Kentucky fried Clinton

Democratic primaries will be held Wednesday our time in Oregon and Kentucky, which will respectively choose 52 and 51 delegates. Below is another race associated with the latter state, which this year ended with runner-up Eight Belles having to be put down*. Does the knackers’ yard beckon for a certain second-placed Democratic nag? Discuss.

* Unfortunately for my metaphor, Clinton in fact holds a handy 30.5 per cent lead in Kentucky, according to Real Clear Politics. Obama however leads by 12.4 per cent in Oregon.

Author: William Bowe

William Bowe is a Perth-based election analyst and occasional teacher of political science. His blog, The Poll Bludger, has existed in one form or another since 2004, and is one of the most heavily trafficked websites on Australian politics.

2,133 comments on “Kentucky fried Clinton”

Comments Page 43 of 43
1 42 43
  1. Following on from Judge Growlers questions aboout polls vs US prediction markets, I’ve hunted a few bits down.

    Of the “964 polls conducted over five presidential elections since 1988, the [Iowa Electronic Market] exchange has been closer to the real outcome 74 percent of the time.”
    http://www.lasvegassun.com/news/2008/jan/16/markets-pick-em-better/

    Intrade was 100% accurate with all the mid term Senate races last time and nailed every State at the 2004 presidential election.

    Heading over here:
    http://www.biz.uiowa.edu/faculty/jberg/papers/ResultsIEM_2003Mar.pdf
    (66KB file)

    Some handy graphs in there that look at poll vs market performance. One thing it shows is that the Euro markets are like Australian ones – not particularly good. Take note of the very last graph of on the projected size of the clinton winning margin vs what the polls were saying across the period of 1996.

    There’s still a graph I’m hunting down that I can remember seeing somewhere which deals with polls v market prices for US elections over the period of 12 months leading into an election.

  2. be great to have a new thread do we dont have to scroll to the bottom of the homepage, pretty please William

  3. Possum 2102

    These guys say that polls are more accurate than market price.

    Here is the abstract

    Election markets have been praised for their ability to forecast election outcomes, and to forecast better than trial-heat polls. This paper challenges that optimistic assessment of election markets, based on an analysis of Iowa Electronic Market (IEM) data from presidential elections between 1988 and 2004. We argue that it is inappropriate to naively compare market forecasts of an election outcome with exact poll results on the day prices are recorded, that is, market prices reflect forecasts of what will happen on Election Day whereas trial-heat polls register preferences on the day of the poll. We then show that when poll leads are properly discounted, poll-based forecasts outperform vote-share market prices. Moreover, we show that win-projections based on the polls dominate prices from winner-take-all markets. Traders in these markets generally see more uncertainty ahead in the campaign than the polling numbers
    warrant—in effect, they overestimate the role of election campaigns. Reasons for the performance of the IEM election markets are considered in concluding sections.

    Are Political Markets Really Superior to Polls as Election Predictors?
    http://palmdesert.ucr.edu/conferences/economica2007/erikson-gdi.pdf

  4. Dio – those guys are exactly right! But we’ve got to look at what they’re doing.

    They’re saying that if you carefully select good polling data, build quite sophisticated regression models based on that polling data and forecast into the future – it performs slightly better than the Iowa markets.

    But how many of us actually build regression models like that?

    They are treating polls exactly as they’re meant to be treated – as snapshots – then using time series econometric models to gather all of those snapshots up and analyse where the mean and variance of those polls are heading.

    But unfortunately – still – that’s not the way most people treat polls (most Bludgers excepted since we’ve all done 101,201 and 301 courses in practical polling stats from the last election! OK, not you Dennis – but there’s always one village idiot)

    They also, unfortunately, don’t use Intrade data, which is a little better by my reckoning than the Iowa markets.

    So those guys are right – robust econometric modeling of the polls give prediction markets a run for their money in the US.

    But single polls or any given poll doesn’t because they just aren’t meant to.

