08 February 2012

Idiots and Statistics

Here’s a perfect example of the two combining:

Spanking children can cause long-term developmental damage and may even lower a child's IQ, according to a new Canadian analysis that seeks to shift the ethical debate over corporal punishment into the medical sphere.
"What people have realized is that physical punishment doesn't only predict aggression consistently, it also predicts internalizing kinds of difficulties, like depression and substance use," said Durant. [Emphasis added.]

In the first place, the mere presence of the words “may” and “can” couched around the conclusions points rather obviously to the study establishing simple correlation, and nothing more.  I haven’t got my hands on a copy of the original study, but I’d bet $50 that all the study proves is basic correlation.  And, as anyone who ever paid attention in a statistics class knows, correlation is not causation.

In the second place, causation cannot be established for the very simple reason that one simply cannot know how the subjects would eventually behave had they not been spanked as children.  Spanking is a binary condition (either one is or is not spanked as a child), and so it is impossible to say how much better or worse behaved one would be had he been spanked or not been spanked as a child.  Quite simply, there is no way to adequately compare.

Finally, the researchers appear to have neglected to ask one very simple question:  why do parents spank their children?  Most parents, if asked, would say that they spank their children in order to punish their children for misbehavior, generally in the hopes of preventing that behavior from occurring again.

Now, assuming this is the case, and also assuming that the frequency of such punishment directly correlates to the frequency of misbehavior, then it should stand to reason that the more inclined a child is to misbehave, the more spankings he would receive.  As such, it is plausible—though not guaranteed—that one explanation for the correlation between spankings and depression and drug abuse later in life is that children aren’t spanked enough when they are young.

Therefore, there are at least two possible explanations for the correlation between being spanked as a youth and having problems later on in life:  either spanking is counterproductive or most parents do not use it effectively.  Thus, the conclusion of the research paper is more indicative of the researchers’ biases than of reality.

No comments:

Post a Comment