The Washington Post’s Excellent Foray into Intensity of Public Opinion: Will All Serious Media Follow Suit?
–Richard E. Vatz
For years I have been writing about the invalid media public opinion polls which preponderantly inhabit our political landscape. People wonder why so many polls are wrong beyond their margins of error. Neil Newhouse of the NBC-Wall Street Journal poll stated after the 2008 New Hampshire Democratic polling debacle that “We just didn’t see it coming” and “…any shred of reputation that pollsters have for being accurate barometers of public opinion goes out the window.”
Polls are typically error-prone for a number of reasons, including the polling of unrepresentative samples of respondents, the wording of polling questions, and, in the case of pre-election polling, the lack of attention to the intensity of respondents. Some polls have included this information in unpublished or hard-to-find “internals,” but few have ever given such salience to the strength of support for candidates as does today’s (August 16, 2009) Washington Post.
Why is intensity so important? Well, for accuracy in interpretation of public opinion, of course. But there are important consequences for candidates as well.
Say I am running against Kurt Schmoke (whom I like, parenthetically – no subtle negativism implied here) for the mayoralty in Baltimore, and a poll is taken in January of the election year in which he is not so popular; he may win 90%-10% regardless. As I become better known and people want “change,” to coin a slogan, this 80% difference may shrink significantly, as the difference did shrink in 1986-1987 when Clarence H. “Du” Burns ran against him for the Democratic nomination.
But by then it was too late, as Burns had problems financing his campaign, due in large part to supporters’ losing confidence in his viability due to his low standing in the polls.
The serial failings in the polls regarding the Democratic presidential nomination in 2008 were probably due to their either not measuring or not publicizing the intensity of the results among many candidates or just then-Senator Barack Obama and then-Senator Hillary Rodham Clinton. Predictably, some – actually, many – Obama supporters worried about polling inaccuracy due to the mythical “Bradley Effect,” which alleged phenomenon argued that a racially biased electorate would tell pollsters one thing to show themselves racially bias-free, while in the privacy of the voting booth they would turn around and reveal their manifest racism.
Today, The Washington Post’s lead article on Virginia’s gubernatorial contest, “McDonnell Ahead in Governor’s Race,” published on their front page, provides evidence of their critically important sub-head “Va. Voters Aren’t Locked In, Poll Says….”
The poll results are depicted in a pie chart on the front page with the number of “definite” votes, “undecided /no opinion,” and the proportion of voters (almost 40%) who “could change [their] mind” for candidates Republican Robert F. McDonnell and Democrat R. Creigh Deeds. Thus, very significant information is provided regarding the strong lack of certainty of the outcome, information that would be less clear if the chart indicated only that McDonnell had a 7-point lead.
Interesting, it was The Washington Post which, apropos of American-sponsored polls which found Mahmoud Ahmadinejad ahead prior to Iran’s recent national election, printed an op-ed piece defending as legitimate the outcome of Iran’s electing him as president. The problem with that article’s analysis? The polls didn’t take into account the lack of intensity of Ahmadinejad’s support.
Some polling critics have written for years that polling results in major media need to include information regarding the intensity of the results (see, as just one example my “Of What Value are Public Opinion Polls,” USA Today Magazine, May 2006). Today’s Post poll also measures how “enthusiastic” polling respondents are and how much voters claim they know and how closely they are following the election. Also, this poll includes both “conventional and cellular telephones” – a significant, new improvement in phone polls.
(In addition to articles on the invalidity of political polling, I have written specifically regarding the ignoring of the intensity dimension to the Post on innumerable occasions over the years.)
Let’s hope the latest focus on this crucial component of polled opinion continues and is adopted by other media consistently.
Professor Vatz teaches political rhetoric and a course in “Media Criticism” at Towson University