August 29, 2013

How Pew Research calculates broadband adoption

The Pew Research Center released its latest numbers on home broadband adoption this week, reporting that seven in ten American adults now have a broadband connection at home. The report also found that one in ten Americans own a smartphone but lack home broadband.

Since the report was published, we’ve received a few questions about how we measure and report broadband adoption in America.

How do you define “broadband”?

Our broadband question has historically tried to distinguish between dial-up users and those with higher connection speeds. It gives a general measure of “broadband” users in the United States, and tracks pretty well with government data on broadband adoption.

Because our research is based on nationally-representative phone surveys, we rely on self-reported data—in other words, we just ask. Specifically, we ask adults who say they use the internet at home: “At home, do you connect to the Internet through a dial-up telephone line, or do you have some other type of connection, such as a DSL-enabled phone line, a cable TV modem, a wireless connection, or a fiber optic connection such as FIOS?”

We have asked about home broadband since 2000, and have asked a version of this question since 2003. This allows us to see changes in how people connect to the internet during the last decade, as shown in the accompanying chart.

As of May 2013, 70% of American adults ages 18 and older say they have some type of broadband connection at home, and 3% say they have dial-up. (Note: We suspect that many “wireless” respondents in our home broadband number include people with wired DSL or cable connections via wireless routers, as opposed to strictly “wireless” satellite or Aircard users.)

What type of connection speeds do you measure?

The short answer is: We don’t. It is often very difficult for people to answer detailed questions about their internet connection type and speed. In 2010, for instance, we asked about “premium” vs “basic” service, but about 13% of broadband users weren’t sure of what type they had. When we’ve asked specifically about broadband users’ connection speed, the vast majority (about 90%) of respondents said they did not know.

One source of detailed data on this topic is the National Broadband Map, an interactive from the National Telecommunications and Information Administration (NTIA). It has lots of information about types of connection, broadband speeds (including advertised speeds, typical speeds, and the results of speedtests), and general availability by state, county, metropolitan area, and more.

Do you count Americans who rely on smartphones for home internet access as “broadband users”?

No. There is no widespread consensus as to whether 3G or 4G smartphones qualify as “broadband” speed, and many would question whether they offer the same utility to users as a dedicated home internet connection—activities such as updating a resume, filing taxes, or viewing educational content are certainly more challenging on a smartphone operating over a cell phone network, than on a broadband-connected home computer. For these reasons, smartphones are qualitatively distinct enough that we do not include them in our standard definition of what constitutes a “broadband user.”

In other words, when we write that 70% of adults have home broadband, that does NOT include the 10% of adults who do not have broadband but do own a smartphone.

Topics: Broadband, Internet Activities

  1. Photo of Kathryn Zickuhr

    is a Research Associate at the Pew Research Center’s Internet & American Life Project.


  1. Todd O’Neill4 years ago

    Without asking about, or determining, the actual speed of the broadband how do you ascertain that these people have broadband?
    How was mobile broadband determined? Are you referring to smartphone based broadband or technologies like WIMAX?
    Were your respondents low tier customers, high tiered customers, 3G or LTE customers, and from what companies or carriers?
    If “wireless” broadband were removed from the equation (33% as I read it) then it looks like wired, broadband penetration is 37%.
    It doesn’t seem fair, at this time, to compare wireless broadband technologies since the speed comparison between wired/fiber and wireless/mobile will spread in the coming years especially in light of a finite set spectrum allocations.
    I appreciate the research that Pew does but this seems to be missing some key data points for it to be held as currently fully accurate.

    1. Kathryn Zickuhr4 years ago

      Hi Todd,

      I believe that most of your questions are addressed in the post, but ultimately it boils down to how you define “broadband.” Many people define broadband differently, and sometimes these definitions change over time. In the early days of the Pew Internet Project, we focused on the differences between broadband users and those who went online via dial-up connections, and “broadband” was widely understood to be one of the new types of services that offered a faster connection than 56-kps modems. When we discuss broadband generally, most people still understand broadband by its relation to dial-up or by their knowledge of the enterprise to which they pay their monthly bills.

      Today, as broadband users vastly outnumber those with dial-up, we continue to measure the broad outline of types of internet access. You and others are correct that we don’t factcheck our respondents or independently assess their internet speeds. That would take more than double our annual research budget for everything we do at Pew Internet and just isn’t feasible. We take respondents’ word as they describe the type of connection they use. Of course, we know there are major variations in broadband connection types and speed, however, as the NTIA’s National Broadband Map demonstrates. However, that’s a different discussion that is best addressed with a different set of methodologies from ours. For now, we are still using the widely understood term “broadband” as a general measure of a broad category of connections and we plan to continue to gather trends this way until another broadly understood term comes into general use.

  2. bill kingsbury4 years ago

    Wow! increasingly pew is exposing themselves ad doing sloppier and sloppier polling!
    first it was the (too early) poll on citizens who don’t mind giving up personal info (regarding nsa spying), but the fallout of personal postings on sites like facebook haven’t reverberated until they graduate college and try to get a job with that beer party picture haunting them. (eg:if aspiring politicians can be swiftboated by digging up 30 year ago past pronouncements, imagine how much will be available if our 5th graders start providing damning evidence from now).
    On the poll of broadband, while some is attributable to lack of adoption by elderly, by FAR the most salient determinant is the high cost of internet access fees. Lack of adoption is more highly correlated with household SIZE than age or race. At $50/mo, it’s easier to amortize the cost in a household with 2 adults and 3 children, than it is with a lone adult. Have a read of david cay johnson’s book on how s korea has more broadband access at LOWER COST and vastly outshines “the great inventors and innovators”.

    1. Mark3 years ago

      To be fair, S Korea is a vastly smaller and more densely-populated country than the US. More reasonable comparisons might be to Canada, Australia, and to a lesser extent, the BRIC countries.

      Also, when comparing cost, it very much matters how a person gets their broadband. Most, I suspect get it via their cable provider. If so, the cost is likely bundled up with TV and maybe phone service. Pinning down how much broadband costs in that bundling is impossible. Additionally, if one gets DSL via a company like NetZero, while the cost may be low but you separately pay for a phone line.

      Finally, comparing costs is tricky because you have to account for both what the advertised Mbps for the price is, but also what the actual delivered speed is (which depends on many factors). I ran a test on my supposed 105 Mbps internet this morning and got a reading of 24.