But whom does it resemble kissing?

Slate.com's political gabfest has solicited sports metaphors for the current state of the race for the Democratic Presidential nomination. I propose that it's like the old college football bowl system and that the analogy has a lot to tell us about how we perceive close contests.

As a method of ranking teams, the old college football bowl system worked well. It worked for exactly the reason everybody else the world says it didn't work: it was messy and indecisive--and therefore accurate. At the end of a season, sometimes you had two or even three or four teams who could make credible claims to be the best in the country. Perhaps one team finished the strongest but lost a game early, another went undefeated against a weaker schedule, another lost two games while its best player was hurt. If these teams did not play one another in bowl games, analysts and fans could build different narratives supporting the claims of different teams to be the true national champions; the official rankings never settled the arguments. Leaving those arguments unsettles was the great strength of the bowl system as an evaluative tool. It was also the great weakness of the system as a mechanism of producing drama. The NCAA basketball tournaments, by contrast, create situations where, say, one referee's decision at the end of a game creates a durable consensus about the relative merits of the two teams playing the game. Tournament play takes contests that are essentially tied and forces them to a decisive result; tournaments produce wonderful drama by denying the messiness of rankings.

The race for the Democratic Presidential nomination will ultimately be a tournament: either Hillary Clinton or Barack Obama will become the nominee, and the narrative of both candidates success or failure will be shaped dramatically by the result. But for now, we have a process, like the old bowl system, that is driving analysts crazy by failing to impose false decisiveness on the contest. Obama does better in caucuses, Clinton in primaries. Obama does better with some demographics and Clinton with others, but most of these differences are relatively small and shift in magnitude from state to state. By awarding delegates proportionally and gradually, the system has done an admirable job of reflecting these uncertainties.

Of course, the nomination process is unlike the college football season in that the Democratic party, for all sorts of structural reasons, really does have to settle on a winner. But the drive for creating narratives of decisive wins--the narratives that produce drama at the expense of accurately messy judgment--has controlled media coverage of the race throughout. Consider the weight placed on Clinton winning the popular vote in New Hampshire, or Obama in Missouri, or Clinton in Texas. But come on: they basically tied. And the gloriously messy, generally proportional methods of delegate allocation have admirably reflected those ties, creating a calm and quiet story of ties and narrow victories while the campaigns and the media shout about decisiveness.

A Presidential nominating process, like the NCAA basketball tournaments, must eventually produce one winner, and the name of that winner will forever distort the narrative of the contest. Our memories will select the facts that make the result seem inevitable.

For the moment, however, let us give thanks for delegate allocations that muck things up, that let similar performances accrue similar benefits without forcing one referee's whistle, or one butterfly ballot design, to take on the force of destiny.

Clinton or Obama: Which Democratic Presidential candidate do the markets like in the general election?

Supporters of Hillary Clinton and Barack Obama both argue that their candidate is better equipped to beat the Republican nominee for President in a general election. Alex Tabarrok at Marginal Revolution first alerted me to the fact that when the Intrade political market contracts for Clinton and Obama to win the nomination were swinging wildly in response to Iowa and New Hampshire results, the contract price for the Democratic party nominee winning the Presidency remained calmly in the low 60s.

(For non-Intrade junkies: the price of the contract is for a share that will pay $100 if the contracted event comes true. That is, if you buy a share of the Democratic nominee for President at $65 and the Dem wins, you get $100, but if the Dem loses, you get nothing. Therefore, the price functions as the market's estimate of probability: a $65 price implies a collective judgment of a 65% probability of the Dem nominee winning the Presidency.)

Since Tabarrok made his post, the probability of the generic Democrat winning the general election has climbed above 65, but many factors could explain the move: Obama's shift to frontrunner status, McCain's emergence as the Republican nominee, increased worries about the economy relative to national security, and so forth. Therefore, I did a snapshot analysis earlier today that derives the answer to this question: according to the markets, would Obama or Clinton give the Democrats a better chance to win the Presidency?

