Please cite use of this data as:
Tolbert, Caroline, and Mossberger, Karen, 2015, “U.S. Current Population Survey & American Community Survey Geographic Estimates of Internet Use, 1997-2014.”
The county-level estimates between 1997 and 2012 available here are unique, as county-level estimates were not provided by the U.S. Bureau of the Census until the 2013 American Community Survey. Additionally, we provide more detailed estimates of county Internet use for 2013 and 2014, including estimates for demographic groups within the counties.
Because counties include both rural and urban areas, there are important differences apparent across counties. Rural counties tend to have experienced later growth in broadband at home, and many still lag behind in broadband and mobile. The graphs below show summary figures for the county data. The data and graphs facilitate comparisons across counties, within counties over time, and for rates of change across counties. Maps also show regional patterns as well as overall differences.
Using Summary Tables of the 2013 and 2014 American Community Survey available in FactFinder, County estimates were created for home Internet access and home broadband by race, ethnicity, education, age, and employment status.
Using microdata from the 2013 American Community Survey, County estimates were developed for home Internet access, home broadband, mobile Internet, and fully connected households broken down by race, ethnicity, education, age, family income, and language skill.
See Codebook here for a more complete description of the datasets, sources of data, survey questions, and methods. CodebookMethodsNSF_BB.pdf
Summary figures (graphs) for counties are available here. These show by county and by overall county averages three-year moving averages, one-year time series estimates, and rates of change for Internet access, broadband at home, mobile use, & fully-connected households. County_Figures.pdf
U.S. County Total Internet Access 2013-14 (user chosen variables for maps) - choices include broadband access and home access for total population, white, black, Latino, Asian, education, age, employed/unemployed
Users can access additional information by rolling the mouse over areas of the map in Tableau. Regional patterns are apparent, with lower rates of Internet use in the Midwest, South, and Southwest, especially in rural areas.
This County Time Series data set contains the yearly estimates of the percentage of Internet use in U.S. counties along with their populations in 2014. The percentage of home Internet access goes back to 1997 from 2014, home broadband to 2000 from 2014, and mobile Internet and fully connected household to 2011 from 2014.
The CPS data is disaggregated with household weights to create the Internet estimates for 1997, 1998, 2000, 2001, 2003, 2007, 2009, 2010, 2011, and 2012. The ACS summary tables of metropolitan areas from American FactFinder provide the Internet estimates for 2013 and 2014 and the estimates for the missing years are imputed via linear interpolation.
This County Three Year Average data set takes the three-year moving average of the Internet estimates. The data set contains the yearly estimates of the percentage of Internet use in U.S. counties. The percentage of home Internet access is estimated between 1998 and 2013, and home broadband between 2001 and 2013. Because of the very short time series, three-year moving averages are not estimated for mobile Internet and fully-connected household.
This County Time Services Change Rate data set is based on CountyTimeSeries.xls, and it estimates the yearly rate of change in the percentage of home Internet access, home broadband, mobile Internet, and fully-connected households in U.S. counties.
This County Three Year Average Change Rate data set is based on countyThreeYearAverage.xls, and it estimates the yearly rate of change in the percentage of home Internet access and home broadband in U.S. counties.
This County Yearly Summary Statistics data contains the summary statistics of the percentage of Internet use in counties across years. It summarizes home Internet access, home broadband, mobile Internet, fully connected household for single-year county estimates (countyTimeSeries.xls), and home Internet access and home broadband for three-year moving averages (countyThreeYearAverage.xls).
This County 2013 ACS Fact Finder data set contains the 2013 estimates of the percentage of Internet use in U.S. counties along with their 2013 populations. It is based on the summary tables of 2013 American Community Survey from American FactFinder. For home Internet access and home broadband, the estimates are broken down by race, ethnicity, education, age, and employment status.
This County 2013 ACS Microdata data set contains the 2013 estimates of the percentage of Internet use in U.S. counties. The estimates are generated by disaggregating the 2013 American Community Survey 1-Year Public Use Microdata Sample which was obtained through Minnesota Population Center. For home Internet access, home broadband, mobile Internet, and fully connected household, the estimates are broken down by race, ethnicity, education, age, family income, and language skill.
