What happens if you are running a conference and a speaker doesn't show? Don't panic: just let one of the other star turns - in this case Professor Ann Oakley - run on a bit, and then fill in the gap yourself.

So it was with David Torgerson, Director of the York Trials Unit, on the second day of this year's conference when he filled in with a spirited ad hoc history of randomized controlled trials in the social policy world. Ann Oakley herself had given us a healthy injection of history, recalling work in the 1960s by Tilda Goldberg - with a foreword by her father Richard Titmuss - and an early systematic review by Barbara Wootton. Ann Oakley's working life involved a spell with Iain Chalmers, instigator of Cochrane Reviews, who was responsible for the revolution in medicine that some of us seek for children's services.

There is no definitive history, but Torgerson considers an experiment to 'get out the vote' in Chicago in 1924 to be the first RCT. It is striking that although early random allocation studies seldom mention power estimates, samples tended to be much larger than in today's trials. Torgerson gave an example of work by Downing and Jones, published in Educational Research in the 1960s, that included over 1,000 children. Few RCTs have this kind of sample size today.

History is exciting but it can also be depressing. Richard Clark Cabot evaluated the Cambridge Somerville study in Massachusetts. It was one of the most exciting and innovative preventative programmes for antisocial behavior. But it found relatively few effects – and the ones it did find all favored the control group, who committed fewer crimes, were less likely to become alcoholic and more likely to get a good job.

Some members of the audience were able to recall how resistant the medical profession had been to the idea of experiments in the 1960s. Now few dare to object. At the time, it was agricultural scientists who had the expertise in RCTs, and some moved to medicine. It is tempting to think that today's health scientists might extend the process and move to the world of child development and children's services.

But, as Ann Oakley pointed out, some of the residue of the language of medical trials is potentially more of a hindrance than a help. Does it make sense, she asked, to talk about 'intention to treat' or 'dose response' in the context of a study about the effectiveness of programs to improve adolescent health? Curiously, Tilda Goldberg's classic study Helping the Aged made no mention of RCTs, but that’s what she carried out. She called it a 'field experiment in social work'.