Identification Of Causal Effects Research Articles

Causal inference with hidden mediators

Summary Proximal causal inference was recently proposed as a framework to identify causal effects from observational data in the presence of hidden confounders for which proxies are available. In this paper, we extend the proximal causal inference approach to settings where identification of causal effects hinges upon a set of mediators that are not observed, yet error prone proxies of the hidden mediators are measured. Specifically, (i) we establish causal hidden mediation analysis, which extends classical causal mediation analysis methods for identifying natural direct and indirect effects under no unmeasured confounding to a setting where the mediator of interest is hidden, but proxies of it are available; (ii) we establish a hidden front-door criterion, criterion to allow for hidden mediators for which proxies are available; (iii) we show that the identification of a certain causal effect called the population intervention indirect effect remains possible with hidden mediators in settings where challenges in (i) and (ii) might co-exist. We view (i)–(iii) as important steps towards the practical application of front-door criteria and mediation analysis as mediators are almost always measured with error and, thus, the most one can hope for in practice is that the measurements are at best proxies of mediating mechanisms. We propose identification approaches for the parameters of interest in our considered models. For the estimation aspect, we propose an influence function-based estimation method and provide an analysis for the robustness of the estimators.

RoA: visual analytics support for deconfounded causal inference in observational studies

The gold standard in medical research to estimate the causal effect of a treatment is the Randomized Controlled Trial (RCT), but in many cases these are not feasible due to ethical, financial or practical issues. Observational studies are an alternative, but can easily lead to doubtful results, because of unbalanced selection bias and confounding. Moreover, RCTs often only apply to a specific subgroup and cannot readily be extrapolated. In response, we present Rod of Asclepius (RoA), a novel visual analytics method that integrates modern techniques designed for identification of causal effects and effect size estimation with subgroup analysis. The result is an interactive display designed to combine exploratory analysis with a robust set of techniques, including causal do-calculus, propensity score weighting, and effect estimation. It enables analysts to conduct observational studies in an exploratory, yet robust way. This is demonstrated by means of a use case involving patients undergoing surgery, for which we collaborated closely with clinical researchers.

Learning about treatment effects in a new target population under transportability assumptions for relative effect measures.

Investigators often believe that relative effect measures conditional on covariates, such as risk ratios and mean ratios, are "transportable" across populations. Here, we examine the identification of causal effects in a target population using an assumption that conditional relative effect measures are transportable from a trial to the target population. We show that transportability for relative effect measures is largely incompatible with transportability for difference effect measures, unless the treatment has no effect on average or one is willing to make even stronger transportability assumptions that imply the transportability of both relative and difference effect measures. We then describe how marginal (population-averaged) causal estimands in a target population can be identified under the assumption of transportability of relative effect measures, when we are interested in the effectiveness of a new experimental treatment in a target population where the only treatment in use is the control treatment evaluated in the trial. We extend these results to consider cases where the control treatment evaluated in the trial is only one of the treatments in use in the target population, under an additional partial exchangeability assumption in the target population (i.e., an assumption of no unmeasured confounding in the target population with respect to potential outcomes under the control treatment in the trial). We also develop identification results that allow for the covariates needed for transportability of relative effect measures to be only a small subset of the covariates needed to control confounding in the target population. Last, we propose estimators that can be easily implemented in standard statistical software and illustrate their use using data from a comprehensive cohort study of stable ischemic heart disease.

S-ID: Causal Effect Identification in a Sub-population

Causal inference in a sub-population involves identifying the causal effect of an intervention on a specific subgroup, which is distinguished from the whole population through the influence of systematic biases in the sampling process. However, ignoring the subtleties introduced by sub-populations can either lead to erroneous inference or limit the applicability of existing methods. We introduce and advocate for a causal inference problem in sub-populations (henceforth called s-ID), in which we merely have access to observational data of the targeted sub-population (as opposed to the entire population). Existing inference problems in sub-populations operate on the premise that the given data distributions originate from the entire population, thus, cannot tackle the s-ID problem. To address this gap, we provide necessary and sufficient conditions that must hold in the causal graph for a causal effect in a sub-population to be identifiable from the observational distribution of that sub-population. Given these conditions, we present a sound and complete algorithm for the s-ID problem.

