April 29, 2010 | By Eric Sorensen | 2 Comments »
Categories: Performing arts, Sociology
Tags: American Idol, Danna Moore, Eric Sorensen, Facebook, Idol Fanatic, John Tarnai, polling, research, Science, SESRC, Social and Economic Sciences Research Center, Sociology, Washington State University, WSU
Loyal readers will recall the Discovery blog last week extolling the wonders of a Facebook application accurately predicting the losers of “American Idol” for three weeks straight. John Tarnai and Danna Moore, leaders of the Washington State University Social and Economic Sciences Research Center, explained that this made sense. Whether they voted by phone during the show or by web on a Facebook fan page, said Tarnai and Moore, the survey’s self-selected participants are coming from the same highly motivated pool of “Idol” fans.
Now comes the upset.
Based on 1,700 votes cast through Facebook’s Idol Fanatic fan page, this seemed like the outcome after Tuesday’s Shania Twain-themed performances:
Crystal – 30.9%
Lee – 19.9%
Casey – 14.4%
Aaron – 13.6%
Siobhan – 12.5%
Michael – 8.6%
But on Wednesday’s results show, the bottom three were Casey–not Aaron–plus Michael and Siobhan. Siobhan–not Michael–was voted off.
We asked John Tarnai if the difference between Michael and Siobhan, or Casey and Aaron, might be accounted for by the survey’s margin of error.
“This is very interesting, especially that the bottom three on the show were also part of the bottom four on the Facebook survey. Unfortunately, since neither the Facebook sample nor the voting on the show is from a probability sample, you cannot use the margin of error to explain the difference. The margin of error is only accurate for probability samples. For self-selected samples such as Facebook and the Idol show, there is no way to assess the true margin of error.
“However, if you did use margin of error, the appropriate one to use would be the total 1,700 vote sample, or plus/minus 2.3 percent. It’s possible that the voting on the show came closer to the actual performance than the Facebook votes, which may have caused some people to switch their votes. That’s my take on it.”
If it’s any consolation, the Idol Fanatic-meter proved more reliable than the dozen prognosticators tracked by the Los Angeles Times “American Idol” Buzzmeter. Only one of them, Steve Gidlow of In Touch Weekly, saw Siobhan heading for the door. Eight predicted Michael would be gone.
Update: One possible explanation for the Siobhan upset is a snafu in which the wrong phone number to vote for her was posted on her Facebook page. It turns out those votes went to Aaron. The error has launched a petition drive to get her reinstated. Chances are justice will come in the form of Aaron getting voted off after Sinatra night. Here are the latest numbers: