Digital Divide, Digital Capital, and Educational Dropout Risk: Structural and Psychological Pathways of Digital Inequality

Educational inequality in the digital era extends beyond traditional resource disparities to encompass systematic digital divides that mediate socioeconomic vulnerability into educational failure. Existing research faces three critical limitations: (1) treating digital access as a binary gap rather than a multi-dimensional capital system with nonlinear effects, (2) failing to account for psychological mechanisms through which digital inequality translates into educational outcomes under varying socioeconomic contexts, and (3) lacking interpretable models to decompose the compound pathways connecting digital capital deficits to dropout risk. We propose Digital Inequality Pathway Analysis (DIPA), a SHAP-enhanced machine learning framework integrating three innovations matched to these gaps: (1) a composite digital capital index combining financial resources and connectivity to capture threshold dynamics, (2) stratified nonlinear effect quantification revealing context-dependent returns and paradoxical high-access risks in vulnerable populations, and (3) individual-level SHAP decomposition uncovering how digital capital mediates between structural disadvantage and dropout risk through psychological pathways including relative deprivation and technology appropriation barriers. Experiments on 1,000 schools demonstrate that digital capital operates as a critical threshold mediator, with schools below 25% internet access facing 10%+ dropout rates while those above 62.5% maintain sub-8% rates, and that minority composition and low-income percentage channel digital inequality into educational failure through asymmetric, non-monotonic mechanisms resistant to simple connectivity interventions.