st: Re;pwcompare, repeated measures and xtmixed

[Arti Rayit <arti.rayit@yahoo.com>]

Hello,

I am trying to get to grips with pwcompare and it's use with repeated measures with anova and xtmixed in Stata 12.1. On a simple repeated measures design I get the following:
anova f i.j pat,rep(j)
. pwcompare j,eff
Pairwise comparisons of marginal linear predictions
Margins      : asbalanced

------------------------------------------------------------------------------
             |                            Unadjusted           Unadjusted
             |   Contrast   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
             j |
     2 vs 1  |     -3.875   1.351256    -2.87   0.009    -6.685091   -1.064909
     3 vs 1  |         -4     1.351256    -2.96   0.007    -6.810091   -1.189909
     4 vs 1  |     -2.375   1.351256    -1.76   0.093    -5.185091    .4350907
     3 vs 2  |      -.125    1.351256    -0.09   0.927    -2.935091    2.685091
     4 vs 2  |        1.5     1.351256     1.11   0.280    -1.310091    4.310091
     4 vs 3  |      1.625   1.351256     1.20   0.243    -1.185091    4.435091
------------------------------------------------------------------------------


Yet when I use the following:
xtmixed f i.j, || pat:,cov(id)

. pwcompare j,eff
Pairwise comparisons of marginal linear predictions
Margins      : asbalanced
------------------------------------------------------------------------------
             |                            Unadjusted           Unadjusted
             |   Contrast   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
f            |
           j |
     2 vs 1  |     -3.875   1.154982    -3.36   0.001    -6.138724   -1.611276
     3 vs 1  |         -4     1.154982    -3.46   0.001    -6.263724   -1.736276
     4 vs 1  |     -2.375   1.154982    -2.06   0.040    -4.638724   -.1112761
     3 vs 2  |      -.125    1.154982    -0.11   0.914    -2.388724    2.138724
     4 vs 2  |        1.5     1.154982     1.30   0.194    -.7637239    3.763724
     4 vs 3  |      1.625   1.154982     1.41   0.159    -.6387239    3.888724
------------------------------------------------------------------------------

and the probability for the 4 - 1 contrast goes from non significant to significant.

Have I misused pwcompare or is my specification for xtmixed incorrect?

Thanks.

Arti.

*
*   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/