Search

Dropout often threatens the validity of causal inference in longitudinal studies. While existing studies have focused on the problem of missing outcomes caused by treatment, we study an important but overlooked source of dropout, selective eligibility. For example, patients may become ineligible for subsequent treatments due to severe side effects or complete recovery. Selective eligibility differs from the problem of ``truncation by death'' because dropout occurs after observing the outcome but before receiving the subsequent treatment. This difference makes the standard approach to dropout inapplicable. We propose a general methodological framework for longitudinal causal inference with selective eligibility. By focusing on subgroups of units who would become eligible for treatment given a specific treatment history, we define the time-specific eligible treatment effect (ETE) and expected number of outcome events (EOE) under a treatment sequence of interest. Assuming a generalized version of sequential ignorability, we derive two nonparametric identification formulae, each leveraging different parts of the observed data distribution. We then derive the efficient influence function of each causal estimand, yielding the corresponding doubly robust estimator. Finally, we apply the proposed methodology to an impact evaluation of a pre-trial risk assessment instrument in the criminal justice system, in which selective eligibility arises due to recidivism.

In this paper, we demonstrate how to enhance the validity of causal inference with unstructured high-dimensional treatments like texts, by leveraging the power of generative Artificial Intelligence. Specifically, we propose to use a deep generative model such as large language models (LLMs) to efficiently generate treatments and use their internal representation for subsequent causal effect estimation. We show that the knowledge of this true internal representation helps separate the treatment features of interest, such as specific sentiments and certain topics, from other possibly unknown confounding features. Unlike the existing methods, our proposed approach eliminates the need to learn causal representation from the data and hence produces more accurate and efficient estimates. We formally establish the conditions required for the nonparametric identification of the average treatment effect, propose an estimation strategy that avoids the violation of the overlap assumption, and derive the asymptotic properties of the proposed estimator through the application of double machine learning. Finally, using an instrumental variables approach, we extend the proposed methodology to the settings, in which the treatment feature is based on human perception rather than is assumed to be fixed given the treatment object. We conduct simulation studies using the generated text data with an open-source LLM, Llama3, to illustrate the advantages of our estimator over the state-of-the-art causal representation learning algorithms.

Political actors frequently manipulate redistricting plans to gain electoral advantages, a process commonly known as gerrymandering. To address this problem, several states have implemented institutional reforms including the establishment of map-drawing commissions. It is difficult to assess the impact of such reforms because each state structures bundles of complex rules in different ways. We propose to model redistricting processes as a sequential game. The equilibrium solution to the game summarizes multi-step institutional interactions as a single dimensional score. This score measures the leeway political actors have over the partisan lean of the final plan. Using a differences-in-differences design, we demonstrate that reforms reduce partisan bias and increase competitiveness when they constrain partisan actors. We perform a counterfactual policy analysis to estimate the partisan effects of enacting recent institutional reforms nationwide. We find that instituting redistricting commissions generally reduces the current Republican advantage, but Michigan-style reforms would yield a much greater pro-Democratic effect than types of redistricting commissions adopted in Ohio and New York., Comment: 20 pages, 8 figures, 1 table, plus references and appendices

A century ago, Neyman showed how to evaluate the efficacy of treatment using a randomized experiment under a minimal set of assumptions. This classical repeated sampling framework serves as a basis of routine experimental analyses conducted by today's scientists across disciplines. In this paper, we demonstrate that Neyman's methodology can also be used to experimentally evaluate the efficacy of individualized treatment rules (ITRs), which are derived by modern causal machine learning algorithms. In particular, we show how to account for additional uncertainty resulting from a training process based on cross-fitting. The primary advantage of Neyman's approach is that it can be applied to any ITR regardless of the properties of machine learning algorithms that are used to derive the ITR. We also show, somewhat surprisingly, that for certain metrics, it is more efficient to conduct this ex-post experimental evaluation of an ITR than to conduct an ex-ante experimental evaluation that randomly assigns some units to the ITR. Our analysis demonstrates that Neyman's repeated sampling framework is as relevant for causal inference today as it has been since its inception.

