Handbook of Matching and Weighting Adjustments for Causal Inference
... First, to validate the PSM model configuration, we conducted a placebo test and performed multiple linear regressions on the matched samples. Following the methodologies of Imbens (2014), Smith and Todd (2001), and Zubizarreta et al. (2023), we randomly shuffled all samples from the treatment group and then randomly selected half of them to serve as a pseudo-control group. Using the original PSM configuration, we recalculated the green premium results to determine if the observed treatment effects were due to statistical noise, coincidence, or model specification errors, which might lead to significant treatment effects even with a randomly constructed pseudo-control group. ...
Impact investing urges investors to weigh the social and environmental impacts of their investment decisions. However, in practice, it remains unclear whether investors in financial products are driven by ethical motivations, such as environmental considerations, and what factors influence their trust in the non-financial aspects (e.g., green attributes) of these investments. This study investigates the ethical motivations behind investors’ decisions to invest in green bonds using a machine learning-assisted causal inference framework based on data collected on all green and conventional bonds issued worldwide from 2007 to 2022. It also explores the underlying factors contributing to investors’ trust in green bonds by examining four perspectives: the issuer’s environmental performance, the transparency and governance of environmental disclosure related to bonds, and the financing purpose of the bonds. The results indicate that (1) investors are willing to forgo financial gains for environmental causes, with this sacrifice quantified as an 18 basis points (bps) green premium, demonstrating a clear ethical motivation; (2) the credibility of bond information, financing purposes, and issuer’s greenhouse gas emission intensity directly influence investors’ trust in green bonds, while bond information disclosure and the issuer’s environmental (E) scores have only indirect effects; and (3) companies with weaker environmental performance often adopt proactive disclosure or certification policies to bolster investors’ green trust. This study is the first to explore the factors influencing investors’ trust in green bonds and to analyze the overall causal transmission mechanism among these factors using a causal inference framework.
... Cardinality matching differs from minimum distance matching, in which the size of the design is fixed in advance and a minimum distance match of that size is constructed. Both are discrete optimization problems, but in cardinality matching the sample size is maximized subject to constraints on covariate balance, while in minimum distance matching a total within-block covariate distance is minimized subject to constraints on sample size and covariate balance; see Rosenbaum & Zubizarreta (2023). ...
In experimental design, aliasing of effects occurs in fractional factorial experiments, where certain low order factorial effects are indistinguishable from certain high order interactions: low order contrasts may be orthogonal to one another, while their higher order interactions are aliased and not identified. In observational studies, aliasing occurs when certain combinations of covariates -- e.g., time period and various eligibility criteria for treatment -- perfectly predict the treatment that an individual will receive, so a covariate combination is aliased with a particular treatment. In this situation, when a contrast among several groups is used to estimate a treatment effect, collections of individuals defined by contrast weights may be balanced with respect to summaries of low-order interactions between covariates and treatments, but necessarily not balanced with respect to summaries of high-order interactions between covariates and treatments. We develop a theory of aliasing in observational studies, illustrate that theory in an observational study whose aliasing is more robust than conventional difference-in-differences, and develop a new form of matching to construct balanced confounded factorial designs from observational data.
... In this case, it is necessary to find counterfactual observations, i.e. similar observations (good with similar characteristics) with a different status, to proxies what could have been observed as outcomes if the status were different. There exist different ways to define counterfactuals (see Zubizarreta et al., 2023), but one of the most famous is based on the propensity score analysis (Figure 1). ...
Purpose- This paper aims to estimate the price premium for a sea view on room rent in a Nordic context, i.e. where proximity to the sea is not valued for the presence of swimmable beaches and suntanning activities. The analysis also explores regional and seasonal variations in price premiums.
Design/methodology/approach- To do so, the study uses information from a Web search of room rents during winter and summer peak seasons. The investigation is based on hotels located along the St. Lawrence River in the Province of Quebec (Canada), where about 40 to 60 km separate both shores. A matching procedure and hedonic pricing models are used to identify the causal impact of a sea view on individual room rents.
Findings- Results suggest that the view price premium varies between 0% and 20%. It is relatively stable on the North Shore, but varies highly on the South Shore, where touristic activities are mainly operating in summertime. The estimation suggests a median local economic benefit of about $30.1M/year.
