some posts… no shit

some posts maybe you are never ready to write. nevertheless, the news of marc roberts‘s death over the weekend seems to warrant both an immediate response and the response that is just right. he seemed to usually be able to manage these simultaneously but, given the sense of time that follows a completion of a life well-lived and well-said, i’ll err on the side of the former.

i won’t claim to have known marc as well as many but i knew him well enough to respect him, which means, perhaps, seeing past rough first impressions. he pronounced himself a reformed economist at some point early enough to influence me: politics and the realities of implementation and the curves of ethics-in-real-life became the subject of his writing and his teaching and we are better for it.

marc had a standard line — a bit of a trap — that he would lead you  into (funnier to watch others go than to realize you had followed in). you might make a comment; maybe even one you thought useful. then he would start. he grew up in jersey. [fill in a few lines about the roughness of growing up in a steel town in jersey.] they had a saying back then, he’d say, that would apply to the point you’d just raised.

no shit.

familiar and biting each time (after the first, which was less pleasant). what always made it ok was the sense that he was, and wanted you to be, in pursuit of the right questions. he raised questions of distribution when everyone else was looking at average treatment effects. he was a reformed economist when the economist profession was booming. he wanted to know about implementation when everyone was looking at theoretical equations. and wanted to know about practical theory when everyone was looking at the sexy result of the moment.

we were through “pinning butterflies,” i was told indirectly by marc. categorizing of treatments or results wasn’t what we needed — we needed to explain things and try to make sense of them.

and then to do better.

some posts you are never ready to write. but some some are scratched in before you even sit down to it and some give you a sense that you shouldn’t wait. with marc, the gist sank in early, so one doesn’t have to do much work to imagine he’s still around. which is quite a good thing.

we need his voice. it’ll be missed but, as with all good teachers, it, with its gruff accent, is hardly gone.

thank you, marc. (



data systems strengthening

i have been saying for some time that my next moves will be into monitoring and vital registration (more specifically, a “poor richard” start-up to help countries to measure the certainties of life: (birth), death, and taxes. (if village pastors could get it done with ink and scroll in the 16th c across northern Europe, why aren’t we progressing with technology??! surely this is potentially solid application of the capacity of mobile phones as data collection and transmission devices?).

i stumbled onto a slightly different idea today, of building backwards from well-financed evaluation set-ups for specific projects to more generalized monitoring systems. this would be in contrast to the more typical approach of skipping monitoring all together or only working first to build monitoring systems (including of comparison groups), followed at some point by an (impact) evaluation, when monitoring is adequately done.

why don’t more evaluations have mandates to leave behind data collection and monitoring systems ‘of lasting value,’ following-on an impact or other extensive, academic (or outsider)-led evaluation? in this way, we might also build from evaluation to learning to monitoring. several (impact) evaluation organisations are being asked to help set up m&e systems for organizations and, in some cases, governments. moreover, many donors talk about mandates for evaluators to leave behind built-up capacity for research as part of the conditions for their grant. but maybe it is time to start to talking about mandates to leave behind m&e (and MeE) systems — infrastructure, plans, etc.

a potentially instructive lesson (in principle if not always in practice) is of ‘diagonal’ health interventions, in which funded vertical health programs (e.g. disease-specific programs, such as an HIV-treatment initiative) be required to also engage in overall health systems strengthening (e.g.).

still a nascent idea but i think one worth having more than just me thinking about how organisations that have developed (rightly or not) reputations for collecting and entering high-quality data for impact evaluation could build monitoring systems backwards, as part of what is left behind after an experiment.

(also, expanding out from DSS sites an idea worth exploring.)