The use of Artificial Intelligence (AI), or more generally data-driven algorithms, has become ubiquitous in today's society. Yet, in many cases and especially when stakes are high, humans still make final decisions. The critical question, therefore, is whether AI helps humans make better decisions compared to a human-alone or AI-alone system. We introduce a new methodological framework to empirically answer this question with a minimal set of assumptions. We measure a decision maker's ability to make correct decisions using standard classification metrics based on the baseline potential outcome. We consider a single-blinded and unconfounded treatment assignment, where the provision of AI-generated recommendations is assumed to be randomized across cases with humans making final decisions. Under this study design, we show how to compare the performance of three alternative decision-making systems--human-alone, human-with-AI, and AI-alone. Importantly, the AI-alone system includes any individualized treatment assignment, including those that are not used in the original study. We also show when AI recommendations should be provided to a human-decision maker, and when one should follow such recommendations. We apply the proposed methodology to our own randomized controlled trial evaluating a pretrial risk assessment instrument. We find that the risk assessment recommendations do not improve the classification accuracy of a judge's decision to impose cash bail. Furthermore, we find that replacing a human judge with algorithms--the risk assessment score and a large language model in particular--leads to a worse classification performance.

We introduce the "cram" method, a general and efficient approach to simultaneous learning and evaluation using a generic machine learning (ML) algorithm. In a single pass of batched data, the proposed method repeatedly trains an ML algorithm and tests its empirical performance. Because it utilizes the entire sample for both learning and evaluation, cramming is significantly more data-efficient than sample-splitting. The cram method also naturally accommodates online learning algorithms, making its implementation computationally efficient. To demonstrate the power of the cram method, we consider the standard policy learning setting where cramming is applied to the same data to both develop an individualized treatment rule (ITR) and estimate the average outcome that would result if the learned ITR were to be deployed. We show that under a minimal set of assumptions, the resulting crammed evaluation estimator is consistent and asymptotically normal. While our asymptotic results require a relatively weak stabilization condition of ML algorithm, we develop a simple, generic method that can be used with any policy learning algorithm to satisfy this condition. Our extensive simulation studies show that, when compared to sample-splitting, cramming reduces the evaluation standard error by more than 40% while improving the performance of learned policy. We also apply the cram method to a randomized clinical trial to demonstrate its applicability to real-world problems. Finally, we briefly discuss future extensions of the cram method to other learning and evaluation settings.

A growing number of researchers are conducting randomized experiments to analyze causal relationships in network settings where units influence one another. A dominant methodology for analyzing these experiments is design-based, leveraging random treatment assignments as the basis for inference. In this paper, we generalize this design-based approach to accommodate complex experiments with a variety of causal estimands and different target populations. An important special case of such generalized network experiments is a bipartite network experiment, in which treatment is randomized among one set of units, and outcomes are measured on a separate set of units. We propose a broad class of causal estimands based on stochastic interventions for generalized network experiments. Using a design-based approach, we show how to estimate these causal quantities without bias and develop conservative variance estimators. We apply our methodology to a randomized experiment in education where participation in an anti-conflict promotion program is randomized among selected students. Our analysis estimates the causal effects of treating each student or their friends among different target populations in the network. We find that the program improves the overall conflict awareness among students but does not significantly reduce the total number of such conflicts.

While there now exists a large literature on policy evaluation and learning, much of prior work assumes that the treatment assignment of one unit does not affect the outcome of another unit. Unfortunately, ignoring interference may lead to biased policy evaluation and ineffective learned policies. For example, treating influential individuals who have many friends can generate positive spillover effects, thereby improving the overall performance of an individualized treatment rule (ITR). We consider the problem of evaluating and learning an optimal ITR under clustered network interference (also known as partial interference) where clusters of units are sampled from a population and units may influence one another within each cluster. Unlike previous methods that impose strong restrictions on spillover effects, the proposed methodology only assumes a semiparametric structural model where each unit's outcome is an additive function of individual treatments within the cluster. Under this model, we propose an estimator that can be used to evaluate the empirical performance of an ITR. We show that this estimator is substantially more efficient than the standard inverse probability weighting estimator, which does not impose any assumption about spillover effects. We derive the finite-sample regret bound for a learned ITR, showing that the use of our efficient evaluation estimator leads to the improved performance of learned policies. Finally, we conduct simulation and empirical studies to illustrate the advantages of the proposed methodology.

Across a wide array of disciplines, many researchers use machine learning (ML) algorithms to identify a subgroup of individuals who are likely to benefit from a treatment the most (``exceptional responders'') or those who are harmed by it. A common approach to this subgroup identification problem consists of two steps. First, researchers estimate the conditional average treatment effect (CATE) using an ML algorithm. Next, they use the estimated CATE to select those individuals who are predicted to be most affected by the treatment, either positively or negatively. Unfortunately, CATE estimates are often biased and noisy. In addition, utilizing the same data to both identify a subgroup and estimate its group average treatment effect results in a multiple testing problem. To address these challenges, we develop uniform confidence bands for estimation of the group average treatment effect sorted by generic ML algorithm (GATES). Using these uniform confidence bands, researchers can identify, with a statistical guarantee, a subgroup whose GATES exceeds a certain effect size, regardless of how this effect size is chosen. The validity of the proposed methodology depends solely on randomization of treatment and random sampling of units. Importantly, our method does not require modeling assumptions and avoids a computationally intensive resampling procedure. A simulation study shows that the proposed uniform confidence bands are reasonably informative and have an appropriate empirical coverage even when the sample size is as small as 100. We analyze a clinical trial of late-stage prostate cancer and find a relatively large proportion of exceptional responders.

