Interobserver reliability in clinical research: Current issues and discussion of how to establish best practices.

The study of a wide range of topics covered by clinical research studies relies on data obtained by observational measures. These include observations of individuals (adults, children, but could also be animals), couples, parents and their child(ren), or groups observed for varying amounts of time in settings that range from naturalistic to structured. Data from observations, whether expressed affect, behavior, or verbalizations, are typically rated or coded. Regardless, interobserver reliability is a key criterion with which one evaluates the quality of evidence from observations. In this article, we discuss the complexities involved in designing, implementing, testing, and evaluating the adequacy of interobserver reliability. We offer options for addressing each complexity and discuss advantages and disadvantages of different approaches. We leave readers sufficiently well informed to make decisions about approaches to interobserver reliability regardless of their area of study within clinical research. (PsycINFO Database Record (c) 2019 APA, all rights reserved).