I was reproached lately on one of my mail lists:
I know everyone says the Dems have essentially tailed the Reps in the rightward move, but in fact the partisan divide in Congress now is far greater than it was 30 or 40 years ago, a fact that people like you never bother to address.One often hears this kind of thing, couched in terms like 'polarization' and 'partisanship'. Now since I personally am most impressed by the extent of complete agreement between the two 'parties', this talk confuses me, and I said so.
Doug Henwood obligingly provided some links(*) to research on the subject. As I rather suspected, the research shows something a bit different from what you'd expect.
The exact procedure is not quite clear to me, but it looks like they did something like this:
1) For every roll-call vote (which of course excludes the votes by acclamation that frequently determine the outcome), the researchers decided whether a Yea or a Nay on that particular vote was more 'liberal' or 'conservative'.
2) They then accumulated scores on individual 'legislators' in Congress, these scores being essentially a ratio of the times each individual Solon cast the 'liberal' vote over the times s/he cast the 'conservative' vote. Perhaps the individual votes were weighted somehow, but just how does not clearly appear, after an afternoon's reading.
3) Do a time series by party. Voila, it turns out that party-line voting is becoming more common -- a dire state of affairs which the authors ominously refer to as 'parliamentary'. What this means is that you're seeing fewer and fewer Democrats whose lifetime average falls farther 'conservative' than the most 'liberal' Republican, and contrariwise.
5) None of this has any bearing on the question of whether the parties are closer together or farther apart in a substantive way, or whether the 'center' has moved to the right or to the left -- much less how. It simply tells us whether the parties have dissimilated more or less; whether the region of overlap -- of average voting records, NB, not individual votes -- has expanded or shrunk.
What the research shows is simply that if you add up the the votes, rated on a liberal/conservative scale, then even a Blue Dog has a lower lifetime right-wing batting average than the most liberal Republican.
Non-overlap on cumulative voting records doesn't seem like a very informative proxy for anything, except perhaps the character of the party system ('parliamentary' or otherwise).
And it doesn't say anything at all about the actual effective difference between the parties, nor about how 'policy' is actually made. Right-wing Democrats can dependably bolt across the aisle on every vote of consequence and still have lifetime averages that lie slightly to the 'liberal' side. And both parties can be moving steadily rightward the whole time, an effect which is of course normalized out by the method chosen.
'Polarization' seems a misleading word for this, though. It suggests that the parties are off on opposite trajectories, approaching the respective 'poles' of -- what? New Deal liberalism on the one hand, and Falangism on the other?
Whereas the reality is, I would say, that both parties are moving right, and drawing closer to each other if anything; certainly not farther apart. Those movements are quite consistent with the one-dimensional sorting-out and stratification and increasingly coherent 'branding' of the two gangs, noticed in this research.
You know the old joke about the unfortunate statistician? He drowned in a creek that was six inches deep -- on average.
(*) To wit: