Report on contextually sensitive framework for defining QRPs

Authors

Fanelli, Daniele, Voodla, Alan, Uusberg, Andero, Andres, Siim, Iordanou, Kalypso & de Vos, Maura

Description

This document addresses the conceptual and empirical challenges of defining and studying Questionable Research Practices (QRPs), which are inherently context-dependent and difficult to classify. To tackle these limitations, it introduces a theoretical framework aimed at better defining and contextualising QRPs and empirically examining how contextual features influence their impact. The report synthesises three studies: the first reclassified existing survey-based definitions of QRPs into a structured scheme, identifying 27 distinct types. It found that selective reporting and HARKing are the most frequently self-reported QRPs and are also the most sensitive to contextual factors, making them the focus of further analysis.

Building on these findings, the second study applied Information Compression Theory to explain how the impact of QRPs varies across research fields, depending on information yield, replicability, and existing bias. These dimensions informed a third study that tested whether researchers consider context when judging QRP severity. Using an experimental vignette focused on HARKing, the study found that researchers generally judged the non-questionable option more favourably, regardless of manipulated contextual factors. However, researchers who believed contextual factors mattered evaluated HARKing less negatively, suggesting that perceptions of context influence how QRPs are morally assessed, even when such context is not systematically applied in practice.