Multiple testing when many $p$-values are uniformly conservative

with application to testing qualitative interaction in educational interventions

Abstract

Qualitative interaction is an extreme form of treatment effect heterogeneity where the treatment can be beneficial for some but harmful for others. We formulated this question as a global testing problem with many conservative null $p$-values and proposed a simple technique—conditioning—to greatly improve the statistical power.

Publication
Journal of the American Statistical Association (2019)
comments powered by Disqus