Algorithmic decisions and recommendations are used in many high-stakes decision-making settings such as criminal justice, medicine, and public policy. We investigate whether it would have been possible to improve a security assessment algorithm employed during the Vietnam War, using outcomes measured immediately after its introduction in late 1969. This empirical application raises several methodological challenges that frequently arise in high-stakes algorithmic decision-making. First, before implementing a new algorithm, it is essential to characterize and control the risk of yielding worse outcomes than the existing algorithm. Second, the existing algorithm is deterministic, and learning a new algorithm requires transparent extrapolation. Third, the existing algorithm involves discrete decision tables that are difficult to optimize over. To address these challenges, we introduce the Average Conditional Risk (ACRisk), which first quantifies the risk that a new algorithmic policy leads to worse outcomes for subgroups of individual units and then averages this over the distribution of subgroups. We also propose a Bayesian policy learning framework that maximizes the posterior expected value while controlling the posterior expected ACRisk. This framework separates the estimation of heterogeneous treatment effects from policy optimization, enabling flexible estimation of effects and optimization over complex policy classes. We characterize the resulting chance-constrained optimization problem as a constrained linear programming problem. Our analysis shows that compared to the actual algorithm used during the Vietnam War, the learned algorithm assesses most regions as more secure and emphasizes economic and political factors over military factors.

The United States Census Bureau faces a difficult trade-off between the accuracy of Census statistics and the protection of individual information. We conduct the first independent evaluation of bias and noise induced by the Bureau's two main disclosure avoidance systems: the TopDown algorithm employed for the 2020 Census and the swapping algorithm implemented for the three previous Censuses. Our evaluation leverages the Noisy Measure File (NMF) as well as two independent runs of the TopDown algorithm applied to the 2010 decennial Census. We find that the NMF contains too much noise to be directly useful, especially for Hispanic and multiracial populations. TopDown's post-processing dramatically reduces the NMF noise and produces data whose accuracy is similar to that of swapping. While the estimated errors for both TopDown and swapping algorithms are generally no greater than other sources of Census error, they can be relatively substantial for geographies with small total populations., Comment: 25 pages, 6 figures, 2 tables, plus appendices

Conditioning on variables affected by treatment can induce post-treatment bias when estimating causal effects. Although this suggests that researchers should measure potential moderators before administering the treatment in an experiment, doing so may also bias causal effect estimation if the covariate measurement primes respondents to react differently to the treatment. This paper formally analyzes this trade-off between post-treatment and priming biases in three experimental designs that vary when moderators are measured: pre-treatment, post-treatment, or a randomized choice between the two. We derive nonparametric bounds for interactions between the treatment and the moderator under each design and show how to use substantive assumptions to narrow these bounds. These bounds allow researchers to assess the sensitivity of their empirical findings to priming and post-treatment bias. We then apply the proposed methodology to a survey experiment on electoral messaging., Comment: 28 pages (main text), 22 pages (supplementary materials), 5 figures

The U.S. Census Bureau collects and publishes detailed demographic data about Americans which are heavily used by researchers and policymakers. The Bureau has recently adopted the framework of differential privacy in an effort to improve confidentiality of individual census responses. A key output of this privacy protection system is the Noisy Measurement File (NMF), which is produced by adding random noise to tabulated statistics. The NMF is critical to understanding any errors introduced in the data, and performing valid statistical inference on published census data. Unfortunately, the current release format of the NMF is difficult to access and work with. We describe the process we use to transform the NMF into a usable format, and provide recommendations to the Bureau for how to release future versions of the NMF. These changes are essential for ensuring transparency of privacy measures and reproducibility of scientific research built on census data., Comment: 9 pages, 2 figures

