aliens calling tinfoil hat... come in, tinfoil hat...
The 5 Couple Week - The Massive Bias Against The Public Vote
2 of 2
Originally Posted by davegold:
“Let us consider a horse race at Newmarket instead of Strictly with 5 horses. It is complete fantasy to say that all outcomes of the horse race are equally likely because the race has not been won yet. Punters, bookies, and spectators all know that the the horses and jockeys are not all the same. In Strictly we know that the couples are not all the same as well. For you to say that it is just as likely for the horses to be any position, and for you to say that is just as likely for the couples to have any popularity, would not be a justified assumption.
Let us consider also the situation that everyone followed your advice. If they did then the number of votes going to the first two couples would drop to near zero. Those couples would certainly then be at risk of the dance off and the 'likelihoods' would not be applicable. This shows that voting patterns amongst the public stop are a good reason not to assume that each couple have the same chance of being in each position.
Let us consider also that if a couple are regularly in the bottom two but always avoid the dance off then we have extra information on them. We know that are not in the lower positions in the public vote, even after one couple has gone. This means they are not going to have the same likelihood of being bottom of the vote next week as someone who has survived the dance off - unless you are also assuming that the voting is independent from week to week (which you presumably are not).
The public will know all of the above as common sense, to some degree. The public know that the couples are likely to get more votes with certain types of performances than others. The public know that people do give votes to couples at the bottom of the leader board and ignore those in the middle. The public know that a lot of couples keep their popularity from week to week. All this information is discarded within your analysis (by assuming all outcomes are equally likely). The public are using information that your analysis does not, so please do not presume to tell them how to vote.
Others posters have told you this as well. Please listen to them even if you are determined to prove me wrong!”
“Let us consider a horse race at Newmarket instead of Strictly with 5 horses. It is complete fantasy to say that all outcomes of the horse race are equally likely because the race has not been won yet. Punters, bookies, and spectators all know that the the horses and jockeys are not all the same. In Strictly we know that the couples are not all the same as well. For you to say that it is just as likely for the horses to be any position, and for you to say that is just as likely for the couples to have any popularity, would not be a justified assumption.
Let us consider also the situation that everyone followed your advice. If they did then the number of votes going to the first two couples would drop to near zero. Those couples would certainly then be at risk of the dance off and the 'likelihoods' would not be applicable. This shows that voting patterns amongst the public stop are a good reason not to assume that each couple have the same chance of being in each position.
Let us consider also that if a couple are regularly in the bottom two but always avoid the dance off then we have extra information on them. We know that are not in the lower positions in the public vote, even after one couple has gone. This means they are not going to have the same likelihood of being bottom of the vote next week as someone who has survived the dance off - unless you are also assuming that the voting is independent from week to week (which you presumably are not).
The public will know all of the above as common sense, to some degree. The public know that the couples are likely to get more votes with certain types of performances than others. The public know that people do give votes to couples at the bottom of the leader board and ignore those in the middle. The public know that a lot of couples keep their popularity from week to week. All this information is discarded within your analysis (by assuming all outcomes are equally likely). The public are using information that your analysis does not, so please do not presume to tell them how to vote.
Others posters have told you this as well. Please listen to them even if you are determined to prove me wrong!”
Or as Discworld maths have it "Million to one chances happen nine time out of ten".
I think you have to give all outcomes an equal weighting when calculating probability of the lowest scored being in the dance off because the public vote can be so fluid and can change during the series. Once you build in personal views of who is likely to poll more votes it becomes less objective and open to challenge. I'd only caveat that and say that if there is overwhelming evidence that someone is polling high then you would probably build that into the model. I don't think there is overwhelming evidence this year.
2 of 2
