One of my goals with this blog is to help people think more systematically and logically about how society really works. I've mentioned that thinking like a sociologist involves employing some of the scientific tools that sociologists use. This post focuses on statistics. If numbers make you queasy, no need to worry. For this example, I don't really need numbers.
Numbers Aren't Always Enough:
Using statistics goes beyond looking at numbers. You have to understand the relationships between different numbers. If you understand how to look at the relationship between two numbers, you will be a smidgen better prepared to understand society.
So, what do I mean by claiming that ice cream causes crime? How does that headline relate to understanding the relationships between numbers, to better understand how society works?
When Being Together Isn't Enough:
I'm not going to change course and talk about faltering relationships. In the sciences, we use the term "correlation" to describe a situation where one numerical value changes in some relationship to changes in another numerical value. In the ice cream example, ice cream sales tend to increase as burglaries increase. There really is a correlation. Why? Let us ignore the strength of the correlation for this example.
Does it make sense that ice cream causes burglaries? Of course not! You can sense that something else must be going on. Well, ice cream increase in summer, and many people go out of town on vacation. The correlation is meaningless in this case. Time of year explains the seemingly related increase in both ice cream sales and burglaries.
That section heading also applies to relationships between one thing we observe and another thing we observe. Correlation does not equal causation! Just because two things seem to have some relationship, it doesn't mean one causes the other.
To prove that ice cream causes burglaries we would have to do three things. First, we would have to establish a time order of events, so increased ice cream eating has to come before increasing burglaries. Second, there must be a correlation between ice cream sales and burglaries. Lastly, we would have to exclude all other factors that might influence ice cream sales or burglaries, or both.
Those three conditions are actually quite difficult to satisfy. It happens more often in the physical sciences than in sociology. Sociological research underlies many claims you see in the news. While straight news sources tend to do a decent job of presenting social research, this isn't necessarily so for politicians and talk show hosts.
Abusing Cause-and-Effect Thinking:
Stay alert for claims that X causes Y. X will be some government policy, or social trend, or behavior. Y is always some real or imagined social problem. Consider this made-up example: "Strict gun control laws cause more violent crime." X and Y in this case should be clear.
What if the crime rate does go up? Handguns are banned, then crime goes up. Seems like an obvious connection, right? Wrong. You should ask more questions about the situation. What was the trend in crime rates before the ban? Did the city's economy suffer a severe blow in 2002 or 2003? Was there a change in how violent crimes are reported?
Next time you listen to a talk show host or political analyst or politician explain how one thing "causes" another, you will be better prepared to know if you are being bamboozled.
(The ice cream example came from one of my graduate courses and I never had a source for it.)
No comments:
Post a Comment