The U.S. Electorate Since 1982, In 12 Graphs
To: Our readers
From: David FreddosoThis week:
- The 2014 election and the long-term trends
- Whither the Hispanic vote?
- Catholics drop out of the Democratic coalition.
Amid the tangled lines above, we see the trends of America’s voters over the last 32 years.
The chart measures something very specific: How much larger (or smaller) a Democratic margin of victory has grown in each of several different voting groups for U.S. House in the last 17 elections.
For example: If Democrats won women by three points in an election Republicans won by six points, women were nine points more Democratic than the average voter in that election. And if Republicans won white women by four points while winning the election by six, then white women are two points more Democratic than the average.
Look at this trend over time, and you can see what various groups are doing while controlling for some other factors.
Before breaking out some of the key groups in their own charts, let’s discuss the rationale behind studying the electorate this way in the first place.
Why House elections?
They take place once every two years, and thus provide the most consistent thermometer for the national political climate in every single election. To identify the median voter for U.S. House is to learn a lot about where the national mood is. And this thermometer is not overly affected by just one super-popular or super-unpopular candidate, the way presidential elections are, because it’s spread out among 435 individual races.
Why look at the deviation from the average voter?
If you just look at margins from past elections, you’re going to lose a lot of insights among the noise — the natural ups and downs that national political circumstances produce. You just can’t get an apples-to-apples comparison when you hold up a landslide election next to a squeaker unless you adjust for those circumstances, and this method makes that adjustment for you.
If you’re trying to study one group of voters or another, the proper way to do it is to study how much more Republican are men or whites or Evangelical Christians than the average voter. Electoral performance is always relative, and this method de-emphasizes the changes of performance in any one particular election. It might be useful to know that Latino voters lean Democratic or that senior citizens lean more Republican, but it is more useful to understand how much more Democratic or Republican they are than the average voter, and whether this habit has changed over time. If it changes significantly over time, it signifies a trend that transcends the natural ups and downs suffered by the respective parties in specific election years.
Why go back to 1982?
We were inspired to try this when we saw that The New York Times had gathered all the data from that year forward in a single chart after the 2010 election. We added the exit polling data from the next two elections worth of data, did some relatively simple spreadsheet operations, to produce the graphs you see here.
In short, this is a helpful way of studying the trends in habits among different voting groups. As you will see, it can provide a few really neat revelations. But a few caveats also apply.
The charts you are about to see require you to bring some prior knowledge with you if you are to interpret them correctly. For example, they are silent as to which groups are gaining larger shares of the electorate (over 60, Hispanic, family income over $50,000) and which ones are losing ground (whites, union households, family income under $30,000), so you have to bring prior knowledge with you to understand which trends are more important and which ones less so.
These charts require you to understand in advance that different age groups and income groups are composed of different people as time passes — for example, one reason senior citizens have become more Republican recently is the dying out of voters who came of age in the FDR era. And three decades of inflation and increases to the standard of living have transformed the “family income of $30,000 to $50,000” group from a middle-class group into a low-income group, which might change its voting habits without necessarily signifying anything politically significant.
With all this in mind, let’s look at a few important trends:
The Democratic Coalition:
These are not its only components, but these are probably most of the important ones:
These are all groups that have traditionally or in several consecutive elections voted more Democratic than the average voter. You’ll notice that two of them have dropped below zero, suggesting they are no longer part of the Democrats’ coalition but are instead pulling the national vote in the opposite direction.
One of these (in yellow) is those with only a high school education — about one-fifth of the electorate, and a group that tends to drop off a bit in non-presidential years. This block was unusually Republican in 2014, but it’s just one election. It’s not clear whether this represents a trend, although it could.
But then there’s the pink line — Catholics, once the essential Democratic bloc, which as late as the 1980s could once be counted on to vote more than ten points more Democratic than the average voter. Catholics, who make up nearly a quarter of the electorate, may have left the Democratic fold as part of a long-term trend. The Catholic voter has now been more Republican than the average voter for three elections in a row — possibly for the first time in American history. That’s a huge and largely unremarked development.