    Polls arent predictive, they’re explanatory. They don’t become predictive unless you build a model that enables their data to be projected ahead into the future.

    Market data on the other hand is predictive from the outset – people are betting on the outcome.

  5. Possum Comitatus at 2108
    You mentioned back at your 1883 comment that the scenario/model you presented has an inbuilt implied probability of an Obama nomination of 92%. I’m not all together sure what that means relative to a general election projection – could you explain? I would have assumed that a projection of an Obama/McCain forecast would be based on an implicit 100% probability of an Obama nomination of 92.

  6. woops – when I said
    100% probability of an Obama nomination of 92.
    I really meant to say
    100% probability of an Obama nomination.

  7. 2108
    Possum Comitatus

    Ah, all those people plonking down their hard earned really do think about it, don’t they? LOL

    As opposed to those who just shoot off their mouths to pollsters at 6pm whilst cooking the dinner!

    I would have thought the more ‘free market’ (whatever that really means!) types would ‘follow the money’, but no, they seem rather attached to their losing candidate or McCain. Consolation prizes for consistency?

    Naught so queer as folk, as the Yorkshireman once said.

  8. G G

    Possums betting argument

    Seeing you aee interested but mot persuaded you will find some spiffy graphs to support the notion of betting is an election predictor , but it a notion only , a statistical almost guaranteed coincidence , not necessarily a predictor.There are in fact numerous “economic election predictors” who have accurately picked the winner margin size based on certain economic criteria (some the same criteria between them , some different). These “economist election predictors” say I’ve been proved correct so my methodology is proved , so prove why I’m wrong. Well of course you cann’t disprove cause & effect and concurrently disprove effect and cause. Same with betting markets accuracy. (the theories abound from astrologers to ones wife liking the candidates ‘good dress sense’)
    Nothing is proved by either betting or ‘economic election predictors” nor by any other method irrespective of the alleged %accuracy , statiscal coincidences are everywhere. but below is a intellectual base using judgement factors never perfect including using computers & money but not as the only criterias.

    This is only a 2 horse race , so it’s a 50/50 prediction any way and the same with margin (with a little historical checking). Whilst the betting is a factor that should be considered , so are economic election predictions etc etc they can be simply statistically coincidence as stated. Polls , trends , demographics in their entirety , historical precedents & other psephological factors , turnout rates & mobilisation options , policys , incubancy , the forcast economic election month economic conditions then prevailing , election themes etc are a sounder base to predict , so I do not agree (for all the above reasons) with my marsupials conclusion I’m absurb unless I’m are allergic to historical accuracy.

    My absurbness and historic accuracy description has now been since played with so perhaps I will also addrss to yous regular blogers . There is one historic accuracy about this site the blogs forever do record historically accurately , the barbarian puts views with reasons not one liners (barbarianly correct) on one island and there are arrayed an army of 20 odd , lead by Generals ranging from marsupial chartists of repute , history Professors marxist armed literary prose exquisite , Lawyers , Economists , Uni Lecturers , 2 doctors & a brain surgon , 2 historic Philosophers , a greenie , foot solders & snipers , a Debating expert , some of whom are latte sipping elitist butterflys , the majority throw their barbs of dismissal & poor english constuction or “move on” as there intellectual bullets whilst the barbarian’s beliefs remain re Hillary by far better “electability” argument & her better universal healthcare with fixed % income cap labor type policy (presently badly affecting over 70 million poor/.loower middle class) and “ticker” & experience stand against

