High Dimensional Propensity Score Estimation via Covariate Balancing
|
|
- Alison Garrett
- 6 years ago
- Views:
Transcription
1 High Dimensional Propensity Score Estimation via Covariate Balancing Kosuke Imai Princeton University Talk at Columbia University May 13, 2017 Joint work with Yang Ning and Sida Peng Kosuke Imai (Princeton) Covariate Balancing Propensity Score Columbia (May 13, 2017) 1 / 20
2 Motivation Covariate adjustment in observational studies: 1 Identify and measure a set of potential confounders 2 Weight or match the treated and control units to make them similar Practical questions: 1 Which covariates should we adjust? 2 How should we adjust them? 3 What should we do if we have many potential confounders? Major advances over the last several years in the literature: Tan, Hainmueller, Graham et al., Zubizarreta, Belloni et al., Chan et al., Farrell, Chernozhukov et al., Athey et al., Zhao, etc. Covariate Balancing Propensity Score (CBPS) Imai & Ratkovic JRSSB; Fan et al. Estimate propensity score such that covariates are balanced Extensions: continuous treatment (Fong et al.), dynamic treatment (Imai & Ratkovic JASA), high-dimensional setting (Ning et al.) Kosuke Imai (Princeton) Covariate Balancing Propensity Score Columbia (May 13, 2017) 2 / 20
3 Overview 1 Review of propensity score and CBPS Inverse probability weighting (IPW) Propensity score tautology covariate balancing Optimal covariate balancing for CBPS 2 High-dimensional CBPS (HD-CBPS) Impossible to balance all covariates trade-off between covariates Sparsity assumption + Regularization Balance covariates that are predictive of outcome Remove bias (due to regularization) by covariate balancing Kosuke Imai (Princeton) Covariate Balancing Propensity Score Columbia (May 13, 2017) 3 / 20
4 Propensity Score and CBPS T i {0, 1}: binary treatment X i : d-dimensional pre-treatment covariates Identification assumptions: SUTVA, overlap, unconfoundedness Dual characteristics of propensity score: 1 Predicts treatment assignment: π(x i ) = Pr(T i = 1 X i ) 2 Balances covariates: { } Ti E π(x i ) f (X i) { } 1 Ti = E 1 π(x i ) f (X i) = E{f (X i )} Propensity score tautology CBPS: estimate propensity score such that balance is optimized Covariate balancing as estimating equations Kosuke Imai (Princeton) Covariate Balancing Propensity Score Columbia (May 13, 2017) 4 / 20
5 Optimal Covariate Balancing Equations Which covariates and what functions of them should we balance? The outcome model matters Outcome model: E(Y i (t) X i ) = K t (X i ) for t = 0, 1 Possibly misspecified propensity score π β (X i ) π β o(x i ) Inverse probability weighting (IPW) estimator: 1 n { n i=1 T i Y i π ˆβ (X i) (1 T i)y i 1 π ˆβ (X i) Asymptotic bias for the ATE: [ ( Ti E π β o(x i ) 1 T ) ] i {π β o(x i )K 0 (X i ) + (1 π β o)k 1 (X i )} 1 π β o(x i ) }{{} optimal f (X i ) [ ( = E 1 1 T ) ( ) i Ti ] K 0 (X i ) + 1 π β o(x i ) π β o(x i ) 1 K 1 (X i ) }{{}}{{} balance control group balance treatment group Kosuke Imai (Princeton) Covariate Balancing Propensity Score Columbia (May 13, 2017) 5 / 20 }
6 Covariate Balancing in High-Dimension We focus on the estimation of E(Y i (1)) A similar estimation strategy can be applied to E(Y i (0)) Low dimension strong covariate balancing 1 n n i=1 ( ) Ti 1 X i = 0 ˆπ i asymptotic normality, semiparametric efficiency, double robustness (Fan et al.) High dimension weak covariate balancing 1 n n i=1 ( ) Ti 1 α X i = 0 ˆπ i Kosuke Imai (Princeton) Covariate Balancing Propensity Score Columbia (May 13, 2017) 6 / 20
7 Assumptions for Estimating µ 1 = E(Y i(1)) 1 Linear outcome model: 2 Logistic propensity score model: K 1 (X i ) = α X i Pr(T i = 1 X i ) = π(β X i ) = exp(β X i ) 1 + exp(β X i ) 3 Sparsity of both models: max(s 1, s 2 ) log d log n/n = o(1) where s 1 = #{β j > 0} and s 2 = #{α j > 0} 4 Tuning parameters for Lasso: λ = a log(d)/n and λ = a log(d)/n for some unknown constants a and a Kosuke Imai (Princeton) Covariate Balancing Propensity Score Columbia (May 13, 2017) 7 / 20
8 The Proposed Methodology 1 Fit the penalized logistic regression model for propensity score: 1 ˆβ = argmin β R d n n i=1 { } T i (β X i ) log(1 + exp(β X i )) + λ β 1, 2 Fit the penalized linear regression model for the outcome: 1 n α = argmin T i {Y i α X i } 2 + λ α 1, α R d n i=1 3 Calibrate the estimated propensity score by balancing covariates: 1 n γ = argmin T 2 i n π(γ X i + ˆβ S Sc X i S ) 1 X i S c γ R S where S = {j : α j > 0} i=1 4 Use the estimated propensity score π i = π( γ X i S + ˆβ Sc X i S c ) for the IPW estimator ˆµ 1 = 1 n n i=1 T iy i / π i Kosuke Imai (Princeton) Covariate Balancing Propensity Score Columbia (May 13, 2017) 8 / 20 2
