Thinking About Swing State Demography

|
October 4, 2016 9:29 a.m.

Through all the swings and vicissitudes of this campaign and the often fairly dramatic shifts in the polls there have been a number of persistent patterns that have held my attention. One of those is what I’ve referred to as the ‘Clinton Wall’, the fact that while the margin separating the two candidates has swung wildly, Trump has never been able to move into a lead over Clinton. He keeps running at that wall. But he bounces off each time. Another pattern is the shifting electoral map itself. Clinton has had much more persistent and more robust leads in a series of states that weren’t even considered swing states a decade ago, even as she’s struggled (at least in relative terms) in some fairly reliable blue states.

The main pattern is increasing Democratic strength in Upper South states like Virginia and North Carolina and Western states like Colorado and then relative weakness in the Industrial Midwest. In most cases this is relative weakness. So Clinton has held consistent leads in Michigan, Wisconsin and Pennsylvania but they’ve been more tenuous than her lead in Virginia which wasn’t even considered a swing state a dozen years ago. You have the extreme case of Iowa, a swing state which has nonetheless gone Republican only once since 1988 showing a consistent Trump lead for months.

Not surprisingly, the underlying patterns separating these states are the same one we see so clearly dividing the country in the national polls: race, education and age. Take Iowa again. Despite its recent electoral history, Iowa is very white (86.7%), relatively old and has relatively fewer people with college degrees than the country at large. When you break it down like that, it’s less surprising to see Trump doing well there both in absolute terms and relative to the rest of the country and other swing states. On the other hand, Virginia has about the same number of white voters as the country at large. It has significantly more college educated voters and slightly fewer older voters than the rest of the country.

Race is the big factor which unlocks this new electoral map. But age and education also play a clear role. So tonight I put together a chart to pull together what I’ve been mulling for a few months. I pulled the most recent Census data for each of these states on those three demographic markers: the percentage of people who are white and have no Hispanic/Latino origin, the percentage of people with college degrees and the percentage of people over age 65.

Newsletters
Get TPM in your inbox, twice weekly.
Your subscription could not be saved. Please try again.
Your subscription has been successful.

Before proceeding, let me anticipate some obvious and valid criticisms of what I’ve done here. I’ve piled several different demographic numbers into a sort of composite score even though they’re completely unlike variables, apples and oranges. These aren’t meant to be a definitive measure of some specific social reality, more an exercise in seeing the relationship between certain demographic measures and shifting electoral patterns.

So, with that out of the way, I’ve taken 9 swing states and pulled their numbers on those three measures. I’ve then measured these against the national average and expressed the difference as a positive number if they diverge in the direction that has generally favored Clinton and a negative number if they diverge in the direction favoring Trump. Then I add all three up as a “demo score.”

I’ve also included the vote margin for 2012, with the margin for the Democrat expressed as a positive number, Republican, a negative number. I’ve then added the current PollTracker Average for that state and then included a four year trend, to see which direction they’ve moved.

As I said, adding a like of different things together there. But again, this is just a general experimentation.

What’s important to note is that scoring ‘high’ on these numbers doesn’t necessarily make a state more Democratic. I ran the numbers for Georgia too. And it’s higher than any of these swing states. What these numbers do do is give us a decent indication of how swing states that we’ve mainly put in one bucket have diverged – often in surprising directions – in the era of Trump. And the composite score does a pretty good job of indicating which states have and haven’t trended in a more Democratic direction. Lower composite score, the more they’ve trended away from the Democrats and vice versa. So for instance, Iowa, the big pro-Trump stand-out, has by far the highest composite score and the highest move away from the Dems over four years.

Now even if you don’t have much use for my composite score, just looking at the divergences from the national average on these three measures is pretty instructive. And it’s handy to have the data together in one place. We discussed Iowa. But why does Clinton have a tighter than expected race in Wisconsin and possibly losing Ohio? These three numbers explain why they’re trending away while states like Virginia, Colorado and North Carolina are moving in the opposite direction.

The one state that kind of breaks the model is Nevada. Clinton is up there right now. But it’s been almost as much of a Trump hold-out as Iowa for the last couple months even though it’s an extremely racially diverse state. But here you have a possible indication of why: Nevada has the lowest percentage of people with college educations (and yes, this is controlled for age) of any of these states. The data I was using isn’t broken down by college education and race. But this probably means that even though there are fewer whites in the state, a higher proportion of them don’t have a college degree and thus lean hard for Trump.

So there it all is. I find this to be one of the most fascinating stories of the cycle. These are clearly trends that go beyond this election cycle. My question is whether they will be this pronounced in future cycles where the same trends are in place but – presumably? – you don’t have someone as personally and openly incendiary and racist as Trump.

Latest Edblog
Masthead Masthead
Founder & Editor-in-Chief:
Executive Editor:
Managing Editor:
Associate Editor:
Investigations Desk:
Reporters:
Newswriters:
Director of Audience:
Editor at Large:
General Counsel:
Publisher:
Head of Product:
Director of Technology:
Associate Publisher:
Front End Developer:
Senior Designer: