Matching as Nonparametric Preprocessing for Reducing Model Dependence in Parametric Causal Inference

Kosuke Imai Department of Politics, Princeton University, Princeton, NJ 08544 e-mail: kimai{at}princeton.edu Gary King Department of Government, Harvard University, 1737 Cambridge Street, Cambridge, MA 02138 Elizabeth A. Stuart Departments of Mental Health and Biostatistics, Johns Hopkins Bloomberg School of Public Health, 624 North Broadway, Room 804, Baltimore, MD 21205 e-mail: estuart{at}jhsph.edu e-mail: king{at}harvard.edu (corresponding author) Although published works rarely include causal estimates frommore than a few model specifications, authors usually choosethe presented estimates from numerous trial runs readers neversee. Given the often large variation in estimates across choicesof control variables, functional forms, and other modeling assumptions,how can researchers ensure that the few estimates presentedare accurate or representative? How do readers know that publicationsare not merely demonstrations that it is possible to find aspecification that fits the author's favorite hypothesis? Andhow do we evaluate or even define statistical properties likeunbiasedness or mean squared error when no unique model or estimatoreven exists? Matching methods, which offer the promise of causalinference with fewer assumptions, constitute one possible wayforward, but crucial results in this fast-growing methodologicalliterature are often grossly misinterpreted. We explain howto avoid these misinterpretations and propose a unified approachthat makes it possible for researchers to preprocess data withmatching (such as with the easy-to-use software we offer) andthen to apply the best parametric techniques they would haveused anyway. This procedure makes parametric models producemore accurate and considerably less model-dependent causal inferences. Authors' note: Our thanks to Dan Carpenter and Jeff Koch fordata; Alberto Abadie, Neal Beck, Sam Cook, Alexis Diamond, BenHansen, Guido Imbens, Olivia Lau, Gabe Lenz, Paul Rosenbaum,Don Rubin, and Jas Sekhon for many helpful comments; and theNational Institutes of Aging (P01 AG17625-01), the NationalInstitute of Mental Health (MH066247), the National ScienceFoundation (SES-0318275, IIS-9874747, SES-0550873), and thePrinceton University Committee on Research in the Humanitiesand Social Sciences for research support. Software to implementthe methods in this paper is available at http://GKing.Harvard.edu/matchitand a replication data file is available as Ho et al. (2006).     

Differential outcomes of court-supervised substance abuse treatment among California parolees and probationers

To explore the effectiveness of court-supervised drug treatment for California parolees, offender characteristics, treatment experiences, and outcomes were examined and contrasted to those of probationers. The analysis used statewide administrative data on 4,507 parolees and 22,701 probationers referred to treatment by Proposition 36 during fiscal year 2006-2007. Compared with probationers, parolee problems were more severe at treatment entry, more were treated in residential settings, treatment retention was shorter, and fewer completed treatment. Regarding outcomes, fewer parolees were successful at treatment discharge and more recidivated over 12-months post admission. Both groups improved in many areas by treatment discharge, but improvements were generally smaller among parolees. Significant interaction effects indicated that parolees benefited from residential care and more treatment days, even after controlling for covariates. Court-supervised drug treatment for parolees can "work;" however, parolees have more frequent and diverse needs, and their outcomes are enhanced by more intensive treatment. Findings suggest methods for optimizing the effectiveness of criminal-justice-supervised programs for treating drug-dependent offenders.     

An Evaluation of Six Brief Interventions that Target Drug-Related Problems in Correctional Populations

Finding brief effective treatments for criminal justice populations is a major public need. The CJ-DATS Targeted Intervention for Corrections (TIC), which consists of six brief interventions (Communication, Anger, Motivation, Criminal Thinking, Social Networks, and HIV/Sexual Health), were tested in separate federally-funded randomized control studies. In total, 1,573 criminal justice-involved individuals from 20 correction facilities participated (78% males; 54% white). Multi-level repeated measures analyses found significant gains in knowledge, attitudes, and psychosocial functioning (criteria basic to Knowledge, Attitude, and Practices (KAP) and TCU Treatment Process Models). While improvements were less consistent in criminal thinking, overall evidence supported efficacy for the TIC interventions.     

