Dog Days

Between all the fiscal cliff-hanging and the preparations for the inauguration later this month, nobody inside the Beltway is paying much attention to the burgeoning political-science literature on the electoral significance of presidential dog ownership.

Well, official Washington has its priorities, and I have mine. A paper on the topic appears in the January issue of the American Political Science Association’s journal PS: Political Science & Politics. “Burgeoning” is something of an overstatement, but it’s the second time an article on dogs and the presidency has appeared there in a couple of years. So, close enough.

The work of Matthew L. Jacobsmeier and Daniel C. Lewis, two assistant professors of political science at the University of New Orleans, “Barking Up the Wrong Tree: Why Bo Didn’t Fetch Many Votes for Barack Obama in 2012,” is full of statistics and (let its title be fair warning) puns. Their argument builds on the work of Diana C. Mutz (right, I know) whose paper “The Dog That Didn’t Bark: The Role of Canines in the 2008 Campaign” appeared in the October 2010 issue of PS. It would be more accurate to say that Jacobsmeier and Lewis undermine and overturn her analysis, but at least they are friendly about it. (Mutz is a professor of poli sci at the University of Pennsylvania and Princeton University.) wait -- she's at both universities? dl

Documentation of the role of pets in the history of the executive branch already existed when Mutz set to work, though it was, for the most part, anecdotal. But she could cite a survey from 2006 indicating that, in local elections at least, not quite 99 percent of dog owners responded that “a candidate’s position or track record on issues such as breed discrimination, breed bans, or leash laws played a significant role in their electoral choice.”

That statistic is at least somewhat questionable, coming as it does from My Dog Votes™, identified as “the world’s only company with a mission of Saving Dogs and Democracy … [by means of] clothing, accessories, and real campaign gear.” And the effect of dog-related issues on voter behavior during national elections remains very much an understudied question. Be that as it may, Mutz ventured a significant interpretation of the 2008 election -- which, while historic, was short of the landslide many expected.

She wrote: “Early in his run for the presidency, Obama made a widely publicized promise to get his daughters a dog after the election, regardless of the outcome. This gesture may have been superficially endearing, as promises go. However, I argue that in the end, this promise backfired on Obama by raising the salience of his family’s doglessness and thus alienating a significant proportion of the population.”

Mutz drew on data collected by the National Annenberg Election Study, a poll that “tracked a large, randomly selected sample of respondents throughout the 2008 presidential campaign.” Pet ownership was one of numerous characteristics recorded in the survey, along with gender, income, educational level, size of household, gun ownership, party identification, and the respondent’s perception of whether the economy was improving or worsening.

The problem was to determine how much weight dog ownership had as a variable affecting voters’ feelings about whether they would be likely to support a candidate. That means taking into account, through regression analysis, the strength of the other factors (gender, income, etc.) and any confluence between them. The results varied across Mutz’s four models, but dog ownership consistently proved to be a negative predictor for an Obama ballot for 1.7 to 5 percent of those surveyed – and among subjects who reported their votes, “the odds decreased by 16 percent if the respondent was a dog owner.”

Mutz offered two possible explanations for this remarkable gap. One was the failure of group identification: “The minimal group paradigm suggests that in-group favoritism can be stimulated even by very weak, transient, and meaningless group identifications.... Whether for symbolic or imputed substantive reasons, group identification theory suggests that, all else being equal, dog owners should be drawn to dog-owning candidates.”

An alternative (not mutually exclusive by any means) was the “congruity-oriented theory” that owners of a particular sort of pet will prefer candidates with similar characteristics, such as “emotional transparency and straightforward displays of emotion” in the case of dogs. That would present difficulties for an altogether feline politician such as Obama.

The scholarship has advanced considerably since the says of Gibbs Davis’s Wackiest White House Pets (2004) and it should come as no surprise to learn that others have revisited Mutz’s data from an alternative perspective. While admiring her analysis as “particularly elegant and compelling,” Jacobsmeier and Lewis challenge it on the basis of “our graduate school experiences [which] included Pavlovian training in the detection of omitted variable bias.”

The omitted variable, in this case, is region. The American Veterinary Medical Association reports that 37.2 percent of American households included a dog in 2006, but they are not evenly distributed. “Rates of dog ownership clearly vary with geographical location,” write Jacobsmeier and Lewis. “Using census region as the geographical unit, dog ownership is most common in the South and least common in the Northeast.”  

The data also shows that “a large gap in dog ownership exists between black and white respondents” -- with whites having the higher rates, as do gun owners, home owners, and people living in rural areas. Mutz’s regression models took into account respondents’ party affiliations and how strongly they identified themselves as liberal or conservative, and tried thereby to isolate dog ownership as an independent factor. Instead, it proves to be a kind of proxy for “red state”-ism.  

Among the pools of data the authors tapped into while doing their research, evidently, was an exhaustive collection of canine-pertinent verbs, images, sayings, etc., every single one of which was then incorporated into the paper. It seems like something best done with monomaniacal thoroughness, if you’re going to do it at all. I have managed to keep most of them out of this column, but you can find them all -- and many other interesting points scanted here -- in a prepublication copy of the paper here.