Reproducible Aggregation of Sample-Split Statistics

Statistical inference is often simplified by sample-splitting. This simplification comes at the cost of the introduction of randomness that is not native to the data. We propose a simple procedure for sequentially aggregating statistics constructed with multiple splits of the same sample. The user specifies a bound and a nominal error rate. If the procedure is implemented twice on the same data, the nominal error rate approximates the chance that the results differ by more than the bound. We provide a non-asymptotic analysis of the accuracy of the nominal error rate and illustrate the application of the procedure to several widely applied statistical methods.