Re: st: rmttest for 5:1 matching of controls to case?

[Roger Harbord <roger.harbord@bristol.ac.uk>]

-rmttest- just does a paired t-test and therefore only works for paired 
data. I think you'd be better off with -xtreg, fe-, e.g.:

. xtreg duration disease, i(matchvar) fe

This gives a test of your question (1) below.

Roger.

--On 06 February 2006 12:34 +0000 Marieke Schoonen 
<Marieke.Schoonen@lshtm.ac.uk> wrote:


> Hello,
> I tried contacting the creator of the rmttest command but his email
> address doesn't work.
> Below is the message I would like to send to John R. Gleason at
> Syracuse university - but perhaps someone else can help me out here? (or
> provide me with his current email address)
>
> I have a matched case-control dataset with up to 5 controls per case.
> I would like to know whether there is a difference between cases and
> matched controls for the continuous variable <duration>.
>
> > From the examples in the help-file it appears that the rmttest should
> be used when testing repeated measures within the same person.
> Is it appropriate to use rmttest to answer the above question?
> If so, I am wondering whether I am testing
> 1) if the mean <duration> of all matched (up to 5) controls is equal to
> the single value of the matched case
> or
> 2) if each control has the same duration as the matched case; i.e. each
> case is tested up to 5 times
>
> The do file would look like this:
> *** outcome variable: "disease"  (0=control, 1=case), there are N=1220
> cases and N=4723 controls.
> *** exposure variable: "duration" (continuous)
> *** matching variable: "matchvar" (numerical)
>
> rmttest duration, by(disease 0 1) id(matchvar)
>
>
> Thanks in advance for your help,
> Marieke Schoonen

*
*   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/