This County 2014 ACS Fact Finder data set contains the 2014 estimates of the percentage of Internet use in U.S. counties along with their 2014 populations. It is based on the summary tables of 2014 American Community Survey from American FactFinder. For home Internet access and home broadband, the estimates are broken down by race, ethnicity, education, age, and employment status.
NOTE: The Catalist data posted here as a measure of broadband for 3,000 counties is a “high technology household indicator” which includes personal computers and a home broadband subscription. It does not include mobile internet use, as the American Community Survey does. This is a 1% sample of commercial data blended with modeled data, for a population of 240 million individuals.
We make this secondary data available because of gaps in the census data, which covers only 820 of the 3,000 counties in the U.S. As discussed in more detail in this document, we do not have exact information about how this commercial data was collected or modeled, and so this should be used to indicate patterns of difference across counties that are not visible in the census. This should not be used as an exact measure of broadband adoption in individual counties.
Use of Census data to estimate Internet access for local communities, such as counties smaller than 60,000 people and urban neighborhoods is difficult, as the Census does not release geographic identifiers for respondents residing in these local communities in their public use files for confidentiality reasons. Very small sample sizes make it difficult to estimate Internet access for sparsely populated rural counties or urban neighborhoods.
Catalist is a new national dataset available for commercial purposes that provides an opportunity for creating local geographic estimates of Internet use. This project used a 1% sample of Catalist data (www.catalist.com) of 240 million Americans for 2015-16. These data provide county level estimates reported below.
The Catalist effort grew out of President Obama’s 2008 presidential campaign and includes data on 185 million registered voters of all party affiliations—a unified national voter file—from Election Officials in all 50 states and the District of Columbia. Commercial data adds in another 55 million unregistered individuals. Census data, consumer data, specialty data and media market data enhance the database with hundreds of fields, including household attributes, purchasing profiles, donation behavior, occupational information, and Internet access and media consumption and even engagement with civic and community groups. http://www.catalist.us/data/
The Catalist database has information on 240 million unique voter age citizens (voters and non-voters, Democrats, Independents and Republicans). As part of the academic subscription we purchased the 1 percent sample from this file includes hundreds of variables (for nearly 3 million people). The advantage of these data is the ability to create geographic estimates for small geographic areas, such as rural counties.
We use the 10 questions on Internet access for this part of the study (home broadband access, early Internet adopter, heavy Internet user, and measures of online activities, including online banking, consumer purchases, travel, occupation, etc.) to predict Internet use and skills for all 3100+ U.S. counties using simple disaggregation (and in the future, urban neighborhoods). The Census Factfinder only provides data for 820 counties with populations of 65,000 or greater. Using simple disaggregation, Figure 11 presents the estimates of the percent of the population with a home computer and home broadband (measured by a subscription with an Internet service provider) using Catalist.
County level interactive Tableau map of a Home Computer and Home Broadband (ISP) - rollover with mouse in Tableau to see individual data for counties.
The figure above shows expected patterns, with sparsely populated rural and Western counties with the lowest access rates, as well as poor and high minority areas. The Catalist data also allow us to create propensity scores of engaging in online activities for the nation’s 3100 counties. Figure 12 maps one of many variables—the propensity of the population to use online banking. Higher values indicate places where residents are more likely to be engaged in online banking, purchase products online, or be heavy Internet users. Such variables can serve as a more nuanced measure of Internet use beyond simple access or home broadband adoption.
County level interactive Tableau maps with multiple online activities by county - rollover with mouse in Tableau to see individual data for counties.
The above figure is a map of the propensity of county populations to purchase airline tickets online, use Amazon, purchase books online, use the Internet for purchases, be an early adopter of the Internet or be a frequent user of the Internet.
County level interactive Tableau maps of Internet access seen above (2013-2014 ACS) - rollover with mouse in Tableau to see individual data for counties.