Tuesday, October 28, 2008

Intrade Election Market: No Bradley Effect

Analysts of the 2008 election have repeatedly cited the Bradley effect as a reason to discount polling data in elections pitting white candidates against black candidates, arguing that some voters tell pollsters they would vote for a black candidate when they intend to vote for the white candidate. Yet analyzing the data, it appears that Obama shares are performing as well as (or perhaps outperforming) his RealClearPolitics (RCP) polling data, indicating that Intrade investors believe that the RCP polls understate Sen. Obama's margin of victory.

The Bradley effect refers to the California gubernatorial election of 1982, in which the black mayor of Los Angeles, Tom Bradley, lost an election he had been expected to win. In a number of other elections featuring black candidates versus white candidates, the black candidate appeared to fare worse on Election Day than expected (see here for anecdotes from elections from the 1980s through 2006). While there is still active debate about whether the Bradley effect may have disappeared as American social attitudes changed in recent years, the evidence from Intrade markets is clear: Intrade market participants completely discount the Bradley effect.

If Intrade investors believed in the Bradley effect we would expect to see the price of Obama shares below 50 when the polls were even, only rising above 50 when Sen. Obama had a lead in the polls greater than the expected Bradley effect. For example, if 2% of the voters told pollsters they would vote for Sen. Obama when in fact they were supporters of Sen. McCain, then Obama shares would trade at 50 only when he had a lead greater than 2% in the polls.

Since Sen. Obama has had a large lead in the RCP polls in recent weeks, it is difficult to know whether the price of Obama shares accurately reflects the polling data. But for a brief period in early September the polls were very close. On September 2, after Sen. McCain cancelled the first day of the Republican Convention, Sen. Obama led in the RCP polls by 6.4% and his Intrade shares were trading around 60 cents. Then, after the surprise announcement of Gov. Sarah Palin as VP and a reasonably successful convention, Sen. McCain started to rise in the polls.

Sen. McCain actually led in the RCP polls for a short period; he had a small lead from September 7 until Sen. Obama came back to even in the polls on September 17 (and Obama has maintained a lead ever since).

Sen. Obama's closing Intrade price was 57 on September 7 and his shares remained above 50 through September 10, despite Sen. McCain's small and growing lead in the RCP poll. Only when Sen. McCain's lead reached 2.5 points on September 11 did Obama prices drop below 50. When the RCP polls were tied on September 17, Obama shares closed at 48.9, but they rallied to 51.8 on September 19 as Obama gained a 1.9 point in the RCP poll.

In other words, Intrade traders clearly believe that the Bradley effect will not be a factor in the 2008 Presidential election. Prices of Obama contracts remained above 50 at a time when he was behind in the polls, indicating that Intrade traders believed that the polls actually underestimated Obama's likely election performance. When Sen. McCain's lead in the polls grew, the price of Obama shares briefly dropped below 50 but they quickly recovered with the evidence of improvement in the polling data.

Political scientists will spend the next decade analyzing the 2008 polling data and election returns for evidence of a Bradley effect but Intrade investors have already cast their vote.

35 comments:

Intrade said...

Chicago Paul said :

    More sage wisdom from Mr. Kendall

Intrade said...

Alleghany said :

    Very nice indeed

Intrade said...

Philosophical observer said :

    I'm wondering about a reverse Bradley effect...How many conservatives would rally behind the White, but behind closed curtain vote regardless of color. It's gut intuition, the polls aren't taking that into effect

Intrade said...

Suspicious said :

    Problem with your analysis is that the Obama/McCain contract prices were adversely affected by one trader during that time period.

http://www.cqpolitics.com/wmspage.cfm?parm1=1&docID=news-000002976265

Intrade said...

sceptistat said :

    Problem is the popular vote (or national polls) does not elect presidents. One would like to know if his probability of being elected moved closely with his (state by state) polling electoral total or if it was lagging there.

Intrade said...

Coleman Kendall said :

    In response to "Suspicious": Thanks; I just read the CQ Politics article and it was not obvious to me what time period they were talking about. I used the "closing" prices of the contracts as defined by Intrade.