9 Theoretical Properties 1 n-consistency ˆµ 1 µ 1 = 1 n n [ Ti π i=1 i ] (Y i (1) α X i ) + α X i µ 1 + o p (n 1/2 ) 2 asymptotic normality and semiparametric efficiency n(ˆµ1 µ 1 ) [ ]) D 1 N (0, E π E(ɛ2 i1 X) + (α X i µ 1 )2 valid under K -fold cross-validation 3 sample boundedness ˆµ 1 min i:t i =1 Y i n n i=1 T i π i = min i:t i =1 Y i and similarly, ˆµ 1 max i:ti =1 Y i Kosuke Imai (Princeton) Covariate Balancing Propensity Score Columbia (May 13, 2017) 9 / 20
10 Robustness and Generalizations 1 Multi-valued treatment regime use the penalized multinomial logistic regression balance selected covariates for each treatment group 2 Generalized linear models for the outcome exponential family p(y X) = h(y, ψ) exp[{y α X b(α X)}/a(ψ)] balance f (X i ) = (b ( α X i ), b ( α X i )X i S) b (α X i ) b ( α X i ) + b ( α X i )(α α)x i 3 Misspecified propensity score model theoretical properties hold so long as the outcome model is correct sure screening property must hold for S in the outcome model Kosuke Imai (Princeton) Covariate Balancing Propensity Score Columbia (May 13, 2017) 10 / 20
11 Comparison with Other Methods 1 Augmented inverse probability weighting (AIPW) Belloni et al. (2014), Farrell (2015), Chernozhukov et al. (2016) Extends Robins doubly-robust estimator to high-dimension 1 n n { ˆm 1 (X i ) + T } i(y i ˆm 1 (X i )) ˆπ i (X i ) i=1 Double selection method 2 Residual balancing (RB) Athey, Imbens, and Wagner (2016) Assumes sparsity only for the outcome model Critically relies on linearity assumption Kosuke Imai (Princeton) Covariate Balancing Propensity Score Columbia (May 13, 2017) 11 / 20
12 Simulation Studies Inspired by Kang and Schafer (2007) Covariates: X i i.i.d. N (0, Σ) where Σ jk = ρ j k Propensity score model: Pr(T i X i ) = logit 1 ( X i X i2 0.25X i3 0.1X i4 ) Outcome model: Y i (1) = X i X i X i9 + ɛ 1i Y i (0) = X i X i X i X i X i9 + ɛ 0i where ɛ ti i.i.d. N (0, 1) for t = 0, 1 Add irrelevant covariates so that a total number of covariates, d, varies from 10 to 2000 Comparison with Residual Balancing (RB) and regularized augmented inverse probability weighting (AIPW) Kosuke Imai (Princeton) Covariate Balancing Propensity Score Columbia (May 13, 2017) 12 / 20
13 Both Models are Correct Standardized root-mean-squared error E(ˆµ µ ) 2 /µ n = 100 n = 1000 d HD-CBPS RB AIPW HD-CBPS RB AIPW Kosuke Imai (Princeton) Covariate Balancing Propensity Score Columbia (May 13, 2017) 13 / 20
14 Propensity Score Model is Misspecified Misspecification through transformation: X mis = (exp(x 1 /2), X 2 (1 + exp(x 1 )) , (X 1 X 3 / ) 3, (X 2 + X ) 2, X 5,, X d ) n = 100 n = 1000 d HD-CBPS RB AIPW HD-CBPS RB AIPW Kosuke Imai (Princeton) Covariate Balancing Propensity Score Columbia (May 13, 2017) 14 / 20
15 Both Models are Misspecified Outcome model misspecification through another transformation: X mis = (exp(x 1 /2), X 2 (1 + exp(x 1 )) , (X 1 X 3 / ) 3, (X 2 + X ) 2, X 6, exp(x 6 + X 7 ), X 2 9, X , X 9,, X d ) n = 100 n = 1000 d HD-CBPS RB AIPW HD-CBPS RB AIPW Kosuke Imai (Princeton) Covariate Balancing Propensity Score Columbia (May 13, 2017) 15 / 20
16 Coverage of 95% Confidence Intervals Both models are correctly specified Average length of confidence intervals in parentheses d n = 100 n = 500 n = (0.8550) (0.3674) (0.2766) (0.7649) (0.3767) (0.2747) (0.8027) (0.3800) (0.2617) (0.7268) (0.3718) (0.2697) (0.7882) (0.4000) (0.2609) Kosuke Imai (Princeton) Covariate Balancing Propensity Score Columbia (May 13, 2017) 16 / 20
17 Logistic Outcome Model Pr(Y i (1) = 1 X i ) = logit 1 (1 + exp( X i X i X i3 )) Pr(Y i (0) = 1 X i ) = logit 1 (1 + exp( X i X i X i X i X i5 )) n = 100 n = 500 n = 1000 d HD-CBPS AIPW HD-CBPS AIPW HD-CBPS AIPW Kosuke Imai (Princeton) Covariate Balancing Propensity Score Columbia (May 13, 2017) 17 / 20
18 Empirical Illustration Effect of college attendance on political participation (Kam and Palmer 2008 JOP) Data: Youth-Parent Socialization Panel Study (n = 1051) Outcome: a total number of (up to 8) participatory acts Covariates: 81 pre-adult characteristics Propensity score: a total of 204 predictors The authors found little effect of college attendance Henderson and Chatfield (2011) and Mayer (2011) use genetic matching and find a positive and statistically significant effect Kosuke Imai (Princeton) Covariate Balancing Propensity Score Columbia (May 13, 2017) 18 / 20
19 Empirical Results Propensity score: logistic regression with 204 predictors Outcome model: binomial logistic regression with 204 predictors 27 covariates are selected by HD-CBPS Overall (ATE) Overall (ATT) Whites (ATE) HD-CBPS CBPS AIPW (0.1247) (0.2380) (0.1043) (0.1420) (0.3094) (0.1279) (0.1089) Kosuke Imai (Princeton) Covariate Balancing Propensity Score Columbia (May 13, 2017) 19 / 20
20 Concluding Remarks Covariate balancing as an efficient and robust way to estimate propensity score Challenges in high-dimension = cannot balance all covariates High-dimensional covariate balancing propensity score (HD-CBPS) sparsity assumption + weak covariate balancing efficient and robust to misspecification of propensity score model more reliance on the outcome model specification loss of double-robustness HD-CBPS is implemented in the R package CBPS Kosuke Imai (Princeton) Covariate Balancing Propensity Score Columbia (May 13, 2017) 20 / 20
