The Replicability Crisis in Social Sciences: Do We Need a New Paradigm?

In 2016, Baker reported findings from a Nature-published study surveying 1,576 researchers at leading universities about replication challenges in their respective fields. About 70% of them reported that they had failed at least once to replicate the results of other authors' research, and about half of them had failed at least once to replicate the results of their own previously published research. However, what particularly sparked discussion in the scientific community was the fact that only 23% of them had attempted to publish their failed replications in scientific journals, while half of them were unsuccessful in their attempts—the scientific journals rejected them.The replication crisis, while present in natural sciences, has garnered more attention in social sciences and medicine, with studies revealing alarmingly low rates of reproducibility. Notably, research in fields such as psychology, economics, and various business disciplines has demonstrated significant challenges in replicating original findings, with success rates often falling below 50%. These studies have not only highlighted the pervasiveness of the replication issue across disciplines but also revealed substantial reductions in effect sizes when experiments are reproduced, underscoring the need for improved research practices and increased scrutiny of published results. These and numerous other findings have led to many questions in science, from conceptual issues to those concerning academic fraud, biases in interpreting results, post-hoc hypothesis formation (HARKing), and increasing pressure to publish. This literature review aims to examine the replicability crisis in social sciences, while also critically analyzing the role of scientific journal policies and publication pressures in exacerbating the crisis, ultimately advocating for a paradigmatic shift in research methodologies and our understanding of social phenomena.