Practical implications- The analysis reveals that a hedonic pricing model might fail to identify causal effects, especially if it does not account for hotel characteristics. A multiple linear regression model does not ensure a causal interpretation if it neglects unobserved characteristics correlated with the view.
Originality/value- The paper proposes a matching identification procedure accounting for spatial confounding to retrieve the causal impact of the view of the sea on hotel room rents. A heterogeneity analysis suggests that view price premium on room rent can vary within seasons but mainly across regions, even for the same amenities.
... This study uses propensity score matching (PSM) to address the selection bias that occurs because of comparisons between treated and control units. Propensity scores serve as a helpful method for accounting for observed covariates in observational studies focused on determining causal effects [10]. ...
Slum dwellers often fail to expend enough on waste disposal to have a healthy life, and remittances provide funds for a household to expend on the betterment of life conditions. This paper examines the impact of remittances on total monthly waste disposal expenditures and the impact of better waste disposal on monthly health expenditures among slum households in Bangladesh. Propensity score matching was adopted as an identification strategy to reduce selection bias. In this study, remittance includes both remittances received from within and outside Bangladesh. Remittance receipt equals 1 if a household receives remittances from within and/or outside of Bangladesh. Our results show that remittances increase expenditures on waste disposal by 28.77% to 32.74% among slum households in Bangladesh. Waste disposal expenditure is considered as an indicator of better waste disposal. Furthermore, we find that better waste disposal reduces total outpatient expenditures. A reduction in outpatient expenditure indicates that waste disposal results in better health conditions for slum dwellers. The findings of this study can be connected to Sustainable Development Goal 11, which targets sustainable cities and communities and suggests that remittances are a bottom-up financial mechanism for improving waste disposal at the micro level to improve health status.
Graphical abstract
... Various methods have been proposed to assess the impact of omitted variable bias when all relevant covariates are not included. However, implementing all the methods presented in Fig. 3 may not be practical [21][22][23]. ...
Recently, there has been considerable progress in developing new technologies and equipment for the medical field, including minimally invasive surgeries. Evaluating the effectiveness of these treatments requires study designs like randomized controlled trials. However, due to the nature of certain treatments, randomization is not always feasible, leading to the use of observational studies. The effect size estimated from observational studies is subject to selection bias caused by confounders. One method to reduce this bias is propensity scoring. This study aimed to introduce a propensity score matching process between two groups using a practical example with R. Additionally, Rex, an Excel add-in graphical user interface statistical program, is provided for researchers unfamiliar with R programming. Further techniques, such as matching with three or more groups, propensity score weighting and stratification, and imputation of missing values, are summarized to offer approaches for more complex studies not covered in this tutorial.
... Correcting for such possibilities can be done using two distinct avenues. A first one is to implement, before any analysis, a matching approach (Rosenbaum, 2023;Rosenbaum and Rubin, 1983;Angrist and Pischke, 2009;Heckman and Todd, 2009;Stuart et al., 2014) to select transactions with characteristics that are similar to the global stock. The propensity score matching is one of the more popular methods of matching. ...
Many empirical applications aim to isolate the impact of implementing new public transport on real estate uplift value. While their conclusions generally point to positive impact, due to reduction in transportation cost, the methodological framework to investigate uplift value has largely evolved over time. This paper reviews the different methodological challenges in measuring causal uplift value and proposes an adjusted parametric approach inspired from the Alonso-Muth-Mills model, returning a complex 2-D price premium function allowing for spatial heterogeneous patterns of the average treatment effect. The proposed framework also accounts for other methodological challenges underlined by literature such as spatial autocorrelation, selectivity and representativity issues, and possible anticipation effects. To illustrate the importance of methodological choices on estimation results, the framework is applied to the case of the implementation of a bus rapid transit (BRT) system in Québec City, a medium-size Canadian city, as a specific case study.
... To illustrate the studied methods for PS weighting with clustered data structure and missing data on covariates, we use a sample from the Education Longitudinal Study (ELS; Ingels For illustrative purposes, a total of 14 student-level covariates were selected for the PS analysis, considering that they might influence the treatment assignment and outcome (Austin, 2011;Brookhart et al., 2006;Zubizarreta et al., 2023). The selected covariates were all collected in the base year. ...
