X Factor 2012 Review Part 1: How many people vote on X Factor?

Sofabet commenter EM emails: “I’ve always wanted to know the real voting numbers as well as the percentages each week.” The show, sadly, has always been coy about revealing this. After the 2010 series, they let slip that there had been over 15 million votes throughout the series. This year, in week 9 Dermot said there had been over a million votes, and in the final over 3 million. As far as we’re aware, these are the only three data points in the public domain.

EM continues: “Using Dermots “million votes” in week 9, correlating that to each weeks viewing figures and an assumption that the percentage of viewers voting increases each week, what you get is”:

EM’s email made us wonder if there might be alternative ways of approaching the question to see how they compared. In our post-series chat with YouGov’s Joe Twyman (to be published later in this review series), we’d asked if their polling gives them a sense of how many more people vote in the final than the early weeks. Joe said: “I wouldn’t like to guess at a precise number, but it’s definitely a lot more. In previous years, we usually saw it picking up in semi week. But this time, we picked up more people voting in week 8 – the week after Ella and James’s bottom two.”

We can learn something from the published YouGov polling numbers, too. About 11% of the viewers polled say they voted in the semi. This seems to tie in pretty well to the viewing numbers. Dermot said there were a million votes, and the viewing figures were in the region of 8.4 million. 11% of 8.4 million takes us past 900,000; throw in some multiple voters, and we’re definitely in the right ballpark.

YouGov also asked people if they would vote in the final. 11% of viewers said they “definitely” would (oddly, not quite the same 11% who said they voted in the semi). A further 17% said they “may” vote in the final, and the other 72% said they definitely wouldn’t.

The published percentages show (assuming nobody voted for Christopher after he was eliminated) that approximately 44% of Dermot’s “over 3 million” votes in the final were cast before the Saturday vote freeze. That’s at least 1.32 million before the freeze and 1.68 million after. Viewing figures averaged about 10.2 million on Saturday and 11.1 million on Sunday, which gives us roughly 13% of Saturday viewers voting before the freeze and 15% of Sunday viewers voting after it. What we don’t know, of course, is the extent of overlap between those groups, or of multiple-voting. But, again, the figures are clearly in the right ballpark with YouGov’s definitely-maybe breakdown.

What about earlier weeks? Here we have to get much more speculative. We can obviously read something into Dermot mentioning a million votes in week 9 but not in any earlier weeks – this must be the first time that threshhold had been crossed in the series. And we can factor in Joe Twyman’s observation that YouGov’s polling finds the numbers of people voting usually increases in week 9, semi week, although this year a big increase came in week 8.

What happens if we – entirely unscientifically – guess that the voting approximately doubles from the early weeks to the semi? That would give us a nice round figure of 500,000 votes per week having been cast every week from 1 to 7. We’ll call it 750,000 in week 8, then Dermot’s 1 million in week 9 and 3 million in the final. We asked EM to plug those figures into his chart to see how they compared, and here’s what comes out:

As EM says in response, “Through entirely different reasoning we arrived at pretty much the same numbers”.

Can we attempt to approach these very rough guesses in a third way? Well, we’ve got that figure of 15 million for the whole series in 2010. What happens if we apply our rough guess that x people vote in weeks 1-7, 1.5x in week 8, 2x in week 9, and 6x in the final? That gives us just over 900,000 votes a week in the early weeks. In other words, for our figure of 500,000 a week to be correct this year, voting would have to have declined by about 45% from 2010 to 2012.

Wikipedia suggests that ratings have declined by somewhere in the region of 35-40% – so if we can assume that voting figures have declined roughly proportionally, that, again, would put us in the right ballpark.

All very interesting as an academic curiosity, but what might we conclude from such speculation? We have a few suggestions. First, it could shed light on Christopher Maloney’s treatment this series.

In the last couple of weeks of the series, a debate raged among Sofabet commenters. Those backing Christopher Maloney for the win argued that his fans would not desert him. Others argued that the question of Chris’s fans deserting him was irrelevant – what mattered was whether he could pick up new voters in the final week or two.

EM’s charts suggest that the latter argument was right. They suggest a fairly solid level of absolute vote totals for Christopher throughout the series – a core fanbase that was enough to see him win the first seven weeks, but not on its own to get him anywhere near winning the final.

