help pwmc


pwmc -- Pairwise multiple comparisons (unequal variances)


Syntax for pwmc

pwmc varname , over(varname) [options]

Immediate command

pwmci stats1 stats2 stats3 [ ... statsk ] [ , options ]

where stats is

#obs #mean #sd


pwmc performs pairwise comparisons of means. It computes pairwise differences of the means of varname over the levels of over(varname). Confidence intervals are derived using procedures allowing for unequal variances across groups.

pwmc implements Dunnett's C, GH and T2 procedures, as proposed by Dunnett (1980), Games and Howell (1976) and Tamhane (1979).

pwmci is the immediate form of pwmc. See immed.


over(varname) is required and specifies that means are computed for each level of varname.

procedure(proc) specifies the procedure to be used to derive confidence intervals. proc may be any combination of C, GH and T2 and case does not matter. Default is to report confidence intervals using each of the procedures.

level(#) specifies the confidence level, as a percentage, for confindence intervals. The default is level(95). See set level.

varlabel displays variable labels instead of variable names.

vallabel displays value labels instead of numeric codes.

cformat(%fmt) specifies how to format means, standard errors, and confidence limits.


. pwmc wage , over(race)

. pwmci 1637 8.08 5.96 583 6.84 5.08 26 8.55 5.21

Saved results

pwmc saves the following in r():

Scalars r(k) number of levels of over(varname) r(ks) number of differences

Macros r(depvar) varname from which means are computed r(over) varname in over(varname)

Matrices r(diff) differences of means r(Var) variances of differences r(nuhat) degrees of freedom r(ll_proc) lower limit of confidence interval using procedure proc r(ul_proc) upper limit of confidence interval using procedure proc r(A_proc) critical values using procedure proc (cf. Dunnett, 1980)


Dunnett, Charles W. (1980). Pairwise Multiple Comparisons in the Unequal Variance Case, Journal of the American Statistical Association, 75(372), 796-800.

Games, Paul A., Howell, John F. (1976). Pairwise Multiple Comparison Procedures with Unequal N's and/or Variances: A Monte Carlo study, Journal of Educational Statistics, 1(2), 113-125.

Tamhane, Ajit C. (1979). A Comparison of Procedures for Multiple Comparisons of Means with Unequal Variances, Journal of the American Statistical Association, 74(366), 471-480.


Matthew K. Lau's 'DTK' package for R was helpful developing pwmc.


Daniel Klein, University of Kassel,

Also see

Online: pwmean, oneway

if installed: dunnett, prcomp