How (if at All) do Perceptions of Supervisor’s Listening Differ from General Relationship Quality?: Psychometric Analysis

Abstract

Employees who perceive their supervisors to listen well enjoy multiple benefits, including enhanced well-being. However, concerns regarding the construct validity of perceived-listening measures raise doubts about such conclusions. The perception of listening quality may reflect two factors: constructive and destructive listening, which may converge with desired (e.g., humility) and undesired (e.g., rudeness) supervisor-subordinate relationship behaviors, respectively, and both may converge with relationship quality (e.g., trust). Therefore, we assessed the convergent validity of four perceived listening measures and their divergent validity with eight measures of supervisor-subordinate relationship behaviors, eight relationship-quality measures, and a criterion measure of well-being. Using data from 2,038 subordinates, we calculated the disattenuated correlations and profile similarities among these measures. The results supported convergent but not divergent validity: 58.7% (12.6%) of the correlations expected to diverge had confidence intervals with upper limits above 0.80 (0.90), and 20% of their profile-similarity indices were close to 1. To probe these correlations, we ran a factor analysis revealing good and poor relationship factors and an exploratory graph analysis identifying three clusters: positive and negative relationship behaviors and relationship quality. A post-hoc analysis indicated that relationship-quality mediates the effect of the positive and negative behaviors on well-being. The results demonstrate the challenge of differentiating the perception of listening from commonly used supervisor-subordinate relationship constructs, and cast doubts on the divergent validity of many constructs of interest in Organizational Behavior. However, using the “sibling” constructs framework may allow disentangling these highly correlated relationship constructs, conceptually and empirically.