Pollster.com

Articles and Analysis

 

NH Poll: Horserace or Betting Line?

Topics: 2008 , The 2008 Race

Yesterday, I wrote about an aspect of the way the media has been covering the campaign that "makes me want to scream." Today, we have a story about a new poll in New Hampshire that may turn me into Howard Beale.

Via The Page we learn of a new telephone survey of just 401 "likely primary voters" conducted November 1-4 by Boston/New Hampshire television station WBZ and Franklin Pierce University (story, results, tables). Given the small sample size (which includes likely voters for each primary), the initial intent may have been to focus on issues of interest to all primary voters rather than the usual trial-heat results. Issues were the focus of the poll story that WBZ broadcast last night. But that is not the way it worked out in their online article.

The headline on the wbstv.com article proclaims:

11-06%20wbz%20headline.png

The lead:

Former Massachusetts Gov. Mitt Romney and New York Senator Hillary Clinton continue to hold on to their lead in the latest WBZ/Franklin Pierce University New Hampshire Primary Poll.

The story also characterizes Hillary Clinton's "lead" as "very strong."

The problem? The WBZ/Franklin Pierce poll did not ask a question about vote preference (at least not that was referenced in the story or any of the materials posted online, and our calls to the number provided in the PDF were not answered). Here are the two questions they asked that were referenced in the story:

Q07: Of all of the Republican candidates running for President, which one do you think is most likely to win the New Hampshire Presidential primary?

Q08: Of all of the Democratic candidates running for President, which one do you think is most likely to win the New Hampshire Presidential primary?

Memo to WBZ: The New Hampshire primary works differently than the Power of 10. When voters go the polls on primary day, the ballot will ask for their vote preference, not the candidate they expect to win.

The difference between the candidate that voters support and the candidate they believe will win can also be huge. For example, on the CNN survey conducted earlier this month, 51% of Democrats preferred Hillary Clinton (in a race without Al Gore), but 64% believed she was the candidate "most likely to win the Democratic nomination. Among Republicans, 27% said they would be most likely to support Rudy Giuliani, but 50% believed he was most likely to win (numbers from CNN release).

So if you are going to cover the horse race, please, cover the horse race. Not the betting lines.

 

Comments
bob:

Ugh, a new low water mark.

____________________

Aditya S.:

As much as this poll may be driving you crazy, I have long thought that this approach (asking people who they thought would win--not who they were voting for) could be a much more accurate predictor of outcomes than traditional polling.

This type of polling taps more deeply into the wisdom of crowds and I think gleans much more information than a conventional poll. In a regular poll, someone who is voting for a Mike Gravel or Dennis Kucinch isn't really giving us any useful information about the winner. But in this poll, where presumably some of the voters for those trailing candidates go in knowing they are voting for a losing candidate and will help give data about perceptions of who the real winner will be.

I actually asked James Surowiecki (author of "The Wisdom of Crowds") about this kind of polling at a talk he gave about a year ago and he believed that it would be more accurate, and that it was becoming more common in European elections.

____________________

DTM:

I'd like to see hard data on the predictive abilities of such polls, but I am personally skeptical about whether they will be strongly predictive. As other polls have shown, most people do not closely follow politics and are not aware of many facts (current and historical) that would be relevant to predicting an outcome. So it seems a little odd to assume the public would necessarily be making a well-informed prediction (more likely, I think, is that they would be reflecting the conventional wisdom in the most popular mass media outlets).

Moreover, predictions right now are only as good as the available information, and we lack information about key matters, such as all the events which may transpire between now and voting. So, even the best-informed predictor at this point lacks much of the information he or she would need to make a reliable prediction (absent a functioning crystal ball).

By the way, in most circumstances it turns out that crowds are not all that well-informed. For example, the efficiency of the financial markets (the classic example of this theory) is actually driven by a small number of people who are very smart, very well-informed, and very well-funded, and who seek arbitrage opportunities in the markets. These few people end up making financial markets efficient by competing among themselves for such arbitrage opportunities, and so market prices very quickly get driven to the efficient point.

But that arbitrage process is not at all the same as process in which, say, the market price was set by polling all the market participants on the right price. And that would actually be the closest analogy to this poll.

And the same quality of information point applies: efficient markets process all the information available at the time, but that doesn't mean they are never wrong in hindsight. And that is because there is much information about the future that is unavailable to anyone, even the smartest and best-informed people around.

____________________

Mark:

Would it be too much to ask for that poll to be struck and not used?

____________________

Rich B:

This reminds me of the exit polls that were conducted AND REPORTED ON during the Presidential election of 2004. The poll was skewed to make it appear John Kerry was leading. It was an obvious attempt by the MSM to influence the election results. These ****-stains never fail to amaze me with their tactics. No bias in the MSM. Yeah right!

____________________



Post a comment




Please be patient while your comment posts - sometimes it takes a minute or two. To check your comment, please wait 60 seconds and click your browser's refresh button. Note that comments with three or more hyperlinks will be held for approval.

MAP - US, AL, AK, AZ, AR, CA, CO, CT, DE, FL, GA, HI, ID, IL, IN, IA, KS, KY, LA, ME, MD, MA, MI, MN, MS, MO, MT, NE, NV, NH, NJ, NM, NY, NC, ND, OH, OK, OR, PA, RI, SC, SD, TN, TX, UT, VT, VA, WA, WV, WI, WY, PR