|
A little over a week ago, Chris wrote a diary, Concern Troll Demographers, taking off on an article at The Politico, Dems must woo white men to win . In response, Chris pointed out that that the "white men" being lost were actually a narrower demographic, "the white, straight, Christian, non-union male vote," which "forms about 25-30% of a shrinking percentage of the electorate." Which, of course, is quite true, as Chris has pointed out several times over the past few years. But for me, it brought to mind something else as well, the following three charts which I generated last year, which show that Democrats have basically stabilized their support outside the White South since the Bush I years, while only the Southern White conservative Republican demographic has continued to swell:



This got me to thinking about what it would look like to examine party ID by income, comparing the South to the rest of the country. Which I do on the flip...
|
| A Brief Reprise: Chris on Dems and White Men
Before plunging into the income-based numbers, I want to briefly reprise a few points Chris made:
- Democrats are actually losing the white, straight, Christian, non-union male vote. And yes, Democrats are getting thoroughly trounced in this demographic, and are lucky to score more than 30% of the vote here in any given national election. This group forms about 25-30% of a shrinking percentage of the electorate, as both self-identified Christians and self-identified whites are dropping precipitously as a percentage of the electorate. These drops easily cancel out the drop in union membership.
- If the Democratic electoral problem is actually among white, straight, Christian, non-union men (termed WSCNUM from now on), then doesn't the cause of this problem become pretty obvious? It is a combination of multi-faceted identity backlash and the collapse of union density in America.
He went on to conclude:
My point is all of this is two-fold. First, Democrats are not losing all types of white men, and are actually doing quite well among some groups. Second, the groups they are doing the worst among are also shrinking relative to the rest of the electorate.
With that in mind-as well as the charts above, I wanted to see what things look like if we look at income instead of race, and if we take the South out of the equation. This approach-leaving race out of the equation for now-makes sense in terms of geographical/electoral college logic, and also in terms of highlighting the importance of economics over a time period in which conservatives have done everything imaginable to try to take our eye off the ball.
The National Picture
The data I'm using here comes from the American National Election Studies (ANES) Cumulative Datafile. I use that instead of the General Social Survey (GSS), the source of the data for the charts I generated, because the ANES has a consistent breakdown of income groups by percentile, while the GSS does not.
Nationally, we can see that Democratic Party identification remains ahead of Republican Party identification, although by a smaller margin after the 1970s. Both party loyalty and turnout are lower among Democrats, which is the only reason, even today, that Republicans remain competitive nationwide:
Here's the breakdown by individual election-year surveys:
