The Invalid and Likely Incorrect Washington Post Poll on the Maryland Race for Governor
–Richard E. Vatz
Gov. Robert L. Ehrlich is quoted in The Washington Post as calling the stunning Postian Maryland gubernatorial poll which showed him down 11 points “out of whack” with any other poll he has seen during the campaign, and it is indisputably an outlier. Just one week earlier, a Rasmussen poll found a 3-point difference between Gov. Martin O’Malley and Gov. Ehrlich. Moreover, all of the recent polls have found the race to be within the margin of error.
Again, this poll finds Gov. O’Malley has an 11-point lead.
How does the Post cover the poll and how should the Post cover it?
It is covered as if it were gospel.
The Post analysis optimally would have at least some – some – skepticism. In its piece on the poll (updated today, September 30) reporters Aaron C. Davis, John Wagner and Jon Cohen report the polling results as if they are immutable facts.
Typical of their certitude is this section: “Instead [Gov. Ehrlich’s attempt to gain cross-party appeal], despite widespread concern among Marylanders about the economy and direction of the state, Democrats have moved solidly behind O’Malley. And he is more popular now than at any time a Post poll has been taken since 2004.“
There is in this major, front-page lead article no hedging about the interpretation of this poll and no mention of its incongruity with other polls.
The poll should be covered with some self-awareness of the imprecision of polls in general and this one in particular.
Mention the atypicality of the poll. Mention the possible instability of the responses and the design irregularities of the poll
Public opinion polls of an election during a recession, which often depresses turnout and maximizes difficulty in determining who are likely voters, are notoriously imprecise, especially if they are at major variance with other polls taken at similar times by reputable pollsters.
This poll offers a mishmash of likely and registered voters, without saying precisely what, beyond the claims of the voters themselves, makes them likely to vote.
Davis, who has generally written fairly on the race, at least has a blog following this uncritical account of the poll and its implications in which he asks and then answers a question that attenuates the certainty of the poll’s implications: “Does that mean the race is over? No – far from it.”
It is important to remember that polls are more often than realized incorrect, despite what their purported accuracy and claimed margins of error would lead you to believe.
For just one example, Hillary Clinton’s support was polled at 9% below Barack Obama’s in polls immediately preceding the 2008 Democratic New Hampshire primary, but she won by 3%. Neil Newhouse of NBC-Wall Street Journal polling said, “With Hillary Clinton’s victory last night, any shred of reputation that pollsters have for being accurate barometers of public opinion goes out the window.”
Pollsters and their journalistic enablers were embarrassed – for about a day.
Again, since there is no way to prove polls’ inaccuracy — since elections do not occur simultaneously – they enjoy a believability they do not deserve.
The danger is that such errancy can affect voting patterns and financial support for those not polling well in an election.
With the Post’s general negative disposition toward Gov. Robert L. Ehrlich, they should be more sophisticated when reporting an unusually negative poll at odds with other polls about the former governor’s support.
Professor Vatz teaches Media Criticism and Political Rhetoric at Towson University