Election Aftermath: Can Internet Kill the Telephone Poll?
Aside from the candidates, the big stars of this election cycle were the pollsters — most notably Nate Silver of the FiveThirtyEight blog.
By aggregating polls and adding his own special sauce, Silver correctly predicted the electoral outcome in all 50 states. But you know who else came close? SurveyMonkey.
SurveyMonkey is a Silicon Valley company that specializes in helping individuals and businesses gather data.
The company says it can reach more than 10 million respondents with its surveys. CEO Dave Goldberg and his team, in an effort to test the chops of their new SurveyMonkey Audience product, decided weeks ago to see how good it would be at predicting election results.
In the end? Pretty good.
SurveyMonkey says the raw data from its surveys nailed 48 out of 50 states, missing only Florida and Ohio (two of the closest swing states). Of course, if Florida and Ohio had gone Republican, the Electoral College would have been a bit of a nail-biter at 285 for Obama and 253 for Romney; in the end it was 332 to 206. (Read More: 'Government Surveillance Is on The Rise' )
"We get 60 million completed responses a month, about half of those are in the U.S.," Goldberg said. At the back of other people's surveys, "We said thank you for taking the survey, and we'd like to ask you a question or two about the election."
Still, I found the idea of accurate online polling to be fascinating. After all, we've traditionally thought of Internet access as skewing toward the affluent. But now that PC penetration is so high — and there's probably even more correlation between being a likely voter and having Internet access — could online surveys beat the phone.
Goldberg thinks it's possible. He points out that SurveyMonkey's election surveys got a 30 percent response rate, far better than the nine percent that phone polling gets these days. I asked him whether he's thinking about selling this data, maybe creating a new revenue stream?
It's too early to talk about that, he said. So far, he's looking at it as proof that the company's new Audience tools can deliver good data when it matters.