This is more significant than even meets the eye, because the fastest-growing segment among Catholics is a more Democratic-leaning group — Hispanics. White Catholics voted about the same as Protestants of all races in 2014 (60 percent Republican). But even Hispanics have been trending just ever-so-slightly Republican over time (more on this later). A dramatic shift in white Catholic opinion, combined with a glacial shift by Hispanics toward the GOP, is countervailing the change in the composition of the universe of Catholic voters.
The Republican Coalition:
This is a coalition mostly of larger groups that give smaller margins. But not all of them — white Evangelical Christians, for example, made up 26 percent of the electorate in both 2012 and 2014.
With the departure of Catholics from the Democratic coalition, the trend among religiously-identifying voters seems rather one-sided.
Jews remain Democratic, but they are also a very small component of the voting population, even in the few states where they have their largest presence. Democrats have mostly made up this ground among the religiously unaffiliated, whose share of the vote has grown.
Note also that Jews reached their modern Democratic peak in 2006 — very interesting because the midterm voter of 2006 was already quite Democratic to begin with, and the moved further to his left. Jewish voters remain a very Democratic constituency and well short of their most Republican point in recent history (1988), but they have moved in only one direction since 2006 — toward the GOP.
The Hispanic Vote:
Yes, here is the group everyone is talking about — the fastest growing segment in American politics. There are several things to say about the Hispanic vote, but here’s one you won’t hear much about in your typical election commentary.
It has only happened recently, but the gender gap has suddenly intruded among Hispanics. For the last few elections, there hasn’t been just one Hispanic vote — there have been two distinct Hispanic votes. For three elections in a row now, Hispanic men have voted just to the north or south of 20 points more Republican than Hispanic women:
What you see in this chart does not support the conventional wisdom about Hispanic voters at all — neither the Democrats’ wishful thinking nor Republicans’ expressions of hopelessness. Democrats’ built-in advantage among Hispanic men has been cut nearly in half since Reagan’s day, and that’s amid huge growth in the Hispanic voting population. In the last two elections, Hispanic men could only be counted on to deliver Democrats a margin in House races 22 points greater than the electorate as a whole. (In 2014, this meant they went 57-41 percent Democratic when the overall electorate polled at 53-47 Republican.) And it appears that the voters who propelled Republicans to their 2014 victories were, on the whole, a bit browner than the ones who made 2010 possible.
Hispanic women also shifted toward Republicans in 2014, but it wasn’t part of a clear trend in that direction. They had voted more or less the same as Hispanic men before 2010, at which point they went in a dramatically more Democratic direction. The precise reasons are open to interpretation, but there is no obvious answer. Recall that immigration was not a big partisan issue in that election. It has to be remembered also that the sample sizes with subgroups like these are small enough that the final numbers for any individual election year can be subject to greater error than for larger voting populations.
Also worth noting — the overwhelming and disproportionate rejection by Latino voters of Mitt Romney in 2012 does not seem to have translated to a dramatic shift in down-ticket races among either Hispanic men or women. The median Hispanic male voter actually grew closer to the median voter (thus more Republican) that year in his choice for U.S. House.
You hear so much about the gender gap, including the above, but it’s insignificant in understanding American politics. The enormous and widening gap in racial voting patterns is far more determinative of outcomes:
White women, like white men, have been trending more Republican since the mid-1990s — so if you want to talk about so-called “Soccer Moms” (see also next section), Democrats have been losing that war pretty consistently, even if perhaps a few Democratic candidates have been winning them.
Black women, typically the most Democratic of all voting blocs, have only become more so, as have black men. And we discussed Hispanic voters of both sexes above. Whites and blacks vote pretty consistently along a trend line. Hispanic men and women seem a bit more volatile in their voting habits.
America’s politics are determined much more by race than by gender or economic class — even if that is not a good thing.
A few myths can be exploded here, at least on a national level. Based on specific results and the Blue-ing of certain key counties, much of the commentariat has discussed the Blue-ing of the suburbs. But it is perhaps not as significant or widespread as believed.
