4 min read
Banerjee, Duflo, and Kremer winning the Nobel prize for Economics has been a triumph not just for empirical scholarship on poverty, but for an entire sector that strives to use data and evidence to effect policy change and social impact. The spotlight on the indispensability of robust public systems data has been reinstalled, and this could prove a turning point in achieving development targets and goals using evidence.
To generate large scale data-led transformation, however, we need to carefully reevaluate the nature and quality of data being used as well as the infrastructure to harness it. A question that needs addressing is whether we are foundationally equipped to scale data-backed processes and have a culture in place to enable it at all levels.
The challenge with building a foundation for data
Harnessing data from public systems is often challenging due to structural and resource constraints. A typical example of this is in public hospitals where record-keeping and data monitoring systems become exceptionally difficult to implement.
Take for instance the partograph, a graphical recorder that monitors the progress of women’s labor during childbirth to prevent any complications. This can be an immensely rich source of data on maternal health. However, it is rarely used in rural hospitals despite the World Health Organization’s attempts to promote it. Nurses and doctors find it difficult to carry out such labor-intensive and time-consuming tasks in addition to their existing workload and are not always given the requisite training to utilize the tool.
Therefore, while data can be instrumental in decision-making, resource and behavioral challenges make it difficult to establish the groundwork needed to inform change at the highest levels.
Limits to alternatives
Difficulties in accessing robust data from public systems are pushing development practitioners to find alternate sources. Criticism by academics over the integrity of public statistics and the government’s tendency to suppress uncomfortable data have added to this shift.
As an alternative to surveys, data from satellites, IoT, and mobile phones is increasingly being utilized as a proxy for behavioral indicators. For example, satellite data on mobile phone locations is being used to study rural-urban migration and the spread of infectious diseases; the types of roofs in homes are being used as a proxy-indicator for poverty; research using online search data to study migration flows is also underway.
Such studies and methodologies are undoubtedly novel and can prove instrumental. However, one cannot treat them as a replacement for primary, survey-backed behavioral insights from the grassroots, which are ultimately key in making impactful decisions for policy makers and development entities alike. Robust statistics and data points, ethically and meticulously sourced, must serve as the foundation for advanced analytics such as AI and machine learning which are gradually gaining traction within development and policy circles.
Call to Action
To push for greater usage of accurate and reliable data, we need to build innovative systems for data collection and analysis. Such innovations could include gathering primary data in real-time or leveraging remote data collection interfaces that are user-friendly for decision-makers. Going back to the case of the partograph, if rural hospitals can utilize electronic or mobile devices for data entry instead of graphically plotting it, human effort will be minimized and adoption of partograph monitoring will increase.
Building fundamental data infrastructure calls for a concerted effort from a host of institutions, including tech innovators, researchers, development practitioners, and public bodies alike. Impact investors will also have to start viewing data as a cause worth funding.
Above all, we need to initiate a shift in thinking by installing a culture for data and evidence across all levels. Technology innovation designed with development goals and requirements at the center will prove to be pivotal in facilitating this.
