How Long? How Many? How Much? Evidence of Convergent Validity Among Thin-Slice Behavioral Coding Metrics

Abstract Despite the ubiquity of thin-slice coding for behavioral measurement, there exists relatively little systematic research into the convergent validity of thin-slice coding metrics for nonverbal behaviors when using human coders. This study utilized five previous datasets to measure four commonly-measured nonverbal behaviors (gaze, gestures, nods, smiles) using three different coding metrics (duration, frequency, rating) coded in 2 or 3 min slices. Convergent validity was measured by comparing a given behavior coded with at least two different metrics. Meta-analytic assessments across studies, behaviors, and metrics indicated strong convergent validity for various metrics for each behavior. Results provide confidence to researchers on the validity of using these thin-slice coding metrics for nonverbal behaviors.