RE: st: Unbalanced repeated measures analysis question

["Ploutz-Snyder, Robert (JSC-SK)[USRA]" <robert.ploutz-snyder-1@nasa.gov>]

Karin,

You could define your model the way you suggested, yes, however mixed models can be specified a number of different ways depending on your research goals and how you want to consider the nesting of your repeated measures factors (i.e. random terms).  

There are a number of excellent books on this type of analysis, going by names including mixed-effects modeling, mixed modeling, higher level modeling (HLM), multi-level modeling (MLM) and probably a few other terms... If you are interested in a more Applied book that uses Stata in particular, Rabe-Hesketh and Skrondal put together a nice one book called Multilevel and Longitudinal Modeling Using Stata.  I think you might do well to take a course in MLM if you can to at least wrap your brain around the theory. But if you want to jump right in then a book like this one could get you going in the right direction.

Rob



-----Original Message-----
From: owner-statalist@hsphsun2.harvard.edu [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of K Jensen
Sent: Thursday, July 22, 2010 5:33 AM
To: statalist@hsphsun2.harvard.edu
Subject: Re: st: Unbalanced repeated measures analysis question

Thanks to Robert and David for your helpful comments. Sorry to sound
stupid here but mixed models are entirely new to me. I have been
reading up on them.

I have the variables outlined below:
SubjectID MeasurerID MeasurerType Result GoldStandard
where MeasurerID is always a certain MeasurerType (1-3)

SubjectID and MeasurerID should be random effects and MeasurerType
fixed?  How would you specify that in the xtmixed syntax?  I am
confused about having two grouping variables for the random effects.

Karin

On 21 July 2010 22:37, Ploutz-Snyder, Robert (JSC-SK)[USRA]
<robert.ploutz-snyder-1@nasa.gov> wrote:
> Karin,
> I feel your pain RE Stata's anova syntax for repeated measures...
>  But I also agree with David that I think your better bet is
> probably to use -xtmixed- and then apply -margins- for  your
> post-hoc comparisons, given the imbalance issue.  You can use
> -margins- to compare each of the three measures to the gold
> standard--akin to simple effect contrasts.
>
> If you wish to remain in the anova syntax, you might want to
> check out the user written -anovalator- command, thanks to Phil
> Ender from UCLA.  But from the sounds of your imbalanced design,
> I would tend to lean more to -xtmixed- with -margins-
>
> (BTW--the Phil's website at UCLA has some nice walk-throughs
> of all of this.)
>
> Rob
>
> -----Original Message-----
> From: owner-statalist@hsphsun2.harvard.edu [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Airey, David C
> Sent: Wednesday, July 21, 2010 4:07 PM
> To: statalist@hsphsun2.harvard.edu
> Subject: re: st: Unbalanced repeated measures analysis question
>
>
> I think when you have comparisons to a gold standard, or all
> comparisons with one control, that there are specific ANOVA
> post-hoc tests that perform better than all possible or all
> pairwise comparisons procedures.
>
> There is the complication that you are testing for equivalence,
> as you say.
>
> The Stata command -xtmixed- can do what -anova- can. Sometimes
> -manova- or -mvtest- is useful with repeated measures too.
>
> It is hard to understand how your design is unbalanced without
> seeing the data cross-tabs, etc.
>
>> Hi
>>
>> I have data on measuring a biological property for three different
>> methods plus a gold standard. Different people were trained in each
>> method (1,2 or 3) and measured the same subjects during different
>> sessions, together with the gold standard measurement.
>>
>> So the data look like
>> SubjectID MeasurerID MeasurerType Result GoldStandard  Diff
>> 1         1          1            95     99            -4
>> 1         2          3            102    99            +3
>> 1         3          2            92     99            -7
>> ...
>> 1        10          3            105    99            +6
>> 2         1          3             98   100            -2
>> ...
>>
>> Sometimes patients would be called in to see the consultant and so
>> missed for a particular measurer, but otherwise all the measurers
>> would measure all the patients seen in a particular session. Different
>> sets of measurers (but all trained by methods 1,2 or 3) were used on
>> each session (individual measurers 1-10 on session 1, 11-20 on session
>> 2 etc).
>>
>> The gold standard measurements on each session are roughly normally
>> distributed, as are the differences from the gold standard. We are
>> interested in the accuracy of each of the three methods.
>>
>> Is it OK to do some sort of repeated measures ANOVA here, with an
>> unbalanced design? If it is what would be the syntax (Stata 10)? Sorry
>> to sound pathetic but I just can't get the anova command with the
>> repeated option to work here.
>>
>> Is there a better measure to use than the difference to reflect the
>> fact that we are interested in a comparison with a gold standard?
>>
>> Thankyou
>> Karin

*
*   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/