Shortly after the recent AidData/CCAPS/WBI event, the World Bank’s Shanta Devarajan reviewed four critiques of the open data movement. His summary is insightful and we commend it to readers. Here we would like to push a few of his ideas a bit further. In particular, his third point — “there is limited evidence that information improves outcomes” — deserves further consideration.
The evidence that open data leads to better outcomes is indeed limited and contradictory. For example, one innovative study found that community-level information dramatically improved health outcomes in Uganda, while a concurrent Indonesian study found that an informed public had little effect on perceived government corruption.
Dr. Devarajan writes “the main reason for the paucity of evidence on outcomes is that the underlying service delivery failure is political, and it is difficult to disentangle the effects of information on the political system.”
We agree. In many developing countries, domestic political barriers likely "short-circuit" the relationship between better information and better development outcomes. After all, how can informed citizens get better development outcomes if their demands go unheard?
Two studies are under construction at AidData that will address this issue head-on. These randomized control trials will introduce interventions that (1) solicit citizen feedback on development outcomes and (2) take the feedback directly to the desks of relevant policy-makers.
In the first experiment, we plan to test how well user-level feedback can improve a government’s delivery of needed services. In certain treatment groups, the user-level feedback on service shortages will be sent directly to the government officials in charge of distribution. By comparing outcomes from this group with the control-group results, we will see whether direct information delivery can significantly enhance outcomes. In the second study, user-level feedback on specific projects will be sent directly to the donor-level project managers.
The purpose of these experiments is to learn whether, when, and how user-level information can be used to improve service delivery among both governments and donors. Our hope is that champions of the open data movement will use – and expand on – our results to sharpen their strategies for using user-level aid data.
Information is power, and it may improve development outcomes – if applied correctly. Political barriers are not insurmountable, but innovative approaches to using open development data will be essential. Our latest work will take several steps in this direction. We urge others, especially at the user level, to take on this important question: How can we make open-source data really work for the poor?