Covariate Balancing Propensity Score for General Treatment Regimes
Covariate Balancing Propensity Score for General Treatment Regimes Kosuke Imai Princeton University October 14, 2014 Talk at the Department of Psychiatry, Columbia University Joint work with Christian
More informationWeighting Methods. Harvard University STAT186/GOV2002 CAUSAL INFERENCE. Fall Kosuke Imai
Weighting Methods Kosuke Imai Harvard University STAT186/GOV2002 CAUSAL INFERENCE Fall 2018 Kosuke Imai (Harvard) Weighting Methods Stat186/Gov2002 Fall 2018 1 / 13 Motivation Matching methods for improving
More informationDouble Robustness. Bang and Robins (2005) Kang and Schafer (2007)
Double Robustness Bang and Robins (2005) Kang and Schafer (2007) Set-Up Assume throughout that treatment assignment is ignorable given covariates (similar to assumption that data are missing at random
More informationPrimal-dual Covariate Balance and Minimal Double Robustness via Entropy Balancing
Primal-dual Covariate Balance and Minimal Double Robustness via (Joint work with Daniel Percival) Department of Statistics, Stanford University JSM, August 9, 2015 Outline 1 2 3 1/18 Setting Rubin s causal
More informationCausal Inference Lecture Notes: Causal Inference with Repeated Measures in Observational Studies
Causal Inference Lecture Notes: Causal Inference with Repeated Measures in Observational Studies Kosuke Imai Department of Politics Princeton University November 13, 2013 So far, we have essentially assumed
More informationCombining multiple observational data sources to estimate causal eects
Department of Statistics, North Carolina State University Combining multiple observational data sources to estimate causal eects Shu Yang* syang24@ncsuedu Joint work with Peng Ding UC Berkeley May 23,
More informationModification and Improvement of Empirical Likelihood for Missing Response Problem
UW Biostatistics Working Paper Series 12-30-2010 Modification and Improvement of Empirical Likelihood for Missing Response Problem Kwun Chuen Gary Chan University of Washington - Seattle Campus, kcgchan@u.washington.edu
More informationApproximate Residual Balancing: De-Biased Inference of Average Treatment Effects in High Dimensions
Approximate Residual Balancing: De-Biased Inference of Average Treatment Effects in High Dimensions Susan Athey Guido W. Imbens Stefan Wager Current version February 2018 arxiv:1604.07125v5 [stat.me] 31
More informationApproximate Residual Balancing: De-Biased Inference of Average Treatment Effects in High Dimensions
Approximate Residual Balancing: De-Biased Inference of Average Treatment Effects in High Dimensions Susan Athey Guido W. Imbens Stefan Wager Current version November 2016 Abstract There are many settings
More informationarxiv: v3 [stat.me] 14 Nov 2016
Approximate Residual Balancing: De-Biased Inference of Average Treatment Effects in High Dimensions Susan Athey Guido W. Imbens Stefan Wager arxiv:1604.07125v3 [stat.me] 14 Nov 2016 Current version November
More informationCovariate Balancing Propensity Score for a Continuous Treatment: Application to the Efficacy of Political Advertisements
Covariate Balancing Propensity Score for a Continuous Treatment: Application to the Efficacy of Political Advertisements Christian Fong Chad Hazlett Kosuke Imai Forthcoming in Annals of Applied Statistics
More informationRobust Estimation of Inverse Probability Weights for Marginal Structural Models
Robust Estimation of Inverse Probability Weights for Marginal Structural Models Kosuke IMAI and Marc RATKOVIC Marginal structural models (MSMs) are becoming increasingly popular as a tool for causal inference
More informationAn Experimental Evaluation of High-Dimensional Multi-Armed Bandits
An Experimental Evaluation of High-Dimensional Multi-Armed Bandits Naoki Egami Romain Ferrali Kosuke Imai Princeton University Talk at Political Data Science Conference Washington University, St. Louis
More informationCausal Inference with General Treatment Regimes: Generalizing the Propensity Score
Causal Inference with General Treatment Regimes: Generalizing the Propensity Score David van Dyk Department of Statistics, University of California, Irvine vandyk@stat.harvard.edu Joint work with Kosuke
More informationStatistical Analysis of Randomized Experiments with Nonignorable Missing Binary Outcomes
Statistical Analysis of Randomized Experiments with Nonignorable Missing Binary Outcomes Kosuke Imai Department of Politics Princeton University July 31 2007 Kosuke Imai (Princeton University) Nonignorable
More informationWhen Should We Use Linear Fixed Effects Regression Models for Causal Inference with Panel Data?
