To field test how machine-learning algorithms compare with traditional techniques for estimating heterogeneous effects in behavioral experiments

  • Amount $189,873
  • City New York, NY
  • Investigator Josh Wright
  • Year 2018
  • Program Economics
  • Sub-program Economic Institutions, Behavior, & Performance

Funds from this grant support research by Josh Wright, working in concert with economists Sendhil Mullainathan of the University of Chicago and Susan Athey of Stanford, to test innovative new machine learning techniques in economics field experiments. The group intends to investigate whether machine learning can improve randomly controlled trials in two ways. First, can machine learning enhance the assignment of subjects to control and treatment groups in ways that can lower necessary sample size without sacrificing rigor? Second, can machine learning techniques expand our ability to identify and analyze heterogenous treatment effects? Wright and his team will deploy state-of-the-art machine learning techniques in a series of actual economic field experiments and then share their findings via conferences, talks, and papers.

Back to grants database