New article on exploratory and confirmatory research

A new article concerning the blending of exploratory and confirmatory research by Michael Höfler has been published. Data exploration holds enormous potential for discovering novel scientific findings, as well as for creating new and for modifying existing hypotheses. However, it is often misused in papers to cherry-pick from a variety of results, which are then framed as evidence for hypotheses, which did not necessarily exist before data were collected. Nonetheless, if data exploration is transparent, it may lead to new findings, but these findings requires new data to be tested confirmatively. In this first of two consecutive papers, we describe how confirmatory (hypothesis testing) and exploratory research (hypothesis generation and modification) are blended together, but how they can also be strictly separated by a concept called "severe testing". In Part II, suggestions are made on how to actually plan and conduct data exploration, and how to implement more transparent exploratory research in psychology.

You can read the full article here.