When Should We Use Linear Fixed Effects Regression Models for Causal Inference with Panel Data? Kosuke Imai Department of Politics Center for Statistics and Machine Learning Princeton University Joint
More informationA Sampling of IMPACT Research:
A Sampling of IMPACT Research: Methods for Analysis with Dropout and Identifying Optimal Treatment Regimes Marie Davidian Department of Statistics North Carolina State University http://www.stat.ncsu.edu/
More informationCOVARIATE BALANCING PROPENSITY SCORE FOR A CONTINUOUS TREATMENT: APPLICATION TO THE EFFICACY OF POLITICAL ADVERTISEMENTS
The Annals of Applied Statistics 2018, Vol. 12, No. 1, 156 177 https://doi.org/10.1214/17-aoas1101 Institute of Mathematical Statistics, 2018 COVARIATE BALANCING PROPENSITY SCORE FOR A CONTINUOUS TREATMENT:
More informationEstimating the Marginal Odds Ratio in Observational Studies
Estimating the Marginal Odds Ratio in Observational Studies Travis Loux Christiana Drake Department of Statistics University of California, Davis June 20, 2011 Outline The Counterfactual Model Odds Ratios
More informationIdentification and Inference in Causal Mediation Analysis
Identification and Inference in Causal Mediation Analysis Kosuke Imai Luke Keele Teppei Yamamoto Princeton University Ohio State University November 12, 2008 Kosuke Imai (Princeton) Causal Mediation Analysis
More informationENTROPY BALANCING IS DOUBLY ROBUST QINGYUAN ZHAO. Department of Statistics, Stanford University DANIEL PERCIVAL. Google Inc.
ENTROPY BALANCING IS DOUBLY ROBUST QINGYUAN ZHAO Department of Statistics, Stanford University DANIEL PERCIVAL Google Inc. Abstract. Covariate balance is a conventional key diagnostic for methods used
More informationENTROPY BALANCING IS DOUBLY ROBUST. Department of Statistics, Wharton School, University of Pennsylvania DANIEL PERCIVAL. Google Inc.
ENTROPY BALANCING IS DOUBLY ROBUST QINGYUAN ZHAO arxiv:1501.03571v3 [stat.me] 11 Feb 2017 Department of Statistics, Wharton School, University of Pennsylvania DANIEL PERCIVAL Google Inc. Abstract. Covariate
More informationCausal Inference Lecture Notes: Selection Bias in Observational Studies
Causal Inference Lecture Notes: Selection Bias in Observational Studies Kosuke Imai Department of Politics Princeton University April 7, 2008 So far, we have studied how to analyze randomized experiments.
More informationExtending causal inferences from a randomized trial to a target population
Extending causal inferences from a randomized trial to a target population Issa Dahabreh Center for Evidence Synthesis in Health, Brown University issa dahabreh@brown.edu January 16, 2019 Issa Dahabreh
More informationWhen Should We Use Linear Fixed Effects Regression Models for Causal Inference with Longitudinal Data?
When Should We Use Linear Fixed Effects Regression Models for Causal Inference with Longitudinal Data? Kosuke Imai Department of Politics Center for Statistics and Machine Learning Princeton University
More informationComment: Understanding OR, PS and DR
Statistical Science 2007, Vol. 22, No. 4, 560 568 DOI: 10.1214/07-STS227A Main article DOI: 10.1214/07-STS227 c Institute of Mathematical Statistics, 2007 Comment: Understanding OR, PS and DR Zhiqiang
More informationWeb-based Supplementary Materials for A Robust Method for Estimating. Optimal Treatment Regimes
Biometrics 000, 000 000 DOI: 000 000 0000 Web-based Supplementary Materials for A Robust Method for Estimating Optimal Treatment Regimes Baqun Zhang, Anastasios A. Tsiatis, Eric B. Laber, and Marie Davidian
More informationStatistical Analysis of List Experiments
Statistical Analysis of List Experiments Graeme Blair Kosuke Imai Princeton University December 17, 2010 Blair and Imai (Princeton) List Experiments Political Methodology Seminar 1 / 32 Motivation Surveys
More informationUse of Matching Methods for Causal Inference in Experimental and Observational Studies. This Talk Draws on the Following Papers:
Use of Matching Methods for Causal Inference in Experimental and Observational Studies Kosuke Imai Department of Politics Princeton University April 13, 2009 Kosuke Imai (Princeton University) Matching
More informationBootstrapping Sensitivity Analysis
Bootstrapping Sensitivity Analysis Qingyuan Zhao Department of Statistics, The Wharton School University of Pennsylvania May 23, 2018 @ ACIC Based on: Qingyuan Zhao, Dylan S. Small, and Bhaswar B. Bhattacharya.