Many networks in political and social research are bipartite, with edges connecting exclusively across two distinct types of nodes. A common example includes cosponsorship networks, in which legislators are connected indirectly through the bills they support. Yet most existing network models are designed for unipartite networks, where edges can arise between any pair of nodes. However, using a unipartite network model to analyze bipartite networks, as often done in practice, can result in aggregation bias and artificially high-clustering -- a particularly insidious problem when studying the role groups play in network formation. To address these methodological problems, we develop a statistical model of bipartite networks theorized to be generated through group interactions by extending the popular mixed-membership stochastic blockmodel. Our model allows researchers to identify the groups of nodes, within each node type in the bipartite structure, that share common patterns of edge formation. The model also incorporates both node and dyad-level covariates as the predictors of group membership and of observed dyadic relations. We develop an efficient computational algorithm for fitting the model, and apply it to cosponsorship data from the United States Senate. We show that legislators in a Senate that was perfectly split along party lines were able to remain productive and pass major legislation by forming non-partisan, power-brokering coalitions that found common ground through their collaboration on low-stakes bills. We also find evidence for norms of reciprocity, and uncover the substantial role played by policy expertise in the formation of cosponsorships between senators and legislation. We make an open-source software package available that makes it possible for other researchers to uncover similar insights from bipartite networks., Comment: 41 pages (main text), 6 pages (appendix), 19 pages (online SI)

The estimation of racial disparities in various fields is often hampered by the lack of individual-level racial information. In many cases, the law prohibits the collection of such information to prevent direct racial discrimination. As a result, analysts have frequently adopted Bayesian Improved Surname Geocoding (BISG) and its variants, which combine individual names and addresses with Census data to predict race. Unfortunately, the residuals of BISG are often correlated with the outcomes of interest, generally attenuating estimates of racial disparities. To correct this bias, we propose an alternative identification strategy under the assumption that surname is conditionally independent of the outcome given (unobserved) race, residence location, and other observed characteristics. We introduce a new class of models, Bayesian Instrumental Regression for Disparity Estimation (BIRDiE), that take BISG probabilities as inputs and produce racial disparity estimates by using surnames as an instrumental variable for race. Our estimation method is scalable, making it possible to analyze large-scale administrative data. We also show how to address potential violations of the key identification assumptions. A validation study based on the North Carolina voter file shows that BISG+BIRDiE reduces error by up to 84% when estimating racial differences in party registration. Finally, we apply the proposed methodology to estimate racial differences in who benefits from the home mortgage interest deduction using individual-level tax data from the U.S. Internal Revenue Service. Open-source software is available which implements the proposed methodology., Comment: 28 pages, 9 figures, plus references and appendices

In "Differential Perspectives: Epistemic Disconnects Surrounding the US Census Bureau's Use of Differential Privacy," boyd and Sarathy argue that empirical evaluations of the Census Disclosure Avoidance System (DAS), including our published analysis, failed to recognize how the benchmark data against which the 2020 DAS was evaluated is never a ground truth of population counts. In this commentary, we explain why policy evaluation, which was the main goal of our analysis, is still meaningful without access to a perfect ground truth. We also point out that our evaluation leveraged features specific to the decennial Census and redistricting data, such as block-level population invariance under swapping and voter file racial identification, better approximating a comparison with the ground truth. Lastly, we show that accurate statistical predictions of individual race based on the Bayesian Improved Surname Geocoding, while not a violation of differential privacy, substantially increases the disclosure risk of private information the Census Bureau sought to protect. We conclude by arguing that policy makers must confront a key trade-off between data utility and privacy protection, and an epistemic disconnect alone is insufficient to explain disagreements between policy choices., Comment: Version accepted to Harvard Data Science Review

We consider the estimation of average treatment effects in observational studies and propose a new framework of robust causal inference with unobserved confounders. Our approach is based on distributionally robust optimization and proceeds in two steps. We first specify the maximal degree to which the distribution of unobserved potential outcomes may deviate from that of observed outcomes. We then derive sharp bounds on the average treatment effects under this assumption. Our framework encompasses the popular marginal sensitivity model as a special case, and we demonstrate how the proposed methodology can address a primary challenge of the marginal sensitivity model that it produces uninformative results when unobserved confounders substantially affect treatment and outcome. Specifically, we develop an alternative sensitivity model, called the distributional sensitivity model, under the assumption that heterogeneity of treatment effect due to unobserved variables is relatively small. Unlike the marginal sensitivity model, the distributional sensitivity model allows for potential lack of overlap and often produces informative bounds even when unobserved variables substantially affect both treatment and outcome. Finally, we show how to extend the distributional sensitivity model to difference-in-differences designs and settings with instrumental variables. Through simulation and empirical studies, we demonstrate the applicability of the proposed methodology.

