Setting Statistical Hurdles for Publishing in Accounting

Abstract

Abstract Ohlson (2023) argues that researchers tacitly avoid raising statistics-related ‘elephants’ that could undermine inferences. We offer a balanced perspective, first applauding the remarkable progress made in deriving testable predictions, leveraging modern statistical techniques, and tapping alternative Big Data sources to address issues relevant to practitioners, regulators and academia. While we concur with Ohlson’s elephants, we caution against over-criticism based on statistical design choices, as it risks creating new elephants. Our key lessons: focus on meaningful hypotheses, recognize merits of descriptive studies, balance Type I and II errors in data handling and journal reviewing, employ proper context when interpreting statistical significance and consider economic significance. Overall, though empirical accounting research faces challenges, criticism should not deter innovative research (Type II error in journal reviewing).