The Instability of Polls: The Gubernatorial Race Between Martin O’Malley and Robert L. Ehrlich, Jr.
–Richard E. Vatz
Regarding the upcoming race for governor in Maryland, Rasmussen Reports says flatly that according to their polling, “The two men (O’Malley and Ehrlich) are now tied…”
Last fall I was involved in a dispute played out in The Sun with Gonzales Research & Marketing Strategies regarding their polling which had indicated that former Maryland Governor Robert L. Ehrlich, Jr. was trailing Governor Martin O’Malley by an apparently not insignificant 11 points, 49-38 percent. By January of this year, their results were 48-39 in the same direction.
I criticized harshly the fact that the fall poll ignored the critical dimension of “intensity,” a measure that related to the stability of the outcome. The pollsters made no attempt to measure how strongly respondents felt about their choices. Therefore, Gonzales’ interpretations could not be assumed to be valid.
Relative voter support, even of leaders who are well known, is often soft until candidacies are announced, and even then they may remain significantly variable until the race begins in earnest in the last several months of the campaign. (When candidates are not well known, the instability can be much more pronounced – see, Palin, Sarah, Vice Presidential race, 2008.)
Consistent with these early misgivings regarding the Gonzales poll, Rasmussen Reports reported days ago that their poll of 500 likely voters indicated a dead heat, 45%-45%, between O’Malley and Ehrlich with a margin of error of +/- 4.5 percentage points, with a 95% level of confidence. Intensity in terms of strength of “favorability” is measured in this poll and found to be roughly equivalent between the two candidates.
Five hundred respondents is an insufficient number for a poll of Maryland voters to be very precise, and the internals showed that Rasmussen determined “likely voters” through undefined criteria of “voting history, interest in the current campaign, and likely voting intentions.” These sound like simply the unverified claims of the respondents in a phone survey. I should add, parenthetically, that margins of error are impossible to validate, as are levels of confidence.
Regardless, the “dead heat” finding, even with its inescapable and escapable invalidities, is not irrelevant: it means that the race is probably up for grabs, an inference at variance with an 11-point differential last year by the Gonzales pollsters, the meaninglessness of which some of us articulated even then.
The Rasmussen poll reported a differential between the two then-likely gubernatorial opponents earlier this year of 6 points, followed by 3 points later in the year, followed by this even closer race.
The poll has some problematic components, such as contaminating questions – a problem with the earlier Gonzales polls as well — referencing President Barack Obama, oil spills and immigration, all of which questions, along with others, are cues to respondents which may affect their gubernatorial choices in the same poll.
The bottom line is that we really don’t know even within 5 points what the vote for Maryland Governor would be if it were held today, but the poll along with the proliferation of signs and intuitive criteria all point to an election which no very serious observer can predict with confidence.
Every move is consequential now, gentlemen. Who will be the Republican Lieutenant Governor nominee?
Maybe the choice will make the “gentlemen” reference anachronistic and shake the polls — and voter preference — as well.
–Professor Vatz teaches political rhetoric at Towson University