Estimating the Proportion of True Null Hypotheses: A Likelihood Approach
Many estimators for the proportion of the true null hypotheses in a multiple testing problem have been proposed in literature. Motivated from the work on the histogram approach, in this article we propose a new estimator based on the likelihood function with an approximating alternative histogram. AIC is used to select the number of bins for the histogram. Simulation study demonstrates that the new estimator outperforms and substantially improves existing methods including Storey estimators, convex density estimator, and histogram estimator. The new method is applied to a real-life data set of breast cancer.
