I hadn’t planned on getting to a “part 3″, but neither is it surprising that this turns out to be a large subject. It’s counter-intuitive to say Democrats need to win more white voters when so much discussion is about the shrinking white majority and the nation’s growing demographic groups (growing as a portion of the population — they’re all growing in absolute numbers) being Democratic-leaning. In the first part I addressed why we need to do better and summarized in part two, so the even briefer version, we need more whites because the presidential margin was thin and we’re losing too many districts to win majorities. The second part was an overview of the variables we’re working with. One of those variables is population density, which, while trying to be brief, I explained was so important and more complex than could be dealt with briefly, that I planned a future post on it. Here we are.
And here’s that chart which, judging from the fact this is the third or fourth post where I’ve used it, I must like a lot.
There in one chart is a trend most readers have probably seen in maps showing even blue states looking all red with spots of blue, or inferred when noticing which presidential candidate won which states. Mostly for the benefit of those who can’t see the chart, the density of a precinct made little difference from the sparsest precincts until what looks like about 300 people per square mile. Romney had a large lead over Obama. There’s a crossover around 800, and then a massive Obama advantage in more densely populated areas. Though it measures just the presidential race, when looking at who won which congressional districts, it seems clear the chart would look the same. Legislative maps tend to have the same look of blue islands in a red sea.
Is density the real problem, and not race? Do whites vote equally Republican regardless of where they live, and Democratic-leaning demographic groups (DLDGs) equally Democratic regardless of where they live? Given how whites in urban areas I’m familiar with tend to vote Democratic (admittedly not a scientific sample, but what I have for now), I’m thinking density rather than race is the key, except that begs the question of whether DLDGs vote as Democratic in rural areas as they do in cities (keep in mind DLDGs aren’t just non-white, but can include young people, non-religious, etc.) Again, the instances of rural DLDGs I know of, not an exhaustive survey, vote Democratic like in urban areas. So looks like neither race nor density is determinative — but yet there’s that graph. Maybe DLDGs in rural areas vote Democratic, but there aren’t enough of them to make much difference, while whites in urban areas have come to vote more Democratic like other urbanites. That crossover at 800 people per square mile certainly suggests density is itself a factor.
Being aware that correlation doesn’t equal causation, we don’t just assume density makes people liberal or sparsity makes people conservative — though we don’t rule that out either. Maybe some personality traits cause people to prefer aspects of urban life, and to tend toward liberalism. Maybe some personality traits lead people to prefer small town and rural life, and to prefer conservatism. Another possibility is people are self-selecting. I doubt more than a tiny portion of the population would pick where they live according to the party likely to win the local elections, but maybe they’re moving to be with people who are more like themselves.
Essentially, we’re looking at something important, but cause and effect are not obvious. So we’re looking at possibilities and likelihoods, and at some point, what can we do with the information?
Let’s start with the explanation for this correlation that strikes me as easiest to dismiss, that people have self-selected according to political inclinations. Politically active people might let party or ideology be a factor in where they choose to live, but most people aren’t that engaged and don’t have a strong party identification, and maybe don’t even know what “ideology” means. In this article from The New York Times Magazine about how the parties have adapted to changing circumstances, there was a lengthy section about a Republican pollster asking focus groups about impressions of the two parties. It was striking how the groups thought of themselves generally as conservatives, yet generally agreed with liberal attitudes and positions on current issues. The words they associated with each party were mostly positive about Democrats and negative about Republicans. I speculate that the only reason they think of themselves as conservative is the decades-long successful demonization of the word “liberal”, and no pollster has to my knowledge ever used the word “progressive”. Even among political activists, I’ve seen surprise when shown electoral maps of a state divided by county or legislative district. That’s anecdotal, but if even activists couldn’t have predicted what the map of election results would look like in their own state, I have trouble believing the average voter has any awareness of it at all.
In short, I’m just not seeing any evidence people are choosing where they live based on politics, and there seem to be stronger explanations.
