The headlines screamed the results of this Columbia University poll over the weekend:

“Here’s another reason to fear Facebook”

“New survey links drug abuse to Facebook”

“Are Social Networking Sites Turning Teens into Substance Abusers?”

“Facebook Linked to Teen Drug Use”

This annual poll by The National Center on Addiction and Substance Abuse has ruffled lots of feathers for a single reason: because most people don’t know how to read studies.

Here, the researchers have stumbled onto a correlation between drug use (tobacco, marijuana, alcohol) and use of social networking sites, and the general public has assumed – with the help of headlines like these – that the study is claiming that the use of Facebook is causing drug use/abuse. They’ve read the headlines and gotten scared into thinking that Facebook will cause teens to drink alcohol or smoke weed or cigarettes.

That’s not what the data says.

The results simply say that the two activities are correlated. Instead of saying, ‘Facebook causes people to use drugs’, we could say ‘Drug abuse causes people to use Facebook’, and it would be no more incorrect than the way it’s being reported.

The first rule in analyzing raw statistical data like this is to remember the maxim “Correlation does not imply causation“. Instead of assuming that one of the correlated things causes the other, we should consider the possibility that a third factor causes both of them.

In this case, might I suggest the existence of the ‘Strictness of Parents’ factor?

Social networking is a norm now among teenagers. Almost three-quarters of them are on Facebook, mySpace or something similar. So, my speculation – which would require further study – is that this study says more about the teens who are not using social networking sites than it says about those who are.

Could it be that a high percentage of the one-quarter of teens not using social networking are being raised by very strict parents who won’t allow them to join? Such teens would also be statistically the least likely to be using drugs, wouldn’t they? A poll targeting teens who are not members of social networking sites may also find that a high percentage of their parents are evangelical Christians, for example, or with higher-than-average representation of Asian-American parents of the ‘tiger mother’ mold; both demographics strongly disapprove of drug use and tend to separate themselves from the culture in which drug use is more common.

As I said, my hasty little theory would require actual study, which is what separates the story told by science from the story told by everyone else. In any case, there’s one thing for sure: this study has not proved that Facebook leads to a higher probability of drug abuse.