Statalist


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

Re: st: RE: RE: Covariates in ANOVAs


From   David Airey <david.airey@Vanderbilt.Edu>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: RE: RE: Covariates in ANOVAs
Date   Fri, 15 May 2009 08:38:16 -0500


There are two ways to control age. One is as a continuous covariate where the age of each individual is used (eg ANCOVA). The other is as a blocking variable where age is grouped and variation associated with each age bracket is modeled separate from the error term (eg randomized block). Both of these assumes the age is not changing across the design relative to the other variables.

-Dave

On May 14, 2009, at 11:30 PM, Miss Gina Micke wrote:

Thanks for your input. Maybe I didn't make myself clear enough re "age range". Age range refers to actual age in days at each measurement. As all subjects were measured on the same day at each measurement event, the difference between subject ages is the same at each measurement event, so age at first measurement would be the data in the "age range" variable. Do you still think this should be inlcuded as a categorical variable?
________________________________________
From: owner-statalist@hsphsun2.harvard.edu [owner-statalist@hsphsun2.harvard.edu ] on behalf of Joseph Coveney [jcoveney@bigplanet.com]
Sent: Friday, 15 May 2009 11:48 AM
To: statalist@hsphsun2.harvard.edu
Subject: st: RE: Covariates in ANOVAs

Miss Gina Micke wrote:

I would like to know how to control for an independent continuous variable when using a repeated measures ANOVA. I have a 2x2 factorial design experiment with males and females in each of the treatment groups (T1 & T2) measured at multiple "events", each subject assigned an "id". The subjects are all different ages so
I would like to be able to control for the effect of "age range".

If I didn't have "age range" as a covariate my model would be as follows:

anova x T1 T2 sex T1*T2 T1*sex T2*sex T1*T2*sex/id|T1*event T2*event T1*T2*event sex*event sex*T1*event sex*T2*event T1*T2*sex*event, repeated(event) partial
bse(id|T1*T2*sex)


Do I use the ANCOVA and cont(age range) command? Is this output the effects of
my fixed factors taking into account the effect of age range?

--------------------------------------------------------------------------------

First, the command line text shown doesn't look at all like correct syntax. Second, by the name of it, a variable called "age range" would be categorical and so you would not be using the -continuous()- option for it in - anova-.

Unless you have balanced data (equal representation of age ranges within each factorial cell, as well as equal sex-by-treatment group cell sizes maintained throughout the observation period), try something like the following to start
with:

xi3 i.treatment_group*i.age_group*i.sex*i.observation_interval
xtmixed response _I* || patient_id: , nolrtest variance reml

You'll need to install the user-written command, -xi3-, from SSC if you haven't
already.

Joseph Coveney



*
*   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/

*
*   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/

*
*   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