The Times\'s Washington Bureau Chief, and Legions of Others, in Defense of Nate Silver

Nate Silver, author of the FiveThirtyEight blog on NYTimes.com, may be under attack from some people, as I noted in a post on Thursday but he also has many defenders.

Hundreds of them wrote to me in e-mails, in Twitter messages and in comments on the blog to say that they vehemently disagreed with my criticism of Mr. Silver's offer of a wager to a talk show host on the outcome of the presidential election.

Some questioned my intelligence, sanity or sense of fun. Some said that by criticizing the wager offer, I encouraged the unfair critics of his overall methods.

Others questioned the logic of my specific complaint. I'll address the latter here, then turn over the floor to the Times's Washington bureau chief, David Leonhardt.

First off, I want to state clearly that I see nothing in Mr. Silver's writing that suggests his work has partisan motivations. I also will repeat that those who choose to equate probability and a close political race are wil dly missing the point. To use everyone's favorite new word, they are innumerate.

But here is the problem: Mr. Silver's offering a wager could be interpreted, by critics who already paint him as partisan, as evidence that he has a rooting interest in a particular outcome. Yes, even though the winnings would go to charity and even though he was betting to make a point about his model. There may not be a true conflict of interest, but there is an appearance of one. And appearance matters - it affects credibility, which is at the heart of good journalism. (There is a school of thought that rejects this idea and many people articulated that well on Friday.)

Put more broadly: Journalists shouldn't bet on the outcome of news events they cover because betting raises the reasonable idea that they have a stake in how those news events turn out - or that they even might try to make the events take a particular course. That's why business repo rters are not allowed to trade stock on companies they cover.

The Wall Street Journal, for example, has a strict anti-betting policy.

One of the most thoughtful respondents to my post was Mr. Leonhardt. While not defending the wager offer, he pointed out that Mr. Silver has a public record of taking a nonpartisan approach to polling analysis. He wrote:

When liberals were claiming momentum in the final days of the recall campaign against Scott Walker in Wisconsin, Nate threw cold water on the argument and said Walker remained a heavy favorite. Walker won, of course. Nate's model also suggested that the G.O.P. House takeover in 2010 and Scott Brown's win in 2010 were likely to happen.

None of this means Obama will win. A 20 percent chance is a serious chance. The system would be flawed if 80 percent favorites won 100 percent of the time. To put it another way, a pair of dice isn't broken if it rolls a seven.

The larger point is that Nate's work has long earned the benefit of the doubt about its goal: to produce the best polling analysis possible, given the inherent noise in polls and the inherent uncertainty of life.

Later, after conferring with Mr. Silver, he also offered examples of cases in which underdogs in his model have gone on to win, noting that these also reflect no partisan pattern. (As Mr. Leonhardt notes, underdogs “should” win sometimes, “just as baseball batters sometime get hits.” ) The examples follow:

In 2008, Mr. Silver had John McCain, Republican, favored in Indiana, and Barack Obama won.

In 2009, he had gay marriage favored to pass in Maine, and it did not.

In 2010, he had Sharron Angle, Republican, favored in Nevada, and Harry Reid won.

In 2010, he had Ken Buck, Republican, favored in Colorado, and Michael Bennet won.

In 2010, he had the Tea Party Republican, Joe Miller, favored in the Alaska Senate race, and the moderate Republican (Lisa Murkowski) won.

In 2010, he had Bill Brady, Republican, favored in the Illinois governor's race, and Pat Quinn won.

In 2010, he had Republicans projected to win 55 House seats, and they won 63.

There were three cases in which Mr. Silver had Mitt Romney favored in the primaries, but Rick Santorum won.

Mr. Leonhardt adds this praise to his commentary on Mr. Silver:

Nate has done a public service through his work. He didn't cause people to start paying attention to polls. He instead helped people who were already paying attention to polls understand that individual polls can be so noisy as to be directionally misleading - and yet even flawed individual polls often have both noise and information. There is no pollster, no political scientist and no other writer who has a better recent public record of analyzing elections data than Nate. It's not perfect, as he himself tries to convey with his emp hasis on uncertainty and humility. But it's serious, impressive and nonpartisan, and the same is not true of many of his critics' claims.

It's well said and I find much to agree with there. I haven't changed my mind about the wisdom of publicly offering a wager to a television talk show host, but I'll admit that it's a quibble in the overall picture.

And, on a personal note, I want to acknowledge Mr. Silver's generous response to me on Twitter on Friday morning. Clearly, this is someone who understands what it feels like to be under siege.