One of the many topics that Pew Research releases polls about is the subject of political party identification (or “party ID”). Every even-numbered year before a national election, you can count on them surveying on that. Party ID is a little different than just counting up the number of people registered with each party in each state, which is also an important measure if you’re looking at a specific state. But “party ID” is more helpful than that. For one thing, a number of states don’t even have registration by party, so you can never pin down a truly national number.
For another thing, it also lets us know how independents really line up (and for that matter, Ds and Rs line up). Pew’s party ID survey asks people whether they identify as Democratic, Republican, or independent, which potentially skirts around a problem that we saw in the special election in Pennsylvania’s 18th district: the one of “ancestrally Democratic” (or “ancestrally Republican”) areas where people tend to cling to their registration for the once-dominant party (in PA-18’s case, the Democrats) while usually voting for the other party in general elections.
Pew also asks independents whether they lean to one party or the other, which is important because most independents aren’t truly independent; in some cases, they’re better partisans than party members (in other words, they choose to be “independent” because their preferred party is too impure or too feckless, but they’d certainly never vote for the other party; their choice is voting for one party or staying home). The pool of “swing voters,” who’ll freely pick and choose between both parties based on candidates’ individual merits, is actually fairly small.
The main story from Pew’s 2018 survey—and, as you can probably guess, it is very good news—is that Democratic party identification, if you include leaners, is rapidly increasing. The topline tells us that independents are the group that’s gaining the most, at the Republicans’ expense. The nation’s current breakdown is 37 percent independent, 33 percent Democratic, and 26 percent Republican, which sees I’s up and Rs down from when the series started in 1994 (when it was 33D, 33R, and 30I). But when you push leaners, there’s a sizable Democratic edge: 50 Democratic, 42 Republican (compared with 46 R, 44 D in 1994). That’s a large gap by historical standards, though not quite as large as what we saw in 2008 (when it was 51 D, 39 R), when the Democrats won a pretty decisive victory. The gaps are even larger when you start prying the numbers apart by different demographic categories, which we’ll explore further.
In retrospect, Pew’s 2016 survey on party ID should have been a wake-up call for a lot of people. It wasn’t, because almost all pollsters (including Pew themselves) were seeing large leads for Democrats in the national vote for the presidential race and people were feeling content about that. It turns out, of course, that pollsters weren’t really measuring the right thing: they weren't contacting enough non-college white voters and weren’t correctly weighting the ones they did contact, which screwed things up in state-level polls in states where non-college whites make up a disproportionately large part of the electorate. Pew’s 2016 party ID found Democratic identification falling off nationally (down to 48 D-44 R, way off from the heady days of 2008)—which, by itself, isn’t that unusual, as it happens to any party in power regardless of how well they perform in office (it’s what we call “thermostatic public opinion”)—but with the vast bulk of that erosion happening among white males, white senior citizens, and whites without college degrees. Knowing that, and weighting for that, could have saved a lot of pollsters a lot of grief in 2016.
So, today, let’s look at the flip side of who moved away from the Democrats in 2016, and who is currently leading the flight away from the Republicans in 2018: women, young people, and people with college degrees. (Race, interestingly, isn't as much of a factor, simply because non-white voters didn’t move at all toward the Republicans in 2016, and remain maxed out enough that there isn’t the same level of even-more movement to the Democrats now. There’s one sort-of exception for Asians, who were close to a swingy demographic in the 1990s but are now voting Democratic at nearly the same rate as Latinos, breaking 65 D and 27 R in Pew’s 2018 study.)
The gender gap, for instance, is larger than ever, with women identifying 56 percent Democratic and 37 percent Republican, a 19-point gap (though it’s on a similar level with where we were in 2008). That’s a significant improvement from 1994, when the Dems had only a 6-point lead among women. Note, also, some significant improvement among men, not just from 1994 (pretty much the Dems’ low water mark in every category) but even from 2016.
And, again, the “leaner” numbers show the importance of not just looking at independents but asking how they break. If you just looked at the raw data, you’d assume that things were flat for the Democrats among women; it was 37 D in 1994 and 39 D now, though with independents moving ahead of the GOP recently. But by pushing independents who lean, you can see that “independent” women actually break for the Democrats by an almost 2-to-1 margin.
One of the major story lines of the 2016 election was the sudden and looming presence of an education gap, one that was nearly as significant as the long-understood gender gap. You could certainly see it when you looked at county-level results, where the Democrats collapsed in longtime strongholds in blue-collar parts of the upper Midwest while picking up iconic, historically Republican suburban counties across the Sun Belt.
Pew’s 2018 numbers show even further acceleration of that trend, especially when you narrow the view down to just white voters. Non-white voters vote heavily Democratic enough that you don’t see a lot of difference regardless of education level. Also, looking at, for instance, the “high school or less” numbers across all races wouldn't really tell you that there’s a huge problem for Democrats among white voters there; there are enough black and Latino voters in the “high school or less” pool that, overall, the Democrats nearly draw even, at 47 R, 45 D.
