Purpose This study aims to explain how transformational leadership (TL) relates to team innovation by specifying two concurrent mechanisms – knowledge sharing (resource pathway) and role conflict (hindrance pathway) – and by testing when each pathway is stronger or weaker through the boundary role of collaborative team climate. Design/methodology/approach Multisource data were collected from 75 teams (330 members) across organizations in the United Arab Emirates. Managers rated TL and team innovation; team members rated collaborative climate, knowledge sharing and role conflict. Hypotheses were tested using multilevel structural equation modeling (Mplus v8.9) with standardized variables. Conditional indirect effects and indices of moderated mediation were estimated with a complementary Bayesian specification (5,000 draws). Findings Transformational leadership exhibits a positive direct association with team innovation. Indirectly, it promotes innovation through greater knowledge sharing while potentially dampening it via heightened role conflict. Collaborative team climate operates as a first-stage moderator, strengthening the leadership–knowledge sharing pathway and weakening the leadership–role conflict pathway. Under stronger collaborative climates, the resource pathway becomes more pronounced and the hindrance pathway attenuates, consistent with moderated mediation on both routes. Practical implications To convert TL into innovation while minimizing coordination losses, leaders should pair inspirational and developmental behaviors with collaboration scaffolds (role charters, decision-rights matrices and cadence agreements) and invest in routines that build collaborative climate and psychological safety. Originality/value The paper positions TL within the job demands–resources framework as a dual-pathway influence, clarifies construct boundaries between leadership and team processes and identifies collaborative team climate as a first-stage moderator – not a mediator – thereby integrating leadership and climate in a single conditional process model.
Khassawneh et al. (Thu,) studied this question.