Teaching Infant Mortality
I’m a data geek, with a background in data mining and analytics. So when I get a weird notion that I’m interested in checking out, I have the skillset to be able to ferret out information and analyze it. This means that when the idea popped into my head about whether teacher salary implies anything about infant mortality, I now how to start digging in. The first step is usually a correlation analysis, to uncover whether it’s even worth digging more deeply. That’s what this paper is: a correlation analysis of Teacher Salary versus Infant Mortality.
My data come from multiple sources. For Teacher Salary, I pulled the median teacher data from World Population Review. I vetted this against the data from the nation’s largest teacher labor organization, the National Education Association. I also adjusted this by the cost of living, which I pulled from the Missouri Economic Research and Information Center. For infant mortality, I went directly to the CDC. The principal data sources are where the QR codes in each image take you.
What do the Data Show?
The first thing I noticed, looking at a histogram of the teacher salary data, was how skewed these data are. I would have expected more of a normal distribution (I don’t know why, but that’s what I was thinking stepping into this). In fact, it’s not that way at all. The majority of the states pay teachers on the low end of the spectrum. There are some outliers on the far right, which would be California and Washington State. There’s something causing this, a policy that pushes salaries lower in most states. I’m curious about that, but I haven’t yet dug into it.
A histogram showing 5 bins of teacher salaries, indexed. The important thing about the graphic is that the histogram is skewed to the left, with the lower 50% of the data including almost forty of the fifty states.
I wanted to see what that looks like on a graphic (being a very visual learner). This is what I produced below. The West Coast and the North East regions both seem to pay their teachers better than the other states, but these are not the majority. Generally, it appears that once cost of living is taken into account, most states pay their teachers toward the bottom of the pay scale. Keep in mind that this teacher salary scale only includes K-12, and college is not accounted for in this graphic.
Teacher Salary by State, normalized and adjusted for Cost of Living. The important details here are that most of the states are in the middle of the salary range, even adjusted by cost of living. This is something we’d expect. Two states stand out on the high end: California and Washington State. On the low end, ten states stand out: Mississippi, South Dakota, West Virginia, North Carolina, Arkansas, Missouri, Oklahoma, Montana, Louisiana, and North Dakota.
Infant Mortality by State.The salient information here is how well the graphic lines up to the one above. First, we see that again the West Coast and the North East have the lowest infant mortality rates. There’s also an interesting band across the top of the midwest where infant mortality rates are on the lower side as well. I plan to dig into that later in a different post.
Observations
As we can see above, the maps look very close. In fact, I did a correlation analysis, and they are pretty aligned, though not as strongly correlated as it might first appear. The number, if you’re curious, is -0.55. A good data scientist will tell you that you really want to get to 0.6 for a strong correlation, but this is very close. The negative indicates that as teacher salaries go up, infant mortality goes down. What does this mean though? Does it mean that the magic key to lowering infant mortality is to pay teachers more?
No. What it means can be multiple things. First, and most clearly, there appears to be a correlation between the two, if a weak one. That’s the most basic thing we can say. We can’t say why that correlation exists, but knowing of the two domains, I’d say it’s pretty unlikely that simply bumping up the teacher salary will impact infant mortality directly. What’s more likely is that there’s something about states that pay their teachers more, that also drives down infant mortality. In other words, there’s a common cause for both of these. Another thing sticks out as well. The cost of living in the states with the highest teacher salaries (adjusted for cost of living), is much higher in the states with lower infant mortalities. Could this be it? Higher cost of living yields lower infant mortality? Again, dubious.
Another interesting observation is that when we consider average salaries, then the correlation is very weak (-0.22). So that means that it’s not about paying a few teachers well, rather about paying most teachers a decent wage compared to the cost of living. This highlights a significant limitation of the average as a value when dealing with things like salaries: gross inequalities skew the numbers.
Next Steps
I want to dig into policy in the 2 highest and 2 lowest states next, and see if I can understand better what it is that these states are doing on the teacher salary and the infant mortality fronts to come to such positive outcomes. I’m also curious what other policy differences track with teacher salary, as that can yield policy and administrative clues as to what seems to be working, and what perhaps isn’t working.
