Bad Social Science from Ben Goldacre

This blog is a great fan of “Bad Science”. But this week, Ben Goldacre has pretty well gone out of his depth when he dipped his toes in the murky pool of social science research.

He applauds what he believes to be the first randomised controlled trial in social research. So far so good. An excellent idea to apply scientific methods to social sciences…. He makes some reasonable specific points but, overall, he is completely missing the point.

For instance, he suggests there should have been randomised trials for Drug Treatment Orders as alternatives to prison, with prisoners allocated at random to one or the other. In this specific case, he shows a lack of social understanding that seems almost wilfull for a doctor.

Firstly, DTTOS were introduced because they are cheaper. Economists are constantly making decisions on pretty scientific grounds.

Secondly, they are more humane than prison. Almost anything is more humane than prison.

So unless DTTOs can be shown to actively increase levels of reoffending (unlikely, given the fair amount of evidence that no sentencing policies make much difference anyway) , they would have to be a better option, being both cheaper and more humane.

Thirdly, the idea of randomising sentencing is truly chilling. Sentences are supposed to reflect the nature of the crime and the situation of the defendent. Judges and probation officers are traine dto consider individuals and make hard decisions.

Most crucially, there is already a HUGE and ever-expanding body of scientific research in the social sciences. Results may be contradictory; some research is suspect in its purposes or flawed in its premises but methodologies are infinitely varied and it has a moral dimension. And so on. ( A bit like hard science really. )

Ben, there are libraries full of works that evaluate social policies. Didn’t they have them in your university? It’s not that the research doesn’t exist.

Social science research covers many issues where solutions aren’t limited to two easily testable alternatives. With regards to social policy, the application of research is completely constrained by social factors – the will of politicians, the vested interests of social groups, the beliefs of the public, the interpretations of the media.

The fact that results of mass social surveys – like the British Crime Survey – are misrepresented, ignored and apparently not even understood in Parliament, shows that, even when there is evidence, it may be the last thing that’s important when it come sto policy-making.

Goldacre doesn’t seem to realise that social policy is created through social action, not imposed by experts. This might be a bad thing in many cases, but as one commenter observed, a random controlled experiment in fascist politics might indeed make the trains run on time – but you would hope any such project would be constrained by people’s awareness of other issues. Social science research has to have ethical concerns at its centre. But social scientists can’t be trusted to make all the moral decisons.

I know little about medical research. Well, only marginally more than nothing, to be honest. But I suspect that a fair bit of it would benefit from some awareness of the interplay of social forces. Medical resarch is not in itself ethically neutral.

If doctors don’t understand all the social complexities of the funding of research, the role of drug companies, the sourcing and testing of ingredients, the political economy of healthcare, national and international inequalities, social priorities, and so on, ad inf – well that’s what social science is for,