Nonparametric identification of causal effects in clustered observational studies with differential selection

Abstract The clustered observational study (COS) design is the observational counterpart to the clustered randomized trial. COSs are common in both education and health services research. In education, treatments may be given to all students within some schools but withheld from all students in other schools. In health studies, treatments may be applied to clusters such as hospitals or groups of patients treated by the same physician. In this paper, we study the identification of causal effects in COS designs. We focus on the prospect of differential selection of units to clusters, which occurs when the units’ cluster selections depend on the clusters’ treatment assignments. Extant work on COSs has made an implicit assumption that rules out the presence of differential selection. We derive the identification results for designs with differential selection and that contexts with differential cluster selection require different adjustment sets than standard designs. We outline estimators for designs with and without differential selection. Using a series of simulations, we outline the magnitude of the bias that can occur with differential selection. We then present 2 empirical applications focusing on the likelihood of differential selection.

Active causal effect identification with expert knowledge

Active causal effect identification with expert knowledge

Causal Effect Identification in Cluster DAGs

Reasoning about the effect of interventions and counterfactuals is a fundamental task found throughout the data sciences. A collection of principles, algorithms, and tools has been developed for performing such tasks in the last decades. One of the pervasive requirements found throughout this literature is the articulation of assumptions, which commonly appear in the form of causal diagrams. Despite the power of this approach, there are significant settings where the knowledge necessary to specify a causal diagram over all variables is not available, particularly in complex, high-dimensional domains. In this paper, we introduce a new graphical modeling tool called cluster DAGs (for short, C-DAGs) that allows for the partial specification of relationships among variables based on limited prior knowledge, alleviating the stringent requirement of specifying a full causal diagram. A C-DAG specifies relationships between clusters of variables, while the relationships between the variables within a cluster are left unspecified, and can be seen as a graphical representation of an equivalence class of causal diagrams that share the relationships among the clusters. We develop the foundations and machinery for valid inferences over C-DAGs about the clusters of variables at each layer of Pearl's Causal Hierarchy - L1 (probabilistic), L2 (interventional), and L3 (counterfactual). In particular, we prove the soundness and completeness of d-separation for probabilistic inference in C-DAGs. Further, we demonstrate the validity of Pearl's do-calculus rules over C-DAGs and show that the standard ID identification algorithm is sound and complete to systematically compute causal effects from observational data given a C-DAG. Finally, we show that C-DAGs are valid for performing counterfactual inferences about clusters of variables.

Estimating the Effect of a Treatment When There Is Nonadherence in a Trial.

Randomized trials offer a powerful strategy for estimating the effect of a treatment on an outcome. However, interpretation of trial results can be complicated when study subjects do not take the treatment to which they were assigned; this is referred to as nonadherence. Prior authors have described instrumental variable approaches to analyze trial data with nonadherence; under their approaches, the initial assignment to treatment is used as an instrument. However, their approaches require the assumption that initial assignment to treatment has no direct effect on the outcome except via the actual treatment received (i.e., the exclusion restriction), which may be implausible. We propose an approach to identification of a causal effect of treatment in a trial with 1-sided nonadherence without assuming exclusion restriction. The proposed approach leverages the study subjects initially assigned to control status as an unexposed reference population; we then employ a bespoke instrumental variable analysis, where the key assumption is "partial exchangeability" of the association between a covariate and an outcome in the treatment and control arms. We provide a formal description of the conditions for identification of causal effects, illustrate the method using simulations, and provide an empirical application.

Recent Developments in Causal Inference and Machine Learning.

This article reviews recent advances in causal inference relevant to sociology. We focus on a selective subset of contributions aligning with four broad topics: causal effect identification and estimation in general, causal effect heterogeneity, causal effect mediation, and temporal and spatial interference. We describe how machine learning, as an estimation strategy, can be effectively combined with causal inference, which has been traditionally concerned with identification. The incorporation of machine learning in causal inference enables researchers to better address potential biases in estimating causal effects and uncover heterogeneous causal effects. Uncovering sources of effect heterogeneity is key for generalizing to populations beyond those under study. While sociology has long emphasized the importance of causal mechanisms, historical and life-cycle variation, and social contexts involving network interactions, recent conceptual and computational advances facilitate more principled estimation of causal effects under these settings. We encourage sociologists to incorporate these insights into their empirical research.

Educated politicians and government efficiency: Evidence from Norwegian local government

The paper studies the effect of politicians’ education levels on government efficiency. Using data on Norwegian local government, the paper measures efficiency by DEA-productivity indicators and comprehensive indicators of service production. The identification of causal effects exploits close (within-list) elections to design an instrumental variable for council shares with higher education.Consistent with survey data, the estimates show that better-educated politicians induce higher levels of efficiency, particularly for old-age care services. A similar empirical strategy indicates that political experience has small efficiency effects. Educational misrepresentation has modest effects on fiscal allocations and is mostly canceled out by the efficiency gains.

