wyoung 2.0 released

[Julian Reif]

The Stata command wyoung is updated to version 2.0 with new support for permutation-based resampling. Install with:

ssc install wyoung, replace

wyoung controls the family-wise error rate—the probability of rejecting at least one true null hypothesis—when testing multiple hypotheses that arise from considering multiple outcomes, subgroups, or parameters. This method leverages resampling techniques, such as bootstrapping (sampling with replacement) or permutation (shuffling), to adjust the standard p-values obtained from model estimation. The command also computes the Bonferroni-Holm and Sidak-Holm adjusted p-values. Detailed documentation, including simulation results, is available on Github. Syntax and usage instructions can be accessed directly in Stata by typing help wyoung at the command prompt.


Example. Perform the Westfall-Young adjustment using permutation with a stratified random sample (3 hypotheses).

sysuse auto.dta, clear
set seed 20
gen stratum = floor(mpg/11)
gen treat = foreign
local yvars "mpg headroom turn"
wyoung `yvars', cmd(regress OUTCOMEVAR treat) familyp(treat) permute(treat) strata(stratum)


Estimating family-wise adjusted p-values for 3 hypothesis tests

familyp: treat
    regress mpg treat
    regress headroom treat
    regress turn treat

Performing 100 permutations...

     +------------------------------------------------------------------------------------------------------------------------+
     | k                    model    outcome   familyp         coef      stderr           p   pwyoung       pbonf      psidak |
     |------------------------------------------------------------------------------------------------------------------------|
  1. | 1        regress mpg treat        mpg     treat    4.9458042   1.3621621   .00052542       .08   .00105083   .00105056 |
  2. | 2   regress headroom treat   headroom     treat   -.54020979   .20708836   .01104863       .08   .01104863   .01104863 |
  3. | 3       regress turn treat       turn     treat   -6.0332168   .87394304   1.665e-09         0   4.994e-09   4.994e-09 |
     +------------------------------------------------------------------------------------------------------------------------+