3 weeks in, getting the hang of this twitter thing now. Last week I got the bottom 2 spot on, almost certainly in order, so starting to believe there might be something in this! This time the semi-final. No bottom 2 so it was all about spotting who was bottom of the vote.
I received figures as usual from my source and plugged them into my model, before predicting on sofabet percentages of:
| Contestant | Vote Share |
|---|---|
| Little Mix | 41.7% |
| Marcus | 32.1% |
| Amelia | 14.2% |
| Misha | 12.0% |
So I called the elimination for Misha, but was nervous as it seemed that Amelia was only just ahead. Dermot announced the results on Sunday night, and chalk another one up - Misha was indeed revealed as bottom in the vote. In the betting, the odds on Amelia plunged - but I have her only just surviving. One of these has to be wrong. But which one?
The only concrete evidence about how Amelia is doing is that she was in the bottom 2 in week 6, apart from a newspaper report that she topped the public vote in week 5 when she returned. I can measure her twitter support compared to this - not the raw numbers themselves, but I compare her support to the other contestants week by week, factoring in how much of each contestant's fanbase is on twitter.
If contestant A has 300 tweets one week and 700 the second, that tells me nothing. But if we also know contestant B had 400 in week 1 and 800 in week 2, although they've both gone up by 400, I'd say it's better for A. Why? A has 75% of B's tweets in week 1, and 88% in week 2, so A is proportionally catching up on B. Now if I know contestant B is in the bottom 2 in week 1, and I've worked out that A has improved their proportion against B in week 2, I'd be pretty confident that A isn't in the bottom 2 second time around.
And that's all true for Amelia; we know she was behind Little Mix and Marcus in the vote on week 6, and comparing their twitter figures that week to this week, Amelia is now further behind Marcus and Little Mix than when she was in the bottom 2. I can't weight her any lower, as it would take her below Craig in week 6, when we know she was ahead (on deadlock - I think it was very close.) And any higher would have her going above Misha and out of the bottom 2. So my results suggest that the Amelia hype isn't based on reality, and that she is way, way off the pace heading into the final.
A late change was make on anlytk to drop support for Marcus. (due to blending with the American Marcus. Contestants with the same name in different countries? Bah! What's a twitter analyst to do? ;) ) I've chosen not to take this into account in my figures - as this is all based on trends, any changes to the underlying data will skew my results (of course I totally understand that my source wants to improve their data all the time as do I!) Just for fun, If I had gone with the amended results, I'd have indicated:
| Contestant | Vote Share |
|---|---|
| Little Mix | 46.1% |
| Marcus | 25.4% |
| Amelia | 15.3% |
| Misha | 13.2% |
If this is anywhere near the mark, Little Mix already have it in the bag, unless they forget the lyrics on the Wembley stage. Here is the updated results graph based on the top set of figures. Even without the downgrade on Marcus, it's grim reading for his fans, and it looks like a huge week for Little Mix. I've taken off the first week's voting as there are too many weeks here to make it readable: