You’ve probably heard all the alleged reasons why the political process in North Carolina is broken. Apparently, our politicians are craven, too indebted to special interests and “the wealthy” to do the bidding of “working people” (hint: virtually no household with wealth is lacking a very hard-working person), and too fixated on public policy opinions and the next election to make tough, long-term decisions about the future of the state.
I have a different diagnosis. It includes the observation that politicians and those who comment on them lack a working understanding of statistics.
Notice I didn’t say we had too many participants in the debate who can’t do calculus. I’m not talking about high-level mathematics here. I’m talking about a basic familiarity with statistical comments such as significance, probability, and causality.
I think that the worst argument you will hear in political discourse starts with phrases like “history has shown” and “the numbers don’t lie.” All too often, those who begin sentences this way end them with egregious errors, faulty conclusions, or, well, numbers that lie.
For example, there’s the old problem of confusing correlation with causality. Just because one thing happens at the same time another thing happens doesn’t mean that one caused the other. Consider the example of the sale of cold-weather clothing. Suppose that you observe a strong upswing in the sale of coats, scarves, and mittens. Suppose you also observe that the sun is setting earlier and earlier in the day. Does that mean that the sale of winter clothes has an impact on the relationship between the earth and the sun? Of course not. The opposite statement exposes the causality. The onset of winter increases the demand for warm clothes. Similarly, the sale of winter clothes may rise at the same time that heating-oil sales rise. But one does not cause the other. Each has a common cause, the weather.
Dr. John Station, a scientist at Duke University and a longtime friend, once provided an example of a related problem, which is the suggestion that if A and B are strongly correlated, and B and C are strongly correlated, then there must be a strong correlation between A and C. Station pointed out that while age is associated with height, at least to a point, and age is also associated with gray hair as the years go on, height and gray hair demonstrate little or no statistical relationship.
What does all this have to do with politics? Unfortunately, you often see the political versions of “if you’ll just buy more mittens, the sun will set faster” and “my new hair-graying program will increase average heights and basketball victories through the land.” Some argue, for instance, that North Carolina’s ABCs accountability program for public schools has “caused” test scores in the state to rise. But as you can see from the graph on the bottom of this page test scores were already rising markedly before the advent of the ABCs program in 1996 – and other policies, such as teacher-pay increases, the statewide Smart Start program, charter schools, also kicked in at roughly the same time. So you have no significant change in the trend line, and even if you did there are several potential causes.
I guess my point today is that before we come up with grandiose plans to “fix” politics in North Carolina – including campaign-finance rules that infringe on basic constitutional rights or longer terms for politicians so they can tax, spend, and borrow with impunity – let’s consider some lower-cost options. How about buying every legislator, mayor, commissioner, and council member a statistics textbook and a copy of Thomas Sowell’s Basic Economics?
Hood is president of the John Locke Foundation and publisher of Carolina Journal.