More informationWhen Should We Use Linear Fixed Effects Regression Models for Causal Inference with Longitudinal Data?
When Should We Use Linear Fixed Effects Regression Models for Causal Inference with Longitudinal Data? Kosuke Imai Princeton University Asian Political Methodology Conference University of Sydney Joint
More informationPropensity Score Weighting with Multilevel Data
Propensity Score Weighting with Multilevel Data Fan Li Department of Statistical Science Duke University October 25, 2012 Joint work with Alan Zaslavsky and Mary Beth Landrum Introduction In comparative
More informationStable Weights that Balance Covariates for Estimation with Incomplete Outcome Data
Stable Weights that Balance Covariates for Estimation with Incomplete Outcome Data José R. Zubizarreta Abstract Weighting methods that adjust for observed covariates, such as inverse probability weighting,
More informationarxiv: v1 [stat.me] 15 May 2011
Working Paper Propensity Score Analysis with Matching Weights Liang Li, Ph.D. arxiv:1105.2917v1 [stat.me] 15 May 2011 Associate Staff of Biostatistics Department of Quantitative Health Sciences, Cleveland
More informationLinear Regression Models P8111
Linear Regression Models P8111 Lecture 25 Jeff Goldsmith April 26, 2016 1 of 37 Today s Lecture Logistic regression / GLMs Model framework Interpretation Estimation 2 of 37 Linear regression Course started
More informationCOVARIATE BALANCING PROPENSITY SCORE BY TAILORED LOSS FUNCTIONS. By Qingyuan Zhao University of Pennsylvania
Submitted to the Annals of Statistics COVARIATE BALANCING PROPENSITY SCORE BY TAILORED LOSS FUNCTIONS By Qingyuan Zhao University of Pennsylvania In observational studies, propensity scores are commonly
More informationProgram Evaluation with High-Dimensional Data
Program Evaluation with High-Dimensional Data Alexandre Belloni Duke Victor Chernozhukov MIT Iván Fernández-Val BU Christian Hansen Booth ESWC 215 August 17, 215 Introduction Goal is to perform inference
More informationDiscussion of Papers on the Extensions of Propensity Score
Discussion of Papers on the Extensions of Propensity Score Kosuke Imai Princeton University August 3, 2010 Kosuke Imai (Princeton) Generalized Propensity Score 2010 JSM (Vancouver) 1 / 11 The Theme and
More informationInstrumental Variables
Instrumental Variables Kosuke Imai Harvard University STAT186/GOV2002 CAUSAL INFERENCE Fall 2018 Kosuke Imai (Harvard) Noncompliance in Experiments Stat186/Gov2002 Fall 2018 1 / 18 Instrumental Variables
More informationOn the Use of Linear Fixed Effects Regression Models for Causal Inference
On the Use of Linear Fixed Effects Regression Models for ausal Inference Kosuke Imai Department of Politics Princeton University Joint work with In Song Kim Atlantic ausal Inference onference Johns Hopkins
More informationStatistical Analysis of Causal Mechanisms
Statistical Analysis of Causal Mechanisms Kosuke Imai Luke Keele Dustin Tingley Teppei Yamamoto Princeton University Ohio State University July 25, 2009 Summer Political Methodology Conference Imai, Keele,
More informationStatistical Analysis of Causal Mechanisms
Statistical Analysis of Causal Mechanisms Kosuke Imai Princeton University November 17, 2008 Joint work with Luke Keele (Ohio State) and Teppei Yamamoto (Princeton) Kosuke Imai (Princeton) Causal Mechanisms
More informationStatistical Analysis of the Item Count Technique
Statistical Analysis of the Item Count Technique Kosuke Imai Department of Politics Princeton University Joint work with Graeme Blair May 4, 2011 Kosuke Imai (Princeton) Item Count Technique UCI (Statistics)
More informationA General Framework for High-Dimensional Inference and Multiple Testing
A General Framework for High-Dimensional Inference and Multiple Testing Yang Ning Department of Statistical Science Joint work with Han Liu 1 Overview Goal: Control false scientific discoveries in high-dimensional
More informationAchieving Optimal Covariate Balance Under General Treatment Regimes
Achieving Under General Treatment Regimes Marc Ratkovic Princeton University May 24, 2012 Motivation For many questions of interest in the social sciences, experiments are not possible Possible bias in
More informationResidual Balancing: A Method of Constructing Weights for Marginal Structural Models
Residual Balancing: A Method of Constructing Weights for Marginal Structural Models Xiang Zhou Harvard University Geoffrey T. Wodtke University of Toronto March 29, 2019 Abstract When making causal inferences,
