Analyzing county-level adoption disparities between low-and upper-income households
Bowling Green, Ky. (May 6, 2025) - The American Community Survey (ACS), run by the U.S. Census, provides us with a treasure trove of data about the American public. Among that data, we have information about household internet adoption, broken down by various demographics, socioeconomic indicators, and technology types. Exploring and analyzing this data can be invaluable in understanding the needs of communities and what sorts of interventions could help close the Digital Divide.
For this exercise, I chose to dive into the household internet adoption data at the county level, specifically disaggregated by household income. I rely on the 2019-2023 five-year estimates from the ACS. At the household level, the survey estimates the overall number (and percent) of households without a home internet connection and broken down by household income – those earning under $20,000 per year and those earning over $75,000 per year – then aggregates these numbers for different geographic areas. With this information, I calculated what Roberto Gallardo calls the internet income ratio (IIR) for each county, which divides the percentage of households making under $20,000 a year without a home internet subscription by the percentage of households making over $75,000 a year. This metric approximates how likely a household making over $75,000 a year is to have home internet compared to a household making less than $20,000. I also rely on the ACS estimates of the percentage of households without a home internet subscription at the county level.
Looking at these two metrics reveals a humbling picture of how much farther we need to go to close the Digital Divide in this country. At the national level, 10.1% of households do not subscribe to home internet service. Low-income households (making less than 20k per year) are less likely to have a home internet connection (27.9%) than high-income households (making more than 75k per year), of which only 3.8% lack a connection. These numbers indicate an IIR of 7.34, meaning high-income households are 7.34 times more likely to have a home internet connection than low-income households.
The scatterplot below graphs these two metrics at the county level, with dashed lines indicating the national averages. This depiction shows us where the counties fall compared to the country overall.
In the chart above, light blue dots depict counties where the IIR is higher than the national average but the non-adoption rate is lower than the national average – meaning most people are connected, but there are large disparities in adoption rates based on income. About 10.7% of counties fall into this group. The dark blue dots illustrate counties where the non-adoption rate is higher than the national average but the IIR is lower than the national average. These counties face challenges getting online and do not face large disparities depending on household income. The majority of counties (61.1%) fall into this group. The red dots show counties where the IIR is higher than the national average AND the non-adoption rate is higher than the national average. This group (19.1% of all counties) faces some of the greatest challenges getting households online. Finally, the light grey dots represent counties where the IIR and the non-adoption rate are below the national average; only 9.1% of counties fall into this camp.
The map below illustrates the four groups from the scatterplot across the country.
By and large, urban counties are colored either light grey or light blue – indicating that their non-adoption rates are lower than the national average. Some notable exceptions include Cook County, IL (Chicago), Bernalillo County, NM (Albuquerque), Milwaukee County, WI (Milwaukee), Wayne County, MI (Detroit), Philadelphia County, PA (Philadelphia), and Baltimore (city), MD. Meanwhile, most rural counties are colored either dark blue or red – indicating that their non-adoption rates are higher than the national average. This split reflects the broader divide between urban areas with robust broadband infrastructure and rural areas with less access. IIR varies considerably across the country, but some trends emerge, mainly among counties with lower non-adoption rates. Much of New England is light blue, indicating larger disparities in internet adoption among low and high-income households. Other metropolitan areas with higher cost of living also follow this trend, including Seattle, WA, Portland, OR, and the Bay Area of California.
More than anything, this exercise illustrates that we still have a long way to go in closing the Digital Divide across the United States. Significant gaps in internet adoption persist – both within and between counties. Highlighting the data allows us to see which areas have the greatest need. As we continue to expand access in the coming years, collecting and analyzing data about internet adoption and usage will be critical to ensuring that investments are effective in getting households online.
Everyone belongs in a Connected Nation.
About the Author: David Nunnally is a Connected Nation Research Analyst. David is responsible for using qualitative and quantitative techniques to interpret survey data, in addition to collecting data from secondary sources to help support those findings. David works with internal and external stakeholders to help develop research and provide critical information in support of the Connected Nation mission.