medical research is true

A common model for p-values from multiple hypothesis tests is the “two-groups” model[1], where the p-values are assumed to be drawn from the mixture distribution: p ∼ π0f0 + (1− π0)f1 where π0 is the proportion of p-values corresponding to tests where the null hypothesis is true, f0(·) is the density of the p-values under the null hypotheses and f1(·) is the density of p-values under the alternative hypothesis. In general, each test may have a corresponding null and alternative distribution, in which case both f0 and f1 may be represented by mixture distributions. This model is frequently used in genomics [2], or brain imaging studies [3], where thousands of variables are measured simultaneously with the goal of estimating the statistical significance of a group of variables [4]. This model simultaneously estimates π0, the fraction of results that are drawn from the null distribution. Alternatively, 1 − π0 represents the fraction of p-values from the alternative distribution.