Reflections on Managing a Data Lab in Haiti
By Castelline Tilus
In a new blog series on staff perspectives, TReNDS’ new Research Manager, Castelline Tilus, describes her experiences working in a data science lab in a Global South context.
Originally trained in international policy and development, my interest in data science began in 2017 after conducting years of survey research in Haiti under the guidance of Dr. Mark Schuller, a professor of Anthropology and Nonprofit Studies at Northern Illinois University and an affiliate professor at the State University of Haiti. As part of my field research, I worked to assess public views of state and non-for-profit institutions in the South of Haiti to determine their perceived levels of effectiveness in addressing community identified problems. This project awakened a dormant interest in using data for development, particularly in using data to better identify community needs and to monitor progress towards addressing those needs.
My experience conducting survey research was an integral part of what led me to co-found a data laboratory, Ayiti Analytics, in Haiti in 2020 to service the analytical needs of governments, businesses, and international non-profit organizations in the country. Headquartered in the capital city of Port-au-Prince, Ayiti Analytics’ core mission is to utilize data and technology to improve human life, and during my tenure at the lab, I collaborated with stakeholders from diverse backgrounds to conduct data-driven research for the public good. For instance, in our inaugural year, Ayiti Analytics conducted a data-driven analysis of Haiti’s healthcare system. We worked in partnership with public health professionals and Haitian epidemiologists to apply machine learning models to rank the health system capacity of different regions in Haiti to respond to public health emergencies, and this work proved to be an essential public resource during the COVID-19 pandemic.
Data Challenges and Opportunities in Haiti
There are a number of challenges to fully harness data for the public good in a Global South context. In Haiti, in particular, roadblocks to data production are varied, but include a lack of human and financial resources for data collection and inadequate digital infrastructure for collecting greater volumes and variety of data, especially within the public sector. Furthermore, digital transformation is still in the nascent stage. Along with relatively low levels of internet penetration (41% in Haiti compared to the regional average of 68%), there is a lack of interconnectivity between and within public entities, hindering the ability of public institutions to provide online services to the population. Additionally, non-traditional digital forms of data are few and far between for higher-level engagement and analysis among institutions.
This has resulted in a heavier reliance on traditional sources of data for decision-making within the country, including the population census and nationally representative surveys collected by international development organizations and businesses for operations and program monitoring purposes. The extent to which these data are made publicly available varies from institution to institution, which has further inhibited data use in Haiti. Take for example the case of Ayiti Analytics. During the COVID-19 pandemic, the team wanted to examine disparities within the existing healthcare system. But we initially struggled to find data to conduct the analysis. Fortunately, a public health professional referred us to the Demographic and Health Survey (DHS), a USAID-funded program that collects, analyzes, and disseminates data on population, health, and nutrition in over 90 countries, enabling us to proceed with our analysis and subsequent publication. However, we were unable to find a comparable dataset with information on Haiti’s 1,033 health facilities across varied residential, commercial, rural, and urban settings collected and disseminated by the Haitian government. Additional roadblocks to data use in Haiti include diminished trust in government data and a lack of awareness regarding the potential impact of data across industries, among other challenges.
Despite the many challenges, they also present opportunities to further digital development in Haiti by addressing these key blockages to data collection, analysis, dissemination, and use. For instance, Haiti has a growing mobile-connected population, with 60 mobile phone subscriptions per 100 inhabitants. Haitians are also increasingly transacting online through mobile phone-based wallets. As such, telecommunications companies and digital financial service providers can partner with government entities to open source data for public use. This is a largely unexplored territory within the country, but it is an area that Ayiti Analytics is working to advance through individual consultations with private sector partners. Additionally, by facilitating computer usage and encouraging local students to develop technology skills, Ayiti Analytics has graduated 26 students from its flagship Port-au-Prince Data Science Bootcamp and endeavors to fill the remaining resource gap within public and private sector institutions in the country. There are also a host of other actors within Haiti’s technology ecosystem that are working to address these challenges, including ICT universities, emerging tech hubs, and incubator programs for technology-focused startups.
Lessons Learned for Future Work
The country-level work in Haiti provides useful context for assessing global challenges and opportunities for advancing the data for development agenda. Although problems are not perfectly transferable across countries, their root causes may be. My experience in Haiti has taught me the importance of developing multi-stakeholder partnerships - including with the public and private sector - to build a robust data science community. I have also learned the importance of filling existing skills gaps amongst future generations of leaders to address ongoing data literacy challenges in the long-term that are often widespread in low and middle income countries. I hope that the knowledge gained from the past five years of field experience will allow for in-depth comparative analysis where applicable to address data production or data use challenges in similar country contexts in my work with TReNDS.