Re: st: Comparing trends between proportions for two groups

[Maarten buis <maartenbuis@yahoo.co.uk>]

--- On Thu, 25/3/10, roland andersson wrote:
> I am analysing the results of a randomised trial comparing
> healing rates of wounds for two treatments. Follow up a 2
> weeks, 3 months and 1 year. At each moment there is no
> statistical difference, but there is a constant trend that
> one treatment is better. I would like to compare the trends
> for the healing rates between the two treatments as they
> may come up with a significant difference. How can I do that?

You haven't got any feedback yet, so let me try to give some
comments that may or may not be useful:

This setup reminds of a discrete time survival dataset, and
your hypothesis would than be about differences in steepness
of the survival curve. The strength of such a model is that
they usually include some (more or less) reasonable 
assumptions which allow you to summarize the differences in
healrates for each follow up in one number, thus using more
information from your data in estimating the difference 
between treatments. This would hopefully improve the power
of your test, which seems to be the problem. The price is
ofcourse that some of this "extra information" is actually
comming from the assumptions rather than the data, so as
always you'll need to check those.

For (discrete time) survival analysis I like this site:
http://www.iser.essex.ac.uk/study/resources/module-ec968

Hope this helps,
Maarten

--------------------------
Maarten L. Buis
Institut fuer Soziologie
Universitaet Tuebingen
Wilhelmstrasse 36
72074 Tuebingen
Germany

http://www.maartenbuis.nl
--------------------------


      

*
*   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/