But, in any event, if a trader pushed up the McCain prices, that makes the anti-Bradley effect even stronger, because even when McCain was leading in the polls his shares might (in the absence of the "unusual" trading) remained below 50.

Please let me know if you have more specific information about the dates of the "unusual" price swings, and if they affected the "closing" prices as defined by Intrade.

Intrade said...

Coleman Kendall said :

    In response to "sceptistat": Yes, popular vote does not elect presidents. But in most elections (except, of course 2000 and 1876), the popular vote winner wins the Electoral College. Surely the odds of winning the overall election is increasing with the margin in the popular vote.

Intrade said...

Coleman Kendall said :

    In response to "Philosophical observer": Intrade market participants seem to favor your hypothesis.

And thanks to those who provided encouraging comments.

Intrade said...

RG said :

    There are other possibilities that this analysis glosses over. First, people could expect a Bradley effect, but expect it to be smaller than the edge that Obama will gain from his superior GOTV effort and from the under-polling of cell-phone-only users (who favor Obama by 2-1 and who are expected by most analysts to add a point or more to Obama's total over the polls). Second, it's quite possible that people don't trust the RCP averages. Nate Silver at 538 did a piece a couple weeks ago showing RCP apparently excluding some pollsters when they presented unfavorable results and apparently including those same pollsters when they included favorable results. RCP last I checked does not have a policy on which polls are included/excluded, and that lends itself to this kind of problem. Third, things appear to be quite different now, with traders at this site massively overestimating McCain's chances (as compared to, say, election.princeton.edu's analysis or 538's analysis). Part of the story of change over time here may be that some intraders back in September thought that even though McCain went up in the polls briefly, his Palin choice would eventually sink him, and their trading reflected that idea. The explanation for that part of the story, in other words, could have nothing to do with voters doubting the Bradley effect. Finally, if traders on this site are estimating McCain at 15% probability while other often-viewed sites are estimating his chances at somewhere between 1% and 5%, and Obama is up 10 points in the national polls and at something like 375 EV's in state polls ... that sounds like some intraders might be expecting a possible Bradley effect. (Though this post ignored the overperform/underperform the polls contracts, which could also shed light on this issue.)

Intrade said...

Coleman Kendall said :

    In response to RG: First, thanks for the detailed response. As to your points:

1. You are correct that it is possible that Intrade participants view the Bradley effect as small relative to other factors. That seems consistent with my general point.

2. If you go to the RCP site, they list all the polls they use. While they may select a bit, if you look at polls starting September 1, McCain led or was tied during many of the polls over the next two weeks. It is not obvious to me that these results are dependent on RCP's selection of particular polls. If you look at the 538 article from September 10, Nate Silver had McCain as a slight favorite to win the Electoral College (tough there was a caveat that he thought the polls would bounce back for Obama). But in any event the race was close in the polls and Obama's price was close to 50.

3. I agree that now is different; it is very difficult to translate the current polling data into share prices, and it is not surprising that different places do it differently. That is why I focused on share prices at a time when the polls were closer.

4. Finally, there are certainly other explanations of why Intrade participants would trade at prices that were not reflective of the recent polling data. But while Obama has run a much better campaign in the last six weeks he has also benefited from the financial crisis. If Intrade participants all anticipated that the stock market would crash then they would have been better off speculating in other markets.

5. I think this is a restatement of point 3 and I agree.

Intrade said...

Coleman Kendall said :

    And for RG's coda, the over/under relative to the poll contracts do not seem to have enough volume to interpret reliably.

Intrade said...

Dvorkin in Denver said :

    Ever since discovering Intrade and its remarkable record of accuracy, we have been far more optimistic about the outcome of this race. We have sent the link to the Intrade site to many friends and relatives, and they all find this fascinating. Let's just hope you guys are right, and that it will be Obama in a landslide. Things are looking pretty good so far. GOBAMA!
Leonore and David Dvorkin, Denver, CO www.dvorkin.com

Intrade said...

