Each week, we will be publishing layman’s abstracts of new articles from our prestigious portfolio of journals in statistics. The aim is to highlight the latest research to a broader audience in an accessible format.

The article featured today is from Statistics in Medicine, with the full Open Access article now available to read here.

Montgomery, RN, Mahnken, JD. A prediction‐based test for multiple endpoints. Statistics in Medicine. 2020; 39: 4267– 4280. https://doi.org/10.1002/sim.8724

Contemporary research typically entails the collection of many measurements, often on very few subjects or “observational units.”  Most of the ways to assess studies are set up for a completely different setting, that of studies with few, or only one, measurement collected on many subjects.  Our new approach takes advantage of these common study settings to draw conclusions and make decisions.  In studies with many things measured, it would seem unusual to predict even simple changes—like an increase or a decrease by any amount—for all or even most of those things measured simply because there are so many to predict. So, if an investigator can accurately predict most or all of these simple changes in advance, then these predictions can provide strong support that the investigator understands what is happening and therefore that their underlying theory is correct.  And as the number of measurements collected increases, it would become more unlikely to simply guess most or all of them correctly.  Thus, the common research setting of collecting many study measures can be leveraged—regardless of the number of subjects—to advance research theories, whereas current methods make this burden of proof much harder.