Chris Kennedy

CONTACT INFORMATION
Office:
Barrows 350A
Chris Kennedy's picture

Graduate Students

Personal Statement: 

Chris Kennedy is a third-year PhD student specializing in methodology and American politics. His research interests include voting & elections, interest groups, and climate change politics, as well as precision medicine. He develops and implements methods for randomized trials, machine learning, text analytics, and survey research. He is affiliated with D-Lab, Alan Hubbard's trauma group, the Institute of Governmental Studies, and the Yale Project on Climate Change Communication, and is a member of the Analyst Group.

Prior to coming to Berkeley, he spent 7 years in DC working with progressive organizations on campaign field experiments, predictive models of political behavior & attitudes, and advocacy using digital technologies. He has worked for the Voter Participation Center, Rock the Vote, Al Gore's Alliance for Climate Protection, and Avaaz, and continues to consult as a data scientist for a variety of organizations. He holds a master of public affairs and a B.A. in government & economics from the University of Texas at Austin.

SELECTED WORKING PAPERS

  • "The Political Consequences of Violence" 
    • With Amy Lerman. Presented at Law and Society Assocation's 2016 annual meeting
  • "Registration, Turnout, and Residential Mobility in the  2012 U.S. Elections: A Field Experiment"
    • Presented at West Coast Experiments Conference 2016
  • "Learning American Climate Activism."
    • Presented at MPSA 2015.
  • "Impacts of Personalizing Election Administration: Mobilization via Requesting Mail Ballots by Phone."
    • With Christopher B. Mann. Presented at MPSA 2014.

SOFTWARE

  • varImpact - R package to estimate variable importance using causal inference statistics (targeted minimum loss estimation - TMLE). With Alan Hubbard. View on Github
  • Randomize.ado - Stata random assignment algorithm for clinical trials, including balance checking, blocking, and automated rerandomization for optimal covariate balance. View on Github
  • Htestimate - R package to calculate unbiased treatment effect estimates for RCTs when assignment probabilities are correlated across units, such as under clustered designs or rerandomization. With Joel Middleton. View on Github

ACADEMIC ADVISORS:
Joel Middleton
Alan Hubbard
Jasjeet Sekhon
Amy Lerman

Primary Subfield: 
Methodology & Formal Theory
Secondary Subfield: 
American Politics