Top 25 Projections

As irritating as it is, I was not able to include pre-exposure in the numbers to come. In the end, Dialidol and WNTS scores have better statistical significance, and if I have to choose two variables, those are them. Funky things start to happen if you try to overfit. Perhaps next year.

There are certainly some odd things happening out there. To look at those two sites, it’s hard to make a call. Eben Frankewitz scored a dismal 10/100 on the WNTS approval ratings, but registered a 44.24 on Dialidol. That is a tremendous score, and despite such an execrable approval rating, the model regards him as the top vote getter. After that are Chase Likens and Adam Brock, who also got low approval, but substantial Dialidol numbers. Colton Dixon had a very low Dialidol score, but got relatively high approval ratings. Nevertheless, he is roughly half as likely to advance as Phillip Phillips.

For the women, the Dialidol numbers were small across the board. The highest vote getter was also the contestant with highest approval, Jessica Sanchez. Elise Testone, despite a small 0.12 Dialidol score, is regarded by the model as the next most likely to advance on the strength of her nearly universally praised performance (approval rating 84/100). To the chagrin of Worsters, Haley and Baylie are both at the bottom.

Finally, Brielle Von Hugel is predicted not to make it, despite having the highest Dialidol score, because of performance approval.

Now, note that these are probably going to be wrong. Hell, they might even be very wrong. It’s just a model, and weird things happen all the time on Idol. If picked at random, one should get 11/25 = 44% by just dumb luck.

Men

Name Pre-exposure
(seconds)
WNTS
Approval
Dialidol Probability
of Advancing (%)
Eben Frankewitz 718 10 44.24 71.1
Chase Likens 190 32 20.99 70.2
Adam Brock 947 32 19.74 69.8
Jermaine Jones 1014 51 14.57 68.5
Joshua Ledet 466 81 10.9 68.4
Phillip Phillips 1058 67 6.84 60.8
Colton Dixon 728 63 0.45 35.8
Creighton Fraker 1036 55 0.54 31.9
Jeremy Rosado 441 55 0.2 30.4
Aaron Marcellus 450 52 0.04 28.1
Reed Grimm 1158 36 0.52 22.2
Heejun Han 1424 38 0.24 22
Deandre Brackensick 439 37 0.04 20.8

Women

Name Pre-exposure
(seconds)
WNTS
Approval
Dialidol Probability
of Advancing (%)
Jessica Sanchez 601 82 3.08 57.2
Elise Testone 420 84 0.12 54.5
Hollie Cavanagh 210 76 2.52 54.2
Skylar Laine 726 74 2.43 53.3
Shannon Magrane 738 62 2.22 47
Erika Van Pelt 570 63 0.01 43.7
Hallie Day 317 58 1.32 43.2
Jen Hirsh 812 53 1.38 40.2
Brielle Von Hugel 537 24 9.85 38.4
Chelsea Sorrell 198 40 2.99 35.2
Haley Johnsen 346 12 2.08 17.2
Baylie Brown 366 13 0.88 15.9

Bookmark the permalink.
  • Lurkher

    Would it be easy to manually adjust the model to give early Dialidol scores less weight?

    • Reuben

      Yes, that should already be in there. The data used was only first voting round (usually with 24 people), so the strength is already appropriately adjusted for this point in the game.