More informationCalibration Estimation of Semiparametric Copula Models with Data Missing at Random
Calibration Estimation of Semiparametric Copula Models with Data Missing at Random Shigeyuki Hamori 1 Kaiji Motegi 1 Zheng Zhang 2 1 Kobe University 2 Renmin University of China Econometrics Workshop UNC
More informationCalibration Estimation of Semiparametric Copula Models with Data Missing at Random
Calibration Estimation of Semiparametric Copula Models with Data Missing at Random Shigeyuki Hamori 1 Kaiji Motegi 1 Zheng Zhang 2 1 Kobe University 2 Renmin University of China Institute of Statistics
More informationRobustness to Parametric Assumptions in Missing Data Models
Robustness to Parametric Assumptions in Missing Data Models Bryan Graham NYU Keisuke Hirano University of Arizona April 2011 Motivation Motivation We consider the classic missing data problem. In practice
More informationCausal Interaction in Factorial Experiments: Application to Conjoint Analysis
Causal Interaction in Factorial Experiments: Application to Conjoint Analysis Naoki Egami Kosuke Imai Princeton University Talk at the University of Macau June 16, 2017 Egami and Imai (Princeton) Causal
More informationCausal Inference Basics
Causal Inference Basics Sam Lendle October 09, 2013 Observed data, question, counterfactuals Observed data: n i.i.d copies of baseline covariates W, treatment A {0, 1}, and outcome Y. O i = (W i, A i,
More informationPropensity-Score Based Methods for Causal Inference in Observational Studies with Fixed Non-Binary Treatments
Propensity-Score Based Methods for Causal Inference in Observational Studies with Fixed Non-Binary reatments Shandong Zhao Department of Statistics, University of California, Irvine, CA 92697 shandonm@uci.edu
More informationMatching and Weighting Methods for Causal Inference
Matching and Weighting Methods for ausal Inference Kosuke Imai Princeton University Methods Workshop, Duke University Kosuke Imai (Princeton) Matching and Weighting Methods Duke (January 8 9, 23) / 57
More informationFall 2017 STAT 532 Homework Peter Hoff. 1. Let P be a probability measure on a collection of sets A.
1. Let P be a probability measure on a collection of sets A. (a) For each n N, let H n be a set in A such that H n H n+1. Show that P (H n ) monotonically converges to P ( k=1 H k) as n. (b) For each n
More informationVariable selection and machine learning methods in causal inference
Variable selection and machine learning methods in causal inference Debashis Ghosh Department of Biostatistics and Informatics Colorado School of Public Health Joint work with Yeying Zhu, University of
More informationCausal Inference in Observational Studies with Non-Binary Treatments. David A. van Dyk
Causal Inference in Observational Studies with Non-Binary reatments Statistics Section, Imperial College London Joint work with Shandong Zhao and Kosuke Imai Cass Business School, October 2013 Outline
More informationCausal Interaction in Factorial Experiments: Application to Conjoint Analysis
Causal Interaction in Factorial Experiments: Application to Conjoint Analysis Naoki Egami Kosuke Imai Princeton University Talk at the Institute for Mathematics and Its Applications University of Minnesota
More informationPropensity-Score Based Methods for Causal Inference in Observational Studies with Fixed Non-Binary Treatments
Propensity-Score Based Methods for Causal Inference in Observational Studies with Fixed Non-Binary reatments Shandong Zhao David A. van Dyk Kosuke Imai July 3, 2018 Abstract Propensity score methods are
More informationStatistical Analysis of Causal Mechanisms
Statistical Analysis of Causal Mechanisms Kosuke Imai Princeton University April 13, 2009 Kosuke Imai (Princeton) Causal Mechanisms April 13, 2009 1 / 26 Papers and Software Collaborators: Luke Keele,
More informationUsing Estimating Equations for Spatially Correlated A
Using Estimating Equations for Spatially Correlated Areal Data December 8, 2009 Introduction GEEs Spatial Estimating Equations Implementation Simulation Conclusion Typical Problem Assess the relationship
More informationSelective Inference for Effect Modification
Inference for Modification (Joint work with Dylan Small and Ashkan Ertefaie) Department of Statistics, University of Pennsylvania May 24, ACIC 2017 Manuscript and slides are available at http://www-stat.wharton.upenn.edu/~qyzhao/.
