You should all read Ben Goldacre's latest Bad Science article. Ben is one of the leading science warriors of the day - laying waste to pseudofacts and junk science like a level 34 Paladin amongst kobolds.
The basic premise is that social policy - something governments spend hundreds of billions on - is very rarely tested with scientific methods - particularly with controlled, randomized trials. These trials would be easy to do, and not very expensive. They could tell us whether a social policy actually works. If you're going to spend hundreds of billions of pounds/dollars on untested social programs, doesn't it make sense to spend mere millions to see what effect they actually have? Of course, sometimes controlled trials are not feasible for practical or ethical reasons - we can't randomly assign innocent people to prison to see what happens - but we can randomly assign people already in prison to various rehabilitation programs and monitor the varying outcomes of these programs. As always, strict watchdogs will need to ensure the scientific testing process remains ethical, but I believe any risk associated with these trials is far outweighed by the existing danger of present programs. Take DARE, for example. After more than 20 years, many billions in government dollars, and implementing the program on 36 million children, the effects of the program were finally tested. Instead of reducing drug use, it was found the DARE program increased drug use (including alcohol, tobacco, and illegal drugs) and lowered students' self esteem. Great stuff, huh? Now that is unethical. We could have saved a lot of money and a lot of grief if the program had actually been tested at some point previously in those 20 years. What was the conversation like? "Hey Frank, we've used this program on 36 million kids, maybe we should see what effect it has?" "Yeah Bob, you may be on to something."
Ben makes the point that in the recent past doctors opposed using controlled trials to test drugs and medical techniques. When trials finally were introduced it was found that many drugs of the past were in fact useless (hey, remember blood letting?), and often caused horrible unnecessary side-effects. Now it is unthinkable to introduce a new medication without testing its effects. Why are we not testing social policies? They are just as universal and societally relevant as medicine.
The fact that we don't already test social policies is frankly bizarre. It's like we don't want to know whether these programs work or not. Like endorsing a social policy is a matter of faith rather than deliberation. Like we don't want to find out the experts might have been wrong all this time. Thanks, but I'll go for the controlled trials.