Significance Testing in Empirical Finance : A Critical Review and Assessment

This paper critically reviews the practice of significance testing in modern finance research. Employing a survey of recently published articles in four top-tier finance journals, we find that the conventional significance levels are exclusively used with little consideration of the key factors such as the sample size, power of the test, and expected losses. We also find that statistically significant results reported in many surveyed papers become questionable, if Bayesian method or revised standards for evidence were instead used. We observe strong evidence of publication bias in favour of statistical significance. We propose that substantial changes be made to the current practice of significance testing in finance research, in order to improve research credibility and integrity