| Party ID (with leaners) By Year: Nationwide, 1952-2004 | Margins Among Income Groups By Percentiles | | Year | Dems | Ind | Reps | All | 0 to 16 | 17 to 33 | 34 to 67 | 68 to 95 | 96 to 100 | | 1952 | 57.2 | 8.5 | 34.3 | 22.8 | 28.3 | 27.0 | 27.9 | 20.4 | -31.3 | | 1954 | 55.8 | 11.1 | 33.1 | 22.7 | 33.0 | 25.5 | 24.4 | 24.8 | -31.5 | | 1956 | 50.5 | 12.6 | 37.0 | 13.5 | 15.4 | 21.0 | 16.7 | 13.0 | -22.4 | | 1958 | 56.3 | 10.9 | 32.8 | 23.5 | 16.7 | 33.7 | 30.5 | 19.5 | -31.6 | | 1960 | 51.5 | 12.4 | 36.1 | 15.4 | 9.1 | 23.4 | 20.4 | 17.5 | -44.0 | | 1962 | 53.1 | 11.4 | 35.5 | 17.6 | 13.1 | 10.8 | 26.2 | 15.0 | -11.3 | | 1964 | 61.1 | 8.7 | 30.2 | 30.9 | 41.1 | 42.0 | 40.8 | 17.0 | -3.8 | | 1966 | 55.2 | 13.0 | 31.7 | 23.5 | 25.5 | 38.8 | 22.2 | 24.1 | -4.2 | | 1968 | 55.0 | 12.1 | 32.9 | 22.0 | 38.6 | 25.4 | 21.1 | 17.3 | -11.4 | | 1970 | 54.4 | 13.3 | 32.3 | 22.1 | 36.0 | 27.6 | 24.6 | 15.5 | -16.7 | | 1972 | 51.7 | 14.1 | 34.2 | 17.4 | 27.3 | 22.4 | 22.5 | 7.7 | -17.4 | | 1974 | 52.2 | 17.5 | 30.3 | 21.9 | 28.3 | 37.1 | 27.6 | 8.4 | -36.6 | | 1976 | 51.8 | 15.3 | 32.9 | 19.0 | 34.5 | 32.6 | 24.4 | 8.8 | -44.9 | | 1978 | 54.4 | 15.5 | 30.1 | 24.3 | 35.2 | 34.1 | 26.6 | 15.0 | -0.7 | | 1980 | 52.2 | 14.6 | 33.3 | 18.9 | 35.6 | 36.5 | 17.5 | 7.2 | -20.8 | | 1982 | 55.9 | 12.1 | 32.0 | 23.9 | 48.1 | 35.8 | 27.5 | 11.2 | -23.2 | | 1984 | 48.2 | 12.1 | 39.7 | 8.4 | 29.9 | 24.6 | 7.8 | -5.3 | -33.3 | | 1986 | 50.3 | 13.4 | 36.3 | 14.0 | 35.8 | 21.5 | 15.2 | 0.0 | -19.8 | | 1988 | 47.0 | 11.7 | 41.3 | 5.7 | 25.9 | 16.9 | 5.0 | -6.0 | -57.4 | | 1990 | 51.8 | 11.6 | 36.6 | 15.2 | 29.3 | 29.5 | 13.1 | 11.1 | -50.6 | | 1992 | 49.9 | 12.3 | 37.8 | 12.1 | 27.9 | 27.4 | 17.3 | -3.2 | -30.5 | | 1994 | 46.9 | 10.8 | 42.4 | 4.5 | 38.4 | 29.1 | 0.9 | -16.9 | -37.7 | | 1996 | 51.9 | 10.0 | 38.1 | 13.8 | 36.9 | 30.2 | 17.4 | -12.0 | -12.5 | | 1998 | 51.0 | 12.5 | 36.5 | 14.5 | 23.1 | 31.9 | 17.6 | 3.5 | -18.5 | | 2000 | 50.4 | 13.2 | 36.4 | 14.1 | 39.4 | 19.0 | 7.3 | 10.7 | -18.3 | | 2004 | 50.1 | 8.9 | 41.0 | 9.1 | 25.0 | 18.8 | 7.5 | -0.4 | -5.0 | | TOT | 52.3 | 12.6 | 35.1 | 17.2 | 29.6 | 28.1 | 19.4 | 8.4 | -23.5 |
And here's the breakdown by cumulative decade totals:
| Party ID (with leaners) By Decade: Nationwide, 1952-2004 | Margins Among Income Groups By Percentiles | | Decade | Dems | Ind | Reps | All | 0 to 16 | 17 to 33 | 34 to 67 | 68 to 95 | 96 to 100 | | 1950s | 54.9 | 10.8 | 34.3 | 20.6 | 23.4 | 26.8 | 24.9 | 19.4 | -29.2 | | 1960s | 55.2 | 11.5 | 33.3 | 21.9 | 25.5 | 28.1 | 26.1 | 18.2 | -15.0 | | 1970s | 52.9 | 15.1 | 32.0 | 20.9 | 32.3 | 30.8 | 25.1 | 11.1 | -23.3 | | 1980s | 50.7 | 12.8 | 36.5 | 14.2 | 35.1 | 27.0 | 14.6 | 1.4 | -30.9 | | 1990s | 50.3 | 11.4 | 38.3 | 12.0 | 31.1 | 29.6 | 13.3 | -3.5 | -30.0 | | 2000s | 50.3 | 11.0 | 38.7 | 11.6 | 32.2 | 18.9 | 7.4 | 5.1 | -11.7 | | TOT | 52.3 | 12.6 | 35.1 | 17.2 | 29.6 | 28.1 | 19.4 | 8.4 | -23.5 |
Because of the relatively small sample size (just 201 for this decade), even the decade totals for the top 96-100 percentile are rather noisy. Thus, the sharp drop in the GOP margin among this group from the 1980s and 1990s to the 2000s could well be significantly overstated. Furthermore, not all respondents give their incomes, and this can skew the data somewhat. Still, the big picture trends are fairly clear, which enhances our confidence in them.