The regression discontinuity (RD) design is widely used for program evaluation with observational data. The primary focus of the existing literature has been the estimation of the local average treatment effect at the existing treatment cutoff. In contrast, we consider policy learning under the RD design. Because the treatment assignment mechanism is deterministic, learning better treatment cutoffs requires extrapolation. We develop a robust optimization approach to finding optimal treatment cutoffs that improve upon the existing ones. We first decompose the expected utility into point-identifiable and unidentifiable components. We then propose an efficient doubly-robust estimator for the identifiable parts. To account for the unidentifiable components, we leverage the existence of multiple cutoffs that are common under the RD design. Specifically, we assume that the heterogeneity in the conditional expectations of potential outcomes across different groups vary smoothly along the running variable. Under this assumption, we minimize the worst case utility loss relative to the status quo policy. The resulting new treatment cutoffs have a safety guarantee that they will not yield a worse overall outcome than the existing cutoffs. Finally, we establish the asymptotic regret bounds for the learned policy using semi-parametric efficiency theory. We apply the proposed methodology to empirical and simulated data sets.

We provide the largest compiled publicly available dictionaries of first, middle, and last names for the purpose of imputing race and ethnicity using, for example, Bayesian Improved Surname Geocoding (BISG). The dictionaries are based on the voter files of six Southern states that collect self-reported racial data upon voter registration. Our data cover a much larger scope of names than any comparable dataset, containing roughly one million first names, 1.1 million middle names, and 1.4 million surnames. Individuals are categorized into five mutually exclusive racial and ethnic groups -- White, Black, Hispanic, Asian, and Other -- and racial/ethnic counts by name are provided for every name in each dictionary. Counts can then be normalized row-wise or column-wise to obtain conditional probabilities of race given name or name given race. These conditional probabilities can then be deployed for imputation in a data analytic task for which ground truth racial and ethnic data is not available.

Congressional district lines in many U.S. states are drawn by partisan actors, raising concerns about gerrymandering. To separate the partisan effects of redistricting from the effects of other factors including geography and redistricting rules, we compare possible party compositions of the U.S. House under the enacted plan to those under a set of alternative simulated plans that serve as a non-partisan baseline. We find that partisan gerrymandering is widespread in the 2020 redistricting cycle, but most of the electoral bias it creates cancels at the national level, giving Republicans two additional seats on average. Geography and redistricting rules separately contribute a moderate pro-Republican bias. Finally, we find that partisan gerrymandering reduces electoral competition and makes the partisan composition of the U.S. House less responsive to shifts in the national vote., Comment: 10 pages, 4 figures, plus references and appendix

Data-driven decision making plays an important role even in high stakes settings like medicine and public policy. Learning optimal policies from observed data requires a careful formulation of the utility function whose expected value is maximized across a population. Although researchers typically use utilities that depend on observed outcomes alone, in many settings the decision maker's utility function is more properly characterized by the joint set of potential outcomes under all actions. For example, the Hippocratic principle to ``do no harm'' implies that the cost of causing death to a patient who would otherwise survive without treatment is greater than the cost of forgoing life-saving treatment. We consider optimal policy learning with asymmetric utility functions of this form. We show that asymmetric utilities lead to an unidentifiable social welfare function, and so we first partially identify it. Drawing on statistical decision theory, we then derive minimax decision rules by minimizing the maximum regret relative to alternative policies. We show that one can learn minimax decision rules from observed data by solving intermediate classification problems. We also establish that the finite sample regret of this procedure is bounded by the mis-classification rate of these intermediate classifiers. We apply this conceptual framework and methodology to the decision about whether or not to use right heart catheterization for patients with possible pulmonary hypertension.

This article introduces the 50stateSimulations, a collection of simulated congressional districting plans and underlying code developed by the Algorithm-Assisted Redistricting Methodology (ALARM) Project. The 50stateSimulations allow for the evaluation of enacted and other congressional redistricting plans in the United States. While the use of redistricting simulation algorithms has become standard in academic research and court cases, any simulation analysis requires non-trivial efforts to combine multiple data sets, identify state-specific redistricting criteria, implement complex simulation algorithms, and summarize and visualize simulation outputs. We have developed a complete workflow that facilitates this entire process of simulation-based redistricting analysis for the congressional districts of all 50 states. The resulting 50stateSimulations include ensembles of simulated 2020 congressional redistricting plans and necessary replication data. We also provide the underlying code, which serves as a template for customized analyses. All data and code are free and publicly available. This article details the design, creation, and validation of the data., Comment: 11 pages, 3 figures