Propensity score (PS) analyses are increasingly popular in behavioral sciences. Two issues often add complexities to PS analyses, including missing data in observed covariates and clustered data structure. In previous research, methods for conducting PS analyses with considering either issue alone were examined. In practice, the two issues often co-occur; but the performance of methods for PS analyses in the presence of both issues has not been evaluated previously. In this study, we consider PS weighting analysis when data are clustered and observed covariates have missing values. A simulation study is conducted to evaluate the performance of different missing data handling methods (complete-case, single-level imputation, or multilevel imputation) combined with different multilevel PS weighting methods (fixed- or random-effects PS models, inverse-propensity-weighting or the clustered weighting, weighted single-level or multilevel outcome models). The results suggest that the bias in average treatment effect estimation can be reduced, by better accounting for clustering in both the missing data handling stage (such as with the multilevel imputation) and the PS analysis stage (such as with the fixed-effects PS model, clustered weighting, and weighted multilevel outcome model). A real-data example is provided for illustration.
This study develops methods for evaluating a treatment effect on a time-to-event outcome in matched-pair studies. While most methods for paired right-censored outcomes allow determining an overall treatment effect over the course of follow-up, they generally lack in providing detailed insights into how the effect changes over time. To address this gap, we propose time-specific and overall tests for paired right-censored outcomes under randomization inference. We further extend our tests to matched observational studies by developing corresponding sensitivity analysis methods to take into account departures from randomization. Simulations demonstrate the robustness of our approach against various non-proportional hazards alternatives, including a crossing survival curves scenario. We demonstrate the application of our methods using a matched observational study from the Korean Longitudinal Study of Aging (KLoSA) data, focusing on the effect of social engagement on survival.
We introduce a new convexified matching method for missing value imputation and individualized inference inspired by computational optimal transport. Our method integrates favorable features from mainstream imputation approaches: optimal matching, regression imputation, and synthetic control. We impute counterfactual outcomes based on convex combinations of observed outcomes, defined based on an optimal coupling between the treated and control data sets. The optimal coupling problem is considered a convex relaxation to the combinatorial optimal matching problem. We estimate granular-level individual treatment effects while maintaining a desirable aggregate-level summary by properly constraining the coupling. We construct transparent, individual confidence intervals for the estimated counterfactual outcomes. We devise fast iterative entropic-regularized algorithms to solve the optimal coupling problem that scales favorably when the number of units to match is large. Entropic regularization plays a crucial role in both inference and computation; it helps control the width of the individual confidence intervals and design fast optimization algorithms.
Background
There is a lack of evidence associating intraoperative transoesophageal echocardiography (TOE) use with improved outcomes among coronary artery bypass graft (CABG) surgery subpopulations.
Methods
This matched retrospective cohort study used a US private claims dataset to compare outcomes among different CABG surgery patient populations with vs without TOE. Statistical analyses involved exact matching on pre-selected subgroups (congestive heart failure, single vessel, and multivessel CABG) and used fine and propensity-score balanced techniques to conduct multiple matched comparisons and sensitivity analyses.
Results
Of 42 249 patients undergoing isolated CABG surgery, 24 919 (59.0%) received and 17 330 (41.0%) did not receive TOE. After matching, intraoperative TOE was significantly associated with a lower, 30-day mortality: 2.63% vs 3.20% (odds ratio [OR]: 0.81; 95% confidence interval [CI]: 0.71–0.92; P=0.002). In the subgroup matched comparisons, intraoperative TOE was significantly associated with a lower, 30-day mortality rate among those with congestive heart failure: 4.20% vs 5.26% (OR: 0.78; 95% CI: 0.66–0.94; P=0.007) and among those undergoing multivessel CABG with congestive heart failure: 4.23% vs 5.24% (OR: 0.80; 95% CI: 0.65–0.97; P=0.025), but not among those undergoing multivessel CABG without congestive heart failure: 1.83% vs 2.15% (OR: 0.85; 95% CI: 0.70–1.02; P=0.089, nor any of the remaining three subgroups.
Conclusions
Among US adults undergoing isolated CABG surgery, intraoperative TOE was associated with improved outcomes in patients with congestive heart failure (vs without) and among patients undergoing multivessel (vs single vessel) CABG. These findings support prioritised TOE allocation to these patient populations at centres with limited TOE capabilities.
ResearchGate has not been able to resolve any references for this publication.