When you pick cherries, beware of the pits

The Atlantic brings us an article about the psychological basis for our nation's (or any nations) political polarization.  As evidence of this polarization, Lane Wallace points to another recent column by Charles Blow of the New York Times, using Blow's arguments to counter beliefs expressed by President Obama that most voters are not merely partisan automatons:

In his press conference following last week's election, President Obama told a questioning reporter that he "didn't believe people carried around with them a fixed ideology"--that if you'd asked most people on Election Day, they would have said that there were some things they agreed with Democrats on and some things they agreed with Republicans on.

Wallace then cites Blow's article as evidence that the President is incorrect in this belief.  Unfortunately (or fortunately, perhaps), Blow's alarmist thesis about the danger of our partisan divide is not very convincing once you look at his data.

Blow supports his arguments with two bits of data.  The first is shown in a graph found here, which illustrates a downward trend in ideological crossover voting in Congressional elections over the last 30 years.  In other words, self identified liberal voters have been less likely to vote for Republican House candidates.  Likewise, conservatives have been voting for Democrats in decreasing numbers. Certainly a possibly sign of polarization, but scale is also important.  As you'll notice, the big drop on the liberal side is from 20% to 10% since 1982.  For conservative voters, the drop is larger, from the low 30s to mid teens.  Both trends indicate a higher degree of partisan sorting, but the small magnitude is not so alarming as Blow makes it out to be and the conclusions of hostile partisan conflict do not necessarily follow from the evidence.  The alarmist could point out that crossover voting has declined 50% (I'm surprised Blow didn't given the tone of his piece), but its just as accurate to point out that crossover was never that likely and has only become slightly less likely.

What's happening here?  There are many possibly explanations, mostly having to do with partisan sorting and the realigment of the parties over this period.  Its no secret that the old Southern Democrats have all but vanished, replaced by Southern Republicans who are of a different party but occupy much of the same ideological space. Likewise the moderate Republican may also be going extinct in some areas like the Northeast.  This has created more polarization in Congress, as measured by roll call votes.

We must remember, however, that the underlying policy preferences of the electorate need not be changed.  The same district which would have elected a Southern Democrat 40 years ago may now elect a Republican instead, but this need not lead to any significantly more gridlock or polarization than it did then.  The voting record may be more conservative overall, but the policy preferences of the district are may be basically unchanged.  If the behavior of the Representative follows these preferences than the policy outcomes need not be necessarily different.

Another thing to remember is that this is a sample of voters, who we might expect to be more loyal to a given party.  This is especially true for midterm elections, which are lower profile affairs where motivated partisans are far more likely to turn out in order to support their chosen political party. 

Finally, there a huge piece of the puzzle left out: moderates.  Self described moderate voters are the largest group in out body politic, and as they go so goes any election.  Obama won because he won more moderate support.  Republicans won this month because they flipped that around. The plurality of voters, by very definition, favor moderation. 

Blow's other piece of data is even less convincing.  He writes:

And the new Republican majority in the House comes to power with a sour sentiment from their electorate: make no deals and take no prisoners. A May poll released by the Pew Research Center found that a plurality of Republican voters said that they were less likely to vote for a candidate who “will compromise with people they disagree with.” They want either steamrollers or roadblocks, not consensus-builders.

Fortunately for us, and unfortunately for his argument, Blow helpfully links to the poll in question.  In looking at it, we see that what he says is technically true: 40% of self identified Republican voters say that a willingness to compromise does not appeal to them when casting their Congressional vote.  Sour, indeed.  But looking closer, we see that almost equal number of that same group, 35%, say that a willingness to compromise is actually a positive for them.  Additionally, 20% are unsure.  So if we take the glass is (more than) half full approach,  60% of Republicans did not express an overt dislike for compromise. 

Further hurting this story is the fact that the other two relevant groups, Democrats and Independents (both of which are larger than Republicans), willingness to compromise won out among voters.  For the whole sample, 42% said that ability to compromise would effect their vote positively, while 29% were indifferent to the quality.  Bringing up the rear were those who disliked compromise in a candidate, with only 22% rating it as a negative.

The funniest thing is that Blow somehow missed the title of Pew's press release for the poll: 

"Willingness to Compromise a Plus in Midterms".

He must have needed a whole bucket for all the cherries he was picking.

One even larger issue with Blow's whole thesis is the fact that he never really addressed what President Obama said about voters!  The President was referring to policy preferences, not merely to partisan vote choice.  Its been proven by smarter men than me that the average citizen is capable of holding a variety of policy opinions which cross what we would assume to be the normal partisan lines.  Context, nuance, world conditions and random chance all factor into citizen's expressed preferences at any given time, be it Election day or any other day of the year.  Furthermore, there is much that strong majorities of Americans do agree on when asked about their individual policy preferences.  This agreement cuts across ideologies, partisanship, and generations.

Political polarization is real.  Its likely growing, especially among elites, and it does create real problems to efforts to create good policy.  But hyperbole and bad science don't help the situation.  Let's not forget that this nation of ours was once so divided that it literally divided, and the results were not pretty.  With that in mind I have to disagree with Blow when he declares that we are in the "twilight of American moderation".