Last year, we published our Rethinking Evaluation blog series, which generated a lot of reader interest. Over the past 18 months, people have asked me why I wrote the Rethinking Blog Series. In particular, did I have in mind that it would evolve into the global consultation process recently commissioned by the DAC Network on Development Evaluation.
I had – mistakenly – assumed that the rationale for the “Big 5” evaluation criteria was well understood. After so many years of use, it became almost second nature to use these criteria, as opposed to other options, in our evaluations. [Recent] discussions revealed that while the DAC evaluation criteria were indeed well known, the underlying rationale for using them was not.
We need to develop methods to assess whether individual interventions produce more together than on their own. That is necessary already in the current context when focusing on country programs, but will be needed even more in cases of "aid-trade-diplomacy" or for the SDGs where some are mutually reinforcing and others will require tough trade offs.
I was pleased to hear that the DAC Secretariat was committed to revisiting the criteria. Not to start from scratch but build on the successful foundation that the criteria had established. But, also in recognition of the need for change in changing times.
The growing interest in strengthening development outcomes has stirred increasing debate about evaluation effectiveness. Today, many development institutions subscribe to what has come to be known as the DAC evaluation criteria. Specifically, these are five criteria – relevance, effectiveness, efficiency, impact, and sustainability; in short R/E/E/I/S – that underpin most evaluations in international development.
Taken together these dimensions of sustainability – economic, fiscal, environmental, and social – are complex. It will be hard and costly to try to address them systematically in all evaluations. At the same time, we evaluators cannot afford turning up with empty hands and concerns about missing data.