Social Media Exposure and Dietary Quality in University Restaurant Consumers: A PLS-SEM Approach from Northern Peru

University students face dietary transitions shaped by time constraints, campus food environments, and intensive exposure to food-related content on social media, yet the mechanisms linking digital exposure to observable food choices and overall diet quality remain insufficiently modeled in Latin American contexts. This study examined whether social media-driven food norms (NI) and in-restaurant food choices (CD) sequentially mediate the effect of Instagram (IG) and TikTok (TK) exposure on overall diet quality (Y), while incorporating physical activity (PA) as an independent predictor. This is a quantitative cross-sectional study based on a paper questionnaire administered face-to-face to 615 university students (53.2% women; M = 21.7 years; 39.2% public, 60.8% private universities) eating in campus restaurants in La Libertad, northern Peru. Data were analyzed through PLS-SEM (SmartPLS 4) with 5000 bootstrap resamples and BCa 95% confidence intervals; Y was operationalized through a culturally adapted KIDMED index. All five structural hypotheses were supported: TK → NI (β = 0.479) exceeded IG → NI (β = 0.349); NI → CD (β = 0.473) and PA → CD (β = 0.216) operated as independent pathways; and CD → Y (β = 0.255) confirmed the distal link. NI fully mediated both digital pathways toward food choices. Diet quality in university restaurants is reconfigured primarily through normative, not informational, digital mechanisms, suggesting norm-based interventions over nutrition-information campaigns. Reflective measures showed adequate internal consistency for IG, TK, and NI (Cronbach's α = 0.874-0.889; CR = 0.907-0.923) and convergent validity (AVE = 0.620-0.751); the structural model explained 55.5% of the variance in NI, 30.8% in CD, and 34.8% in overall diet quality.