st: rm anova different in Stata 9.2/10 versus Stata 11/12?

[Pieter-Jan van Ooij <duikerarts@home.nl>]

Dear Statalister.

Since a few years I perform my statistics using Stata 9.2. As this is a
relatively "old" version I tried Stata 12 version from a acquaintance to
see if it has any benefit for me. I used a recent database to compare
Stata 9.2 and Stata 12 and used RM anova. When I compared the results it
appeared that in Stata 9.2 this result was non-significant (p=0,0570)
while in Stata 12 with the same database and command line it was
significant (p=0.0480). When I checked the output Stata generated it
appeared that in Stata 9.2 only 15 out of 25 subjects were taken (no
warning of dropped cases) in this anova while in Stata 12 all 25 were
included. When I "downgraded" Stata 12 to Stata 9.2, Stata 10 and Stata 11
using the "version" command it appeared that what happened in Stata 9.2
also happened in Stata 10 whereas Stata 11 generated the same result Stata
12. I tried to find out what the reason is for this discrepancy but I
couldn't find any explanation. Does anyone of you, statalisters, know what
the reason is or how you can correct for this? For the sake of
completeness the database consist of 25 subjects measured during 2
separate days at 6 time points a day.

Kind Regards,

Pieter-Jan van Ooij


*
*   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/