Prediction of individual's race and ethnicity plays an important role in social science and public health research. Examples include studies of racial disparity in health and voting. Recently, Bayesian Improved Surname Geocoding (BISG), which uses Bayes' rule to combine information from Census surname files with the geocoding of an individual's residence, has emerged as a leading methodology for this prediction task. Unfortunately, BISG suffers from two Census data problems that contribute to unsatisfactory predictive performance for minorities. First, the decennial Census often contains zero counts for minority racial groups in the Census blocks where some members of those groups reside. Second, because the Census surname files only include frequent names, many surnames -- especially those of minorities -- are missing from the list. To address the zero counts problem, we introduce a fully Bayesian Improved Surname Geocoding (fBISG) methodology that accounts for potential measurement error in Census counts by extending the naive Bayesian inference of the BISG methodology to full posterior inference. To address the missing surname problem, we supplement the Census surname data with additional data on last, first, and middle names taken from the voter files of six Southern states where self-reported race is available. Our empirical validation shows that the fBISG methodology and name supplements significantly improve the accuracy of race imputation across all racial groups, and especially for Asians. The proposed methodology, together with additional name data, is available via the open-source software WRU.

Researchers are increasingly turning to machine learning (ML) algorithms to investigate causal heterogeneity in randomized experiments. Despite their promise, ML algorithms may fail to accurately ascertain heterogeneous treatment effects under practical settings with many covariates and small sample size. In addition, the quantification of estimation uncertainty remains a challenge. We develop a general approach to statistical inference for heterogeneous treatment effects discovered by a generic ML algorithm. We apply the Neyman's repeated sampling framework to a common setting, in which researchers use an ML algorithm to estimate the conditional average treatment effect and then divide the sample into several groups based on the magnitude of the estimated effects. We show how to estimate the average treatment effect within each of these groups, and construct a valid confidence interval. In addition, we develop nonparametric tests of treatment effect homogeneity across groups, and rank-consistency of within-group average treatment effects. The validity of our methodology does not rely on the properties of ML algorithms because it is solely based on the randomization of treatment assignment and random sampling of units. Finally, we generalize our methodology to the cross-fitting procedure by accounting for the additional uncertainty induced by the random splitting of data.

Conjoint analysis is a popular experimental design used to measure multidimensional preferences. Researchers examine how varying a factor of interest, while controlling for other relevant factors, influences decision-making. Currently, there exist two methodological approaches to analyzing data from a conjoint experiment. The first focuses on estimating the average marginal effects of each factor while averaging over the other factors. Although this allows for straightforward design-based estimation, the results critically depend on the distribution of other factors and how interaction effects are aggregated. An alternative model-based approach can compute various quantities of interest, but requires researchers to correctly specify the model, a challenging task for conjoint analysis with many factors and possible interactions. In addition, a commonly used logistic regression has poor statistical properties even with a moderate number of factors when incorporating interactions. We propose a new hypothesis testing approach based on the conditional randomization test to answer the most fundamental question of conjoint analysis: Does a factor of interest matter in any way given the other factors? Our methodology is solely based on the randomization of factors, and hence is free from assumptions. Yet, it allows researchers to use any test statistic, including those based on complex machine learning algorithms. As a result, we are able to combine the strengths of the existing design-based and model-based approaches. We illustrate the proposed methodology through conjoint analysis of immigration preferences and political candidate evaluation. We also extend the proposed approach to test for regularity assumptions commonly used in conjoint analysis. An open-source software package is available for implementing the proposed methodology.

Estimation of heterogeneous treatment effects is an active area of research. Most of the existing methods, however, focus on estimating the conditional average treatment effects of a single, binary treatment given a set of pre-treatment covariates. In this paper, we propose a method to estimate the heterogeneous causal effects of high-dimensional treatments, which poses unique challenges in terms of estimation and interpretation. The proposed approach finds maximally heterogeneous groups and uses a Bayesian mixture of regularized logistic regressions to identify groups of units who exhibit similar patterns of treatment effects. By directly modeling group membership with covariates, the proposed methodology allows one to explore the unit characteristics that are associated with different patterns of treatment effects. Our motivating application is conjoint analysis, which is a popular type of survey experiment in social science and marketing research and is based on a high-dimensional factorial design. We apply the proposed methodology to the conjoint data, where survey respondents are asked to select one of two immigrant profiles with randomly selected attributes. We find that a group of respondents with a relatively high degree of prejudice appears to discriminate against immigrants from non-European countries like Iraq. An open-source software package is available for implementing the proposed methodology., Comment: Major revision; added Propositions 1 and 2. 27 pages (main text); 34 pages (supplementary information)

