X Factor: I’m sorry, Twitter told me to…
By Katy King[Our open door writing policy is in full effect again today with Strategy Executive, Katy King, picking up the blogging reigns to talk about a recent piece of extra-curricular activity she indulged in around that most talked about of modern media events: X Factor - JW]
Back in early October, when the UK’s latest series of the talent/music competition ‘X Factor‘ was kicking off, I set myself the challenge of trying to use Twitter to correctly predict who would be eliminated each week. I knew that the ‘heads had completed a similar piece of work with their Haye vs Klitschko analysis a while back and I wanted to give it a shot myself.
To make things a little bit more interesting (and without endorsing gambling in any way, shape or form!) I decided place a £5 bet on each prediction. Naturally, I set up a blog to track my progress, and resigned myself to the possibility of losing up to £50 over the next three months. Ah well.
Each week I measured the amount of conversation each contestant generated on Twitter and then, on top of that, I analysed how positive or negative that conversation was. It was with this information that I would predict which contestant would gain the least votes, lose the sing off and ultimately be eliminated.
However, I didn’t have an exact formula.
For example, in Week Six of the competition, I had to make a decision whether to bet on Misha B; a good contestant who had generated 27,000 tweets (which were 52% positive) and Kitty Brucknell, an equally well-talented [if slightly unhinged] girl who had generated less than 13,000 tweets, but were slightly more positive (57% positive). In this case, I decided that in total there was less advocacy for Kitty, and therefore she was more likely to be eliminated. My judgement proved to be correct, but this certainly wasn’t the case every week.
My project got off to a good start. I correctly predicted mismatched boy band “Nu Vibe” would lose the first public vote after finding that 58% of Twitter conversation about them was negative. The week after, the fact that there was very little Twitter conversation about Sami Brookes suggested to me that there wouldn’t be many people voting for her either. My hunch proved to be right, and I won again.
However, even when I was getting it right, it became clear that Twitter conversation was not an entirely accurate representation of the public’s voting intentions.
In Week Two, even though resident bad boy [and soon to be disqualified] contestant, Frankie Cocozza, generated the most Twitter conversation (much of it positive), he still appeared in the bottom two of the public vote. In fact, even though Frankie generated the most tweets of any contestant for the first four weeks, he actually never reached higher than 6th place in any of the public votes.
Similarly, contestants such as Craig Colton, Sophie Habibis and ‘The Risk’ were all eliminated when they appeared to have a healthy amount of support on Twitter.
In the chart above, you can see that Frankie (red line) generated higher levels of conversation than the other contestants, before he was sent home for bad behaviour in mid-November. In contrast, the amount of tweets about Little Mix (the light blue line) grew steadily over the weeks, leading ultimately to their overall victory.
Oh, and in case you were wondering why tweets about Frankie Cocozza reappeared right at the end of the time line, it’s because he accused Wayne Rooney of being smelly. And Wayne Rooney replied.
Hilarious.
In total, I correctly predicted who would be eliminated 6 out of 9 times using Twitter as a guide. I ended the project £5 up overall as well as a healthy reality check about how seriously we should take Twitter conversation. What I learnt was something I should have already known, positive conversation (advocacy) does not always lead to action (in this case, voting).
See also: slacktivism / clicktivism
It’s been claimed that Twitter can predict box office hits and even the stock market. After undertaking this experiment, and having learnt from 1000heads’ previous analyses of Twitter conversation, it’s clear that analysing Twitter conversation is a useful and interesting exercise, but it’s by no means a fool proof way of predicting the future.
Merry Christmas everyone, I’m off to spend my winnings…
Like this?
-
Bandro
