For quite a while now (pre-dating Obama, but more frequently since he was elected), I’ve been reading about the Democrats’ troubles with “the white working class”. In some ways, this is unsurprising. In every country with which I’m familiar, a substantial proportion of the working class votes for the more conservative/rightwing party. And, even compared to the most wishy-washy of social democratic and labor parties elsewhere, the Dems aren’t exactly fervent champions of the worker. Still, the Repubs are even worse, so it seemed surprising to read that they regard the white working class as their base. Other things I read (sorry can’t find links now) made things even more puzzling. On the one hand, in the US as elsewhere, higher incomes are correlated with voting for the conservative/rightwing party, which seems to cut against the thesis. On the other hand, I’ve read that the average income of the US working class is the same as that of the population as a whole, which goes against the whole idea of “working class” as I understand it.
All became clear(or, at least, clearer) when I discovered that US political discussion uses two very different (though correlated) concepts of “working class”. The first is the more or less standard one – people who depend on wage labor (normally in manual or low-status service occupations) for their income. The second, specific to the US, and standard in most political polling, is “people without a 4-year college degree”, a class which includes such horny-handed sons and daughters of toil as Bill Gates and Paris Hilton. More prosaically, it includes lots of small business owners, and (since college graduation rates were rising until relative recently), over-represents the old.
Data on US voting patterns is surprisingly scarce, but Andrew Gelman has a big data set confirming the point that Republican voting rises with income. Andrew kindly sent me the data, which classifies voters by education (5 levels), income (5 categories) and race/ethnicity(4), for a total of 100 categories, and gives, for each group the proportion voting Republican. I’ve used this to look at an income-based definition of working class, encompassing everyone with an income less than $40 000. I’m not sure of the exact definition of this variable, but it seems pretty clear that people with income at this level are unlikely to be living on income from capital or a high-status job. To focus on the claim about the white working class, I’ve divided the 100 categories into four roughly equal-sized groups: working class whites (income less than 40K), middle/high income whites with and without college degrees, and all non-whites. Then I’ve looked at how many votes the Republicans got from each group in 2008.
As the pie chart below illustrates, the biggest group in the Republican voting base, and the group with which they do best is that of middle/high income whites without college degrees (the percentage after the group name gives the Republican share of the vote for that group). There’s nothing surprising in this, since all three variables are correlated with Republican voting. It’s the practice of calling this group “working class” that causes the confusion.
Disaggregating, the extreme case is that of high-school educated whites with incomes over $150K, 81.7 per cent of whom supported the Republicans in 2008. They’re a small group of course, but not negligible at about 1 per cent of the sample (155 out of 19170).
The two remaining groups of white voters are split pretty evenly between Reps and Dems, while, as is well known, non-white voters strongly favor the Dems.
The Republican voting base
(percentages after each group give proportion of that group voting R).
[click to continue…]