Inter-Rater Agreement, Data Reliability, and The Crisis of Confidence in Psychological Research

Abstract

In response to the crisis of confidence in psychology, a plethora of solutions have been proposed to improve the way research is conducted (e.g., increasing statistical power, focusing on confidence intervals, enhancing the disclosure of methods). One area that has received little attention is the reliability of data. We note that while it is well understood that reliability of measures is essential to replicability, there is a failure to apply some measure of data reliability consistently, or to correct for chance when assessing agreement. We discuss the problem of relying on Percent Agreement between observers as a measure of reliability and describe a dilemma that researchers encounter when assessing contradictory indicators of reliability. We conclude with some pedagogical strategies that might make the need for reliability measures and chance correction more likely to be understood and implemented. By so doing, researchers can contribute to solving some aspects of the crisis of confidence in psychological research.

Keywords

• reliability; inter-rater agreement; kappa; percent agreement; research methods
• confidence intervals