Conference Agenda

This paper investigates how the quality and reliability of internet access—conceptualized as under‑connectedness (Katz, 2017)—shape patterns of internet engagement in the United States. While prior research has examined under‑connectedness primarily among lower‑income households (e.g., Katz, 2017; Katz & Gonzalez, 2016), our study is the first to analyze the phenomenon using a nationally representative dataset.

Research Questions

We examine two research questions:

RQ1: What are the demographic characteristics of under‑connected respondents in a representative sample of the U.S. population?

RQ2: How do under‑connected respondents differ in terms of their internet use and online activities?

These questions extend digital divide scholarship by examining qualitative dimensions of access—speed, reliability, device quality, and device availability—rather than relying solely on binary measures of access or device ownership.

Methodology

We used the November 2023 Computer and Internet Use Supplement to the Current Population Survey (CPS), a nationally representative dataset of 36,648 U.S. internet users. The CPS includes five items closely aligned with established under‑connectedness measures (Rideout & Katz, 2016), including whether devices “work poorly,” whether internet service meets needs for “speed and reliability,” and whether respondents “temporarily lost an internet connection due to difficulty paying.” Respondents were classified as under‑connected if they reported at least two such problems.

We ran descriptive statistics to compare connected and under‑connected groups, followed by logistic regressions to predict under‑connectedness. As we were interested in how under-connectedness shapes types of internet use and how much online activity respondents engage in, we ran a principal components analysis (PCA) of 23 internet‑use items to identify seven types of online activities: commerce, medical, entertainment, communication, providing services, smart‑device use, and job‑related uses. We measured variety of use by summing the number of online activities performed (0–23).

Disciplinary Fields

The research draws on and contributes to three disciplinary fields. First, we draw on Communication and Internet Studies, specifically digital inequality as our theoretical framework. Second, we use a Sociological lens with a focus on stratification, marginality, and resource disparities. Third, our research is relevant to Public Policy, particularly broadband policy, digital inclusion, and equity‑focused infrastructure planning.

Novelty and Policy Relevance

The paper’s novelty lies in its use of a nationally representative dataset to evaluate under‑connectedness, a concept previously examined only in lower‑income populations. This broader lens reveals that under‑connectedness is not confined to the poorest households, though it remains strongly associated with socioeconomic marginality.

For communications policy, the findings challenge the assumption that near‑universal broadband availability signals the end of the digital divide. Even when households technically “have access,” qualitative deficiencies—slow speeds, unreliable devices, service interruptions— continue to shape meaningful online participation. This underscores the need for policies that move beyond binary access metrics and incorporate quality‑of‑service standards, affordability, and device‑maintenance support.

Results and Conclusions

We find that 9% of U.S. internet users are under‑connected. These individuals are more likely to be Hispanic, unemployed, have less education, and lower incomes. They engage in fewer online activities overall—1.4 fewer on average—and use the internet less for commercial, medical, entertainment, and communication purposes. They also use fewer smart devices.

However, the effect sizes are modest. Under‑connectedness matters, but less than expected. We propose two explanations: First, attitudes toward technology—absent from CPS data—may play a role in explaining variation in internet use variety and different types of internet uses (e.g., Blank, 2013; van Dijk, 2020). Second, some individuals do not value the internet highly so they are not willing to learn how to use it, even though they are required to be online to apply for jobs or government assistance (Allmann & Blank, 2021).

Under‑connectedness is an important but not dominant factor in digital inequality. Structural barriers—income, education, and broader digital divide factors—remain the primary drivers of unequal internet use.