More informationLIKELIHOOD RATIO INFERENCE FOR MISSING DATA MODELS
LIKELIHOOD RATIO IFERECE FOR MISSIG DATA MODELS KARU ADUSUMILLI AD TAISUKE OTSU Abstract. Missing or incomplete outcome data is a ubiquitous problem in biomedical and social sciences. Under the missing
More informationGraybill Conference Poster Session Introductions
Graybill Conference Poster Session Introductions 2013 Graybill Conference in Modern Survey Statistics Colorado State University Fort Collins, CO June 10, 2013 Small Area Estimation with Incomplete Auxiliary
More informationClassification. Chapter Introduction. 6.2 The Bayes classifier
Chapter 6 Classification 6.1 Introduction Often encountered in applications is the situation where the response variable Y takes values in a finite set of labels. For example, the response Y could encode
More informationInteraction Effects and Causal Heterogeneity
Interaction Effects and Causal Heterogeneity Kosuke Imai Princeton University Winter Conference in Statistics Borgafjäll, Sweden March 11, 2015 Joint work with Naoki Egami and Marc Ratkovic Kosuke Imai
More informationCausal Inference with a Continuous Treatment and Outcome: Alternative Estimators for Parametric Dose-Response Functions
Causal Inference with a Continuous Treatment and Outcome: Alternative Estimators for Parametric Dose-Response Functions Joe Schafer Office of the Associate Director for Research and Methodology U.S. Census
More informationUltra High Dimensional Variable Selection with Endogenous Variables
1 / 39 Ultra High Dimensional Variable Selection with Endogenous Variables Yuan Liao Princeton University Joint work with Jianqing Fan Job Market Talk January, 2012 2 / 39 Outline 1 Examples of Ultra High
More informationCalibration Estimation for Semiparametric Copula Models under Missing Data
Calibration Estimation for Semiparametric Copula Models under Missing Data Shigeyuki Hamori 1 Kaiji Motegi 1 Zheng Zhang 2 1 Kobe University 2 Renmin University of China Economics and Economic Growth Centre
More informationTelescope Matching: A Flexible Approach to Estimating Direct Effects *
Telescope Matching: A Flexible Approach to Estimating Direct Effects * Matthew Blackwell Anton Strezhnev August 4, 2018 Abstract Estimating the direct effect of a treatment fixing the value of a consequence
More informationA General Double Robustness Result for Estimating Average Treatment Effects
A General Double Robustness Result for Estimating Average Treatment Effects Tymon S loczyński Jeffrey M. Wooldridge Abstract In this paper we study doubly robust estimators of various average and quantile
More informationPropensity Score Methods for Causal Inference
John Pura BIOS790 October 2, 2015 Causal inference Philosophical problem, statistical solution Important in various disciplines (e.g. Koch s postulates, Bradford Hill criteria, Granger causality) Good
More information,..., θ(2),..., θ(n)
Likelihoods for Multivariate Binary Data Log-Linear Model We have 2 n 1 distinct probabilities, but we wish to consider formulations that allow more parsimonious descriptions as a function of covariates.
More informationModeling the scale parameter ϕ A note on modeling correlation of binary responses Using marginal odds ratios to model association for binary responses
Outline Marginal model Examples of marginal model GEE1 Augmented GEE GEE1.5 GEE2 Modeling the scale parameter ϕ A note on modeling correlation of binary responses Using marginal odds ratios to model association
More informationDOUBLY ROBUST NONPARAMETRIC MULTIPLE IMPUTATION FOR IGNORABLE MISSING DATA
Statistica Sinica 22 (2012), 149-172 doi:http://dx.doi.org/10.5705/ss.2010.069 DOUBLY ROBUST NONPARAMETRIC MULTIPLE IMPUTATION FOR IGNORABLE MISSING DATA Qi Long, Chiu-Hsieh Hsu and Yisheng Li Emory University,
More informationBounded, Efficient, and Doubly Robust Estimation with Inverse Weighting
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 Biometrika (2008), 94, 2, pp. 1 22 C 2008 Biometrika Trust Printed
More informationMatching Methods for Causal Inference with Time-Series Cross-Section Data
Matching Methods for Causal Inference with Time-Series Cross-Section Data Kosuke Imai In Song Kim Erik Wang April 28, 2018 Abstract Matching methods aim to improve the validity of causal inference in observational
More informationUnpacking the Black-Box of Causality: Learning about Causal Mechanisms from Experimental and Observational Studies
Unpacking the Black-Box of Causality: Learning about Causal Mechanisms from Experimental and Observational Studies Kosuke Imai Princeton University February 23, 2012 Joint work with L. Keele (Penn State)
More informationUnpacking the Black-Box of Causality: Learning about Causal Mechanisms from Experimental and Observational Studies
Unpacking the Black-Box of Causality: Learning about Causal Mechanisms from Experimental and Observational Studies Kosuke Imai Princeton University January 23, 2012 Joint work with L. Keele (Penn State)
More informationarxiv: v2 [stat.me] 8 Oct 2018
SENSITIVITY ANALYSIS FOR INVERSE PROBABILITY WEIGHTING ESTIMATORS VIA THE PERCENTILE BOOTSTRAP QINGYUAN ZHAO, DYLAN S. SMALL AND BHASWAR B. BHATTACHARYA arxiv:1711.1186v [stat.me] 8 Oct 018 Department
More informationPropensity Score Analysis Using teffects in Stata. SOC 561 Programming for the Social Sciences Hyungjun Suh Apr
Propensity Score Analysis Using teffects in Stata SOC 561 Programming for the Social Sciences Hyungjun Suh Apr. 25. 2016 Overview Motivation Propensity Score Weighting Propensity Score Matching with teffects
More informationCausal Mediation Analysis in R. Quantitative Methodology and Causal Mechanisms
Causal Mediation Analysis in R Kosuke Imai Princeton University June 18, 2009 Joint work with Luke Keele (Ohio State) Dustin Tingley and Teppei Yamamoto (Princeton) Kosuke Imai (Princeton) Causal Mediation
More informationUniversity of California, Berkeley
University of California, Berkeley U.C. Berkeley Division of Biostatistics Working Paper Series Year 2011 Paper 288 Targeted Maximum Likelihood Estimation of Natural Direct Effect Wenjing Zheng Mark J.
