Smoke and Mirrors: Slate’s Ryan Grim offers an excellent piece on the Bush administration’s evaluation of its own anti-drug advertising campaign. After spending hundreds of millions of dollars, the administration commissioned a piece of research that appears to prove that the ads actually increased the likelihood that young people would smoke pot. Or as the researchers put it, "greater exposure to the campaign was associated with weaker anti-drug norms and increases in the perceptions that others use marijuana."

Grim thinks the research shows that the campaign didn’t work. But I think he’s making a mistaken assumption about the purpose of the campaign.

Some big social marketing campaigns are an honest effort to address societal problems. Anti-smoking efforts like the “truth” campaign fall into that category, as do some of the AIDS education efforts underwritten by the states and by not for profit groups in the late 80s and 90s.

But a great many social marketing campaigns have a different objective: to convince voters that government is advocating “proper” behavior. Abstinence only sex education is a great example. Nobody seriously believes that abstinence-only education will reduce teen pregnancy or prevent the spread of STDs. That’s not the point. The point is to be seen by moralists to be delivering a “positive” moral message to the target audience. If members of the target audience ignore that message, the consequences are their problem.

The anti-marijuana ads are, I suspect, a similar case. They are not designed to discourage kids from smoking pot; they’re designed to make sure kids know that smoking pot is WRONG. So the government sat on the results of the study for 18 months—spending another $220 million on ads it knew were not effective—not because it likes wasting money, but because the money wasn’t wasted. Its supporters, particularly those who believe pot smoking is immoral, want the government to lecture people about the immorality of smoking pot.

The lecture is the point of the exercise, the results are irrelevant.