Granular geographic data present new opportunities to understand how neighborhoods are formed, and how they influence politics. At the same time, the inherent subjectivity of neighborhoods creates methodological challenges in measuring and modeling them. We develop an open-source survey instrument that allows respondents to draw their neighborhoods on a map. We also propose a statistical model to analyze how the characteristics of respondents and local areas determine subjective neighborhoods. We conduct two surveys: collecting subjective neighborhoods from voters in Miami, New York City, and Phoenix, and asking New York City residents to draw a community of interest for inclusion in their city council district. Our analysis shows that, holding other factors constant, White respondents include census blocks with more White residents in their neighborhoods. Similarly, Democrats and Republicans are more likely to include co-partisan areas. Furthermore, our model provides more accurate out-of-sample predictions than standard neighborhood measures., Comment: 34 pages, 11 figures, and supplementary material

Algorithmic recommendations and decisions have become ubiquitous in today's society. Many of these and other data-driven policies, especially in the realm of public policy, are based on known, deterministic rules to ensure their transparency and interpretability. For example, algorithmic pre-trial risk assessments, which serve as our motivating application, provide relatively simple, deterministic classification scores and recommendations to help judges make release decisions. How can we use the data based on existing deterministic policies to learn new and better policies? Unfortunately, prior methods for policy learning are not applicable because they require existing policies to be stochastic rather than deterministic. We develop a robust optimization approach that partially identifies the expected utility of a policy, and then finds an optimal policy by minimizing the worst-case regret. The resulting policy is conservative but has a statistical safety guarantee, allowing the policy-maker to limit the probability of producing a worse outcome than the existing policy. We extend this approach to common and important settings where humans make decisions with the aid of algorithmic recommendations. Lastly, we apply the proposed methodology to a unique field experiment on pre-trial risk assessment instruments. We derive new classification and recommendation rules that retain the transparency and interpretability of the existing instrument while potentially leading to better overall outcomes at a lower cost.

Inverse probability of treatment weighting (IPTW) is a popular method for estimating the average treatment effect (ATE). However, empirical studies show that the IPTW estimators can be sensitive to the misspecification of the propensity score model. To address this problem, researchers have proposed to estimate propensity score by directly optimizing the balance of pre-treatment covariates. While these methods appear to empirically perform well, little is known about how the choice of balancing conditions affects their theoretical properties. To fill this gap, we first characterize the asymptotic bias and efficiency of the IPTW estimator based on the Covariate Balancing Propensity Score (CBPS) methodology under local model misspecification. Based on this analysis, we show how to optimally choose the covariate balancing functions and propose an optimal CBPS-based IPTW estimator. This estimator is doubly robust; it is consistent for the ATE if either the propensity score model or the outcome model is correct. In addition, the proposed estimator is locally semiparametric efficient when both models are correctly specified. To further relax the parametric assumptions, we extend our method by using a sieve estimation approach. We show that the resulting estimator is globally efficient under a set of much weaker assumptions and has a smaller asymptotic bias than the existing estimators. Finally, we evaluate the finite sample performance of the proposed estimators via simulation and empirical studies. An open-source software package is available for implementing the proposed methods.

The US Census Bureau plans to protect the privacy of 2020 Census respondents through its Disclosure Avoidance System (DAS), which attempts to achieve differential privacy guarantees by adding noise to the Census microdata. By applying redistricting simulation and analysis methods to DAS-protected 2010 Census data, we find that the protected data are not of sufficient quality for redistricting purposes. We demonstrate that the injected noise makes it impossible for states to accurately comply with the One Person, One Vote principle. Our analysis finds that the DAS-protected data are biased against certain areas, depending on voter turnout and partisan and racial composition, and that these biases lead to large and unpredictable errors in the analysis of partisan and racial gerrymanders. Finally, we show that the DAS algorithm does not universally protect respondent privacy. Based on the names and addresses of registered voters, we are able to predict their race as accurately using the DAS-protected data as when using the 2010 Census data. Despite this, the DAS-protected data can still inaccurately estimate the number of majority-minority districts. We conclude with recommendations for how the Census Bureau should proceed with privacy protection for the 2020 Census., Comment: 42 pages, 22 figures. New postscript analyzing newly-released DAS-19.61 revision

