The following table summarizes different commands for multi-hypothesis tests. Typically, multiple comparison procedures check the household-wise error rate (FWER) instead, which is the likelihood that there is one or more false positives among all of the hypothesis tests performed.
The original method for the FDR was used for the multiple hypothesis tests for independent test statistics. This 1995 paper showed that the original FDR method controls FDR even in cases where test statistics have a positive regression dependence on each of the test statistics that corresponds to a true null hypothesis. The teams are highly flexible, allowing for each equation to have varying controls, differing samples, having clustered standard errors, etc., but they are testing different assumptions from the above approaches. Instead of tuning every single P-value to fit more tests, it runs a combined test for the Sharpe hypothesis, which is no treatment has an effect, then uses a Westfall-Young approach to check it on all equations.
The ovt command runs the regression specification error (RESET) test on omitted variables. The ovtest command performs one more regression specification test. The linktest command runs the linktest model specification for the one-equation models. It runs a regression analysis and lists the STATA commands that can be used to test heteroscedasticity.
Stata also has avplot, which produces a variable-added plot of all variables, which can be really helpful when you have lots of variables. When you have data that could be considered to be a time series, you should use the dwstats command, which runs a Durbin-Watson test on correlated residuals. The test can be performed either by regressing the regression attrition against a treatment assignment, set of observed values, or observed values that are interacted with treatment assignments using the primary specification. A common approach is to simply check differences in the observed characteristics between the treatment group and comparison group, conditioned on the status of attrition.
That is, if a primary specification bundles the standard errors, and includes fixed effects, weights, and strata, then the attrition test should do so as well. With small numbers of hypotheses tested, controlling for the FWER is helpful, then rwolf2 and wyoung both handle what is needed. Provides STATA commands to calculate the Q-values for multi-test procedures (computes adjusted FDR Q-values).