One positive side effect of the credit crunch has been to concentrate minds on accountability. Evidence-based initiatives proliferate and are increasingly sought after partly because, by definition, they can make a good account of their performance.

So it’s good news for prevention science, but it means policy makers face new responsibilities. They are accountable, too; how are they to choose reliably from the growing stockpile of “proven” programs?

The trouble is that the databases that are being compiled to make those choices easier and more acute have so far tended to apply different standards. Even a trained researcher who lives and breathes the data is likely to be defeated by the variability if not the sheer volume of information.

Enter the US Society for Prevention Research (SPR) which in 2004 resolved to define a set of standards that would help people navigate the maze. 

SPR has since set out three levels of evidence - efficacy, effectiveness and dissemination. 

Efficacy is the bedrock. It refers to the “beneficial effects of a program or policy under optimal conditions of delivery”. In other words, it indicates how well it works if it has good support and is monitored by the developers. 

In the case of a school-based bullying program, for example, it would be delivered in a conditions where teachers and management give it their full support, where resources are adequate and the research team is able to check it is being implemented as instructed. 

As to standards for efficacy, the SPR stipulate that in order to make the grade programs should be tested in at least two rigorous studies, ideally involving randomized controlled trials. Objectives and target groups should be clearly described and outcomes properly measured. Impact should be assessed using trustworthy tests.

It is not sufficient, for example, merely to ask children if they have stopped fighting as a result of taking part in a bullying program. Psychometrically-tested measures that academics will have spent years refining provide more accurate and more useful data. 

Another necessary measure deals with enduring impact. The benefits of an intervention must still be detectable at least six months down the line. 

Causality is another of SPR’s fundamental concerns. Research must be able to show that the program is the cause of any beneficial change and that it could not have happened anyway. The key to this calculation is access to a control group or some equivalent basis for comparison, preferably with random assignment to groups.

So ten-year-olds who receive advice about improving their relationships may get on better a year later, but it might be as a result of their simply growing up or getting to know each other. Without comparing them to a similar group who have not received the help, it will be impossible to tell.   

Being able convincingly to claim that an intervention has an impact on certain outcomes calls for a solid research design based on sound statistics. A randomized controlled trial will be the most robust but PRC standards allow others that can make a similarly convincing case for causality. 

There is no excuse for hiding results. The standards demands that there are reports on all outcomes measured, good bad or indifferent.  

Finally, for a program to demonstrate its potential, others must be able to replicate it. It must be able to offer a manual, materials and training. 

It is already a stringent list but it only amounts to an overview of what SPR standards require. For researchers, it raises the bar quite significantly and requires them to ask awkward questions of their own work. For policy makers it could be a godsend - enabling them to choose among interventions more confident that like has been convincingly compared with like.