Andy Carvin's Waste of Bandwidth

On the one hand, there's the issue of margin of error; when you hear a talking head on TV saying "So-and-so leads the other candidates by a seven-point margin," they don't always follow it up by noting that the poll had a margin of error of four points, making that seven-point lead more like a three-point one.

On the other hand, we can't forget that this is New Hampshire, a place that places much pride in helping kick off what's perhaps the our ritual of civic engagement. Some New Hampshire voters consider the casting of their ballot as a strategic decision, in which they vote for a candidate whose inclusion makes for a better race, if not a winning one.

Jon Greenberg, executive editor of New Hampshire Public Radio, believes that a certain percentage of New Hampshire voters, particularly women voters, decided to vote in a particular way directly because of the wide margin in the polls.

"My thesis is simple," Greenberg told me earlier today. "The polls gave many women a tangible sense of what primary night might be. They didn't like that picture and acted to change it. Put another way, the polls, which we think of as describing reality, became a factor that changed the outcome. Ordinarily, I would not think this possible but never before have we had a primary with a strong, entirely credible, female contender. There are two elements of the NH primary that I think play a key role in explaining the outcome.

"In the traditional framework of the NH Democratic primary, Clinton was not just a woman candidate, but an establishment candidate," he continued. "If you know the work of Dante Scala, the establishment candidate always has the edge with working class voters. In Dante's typology, it is part of the definition of that sort of contender. In that light, Clinton's edge among those voters was typical.

"The other factor that is peculiar to NH is that NH voters can be much more strategic than voters elsewhere. I don't want to exaggerate this, but I don't think you can see the NH electorate during a primary as using the same decision rules as voters elsewhere. There is a systemic difference for some appreciable portion of the population - at least 10% and possibly more. They game the system.

"My theory is that you take these two conditions and throw in gender politics in a way that's never existed before in a presidential race and it's plausible that the polls themselves generated a groundswell reaction among women - and Clinton's edge over Obama would reflect that - that ultimately altered the outcome."

In particular, Greenberg takes note of the voting results from the southeastern portion of the state, and the so-called gender gap. "In terms of polling data, I'd point to the 12-13% margin among women that Clinton had over Obama. She enjoyed equally great margins with lower income and less educated voters but as I said, that's typical of establishment candidates. Less precise but intriguing data comes from the town by town breakdown. Clinton won in the southeastern quadrant of the state where you have many towns with above average household incomes. It's just my guess, but I don't think it was the lower income folks in those towns that put her over the top. I think it was the women."

And then there is the growing number of stories from the voters themselves. Greenberg has spent the better part of the last year involved in Primary Place Online, a community media initiative in which New Hampshire Public Radio created a website for residents of the town of Exeter to publish their thoughts on the election in the months leading up to the primary. Greenberg has examined posts from local women voters on the site, some of who have noted how the polls changed the way they and their friends ended up voting.

One post from a user going by the name Alewife comments on how she's begun to hear from other people who changed their votes due to the wide margin in the polls. "It's noon and I have already met three people who were undeclared voters who chose to vote for McCain so that Romney would not win, but wish now they had voted for Obama," she wrote on the site the day after the primary. "They thought, BECAUSE OF THE POLLS that Obama had it locked up over Clinton." Another commenter replied to her remark by adding, "I have also heard Obama supporters say that because Obama was doing so well IN THE POLLS, they used their vote 'strategically' elsewhere, namely to vote for McCain because they hate Mitt Romney so much."

Another user going by the name MMF explained in great detail how she shifted from being a Clinton supporter to a Richardson fan, then settling as a Clinton voter at the last minute:

Richardson was here on election eve and he was fantastic. I agree squarely with him on almost every issue. I was persuaded to vote for him and keep him in the race too. But then, a student of mine told me of these last polls putting Obama ten points ahead of Clinton. The news confirmed this even yesterday morning. Ultimately, I think it's more important to keep Clinton in the race than Richardson. These polls made me think Clinton might really need my vote and voting for Richardson would help Obama to a big win. That sealed the deal for me. I walked downtown and voted for Clinton.

Another woman named Bricci described how she was one of those voters who didn't make up her mind until the very end:

I myself saw all the candidates speak and studied all the issues that I felt were pertinent. After doing all this I decided to endorse the person most suited for the position of leading our country and that was Christopher Dodd. My second choice was Joe Biden (can anyone see where this is going)? So after the Iowa caucus I was at a loss. I read and reread everything I could get my hands on. I went and saw senator Edwards, governor Richardson and John Mccain. Still as I entered that booth and the curtain closed I was not sure what circle to darken with the sharpie. Taking a deep breath, I filled in the oval next to Hillary Clinton's name. After all the rhetoric, all the speeches, all the reading, I voted with my heart and instincts.

"Are they just anecdotes?" Greenberg asks rhetorically. "Sure, but I think they are relevant." And he's right: three or four examples don't exactly make an empirical data sample. Yet their stories are informative nonetheless.

Assuming there's a connection between the media playing up the wide margin in the polls and some voters changing their mind about how they planned to vote, it raises questions as to what responsibility, if any, the media should have when it comes to how they report their polling results. In some ways, the New Hampshire primary results almost seem like the journalistic equivalent of Heisenberg's Uncertainty Principle, working in combination with the so-called Observer Effect. With the Uncertainty Principle, the more accurately you try to measure the position of a particle, the less precisely you can measure its movement. And in the case of the Observer Effect, the very act of attempting to observe something causes an inadvertent interaction with it, thus leading to an unintended change in its behavior. In this case, the media tried to offer as accurate an assessment as possible regarding potential voting behavior. And by touting both their findings and the consistency of those findings, again and again, they may have directly impacted voter behavior when they punched their ballots.

"The media will continue to pay for polls and they will continue to report the results," Greenberg added. "However, I do think the media have an obligation to present polls more carefully. If I were to make any suggestion for the media, it would be to break the rule of putting the lead of the story first. If the elements that show the uncertainty of the poll were presented first - for example, 45% of voters say they have yet to make up their minds and no difference smaller than 8% should be taken seriously - and the simple comparison numbers were presented second, it might lead the audience to see the polls more realistically. I have a hunch that if the media did something like this, they could effectively encourage the public to treat all polls as hazy shadows of reality rather than as accurate representations of what's really there."

It's certainly an interesting theory. What do you think? -andy