Purpose
Perceived ostracism from supervisor undermines employees' efficacy needs, yet its antecedents remain underexplored. Prior research has linked supervisor ostracism to low leader-member exchange (LMX), which primarily focuses on work-related interactions only. However, employees' evaluation of supervisor ostracism is also developed through non-work interactions, such as shared meals or holiday greetings. This personal, informal, and outside-work exchanges is captured by the concept of Supervisor-Subordinate Guanxi (SSG), which includes affective, hierarchical, and personal involvement aspects of social connections. Drawing on social identity theory, we test whether perceived insider status (PIS) mediates the SSG-supervisor ostracism relationship and whether this indirect effect is moderated by perceived leader-member similarity. We control for LMX to rule out work-related interactions explanations and to assess SSG's incremental value.
Design/methodology/approach
We tested our hypotheses using longitudinal cross-level moderated mediation design with 134 Chinese employees across 30 teams. Data were collected in two waves with a 4.5-month lag.
Findings
Controlling for team-level LMX (T1), individual PIS (T2) fully mediated the negative relationship between team-level SSG (T1) and individual perceived supervisor ostracism (T2). The negative indirect effect was stronger when team members perceived low similarity with their supervisor (T2) compared to high similarity.
Theoretical Contribution
This study identifies novel antecedents and boundary conditions of perceived supervisor ostracism. It expands social identity theory by integrating SSG, perceived insider status, and team member-leader similarity. Additionally, it highlights the spillover of social exchange from private to work domains, as SSG explains incremental value beyond work-related exchanges.
Research limitations/implications
Although the two-wave design supports temporal precedence, future research using larger samples with three-wave longitudinal design is recommended to establish causal ordering more robustly.