Monday Call - July 19 2004 (NEXT POST THURSDAY)
July 18th, 2004
-
Give a political bet as a present!
If you’ve been celebrating or commiserating after Thursday’s by-elections you might be interested in a new gambling service where you can give political bets as presents. It’s all done online and you can give one to yourself.
What’s promising is that the service, Flipem, does well thought out political bets and ran what we think was the only “size of majority” market on the Leicester South by-election, where we came across them. For a unit cost of £2 to you you could have given someone £12 if the Lib Dems had had a majority of 1-2,500.
-
They’ve now got a market on how many seats the Lib Dems will get at the General Election with some reasonable odds.
For 71-75 seats a £2 “gift” will produce £20 for the recipient of your choice if that’s what the party eventually gets. The 45 or less “price” is £32 for the £2 stake which might be popular amongst Labour or Tory supporters! No doubt other political markets will emerge and we will keep an eye on them.
General Election spread-betting gets cheaper. After last week it’s even more tricky to predict the number of seats that each party will get and punters seem to be keeping well away from the spread betting markets. These have now reduced their prices - the spread between “buy” and “sell” - in order to make themselves more attractive but, apart from the Lib Dems, we do not see any value.
The new reduced priced spread markets are:-
LAB 343-350 (+3) : CON 223-230 (-8) : LIBD 59-63 (+1)
The main General Election market move in the opposite direction. The Labour price went from 2/7 to 1/3 on William Hill and this is the current best value. These markets are about Who Wins Most Seats and not on who gets a majority and it’s very hard to see how there can be any other result than Labour coming out on top even if it suffers a huge drop in votes.
TONY BLAIR. He’s still there and while he wants to stay in the job we think that he will remain. Two markets - will he be there on election day and who will be the Labour leader?
There’s even now a market on him being Prime Minister longer than Mrs Thatcher. You can get l!1/2 that he won’t and 6/4 that he will. We wonder whether Gordon Brown has any money on? We do not like markets where you do not know when you will be paid and the prices are not good value.
WHITE HOUSE RACE
Just a week off the Democratic Convention in Boston when John Kerry will be selected, formally, as the party nominee for President in the election on November 2. For those with winning bets on Kerry for the nomination this should mean that we get paid out.
-
The frustrating thing about this market is that it was all resolved in early March yet we have had to wait for so long for the pay-out.
Next time the bookies should define getting the “nomination” when a candidate has secured enough delegates.
The polls have hardened on the Kerry-Edwards ticket and so have the betting prices. Bush, at 9/10, is at his weakest point ever. We stick with our BACK KERRY call.
Other markets have started to emerge including ones on who will win particular states. Included amongst them is Florida where there was all the fuss last time.
POLITICALBETTING.COM We will be away for a few days though reader comments, no doubt, will continue. Next post - Thursday July 24.
Image: http://www.sbac.edu/~tpl/clipart/Holidays/present%2002.jpg
MessageSpace Advertising
You are Mike Smithson, aren\’t you? If so I have just found the following comment you posted on the Anthony Wells blog back in May: \”The affect of Tories who don\’t tell everything to pollsters has been shown at one election after another when large Labour leads have proved by the subsquent results to be wild exaggerations. There is no similar experience with Labour voters so what we have here is Populus making a guess. They might be right - who knows?\” You really ought to check your facts and assumptions before posting portentous comments like that. You are completely wrong. The method we use(ICM use a very similar, though not identical method) to take account of \”Tories who don\’t tell everything to the pollsters\” is not party-specific: it uses weighted ratios based on vast amounts of historical voting data to make a judgement about how many of those who won\’t say how they\’ll vote will end up voting for who they did at the previous election. It applies just as well (or badly) whether the effect is shy Tories (as it was for more than 10 years) or shy Labour (as it has been for around a year). I readily agree it is not a foolproof or scientifically perfect formula - we are totally transparent about its deficiencies as well as about our reasons for doing it - and we keep our methods under constant review (in the belief, after Popper, that \”the wrong view of science betrays itself in the craving to be right\”). I am alsways happy to discuss these issues. Why don\’t you check before lodging confused points into the ether -especially if you\’re going to assert things in such a declamatory way.
