On Tue, 21 Oct 2008, in response to a blurb from Lane Kenworthy et al (summarized by George Packer, dug out by Sam Smith) about Why the Dems lost the White Working Class, Doug Henwood replied by quoting Larry Bartels on the subject:
> <http://www.princeton.edu/~bartels/kansas.pdf>
>
> What's the Matter with
> What's the Matter with Kansas?
>
> Larry M. Bartels
The gist of which is it that we can't talk about the White Working class switching allegiances over the last 40 years -- that what really happened is that the *Southern* White Working class (as well as its professional classes) changed from Democratic to Republican, but the Northern White Working class vote the same now as they did 40 years ago, and may even be a little more Dem than they used to be.
But this was addressed directly in a post from Lane Kenworthy's blog that was posted to LBO-talk by Shane Taylor:
http://lanekenworthy.net/2007/12/17/how-the-democrats-lost-their-class/
and the gist of the answer is: Bartels confined himself to presidential voting patterns, which have problems. Whereas Kenworthy and crew looked at which party self-identification, which is a different dataset, which gave this different result, and which they think is better for the given reasons. Below are the key grafs. There's a chart in the original:
<quote>
Yet the notion that the defection of whites, especially working-class
whites, from the Democrats has been largely confined to the south
paints too simple a portrait. Since 1972 the General Social Survey has
asked American adults about their "party identification," along with a
battery of other questions. People's political preferences ultimately
matter to the extent they influence actual voting choices. But
analyzing presidential voting alone, as Bartels does, can miss part of
the story. Presidential voting is heavily influenced by the particular
candidates the two parties nominate. Arguably, people's underlying
preferences and beliefs are better understood by looking at their party
identification. The following chart shows the share of working-class
whites identifying as strong-Democrat, moderately-strong-Democrat, or
independent-leaning-Democrat since the early 1970s.
[thedemocrats-figure1-test2.png]
The country is split here into three regions: the south, the midwest
and plains states, and the east and west coasts. In the south,
identification with the Democrats fell roughly 20 percentage points --
from 60% to 40% -- between the mid-1970s and the early 1990s and has
held steady since then. This is consistent with the picture offered by
Bartels and Krugman.
Yet the same thing happened in the other two regions -- even on the
coasts, where the most solidly "blue" states are located.
With several graduate students at the University of Arizona, I have
been examining this development ("The Democrats and Working-Class
Whites"). It turns out not to be a function of our measure of party
identification or of the working class. Nor is it specific to men or to
the most religious. And most of those who left the Democrats didn't
become "independents." In fact, since the early 1990s approximately 40%
of working-class whites have identified as Republican -- the same as
the share that identifies as Democrat.
What caused this development? We conclude that it was due in large part
to the crisis of the late 1970s....
<end excerpt>
Michael