Dear 100 Hour Board,
I just read a comment on a New York Times article (this article, comment by "Gfagan," if anyone's curious) that argued that states run by Republican governments are noticeably worse off than states run by Democratic governments. Now, liberal as I may be, I'm rather skeptical of this claim.
I'm also lazy, which is where you come in.
Would you please design a methodology* to determine the general well-being of each state, and then apply that methodology to determine whether there is any correlation between state well-being and the state's party** of governance?
I realize that there's very little chance of finding proof of causation even if correlation is present, so don't worry about that unless you think you can find empirical evidence to put forward. I'd just like to know whether "Gfagan's" argument has any potential basis in fact.
-the one and only lazefulness
*Preferably based on a cross-section of relevant factors.
**I'm less interested in whether the state's governor currently has an R or D next to their name, and more interested in whether the state has consistently been governed by a single party's policies for an extended period of time. I suppose states with mixed governments (ie R legislature and D governor) or with a history of both parties controlling the government could constitute a third category.
Basically this is a really great question that I was excited to answer. In theory, I was going to do all sorts of cool multivariate regressions with tons of control variables so that I could give you a yes or no response with a high degree of confidence. In practice, however, my analysis didn't turn out as thorough as I'd envisioned, partly because (a) finals and (b) the double edged sword of Christmas Break, that peculiar period when I have ALL THE FREE TIME and still manage to get NOTHING DONE. While my results are suggestive, you'd have to put some more thought and work into this project to make sure they're consistent and reliable.
Deciding what constitutes "well-being" can be a bit controversial and arbitrary. For example, does a higher level of debt mean that a state economy is imperiled or simply that money is flowing at a different rate? Does a lower birthrate mean that the state is in danger of a population implosion or that it's helping to improve the environment? The results will be a little bit circular because how people answer these questions in some measure determines how they vote. When most citizens of a state think that a high birthrate is good, for example, that state is likely to elect the party it sees as most friendly to large, traditional families. So while I've listed a handful of factors below, they are in no way the Alpha and Omega of defining state well-being. I've divided these indicators into three categories (political, social, and economic), explained each one below, and justified why I include it as a positive or negative factor.
Integrity—The Center for Public Integrity has put together an index that gives each state a score from 1 (no integrity) to 100 (perfect integrity). The indicators it uses to determine that score fall into twelve categories, including public access to information, political financing, electoral oversight, accountability, budget processes, civil service management, procurement, internal auditing, lobbying disclosure, and ethics enforcement. The organization maintains that this is not a measure of corruption, although they could've fooled me—it looks at many of the elements that constitute corruption and is probably intimately connected. I include this measure because I think a state's political integrity is an important component of its well-being.
Democracy—This figure, which comes from the Center for American Progress Action Fund, is a measure of state' representativeness (how well the government reflects the wishes and opinions of its constituents) and accessibility (how easy it is for the average voter to participate in the political system). It basically indicates whether democracy in the state is working as intended. The organization gave letter grades from A to F in each of three categories, which I converted into numbers and then added up so that each state gets a score between 3 (democratic fail) and 36 (shining paragon of democracy).
Education—I think this is one of the less controversial measures, as most people can agree that higher levels of education are a good thing. There are different ways you could operationalize this—like the number of adults with college degrees or the standardized test scores for elementary school students—but since it was the easiest data to find, I've gone with the percentage of students who graduate from high school within four years, courtesy of the US Department of Education.
Health—Again, physical health is a fairly undisputed indicator of overall well-being, but actually measuring it is a bit more complicated. I've gone with a composite index from the United Health Foundation, which gives each state a score based on smoking, immunizations, infant mortality, cancer deaths, drug use, air pollution, and other important indicators. The index ranges from a score of -1 (poor health) to 1 (good health).
Unemployment—Another fairly obvious measure, although you may question why I include it as a social indicator rather than an economic one. It has to do with how political policy deals with unemployment; it usually implements social legislation, rather than economic legislation, to reduce it. As I'm sure you can guess, low unemployment is considered a good thing. The data for this comes from the Bureau of Labor Statistics.
