An article, last week, in The Washington Post, reports:
"Researchers at Ohio State University garnered little attention in February when they found that youngsters who lose their virginity earlier than their peers are more likely to become juvenile delinquents. So obvious and well established was the contribution of early sex to later delinquency that the idea was already part of the required curriculum for federal "abstinence only" programs.
"There was just one problem: It is probably not true. Other things being equal, a more probing study has found, youngsters who have consensual sex in their early-teen or even preteen years are, if anything, less likely to engage in delinquent behavior later on." [read the article]
The story is not, however, about the advantages of teen sex.
Rather, it's about how we can be misled by data, especially data that seem to support something we'd like to believe.
In the case of teen sex and delinquency, for example, it was not new data which disproved the findings by researchers at Ohio State; it was a new, and more careful, analysis of the same data.
Each of the cases discussed in the article bear some resemblance to Simpson's Paradox, a strange quirk of statistics which I discuss in more detail here.
The basic idea behind Simpson's paradox is that when you're dealing with numbers the big picture sometimes hides the truth.
In the case of the Ohio study, the numbers showed a correlation between early sex and delinquent behavior.
But the second study removed the possibility of genetic influences, by focusing on the pairs of twins in the original data. It turned out teens who had sex earlier than their twin did not exhibit more delinquent behavior than the twin did. In other words, early sex does not cause delinquency.
A third study found that both earlier sex and delinquency had the same cause—a genetic tendency toward risk-taking. This explains the correlation the first study found. It also found that early sex actually decreases the probability of delinquent behavior, other factors being equal.
Apparently there's only so many directions a young person can focus their energy in.
The article goes on to cite similar findings on other questions. The correlation between breast feeding and higher IQs turns out to be because IQ is hereditary, and smart mothers are more likely to breast-feed—not because breast feeding improves intelligence.
The death rate for children born to teenage mothers is higher than that for those born to older women for the same kind of reason: most of the youngest mothers are poor, and poverty (not the age of the mother) causes the high death rate. When economic status is taken into account, it's children born to older mothers who are at greater risk.
What do all of these cases have in common, other than a resonance with Simpson's paradox?
They all involve the struggle, in our culture and in ourselves, between a legal model and a wisdom model of morality. In each of these cases there is a behavior we would like to discourage—that we think of as "wrong" to some degree—early sex, early childbirth, bottle feeding.
The legal model would like evidence that "wrong" behaviors lead to punishment or to greater sin—that early sex leads to delinquency, that early childbearing leads to dead babies, that bottle feeding leads to stupid babies.
When we're operating out of a legal model, we go to the data in order to find support for the right opinion, rather than to gain deeper understanding. We want the traditional position to be supported, whether we come from a conservative tradition or a liberal tradition, from a religious tradition or a secular one.
We prefer "goodness" to truth.
And so, we are tempted to accept the big picture (when it supports our beliefs) and look no deeper.
