Abstract
In psychology, most research findings—and current theories informed by past research—rely on null hypothesis significance testing (NHST), a ubiquitous approach to statistical inference. However, despite its popularity, NHST is a limited and problematic approach, which puts into question the validity of a vast number of past research findings. The goal of the present thesis was to investigate two research practices—p-hacking and statistical misreporting—to determine if they were major issues within seven notable psychology journals. After running all research articles from 1945 to 2020 (spanning every five years) through Statcheck and plotting p-value distributions, p-hacking was observed to be a prevalent issue within particular years of published research and in specific research journals. It was also found that 14.7% of detected p-values were inconsistent with properly re-computed p-values, and that 3% of p-values were inconsistent such that it would alter the decision to reject (or retain) the null hypothesis. These findings suggest that the issue of p-hacking and statistical misreporting is cause for concern, and it warrants re-considering the reliability of past research findings.
