Paper LinkEncouraging low-income Americans to subscribe to home broadband service has proven to be an uphill battle for broadband proponents. While the monthly cost of broadband service is one obvious barrier to adoption within this population, other factors also play a role. Some may lack the necessary digital literacy skills, while others may lack a home computer or rely on other resources like libraries or computing centers for their broadband access. It is therefore important to measure barriers to broadband adoption within this portion of society, as well as determine what steps would have the greatest impact on closing the digital divide in a given geographic area.
This study attempts to measure the marginal impact of various behavioral and demographic variables and their impacts on home and mobile broadband adoption. In addition, those marginal impacts will be used to build a model by which policymakers can estimate the number of low-income households in a given geographic region that do not subscribe to broadband. This procedure will allow policymakers to estimate the number of low-income non-adopters in a given area that might be more responsive to price incentives (potential Lifeline discounts), the number that may require digital skills before adopting (regardless of price incentives), and the population to which more aggressive outreach may be needed beyond price and skill training. The procedure would allow for proper sizing of these various contents of a broadband adoption toolkit.
This study relies on multiple data sources. Early data collected through the Lifeline Pilot project sheds light on low-income respondents and their decision-making process when offered various incentives to subscribe to home broadband service. In addition, using a rich dataset collected through random digit dial telephone surveys with 8,442 low-income respondents across eight heterogeneous states (Iowa, Michigan, Minnesota, Nevada, Ohio, South Carolina, Tennessee and Texas) between 2010 and 2014, this study uses a logistic regression model to predict home and mobile broadband adoption decisions among low-income households. The regression models use binary dependent variables that indicate whether or not the household subscribes to home broadband service, and whether the household uses mobile broadband. The models incorporate independent variables for demographic factors such as the urban/rural location of the household, homeowner age, race, ethnicity, employment status, disability status, and the presence of children in the home. In addition, these models incorporate behavioral factors such as the presence of a computer in the household and use of the Internet at locations outside of the home (such as at work or a library) as independent variables.
The Lifeline Pilot Project data, coupled with early applications of the model, suggest that in addition to the demographic factors that have been shown to have a significant marginal impact on both home and mobile broadband adoption, behavioral factors that had not previously been incorporated into such models also have a significant impact on a low-income individual’s decision whether to subscribe to home or mobile broadband service. This combination of demographic and behavioral factors help provide a more robust image of the low-income non-adopter, allow researchers to estimate the number of such non-adopters within a given area, and provide policymakers design solutions that will address the lower adoption rates within this subset of the population in a more cost effective manner.
