Piloting an experiment in the field

One of my final activities for the summer was assisting in a pilot research project using crowdsourced data. U-report is a UNICEF tool for gathering citizen feedback on what’s happening at the community level. Using this free SMS-based system, U-reporters send in both unsolicited information and responses to polls asking about service provision and soliciting other community information. Development partners and local leaders can then use this crowdsourced data to take citizen voices into account as they make decisions. The hope is that this will help complete the feedback loop between donors, government officials, and community stakeholders.

At the Jinja-Mayuge CDO workshop.

The pilot workshop on crowdsourced data begins for CDOs from Jinja and Mayuge districts.

Here’s where AidData comes in: we want to make crowdsourced data as usable as possible to enable evidence-based decision making by local government and development partners. In order to rigorously evaluate the systems used to collect and distribute crowdsourced data, AidData is interested in piloting randomized control trials (RCTs) to test and measure the effects and impact of giving crowdsourced information to local officials. For our preliminary pilot, we had a group of Community Development Officers (CDOs) from Jinja and Mayuge districts participate in lab-style experiment. Participants were randomly assigned data packets with and without crowdsourced U-report data. Both groups then completed an activity designed to approximate their normal activities, allowing us to compare responses from CDOs with and without a source of crowdsourced data.


CDOs work on the data activity using crowdsourced information from U-report.

There were many lessons learned from piloting an experiment in the field. One takeaway was that timing is key. Giving CDOs enough time to thoroughly examine the data packet was important to ensure they incorporated the data given into their decisions. Some participants took much longer to go over the data and complete the activity than others— it might be a good idea to allot a certain amount of time for each part of the lab. Overall, it was a great experience to see how research actually gets done in the field.


Stay tuned to AidData to find out how this project progresses.