The mystery of maps

MPR’s Curtis Gilbert has a fascinating look at an old theory — that people in the city are DFLers, and people in the ‘burbs are Republican to varying degrees depending on where they live.

The claim appears to be supported by a map lifted from the Institute for Race and Poverty at the U of M.


The story is accompanied by a blizzard of maps showing “election results” for a number of years, ranging from “safe Democrat” to “safe Republican.”

“It’s as if the Republicans are a tribe and they’re living in one part of Minnesota, and Democrats are another tribe living elsewhere,” Humphrey Institute political scientist Larry Jacobs marvels. “It is one of the most striking manifestations of the polarization in our political world today. We are literally living apart.”

The question “what’s wrong with this picture?” can only be answered by first asking another question, “what election are we talking about?” The maps don’t say.

Check the 2006 election results on the MPR Campaign 2006 section.

Here’s the map of the governor’s race, which appears to support the story’s conclusion:


And here is the result of the race for U.S. Senate, which appears to turn the theory on its head:


The political party identification comes from a U of M/MPR poll (pdf) earlier this month in which those surveyed were asked to state which party they most closely identify with. But the survey appeared to find it difficult to correlate that self-identification with other factors:

The inner ring suburbs, which lean Democratic, strongly approve of Governor Pawlenty’s performance (58%). Another surprise is Senator Amy Klobuchar’s 65 percent approval rating in the Republican dominated outer ring suburbs.

At the end of the story, Jacobs acknowledges that people in the suburbs seem more likely to “split the ticket,” meaning they may be as likely to support a DFLer as a Republican (a truism that should make claims that someone is a Republican or a Democrat suspect), just as the maps above show. Why would they do it? Because the issues are different. In the governor’s race, for example, the issues were primarily economic. In the Senate race, the issue was foreign policy, which overshadowed economic concerns. Moreover, these concerns shift over time. The poll, for example, touts that the center city voter is more likely to consider “education” as the top priority, while people in the suburbs lean more to naming the economy.

But you have to be careful with that sort of statement because it’s far too easy to read it as “city voter more likely to be concerned about education than the suburban voter” and there’s really no evidence to support that. The ranking of voter concerns is fraught with peril when it comes to making sweeping judgments based on the rank.

A look at the 2006 District 56 (Woodbury-Lake Elmo) results (here, here, and here)shows the danger of such conclusions. Three DFLers, who ran very strong campaigns that focused on education, knocked off three Republican incumbents, who all ran on economic issues. If you want a lab rat for really understanding the suburban voter, that’s the district to use.

There’s also the danger of using election results as a mirror of what issues are involved. If there’s one thing the cumulative results of Select A Candidate have shown us over the years, it’s that votes for a candidate often stem from factors beyond the issues (and to the extent that it is about issues, it’s interesting to note that Iraq is the #1 issue in the race for president, but health care is the #1 issue in the race for U.S. Senate).

Just as easily, the conclusion could have been: City residents are more likely to be in lockstep with a political party than residents of the suburbs. And woe to the candidate who takes the suburban vote for granted, or attempts to reach a single conclusion on the nature of that voter.

  • There are several issues with those maps, that should be understood before people start trying to parse their meaning, including:

    • All but one represent “winner-takes-color” instead of using shades to represent degrees of leaning on way or another; the world is no more red & blue than it is black & white

    • Area maps distort proportions because we associate mass with quantity. A physically large but sparsely populated county has more visual weight than a smaller, more populous area

    Finally, in my discussions with conservatives and liberals, I’ve found many more DFLers who’ve voted for a Republican and many conservatives who would never vote to a Dem.

    (Collins: Excellent points! But doesn’t the blueness of the Klobuchar victory, and suggest more instances of Republicans voting for a DFLer? But overall, I have questions about how people qualify themselves as being Republican or DFL. They’re not required to register as one party or the other in this state, and clearly people are open to supporting candidates of different stripes. I wonder, whether their definition is based almost exclusively on the top of the ticket?)

  • GregS

    I question using results as a measure of the blue and red of an area. Results are about showing up, i.e. motivation, not values.

    A lackluster candidate, or even a candidate with charisma can affect results without affecting underlying values.

  • Bob Collins

    Yes, that’s somewhat what I meant when I said asking a voter to rank an issue and then saying because education is #1 in the city and because economics is #1 in the ‘burbs means that there’s some sort of gulf between the city and the suburbs.

    I don’t think that by itself proves there is at all, nor do I think people vote strictly on the issue that happens to be #1.

    Suppose the #2 issue (if the pollster had asked) in the city was economics and suppose the #2 issue in the ‘burbs had been education…. therefore we would be able to say that the city voter and the suburban voter say the top two issues are education and economics.

    Suddenly this “gulf” disappears that allegedly makes these two voters so different.

    Couple that with the fact that there’s no standard of measurement for the difference between #1 and #2 from person to person.

    At the heart of this is an attempt to come up with a single definition of values based on geographic location and I think that’s always problematic.

    Especially when how you rank something doesn’t really tell you how you feel about an issue.

  • Bob, There’s no texture to the proportion or quantities of votes represented in the Red/Blue Gov. vs. Senate maps. In the Governor’s race, there were three major candidates, and Pawlenty didn’t win a majority of the vote statewide. He wouldn’t have to win in a county to make it turn red, either.

    Plus, the DFL had a candidate for guv who excited less party support than Klobuchar, while the GOP had a guv candidate who had more appeal than Mark Kennedy. Without looking at actual vote totals, you could not draw any party-crossing conclusions from these maps.