Three critical issues for causal inference that often occur in modern, complicated experiments are interference, treatment nonadherence, and missing outcomes. A great deal of research efforts has been dedicated to developing causal inferential methodologies that address these issues separately. However, methodologies that can address these issues simultaneously are lacking. We propose a Bayesian causal inference methodology to address this gap. Our methodology extends existing causal frameworks and methods, specifically, two-staged randomized experiments and the principal stratification framework. In contrast to existing methods that invoke strong structural assumptions to identify principal causal effects, our Bayesian approach uses flexible distributional models that can accommodate the complexities of interference and missing outcomes, and that ensure that principal causal effects are weakly identifiable. We illustrate our methodology via simulation studies and a re-analysis of real-life data from an evaluation of India's National Health Insurance Program. Our methodology enables us to identify new significant causal effects that were not identified in past analyses. Ultimately, our simulation studies and case study demonstrate how our methodology can yield more informative analyses in modern experiments with interference, treatment nonadherence, missing outcomes, and complicated outcome generation mechanisms.
相關內容
The literature on treatment choice focuses on the mean of welfare regret. Ignoring other features of the regret distribution, however, can lead to an undesirable rule that suffers from a high chance of welfare loss due to sampling uncertainty. We propose to minimize the mean of a nonlinear transformation of welfare regret. This paradigm shift alters optimal rules drastically. We show that for a wide class of nonlinear criteria, admissible rules are fractional. Focusing on mean square regret, we derive the closed-form probabilities of randomization for finite-sample Bayes and minimax optimal rules when data are normal with known variance. The minimax optimal rule is a simple logit based on the sample mean and agrees with the posterior probability for positive treatment effect under the least favorable prior. The Bayes optimal rule with an uninformative prior is different but produces quantitatively comparable mean square regret. We extend these results to limit experiments and discuss our findings through sample size calculations.
The fundamental challenge of drawing causal inference is that counterfactual outcomes are not fully observed for any unit. Furthermore, in observational studies, treatment assignment is likely to be confounded. Many statistical methods have emerged for causal inference under unconfoundedness conditions given pre-treatment covariates, including propensity score-based methods, prognostic score-based methods, and doubly robust methods. Unfortunately for applied researchers, there is no `one-size-fits-all' causal method that can perform optimally universally. In practice, causal methods are primarily evaluated quantitatively on handcrafted simulated data. Such data-generative procedures can be of limited value because they are typically stylized models of reality. They are simplified for tractability and lack the complexities of real-world data. For applied researchers, it is critical to understand how well a method performs for the data at hand. Our work introduces a deep generative model-based framework, Credence, to validate causal inference methods. The framework's novelty stems from its ability to generate synthetic data anchored at the empirical distribution for the observed sample, and therefore virtually indistinguishable from the latter. The approach allows the user to specify ground truth for the form and magnitude of causal effects and confounding bias as functions of covariates. Thus simulated data sets are used to evaluate the potential performance of various causal estimation methods when applied to data similar to the observed sample. We demonstrate Credence's ability to accurately assess the relative performance of causal estimation techniques in an extensive simulation study and two real-world data applications from Lalonde and Project STAR studies.
We propose a doubly robust approach to characterizing treatment effect heterogeneity in observational studies. We develop a frequentist inferential procedure that utilizes posterior distributions for both the propensity score and outcome regression models to provide valid inference on the conditional average treatment effect even when high-dimensional or nonparametric models are used. We show that our approach leads to conservative inference in finite samples or under model misspecification, and provides a consistent variance estimator when both models are correctly specified. In simulations, we illustrate the utility of these results in difficult settings such as high-dimensional covariate spaces or highly flexible models for the propensity score and outcome regression. Lastly, we analyze environmental exposure data from NHANES to identify how the effects of these exposures vary by subject-level characteristics.
Since the average treatment effect (ATE) measures the change in social welfare, even if positive, there is a risk of negative effect on, say, some 10% of the population. Assessing such risk is difficult, however, because any one individual treatment effect (ITE) is never observed, so the 10% worst-affected cannot be identified, while distributional treatment effects only compare the first deciles within each treatment group, which does not correspond to any 10%-subpopulation. In this paper we consider how to nonetheless assess this important risk measure, formalized as the conditional value at risk (CVaR) of the ITE-distribution. We leverage the availability of pre-treatment covariates and characterize the tightest-possible upper and lower bounds on ITE-CVaR given by the covariate-conditional average treatment effect (CATE) function. We then proceed to study how to estimate these bounds efficiently from data and construct confidence intervals. This is challenging even in randomized experiments as it requires understanding the distribution of the unknown CATE function, which can be very complex if we use rich covariates so as to best control for heterogeneity. We develop a debiasing method that overcomes this and prove it enjoys favorable statistical properties even when CATE and other nuisances are estimated by black-box machine learning or even inconsistently. Studying a hypothetical change to French job-search counseling services, our bounds and inference demonstrate a small social benefit entails a negative impact on a substantial subpopulation.
Propensity score weighting is widely used to improve the representativeness and correct the selection bias in the voluntary sample. The propensity score is often developed using a model for the sampling probability, which can be subject to model misspecification. In this paper, we consider an alternative approach of estimating the inverse of the propensity scores using the density ratio function satisfying the self-efficiency condition. The smoothed density ratio function is obtained by the solution to the information projection onto the space satisfying the moment conditions on the balancing scores. By including the covariates for the outcome regression models only in the density ratio model, we can achieve efficient propensity score estimation. Penalized regression is used to identify important covariates. We further extend the proposed approach to the multivariate missing case. Some limited simulation studies are presented to compare with the existing methods.
