Top 3 Forecast: Scotty’s out or he wins

Contestant Today’s Min Today’s Max Today’s Avg Prior Average
(4 weeks)
Probability
of Elimination
Scotty McCreery 41 56 49 53.71 0.613758
Lauren Alaina 33 74 49 63.28 0.207999
Haley Reinhart 41 81 64 72.85 0.115363

Is there a plausible way Scotty can stay in the contest? Sure. Lauren had the worst performance of the night. Haley had a polarizing performance for her last song.

But I wouldn’t put much money on it. If Scotty McCreery is safe tonight, he is overperforming by an outlandish 40 percentage points. If he is not eliminated, basically, there are a huge amount of people voting for him regardless of the caliber of his performance. If that’s the case, there is no plausible scenario where he does not win the entire competition.

Basically, you can find opinions on all sides as to who’s going home. Dialidol (before accounting for the West Coast) calls for Haley. Votefair and Rickey.org think Lauren. DJ Slim and my model say Scotty by a mile.

So, it’s going to go this way. If Scotty gets through,  it doesn’t matter which girl does: he wins. If Scotty is (rightly) eliminated, Lauren and Haley will have a tough competition. Haley did herself a world of good today with her first performance. But Lauren would be a slight favorite to win.

Top 4 projection: Women on top

Disclaimer: The following analysis is more opinion than science. Moreover, it only projects what will happen if this season is like past American Idol seasons. Projected probabilities are just that, probabilities. Anything can happen.

Name Prior Average (4 rounds) Dialidol score Today’s Avg Probability of elimination
Scotty 53.333 24.337 50.5 0.30407495
James 58.666 14.711 57.0 0.21940829
Lauren 63.1666 15.676 63.5 0.14932903
Haley 71.666 11.954 63.0 0.06788917

Nobody tonight had anything close to a disaster. However, in an eyebrow raising move, James Durbin was placed in the statistically safest slots (first and last), had pyrotechnics, and sang a Journey song (previously not cleared for Idol).

Here’s what I think, and this is pure speculation. James isn’t doing nearly as well as people think. The week 2 rounds ago when the Bottom 2 was not revealed (only Casey was revealed to be gone), James was in fact the second-lowest vote getter. So, the model assigning him the second-most likely to go this week is anything but crazy. I buy it.

Whether or not I think Scotty will go home, if this season is like past ones, he should. My personal feeling, though, is that Lauren is in trouble.

Top 7 forecast

Contestant Order WNTS
Rating
Average of
last 3 rounds
Dialidol
score
Probability
of Elimination
Stefano Langone 6 34 45.7 4.72 0.22633591
Jacob Lusk 4 35 47.0 14.81 0.21236367
Scotty McCreery 1 29 57.3 24.71 0.16798150
Casey Abrams 5 53 56.7 7.67 0.10278103
Haley Reinhart 3 77 57.7 8.96 0.05709691
Lauren Alaina 7 60 67.66 12.65 0.05696180
James Durbin 2 66 70.33 12.55 0.04441594

I posted earlier this week about how good of a performance Jacob and Stefano needed to be safe.  Needless to say, they didn’t perform nearly well enough to make this a reality; in fact, the approval ratings put them lower than their previous 3 round average. Stefano Langone, by a hair, is the most likely to go, followed by Jacob. For the first time, Scotty is predicted to be in the bottom 3, though it would not be shocking if it was Casey. Haley, Lauren, and James are likely to be safe.

A bit of explanation is warranted at this point about how the model considers things. There is a significant effect of “momentum” in American Idol at this stage in the finals. This may not be surprising, but the most significant factor determining whether someone is eliminated isn’t how well he or she did that week, but how he or she has been doing recently. In this case, I included the average WNTS rating for the prior 3 weeks. This is not to say that this week’s performance doesn’t matter at all, but it isn’t as significant (either in odds or in statistical significance) as how well the contestant has been doing lately.

Dialidol’s predictions, I’m sorry to say, are just too unreliable right now to include in the model. Dialidol had a truly abysmal record in Season 9, and quite frankly, I’m not seeing that they’ve improved things too much. They have Jacob ranked second, and safe. I don’t buy it.

James Durbin and Lauren Alaina have emerged as the front-runners in the contest. Unless things drastically change, my guess is they will be in the Top 3. As to who is in it with them, your guess is as good as mine.

Another result for the record books

Let’s talk a little about how this never should have happened.

While I personally found it underwhelming, Pia Toscano’s version of “River Deep – Mountain High” rated an 85 in WNTS’s ratings. Dialidol had her at third highest ranked singer of the night, and safe within their margin of error (Dialidol is extremely cautious about calling someone safe). She had one performance before with an even higher score, a 91 for “I’ll Stand By You”. As I’ve noted, very few previous contestants who scored that high in the first few weeks fail to make it to the Top 4.

These scores aren’t theoretical. WNTS score and Dialidol score are statistically significant predictors of the outcome of any given round. While sex is also a major factor, a score like 85 is so high that the logistical model put Pia’s elimination chances at less than 1%.

We have to face some real questions:

1. Could the demographics for Idol be shifting enough to throw all previous years off? With an average viewership drop of only 3% from last year, I find this hard to believe.

2. Is there some systematic reason for the worsening condition of women? I certainly haven’t noticed anything regarding styling, or any other technical reason.

Finally,

3. Is Idol being honest about its results?

I hesitate to raise this question, but this is a major outlier. I’ve noted that in Season 9 Dialidol had some major sampling problems, but overall they are a highly reliable, scaling determiner of the disposition of a contestant’s performance. The average approval polling, however, evidences none of these problems. Pia had zero votes in Zap2It’s Exit Poll every week. Nobody ever thought she was gone, or even near enough the bottom to risk a bet on. It’s also been noted that ratings after a surprising result (e.g., Casey) raise the show’s ratings (about a 10% bump last week from the one previous).

Yes, sometimes a 22-year-old athlete non-smoker has a heart attack. But it’s extremely rare, and these events are getting to be not-rare enough that my eyebrows are firmly raised.

Top 11 post-game redux: assessing the model

The model was somewhat vindicated last night after a disastrous first week. I made significant improvements to the way it considered the data, and it correctly predicted the bottom 3. As I said, Naima was a sure thing, with Paul and Thia in a dead-heat. In the end, Thia was eliminated. What’s interesting to me is just how far off Dialidol was:

The bottom 6 according to Dialidol

On the one hand, Dialidol got 2/3 of the bottom 3. However, it significantly overestimated Naima, which my model had by a factor of 4 the most likely to be eliminated. I also get no sense that Lauren was possibly at risk, but Dialidol ranked her as fourth worst. Does anybody really believe that?

The WNTS approval ratings were quite different: Continue reading