Using Weighted P-Values in Fisher's Method
arXiv:2006.10126v1
Originality Incremental advance
AI Analysis
This work addresses a limitation in statistical analysis for researchers needing to weight p-values differently, but it is incremental as it extends an existing method.
The authors tackled the problem of combining p-values with arbitrary weights in Fisher's method, which previously only allowed equal weights, and they provided a proof for their new method.
Fisher's method prescribes a way to combine p-values from multiple experiments into a single p-value. However, the original method can only determine a combined p-value analytically if all constituent p-values are weighted equally. Here we present, with proof, a method to combine p-values with arbitrary weights.