Birthrate—Including "birthrate" as a measure of social well-being is, I admit, controversial. It's popular these days to frown upon high birthrates because they drive up the earth's human population and therefore the environmental impact of human activity. I do not dispute that this is the case, and in fact I am legitimately worried about the terrible amounts of waste and pollution we produce in using up our planet's natural resources. There is, however, a flip side to that coin. Plummeting birthrates sound attractive for environmental reasons, but especially as life expectancy increases, they tip the balance of the social economy: suddenly you have lots of elderly people who depend on social security to sustain themselves, but social security is itself maintained on an ever-shrinking population of workers. Some countries in Europe, for example, have birthrates as low as 1.4, and they are having to face the very real challenges of dying cities, growing health costs, and a dwindling population. Experts reckon that a birthrate of 2.1 children per woman (which is, thankfully, America's average) is necessary to maintain a population that can support itself economically.
There is another reason that I'm choosing to include high birthrate as a positive indicator of state well-being. Presumably, people who (a) feel secure in their future, (b) feel safe in the place they live, and (c) feel that their communities provide ample support for children and their parents are more likely to have kids. Birthrate therefore acts as a proxy measure for how much people like living in the state. I realize that there are more valid proxies, of course, but under the circumstances this is the best I could do. I pulled the information from the National Vital Statistics Report.
GSPpc—"Do you mean GDPpc?" I hear you ask. No, I actually mean GSPpc—gross state product, as opposed to gross domestic product. Yes, it's a real thing, as reported by the Bureau of Economic Analysis. I used the figures from 2012 because I couldn't find any that were more recent. Basically, "gross state product per capita" refers to the state's total economic output divided by the number of people residing in that state. It is intended to measure average wealth: the wealthier the people in the state, the better off the state as a whole. Of course, we all know that the equation money = happiness is at best incomplete, but we'll let it slide for the purpose of this exercise.
Debt—"Can states even go into debt?" you may be wondering. Indeed they can. Again, some people don't feel that sovereign or state debt is as much of a problem as others make it out to be; they see it as simply a sign of a healthy and flowing economy. However, I think the consensus of the majority is that it's good for states to not be running a deficit. So I'm using debt as a negative measure. I got my data from an organization called State Budget Solutions, which seems pretty reputable, but instead of using the gross sum as listed on that page, I divided each figure by the number of people in that state—because of course California is going to have more debt than Wyoming, but that's because California has the world's eighth largest economy. So the measure we end up with is actually "state debt per capita," which hopefully gives us a more accurate picture.
Trade—Unless you're a protectionist (don't worry, I won't judge)(okay, actually I will judge), trade is generally considered an indicator of a state's self-sufficiency, natural resources, and economic viability. There are many ways of measuring it, but I've gone with the International Trade Administration's data on the value of state exports per capita.
Liberal or Conservative?
Instead of simply labeling each state Republican, Democratic, or swing, I've come up with a more precise variable for measuring party loyalty. Some nice person compiled the information represented in this Wikipedia map, placing the exact numbers for each state in the caption below. The number represents the average margin of victory over the last several presidential elections—in other words, what percentage Democrats or Republicans win by in each state, on average. To make it a single, composite measure, I've put it on a scale from negative to positive, where negative numbers represent Democratic victories and positive numbers represent Republican victories. Therefore, democratic strongholds cluster on the negative end, Republican strongholds on the positive end, and swing states around zero.
Because I wasn't on campus to use campus computers' advanced statistics software, I just went ahead and ran my regressions in Excel. I'm not as familiar with the Excel data analysis tools, however, so if I've made any mistakes in interpreting the results, I'm happy to be corrected. Basically, I just regressed each dependent variable (integrity, democracy, education, health, unemployment, birthrate, GSPpc, debt, and trade) on the independent variable, party, and looked at the p-value, coefficient, and R-squared of each to determine whether there was a significant relationship. If you're interested in seeing the data set or looking at my regression results, here is the Excel document I used to calculate it all.
The correlations between party and four dependent variables (integrity, unemployment, GSPpc, and trade) came out statistically insignificant. This means that we cannot be reasonably certain that there is any relationship between party and any of those indicators, at least not as I've chosen to measure them. Education came out with a p-value of 0.00 and a coefficient of 0.00—in other words, we can be sure that there is no relationship between them. (Admittedly, I had to remove Idaho, Oklahoma, and Montana—three moderately or very Republican states—from the regression because they did not provide data on high school graduation rates.)
