tag:blogger.com,1999:blog-6413493203849579481.post7195425581288708059..comments2023-10-23T18:27:06.378+01:00Comments on PoliticalBetting - Channel 2: WattsUp with YouGov? Haven’t we Been Here Before?Mike Smithsonhttp://www.blogger.com/profile/11961547389548912471noreply@blogger.comBlogger8125tag:blogger.com,1999:blog-6413493203849579481.post-80135629018219839042010-02-28T18:38:57.368+00:002010-02-28T18:38:57.368+00:00Mark,
I didn't know about those sampling fact...Mark,<br /><br />I didn't know about those sampling factors.<br /><br />It is a concern that they have more male over-55s, but again, they're the ones that are likely to vote. As for party ID, this is fluid, although not as fluid as day-to-day preference.<br /><br />So let me propose this thought experiment: if the Tory vote has "hardened", i.e. more and more people positively associate themselves with the Tories although the overall percentage of current Tory supporters remained stable, then this would paradoxically reduce the weighted Tory percentage.<br /><br />You state that internet pollsters have the opposite problem of pro-Tory sampling. Do they? As Flockers illustrated, their downweighting of the Tories has shifted from 0.02% mid-last year to 6.3% now.<br /><br />This is a huge shift, and as Flockers illustrated, the unweighted samples have given a 14-16% lead to the Tories for almost a year.<br /><br />What has changed? I can't find the tables for May 2009, I wanted to compare the group sizes myself, especially over-55s.Richard Mannshttps://www.blogger.com/profile/13780203714224125693noreply@blogger.comtag:blogger.com,1999:blog-6413493203849579481.post-24677637770731274842010-02-28T18:25:04.534+00:002010-02-28T18:25:04.534+00:00Mark Senior. Naturally I apologise for putting th...Mark Senior. Naturally I apologise for putting the Tory range at 38-40 instead of 38-41. And ditto for the LibDem ranges.<br /><br />The point I'm trying to make here is that we know that different pollsters have different weighting criteria.<br /><br />But lately there appear to be some funnies and, in the case of YouGov, combined with a tweak in their methodologies.<br /><br />It's reasonable to wonder why this might be, what effect it might have on the results and whether there are parallels to be drawn with other black-box systems, which [re climategate] we know have been less robust than we first imagined.bunncohttps://www.blogger.com/profile/06768612455017655264noreply@blogger.comtag:blogger.com,1999:blog-6413493203849579481.post-9151983734585341572010-02-28T18:07:02.133+00:002010-02-28T18:07:02.133+00:00Richard , If you look at the detailed data tables ...Richard , If you look at the detailed data tables last page , you will see that Yougov have been oversampling by a large extent males over 55 and respondents with a Conservative Party ID , I would suggest that these 2 facts are linked .<br /> You are correct that the telephone pollsters have invariably to weight Labour downwards because telephone pollsters contact a disproportionate number of Labour voters . Yougov are an internet pollster and they have the opposite pronlem as do AR but seemingly to a smaller extent .Mark Seniorhttps://www.blogger.com/profile/06482009878893889107noreply@blogger.comtag:blogger.com,1999:blog-6413493203849579481.post-46982187595497614512010-02-28T17:45:10.055+00:002010-02-28T17:45:10.055+00:00@ Mark Senior
I agree that the Tories have been f...@ Mark Senior<br /><br />I agree that the Tories have been falling slightly, etc. But historically, samples have had to be weighted to avoid a massive Labour lead, not vice versa.<br /><br />Why has this changed? Bunnco comments on the "black box" of methodology, so it's hard to tell. But some large factor must have kicked in to switch the weighting against Labour; even YouGov's apparent success at reducing the pro-Labour bias through internet polling never converted to a pro-Tory bias, until now (it seems).<br /><br />This is the question. Why? I seem to recall that YouGov froze its sample after the BNP proposed entryism to distort their results. Is this still true? In which case, why the shift in weighting? YouGov says they've changed things to avoid the loss of "older voters", but don't they historically vote Tory? If so, you'd expect any adjustment to be pro-Tory!<br /><br />I propose that it's a pro-incumbency weighting to deal with shy voters last seen in the Tories' 4th victory, as this would be Labour's 4th.Richard Mannshttps://www.blogger.com/profile/13780203714224125693noreply@blogger.comtag:blogger.com,1999:blog-6413493203849579481.post-45516605496771355472010-02-28T17:42:40.108+00:002010-02-28T17:42:40.108+00:00(AnneJGP)
Very nteresting, Bunnco.
