As I’ve mentioned before, I’ve been busy running a survey in one of the many unplanned settlements of Dar es Salaam. It’s my first time doing this, so inevitably there are going to be hiccups. Some tips:
#6 – DON’T use props that might be eaten by hungry respondents. Part of the survey involves respondents tracing out probability distributions using beans or counters (i.e. place more counters here if you think this is more likely to happen). In the slums, we found that groundnuts worked as a cheap and readily available prop for this module. Yet only a day after the launch of the survey my enumerators came asking for more nuts, as the respondents usually favoured eating them over returning them to enumerator. Perhaps asking the poor to use food for anything else than eating is in the same moral territory as Chris Blattman’s soft drink torture.
#23- DON’T try and run a survey from a school. For several days I’ve met my enumerators inside the walled grounds of a local primary school. Yesterday, when I got in my car to drive off, a swarm of children crowded around. They followed me to the gate of the school, closed and defended by an irate guard with a large switch. As he tried to open the gate for me, I realised that the children were using my (white) 4×4 for cover to get past the gate and…. escape school! Several got away – so currently my biggest impact on the community has been to lower the primary enrollment rate.
#41 – DON’T try and design identity cards for the enumerators yourself, or, if you do, make sure you do a good job of it. My enumerators were not impressed with the 5×2 inch ‘super’ ids I created in Microsoft Word, especially the black & white printing. What made things much much worse was a large number of mixed up names with a few misspelling. The biggest gaff: accidentally changing a ‘Godbless’ into a ‘Godless.’
I’m certain there are more to come.


