Thursday, 30 August 2012

Participatory Statistics II

Months ago I promised I would write more about a riveting conference I attended in May 2012, "The Rigour and Potential of Participatory Statistics". (Granted, the title sounds a bit clumsy, but if you don't mention "rigour", you don't get anywhere, nowadays ;-) Unfortunately there seems to be no conference documentation (apart from the invitation) on  the web. But I have my notes!
 
The conference was hosted by SLE (the development studies centre based at Humboldt University in Berlin), and the German Evaluation Association (DeGEval). I particularly enjoyed the interventions by Robert Chambers and Carlos Barahona. They demonstrate that "participation" and "statistics" go together very well. 

Robert has reminded us of the wealth of methods that bring "quantity" into "qualitative" research. People don't need to have any statistics or even literacy skills to quantify their observations and preferences. In participatory rural appraisal, seeds or pebbles can be used for counting and weighing different options. Some of the most common techniques are: 
  • Proportional piling, explained on this FAO page,
  • the "ten seeds technique", described, in great detail, by Ravi Jayakaran here, and
  • matrix ranking - see for example Robert's notes from Kenya and West Bengal.
These exercises can be replicated with large numbers of groups and produce robust data.
Talking about robust data, the focus of Carlos Barahona's presentation was questioning numbers. Which is remarkable, because Carlos is the Deputy Director of the Institute of Statistics at Reading University. His advice to development professionals: "Find a specialist in statistics and tame him or her", to make sure they build into their work (1) attentive listening, (2) opportunities for weaker voices to be heard and (3) self-criticism. Carlos points out that statisticians can deal with sampling errors (by increasing the sample). But "non-sampling errors" such as imprecise measurement tools can cause equal or even more damage. Participatory mapping exercises (such as those listed above) are excellent ways to create sampling frames that reflect local realities. Participatory work can also help to reconstruct absent theories of change (many projects start without an explicit description of their logic), or make up for absent or incomplete baseline surveys.

To really understand policy and its effects, qualitative in-depth research "on the ground" is needed. It's true that pressure and deadlines don't make this easy. But then maybe something can be done against pressure and unrealistic deadlines...

No comments: