Manage LMX differentiation fairly as a leader

Acknowledge that you develop different relationships with different people — and ensure the differences are earned, not arbitrary.

Why it works

Leaders inevitably develop stronger relationships with some team members than others. The question is whether the differentiation is based on performance, trust, and mutual investment (earned) or on similarity, social comfort, and unexamined bias (arbitrary). Arbitrary differentiation produces an in-group that mirrors the leader’s demographics rather than the team’s talent, and an out-group that disengages because the exclusion feels unfair.

How to do it

1. Audit your current team relationships: who are you investing in most? Is that driven by performance or preference?
2. Create formal opportunities for every team member to demonstrate capability — don’t wait for them to self-select into your inner circle.
3. Check whether your high-LMX group reflects demographic similarity to you, and if so, explicitly seek relationship investment with those who are different.
4. Distinguish between high LMX based on performance history (legitimate) and based on shared background (arbitrary).

Evidence

Research on LMX differentiation finds it is associated with team performance when based on competence, but with lower cohesion and fairness perceptions when it appears arbitrary. Martin et al. (2018) clarify conceptually that the outcomes of differentiation hinge on whether members perceive the variation as justified — the precise fairness distinction this practice turns on. (observational)

The LMX differentiation literature is complex and contested; some studies find team LMX diversity beneficial for performance, others do not. Context matters significantly.

Sources

• Martin, R., Thomas, G., Legood, A., & Dello Russo, S. (2018). Leader-member exchange (LMX) differentiation and work outcomes: Conceptual clarification and critical review. Journal of Organizational Behavior, 39(2), 151-168.

Common mistake

Claiming to treat everyone equally while actually differentiating significantly — which is worse than transparent differentiation because it teaches the out-group that their perception is invalid.