More informationBalancing Covariates via Propensity Score Weighting: The Overlap Weights
Balancing Covariates via Propensity Score Weighting: The Overlap Weights Kari Lock Morgan Department of Statistics Penn State University klm47@psu.edu PSU Methodology Center Brown Bag April 6th, 2017 Joint
More informationWhat s New in Econometrics. Lecture 1
What s New in Econometrics Lecture 1 Estimation of Average Treatment Effects Under Unconfoundedness Guido Imbens NBER Summer Institute, 2007 Outline 1. Introduction 2. Potential Outcomes 3. Estimands and
More informationThe Augmented Synthetic Control Method
The Augmented Synthetic Control Method Eli Ben-Michael, Avi Feller, and Jesse Rothstein UC Berkeley November 2018 Abstract The synthetic control method (SCM) is a popular approach for estimating the impact
More informationBayesian regression tree models for causal inference: regularization, confounding and heterogeneity
Bayesian regression tree models for causal inference: regularization, confounding and heterogeneity P. Richard Hahn, Jared Murray, and Carlos Carvalho June 22, 2017 The problem setting We want to estimate
More informationA Bootstrap Lasso + Partial Ridge Method to Construct Confidence Intervals for Parameters in High-dimensional Sparse Linear Models
A Bootstrap Lasso + Partial Ridge Method to Construct Confidence Intervals for Parameters in High-dimensional Sparse Linear Models Jingyi Jessica Li Department of Statistics University of California, Los
More informationLatent Variable Models for Binary Data. Suppose that for a given vector of explanatory variables x, the latent
Latent Variable Models for Binary Data Suppose that for a given vector of explanatory variables x, the latent variable, U, has a continuous cumulative distribution function F (u; x) and that the binary
More informationUse of Matching Methods for Causal Inference in Experimental and Observational Studies. This Talk Draws on the Following Papers:
Use of Matching Methods for Causal Inference in Experimental and Observational Studies Kosuke Imai Department of Politics Princeton University April 27, 2007 Kosuke Imai (Princeton University) Matching
More informationUnpacking the Black-Box: Learning about Causal Mechanisms from Experimental and Observational Studies
Unpacking the Black-Box: Learning about Causal Mechanisms from Experimental and Observational Studies Kosuke Imai Princeton University September 14, 2010 Institute of Statistical Mathematics Project Reference
More informationEmpirical likelihood methods in missing response problems and causal interference
The University of Toledo The University of Toledo Digital Repository Theses and Dissertations 2017 Empirical likelihood methods in missing response problems and causal interference Kaili Ren University
More informationGauge Plots. Gauge Plots JAPANESE BEETLE DATA MAXIMUM LIKELIHOOD FOR SPATIALLY CORRELATED DISCRETE DATA JAPANESE BEETLE DATA
JAPANESE BEETLE DATA 6 MAXIMUM LIKELIHOOD FOR SPATIALLY CORRELATED DISCRETE DATA Gauge Plots TuscaroraLisa Central Madsen Fairways, 996 January 9, 7 Grubs Adult Activity Grub Counts 6 8 Organic Matter
More informationSingle-level Models for Binary Responses
Single-level Models for Binary Responses Distribution of Binary Data y i response for individual i (i = 1,..., n), coded 0 or 1 Denote by r the number in the sample with y = 1 Mean and variance E(y) =
More informationUniversity of California, Berkeley
University of California, Berkeley U.C. Berkeley Division of Biostatistics Working Paper Series Year 2015 Paper 334 Targeted Estimation and Inference for the Sample Average Treatment Effect Laura B. Balzer
More informationData Integration for Big Data Analysis for finite population inference
for Big Data Analysis for finite population inference Jae-kwang Kim ISU January 23, 2018 1 / 36 What is big data? 2 / 36 Data do not speak for themselves Knowledge Reproducibility Information Intepretation
More informationSection 7: Local linear regression (loess) and regression discontinuity designs
Section 7: Local linear regression (loess) and regression discontinuity designs Yotam Shem-Tov Fall 2015 Yotam Shem-Tov STAT 239/ PS 236A October 26, 2015 1 / 57 Motivation We will focus on local linear
More informationConfidence Intervals for Low-dimensional Parameters with High-dimensional Data
Confidence Intervals for Low-dimensional Parameters with High-dimensional Data Cun-Hui Zhang and Stephanie S. Zhang Rutgers University and Columbia University September 14, 2012 Outline Introduction Methodology
More informationHarvard University. Harvard University Biostatistics Working Paper Series
Harvard University Harvard University Biostatistics Working Paper Series Year 2015 Paper 197 On Varieties of Doubly Robust Estimators Under Missing Not at Random With an Ancillary Variable Wang Miao Eric
More informationSupplementary Materials for Residual Balancing: A Method of Constructing Weights for Marginal Structural Models
Supplementary Materials for Residual Balancing: A Method of Constructing Weights for Marginal Structural Models Xiang Zhou Harvard University Geoffrey T. Wodtke University of Toronto March 29, 2019 A.
More informationCausal Inference for Mediation Effects
Causal Inference for Mediation Effects by Jing Zhang B.S., University of Science and Technology of China, 2006 M.S., Brown University, 2008 A Dissertation Submitted in Partial Fulfillment of the Requirements
More information