Randomized control trials thrill some, and frighten others. In the right conditions, they are by far the best way of finding out if a program has an impact on child outcomes.
For a long time there were too few trials or experiments, as they are also described, for policy makers to be able to draw any meaningful conclusions. Now, as Robert Boruch described on Day 1 of the SRCD conference, there are over 15,000.
In the United States policy-making for schools, welfare to work programs and increasingly other areas of child welfare draws heavily on experimental evidence. If a program is going to be funded, there better be some robust evidence to show that it works.
This means that more randomized control trials are being funded, and this means more scientists are interested in seeing how the method can be used to tell us more about child development.
The first advance is to go from 'does this program work?' to 'why does this program work?' In their most basic form, a trial will have a group of kids at one end of the equation. Half the kids, selected at random, receive an intervention, such as a parenting program or a mentor. At the other end of the equation is the result; how much better are the outcomes of children receiving the program than those who did not?
But the question is not only did the program work, but why did it work. What did the parenting program, or mentor, or simple attention given to the child or family by virtue of getting more help do to make the difference?
The second advance is to answer basic questions about child development from experimental trials designed to test program efficacy. One of the workshops on Day 3 was devoted to this issue.
One the one hand, the sophistication of the methods used by US scientists is breath-taking. On the other, one worries that the complexity of the approach will begin to hide the simplicity of the message required for policy makers and practitioners. One also worries that there will be an anti-body reaction in the worlds of science and policy to the amount of experimental work underway.
In many ways, simpler models, well tested, and well communicated might become the order of the day.