All of which might explain the puzzlement some of us felt about why producers kept giving Christopher demo-delighting 80s numbers week after week after week, if they wanted rid of him. The answer, presumably, is that they felt no urgency – they were happy to get rid of him only at the business end of the competition. And, given this, wasn’t making sure he was firmly pigeonholed as an 80s cruiseship crooner the perfect way to ensure that he never picked up new fans?

Presumably producers, knowing from experience that vote totals would pick up in the semi and the final, simply needed to ensure that people who hadn’t loyally voted for Christopher throughout the series wouldn’t start to do so. This they accomplished rather effectively. The charts suggest Christopher got votes from his series-long fanbase in the final, but not many additional votes – whereas James increased his raw vote totals from the early weeks something like 20-fold.

As EM writes, “there is nothing much I can see to suggest James would win it until his week 8 results – up until then he really was in the pack”. It was a similar story with Alexandra Burke in 2008, and reinforces the lesson that an act doesn’t necessarily need to have the biggest of fanbases in the early, crowded field to be able to catch fire when viewers have their minds concentrated on which of the remaining few acts they actually want to win it.

This analysis also casts doubt on another possible explanation put forward in the comments for the puzzle of why producers seemed deliberately to be goading Christopher’s fans to vote, for example by pointing out the demon eyes staging in week 7 and allowing Christopher to give tabloid interviews about how he felt “bullied”. Some commenters wondered if it might be all about increasing phone vote revenue.

These stats suggest Christopher was pulling in somewhere around 120,000 to 150,000 votes in the early and middle weeks. If we generously assume that a third of these were motivated purely by sympathy at the show’s treatment of him, then at 35p a pop we’re talking about £14,000 to £17,500 a week – and let’s not forget that the phone company will presumably take a cut before the show sees that money.

That’s not nothing, but it pales into insignificance against the sums the show makes from advertising – estimated at £85m over the course of the 2012 series, about 30-40 times as much as they will have made from phone votes if our guesstimates are on target. Advertising revenues depend on ratings, so it seems likely that ratings, not phone vote revenue, will be top of programme-makers’ minds. Of course, anything that increases both will be a win-win, but in general it seems unlikely that producers would do anything to chase phone vote revenue if they were remotely concerned that it might jeapordise ratings.

Finally, these charts suggest how close the margins are at the bottom in the early weeks.

In week 1, Carolynne Poole finished bottom with 2.9% of the vote, Rylan joined her in the bottom two with 3.1%, and Melanie Masson narrowly escaped with 3.2%. With rounding, the gap between Rylan and Melanie might have been anywhere between 0.01% and 0.19%, but let’s say it was 0.1%. If our 500,000 votes is in the right ballpark, that’s 500-ish votes.

Think about that. At 35p a vote, a mere £175 worth of votes for Rylan would have given us a bottom two of Melanie versus Carolynne. Melanie would have gone a week earlier, and Carolynne would have bounced probably at least to week 3 or 4 – potentially over Kye, and, with a second save, into the latter stages of the competition.

Producers would surely have been willing to spend £175 to avoid that awkward bottom two of Rylan and Carolynne, two acts – unlike Melanie – whom they clearly wanted to keep around at least until mid-series. And, logistically, it would hardly have been difficult – in theory, half a dozen interns on speed-dial could have saved Rylan in an extended tea break.

Of course, the show would be crazy to take the risk of such a scandal breaking. Still, in our more wildly speculative moments, we have wondered if there might be some “dark ops” scenario for such eventualities – a murky freelance agent with a stash of pay-as-you-go sims, perhaps. We can, however, surely conclude that this isn’t the case if the margin between Rylan and Carolynne was so easily bridgeable and yet remained unbridged.

Which is ironic, because the Rylan-Carolynne singoff led to uncomfortable “FIX!” headlines when producer Richard Holloway was caught on camera chatting to Louis Walsh mid sing-off. As we have often argued on Sofabet, it’s ridiculous to cry “fix” when the judges save an interesting act over a more competent singer – that’s the whole point of the singoff format. A fix would be if the show were trying to interfere with the vote itself – and the very fact that this singoff happened surely proves, if there were any doubt, that the show’s not a fix at all.