#anova

> 
> First, my data is setup up like this --
> 
> ID     grp     outcome     time
> 1         1            15             1
> 1         1            16             2
> 2         2            12             1
> 2         2            20             2
> 3         3            20             1
> 3         3            12             2
> 4         4            20             1
> 4         4            22             2


If your data are really set up like this then ID and grp are causing the conflict, because you have one person in each group. Is that the case? And if you also really have just two time points, a paired ttest would work.

> 
> 
> 1. I am interested in whether the change in the continuous outcome
> over time significantly differs by group.  How would I setup this
> problem in an ANOVA? I would like to obatin the effect of time, group,
> and timeXgroup. I tried using repeated measures ANOVA code and this
> did not work (refer below)
> 
> I wrote -- anova outcome id time grp, repeated(time)
> Received this error msg --  could not determine between-subject basic
> unit; use bseunit() option
> r(422);
> 
> 2. If timeXgroup is significant, how do I perform simple effects
> analysis post hoc in order to understand how the outcome varied by
> each group over time and get a graph showing the trends
> 
> 3. How would I include covariates in the model above essentially?
> 
> 4. If my outcome was binary rather than continuous, I was interested
> in performing xtgee. How would you suggest setting up this code? I
> tried to set it up as below, but I did not get the OR for all groups
> (one of them was omitted). I believe I am specifying the model
> incorrectly.
> 
> xtset id
> xi: xtgee outcome i.grp, family(binomial) link(logit) eform
> 
> 

*
*   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