So is there something in some people’s personalities that make them prefer greater density — and liberal politics? Conversely, some people might have personalities that make them prefer more sparse areas — and conservative politics. Regular readers of liberal blogs have probably come across research on differences between liberal and conservative personalities, and even different brain activity. The author of The Republican Brain, Chris Mooney, recently wrote this article on liberals and conservatives having different preferences as consumers. Essentially, conservatives are inclined to stick with the same brands they’ve been used to, while liberals are more likely to try something unfamiliar. Marketers have learned to try new brands first in liberal areas. Is there a political connection to certain brands? Sometimes, but there’s usually no explicit connection. It seems to be a personality trait that makes people more open to trying new brands and more liberal in their politics. It seems to fit with the idea that a defining difference between liberals and conservatives is their attitude toward change, where liberals drive it, conservatives resist it.
If personality is driving people, then maybe it’s a matter of people who prefer conservative politics also prefer the aspects of life in sparsely populated areas, while other people find themselves drawn to the aspects of life in densely populated areas and, by coincidence, to liberal politics. Perhaps the people living in those areas with the 800 people per square mile, the crossover between Obama and Romney, aren’t just divided fairly evenly between liberals and conservatives, but actually have personalities that make them prefer that level of density and, by coincidence, more centrist politics. This is even before muddying the waters with the other reasons people live where they do, like being near preferred schools or their job, remaining near family, needing mass transit, being unable to sell their house, etc.
What if it’s the other way around however, meaning that the correlation isn’t coincidence, but density makes people liberal and lack of it makes people conservative. We can’t just shift people around for a political science experiment. We do however know from our history that change has almost always been driven from our cities. It’s hard to boil all history down to data points, but looking at other cultures and times, the fact of change being driven by densely populated areas and resisted in sparsely populated areas seems like a constant. That’s counting only internal change, not change enforced on a culture from outside. Being forced to make best guesses from limited information, I think it’s both, and I suspect a feedback is going on that’s driving the stronger partisanship we’ve seen in recent decades. In other words, political preferences are being driven by personality traits that also make people prefer certain densities, but density does affect personality at the same time, with more pushing people towards liberalism and less pushing towards conservatism.
The why do we have conservatives in cities and liberals in some rural areas? Besides whatever is going on the lives of individuals, we’re looking at tendencies, not rules. Remember in this series we’re thinking about winning more white votes, and race is a huge factor in American politics. The idea that urban whites are a bunch of liberals would probably have sounded delusional to non-whites fighting segregation during the civil rights movement, yet looking at it from the present, it seems even during that era, white liberals concentrated in cities. White conservatives left cities for suburbs. Likewise, those rural areas that are predominantly non-white (like Indian reservations or the Southeastern “black belt”) tend to be as Democratic as non-white urban areas or non-white suburbs. So it seems whites vary their politics by density, but non-whites don’t.
Those are multiple possibilities and considerable uncertainty. All we still know for sure is Democrats are winning where populations are dense, and losing where they’re sparse, with exceptions, but the trend only holds up for whites, not for non-whites. Can we apply this to anything useful, like winning elections?
Yes we can. Something jumps out immediately, namely, use density to identify winnable districts. We do that already by looking at results from prior elections, but that seems a bit like trying to understand what happened in a game by looking at the final score. That tells us the result, but not how it came to be. Assuming anything around the 800 people per square mile crossing point should be winnable, that anything to the left on the chart should be red, and anything to the right should be blue, we can probably find election results that don’t match expectations. Maybe the difference is the demographic mix, or an entrenched incumbent defies other factors, or an incumbent had a scandal but stayed in the race, or one local party is much stronger than the other. So use density as an independent factor, apart from recent election results or partisan voting indexes or income or anything else to identify winnable districts, and hopefully find districts we’ve overlooked.
That doesn’t of course get us to solving the problem of winning where, right now, we don’t and shouldn’t. We need to figure out if the exceptions have something unique going on, or something that can be replicated elsewhere, so we can do better where demographics and population density say we should lose. By “better”, I don’t mean just scratching out some more blue votes in red areas to win statewide races, but winning enough districts to win majorities of seats.
So now that we’ve established the why and the overview of winning more while votes, it’s time to dig into the how. Looks like the post that turned into a trilogy is going to have to become a … what’s the word like “trilogy” for something with four parts?