Special issue on non-significant results

Abstract

Academic journals are biased towards publishing results that are statistically significant and this bias can lead to incorrect inferences about economic parameters of interest. This special issue was created as a counter to this bias. The issue contains a mix of articles on important economic questions regardless of whether they document statistically significant results. Some of these articles provide evidence that a well-established result in the literature is not reliably confirmed in existing data, some provide novel empirical evidence that a theorized result is not statistically significant, and some of the articles provide methodological improvements to assist future research. In this preface to the special issue, I provide a brief overview of the literature on publication bias and discuss some of the proposed remedies with a goal of improving inference about important economic questions and encouraging academics to pursue research even if the results are not statistically significant.