The most useful distinctions, in my judgment, come from looking at what happens with the combined lower income 0-16 and 17-33 percentile groups and comparing them with the middle and upper 34-67 and 68-95 percentile groups. The Democratic margin among the former groups actually rises by over 10 percent from the 1950s to the 1970s, where it remains fairly stable for three decades until dipping down 10 points this decade. But the margin among the later groups is stable from the 50s to the 60s, then drops 20% in the 70s, and more than 50% more in the 1980s, and another 40% in the 1990s, before bouncing back slightly in the 2000s. In short, the Democrats loss of partisan edge comes almost entirely from the upper and middle incomes.
But what happens if we look at things regionally, comparing the South with the Non-South?
The Southern Picture
The Southern picture, as one might expect, is one of dramatic decline in Democratic identification. Here's the breakdown by individual election-year surveys:
| Party ID (with leaners) By Year: South, 1952-2004 | Margins Among Income Groups By Percentiles (Data Is Noisy Due To Small Sample Sizes) | | Year | Dems | Ind | Reps | All | 0 to 16 | 17 to 33 | 34 to 67 | 68 to 95 | 96 to 100 | | 1952 | 71.8 | 9.6 | 18.6 | 53.2 | 44.1 | 45.1 | 64.9 | 63.0 | 42.9 | | 1956 | 64.1 | 13.7 | 22.2 | 41.9 | 31.5 | 40.0 | 54.0 | 42.0 | 37.9 | | 1958 | 65.8 | 13.9 | 20.3 | 45.6 | 16.8 | 53.3 | 62.8 | 46.0 | 63.6 | | 1960 | 61.3 | 15.2 | 23.5 | 37.8 | 21.9 | 49.2 | 43.8 | 47.3 | -30.4 | | 1962 | 57.1 | 15.3 | 27.6 | 29.5 | 33.6 | 0.0 | 37.8 | 26.6 | 5.6 | | 1964 | 71.4 | 9.0 | 19.7 | 51.7 | 65.5 | 54.8 | 60.3 | 25.3 | 34.3 | | 1966 | 61.1 | 12.8 | 26.1 | 35.0 | 33.0 | 42.2 | 40.7 | 36.1 | -4.5 | | 1968 | 63.5 | 13.4 | 23.1 | 40.4 | 53.6 | 33.0 | 40.5 | 41.0 | 14.3 | | 1970 | 59.1 | 15.4 | 25.4 | 33.7 | 54.0 | 27.2 | 38.9 | 17.4 | 16.7 | | 1972 | 57.6 | 14.6 | 27.8 | 29.7 | 36.8 | 35.0 | 33.0 | 13.8 | 6.9 | | 1974 | 56.3 | 20.2 | 23.6 | 32.7 | 33.9 | 51.1 | 33.7 | 18.2 | 0.0 | | 1976 | 60.8 | 14.3 | 24.9 | 35.8 | 44.4 | 50.5 | 32.8 | 22.0 | -4.0 | | 1978 | 59.7 | 14.4 | 25.9 | 33.8 | 40.3 | 34.9 | 37.7 | 40.2 | -21.8 | | 1980 | 58.4 | 12.6 | 29.1 | 29.3 | 39.8 | 46.4 | 33.1 | 16.0 | -38.5 | | 1982 | 61.5 | 9.7 | 28.8 | 32.7 | 51.6 | 37.0 | 38.9 | 18.4 | -10.0 | | 1984 | 53.3 | 15.2 | 31.5 | 21.8 | 38.0 | 37.3 | 17.2 | 6.2 | -23.1 | | 1986 | 56.4 | 13.7 | 29.9 | 26.5 | 40.7 | 35.5 | 22.8 | 17.2 | -18.2 | | 1988 | 55.3 | 10.4 | 34.3 | 21.0 | 35.1 | 31.6 | 16.5 | 7.0 | -16.7 | | 1990 | 55.9 | 15.3 | 28.8 | 27.1 | 37.3 | 42.7 | 25.5 | 7.3 | -40.0 | | 1992 | 50.7 | 15.2 | 34.1 | 16.6 | 29.4 | 29.4 | 23.7 | -8.7 | -21.9 | | 1994 | 48.2 | 10.5 | 41.3 | 6.9 | 32.6 | 33.3 | 0.0 | -13.4 | -57.7 | | 1996 | 52.0 | 10.0 | 38.0 | 14.0 | 33.8 | 41.4 | 13.3 | -16.4 | -51.9 | | 1998 | 50.3 | 12.4 | 37.1 | 13.2 | 18.3 | 33.7 | 10.3 | 9.3 | -29.7 | | 2000 | 47.1 | 15.8 | 37.1 | 10.1 | 42.1 | 15.9 | -3.4 | 0.9 | -29.6 | | 2004 | 50.7 | 8.0 | 41.5 | 9.2 | 30.0 | 24.0 | 6.3 | -7.4 | -31.3 | | TOT | 57.6 | 13.6 | 28.8 | 28.9 | 37.4 | 38.3 | 29.8 | 18.5 | -12.3 |
And here's the breakdown by cumulative decade totals:
| Party ID (with leaners) By Decade: South, 1952-2004 | Margins Among Income Groups By Percentiles | | Decade | Dems | Ind | Reps | All | 0 to 16 | 17 to 33 | 34 to 67 | 68 to 95 | 96 to 100 | | 1950s | 67.2 | 12.4 | 20.4 | 46.9 | 30.8 | 46.1 | 60.5 | 50.3 | 48.1 | | 1960s | 62.9 | 13.1 | 24.0 | 38.9 | 41.5 | 35.8 | 44.6 | 35.2 | 3.8 | | 1970s | 58.7 | 15.8 | 25.5 | 33.1 | 41.9 | 39.7 | 35.2 | 22.3 | -0.5 | | 1980s | 57.0 | 12.3 | 30.7 | 26.3 | 41.0 | 37.6 | 25.7 | 13.0 | -21.3 | | 1990s | 51.4 | 12.7 | 35.9 | 15.6 | 30.3 | 36.1 | 14.6 | -4.4 | -40.2 | | 2000s | 48.9 | 11.9 | 39.3 | 9.6 | 36.0 | 20.0 | 1.5 | -3.2 | -30.4 | | TOT | 57.6 | 13.6 | 28.8 | 28.9 | 37.4 | 38.3 | 29.8 | 18.5 | -12.3 |
In the 1950s, the higher income groups give Democrats the highest margins-over 50 percent overall. But by this decade, that advantage is completely wiped out. In contrast, the margins among the lower income groups has shrunk somewhat, but is still substantial. Of course race plays a huge role in keeping those margins up among the lower income groups. But what about the rest of the country?
The Non-Southern Picture
The Non-Southern pictures is strikingly different. Here's the breakdown by individual election-year surveys:
| Party ID (with leaners) By Year: Non-South, 1952-2004 | Margins Among Income Groups By Percentiles | | Year | Dems | Ind | Reps | All | 0 to 16 | 17 to 33 | 34 to 67 | 68 to 95 | 96 to 100 | | 1952 | 51.5 | 8.0 | 40.5 | 11.0 | 11.7 | 19.3 | 16.4 | 10.1 | -40.4 | | 1956 | 44.8 | 12.1 | 43.1 | 1.8 | 1.4 | 12.1 | 3.3 | 5.3 | -39.0 | | 1958 | 52.3 | 9.6 | 38.1 | 14.1 | 16.7 | 24.1 | 18.9 | 10.5 | -54.3 | | 1960 | 46.5 | 11.0 | 42.5 | 4.0 | -6.1 | 9.3 | 10.6 | 6.0 | -49.2 | | 1962 | 51.1 | 9.3 | 39.6 | 11.5 | -1.9 | 17.5 | 21.1 | 9.4 | -18.2 | | 1964 | 56.5 | 8.5 | 34.9 | 21.6 | 23.8 | 34.1 | 32.5 | 14.9 | -22.9 | | 1966 | 52.8 | 13.1 | 34.1 | 18.7 | 20.1 | 36.6 | 14.6 | 21.4 | -4.1 | | 1968 | 51.2 | 11.5 | 37.2 | 14.0 | 28.1 | 21.5 | 13.7 | 9.3 | -20.7 | | 1970 | 51.9 | 12.2 | 35.9 | 16.1 | 20.2 | 27.9 | 16.5 | 14.8 | -26.2 | | 1972 | 48.7 | 13.8 | 37.5 | 11.2 | 19.3 | 14.4 | 17.0 | 5.8 | -24.3 | | 1974 | 49.9 | 16.0 | 34.1 | 15.8 | 22.2 | 28.7 | 24.2 | 4.4 | -46.6 | | 1976 | 47.5 | 