Digital Divide Analysis Using Computational Social Science Models

Abstract

The digital divide -- systematic inequality in access to, use of, and benefit from digital technologies -- is a multi-dimensional phenomenon whose causes, mechanisms, and social consequences resist simple measurement. Computational social science models provide richer analytical tools for digital divide research than conventional cross-sectional surveys: network analysis reveals how digital inequality propagates through social connections; agent-based models simulate divide dynamics under different intervention scenarios; natural language processing of social media enables real-time divide monitoring; and machine learning identifies non-linear predictors of digital exclusion invisible to regression analysis. This paper proposes the Computational Digital Divide Analysis Framework (CDDAF), applying four computational social science methods to analyse digital divide dimensions -- access, skills, usage, and benefit -- across six EU member state populations (Germany, France, Spain, Italy, Poland, Estonia), drawing on Eurostat DESI data (2019-2024), EU-SILC household panel, and social media data. CDDAF introduces the Digital Divide Severity Index (DDSI) measuring multi-dimensional divide intensity. Key results: benefit divide is 2.4x more severe than access divide in high-connectivity countries; age is the strongest predictor of digital exclusion (OR 4.2 for 65+ vs. 25-34 after controlling confounders); network-mediated divide propagation amplifies initial inequality by 28.4% over 10 years. The framework provides evidence-based digital inclusion policy guidance.