Stata power repeated command: calculation of variance

[Roman Mostazir]

Hi,

Please see below the command and results from the 1st example in Stata's power repeated manual. I am trying to figure out from the results how the reported Var_w and Var_we (variance within error) are calculated here. Anyone has any idea?

power repeated 26.4 25.6 21, corr(0.6) varerror(77)

Performing iteration ...

Estimated sample size for repeated-measures ANOVA
F test for within subject with Greenhouse–Geisser correction
H0: delta = 0  versus  Ha: delta != 0

Study parameters:

        alpha =    0.0500
        power =    0.8000
        delta =    0.7426
          N_g =         1
        N_rep =         3
        Var_w =    5.6622
       Var_we =   10.2667
        Var_e =   77.0000
          rho =    0.6000

  Covariance

               |          repeated        
               |       1        2        3
  -------------+---------------------------
  repeated     |                          
             1 |      77                  
             2 |    46.2       77          
             3 |    46.2     46.2       77

Estimated sample sizes:

            N =        20
  N per group =        20

[Rich Goldstein]

I'm not sure it's directly there, but I think the place to start is the "methods and formulas" section of the power repeated part of the manual (pss.pdf)

[Roman Mostazir]

Apologies Rich Goldstein , I wrongly copied the code. The code is now updated with the command in the example.