High-Stakes Gambler said :

    Intrade participants are highly liberal and bet on Obama largely as a fan boy bet. Gotta love the Europeans who see non-stop Obama action on TV and think he's God's gift to America. Thanks guys for the good McCain value.

Intrade said...

Electoral Math said :

    Uh, Coleman, if traders think that 2% say they'll vote for Obama but will really vote for McCain, then Obama would have to lead by 4% (52% to 48%, after adjusting for undecideds), not 2%, in the polls to earn an even market price with McCain. Sure, Obama's got the election locked up now, but it will be interesting to see how the Intrade numbers hold up in the state-by-state vote counts.

Intrade said...

Coleman Kendall said :

    In response to Electoral Math: I think we agree; I looked at a period when the polls were very close (and McCain had a small lead in the polls) and Obama shares were above 50 for much of the time. So Intraders believed that the polls were, if anything, a bit tilted toward McCain.

High Stakes Gambler proposes another explanation of the same phenomenon; Intraders are biased toward Obama for other reasons and hence do not show a Bradley effect.

Intrade said...

fanningtheflames said :

    I think High-Stakes Gambler is interesting, the suggestion is that people may either be deluded in the market, or be trying to influence the election by triggering an avalanche effect for Obama voters and be willing to lose money to do so (the Obama campaign is loaded).

Presumably point 1 should be balanced in the market by delusional Republicans, known I think as Palinistas, or otherwise as the 25% of the electorate that still think Bush is a good President, but the second point could be true.

If you assume High-Stakes Gambler represents the McCain campaign, either formally or just as a supporter, which has been flooding Internet commentary sites with identical comments pumping up the McCain vote (I saw three identical "Obama has blown it, he called Palin a pig" with the same hyperlink on three different sites) then they are attempting the same without financial commitment. Arguably you have a situation where the market could be overpricing Obama, but the fact that Obama supporters are willing to do this may well just demonstrate the greater commitment that Obama supporters have, and thus it may well represent the market pricing this commitment in and thus be reflected in the overall results anyway.

To actually benefit from this intervention in the markets, as High-Stakes Gambler suggest he is doing, McCain supporters would have to put their money where there mouth is and this would show in the market, thus nullifying the effect.

Where is the fault in my logic?

Intrade said...

PanyJamon said :

    It is going to be closer that we think.
http://www.gallup.com/poll/111619/Gallup-Daily-Obama-Maintains-Edge-Over-McCain.aspx

Obama will carry the liberal states by a large margin, but McCain has a chance, Republican votes are always 2-3% underestimated by polls, and leaving aside Bradley, which is totally unpredictable, as voters themselves are generally self-deluded and refuse to accept that they could be prejudiced.

Intrade said...

jhr00000 said :

    High Stakes Gambler - are you taking the McCain trade then?

Intrade said...

High-Stakes Gambler said :

    There are a few contracts where I think traders love for Obama is clouding their judgement. You'll see this more often on some of the more thinly traded contracts (like the states and/or total electoral votes) vs. the main election contract. I think Obama is a bit overvalued on that, though he is and definitely should be the favorite.

I'm not going to say which one of contracts I think the markets gone Obama crazy on because I am betting on these currently and don't want to inadvertently screw myself by persuading Obama bettors to not put up dead money.

The clearest example of one in the past was West Virginia where Obama shot up to 40 on like one BS poll. The Intrade market is so poll happy that they'll bid up a contract rapidly on one poorly done poll, without looking at the fundamentals of a state like West Virginia that:

1. Voted for Hilary 2:1 in the democratic primary (and I promise you it's not because they love Hilary that much)

2. Bush won West Virgnia by 13% in 2004. Given Bush won the general election by 2.5%, it would imply Obama will need to win the general election by at least 11%. Now, relatively speaking, West Virginia doesn't exactly have much of Obama's core fan base, so he'll probably underperform in that state relative to some other states where he may overperform...so a bet on him to WV would essentially imply him winning the general election by 15%.

The WV contract has now come back down to earth, which is why I went through this analysis...but some of the others haven't.

