Social policy is laden with value judgments, and the same evidence can point to two contradictory policies depending upon the values of the person assessing that evidence.

Plus, when it comes to social policy, there tends to be great disagreement about what actually constitutes “evidence”.

First, as Mark says, even if there is agreement on evidentiary matters, there is still the matter of normative statements – “should”s – on which there will be considerable disagreement within the EBP.

Second, the much more complex nature of the social sciences compared with the physical sciences means the design and interpretation of experiments (not to mention the problems with data-mining for “natural experiments”) are much more complicated. It think it was Humphrey Appleby who offered four generic criticisms of any study that could be made without reading it.

Third, evidence is often used as window dressing for normative statements – this happens in every contentious policy debate from guns through crime to abortion. It’s the difference between using statistics as the building block of an argument, and using them to construct a facade.

The best test of how someone is using evidence is to ask, is the evidence were otherwise, would they change their minds? So if they took they cited evidence that “1 in 10 people are affected by bicycle theft”, and it subsequently emerged that 1 in 100 actually were, do they say “bicycle theft wasn’t as big a problem as I thought”, or “you are trivializing the hurt to bicycle owners with your fascist mathematics”?

The pro-legalisation campaigners split into two camps, the ’shoulds’ (who are doing it on basis of personal opinion, which is just as valid or invalid as the personal opinions of the bansturbators) and the EBP, who say that whether or not drugs are harmful or enjoyable is actually irrelevant – the key point is that making them illegal just makes things worse for everybody, not just those who enjoy taking them.