On the reliability of published findings using the regression discontinuity design in political science

The regression discontinuity (RD) design offers identification of causal effects under weak assumptions, earning it a position as a standard method in modern political science research. But identification does not necessarily imply that causal effects can be estimated accurately with limited data. In this paper, we highlight that estimation under the RD design involves serious statistical challenges and investigate how these challenges manifest themselves in the empirical literature in political science. We collect all RD-based findings published in top political science journals in the period 2009–2018. The distribution of published results exhibits pathological features; estimates tend to bunch just above the conventional level of statistical significance. A reanalysis of all studies with available data suggests that researcher discretion is not a major driver of these features. However, researchers tend to use inappropriate methods for inference, rendering standard errors artificially small. A retrospective power analysis reveals that most of these studies were underpowered to detect all but large effects. The issues we uncover, combined with well-documented selection pressures in academic publishing, cause concern that many published findings using the RD design may be exaggerated.

School-age vaccination, school openings and Covid-19 diffusion.

This article investigates the relationship between school openings and Covid-19 diffusion when school-age vaccination becomes available. The analysis relies on a unique geo-referenced high frequency database on age of vaccination, Covid-19 cases and hospitalization indicators from the Italian region of Sicily. The study focuses on the change of Covid-19 diffusion after school opening in a homogeneous geographical territory (i.e., with the same control measures and surveillance systems, centrally coordinated by the Regional Government). The identification of causal effects derives from a comparison of the change in cases before and after school opening in the school year 2020/21, when vaccination was not available, and in 2021/22, when the vaccination campaign targeted individuals of age 12-19 and above 19. Results indicate that, while school opening determined an increase in the growth rate of Covid-19 cases in 2020/2021, this effect has been substantially reduced by school-age vaccination in 2021/2022. In particular, we find that an increase of approximately 10% in the vaccination rate of school-age population reduces the growth rate of Covid-19 cases after school opening by approximately 1%.

An imperfect wealth tax and employment in closely held firms

AbstractFuelled by increasing inequality and rising fiscal deficits, the interest in wealth taxation has grown over recent years, both in the public debate and in academia. A key concern is that the wealth tax may reduce the amount of capital available to closely held firms and drag down their employment. Yet knowledge about the behavioural effects of a wealth tax is limited. A wealth tax is almost by construction imperfect, as the value of some assets is unobserved. In particular, intangible assets held by non‐traded firms are in practice tax‐exempt, giving firm owners an incentive to allocate wealth into their businesses, for example, in the form of (untaxed) human capital investments. We utilize rich Norwegian register data and a series of tax reforms implemented between 2007 and 2017 to study how a net wealth tax imposed on owners of small and medium‐sized businesses affects their firms' employment. Identification of causal effects is based on a saturated control function approach, fully isolating the influence of tax reforms. Our results indicate a positive causal relationship between the level of a household's wealth tax and subsequent employment growth in the taxpayers' closely held firms.

Are richer neighborhoods always better for the kids?

Abstract Based on Norwegian administrative registers, we provide new empirical evidence on the effects of the childhood neighborhood’s socioeconomic status on early educational performance. A neighborhood’s status is measured annually by its adult inhabitants’ earnings ranks within larger commuting zones, and the childhood neighborhood status is the average status of the neighborhoods inhabited from the year after birth to age 15. Identification of causal effects relies on within-family comparisons only. Our results reveal a distinct hump-shaped relationship between the socioeconomic status of the childhood neighborhood and school results at age 15–16, such that the optimal neighborhood is of medium rank.

Nonparametric difference-in-differences in repeated cross-sections with continuous treatments

This paper studies the identification of causal effects of a continuous treatment using a new difference-in-difference strategy. Our approach allows for endogeneity of the treatment, and employs repeated cross-sections. It requires an exogenous change over time which affects the treatment in a heterogeneous way, stationarity of the distribution of unobservables and a rank invariance condition on the time trend. On the other hand, we do not impose any functional form restrictions or an additive time trend, and we are invariant to the scaling of the dependent variable. Under our conditions, the time trend can be identified using a control group, as in the binary difference-in-differences literature. In our scenario, however, this control group is defined by the data. We then identify average and quantile treatment effect parameters. We develop corresponding nonparametric estimators and study their asymptotic properties. Finally, we apply our results to the effect of disposable income on consumption.

