There is quite a bit of talk about this thing called Randomized Control Trials (RCT). Esther Duflo is championing this idea in her very popular TED talk ‘Social experiments to fight poverty’ (see first video below). The 2012 NYU Annual Conference ‘Debates in Development’) featured an opposing view on RTC from Angus Deaton (see second video below). He points out severe methodological problems with RTC. The third video is a Q&A with Deaton and Abhijit Banerjee (Duflo’s co-author). There is an interesting question about the ethical problem of purposefully not providing assistance to half of the control group.
Others have suggested that the ‘Trial and Error’ approach suggested by Deaton could be done in a participative way leading to what is called Participatory Action Research or Participatory Learning and Action (see Participatory Research (Paulo Freire) and Participatory learning (Robert Chambers)).