The suburban vote has grown dramatically since the early 1980s. For the first time in 2014, suburbanites comprised a clear majority of voters (52 percent). But they have also remained surprisingly steady in preferring Republicans by somewhere around 6 points more than the rest of the electorate, with variations here and there. Democratic gains in the suburbs in the late Clinton years were real, but small and temporary. Also, although it is fashionable now to discuss the effect of demographic change in the suburbs, they have not really shifted toward Democrats in recent years. This is in spite of specific suburban areas (such as Northern Virginia and Philadelphia’s collar counties) where this clearly has happened.
Rural America has shrunk as a share of the voting population — from 23 percent of the electorate in 1994 to 13 percent in 2014 — but it has become much more Republican-leaning in the meantime. The last year when rural voters were more Democratic than average was 1988. They didn’t shift much in 2008 — President Obama’s first election — but they have been
The biggest change in 2014 — although it’s not necessarily a long-term trend — came in big cities (greater than 500,000 population), where Republicans lost by substantially smaller absolute margins than in 2012 or 2010 and also by a smaller margin relative to the voting population.
We noted above the Republican shift among the high school-educated. The rest of the picture looks like this:
Non-high-school graduate adults comprise such a small percentage of the voting population (2 percent of it in 2014) that the polling is pretty much useless. The only and significant long-term trend here seems to be the Democratic shift among holders of postgraduate degrees, who have now given Democrats margins 10 points or more above the average for the last three elections in a row. The size of this group has increased amidst the current economic anxiety, with college graduates seeking further refuge from the real world. But bear in mind that this group includes not just gender studies PhDs and the rest of liberal academia, but also lawyers, doctors, CPAs, MBAs, a lot of teachers, and everyone else with an advanced or professional degree.
Interestingly, 2014 marks the first year since 1996 in which more voters said they held college degrees (but not post-graduate degrees) than said they had “some college.” Both groups gave Republicans a ten-point margin, placing both about four points to the right of the average voter.
The income groups are, of course, affected by inflation over time, so this metric might be less useful in studying them over such a long period. If the economy grew consistently and everyone voted according to income or class, then every group would trend Democratic over time, with the higher-income groups growing over time as a share of the voting public while the lower-income groups shrink.
It is worth noting that that the share of voters from union households, a steadily Democratic group, has declined rather dramatically since 2004, from 24 percent to 17 percent of the electorate, which matches the situation of declining unionism in real life.
Democrats’ great hope of demographic change rests on two major trends: The first is that today’s young voters are more Democratic than their elders. The trend in this case is real and relatively new, but not all in one direction:
Young voters have historically been a bit more Democratic than the average voter, but clearly something has been happening since 2000 that has made them much more Democratic. This trend predates Obama, and has actually begun to reverse itself on his watch.
It appears that yesterday’s youngsters — those whose first vote might have been cast for Obama in 2008 — have started moving up into the next age group, causing it to trend more Democratic, even as the youngest group heads back toward the mean. If the progressive pig in the python is indeed moving up to the next age group, that doesn’t quite fit the Democrats’ story about their inevitable rise.
Looking at age by race, there is only one age/racial group with statistically reliable numbers that go back to 1982 whose behavior deviates significantly from a simple racial trend. That is, of course, younger whites. They were at their most Democratic in 2010, but shifted away strongly in the year of Obama’s re-election. They remain more Republican than the mean, but they have trended slightly more Democratic since the turn of the century, defying the trend of every other white age group.
Older whites once resembled the median voter. But the death of the FDR generation has helped accelerate a strong Republican trend among whites that is bringing the entire over-60 age group toward the GOP. Older whites were relatively susceptible to Social Security scare tactics as recently as 2000. Those days seem like a distant memory now. In 2014, older whites became the most Republican age group among whites for the first time since the Lewinsky drama of 1998, and unlike that occasion, it seems to be part of a long-term trend. The age group immediately below them — the baby boomers — seem to be just as Republican for now, on a similar trend-line.
The Bottom Line:
Don’t buy it if someone tells you there is one trend that will determine the future of American politics. There are many trends at any given moment, many of them countervailing one another. Nor are there any hard and fast rules. The parties, for their part, will adjust their messages — and perhaps even their principles, although that’s much rarer — in order to stay in the game.