Do Class Size Reductions Protect Students from Infectious Diseases?

We evaluate the causal effect of class size (number of students in a classroom) on incidence of class closure due to the flu, as an outcome of an infectious disease epidemic. For identification of causal effects, we apply a regression discontinuity design using discontinuous variation of class sizes, around the class size cap set by regulation, to administrative data of public primary and middle school students in one of the largest municipalities within the Tokyo metropolitan area from 2015 to 2017. Most classrooms in Japan are constructed in accordance with a standard of classroom area, 63 square meters; class size reduction improves social distancing among students in a classroom. We find that class size reduction is effective in reducing class closures due to the flu: a one-unit reduction of class size decreases class closure by about 5 percent. Additionally, forming small classes with 27 students at most, satisfying the social distancing of 1.5 meters recommended to prevent droplet infection including influenza and COVID-19, reduces class closure by about 90 percent. Moreover, we find that the older the students, the larger the effects of class size reduction. Our findings provide evidence for the effectiveness of social distancing policy in primary and middle schools to protect students from droplet infectious disease spread, including COVID-19.

A note on identification of causal effects in cluster randomized trials with post-randomization selection bias

Cluster randomized trials are commonly used for evaluating treatment effects of a cluster-level intervention. Because subjects are often recruited after cluster randomization, a main complication is the potential for selection bias due to the intervention affecting which subjects are identified as being eligible for receiving the actual treatment. In the presence of such post-randomization selection bias, direct comparison of the observed outcomes among eligible subjects between arms does not yield a valid causal effect estimate. In this note, we define average and quantile causal estimands using principal stratification in the context of post-randomization selection bias, and provide nonparametric identification assumptions and formulas for these estimands.

Negative Control Exposures: Causal Effect Identifiability and Use in Probabilistic-bias and Bayesian Analyses With Unmeasured Confounders.

Probabilistic bias and Bayesian analyses are important tools for bias correction, particularly when required parameters are nonidentifiable. Negative controls are another tool; they can be used to detect and correct for confounding. Our goals are to present conditions that assure identifiability of certain causal effects and to describe and illustrate a probabilistic bias analysis and related Bayesian analysis that use a negative control exposure. Using potential-outcome models, we characterized assumptions needed for identification of causal effects using a dichotomous, negative control exposure when residual confounding exists. We defined bias parameters, characterized their relationships with the negative control and with specified causal effects, and described the corresponding probabilistic-bias and Bayesian analyses. We present analytic examples using data on hormone therapy and suicide attempts among transgender people. To address possible confounding by healthcare utilization, we used prior tetanus-diphtheria-pertussis (TdaP) vaccination as a negative control exposure. Hormone therapy was weakly associated with risk (risk ratio [RR] = 0.9). The negative control exposure was associated with risk (RR = 1.7), suggesting confounding. Based on an assumed prior distribution for the bias parameter, the 95% simulation interval for the distribution of confounding-adjusted RR was (0.17, 1.6), with median 0.5; the 95% credibility interval was similar. We used dichotomous negative control exposure to identify causal effects when a confounder was unmeasured under strong assumptions. It may be possible to relax assumptions and the negative control exposure could prove helpful for probabilistic bias analyses and Bayesian analyses.

Digital transformation, green innovation and the Solow productivity paradox.

To provide evidence at the micro level for cracking the Solow productivity paradox, this paper deeply studies the impact of enterprise digital transformation on green innovation. In terms of theoretical research, three potential mechanisms are excavated for the first time; considering empirical research, a series of strict causal effect identification strategies are carried out. The results show that enterprise digital transformation can significantly promote green innovation, and it passes a series of robustness tests and endogenous tests. According to the theoretical and empirical results, the policy suggestions mainly include five points: helping enterprises to accelerate digital transformation; strengthening the green innovation ability of enterprises; reducing internal and external costs and promoting the professional division of labor; piloting the digital transformation policy; enhancing corporate social responsibility. It provides a reference of experience and a path for other countries to follow in implementing a digital transformation strategy and green sustainable development strategy.

The effect of higher-achieving peers on major choices and labor market outcomes

This paper investigates how exposure to higher-achieving male and female peers in university affects students’ major choices and labor market outcomes. For identification of causal effects, we exploit the random assignment of students to university sections in compulsory first-year courses. We present two main results. First, studying with higher-achieving peers has no statistically significant or economically meaningful effects on educational choices. Second, we find suggestive evidence that women who have been exposed to higher- achieving male peers end up in jobs in which they are more satisfied.