We should implement several approaches for generating simultaneous confidence intervals covering the ATE across the full set of specified probes. A few ideas:
Bonferroni CIs: \hat{ATE}g \pm Z{1 - 0.05/(2p)} SE_g, where SE_g is the standard error based on the pooled variance estimate
We should implement several approaches for generating simultaneous confidence intervals covering the ATE across the full set of specified probes. A few ideas:
Bonferroni CIs: \hat{ATE}g \pm Z{1 - 0.05/(2p)} SE_g, where SE_g is the standard error based on the pooled variance estimate
Multivariate Normal CIs: an approach based on the variance-covariance matrix of the EIF as implemented in https://github.com/tlverse/tmle3/blob/master/R/utils.R#L55-L57