When you zoom in on just white voters, though, the education gap looks huge. There’s been a bit of snap-back, relative to 2016, among white voters without college degrees (and that’s something that we’ve seen in special elections this cycle, where there’s been even stronger movement in the Dem direction in more rural, Midwestern districts than in suburban areas); the overall gap, though, in favor of Republicans among white voters with “high school or less” is still large, relative to the pre-2016 period. (If you go back to the Bill Clinton era, you can see that the Dems actually had a brief, small edge among non-high school whites!) With leaners included, it’s currently 58 R, 35 D.
Meanwhile, the Dems hit new heights of performance with college-educated whites in 2016, and that trend has only accelerated since then. Among whites with college degrees, the current advantage when leaners are included in 49 D, 46 R (the first Democratic lead in that category in, well, ever). Among those with “postgrad experience,” there’s now actually a gap bigger than the gender gap; the Dem lead is 59 D, 37 R. By comparison, back in the bad old days of 1994, the GOP was the de facto home of whites with college degrees, at 59 R, 34 D, and drew even among post-grads, at 47 R, 46 D. (If you don’t believe me, go back and take a look at county-level maps of elections in the 1980s and 1990s. Guys like Michael Dukakis lost northeastern suburban counties that we’ve taken for granted since then.)
One other thing to keep in mind is that turnout, especially in midterms, is much higher among people at higher education levels. One caveat (and rightly so) that many people have had when looking at large Democratic advantages on the generic House ballot this year, is that Democrats have traditionally had a harder time turning out their base for midterm elections. What’s changed this year is that their base is increasingly consisting of demographics that tend to turn out regularly and, conversely, the Republicans find themselves increasingly reliant on a base that’s less inclined to turn out in off years. (This graphic, by Daily Kos Elections’ Daniel Donner, is based on Census post-election surveys on turnout from 2014, and, when paired with Pew’s latest party ID data, illustrates the GOP’s sudden problem.)
Finally, let’s take a look at the growing generation gap. Pew attracted a lot of attention recently when they finally declared the “Millennial” generation over. (The fuddy-duddys at the Census Bureau don’t name generations, leaving folks like Pew as the main arbiters of how to put people in cleverly-named buckets by age.) Well, the millennials aren’t all dead. What they meant is that all of the millennials have now aged into the electorate — they draw the cutoff at the birth year of 1996, meaning that the youngest millennial is now 21. This also means that Pew is going to need to start surveying respondents from the generation that comes after that, which they admit they haven’t come up with a name for—yet.
And it turns out that the millennials are, by far, the most Democratic-leaning generation. They’ve held that status as long as Pew has been measuring them separately; even in 2004, they were 53 D, 38 R with leaners included. But now they’re at 59 D, 32 R—meaning that the “millennial gap” is even bigger than the gender gap or the education gap.
There starts to be a multiplier effect when you zoom in specifically on millennial women: the gap there is an eye-popping 70 D, 23 R, which is a big gain from 2004, when the gap among millennial women was 54 D, 36 R. (By contrast, millennial men have a gap of merely 49 D, 41 R. That’s significantly better than any other generation, though; Republicans have a significant lead among men from the other three generations.)
Part of the millennial gap, compared with other generations, is based on racial composition. The millennial generation, after all, is much more racially diverse than previous generations (though it’s got nothing on the current up-and-coming generation of current K-12 students, which is already plurality non-white). But even among white millennials alone, there’s a Democratic edge of 52 D, 41 R with leaners included. No other generation has a Democratic gap when you pare it down to just its white members. And when you look at non-white millennials, the gap shoots up to 72 D, 18 R.
It is worth noting that the millennial generation has the biggest share, by far, of people who identify as independents (without leaners included, millennials are 44 I, 35 D, 17 R). Many of those independents, however, lean Democratic. Interestingly, though, more than half of the “independents,” when you push leaners, identify as Republican, suggesting that there’s a large pool of 20-somethings who aren’t on board with the Democrats but who are ashamed to publicly identify as Republican.
Finally, the generational data points to a major problem for the Republicans in coming years: they’re more reliant than ever on a group of people who, increasingly, are permanently leaving the electorate (through death, rather than by choice): the Silent Generation, i.e. people born 1928-1945. This is the only one of the four generations that has moved substantially in the GOP’s direction in the last two decades, going from an actual Dem advantage in 1994 (46 D, 45 R with leaners included) to a big gap in the GOP’s favor today (52 R, 43 D). By contrast, both the Boomers and Generation X have moved from narrow GOP advantages in the 1990s to narrow Dem advantages now.
When you add up all the data, the swing toward the Democrats jibes very well with what we’ve been seeing on the ground, not just with special election results, but also with scholarly investigations of who’s most heavily involved in new pop-up activist groups like Indivisible and Swing Left: it’s disproportionately college-educated women in the suburbs, who weren’t that active in politics but were strongly motivated by the dismaying results of the 2016 election. It also reflects where we’re seeing some of our best pickup opportunities evolving in 2018, in suburban locales like Orange County, California; exurban New Jersey; and suburbs of cities like Dallas, Houston, and Atlanta.