Better luck next time, Bill Tancer, and congratulations, Jordin Sparks

I don't watch American Idol, and I don't really care who wins. But I am interested in the selection process, and more specifically the use of various tools to predict who will be voted off each week.

The most well known of these tools is probably DialIdol.com, which offers software people can download to speed dial votes for their favorite contestants and, more importantly, monitor the busy signal of the various contestants' phone lines to measure their popularity. This data is then collected and used to predict who will be voted off each week. Even though DialIdol only takes dialed votes, not text votes, into consideration, for the most part DialIdol has been pretty accurate.

Another method used to gauge contestants' popularity on American Idol, as well as other shows where audience votes determine the losers each week is by scouring through data of web searches for each of the contestants. An article posted yesterday on the website of Time magazine explains the basics of this, along with predictions for the Season 6 finale of American Idol. Here's the link to the article:

Advantage Blake Lewis by Bill Tancer

While I don't doubt the usefulness of using search engine data to predict winners and losers on audience participation television programs, I think that Bill Tancer's analysis is flawed, mainly because he's focusing solely on the search engine data and ignoring common sense. While the Season 6 winner won't be announced until later tonight, DialIdol's results are in, and they have Jordin Sparks receiving 50% more votes than Blake Lewis. Considering that Tancer's article is crafted mainly around the pitfalls of search engine data analysis, perhaps he could do the same thing next year about why he was so wrong about this season's American Idol finale. Here are a couple things he might include in his anaylsis.

First, the Stacy Keibler Correction Coefficient (SKCC) sometimes, but not always, is an important factor in audience participation television shows. Most of the people searching for her were men who probably didn't watch, and certainly don't vote, on Dancing With The Stars. If they do watch, it's because their wives watch, and when they saw Stacy Keibler, they went into the other room and searched for pictures of her on the computer. When their wives watched the show the following week, they stayed in the room and read the paper until she came on at which point they put the paper down and were glued to the set. Then when her performance was over they resumed reading the paper, or perhaps they once again went on the computer to search for some more pictures of her. Now, Stacy Keibler is an absolute bombshell who is much more attractive than most women, including most contestants on audience participation programs. Still, most contestants on these shows are still quite attractive, such as Jordin Sparks, but she's certainly no Stacy Keibler.

So perhaps next year when considering search engine data to predict the American Idol or Dancing With The Stars winner, Tancer can use the Modified Stacy Keibler Correction Coefficient (MSKCC), which says that the usefulness of search engine data for a certain female contestant is inversely proportional to her attractiveness. When a contestant is so exceptionally attractive that people who would otherwise have no interest in the program start doing searches for her, as was the case with Stacy Keibler, the search engine data is practically useless for prediction purposes. Conversely, when the attractiveness of the contestant is more in line with the attractiveness of the other contestants, and while still attractive, is not enough to garner a significant number of searches from those who don't watch the show, then the search engine data should be pretty useful, even if those searching are more interested in their looks than their talent relevant to the show they're on. Without even looking anything up, it seems pretty likely that the majority of Dancing With The Stars viewers are adult women, while the majority of people doing searches for Stacy Keibler are men, of all ages. In contrast, I'd think the American Idol demographics skew younger, with more teenagers watching and voting on the show. Just because they're more interested in the various contestants looks than their singing doesn't mean they're not voting for them. Indeed, consider how many singers and bands over the years have become millionaires off the disposable incomes of teenagers used to buy their CDs, even if they don't particularly care for their music.

Second, while search engine data undoubtedly holds clues on how people will vote, it must not be used to the exclusion of common sense. Amazingly, Tancer's article doesn't even mention the 3rd place finisher, Melinda Doolittle. Does Tancer think that now that Melinda has been eliminated that those who voted for her are now simply irrelevant and those that voted for her every week will simply not watch or vote on the Season finale? Imagine if in the 2000 presidential election a few weeks prior to the election Al Gore convineced Nader to drop out, perhaps by promising him the opportunity to head the EPA if he won. If that happened, those who had been Nader supporters would have been anything but irrelevant. Last year on American Idol, after Chris Daughtry and Elliott Yamin were eliminated, those who had voted for them broke towards Taylor Hicks in the finale, making him the winner. Similarly this year, those who voted for Melinda previously should be expected to break towards Jordin. Indeed, if you look at this week's DialIdol data compared to last week's DialIdol data, it appears that Melinda voters broke for Jordin 3 to 1, and that Jordin will be named the new American Idol by a pretty comfortable margin.

Better luck next time, Bill Tancer, and congratulations, Jordin Sparks.