Associational Pathways and configurations linking perceived gamification quality and teaching presence to self-reported learning persistence in blended STEM learning

Learning persistence remains a critical challenge in blended STEM education. This study tested a dual-pathway association model linking perceived gamification quality and teaching presence to self-reported learning persistence through motivation, self-regulated learning, and engagement. Of 412 enrolled students, 389 submitted the Week 15 questionnaire; 13 submissions were excluded during data-quality screening, yielding N = 376. Disjoint two-stage PLS-SEM supported all seven hypothesized direct paths and both serial indirect associations, explaining 76.2% of persistence variance. The direct GAM→LP association (β = .274) exceeded the serial GAM→MOT→ENG→LP association (β = .208), whereas TP→SRL→ENG→LP (β = .290) slightly exceeded TP→LP (β = .263). Sensitivity analyses showed that ENG→LP remained stable after removing overlapping indicators (β = .566), while a reduced design-focused gamification specification remained strongly associated with motivation (β = .769). Re-estimated fsQCA identified two high-persistence and three low-persistence configurations, with dominant patterns largely mirrored, indicating limited partial configurational asymmetry. Nonresponders showed lower academic and LMS participation, and gender was strongly confounded with program. Findings therefore provide context-bound, cross-sectional evidence of stable associations rather than verified causal mechanisms or population-wide persistence effects.