15.8 | 36.7 | 10.9 | 23.9 | 22.5 | 20.2 | 4.8 | -53.5 | | 1978 | 51.4 | 16.1 | 32.5 | 18.9 | 30.6 | 33.6 | 21.0 | 3.8 | 12.1 | | 1980 | 49.0 | 15.6 | 35.4 | 13.6 | 32.6 | 30.4 | 10.0 | 3.5 | -11.8 | | 1982 | 52.6 | 13.5 | 33.9 | 18.8 | 45.2 | 35.0 | 21.7 | 7.9 | -29.2 | | 1984 | 45.6 | 10.6 | 43.8 | 1.8 | 22.6 | 17.4 | 3.2 | -9.1 | -36.5 | | 1986 | 46.8 | 13.2 | 40.0 | 6.9 | 30.1 | 12.0 | 11.3 | -6.4 | -20.8 | | 1988 | 42.3 | 12.4 | 45.3 | -3.0 | 17.1 | 6.9 | -1.4 | -11.1 | -63.4 | | 1990 | 49.6 | 9.5 | 40.9 | 8.7 | 21.2 | 18.6 | 6.4 | 12.4 | -52.2 | | 1992 | 49.4 | 10.8 | 39.8 | 9.7 | 25.9 | 25.9 | 14.7 | -1.2 | -33.7 | | 1994 | 46.1 | 10.9 | 43.0 | 3.1 | 44.9 | 27.9 | 1.0 | -18.9 | -25.6 | | 1996 | 51.9 | 9.9 | 38.1 | 13.8 | 40.3 | 20.3 | 19.4 | -9.4 | 8.9 | | 1998 | 51.3 | 12.5 | 36.1 | 15.2 | 26.3 | 30.6 | 21.7 | 0.5 | -12.7 | | 2000 | 52.2 | 11.7 | 36.1 | 16.1 | 36.5 | 20.7 | 12.7 | 14.7 | -12.7 | | 2004 | 49.8 | 9.5 | 40.7 | 9.1 | 19.6 | 15.4 | 8.0 | 1.7 | 4.6 | | TOT | 49.4 | 12.1 | 38.5 | 11.0 | 22.2 | 22.0 | 14.3 | 4.2 | -27.2 |
And here's the breakdown by cumulative decade totals:
| Party ID (with leaners) By Decade: Non-South, 1952-2004 | Margins Among Income Groups By Percentiles | | Decade | Dems | Ind | Reps | All | 0 to 16 | 17 to 33 | 34 to 67 | 68 to 95 | 96 to 100 | | 1950s | 49.5 | 9.9 | 40.6 | 9.0 | 9.9 | 18.5 | 12.9 | 8.6 | -44.6 | | 1960s | 51.6 | 10.7 | 37.7 | 14.0 | 12.8 | 23.8 | 18.5 | 12.2 | -23.0 | | 1970s | 49.9 | 14.8 | 35.3 | 14.6 | 23.2 | 25.4 | 19.8 | 6.7 | -27.7 | | 1980s | 47.3 | 13.1 | 39.7 | 7.6 | 29.6 | 20.3 | 9.0 | -3.0 | -32.3 | | 1990s | 49.7 | 10.7 | 39.6 | 10.1 | 31.7 | 24.7 | 12.7 | -3.3 | -23.1 | | 2000s | 51.0 | 10.6 | 38.4 | 12.6 | 28.0 | 18.1 | 10.4 | 8.2 | -4.1 | | TOT | 49.4 | 12.1 | 38.5 | 11.0 | 22.2 | 22.0 | 14.3 | 4.2 | -27.2 |
Democratic margins in the lowest group are up dramatically since the 1950s-attributable in large part to an increase in minority population percentage-while the others are virtually unchanged, except for the very top, which is probably partly an outlier, with a very small sample size.
In short, the Democrats' loss of partisan ID seems to be heavily concentrated in the South, among middle and upper-income voters-which is another way of saying the White South. This is not to say that they haven't lost some white voters elsewhere throughout the nation. But those losses have been offset by minority gains at virtually every income level.
This may just be a long-winded way of saying, "Chris was right." But given the unrelenting pervasiveness of the "Dems need to appeal to white men" narrative, it's helpful to have the same point made using a variety of different metholodologies. |