Tuesday, August 22, 2006

The Big Brother Final 2006 - Review Of The Voting (Updated);

Greetings to friends and foes of Special Bets.

The conclusion to our now notorious 7pm Big Brother Final post was that;

"The viewing figures of tonights show will be double that of a normal eviction friday. We have been asked a lot of questions about how these extra viewers will affect voting behaviour on the final night. Our track record shows that our polling group has a high degree of accuracy when it comes to Friday night evictions. We have less experience in running final night polls than we do in eviction night polls. There is a risk that our polling methods do not capture the effect of the more casual viewer tuning in tonight and deciding to vote for the first time, and this should be noted by those that follow our analysis."


Pete Bennett, the winner of Big Brother 2006 was clearly the most voted-for housemate among the 8.2mm viewers of Friday night's Big Brother Final show. Third placed Aisleyne Horgan-Wallace, in our opinion was the most popular housemate among the committed viewers who watch the show week-in week out.

Our reliable and responsive polling group that are usually so accurate in calling regular evictions were not representative of the former group only the latter - and the degree to which this is true was so marked that Aisleyne's latecoming Big Brother fan support skewed our results making them completely unhelpful in predicting the order of the final 3.

Further compounding this, second placed Glyn Wise was the beneficiary of 2 important campaigns on finals day which we do not believe would have been influential to the type of committed Big Brother fans that currently form our database of 2000 Big Brother fans.

A full review of the above and of all the internet polling and influencing factors that made up a dramatic finals week which ended in failure for us on a professional level follows.

We hope that Friday's poll fiasco will be akin to YouGov's November 2004 call of the US Presidential Elections for Kerry - an aberration that will benefit us in the long run as a learning experience and a valuable lesson in the need for a truly representative sample.

Thank you to Maloolaboo for your advice, onthehushhush, spindrift, Zoomraker, Joe454 (for your detailed feedback) mills427, haydyn (regional input!), thomasthetanker, theoldkidontheblock, pokerq (hope you enjoyed taking mine and Statman's money!- enjoy), kunnfunna and polkoi for your support us on the betfair forum and also to former Special Bets guest tipster "againstthecrowd" for starting the thread "Special bets Blog--- Your reaction..." thread on the Betfair BB7 forum - its nice to know what everyone thinks.

After much beard stroking (Statman has actually grown one) and discussion between the team and others Statman put together the following review today;

Statman's Special Bets BB7 Review;


1. Push-polling is easier than you might think. In fact you can do it without knowing.

I believe that the way we presented the ongoing poll updates created a self-fulfilling push-poll effect that led to respondents over-emphasising support for Aisleyne. I don't think this was a deliberate attempt by our polling group to mislead, but a genuine combination of the underdog effect (followed by the bandwagon effect). In future, the way that poll updates are published will be changed to avoid this happening.

2. The poll does not always work on its own.

Polls need to be adjusted for external factors. We put ultimate faith in our polling group based on it's eviction prediction record (see Nikki's re-entry, Mikey vs Suzie, Nikki vs Aisleyne) and whilst we did put the disclaimer on the final post about finals night being very different to eviction nights, we did not adjust our analysis accordingly.

This lead to a dismissive attitude towards external factors that would have affected voting, such as what "casual" voters might do, the Welsh factor, and The Sun campaign, because we were unable to measure their effects. we even managed to ignore a useful piece of our own analysis that showed that Aisleyne needed 42% of the initial vote to win. In future, we will adjust the poll for these potential effects.

3.The poll should remain an unbiased gauge of sentiments at all times.

I believe there were media campaigns, and internet fan campaigns, that particpated in our poll as a vehicle for promoting their chosen favourite. In future being associated with these lobby groups is to be avoided.

4. We need a larger and more secure polling group.

Our polls were sent to the same core group of about 2000 people several times. We need to increase the size of the polling group, and poll them seperately in order to accurately gauge trends. We also need to secure our polling group. Whilst we are confident that we did not have any element of "hacking", there are methods by which people could multi vote (though we can track this) or have an impact on the result. In future, we will adapt the polling method so that randomly selected responses are generated, increase the polling group size, and increase levels of security.

5. There are conflicts of interest between running a polling group and being a member of an internet betting community.

We have been accused of using our website to manipulate prices in order to make profits in our own book. These allegations are false and disappointing to hear. Everything that the Special Bets team made over the course of the three months of Big Brother was lost on the final evening, and some more. It was a very disappointing return for three months hard work.

Rather than make these untrue allegations on the internet, I would suggest the accusers instigate an investigation by Betfair into our betting accounts which will show us to be innocent of profiteering from our inaccurate final polls in this series. In future, the polling aspect of this site will be distanced from the betting community in order to maintain its neutral status.

