Mariella Hansen

MIT Department: Abdul Latif Jameel Poverty Action Lab
Undergraduate Institution: Emmanuel College
Faculty Mentor: Anja Sautmann
Research Supervisor: Evelina Smirnitskaya
Website: LinkedIn

2019 Research Poster

Biography

I am an international student from Denmark. I am a rising junior majoring in Economics and Mathematics at Emmanuel College. The importance and relevance of economic-related activities to the modern world have led me to pursue the study of economics at an advanced level. I want to have a direct impact on the field of economics, both during and after graduate school; I hope to help to create a more egalitarian world through research, intercultural understanding, and respect. My interests include reading books, traveling and hanging out with friends.

2019 Research Abstract

Adaptive Treatment Assignment in Experiments for Policy Choice

 Mariella Louisa Hansen1, Sarah Kopper2 and Anja Sautmann2
1Department of Business and Economics, Emmanuel College
2Jameel Abdul Latif Poverty Action Lab, Massachusetts Institute of Technology

Governments in developing countries, international aid agencies, and economists agree that an educated populace is necessary for long-run economic growth and a high standard of living. Given the tight fiscal constraints under which developing country governments operate, a key issue for researchers has been to identify the most cost-effective ways of achieving positive education impacts. This intervention seeks to identify the optimal way of improving student learning outcomes by increasing parental engagement. We test which version of different weekly voice calls to parents most improve student learning outcome, measured by math test scores and attendance. An adaptive approach was developed to assign the subjects to treatment arms through which an algorithm was optimized to find the highest-performing treatment as quickly as possible across several terms in Indian primary schools. The algorithm reduces the number of subjects in the lower-performing treatments in each consecutive wave and increases the number of subjects in the higher-performing treatments. The optimal treatment and stratified information will be used to calculate average performance under each treatment, statistical uncertainty about this performance average, and expected gains from further experimentation. The implementation of this adaptive experimental procedure has the potential to improve policy choice in many different geographical, financially constrained, and field-specific contexts by reducing sample sizes or optimizing the extractable knowledge from a given sample.