Intrade said...

fanningtheflames said :

    High-Stakes Gambler, if you have already bought in those markets the what happens now shouldn't affect you even IF Obama now falls in value and McCain rises, you still profit from buying McCain cheap.

Therefore your concern about Obama buyers suddenly leaving the market when exposed to your great wisdom showing them to be deluded makes no sense, so your refusal to say in what markets the liberal effect lies is illogical.

In fact your comments should be taken into account by the market process of aggregating information. It may be that we are now being told that McCain is overpriced as you try to guess the nature of the markets future values based on your belief that others are buying not on fundamentals but for ideological reasons. This leads you to believe that the futures are incorrectly priced, but you are not pricing based on fundamentals yourself, but on a belief in others believes.

Again, if my logic is at fault please enlighten me somebody.

Intrade said...

Trader said :

    Very nice

Intrade said...

Richard said :

    The Obama campaign is starting to act like Robert Mugabe's ZANU-PF, retaliating against reporters from disloyal newspapers.
The popular vote will tighten but Obama's electoral vote lock will be hard to break. If the election were postponed a month McCain might have a chance; otherwise it's too late. Full disclosure: I will never support McCain because of the war.

Intrade said...

Okie2008 said :

    80% of people approached by a pollster refuse to give an answer!! The highest non-response ever. If Obama is so bloody confident, why is he in Iowa today? It is very unlikely that INTRADE.com will prove to be as accurate as it has it the past.

Intrade said...

High-Stakes Gambler said :

    fanningtheflames, the Indiana contract was perhaps the juiciest of them all. I wish I got more on this one (I'm pretty much done betting for the election now). I didn't want to say anything before since I wanted to be able to get my sell orders on Obama at 43 or better filled. I'm pretty full on Indiana at this point, so I'll go ahead and make a brag post.

A major reason Obama surged on the Indiana contract was a SurveyUSA poll which showed Obama up by 4 points. If you look at the crosstabs of the survey though, you'd notice they surveyed an equal number of 'democrats' and 'republicans.' That's ridiculous given the CNN exit polls of the 2004 election showed Republicans outnumbered Democrats by 14 points (even if there are more Democrats now, having them be equal is downright ridiculous).

That, plus you always have to wonder how 'independent' many of these so-called 'independent' poll responders are.

Intrade bettors often just have drank the Obama kool-aid and will park their money on a bet because of one or two poorly done polls. Thanks for the money.

Bush won Indiana by 20 points. Even if Obama increases his share of the overall popular vote by 10 points, that means he'd have to outperform in Indiana by 20 points. This, in a state where there are few blacks and others that fit into his coalition.

Unless you expect MAJOR outperformance in Indiana relative to other states, you'd have to expect Obama to win the presidency by about 20 points, something that simply won't happen.

Intrade said...

Ask Dr. Stupid said :

    There will be at least a small Bradley effect in '08 but there's no way it is going to lose Obama this election. This is beacuse of what I'm calling the 'Bush Effect' - wherein white, middle-class, centrist, small government Republicans who are disgusted by Bush's performance (and who think McCain is 8 years too late and too old) will be quietly voting Obama despite the McCain-Palin signs on their (our) lawns...

Intrade said...

Scienceprover said :

    The funny thing is, almost all the guys in my circles who trade on intrade.com are Obama supporters. But guess what? They are backing (buying) McCain to win to hedge their disappointement on election night - if Obama loses, they make good money to drown their sorrows in Jack Daniels; if McCain loses, they are emotionally satisfied at the expense of a few dollars that doesn't matter anyway.

