Practical Propensity Score Methods Using R

Leite, Walter L.

doi:10.4135/9781071802854

Cited by 118 publications

(197 citation statements)

References 0 publications

Supporting

Mentioning

188

Contrasting

Unclassified

Order By: Relevance

“…In single-level data, propensity scores are typically estimated with logistic regression. In multilevel data, propensity scores are typically estimated with random or fixed effects logistic regression (Leite, 2016). Propensity scores are often used in matching methods to match treated and control units or to weigh individuals' outcomes via inverse probability weighing.…”

Section: Prior Work and Our Contributionmentioning

confidence: 99%

Hybridizing Machine Learning Methods and Finite Mixture Models for Estimating Heterogeneous Treatment Effects in Latent Classes

Suk¹,

Kim²,

Kang

2020

Journal of Educational and Behavioral Statistics

View full text Add to dashboard Cite

There has been increasing interest in exploring heterogeneous treatment effects using machine learning (ML) methods such as causal forests, Bayesian additive regression trees, and targeted maximum likelihood estimation. However, there is little work on applying these methods to estimate treatment effects in latent classes defined by well-established finite mixture/latent class models. This article proposes a hybrid method, a combination of finite mixture modeling and ML methods from causal inference to discover effect heterogeneity in latent classes. Our simulation study reveals that hybrid ML methods produced more precise and accurate estimates of treatment effects in latent classes. We also use hybrid ML methods to estimate the differential effects of private lessons across latent classes from Trends in International Mathematics and Science Study data.

show abstract

Section: Prior Work and Our Contributionmentioning

confidence: 99%

Hybridizing Machine Learning Methods and Finite Mixture Models for Estimating Heterogeneous Treatment Effects in Latent Classes

Suk¹,

Kim²,

Kang

2020

Journal of Educational and Behavioral Statistics

View full text Add to dashboard Cite

show abstract

“…A key consideration in using weighting estimators in multilevel data is a model for the propensity score e(X ij , W j ). Broadly speaking, in two-level data, there are two main strategies for modeling the propensity score and estimating the ATE: a within-cluster strategy and an across-cluster strategy (Leite, 2016;Steiner et al, 2012). A within-cluster strategy estimates a propensity score model for each cluster with only individual-level covariates.…”

Section: Multilevel Propensity Score Methods Via Weightingmentioning

confidence: 99%

“…Propensity score methods utilize propensity scores to balance covariates between treated and control groups by using matching, stratification, or weighting. Multilevel propensity score methods use propensity scores, but also account for the underlying clustered/hierarchical structure in multilevel data, typically by fitting a multilevel propensity score model; see Section 10 of Leite (2016) for details. If selection is strongly ignorable, i.e., the treatment assignment is as-if random conditional on observed pre-treatment covariates, and the propensity score model is correctly specified, these aforementioned methods can consistently estimate the ATE.…”

Section: Introductionmentioning

confidence: 99%

Random Forests Approach for Causal Inference with Clustered Observational Data

Suk

Kang

Kim

2020

Multivariate Behavioral Research

View full text Add to dashboard Cite

There is a growing interest in using machine learning (ML) methods for causal inference due to their (nearly) automatic and flexible ability to model key quantities such as the propensity score or the outcome model. Unfortunately, most ML methods for causal inference have been studied under single-level settings where all individuals are independent of each other and there is little work in using these methods with clustered or nested data, a common setting in education studies. This paper investigates using one particular ML method based on random forests known as Causal Forests to estimate treatment effects in multilevel observational data. We conduct simulation studies under different types of multilevel data, including two-level, three-level, and cross-classified data. Our simulation study shows that when the ML method is supplemented with estimated propensity scores from multilevel models that account for clustered/hierarchical structure, the modified ML method outperforms pre-existing methods in a wide variety of settings. We conclude by estimating the effect of private math lessons in the Trends in International Mathematics and Science Study data, a large-scale educational assessment where students are nested within schools.

show abstract

“…Relevant covariates for the PS model, meaning those covariates causally related to treatment condition and to the primary outcome of social inclusion, will be selected according to theoretical assumptions and empirical findings [78]. The propensity scores will be obtained by logistic regression [76,79].…”

Section: Baseline Characteristicsmentioning

confidence: 99%

Independent Housing and Support for non-homeless individuals with severe mental illness: randomised controlled trial vs. observational study – study protocol

et al. 2020

View full text Add to dashboard Cite

Background: Social inclusion is essential for an adequate rehabilitation process for people with serious mental illness (SMI). Various supported housing settings aim to promote housing competencies and social inclusion in service users. Nevertheless, there is a strong preference in service users for independent living. We aim to evaluate the effectiveness and efficiency of Independent Housing and Support (IHS) compared to institutionalised residential care settings and other treatment as usual conditions (RCS/TAU) in two cities in Switzerland. Methods: This is a prospective multi-centre, four-arm, non-inferiority cohort study investigating the effectiveness and efficiency of IHS and RCS/TAU for people with SMI. Effectiveness will be measured by a standardised measure of social inclusion as primary outcome as well as by measures of functioning and well-being. Efficiency will be analysed on the basis of service usage and costs associated with the different housing settings. Participants will be consecutively recruited and subsequently enrolled between April 2019 and December 2020 and assessed at baseline and after six, twelve and after 24 months. At one study site, 56 participants will be randomly assigned to one of the conditions; the other study site will be conducted as an observational study investigating 112 admitted participants. Discussion: While the UN Convention of the Rights of People with Disabilities aims to promote the opportunity to choose one's place of residence, the limited supply of alternative forms of housing does not guarantee genuine freedom of choice. Increased diversification and flexibility of housing support is essential. If IHS shows noninferiority in terms of their effectiveness and efficiency, users should be allowed to choose their kind of housing support.

show abstract

Practical Propensity Score Methods Using R

Cited by 118 publications

References 0 publications

Hybridizing Machine Learning Methods and Finite Mixture Models for Estimating Heterogeneous Treatment Effects in Latent Classes

Hybridizing Machine Learning Methods and Finite Mixture Models for Estimating Heterogeneous Treatment Effects in Latent Classes

Random Forests Approach for Causal Inference with Clustered Observational Data

Independent Housing and Support for non-homeless individuals with severe mental illness: randomised controlled trial vs. observational study – study protocol

Contact Info

Product

Resources

About