The Credibility Crisis in Science: Tweakers, Fraudsters, and the Manipulation of Empirical Results

The most dangerous fraudulent practice in science is often overlooked, argue social scientists Plümper and Neumayer (coauthors of Robustness Tests for Quantitative Research) in this thorough but dry examination. The authors demonstrate that “tweaking”—which they define as “the intentional manipulation of model specifications and research designs” typically to increase the likelihood of publication or to get results that align with the authors’ beliefs—is a widespread practice that weakens the validity of empirical results and degrades the body of modern scientific inquiry as a whole. They examine common tweaking strategies, such as modifying the level of statistical significance of the data’s effects and conclusions and selecting model specifications based on the data rather than before the data is collected, and offer examples of fraud, like the case of Francesca Gino, a former Harvard professor accused in 2023 of fabricating data in her behavioral research papers. While the arguments are thorough and the authors offer potential solutions, such as robustness tests selected by journal editors and peer reviewers, it is not clear who the authors intend to reach; the text is largely technical in nature with a few humorous interjections and pop culture quotes interspersed throughout. Still, for those concerned with a loss of credibility in science, this is worth a look. (Mar.)