Intrade said...

skak said :

    This will be the first time intrade is wrong on an election. When there was 3 weeks til election pollsters showed the largest number of undecided voters (18%)ever recorded that late in an election process.
Of that 18% , I can safely say that 0% is black. Therfore the 18% is divided amongst white and hispanic voters.Because hispanics make up 33% of the population I can divide the 18% in 12% white and 6% hispanic voters. If both groups follow normal voting demographics the hispanic vote will break 3.5% for Obama and 2.5% McCain, a net of 1% for Obama. The 12 white voters will break 8% McCain and 4% Obama for a net of 4% McCain. Therefore McCain will gains 3% across the board when undecideds go to the polls. Factor in the fact that Obama continually overpolled in the primaries by 5% and I beleive that over polling continues today as the big 3 networks all show Obama up by 14-15%, which is rediculaus as all of their polls only included 20-30% Republicans in their samples. This leads to unrealistic averages in the oft so quotes realclearpolitics daily averges.
If both facts follow thru then McCain wins the popular vote 51-47, but the elctoral college is a different thing.

Intrade said...

Liberal bias? said :

    McCain leads Obama in the latest one day tracking poll by Zogby, 48-47 (11-1-08).
http://www.zogby.com/main.htm

The polls have been oversampling dems to reps by 7-12 points, higher than what has ever happened historically. The media justifies this because of higher supposed turnout by blacks and youth. Unfortunately, the exit polling data of early voters does not confirm that.

Here are two good places to get daily analysis on the polls.
http://strata-sphere.com/blog/
http://stolenthunder.blogspot.com/

Here is an excellent analysis on gaming the polls.
http://www.zombietime.com/lefts_big_blunder/

And yes, in the spirit of full disclosure, I holding long positions on McCain, and short positions on Obama.

Can't wait til Tuesday.

Intrade said...

atp2007 said :

    I think the so called Bradley Effect was either an LA phenomenon or a misinterpretation of the polls. I have lived in several areas of the country and never met anyone who would not vote for a blcak candidate who was afraid of saying he was voting for the white candidate, so maybe it was an LA thing if it did happen. I think it os more likley that Bradley's affirmative numbers were exagerated by the choices of a % of voters who supported him but did not vote when it came to it. I looked this year at a Pew survey that showed an Obama lead of close to 9% but that included a number of voters who were not registered, had not voted before and probably were not going to bother this time either. If you subtracted them out,Obama's real lead would have looked like it had a Bradley effect. It would not surprise me that if somebody looked over the Bradley race poll numbers again, they would see there was something other than whites not admitting to voting anti-black going on. When biased whites now have a good excuse not to gve for not voting for Obama (i.e. McCain is a hero), why would they say they were going to?

Intrade said...

Racism Light Versus Racism Regular said :

    Racism regular is when a voter will never vote for a black candidate no matter what. Racism light is when voters who are generally prejudiced against blacks will vote for one who they feel is exceptionally brilliant. In fact some racist lights will actually vote for an inspiring black candidate while having a low estimation of blacks in general. I surmise that racism regular is no longer widespread.

Intrade said...

Neihart said :

     A closet racist who buys shares on Intrade knows that he will not vote for a black man, but perhaps assumes that he is the only one with this shameful intent. If he believes that others will vote for Obama because he is the more appealing candidate, he will buy shares in Obama even if he plans to vote for McCain. I don't think the Bradley Effect applies to decision markets.

Intrade said...

SRQ said :

    Could it also be that an Intrade market is susceptible to the same "irrational exuberance" that the regular stock market is? Not saying that it is, its just something interesting to think about. The market of the last few years got way overvalued. Only recently did we discover that the underlying assets were not worth what we thought. Could be the same here.

Intrade said...

Let the hammer fall said :

    I think Obama will overcome any Bradley effect with his large advantage in political contributions... He will get out the vote and convince people he is a viable candidate because of his HUGE financial advantage. That's the difference between Obama and Bradley, or any other "typical" nonwhite candidate. Don't forget... Money = Votes and Obama has no shortage of funds.

Intrade said...

Insult Comic Dog said :

    But he's a nice black man. Like the Cosbys.

Intrade said...

fanningtheflames said :

    highstakesgambler

"Bush won Indiana by 20 points. Even if Obama increases his share of the overall popular vote by 10 points, that means he'd have to outperform in Indiana by 20 points. This, in a state where there are few blacks and others that fit into his coalition."

oops somebody did not understand how the markets work,

very drunk tonight on O'bamas victory (and beer)