    ….vs the O man who this 20 strong diverse amarda with perhaps 47 Uni degrees have never been able to explain Obama’s (a) why ‘change , yes we can “ is not explained in any detail (b) why Obama’s Healthcare is a ‘howard’ policy market choice ie. have it if you can afford it but priced by Insurance Companys premiums free market forces less subsidies (c) Why his economc principles springing from (b) imply “oz wet liberalism” , a petal stick to regulation let free market plunder & no big Govt social involvement is more productive and (d) How a Harvard ‘elitist’ cultural attitude implying policy determination computer clinical not “bitter” people friendly will beeffective (e) a TLC oratory with a daffodil wand to deal with tyrants will work (f) Why his priority “progressive & symbolism based” policys will benefit peple more than priority ‘equity & socially based policys and (g) Why the sell out of his own black race at Philly is so intellectually misunderstood but barbarianly historically will be…..so yes yous can say the lingo isn’t uni standard so what the arguments are their clearly for yous to defend and I’m still here not wilting with my coffee heated in a bronze age wok

    Ps
    Ps
    Ps/ how “informed” the people who get this internal info is , ditto as it passes down the rumour chain to peope who don’t get to see the secret polls , vs what the pollsters do vs the ‘pro and mug’ gamblers can wait

  9. Andrew @ 2097: Very very sensible policy that I came to a week or two ago. Was so disappointed when I found out that he was (most likely) a real-life human being with issues, and not a bot programmed by an idiosyncratic ESL computer geek, that I’ve never recovered.

    And no, EsJ, I have no difficulty in responding to others’ points of view. Including yours. But even with appropriate allowance for the low standards of grammar and rationality demanded by internet forums, there are some people with whom engaging is about as fruitful a use of time as trying to teach a Bhutani goat-herder quantum physics through carrier-pigeon correspondence. Written in Braille.

  10. Ron at 2112
    I’m sorry Ron, but I had a bit of trouble understanding the following:

    I’m absurb unless I’m are allergic to historical accuracy.

    Assuming your absurb actually means absurd, can you clarify if you meant that you are (a) absurd, or the alternative (b) allergic to historical accuracy?

  11. Cat, those Intrade odds are on on which Party will win each States electoral votes and is based on that same market giving Obama a 92% chance of leading the Democrat Party and McCain having a 95% chance of leading the GOP (as that is what the same Intrade market currently rates the chances of Obama and McCain getting their respective party nominations).

    The State markets are based on which Party will win.

    As political prediction markets approach an inevitable event, they start to do funny things at the margins. McCain, for instance, will get the nomination unless he (channeling Hills here) karks it between now and the election – but the chances of that happening are really less than 5%. But the Intrade market still gives a 5% chance of something like that happening.

    The reason things like that happen at the very margins of prediction markets is that some people which hold money on contracts that they are going to lose big on, are willing to lose a few cents more on those contract to keep them a little bit longer, hoping that a miracle happens and they can come out winners. If the contract you purchased was 1000 bucks and is now only worth 10 bucks – if you dont think ten bucks is worth much and you can hold on to it a little more in the very low chance that “something might happen” that can get your money back, many people do.

    KR – prediction markets like Intrade actually have two competing forces working at any given time. On the one hand they have the variance of the news traders, those people that follow every poll and every news story and make bids accordingly.This group created a lot of variance, movement and plain old noise in the Intrade price.

    On the other hand are the informed players, the ones that have a better handle on long term issues and inside information. These guys buy and sell when the variance and noise sends the price too far from where they believe it ought to be (by doing so they make money if they are right), and as a result their actions effectively anchor the noise and variance to operate around where they beleive is the appropriate price.

    It’s also why people cant manipulate these types of betting markets for any period longer than the very short term. There’s speculation that Clinton backers tried it on earlier this year and only ended up getting burnt.

  12. Possum

    Electionprojection and electoral-vote don’t seem to use regression analysis, but 538 does. It’s going to be interesting to see which performs the best. I’m looking forward to the PB sweepstakes.

    The stakes are enormous for the various poll sites. Electoral-vote in 2004 was in the top 10 blogs in the world, with about 700,000 visitors a day! And they will get more in 2008. A recent Daily Telegraph listed the 50 most influential US political pundits and the pollsters didn’t rate a mention, which was remiss.