Procedural justice, routine encounters and citizen perceptions of police: main findings from the Queensland Community Engagement Trial (QCET)

Objectives

To test, under randomized field trial conditions, the impact of police using the principles of procedural justice during routine encounters with citizens on attitudes towards drink-driving, perceptions of compliance, and their satisfaction with the police.

Methods

We conducted the first randomized field trial??the ??Queensland Community Engagement Trial?? (QCET)??to test the impact of police engaging with citizens by operationalizing the key ingredients of procedural justice (neutrality, citizen participation, respect, and trustworthy motives) in a short, high-volume police?Ccitizen encounter. We randomly allocated 60 roadside Random Breath Testing (RBT) operations to control (business-as-usual) and experimental (procedural justice) conditions. Driver surveys were used to measure the key outcomes: attitudes towards drinking and driving, satisfaction with police and perceptions of compliance.

Results

Citizen perceptions of the encounter revealed that the experimental treatment was delivered as planned. We also found significant differences between the experimental and control groups on all key outcome measures: drivers who received the experimental RBT encounter were 1.24 times more likely to report that their views on drinking and driving had changed than the control group; experimental respondents reported small but higher levels of compliance (d?=?.07) and satisfaction (d?=?.18) with police during the encounter than did their control group counterparts.

Conclusions

Our results show that the way citizens perceive the police can be influenced by the way in which police interact with citizens during routine encounters, and demonstrate the positive benefits of police using the principles of procedural justice. Our study was limited by the use of paper-only surveys and low response rate. We also recognize that the experiment setting (RBT road blocks) is limiting and non-reflective of the wider set of routine police?Ccitizen encounters. Future research should be undertaken, using experimental methods, to replicate our field operationalization of procedural justice in different types of police?Ccitizen encounters.     

Why are highly informed citizens sceptical of referenda?

Past work on attitudes toward the use of referenda finds that the most politically informed citizens are more sceptical of its use than their less informed counterparts. An intriguing tension presents itself in that those citizens who may be best equipped for referendum voting, the highly informed and politically sophisticated, are the least likely to support the use of this tool of direct democracy. Using data from Canadian Election Studies, we consider three explanations for relative referenda scepticism among political sophisticates. Our analyses provide evidence to support a confidence in government explanation and a concern for minority rights explanation, but not an incompetent public explanation – a curious finding considering the centrality of citizen competence in the theoretical literature on referenda.     

A Comparative Test of Clinical Judgment Versus Actuarial Prediction of Future Violence

This study tested how graduate level psychology graduate students (n = 20) using the Adapted Violence Risk Appraisal Guide (Adapted VRAG) would do relative to practicing psychologists/psychiatrists (n = 16) using clinical judgment when predicting violence in 10 narratives from the MacArthur Violence Risk Assessment study (Monahan et al., The MacArthur violence risk assessment study. . Retrieved 10 Oct 2005). Results indicated that the practicing psychologists/psychiatrists made significantly more correct predictions than the master’s level students. The professional group demonstrated sensitivity levels of 77.7% and specificity of 96.3%. For the use of the adapted VRAG method by the graduate student group, specificity levels were modest at 54.0%. Sensitivity levels, however, were lower than earlier demonstrated levels at 58.0%. These findings are at variance with earlier reports comparing clinical and actuarial methods. The results may reflect the short amount of time the master’s level students were trained using the Adapted VRAG as well as the small number of participants in this study. Additional research comparing other professions is recommended, as well as examining if experience in the forensics field would affect one’s ability to predict violence.     

Toward a Balanced Approach: Defining Police Roles in Responding to Domestic Violence

Responding to domestic violence poses unique challenges to law enforcement officers. Prior research has focused on issues such as officer safety and arrest policies but overlooked a critical component—the officers’ perception of their role in responding to domestic violence incidents. This study explores how police officers define their roles through the use of focus group interviews with detectives assigned to a domestic violence unit. Findings include the identification of three role perspectives. Recommendations for a more balanced approach of these role perspectives in responding to domestic violence crimes through the use of more effective training, supervision, and evaluation are proposed.