Existing literature on constructing optimal regimes often focuses on intention-to-treat analyses that completely ignore the compliance behavior of individuals. Instrumental variable-based methods have been developed for learning optimal regimes under endogeneity. However, when there are two active treatment arms, the average causal effects of treatments cannot be identified using a binary instrument, and thus the existing methods will not be applicable. To fill this gap, we provide a procedure that identifies an optimal regime and the corresponding value function as a function of a vector of sensitivity parameters. We also derive the canonical gradient of the target parameter and propose a multiply robust classification-based estimator of the optimal regime. Our simulations highlight the need for and usefulness of the proposed method in practice. We implement our method on the Adaptive Treatment for Alcohol and Cocaine Dependence randomized trial.
The growth of wind generation capacities in the past decades has shown that wind energy can contribute to the energy transition in many parts of the world. Being highly variable and complex to model, the quantification of the spatio-temporal variation of wind power and the related uncertainty is highly relevant for energy planners. Machine Learning has become a popular tool to perform wind-speed and power predictions. However, the existing approaches have several limitations. These include (i) insufficient consideration of spatio-temporal correlations in wind-speed data, (ii) a lack of existing methodologies to quantify the uncertainty of wind speed prediction and its propagation to the wind-power estimation, and (iii) a focus on less than hourly frequencies. To overcome these limitations, we introduce a framework to reconstruct a spatio-temporal field on a regular grid from irregularly distributed wind-speed measurements. After decomposing data into temporally referenced basis functions and their corresponding spatially distributed coefficients, the latter are spatially modelled using Extreme Learning Machines. Estimates of both model and prediction uncertainties, and of their propagation after the transformation of wind speed into wind power, are then provided without any assumptions on distribution patterns of the data. The methodology is applied to the study of hourly wind power potential on a grid of 250 by 250 squared meters for turbines of 100 meters hub height in Switzerland, generating the first dataset of its type for the country. The potential wind power generation is combined with the available area for wind turbine installations to yield an estimate of the technical potential for wind power in Switzerland. The wind power estimate presented here represents an important input for planners to support the design of future energy systems with increased wind power generation.
When analyzing data from randomized clinical trials, covariate adjustment can be used to account for chance imbalance in baseline characteristics and to increase precision of the treatment effect estimate. A practical barrier to covariate adjustment is the presence of missing data in covariates, which raise various questions in implementation. In this paper we review several covariate adjustment methods with incomplete data, including complete data analysis and imputation-based methods, and investigate the implications of the missing data mechanism on estimating the average treatment effect in randomized clinical trials with continuous or binary outcomes. We consider both the outcome regression and propensity score weighting adjustment methods. We conduct comprehensive simulation studies to compare these methods. We find that performing adjustment with imputed covariates, regardless of adjustment method or imputation method, generally improves the precision of treatment effect estimates when the proportion of missingness is not too large and the adjusted covariate is associated with the outcome. Furthermore, we underscore the importance of including the missingness indicators as predictors in the outcome model or the propensity score model especially when covariates are missing not at random.
Estimation of conditional average treatment effects (CATEs) plays an essential role in modern medicine by informing treatment decision-making at a patient level. Several metalearners have been proposed recently to estimate CATEs in an effective and flexible way by re-purposing predictive machine learning models for causal estimation. In this chapter, we summarize the literature on metalearners and provide concrete guidance for their application for treatment heterogeneity estimation from randomized controlled trials' data with survival outcomes. The guidance we provide is supported by a comprehensive simulation study in which we vary the complexity of the underlying baseline risk and CATE functions, the magnitude of the heterogeneity in the treatment effect, the censoring mechanism, and the balance in treatment assignment. To demonstrate the applicability of our findings, we reanalyze the data from the Systolic Blood Pressure Intervention Trial (SPRINT) and the Action to Control Cardiovascular Risk in Diabetes (ACCORD) study. While recent literature reports the existence of heterogeneous effects of intensive blood pressure treatment with multiple treatment effect modifiers, our results suggest that many of these modifiers may be spurious discoveries. This chapter is accompanied by 'survlearners', an R package that provides well-documented implementations of the CATE estimation strategies described in this work, to allow easy use of our recommendations as well as reproduction of our numerical study.
The content that a recommender system (RS) shows to users influences them. Therefore, when choosing a recommender to deploy, one is implicitly also choosing to induce specific internal states in users. Even more, systems trained via long-horizon optimization will have direct incentives to manipulate users: in this work, we focus on the incentive to shift user preferences so they are easier to satisfy. We argue that - before deployment - system designers should: estimate the shifts a recommender would induce; evaluate whether such shifts would be undesirable; and perhaps even actively optimize to avoid problematic shifts. These steps involve two challenging ingredients: estimation requires anticipating how hypothetical algorithms would influence user preferences if deployed - we do this by using historical user interaction data to train a predictive user model which implicitly contains their preference dynamics; evaluation and optimization additionally require metrics to assess whether such influences are manipulative or otherwise unwanted - we use the notion of "safe shifts", that define a trust region within which behavior is safe: for instance, the natural way in which users would shift without interference from the system could be deemed "safe". In simulated experiments, we show that our learned preference dynamics model is effective in estimating user preferences and how they would respond to new recommenders. Additionally, we show that recommenders that optimize for staying in the trust region can avoid manipulative behaviors while still generating engagement.
小貼士
登錄享
相關主題
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191