The Redundant Gonzales Gubernatorial Poll of Late July, 2010: Piling on with Feathers
–Richard E. Vatz
Following The Washington Post poll of about two months ago and the Magellan Strategies and Rasmussen Reports polls earlier this month, Gonzales Research and Marketing Strategies reports its stunning findings: “The race for governor as of late July is really close. Incumbent Governor O’Malley has a slight lead in our poll, but it’s clear that either candidate could win in November.”
This must be quite a shock to many political observers, as the Gonzales poll results are reported often without reference to its virtually identical findings with the polls that have preceded theirs. All find the race a toss-up within the mythical and unverifiable “margin of error.”
Since I have no vested interest in polls (full disclosure: I have had some public disputes on polling with Gonzales, but I genuinely do not dislike the personnel there) other than wanting them to provide more valid information, I submit the following questions (and provide some answers) regarding polling and media reactions:
1. How many nearly identical polls must come out before each new one is no longer treated as surprising news and recent such polls are not ignored?
2. What can mildly innovative pollsters do to ensure that their polls provide new information, even if the basic results appear to be merely corroborative of other political polls?
3. What can pollsters do to avoid having their results unnecessarily contaminated?
For question one, all it takes is for journalists -– reporters, editors, and op-ed writers — to recognize that polling is continuous, and regardless of financial ties to one pollster or another, news writers should take the entire context of contemporaneous polls relating to specific races or issues into account in their reportage.
Regarding question number two, in several pieces in The Baltimore Sun last year I was critical of a Gonzales poll measuring the gubernatorial race. I mentioned the “overwhelming publicity advantage of the current governor – and the fact that Mr. Ehrlich is not currently a candidate.” The implicit prediction was that the race would significantly tighten and that the poll was also misleading due to its lack of measuring intensity of respondents’ responses. Several polls now measure intensity as well as specifying different measures for the likelihood of respondents’ voting. Why not Gonzales?
On question number three, Gonzales and some other pollsters insist on including questions about the President (in the past, for example, George W. Bush; in the present, Barack Obama) in their polls on other major races. This may contaminate results by making respondents’ satisfaction or dissatisfaction with the Chief Executive more salient and bleed into their evaluation, for example, of the gubernatorial candidates. Why do they continue to do so?
Horserace polls will, like the poor, always be with us, but there is no reason that journalists cannot cover their significance with some more skepticism and sophistication. Also, if some media ignore some polls because those media are not in a financial relationship with a specific pollster, the reporting on the race eliminates important data for getting at its true picture.
Similarly, if pollsters refuse to make the effort to make polls more informative and valid, we get a false picture of the status of important contests, even though it is never provable that the picture is a chimera.
Professor Vatz teaches political rhetoric at Towson University