Post-estimation

REGO: module that decomposes R2 (share of explained variance) of an OLS model into contributions of (groups of) regressor variables with the help of Shapley or Owen values. The use of “groups” of regressor variables that belong to the same category (such as the variables that belong to a polynomial in age) reduces computational effort in comparison to “classical” Shapley decomposition without groupings. REGO includes a bootstrap option to obtain confidence intervals for the decomposition results. See http://www.uni-leipzig.de/~rego/and http://www.stata.com/meeting/uk12/abstracts/materials/uk12_sunder.pdf.

DOMIN: conducts dominance analysis (Budescu, 1993; Psychological Bulletin) computing general (an additive decomposition of a user supplied fit statistic related to reducing prediction error), and conditional weights as well as complete dominance criteria for user supplied regression model. All dominance analysis statistics are to evaluate a variable’s relative importance in reducing prediction error in the prediction of some (set of) dependent variable(s). Includes 2 wrapper modules to conduct multivariate and linear mixed effect regression dominance analysis. See https://ideas.repec.org/c/boc/bocode/s457629.html.

SHAPLEY: performs (exact additive) decomposition of a sample statistic according to effects on the outcome variable (i.e. predictors) specified in a given list. To perform Shapley decomposition, the effects are eliminated one by one, and marginal effects from each exclusion are weighted according to the stage of exclustion. The weights of the marginal effects are assigned in such a way that all exclusion trajectories have equal
weights.

SHAPLEYX: performs (exact additive) decomposition of a sample statistic according to effects on the outcome variable (i.e. predictors) specified in a given list. To perform Shapley decomposition, the effects are eliminated one by one, and marginal effects from each exclusion are weighted according to the stage of exclustion. The weights of the marginal effects are assigned in such a way that all exclusion trajectories have equal
weights.

SHAPLEY2: performs a Shorrocks-Shapely decomposition of many estimation statistics such as the R-squared in the OLS regression. It provides an additive decomposition of the statistic, allowing you to see the relative contribution of each regressor. The command is thought as a post-estimation command, hence you should use it right after the estimation. Compared to shapley, shapley2 is faster, but provides the same results (numerical differences are possible). However, the computation still takes some
time and the maximum amount of RHS variables is 20. It allows to regroup several regressors into groups and to compute the relative importance of the whole group. This allows as well to accelerate the computation. It is designed as a post-command routine. The model specifications are extracted from the previous estimation (ols, probit). See https://ideas.repec.org/c/boc/bocode/s457543.html.