Voting patterns of America’s whites, from the masses to the elites

Within any education category, richer people vote more Republican. In contrast, the pattern of education and voting is nonlinear. High school graduates are more Republican than non-HS grads, but after that, the groups with more education tend to vote more Democratic. At the very highest education level tabulated in the survey, voters with post-graduate degrees lean toward the Democrats. Except for the rich post-graduates; they are split 50-50 between the parties.

What does this say about America’s elites? If you define elites as high-income non-Hispanic whites, the elites vote strongly Republican. If you define elites as college-educated high-income whites, they vote moderately Republican.

There is no plausible way based on these data in which elites can be considered a Democratic voting bloc. To create a group of strongly Democratic-leaning elite whites using these graphs, you would need to consider only postgraduates (no simple college grads included, even if they have achieved social and financial success), and you have to go down to the below-$75,000 level of family income, which hardly seems like the American elites to me.

The patterns are consistent for all three of the past presidential elections. (The differences in the higher-income low-education category should not be taken seriously, as the estimates are based on small samples, as can be seen from the large standard errors for those subgroups.)

Notes: These graphs show just (non-Hispanic) white voters because (a) most voters are white, (b) minorities tend to vote consistently for Democrats, so there’s less variation in their voting patterns, and (c) sample sizes are smaller for nonwhite groups so it’s hard to see clear patterns amid the noise. Data come from Annenberg pre-election polls for 2000 and 2004 and Pew pre-election polls for 2008; total number of non-Hispanic white respondents: 26161, 36476, and 15212. Graphs show intended vote in presidential election, including only those who expressed a preference for the Democratic or Republican candidate. Income categories are defined as family income less than $20K, $20-40K, $40-75K, $75-150K, 150K+. This work was done in collaboration with Yair Ghitza.

29 thoughts on “Voting patterns of America’s whites, from the masses to the elites

  1. Anyone know why in the “No High School” group the richest cohort turns so strongly non-Republican?

    A thing that tends to be lost on a graph like this is how much larger certain data-points are than others. e.g. What’s the voting power of a “No-HS-mid-income” group versus a “Graduate-degree-high-income” group.

    • “Anyone know why in the “No High School” group the richest cohort turns so strongly non-Republican?”

      As Andrew indicated in his post, we can’t really tell if it does or not because there are few people in this group and so the standard errors are massive.

  2. If you define elites as people holding advanced degrees (e.g., college professors), as many of the pundits on the right do, then they are democratic. In raw vote numbers they may be a small block (you don’t provide sizes for the 5 subgroups) but their degrees lend them more authority, hence they tend to have an outside voice in the political discourse.

    Of course the same could be said about the extremely wealthy (too small a group to show up on your charts) whose wealth also gives them a larger voice in the political discourse.

      • This is quickly turning into a blog about semantics rather than statistics. Not that there’s anything wrong with that, but the problem is that semantics is a social construct, not a scientific undertaking where we can discover what “elite” really means independent of people’s use of the term. (Unless, of course, you believe the philosophers who believe in Platonic “natural kinds”.)

        What you (Andrew) think of as elite probably doesn’t match the meaning of “elite” intended by the “pundits on the right”. I may be taking this too personally, but I think they’re talking about us (Columbia faculty like Andrew and Columbia “think tank” quasi-faculty like me). I don’t think the salary is really an issue. They probably don’t intend to include plumbers or construction workers or farmers or small business owners who make more money than I do. That is, it’s more socio- than economic.

        But I think no matter how we slice the term “elite”, it’s going to include you (Andrew). You write op-ed pieces for the New York Times and are a full professor at Columbia. How more “elite” in the sense intended by the “pundits on the right” could it get? I suppose you could have a column in the Times and hob-nob at Davos.

        • Bob:

          I agree that I’m a member of the elite. But I make more than $75,000, so your statement does not contradict anything that I wrote above. Nowhere did I exclude myself from the “elite” category.

          Or maybe I’m missing something in your comment? The data seem pretty clear to me: high-income people with graduate degrees split 50/50 and high-income people with college degrees lean Republican. I agree that if you slice the data finely enough (e.g., including only people who live in big cities), you can carve out a group of elite Democrats, but then that’s what you’re doing—you’re carving out a group, no longer looking at the country as a whole.

  3. Do we know anything about occupation, I have always seen these demographic characteristics, but occupation might muddle the issue here. I doubt it though of course.

      • I am happy to blame R as well as the author. Using different line patterns and/or point symbols can help the colo[u]r-challenged too, perhaps as the expense of some busy-ness in the graph design.

    • What’s wrong with red & green on a graph? I do this all the time, as they are easy to tell apart (though I prefer dark green). Of course, in most publications I stick to shades of grey.

      • Well, they are easy to tell apart for those who aren’t red-green color-blind. Considering that this is by far the most common form of colorblindness (~6% of all men), having red then green as the default is problematic.

        • That said, this particular red and green should be distinguishable even to the color-blind – the green is light enough and the red dark enough that even in greyscale they would probably be distinguishable shades, even if they looked the “same” color.

        • The fraction of men with red-green colorblindness is actually about 10% (“red-green” symptoms are caused by several types of sex-linked mutations.)

          Unfortunately, The term “red-green” does a bad job of describing which colors are problematic for the colorblind. As someone with a severe form of red-green colorblindness (protanopia), I actually had some trouble distinguishing the red from the black, but no trouble telling red from green. Making the lines thicker helps make the colors more easily distinguishable; at full magnification, all the colors in this plot were easy to tell apart.

  4. Since age and income are highly correlated, I wonder if what is actually driving the increase in Republican voting as income increases, age or income….or likely some interaction of the two?

    • I’ve always wondered the same thing. A few minutes with Google suggests roughly 1.8m primary and secondary school teachers have MA’s (out of 3.6m; 07-08 data:, which unhelpfully categorizes industry by combining ed and health.), while the Census says 16m (2011 data) have MA + prof. deg + doctoral. So fewer than I would have expected, but still a large enough group to make it important to remember when you interpret such things, esp. when you figure there are lots of teachers who A) are unemployed or B) temporarily drop out of the labor force for a few years to raise kids. If the ideological effects of public employment + unions still affects them (and I think we have good reason to expect it does), then something like 15% of the highest education category think like teachers.

  5. Pingback: Same old story « Statistical Modeling, Causal Inference, and Social Science

  6. I’m glad you practice what you preach when it comes to making conclusions beyond the scope of the data, though I haven’t read Red State/Blue State to know for sure. Your posts on this subject are typically confined to the data. You kindly leave it to us/pundits to sensationalize it.

    Two Questions: (1) Is income confounded with tax bracket effect? E.g. the more I earn and the more the government takes out of my paycheck the more I lean Republican/Fiscal Conservative.

    And (2) related to your “Same Old Story” post: how do you define working class? From the graphs above, I suspect that working class would contain a whole bunch of people making mid income with a HS diploma or some college. Taken together that looks like a group that would tend Republican.

  7. Pingback: Models, assumptions, and data summaries « Statistical Modeling, Causal Inference, and Social Science

  8. Pingback: The most educated states are all blue - Page 3 - US Message Board - Political Discussion Forum

Comments are closed.