Recently NYU held a conference entitled "What Would the Poor Say? Debates in Aid Evaluation". You can find many of the ppts of the talks here at William Easterly's new blog "Aid Watch".
I've only looked at a few of them, but I thought I'd highlight two of them who take fairly (although not entirely) contrasting positions. First is Esther Duflo's talk that focuses on the benefits of experimental design for improving aid interventions. The other is William Duggan's short piece arguing against just using statistical evaluation. He suggest two other methodological techniques: historical political economy and qualitative investigation. What I find interesting about both of these is that ultimately they are both asking the same question: what works? but with a very important difference in terms of what the explanation looks like. In other words, they are both looking to test and generate theory, but theory of very different types.
I think Chris Blattman's notion of Evaluation 2.0* nicely captures the means that these two seemingly contrasting approaches can be combined, namely by defining explanation as the identification of the working causal mechanisms. The answer to what works? is the itetification of the active causal mechanism (ideally the intervention), its structure, what its causal tendency is, and how it interacts with particular aspects of the context. In other words, relevant research question is: what works (or not), how, for whom, and in what contexts? From that we can move to the methodological design that provides the best possible answer given the research opportunities the intervention affords.
I've only looked at a few of them, but I thought I'd highlight two of them who take fairly (although not entirely) contrasting positions. First is Esther Duflo's talk that focuses on the benefits of experimental design for improving aid interventions. The other is William Duggan's short piece arguing against just using statistical evaluation. He suggest two other methodological techniques: historical political economy and qualitative investigation. What I find interesting about both of these is that ultimately they are both asking the same question: what works? but with a very important difference in terms of what the explanation looks like. In other words, they are both looking to test and generate theory, but theory of very different types.
I think Chris Blattman's notion of Evaluation 2.0* nicely captures the means that these two seemingly contrasting approaches can be combined, namely by defining explanation as the identification of the working causal mechanisms. The answer to what works? is the itetification of the active causal mechanism (ideally the intervention), its structure, what its causal tendency is, and how it interacts with particular aspects of the context. In other words, relevant research question is: what works (or not), how, for whom, and in what contexts? From that we can move to the methodological design that provides the best possible answer given the research opportunities the intervention affords.
* This notion of a causal mechanism has been around a while in the phil. of social science, but Blattman's example is nice and clear. This type of theory is applicable at both the macro and micro level.
