For my PhD dissertation I conducted a framed field experiment with a sample of 12th grade students studying in constituent schools of a large public university in the East Indian state of Jharkhand. The survey and experiment were funded by the TATA-Cornell Institute, where I am a research scholar. My field- work was a self-led effort and an enormous learning experience for me. Meeting the demands of collecting data of an uncompromising academic quality, in a setting which is resource-constrained and lacks formal channels for facilitating research, has helped me understand better the gap between development research on paper and on-ground. I now feel confident in my ability to lead large-scale surveys/experiments in the future and work with students and researchers who would like to do the same.
My experiment was in field for about 5 months, and we collected two rounds of data from 1525 students, across 9 schools in urban and rural districts around the capital city of Ranchi. My project-coordinator and I spent over a month collecting necessary permits, from school administrators, required to conduct our survey and ensure its smooth completion. I recruited and trained a staff of local survey enumerators who were recent graduates of the state university where we were working.
Even though none of them had prior survey experience and little professional experience, over the course of our work, the team grew tremendously as a well-functioning, goal oriented and committed unit. Some of the challenges we navigated together were, mobilizing student respondents in the face of low rates of school-attendance, re-organizing our entire survey schedule mid-way as classrooms (where we conducted our survey) were converted to polling booths with a sudden announcement of state elections and the mid-survey drop-out of a trained enumerator. In the end, we finished our survey on-time, well within budget, with a very low rate of attrition between rounds and high-quality data.
An important part of my learning process was fielding my survey tool on Open-Data Kit (ODK), a mobile data collection software, now commonly used in almost all large-scale surveys globally. Going digital massively boosted the productivity of the survey. I was able to easily incorporate changes in my survey instrument post “pilot”, build in constraints to ensure sensible survey responses, use data from the first survey round to dynamically update the second round questionnaire posed to each student, and aggregate my data and monitor it’s quality in real-time. Within school, we randomized students into survey sessions of 15 students each, with each session further randomized as a control or treatment session. Even though the survey was self-administered, because our instrument was largely based on eliciting subjective beliefs from respondents, survey questions were challenging and needed enumerator support. We had a team of two enumerators lead each survey session, guiding the respondents through the survey, with slides detailing specifics of each question. For this set-up, we used small, portable Pico projectors, portable screens and even portable batteries for power, since almost all the schools we visited had little stable electricity. We used the same set-up to conduct our information sessions, that were central to the experiment.
In the past, I have also conducted (exploratory) research in other parts of India- in Uttarakhand in the North and in the Vidarbha region of Maharashtra. For two out of three years of my undergrad I worked with a youth-led NGO in Delhi, The YP Foundation, and led their math and life-skills curriculum at one location, designed to transition out-of-school children to school. As a development researcher, I enjoy primary data-collection. The field, for me, is a rich source of research ideas and motivation to continue work in the development sector.
In the past, I have blogged about my top learnings from my survey-experience at the Economics That Really Matters Blog here.