Geographic Patterns and Socio-Economic Influences on Internet Use in U.S. States: A Spatial and Multivariate Analysis
26 Pages Posted: 30 Mar 2017 Last revised: 16 Aug 2017
Date Written: March 28, 2017
Discourse and interest in the digital divide research community is steadily shifting beyond access and adoption to utilization, impact, and outcomes of information and communications technologies (ICTs), particularly the internet. In the United States, studies and surveys conducted by the National Telecommunications and Information Administration (NTIA) indicate increase in internet use in every corner of the country over the last two decades. However, recent surveys on ICT use indicate significant disparities in dimensions of internet use. For example Americans’ use of the internet to pursue e-education, e-health, e-commerce, e-entertainment, and telecommuting has varied significantly – longitudinally as well as geographically. Additionally, internet use habits are rapidly expanding, providing new insights into the emerging internet of things, wearable technologies, and new forms of social media usage. As novel technologies and lifestyles emerge, analysis of new disparities and dimensions of the “usage digital divide” stemming from social, economic, societal, and environmental factors becomes important.
This research examines spatial clusters, geographic disparities, and socio-economic dimensions of existing and emerging dimensions of internet use among the 50 U.S. states. We adapt the Spatially Aware Technology Utilization Model (SATUM) for internet use by positing associations of 20 independent demographic, economic, infrastructural, affordability, innovation, societal openness, and social capital variables with 17 indicators of internet use spanning e-education, e-commerce, e-health, e-education, telecommuting, and emerging forms of internet use. Data on the 17 indicators of internet use are sourced from the July 2015 CPS Supplement on internet use from the U.S. Census. Data on traditional independent correlates are sourced from the same Supplement, U.S. Census of Population, U.S. Economic Census, while data on societal openness, social capital, infrastructure correlates are collected from George Mason University’ Mercatus Center, FCC’s National Broadband Map initiative, and noted political scientist Robert Putnam’s publicly available data on civic engagement.
First, descriptive mapping provides important visual cues about patterns of internet use in U.S. states. Subsequently, K-means cluster analysis of multiple internet use-related factors is conducted to determine agglomerations of states that are most similar in patterns of internet use and outcomes. Subsequently, statistically significant “hotspots” and “coldspots” of internet use and outcomes among U.S. states are identified, followed by spatial autocorrelation analysis of various dimensions of internet usage. A-priori diagnosis of spatial autocorrelation is critical to understand and possibly account for the presence of spatial bias while examining social, economic, societal, and environmental underpinnings of internet usage. Regression residuals are mapped and examined for spatial autocorrelation.
Systematic examination of rapidly evolving dimensions of internet use among U.S. states distinguishes this work. Novelties include thorough analysis of disparities stemming from geography, results showing socio-economic, infrastructural, affordability, civic engagement, and societal openness determinants of the internet “usage digital divide,” and longitudinal analysis of change dimensions. A methodological novelty is diagnosis of spatial autocorrelation of internet use, largely ignored in digital divide literature. Left undiagnosed, this can potentially bias regression-based associations of independent variables associated with internet usage. Finally, the findings of this work have critical policy implications at a time when expanding and stimulating greater variety and intensity of internet use and impacts are well-recognized as aspirations of state and federal policies.
Keywords: Internet Use, Digital Divide, Social Capital, Regression, Spatial Autocorrelation
JEL Classification: C21, C23, L86, L96
Suggested Citation: Suggested Citation