    10 Halperin
    9 Brooks
    8 Stewart
    7 Russert
    6 Drudge
    5 Harris/van deHei (Politico)
    4 Rush
    3 Hannity
    2 Matthews
    1 Rove

    http://www.telegraph.co.uk/news/newstopics/uselection2008/1919961/The-most-influential-US-political-pundits-10-1.html

  13. 2111
    Kirribilli Removal

    Actually I should rephrase that to “those who shoot their mouths off on blogs”!

    The polling is the snapshot, the blogging by those attached to an outcome without evidence are the ones shooting their mouths off!

    And boy, ain’t we seen a lot of that about.

  14. Dio at 2116,

    538 actually uses demographic data for their regressions rather than just polling data – so that adds another twist, although come the election Pablano might realise the dangers of using just demographic data for the presidential election (where it’s far less statistically significant as a driver) compared to the demographic consistency that occurred in the primaries. Each party’s base is demographically consistent, as was the demographic support of potential nominees within their respective party base vote. But the base dont win elections, and certainly don’t produce margins.

    It makes me grimace a bit when I look at the poll sites in the US – there’s not a soul doing anything different (except Pablano at 538 – big cheers for him!), the big polling sites do simple undergraduate polling analysis (but because most of the polls they use aren’t worth a hill of beans, most of it ends up as pure twaddle), none of them that I’ve seen even push a polling average weighted for sample size or use weights based on how far off the pollsters were in predicting the primary results!

    If these guys were in Australia, nearly all of them would go nowhere. It’s pretty funny when you think about it – we few over here on the other side of the big pond actually get spoilt for psephology choice. Billbowe, Mumble, Bryan Palmer, Antony Green, Simon Jackman (although he’s at Stanford), Upperhouse etc.

    In the US they’ve got what? Lots of noise and opinion, a few guys stuffing up poll aggregates and very little else.

    They do have better Youtube vids though – I suppose it’s something.

  15. Possum Comitatus at 2115
    By any chance, are you thinking about running a periodic analysis of the projected general results over the coming months based on the Intrade data? Personally its something I’d be really interested in watching over the coming months.

  16. Ron at 2112 – economic models of election behaviour have a poor to average hit rate, although the Fair model seems to work a little better. But they only predict a YES/NO (Incumbent/Challenger) Party based outcome – not the margin, not the EV’s, and not where the EVs come from.

    Prediction markets have a good record when it come to politics in the US, the proof is in the eating. If you think it’s just coincidence, then it’s a coincidence that defies statistical tests on accuracy – the probability of 50 states being picked correctly at the last election alone by pure random chance is infinitely minute – we’re talking 0.000(insert many more zeros)1 probability here. But getting the mid terms right as well, the mid terms before 2004 to boot, the 2001 election on top etc etc all the way back to 1988 as pure chance? Only if you are walking to the beat of a different statistical dictionary.

    There’s too many states, too many districts, too many Senate races, too many elections in the US where prediction markets have beaten polls (which is to be expected – polls are only predictive if you build a model that forecasts their data). Coincidence was washed away as a viable statistical explanation in 1990 – 18 years later and it’s whistin’ Dixie.

    But prediction markets are flexible, they will change with events, prices will change with the tide of political opinion – as do polls. But unless you start building models for those polls, in the US those polls arent in the predictive ballpark. The uncertainty in those polls kills them before they get out of the gate.

  17. 2128
    Jen

    Being offensive is his only mode of communication. Since Eddy is actually a hyena, it should come as no surprise.

    Eddy’s as big a loser as Gruffy, and don’t they make a lovely couple?

  18. Possum, if you are going to bother compiling all the numbers, maybe you could set up a good site with some google ads for revenue.

    Once the US market discovers you actually know what you are doing, you’d get plenty of traffic.

    You wouldn’t be beholden to the kindness of strangers for donations.

    Only if you’re not averse to ‘going commercial’ that is.

Comments are closed.

Comments Page 43 of 43
1 42 43