Back in the 1950s, Darrell Huff wrote a little book entitled How to Lie with Statistics that is still in print today. While the context of the book is rather outdated (annual salaries of $2000 and the like), the mathematical principles in it are still quite relevant. It is still quite easy to support a cause by playing with statistics.
For example, I recall hearing that the divorce rate in Utah was high. I did a little digging and found that Utah had the 20th lowest rate of divorce in the nation. Not particularly high, but not all that low either, more middle-of-the-pack. But then I actually looked at what was being measured. It was divorces per capita. Wait a minute. What about divorces per marriage? That's what comes to mind when people hear "divorce rate" and I would think Utah would have a pretty high number of marriages. Sure enough, when I ran some calculations, Utah actually had the second lowest rate of divorces per marriage (Wyoming had the lowest, and Idaho was third or fourth, which along with Utah comprise the three most predominately Mormon states).
Another interesting example was the recent report that when looking at the top ten large cities in percent of same sex couples raising children, eight of the ten happen to be in states that have a constitution which defines marriage as opposite gender only. That seems amazing until you realize that 60% of the states have such constitutions. So on average, choose any ten random states and a majority of them will have such constitutions. Suddenly the news doesn't seem like such a big deal after all. So it's hard to draw any real conclusions from such statistics.
Basically, even when the statistics are perfectly accurate, they can be presented in such a way to support an untrue idea (or refute a true idea). So always question the numbers you are fed. If you want to use them to support your thesis, it's always a good idea to know exactly what those numbers measure, and their context in the populations involved.
No comments:
Post a Comment