When should we discard pair dummies in a pair-matched experiment? Should we let LASSO select which pair of dummies to include?
What does "weak instruments" mean in the context of RCTs? Assuming that we have at least some compilers, do the usual Stock et al. weak instruments F-tests provide useful advice for those analyzing field experiments with noncompliance?
Is LASSO an effective way to select covariates?
When should we discard pair dummies in a pair-matched experiment? Should we let LASSO select which pair of dummies to include?
What does "weak instruments" mean in the context of RCTs? Assuming that we have at least some compilers, do the usual Stock et al. weak instruments F-tests provide useful advice for those analyzing field experiments with noncompliance?