Picture of Kyle


87 Trumbull Street, Room C321
Institution for Social and Policy Studies
Yale University

Contact me

Office Hours Sign-up

Working Papers

  • Community Policing and Police Legitimacy: A Field Experiment

    Repeated instances of police violence against unarmed civilians in the United States have drawn worldwide attention to the contemporary crisis of police legitimacy. Community-oriented policing, which encourages officers to engage in positive, non-enforcement contact with the public, has been widely promoted as a policy intervention for building public trust and enhancing police legitimacy. To date, however, there is little evidence that community-oriented policing actually leads to changes in attitudes towards the police. We therefore conducted a randomized trial with a large urban police department. We found that positive contact with police—delivered via brief door-to-door non-enforcement community policing visits—substantially improved citizens' attitudes toward police, including legitimacy and willingness to cooperate. These effects remained large in a 21-day follow-up, and were largest among non-white respondents.
  • Do Survey Measures of Racial Prejudice Predict Racial Discrimination? Experimental Evidence on Anti-Black Discrimination

    with Gregory Huber. Under revision.
    Scholars regularly measure whites' racial attitudes using symbolic racism and, more rarely, overt prejudice. We examine the predictive power of both measures in explaining anti-black discrimination. In Study 1 we obtain a behavioral measure of racial discrimination using the Ultimatum Game (UG). We find that white responders engaged in costly discrimination against black proposers by rejecting offers they would otherwise accept from whites. Overt prejudice predicts which whites discriminate whereas symbolic racism does not. In Study 2, white third-party observers evaluate intergroup interactions in the UG, and overt prejudice predicts racially biased evaluations of the fairness of resource distributions made by black proposers to white responders, but symbolic racism does not. Finally, we re-analyze a published candidate choice experiment and find that overt prejudice predicts discrimination against a black candidate, relative to an otherwise equivalent white one. These results demonstrate the enduring importance of overt prejudice in American politics.
  • Does Trust in Government Affect Support for Redistribution? Results from Randomized Survey Experiments

    Under Revision.
    Why isn't there more support for redistribution in the United States? Scholars have theorized that distrust of government is one potential mechanism. Across three independent survey experiments (N = 3,741), we find that large experimentally induced increases in political trust - enough to eliminate the partisan gap - do not generate more support for redistribution. Even under an optimistic prior, a Bayesian integration of the experimental results suggests the impact on redistribution is approximately zero. To rule out content valence as an explanation for the experimentally induced political trust, we compare these results with two additional survey experiments (N = 1,169) about corruption in the National Football League (NFL), and find treatment effects on political trust and support for redistribution that are indistinguishable from zero. Although the public's trust in government seems to update in response to relevant content, more trust in government does not necessarily imply more support for redistribution.

Statistical Software

  • icsw R package to estimate average treatment effects under non-compliance using inverse compliance score weighting

Recent Publications

  • Enforcing Compulsory Schooling Through Credible Coercion: Lesson's from Australia's Northern Territory Intervention

    The Economic Record, 2018, 94(306): 223–238.
    Australia's Northern Territory Emergency Response and subsequent School Enrolment and Attendance Measure (SEAM) credibly threatened to remove welfare benefits from Indigenous families if their children failed to attend school regularly. A difference‐in‐difference analysis of participation rates in the National Assessment Program – Literacy and Numeracy shows a substantial increase in participation rates the year after SEAM was implemented. However, administrators rarely carried out the threatened sanctions, and these initial gains largely dissipated in subsequent years. This unique episode illustrates the limited scope for promoting Indigenous school participation through conditional cash penalties.
  • What Do We Learn From an Equivalence Study Without Statistical Power?

    The Annals of Emergency Medicine, 2018, 72(2): 223–224.
    We read with interest the article by Linden et al that claims to provide evidence to support the safety of omitting a pelvic examination in women with a confirmed intrauterine pregnancy. In light of previous studies questioning the utility of the pelvic examination, this trial addresses a critical question for patients with threatened abortion. We commend the authors for conducting the largest randomized trial on pelvic examinations to date, but several methodological issues raise questions about whether clinicians can make reliable inferences from this study.
Last updated April 20, 2019.