The Island of Doubt

Did Clinton win the New Hampshire Democratic primary Tuesday because her name came before Obama's on the ballot? SciBlogger Matt "Framing Science" Nisbet has a couple of posts referring us to someone who seems convinced she did. I'm not so sure, but find the mainstream media's reluctance to at least entertain the notion at little curious.

According to social scientist, and presumably survey expert, Jon Krosnick of Stanford University, serious candidates whose names appear higher on ballot lists enjoy a significant advantage:

Our analysis of all recent primaries in New Hampshire showed that there was always a big primacy effect — big-name, big-vote-getting candidates got 3 percent or more votes more when listed first on the ballot than when listed last.

Krosnick explains that NH used to randomize the list of candidates, with different lists in each precinct, precisely to avoid this phenomenom, which is supposely well known. But this year, all ballots were ordered the same way, with Clinton appearing far ahead of Obama on the list (Thus the conspiratorial undertone.)

If true, it would help explain why Obama polled ahead of Clinton in pre-election surveys, which would have randomized their lists, but came behind Clinton on the poll that mattered. But is it real, and if so, how strong is it? I found several mentions of the "primacy" effect, including references to a NH Supreme Court decision that led to the now-abandoned randomizing. But I couldn't find a lot of good science to support it.

Here's the intro from one paper, from a pair of Australian researchers, titled "Are Ballot Order Effects Heterogeneous?"

Past studies of ballot order effects have typically focused on the average benefit to a candidate from being placed at the top of the ballot. But it is possible that this simple average may mask systematic differences in how the ballot order effect varies across candidates and voters. To test this, we analyse all Australian federal elections from 1984-2004, a dataset that is an order of magnitude larger than those used in previous ballot order studies. We find that being placed first on the ballot increases a candidate's vote share by about 1 percentage point.

On Tuesday night, Clinton received 39.1 % of the vote to Obama's 36.5. So even if all of Clinton's extra percentage came at the expense of Obama's she'd still have won. If, that is, the Aussie study is valid. Of course, if Krosnick is right, the numbers would have been reversed, and Obama would have taken the Granite State.

I have a relatively high degree of confidence in the idea that there are a small number of NH voters who are so pathetic when it comes to carrying out their democratic duty that they choose the first name they recognize. It is a state, after all, that puts "Live Free or Die" on their license plates. And I'm sure Krosnick has some interesting data, but I'm not convinced there would be enough to make a 3 % difference. I'd like to see a better analysis before drawing any conclusions, and it is worth investigating.

Incidentally, one thing that almost everyone seems to have overlooked is that both Clinton and Obama were awarded 9 delegates to the convention. So neither actually "won" the state on any level but public perception. And regardless of which researchers are right about the primacy effect, it is hard to see it having any real consequences when it comes to awarding delegate numbers, at least in New Hampshire.