"...a fe...(AnneJGP)<br /><br />Very nteresting, Bunnco.<br /><br /><i>"...a few weeks ago YouGov changed their weighting methodology in preparation for their daily polls, noting that Tory voters tended to respond more quickly to invitations-to-survey and older voters often missed-out because they only check their emails every few days"</i><br /><br />"Older voters often missed out" clearly implies that they responded too late to be included in the survey.<br /><br />But, as long as non-Tory voters responded in time to be included in the survey, why does "Tory voters responding more quickly" matter?<br /><br />Does their model somehow give added weight to a quick response? If so, why?<br /><br />Or do they really mean that non-Tory voters, like older voters, responded too late to be included in the survey? If so, what does that mean for their sample?Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-6413493203849579481.post-41571609992603386732010-02-28T17:01:59.605+00:002010-02-28T17:01:59.605+00:00Usually I read your articles with both respect and...Usually I read your articles with both respect and interest but this one lacks objectivity on a grand scale .<br /> You state that Conservative support is stable in the 38-40% range - False . It has gone down from an average of 41% in early January to an average of 38% now .<br /> You state that it is the LibDems who have lost support and are flatlining at 17/18% - False . LibDem support is a bit lower with Yougov true at 17% but with all other pollsters has crept up this year to 19-21% .<br /> Like many Conservative posters you harp on about the Conservative unweighted lead , This is simply ridiculous an unweighted lead in a heavily pro Conservative sample is meaningless - just as a 1% lead in Mori's unweighted figures is meaningless because their sample will always have a pro Labour bias .<br /> As with every poll you need to look at the base data and the weightings used and then come to a reasoned judgement of what the correct figures may be .Mark Seniorhttps://www.blogger.com/profile/06482009878893889107noreply@blogger.comtag:blogger.com,1999:blog-6413493203849579481.post-699654250813884882010-02-28T16:12:12.005+00:002010-02-28T16:12:12.005+00:00Bunnco,
an excellent post, with all the questions...Bunnco,<br /><br />an excellent post, with all the questions that I would want to ask Mr. Kellner.<br /><br />Clearly OGH has cleaned up from the confusion of this poll.<br /><br />There's a lot riding on the result of the election (including currency / bond speculators betting billions on UK future prospects). If adjustment methodologies appear to be negligent in their design or application I would not want to be in the shoes of the people who implement themfirstlight40noreply@blogger.comtag:blogger.com,1999:blog-6413493203849579481.post-47447997527638351822010-02-28T16:02:28.072+00:002010-02-28T16:02:28.072+00:00I'm no expert, and I don't know much about...I'm no expert, and I don't know much about their methodologies, but I've had a thought:<br /><br />We know that the pollsters suffered from their 1992 failure, but pollsters presumably avoid adding modifiers "we should add 5 points to the Tories"; instead they went for incumbency and "shy x voters".<br /><br />Now that this is the 4th election for Labour, in incumbency terms, it's like 1992. So those "shy Tory voter" modifications will, presumably, now apply equally strongly to Labour and give them a boost.<br /><br />But are Labour voters shy? Speaking as a Tory, I've never known a time where it has been in vogue to be a Tory. Cameron may have "detoxified the brand", but when a new person in my work saw I was a Tory, his response was an incredulous "Why?", as if it were a crime. Meanwhile, watch a recent episode of Argumental, and you'll see a universal derision of right-wingers as "Daily Mail readers" whilst the Guardian is a paragon of civility.<br /><br />I suggest that there is a fundamental difference between swing voters voting for a Tory and voting for Labour, and that is reflected in the attitudes to the parties.<br /><br />I might be wrong. We don't know. There might be a "shy Labour voter". But, as some foreign journalist (according to CasinoRoyale) once said, “You British talk Left but always vote Right.”Richard Mannshttps://www.blogger.com/profile/13780203714224125693noreply@blogger.com