In 2013, more than half of U.S. households with incomes below $20,000 still lacked any kind of home Internet subscription including mobile or dial-up, according to the U.S. Census American Community Survey, or ACS.

The ACS showed much lower percentages of households without Internet in higher income groups.

All told, the ACS data tell us that out of about 30 million U.S. households without home Internet in 2013:

  • more than 11 million had incomes below $20,000
  • more than 18 million had incomes below $35,000.

The nation’s median household income — the income at which half of all households made less and half earned more — was $52,250.

We know that Americans without home Internet give a number of explanations, and cost is only one of them. But looking at the Census results, the strong association between lack of connection and lack of disposable income becomes pretty obvious.

The American Community Survey, a very large annual sample of U.S. households, is the Census’ definitive source of data on Americans’ social, economic and housing conditions. The 2013 ACS was the first to include several questions about household computer ownership and Internet access. Results from these questions, included in ACS data released last Fall, provide the most comprehensive picture to date of Americans’ home Internet access — for the nation as a whole, for states and counties, and for “places” (cities and towns) with 15,000 households or more.

While sub-$20,000 households are only about 18% of the national total, they constitute much bigger percentages in many communities… and the impact of digital disconnection on those communities is correspondingly greater.

In Cleveland, for example, where more than 40% of households had incomes below $20,000 in 2013, the ACS data shows 36% of all homes lacking any kind of Internet subscription, including mobile or dialup.

Other cities with very high levels of poverty — Detroit, Miami, New Orleans, Newark, etc. — also had “no home Internet” rates of a third or more.

The 2013 ACS doesn’t provide its computer and Internet data at the Census tract level, so we can’t use it to compare adoption rates for poorer vs. better-off neighborhoods within cities. Thus may change once the Census gets several years of responses to the new questions.