02 February 2013

Information IS Power (at least in this study area): New paper from Bjorkman et. al.

Community based monitoring of the quality of service delivery is certainly not a panacea.

But some studies have shown that some interventions have been successful in improving outcomes, solely through the introduction of some kind of feedback mechanism.

One of the highest profile studies was by Bjorkman and Svensson in 2009. The study was prominent because it was in one of the top economics journals (the Quarterly Journal of Economics), because it was a randomised controlled trial (RCT) and because of the large effects a community scorecard that publicly rated health facility performance had on public health outcomes.

In fact the effects on health outcomes were so large that it was one of the studies commissioned by 3ie for a reanalysis of the data (no results yet as far as I know).

In a new paper, Bjorkman-Nyqvist, de Walque and Svensson have revisited the 2009 study and have explored two questions: (1) do the effects of the original participation + information intervention hold up in the longer run? and (2) is it the programme design or the context that drives the positive results?

The authors find that the long run effects of the original intervention hold up. Some of the effects are larger (for reducing births), some are similar (reductions in infant mortality, impacts on low weight for age). This is encouraging.

On whether it is context or programme, they can begin to untangle this. They introduce a participation-only intervention and compare it to the control and to the original participation + information intervention. In the participation + information intervention, communities don't just provide feedback on providers, they have access to provider scores.

It turns out that community access to provider scores is vital for improved health outcomes. Without the information on provider scores, participation is disempowered because the accountability mechanisms are absent.

So some aspects of programme design matter a great deal. On whether this effect would be seen in another location, the paper cannot say.

Simplistic notions of participation are discredited by this RCT, which illustrates the importance of community-powered accountability, at least in these 9 districts of Uganda.

Note: The paper I have a hard copy of was dated January 13, 2013 and the paper publicly available is dated March 2012 (although it does not have the results tables included). The conclusions of the two versions are the same.  I have written to the lead author to get permission to publish the January 2013 paper. 

2 comments:

Jane Sherman said...

Many thanks for alerting us to the follow-up paper. The first one was inspiring and this one makes the strategy even more convincing.

Benjamin Wood said...

Thanks for recognizing 3ie's Replication Programme. As a quick point of clarification, Bjorkman and Svensson's influential 2009 publication is on our list of replication eligible papers, but we did not receive any applications to replicate it for our first round of replications. We are planning on opening our second Replication Window for applications on March 1st. Interested researchers are encouraged to apply to replicate this paper or any of the other important impact evaluations on our replication eligible list. More information about 3ie's Replication Programme is available here: http://www.3ieimpact.org/evaluation/impact-evaluation-replication-programme/.

Please stay tuned,
Ben (3ie's Replication Post-Doc)