Is the Internet Representative Enough for Online Qualitative Research?

“BUT, does the Internet really provide a representative sample?”

Ten years ago, this was a constant criticism of online research. Today it is a mere echo, but Internet accessibility remains a concern for some researchers. According to a new Census Bureau report 90-95% of Americans have access to broadband Internet that is fast enough to “handle downloads of Web pages, photos and video or simple video conferencing services.” Also, 68% of households with access to broadband actually subscribe. Therefore, according to the Census Bureau, 61% to 65% of all households have broadband access. (See map of broadband access at right). Clearly, this percentage should continue to rise.

As one would expect, broadband availability is correlated to income. Therefore, online qualitative is less representative for lower income households but is very representative for median income households and above. Indeed, most of the households without access to broadband are poor and/or rural. Otherwise, broadband Internet penetration is strong and samples can be, and are, representative of most populations.

The Obama administration has announced a new initiative to increase broadband penetration. This initiative will only improve the ability of online qualitative research to reach the right respondents.

3 Comments
  1. Do you think that representativity is really what matters? I’m included to think that if your predictions are consistently accurate, your incoming data can be completely unrepresentative. Validity in outcomes all the way.

  2. Obviously validity is much more important than representativeness. However, we hear questions about representativeness from people regarding the internet on a fairly consistent basis (though less than we once did). The point of this post was to show that Internet penetration and, importantly, broadband penetration is getting pretty high. Therefore, the danger of “adopters” being attitudinally different from non-adopters is greatly diminished from what it was just a few years ago.

  3. Jim, I don’t the issue about representativeness has much to do with who has a phone or who is online, as it does with who is would even participate in a phone or an online survey. Self-selection is where the real bias lies.

Leave a reply