Skip to content Skip to navigation

Causal Inference for Social Impact Lab

Several structural barriers make it difficult for government officials and academic researchers to engage collaboratively in evidence-based policymaking. These include a lack of common language for discussing programs, limited methodological strategies for establishing causation without randomized control trials (RCTs), and data-sharing restrictions. The Causal Inference for Social Impact Lab (CISIL) finds solutions to these barriers and enhances academic-government collaboration.

The Lab is led by 2018-19 CASBS fellow Jake Bowers, 2017-18 CASBS fellow Carrie Cihak, and CASBS program director Betsy Rajala.

CISIL has received funding from SAGE Publishing, the Knight Foundation, and the Alfred P. Sloan Foundation.

For more information, please contact CASBS program director Betsy Rajala (Betsy.Rajala@stanford.edu).


Data Challenge

The Causal Inference for Social Impact Lab (CISIL) at the Center for Advanced Study in the Behavioral Sciences (CASBS) invites applications from teams interested in participating in the CISIL data challenge.

The CISIL data challenge has some exciting attributes that differentiate it from Kaggle-style and Atlantic Causal Inference Conference-style challenges:

  • You will use real administrative data on transportation and demographics from King County (Seattle), Washington.
  • The data challenge questions are generated by King County policymakers and are typical of the kinds of questions in evidence-informed public policy evaluations that are growing in prevalence in the USA and around the world.
  • Analyses from data challenge teams will be used by policymakers to advance transportation equity.
  • The data includes a non-randomized study for which there is no one known correct answer.

That is, this is a Data Challenge, not a competition. We are aiming to simulate the kinds of analytic challenges that are typical of real world policy evaluation. Utility of your analyses and responses to decision makers is as important a feature of submissions as the statistical operating characteristics of the estimators and tests used for causal inference.

After data challenge analyses are submitted, we intend to write a paper to which teams may be invited as co-authors about how expert data analysts differ in their approaches to answering the same questions.

Learn more or apply to participate