To answer the question, we need the market's estimate that each candidate will win the party nomination and, separately, the estimate that each candidate will win the Presidency. At an arbitrary moment earlier today, the market gave Obama a 71.0% chance to win the nomination and a 47.2% chance to win the Presidency; for Clinton, the numbers were 29.0% and 18.3%, respectively.

The ratio of the second number to the first is the probability of winning given the nomination. Obama's number is 66.5%, Clinton's, 63.1%. Obama gets an edge at that moment, but I've seen moments over the last couple of days that give Clinton an even tinier edge. I would guess that overall, the market is signaling that it considers Obama the stronger nominee by a tiny margin. What's certain is that the market doesn't care much about the identity of the nominee.

Therefore, the supporters of either candidate who have made the case that their candidate has a clear advantage as a general election contender might want to step back and consider that the arguments put forth by the other side are equally persuasive to the bettors on Intrade. I have made such arguments in support of one candidate (Obama), so I include myself among those who might benefit from reflecting on this data.

Super Tuesday Post: A Brokered Convention?

As I write, during the afternoon of the Super Tuesday Democratic primaries, the identity of the Democratic nominee for the Presidency is breathtakingly uncertain: Josh Marshall points out that two of the most influential polling services have come to incompatible conclusions about voters' opinions, and the Intrade futures for the two candidates are currently trading between 49 and 51, with Clinton just above the 50% line and Obama just below.

The closeness of the race has prompted Chris Bowers, in a widely noted blog post, to say that "the most likely scenario" in the Democratic race is a brokered convention decided by superdelegates rather than primary and caucus results. Bowers presents a brokered convention as nearly inevitable barring a whopping Clinton victory in today's elections.

Given the attention the Bowers argument has received, I think it worth noting that Intrade also has a market in the probability of a brokered convention: shares on the Democratic side last traded at $11.10, implying an underlying probability of only 11.1%, with no eye-catching volatility or trends in the price.

Reader, you choose the moral of the story, as it must be one or both of the following:

1. Something in Bowers's account dramatically understates or underweights the ways in which the party can avoid a brokered convention.

2. Bowers is more or less correct, and there are heaps of money available for the taking on Intrade.

On this blog, I try not to make points that are relevant only in a given news cycle. My intent here is to illustrate the tension between the blog narrative and the market and thereby to wonder, in the new age of prediction markets for increasingly contingent and sophisticated contracts, how frequently some of us will be checking market prices to weigh the importance of a wide range of stories.

The Iowa Democratic Caucus, Education, and Poll Reporting

This morning's Washington Post has a routine article about the paper's latest political poll in Iowa. The main thrust of the story is equally routine: Obama has a small lead, but everything will come down to turnout. As an Iowa caucusgoer myself, I scraped up enough motivation to click to the second page of the story and found this paragraph:

Considering other turnout factors brings no additional clarity. Age and education are two key predictors of caucus participation, with older and more highly educated people disproportionately showing up to vote. While Clinton outpaces Obama among older voters, particularly those aged 65 and up, Obama outperforms her nearly 3 to 1 among those with an education of a college degree or more.

THREE TO ONE? Obama is outpolling Clinton three to one among college grads? I am gobsmacked: I've read a lot of coverage of this race, and I would guess that I've seen a hundred times as much coverage of race and gender as education level. Yet there it is: alongside relatively tiny differences in other areas, an enormous gap based on one variable that almost nobody is talking about. Note that the gap isn't even the main topic of the Post's own paragraph: the gap is presented as a turnout factor, not as the crucial difference between the Iowans who prefer Obama and those who prefer Clinton.

In this race, the education level of voters also seems to work against some of the race's main narratives; for example, given the Clintons' alleged association with cultural elites, would we have heard more about this story if the numbers were reversed? Do we even know how to talk about Hillary Clinton as someone who connects with common people but flops among college graduates? I'm not sure we do.

But I also wonder whether this case illustrates a blind spot in political journalism more generally. I imagine so, at least to some extent. It might be easier, and it seems to me more conventional, to talk about political preference in terms of race, gender, and age than education. If I'm right that there is such a blind spot, does it relate to ways in which we do and don't discuss social inequality in America?