Good Statistics and Bad Inferences
I’ve used this quote more than once here:
“There are three kinds of lies: lies, damned lies, and statistics.”
Sometimes people manipulate statistics to deceive. Statistics are used to get you to give money to shady charities, invest in faulty opportunities, or convince you that what you’re fighting for is wrong. But sometimes, people take statistics and just miss the point completely.
In Cultural Literacy, E.D. Hirsch talks about the Coleman Report, released in 1966, which showed that socioeconomic status was the most prominent determiner of student success, despite the best efforts of schools. Most educational leaders seemed to take that to mean that no matter what schools did, poor kids were going to perform well below wealthy kids. It has become a self-fulfilling prophecy.
But Steven Leavitt and Stephen Dubner pointed out in Freakonmics that, though there is a high correlation between the number of books in the home and child success, it doesn’t mean that the books are important. Their studies showed that there was no correlation connecting student success with being read to everyday. The best explanation (according to the authors) is that parents with a lot of books in the home place a high value on education, which is the actual determiner of success.
In the same way, Hirsch explains that the socioeconomic achievement differences are can likely be attributed to the cultural information children have when they begin school. The literacy levels of six-year-olds are roughly the same, but by seven and eight years of age, they have begun to diverge greatly. He argues that rather than continue the same course and throw up our hands, we should start working to give poor students the cultural information they need for success.
In the continuation of this great tradition of making bad inferences based on sound data, a Reuters wire story announces, “Sex infections still growing in U.S.”:
WASHINGTON (Reuters) – American squeamishness about talking about sex has helped keep common sexually transmitted infections far too common, especially among vulnerable teens, U.S. researchers reported Monday.
Is “American squeamishness” truly to blame? Are Americans even squeamish about talking about sex? I don’t think we are. I’m really only considering TV and movies in my informal analysis, but sex is everywhere. We are definitely not squeamish about talking about it, showing it, selling it, or abusing it. In fact, maybe it’s our lack of squeamishness that is to blame.
It’s easy to blame the problem on abstinence-only education, as those interviewed in the article do. But is that the real problem? For example, what’s the connection between socioeconomic status or subculture and early sexual activity, contraceptive use, and teen pregnancy? What about race, public or private education, religious practice, local cultural norms, etc.? These questions are totally unanswered, while all the problems are set at the door of abstinence only education.
And you know what question is totally ignored? In 1984 just over 400,000 women reported STIs(Sexually Transmitted Infections) to the CDC, compared with over 550,000 men. in 1998, 700,000 women and 300,000 men. Quite a drastic change. But in 2007 the numbers had ballooned to 1.03 million women, and less than 480,000 men. Why are STIs seemingly targeting women over the last 25 years? I can’t wait to see the ridiculous inferences made from that study.