Monday, November 21, 2005

I hate poor causal reasoning... even when it's in the service of good

Radio spots. They're awful about it. I just got out of my car, where I heard a spot by United for Peace and Justice, a faith-based social justice group. They were pointing out that third grade test scores are exactly correlated with the number of incarcerations a few years on down the line, so why don't we focus on education and not building more prisons? I agree. Education > prisons. However, implicit in how they phrased the spot was "low test scores cause future criminality," or, at the very least, "low levels of education cause criminality." Now, undoubtedly, there is a connection. Criminals tend to have lower levels of education (depending on the crime, obviously). But that connection isn't necessarily causal. Could it be that there is a common cause, contributing to both low test scores and high incarceration rates? Poverty, for one. Maybe instead of building new prisons, we should focus on lifting people out of poverty?

Also, I keep hearing a PSA on the radio claiming that if you only have dinner with your family, your kids will be so many times less likely to do drugs. Now, I do not doubt that there is a correlation between having family together time and low rates of drug abuse. But, again, kids do not not use drugs because their family ate together. No, it's something else. Something about families that tend to eat together causes kids to be less likely to do drugs, some other characteristic(s) that are found in families that typically have family-time. Closeness? Openness? Who knows. (I'm sure someone has done the research.) But, while they are in the service of good, they are making claims that simply aren't backed up. Correlation doesn't equal causation.

Another example: MADD (Mothers Against Drunk Driving) has a radio spot telling parents not to give their kids alcohol because they younger that you give it to them, the more likely they are to have an alcohol problem when they grow up. This is based on research showing that the earlier that kids have their first encounter with alcohol, the more likely they are to have problems with it. The problem with their analysis is that when parents give their children alcohol, it's not necessarily the same situation as when a kid gets alcohol from a buddy. I know many parents who let their children sample alcohol at home so as not to create a problem of "forbidden fruit" or whatever. Those kids (I would bet) are less likely to go on to abuse alcohol. At least, that sort of reasoning is behind recent programs at colleges (sorry can't be bothered to find links, I'm lazy) which seek to introduce students to alcohol as a "normal" part of meals, so as to curb binge drinking. But for MADD to say that the earlier a parent gives a child alcohol, the more likely he or she is to abuse it, based on research showing that earlier first consumption leads to abuse, is kind of misleading.

Anyway, that's all I wanted to say. All of these ads: I agree with them. Well, maybe not the MADD one. But certainly the education>prisons and the family-time one. But it seems like it hurts more than helps when they use not-good-reasoning.

No comments: