If someone has invented a successful, innovative, cost-effective social program, doesn’t it seem likely that it would spread quickly to other communities? Susan Evans and Peter Clarke have written a fascinating article detailing why many programs become “orphan innovations” that no one else adopts.
The authors describe a program started by a retired produce wholesaler in Los Angeles, who convinced distributors to donate slightly spoiled produce to food banks. Before long, poor families were receiving fresh fruits and vegetables that would have been dumped in landfills. Evans and Clarke took it upon themselves to make sure that this program was adopted in other cities, but ran into many roadblocks, such as skepticism from overworked local officials that the program would work. Eventually, through sheer determination, they succeeded: the program spread to dozens of communities. But it took 20 years of hard work, creativity, cajoling, and financial support.
Their conclusion? A social program cannot simply be transferred from one locale to another. Instead, it has to be customized at each new location. Unlike a fast food chain, that plops a carbon copy of a restaurant down in every community in American, social programs have to be adapted to the particular staff, clients, and ecology of each setting.
There are valuable lessons to be learned here for those of us interested in social psychological interventions that improve human welfare. There is a growing movement to translate social psychological theory into interventions that help people in the real world, including ones that help people recover from traumatic events, prevent child abuse, reduce adolescent behavior problems, and close the achievement gap in education (as I chronicle in my book Redirect). Critically, these interventions are being tested with well-controlled experiments, to see if they work. This is a huge advance over relying on common sense, which has led to the wide-spread adoption of programs that don’t work or do harm (see my earlier post, Testing, Testing).