The variables that did show up as significant are democracy, health, birthrate, and debt. Here's how party affects each of those factors:
Republican states are insignificantly less democratic. For every additional percent of the vote the Republicans win, on average, democracy drops by 0.147. That score is on a scale of 3 to 36, which means that a very Republican state like Utah (with an average R. margin of a staggering 30.8%) is predicted to score about 8% more democratic (representative and accessible) than a moderately Republican state like Montana (with an average R. margin of 9.7%). This is actually not too surprising—Democrats in general are more likely to support easier access to the polls. The correlation, however, is pretty weak, since the very most Republican state (Utah, unsurprisingly) would only be expected to score 8.3 points less than the very most Democratic state (Massachusetts, unsurprisingly). Nor is it a very strong predictor: a correlation technically exists, but many states don't behave as that correlation would lead you to expect. For example, while the equation predicts that Utah will score 8.3 below Massachusetts, in reality Utah scores one above it, 18 to 17. The conclusion? Meh. Maybe Republican-run governments are slightly less democratic than Democratic-run governments, but the margin is nearly insignificant.
Republican states are significantly less healthy. For every additional percent of the vote the Republicans win, on average, health drops by 0.012. Keeping with our previous examples, we would expect Utah to score 0.24 points lower than Montana and 0.68 lower than Massachusetts. That seems absolutely miniscule, but remember that the scale only goes from -1 to 1, so 0.68 is actually quite significant. Once again, party is only a weak predictor, explaining about 10% of the variation in the quality of health in any given state. There are lots of uncontrolled-for variables that ought to be taken into account. That aside, though, it does appear that Democratic-run states tend to be moderately healthier than Republican-run states.
I'd like to point out that Utah actually completely bucks this trend; even though it's the most conservative state in the nation, it comes in 8th out of 50 for quality of health. Well done, Utah!
Republican states have significantly higher birthrates. For every additional percent of the vote the Republicans win, on average, the birthrate rises by 0.009, meaning that Utah should have a birthrate about 0.18 higher than Montana and 0.51 higher than Massachusetts. That's not too far off from the actual truth: although Utah is something of an outlier, it scores 0.37 higher than Montana and 0.74 higher than Massachusetts. Party is actually a very strong predictor of birthrate, although (as I explained earlier) this relationship obviously has some circularity since the Republican Party is perceived as more friendly to large families.
Republican states have significantly less debt. For every additional percent of the vote the Republicans win, on average, debt drops by $141.45 per person. According to that number, then, Utah should have $2,829 less debt per capita than Montana and $8,034 less than Massachusetts. The relationship isn't super strong—again, a lot of states don't behave as it would predict—but it is pretty significant, considering that the range (the difference between the state with the most debt and the state with the least) is on the order of $34,000 per capita.
The results are mixed. Party affiliation only affects four of the nine indicators of well-being I selected, and even then it's split evenly down the middle: Republican states fare worse in the first two categories and better in the latter two. I haven't made any effort to combine all these indicators into a single composite score that measures "state well-being" as a whole, so I'll leave it to you to decide which categories are the most important. If you're more concerned about democracy and health, then maybe your commenter Gfagan was correct—Democratic states really are better run than Republican states. If birthrates and debt are more important to you, then maybe he's wrong.
I should also mention that I haven't controlled for any additional variables. Party affiliation is only one of many, many things that affect such indicators, so since I'm lacking control variables, my results may be totally off. I may have claimed that relationships exist when they really don't or that they don't exist when they really do. A more rigorous study would take stuff like that into account, but I ran out of time to do so. Again, the variables I chose and the way I operationalized ones like "education" and "health" are kind of arbitrary, and the results may change if you were to use different indexes or different indicators for "state well-being." To yayfulness and others who are sufficiently interested (or sufficiently bored, I guess) to have read this far, I'd welcome emails that question my methodology or suggest improvements, because there may be other factors that I've totally overlooked.
My interpretation of the results? I'm pretty liberal myself, but I don't think you can in any way say that Democratic states are "obviously" better-run than Republican states. It's too nuanced of an issue—sweeping, blanket statements like that ignore all the problems of measurement, operationalization, and bias that make this such a hard question to answer. I think a much more fruitful line of inquiry would be to examine specific policies to determine whether they've been effective and then align your vote with the party or candidate that supports the best ones.
Wishing I could have turned this in as my capstone paper, &c.