Those Final Night Voting Figures

The official published figures from Channel 4 were:

Pete 61%
Glyn 39%
Aisleyne 22%
Richard 9.20%
Nikki 6.50%
Jennie 1%

The problem with these figures is that each percentage is of the remaining housemates at that time, hence they are incomparable to any poll prediction ahead of the show.

In order to try to gauge absolute votes, we need to make some assumptions about voting figures.

In the first five series of big brother, on average 6.8 million votes were cast in the final week. So let's assume for the sake of the exercise 7 million votes this year.

We have some reasonable data for Big Brother 5. In BB5, 60% of the total votes in the final had been cast at the time of the first eviction, so let's use that as a guide for this year.

This means that at 20.30 Jennie had 1% x 60% x 7,000,000 = 42,000 votes.

Now, we will assume that the remaining 40% of votes (2.8 million) were spread evenly (to be straightforward) until Pete's victory 109 minutes later.

Nikki was next to be evicted at 20.45. By this time there were 4.585m votes, and excluding Jennie's 42,000, Nikki scored 6.5%. Hence Nikki actually got 6.5% x (4,585,321 - 42,000) = 295,316 votes.

Using the same logic, this gives the following numbers:

Richard 419,176
Aisleyne 1,040,141
Glyn 2,029,313
Pete 3,174,054

Hence, the question remains how many votes Glyn and Pete had when Aisleyne was evicted. The assumptions above lead us to a figure of 1,515,596 votes cast in the final hour when Glyn went head to head with Pete. If we allocate those votes equally then the standing at the time of Aisleyne's eviction was as follows:

Aisleyne 1,040,141 or 22.0%
Glyn 1,271,515 or 26.9%
Pete 2,416,256 or 51.1%

Under these assumptions Glyn would have had a fairly comfortable lead at the time Aisleyne was evicted. However, it also shows that it could have been close at the time, or leading up to, Aisleyne's eviction - if Glyn had been stronger than Pete in the final hour.

Our polls never showed Glyn fairing better than Pete in a head to head situation, but the bottom line is that we will never know for sure, and it can not be ruled out.

The above is one scenario based on a certain set of assumptions. It implies that 22% of votes were cast in the final hour. Bearing in mind that in Big Brother 3, 16.35% of votes were cast in the final half hour before the winner was announced, this figure of 22% for the last hour this year looks a little low, hence once again this implies more votes for Glyn in the final hour, and that it could have been very close between Aisleyne and Glyn during the final show.

All of which may be interesting in hindsight, but of small comfort to those of you who like us, lost out due to Aisleyne's poor showing in the final relative to our predictions.

To conclude, we'll leave it to the "mystery" Youtube man who was so moved by our analysis that he made this excellent video directed at us in the wee small hours following the show.. frankly we're flattered.

Special Bets WILL be back for X-Factor






Have a free £25 bet with our partners Blue Square
special bets specialbets big brother reality tv betting odds eviction bet pete davina aisleyne vote big brother polls who goes? eviction % prediction
Find the best prices on Special Special Bets' Odds Comparison Site

2 Comments:

Blogger budgiebird said...

Sometimes a losing bet can be very profitable in the long term, if you learn lessons from the mistake you made when placing that bet. Similarly, I believe that although Friday's result was pretty disastrous for you guys, if you can learn from what went wrong, then you will hopefully benefit in the long term.

Statman's self criticism is to be applauded and the possible list of areas where both your poll and maybe your presentation fell down, is refreshingly honest. After all the hard work you have put into your poll and this website, it must have taken a lot to be so self effacing about your end product. Much kudos to all concerned for having the guts to do so.

FWIW, I do think that, at the moment of Aisleyne's eviction, it was very tight between her and Glyn. For me, the reason that her support fell away so badly towards the end was the fact that Nikki, the main focal point of the protest vote which Aisleyne was heavily relying upon, was evicted so early in the evening. Once Nikki had left, I believe that the majority of the protest vote also departed.

Much respect for admitting your failures, congratulations on your good eviction forecasts. Hopefully lessons learned will lead to your finals forecasts matching the success of your eviction forecasts next time around.

2:51 AM  
Blogger Statman said...

Thanks for your comments Budgie.

I guess we will never know exactly how close Ash and Glyn were during the show. The review of the voting proves that it could have been close, however it does not prove that it actually was close. I hope that the people who say that Ash was never in with a shout for second place can digest and appreciate this.

It was a difficult post to make, but it had to be done in order to get ourselves back on our feet and moving again.

Our search for a silver lining appeared futile for the first few days following the final, however I feel we are now a lot wiser and alert to our own limitations than we were this time last week.

Hopefully, as you say, this learning process will evolve to benefit the people like yourself who understand that the only thing we can do is admit we got it wrong, learn from it, and put into place procedures which ensure we never fall foul to the same mistakes again.

Cheers

Statman

12:53 PM  

Post a Comment

<< Home