|History and Causes
Southern realignment is one of the most potent explanations for polarization. With the sharp decrease in white Southern conservative Democrats and the increase in Southern Republicans, the parties have become more ideologically homogeneous and more concentrated in particular geographic regions of the country.
Nationwide, the population share of liberal Democrats barely changed at all from 1972-1984 to 1994-2004, 13.5% to 13.7%, while conservative Republicans increased by over 50%--from 11.2% to 17.1%. Clearly, there was only one extreme that was growing stronger during this time period:
This comparison alone should be enough to conclusively demonstrate that any increased polarization resulting from sorting came from the conservative Republican side, not from "both sides."
Of course, the decline of moderate and conservatives Democrats left the liberals in a stronger position within the party. But just how strong? A comparison of the within-party percentages makes this clear at a glance:
In the last period, liberals were just 39.1% of the Democratic Party, less than the 39.7 who identified as moderates. This contrasts with the Republican Party during the first period, when conservatives constituted almost half of the party--49.1%. In short, liberals during the 1994-2004 period were significantly less dominant within the Democratic Party than conservatives were within the Republican Party during the 1972-1984 period. So, once again, there is simply no evidence that liberal Democrats are driving the polarizaiton process.
But let's take a closer look at how things are working within a specific, very important demographic-the White South. For this is where the once-monolithic Democratic domiance has shifted dramatically to the Republicans.
Here we see the liberal Democratic share declining modestly, from 10.1% to 9.4% (next to very sharp drops in moderata and conservative Democrats), while the conservative Republican share exactly doubles, from 11.8% to 23.6%:
It's hard to see this as anything other than a massive, racially-motivated shift to the Republicans among conservative voters, and from liberal or moderate to conservative, based on the same motivation. This shift was held back for a while, due to fancy footwork by the older generation of local Southern Democratic politicians, but holding back this tide of change became increasingly difficult over time.
Again, the within-party percentages tell a similar tale: the liberal Democratic position in the third period is significantly weaker than the conservative Republican position during the first period. The only difference is that the contrast is significantly stronger than at the nationwide level:
Indeed, conservative Republicans were more than a majority during the first period, and outnumbered liberal conservatives by over 3-1, rising to almost 8-1 in the third period, whereas liberal Democrats barely topped one third of the party makeup in the third period.
Outside the White South, we see a more balanced story, but still one in which Conservative Republicans grew significantly more than liberal Democrats did. Liberal Democrats grew 0.6 of the total population outside the White South--from 14.7% to 15.3%--while counservative Republicans grew 3.7% (more than 6 times more)--from 11.0% to 14.7%:
In this population, both parties lost membership in their other two categories, so the intra-party figures looked as follows:
Again, we see the same pattern: Liberal Democrats were less dominance within the Democratic Party in the third period than Conservatives were dominant during the first period. In the end, liberal Democrats dominated conservative Democrats by just over 2-1, compared to 3-1 by conservative Republicans over liberal Republicans in the first period. By the third period, conservative Republican dominance was just shy of 6-1.
In short, while the parties are closer to mirroring each other than they are in the white South, conservatives clearly dominate the GOP, while liberals hold a bare plurality among the Democrats. Only one party is clearly polarized on an ideological basis, and it is the Republicans.
This point is reinforced by the following two charts. The first, from Robert Altemeyer's online book, The Authoritarians, shows how legilators in various state legislatures scored (group averages) on the RWA (right-wing authoritarian) scale. The degree of divergence between parties is almost certainly under-estimated, Altemeyer explains, since only 1/4 responded overall, and responses diverged more strongly when response rates were higher. Ergo, a 100% response rate would have lead to greater divergence. This was also all done pre-1994. There's been more than a decade for polarization to increase still further. Nonetheless, what we see is instructive--(1) just one body in which Democrats are more authoritarian than Republicans, (2) a wide divergence in the level of RWA among Democratic legislative delegations, and (3) a surprisingly narrow range in the level of RWA among Republican legislative delegations. In short: polarization in just one party, not both:
Finally, from Hacker and Pierson's book Off Center: The Republican Revolution and the Erosion of American Democracy, there is this chart, showing the trajectory of party activists over time, Republicans beoming increasingly polarized, while Democrats turn back foward the center: