Аңдатпа
parents) in Kazakhstan - a post-Soviet setting where institutional trust is characterized by Soviet legacies and post-pandemic coercion - that varied messenger endorsement (Grand Mufti, President, Chief Sanitary Doctor, control) and social norm framing (Muslim, national, local, control), supplemented by qualitative analysis of 829 open-ended responses. All three endorsements reduced vaccination intent by 6-7 percentage points relative to the 76% control baseline, while norm messaging showed null effects overall. Non-Muslims drove the backfire (11-12 percentage point decline); Muslims were insulated. Urban residents responded positively to norms; rural residents did not. Qualitative analysis traced hesitancy to pragmatic safety and efficacy concerns - only two respondents cited religion - indicating the endorsement strategy targeted barriers that did not drive hesitancy in this population. These findings identify trust-dependent boundary conditions for prestige bias theory in health communication: when institutional trust is low and recent coercion has primed reactance, endorsements from prestigious figures trigger resistance rather than deference. The contrast with positive endorsement effects during the pandemic in the same country suggests that effectiveness depends on timing and whether attitudes are still forming. Where institutions have lost credibility, leveraging institutional authority is not just ineffective but counterproductive.