Table 1:

Algorithms for anomalous subgroup detection for binary outcomes

Conditional Automated Stratification Scan (CASS)	Matched Conditional Automated Stratification Scan (mCASS)	Weighted Conditional Automated Stratification Scan (wCASS)
Get treatment group data Data|Z=1 Get comparison group data Data|Z=0 For each subject i in Data|Z=1, estimate the counterfactual outcome as mean outcome in Data|Z=0, i.e. Apply Bias-Scan(X, Y, ) using Data|Z=1 Estimate statistical significance of identified subpopulation using boot-strapped randomization testing
Get the treatment’s propensity scores from the best propensity score model Get the treatment’s logit of the propensity score Get treatment group data Data|Z=1 Get comparison group data Data|Z=0 For each subject i in the Data|Z=1 Identify nearest neighbors in Data|Z=0 as those within 0.2SD of the logit of the propensity score Estimate the counterfactual outcome , as the average outcome among the identified nearest neighbors Apply Bias-Scan(X, Y, ) using Data|Z=1 Estimate statistical significance of identified subpopulation using boot-strapped randomization testing
Get the treatment’s propensity scores from the best propensity score model Compute the average treatment effect on the treated (ATT) weights (wATT) Get treatment group data Data|Z=1 Get comparison group data Data|Z=0 For each subject i in Data|Z=1, estimate the counterfactual outcome Yi as the ATT-weighted mean expected outcome in Data|Z=0 Apply Bias-Scan(X, Y, ) using Data|Z=1 Estimate statistical significance of identified subpopulation using boot-strapped randomization testing