I assumed that the lazy, inaccurate red-state/blue-state meme would fade away as the 2004 election receded from our memories. Instead, it seems to have congealed into conventional wisdom. Hardly a day goes by without a diary alluding to "red state values" or differences in family income, crime rates, birth rates, etc., between the states (the other side is just as bad, of course).
Such state level analysis - a mere artifact of our peculiar electoral system - contradicts public opinion data, spits in the face of statistical literacy, and cheapens discourse on both sides of the blogosphere.
This program is continued on side B of the cassette. . .
Those immoral [other]-staters
The most common usage of red/blue is to attack individual citizens with state-level data. For example, red states have higher rates of teen pregnancy. So much for "family values," eh? The problem is, drawing individual-level inferences from state-level data is what we call the ecological fallacy. It might very well be that the Democrats within each state are the ones getting pregnant. Or not. But the point is, there's no way to know from state-level data.
Mars and Venus?
State-level data is also problematic because intra-state ideological variation is far more pronounced than inter-state variation.
Many diarists, commenters, and friends of mine hold the view that visiting (or God forbid, living in) a red state is akin to crash landing your spaceship into a Dagobah swamp. Perhaps you saw a cowboy hat during your layover in DFW, and suddenly you're an amateur demographer. Unfortunately, the plural of "anecdote" isn't data. But here's some data:
In his book Culture War? preeminent political scientist Morris Fiorina examines red and blue states, and finds them to be quite purple. Here are some examples:
(Question: Red State %, Blue State %)
Democratic self-ID: 32, 40
Republican self-ID: 34, 25
Liberal self-ID: 11, 20
Conservative self-ID: 31, 24
Do you agree with the following statements?
Religion is very important in my life: 74, 62
Gov should ensure fair treatment of blacks in employment: 51, 57
Too much power concentrated in large companies: 62, 64
Homosexual adoption should be allowed: 40, 52
Favor abolition of inheritance tax: 72, 70
Favor government grants to religious organizations: 66, 67
Favor school vouchers: 50, 54
Favor medicare coverage of prescription drugs: 92, 91
Do whatever it takes to protect the environment: 64, 70
Make English official language: 66, 70
Favor death penalty: 77, 70
I'll take Chapel Hill, you can have Barstow.
Does this mean that the country isn't polarized? Not necessarily. But it definitely means that "state" is the wrong unit of analysis.
Most states are diverse enough to encompass the entire political spectrum. Some argue, however, that increasing polarization is found at the county level. For example, a study by the Austin American-Statesman showed that the number of "landslide" counties - those in which one candidate received more than 60% of the vote - has increased rapidly since the 1970s. Others are less convinced, however. See Phillip Klinkner's excellent takedown of the series.
The origins of the red/blue myth
The most annoying aspect of this meme is that it arose merely as a fluky byproduct of our winner-take-all electoral college system. In a presidential election, states can only be red or blue. If California goes 50.001% Democratic, then the Democrat gets all 55 electoral votes. There ain't no purple on Tim Russert's dry-erase board.
But that's it. There is no demographic, geographic, social, or political relevance to state-level public opinion data, except in the spreadsheets of presidential campaign consultants. That's not to say that all state-level political analysis is flawed. In fact, it can be useful if we're talking about institutions that vary discretely by state - legislatures, governors, state policy output, etc. It also has obvious implications for Senate elections. But there is little value in aggregating the American public by state for the purpose of broad public opinion analysis. And there is NO value in making the analysis dichotomous (all or nothing, red or blue).
Implications
Enough with the data. Permit me to revert to defensive red-stater for a moment. I realize that much of the "red states are evil" analysis claims to be in jest. But the sum total of my real-life and Internet interactions leads me to believe that many of y'all are dead serious. When I was a graduate student, I met countless future professors who steadfastly refuse to apply to jobs in "red" areas (and if you know anything about the academic job market, you'd understand how stupid that is).
I can also attest to the importance of intra-state variation over inter-state. Lately I've been splitting my time between Texas (urban) and Maryland. I've seen some rednecks over the last year that would make your hair stand on end. I've seen numerous anti-gay and anti-Muslim bumper stickers as well. And they were ALL in rural Maryland. The urban-suburban-rural split is far more consequential for American politics than the artificial "state" construct.
Finally, I spend all day in the "trenches," by which I mean teaching wealthy, conservative, Southern college students about government and politics. Name an epithet for a liberal, and I've been called it. So the last thing I want to do is log onto a friendly website at the end of the day and bear the brunt of your condescension because at least 51% of my neighbors voted Repbulican in one election (especially if your state has elected an Austrian actor to high office lately).
Second in a series of diaries applying political science insights to the myths of contemporary public discourse. Here's the first.