Top 8 Projection

Note: projection is made using numbers from 1:00AM EST. They may change a bit by the morning.

Contestant Bottom 3 Previously? WNTS Approval Rating Previous Approval Rating Dialidol score (Rank) Probability of Elimination (%)
Hollie Cavanagh Yes 32 49 4.28 (7th) 39.55
DeAndre Brackensick Yes 48 44 0.94 (8th) 38.44
Colton Dixon No 65 52 5.15 (5th) 7.23
Elise Testone Yes 41 91 5.14 (6th) 6.24
Phillip Phillips No 33 72 7.15 (3rd) 5.70
Joshua Ledet Yes 54 64 10.07 (2nd) 1.69
Jessica Sanchez No 63 86 5.51 (4th) 0.91
Skylar Laine Yes 78 65 11.68 (1st) 0.25

Today’s projection has a razor thin margin between Hollie, last week’s pick, and DeAndre. Both suffer from sub-par performances both last week and this week (see below). DeAndre has a very low Dialidol score, but the model uses rank instead of raw number, so that should wash out. Nonetheless, he is dead last on that service’s measurement.

Beginning in the Top 8, there is a significant dependence on past performance on the current result. One bad score like Phillip has today has not historically been enough to doom a contestant, and voters take the previous performance into account almost as much as the current. The model thus assigns a probability in the Top 8 based on today’s approval, last week’s, and the Dialidol ranking for this week.

The model designated Hollie as the most likely to land in the Bottom 3 last week, and although she did, she was not the bottom-vote-getter. We can reasonably assume that those most likely to land in the B3 are also the most likely to be eliminated … right?

There is no significant difference between Colton and Elise in the projection. I would guess Elise will round out the B3.

As I noted in the LiveBlog, this is the first time in my memory that a contestant has run away from a country song in order to garner votes. Skylar Laine switched from a Dolly Parton snoozer (“9 to 5”) to the extremely pop oriented “Wind Beneath My Wings”, and the sampled audience appears to have responded well. Could the scourge of the American Idol bumpkin, that seemingly implacable menace, finally be dead? With Chase Likens’ quick dismissal and Skylar’s trip to the stools last week, one sees hope that it is.

Bookmark the permalink.
  • Lurkher

    WNTS numbers prior to Top 9 week have no observable effect on Top 8 eliminations? Top 9 week by itself could be a decent proxy for overall past performance, but I tend to think that a moving average of an idol’s past songs would do an even better job.

    • Reuben

      This is a really good point. In the later stages I use a 3 episode moving average, but my assumption (and that’s all it is, an assumption) is that during the earlier phases the memory effect of prior stages was limited to nonexistent. Last week, for instance, I could find no significant dependence on prior approval. However, I did not check the moving average. Had I time right now, I would take a look at it. The way I’ve been running is kind of a quick and dirty way to add that memory effect.