Steady on Andrew! Mike calls it as he sees it. We all appreciate his insights, even if we don\’t agree with him. No need to get quite so upset.
I was under the impression that Mike\’s argument was not, as you seem to have interpreted it, that the method you used to determine the % of \”shy Labour\” was arbitrary, rather that it is a guess that \”Shy Labour\” voters exist at all. The concept of a \”Shy Tory\” was, I believe, invented in part as a response to a repeated undervaluing of Tory strength relative to General Elections. Weighting methods were constructed which produced results which, when applied, seemed to negate the \’problem\’ to some extent. For example, people might be asked how they voted in the last General Election, and the \”don\’t knows\” for the main polling question could be adjusted according to any discrepancies in this data. If only 26% claimed they voted Tory when the true figure was 32% then some appropriate number of the “don’t knows” could be altered into probable Tory voters (these would be the “Shy Tories”).
Now it may well be that this effect is now showing itself among supposed Labour voters (eg. Labour support at previous General Elections is now being understated), but whereas in the case of the Tories there is circumstantial evidence (based on the discrepancies between polling and GE outcome) to indicate that a large number of ex-Tories who lie about their past allegiance will then vote Tory anyway, there is no such evidence for the same phenomenon among Labour voters (because there have been no recent elections in which the Labour vote has exceeded that predicted by the polls). Isn’t what we have is a situation whereby Pollsters have come up with a (party-neutral) formula to account for what was an (unproven) single party phenomenon and are now using it to explain away a suspiciously large Tory lead in the raw data?
Using my inexpert method above, who is to say that, whereas Tory liars will return to the party on Election day, Labour liars may not have gone for good? That is I think the “guess” that Mike was referring to.
Alex - while I am no fan of \”adjusting for shy Tories/Labour/whoever\” it\’s more complicated than that.
You\’ve mixed up two different processes slightly. Some pollsters (specifically Populus and YouGov) weight by past vote. In other words, if only 20% of those polled say they voted Tory at the last election then those people\’s answers are weighted so they make up the correct proportion of the final sample (basically, as an example, if former Tories are underrepresented then each former Tory would count as, say, 1.08 of a person to bring it up to correct proportion). It\’s important to note that this won\’t actually be the proportion of the population who *did* vote Conservative at the last election, because people forget about, or lie about how they voted. To start with a hell of a lot of people say they voted Labour who didn\’t really and if these people were cut down to the real level of Labour support the sample would be unrepresentative - you get the idea. Populus\’s technique for accounting for this is that they asked \”a how you voted last time\” question years ago when people\’s recollection was more accurate and used it as their \”base level\” of false recall (the full complexity of this is on their webiste). Suffice to say this weighting isn\’t about shy Tories, it is to try and get an accurate sample. It\’s a system that makes some assumptions - hence the reason that only some pollsters use it and - Populus at least - say they keep it under review.
Adjusting Don\’t Knows on the other hand is indeed all about the \”shy Tories\” (or \”shy Labour\” etc, etc). It was created by ICM to try and solve the problems that the pollsters faced in 1992. In short Don\’t Knows seemed to be disproportionately former Tories. ICM assumed that the Don\’t Knows were most likely to vote with their previous allegiance - reallocating about 60% to the party they previously voted for. The end result to to bump up the level of Tory support. What Mr Cooper has said since then is that the Don\’t Knows now disproportionately include former Labour voters, hence using a method similar to what ICM did in 1997 would now increase the reported level of Labour support.
As you said, this does assume that former Labour voters saying they Don\’t Know will behave in the same manner as former Conservative voters saying they Don\’t Know - a possibly shaky assumption. It is not completely without evidence though, the BES studies in 1997 and 2001 did look at how Don\’t Knows formerly aligned with all parties voted (I cannot tell you the results, I\’m sure Andrew Cooper could) - again people abandoning Labour at the height of their popularity though cannot automatically be assumed to behave the same as those leaving them now.
sinning holding kernel Sloane walnut ridding bring