A primary goal of social science research is to understand how latent group memberships predict the dynamic process of network evolution. In the modeling of international militarized conflicts, for instance, scholars hypothesize that membership in geopolitical coalitions shapes the decision to engage in conflict. Such theories explain the ways in which nodal and dyadic characteristics affect the evolution of conflict patterns over time via their effects on group memberships. To aid the empirical testing of these arguments, we develop a dynamic model of network data by combining a hidden Markov model with a mixed-membership stochastic blockmodel that identifies latent groups underlying the network structure. Unlike existing models, we incorporate covariates that predict dynamic node memberships in latent groups as well as the direct formation of edges between dyads. While prior substantive research often assumes the decision to engage in international militarized conflict is independent across states and static over time, we demonstrate that conflict is driven by states' evolving membership in geopolitical blocs. Changes in monadic covariates like democracy shift states between coalitions, generating heterogeneous effects on conflict over time and across states. The proposed methodology, which relies on a variational approximation to a collapsed posterior distribution as well as stochastic optimization for scalability, is implemented through an open-source software package., Comment: 34 pages (main text), 34 pages (supplementary information), 21 figures

Support vector machine (SVM) is one of the most popular classification algorithms in the machine learning literature. We demonstrate that SVM can be used to balance covariates and estimate average causal effects under the unconfoundedness assumption. Specifically, we adapt the SVM classifier as a kernel-based weighting procedure that minimizes the maximum mean discrepancy between the treatment and control groups while simultaneously maximizing effective sample size. We also show that SVM is a continuous relaxation of the quadratic integer program for computing the largest balanced subset, establishing its direct relation to the cardinality matching method. Another important feature of SVM is that the regularization parameter controls the trade-off between covariate balance and effective sample size. As a result, the existing SVM path algorithm can be used to compute the balance-sample size frontier. We characterize the bias of causal effect estimation arising from this trade-off, connecting the proposed SVM procedure to the existing kernel balancing methods. Finally, we conduct simulation and empirical studies to evaluate the performance of the proposed methodology and find that SVM is competitive with the state-of-the-art covariate balancing methods., Comment: 37 pages, 8 figures

Despite an increasing reliance on fully-automated algorithmic decision-making in our day-to-day lives, human beings still make highly consequential decisions. As frequently seen in business, healthcare, and public policy, recommendations produced by algorithms are provided to human decision-makers to guide their decisions. While there exists a fast-growing literature evaluating the bias and fairness of such algorithmic recommendations, an overlooked question is whether they help humans make better decisions. We develop a statistical methodology for experimentally evaluating the causal impacts of algorithmic recommendations on human decisions. We also show how to examine whether algorithmic recommendations improve the fairness of human decisions and derive the optimal decision rules under various settings. We apply the proposed methodology to preliminary data from the first-ever randomized controlled trial that evaluates the pretrial Public Safety Assessment (PSA) in the criminal justice system. A goal of the PSA is to help judges decide which arrested individuals should be released. On the basis of the preliminary data available, we find that providing the PSA to the judge has little overall impact on the judge's decisions and subsequent arrestee behavior. However, our analysis yields some potentially suggestive evidence that the PSA may help avoid unnecessarily harsh decisions for female arrestees regardless of their risk levels while it encourages the judge to make stricter decisions for male arrestees who are deemed to be risky. In terms of fairness, the PSA appears to increase the gender bias against males while having little effect on any existing racial differences in judges' decision. Finally, we find that the PSA's recommendations might be unnecessarily severe unless the cost of a new crime is sufficiently high.

Two-stage randomized experiments are becoming an increasingly popular experimental design for causal inference when the outcome of one unit may be affected by the treatment assignments of other units in the same cluster. In this paper, we provide a methodological framework for general tools of statistical inference and power analysis for two-stage randomized experiments. Under the randomization-based framework, we consider the estimation of a new direct effect of interest as well as the average direct and spillover effects studied in the literature. We provide unbiased estimators of these causal quantities and their conservative variance estimators in a general setting. Using these results, we then develop hypothesis testing procedures and derive sample size formulas. We theoretically compare the two-stage randomized design with the completely randomized and cluster randomized designs, which represent two limiting designs. Finally, we conduct simulation studies to evaluate the empirical performance of our sample size formulas. For empirical illustration, the proposed methodology is applied to the randomized evaluation of the Indian national health insurance program. An open-source software package is available for implementing the proposed methodology.

Random sampling of graph partitions under constraints has become a popular tool for evaluating legislative redistricting plans. Analysts detect partisan gerrymandering by comparing a proposed redistricting plan with an ensemble of sampled alternative plans. For successful application, sampling methods must scale to maps with a moderate or large number of districts, incorporate realistic legal constraints, and accurately and efficiently sample from a selected target distribution. Unfortunately, most existing methods struggle in at least one of these areas. We present a new Sequential Monte Carlo (SMC) algorithm that generates a sample of redistricting plans converging to a realistic target distribution. Because it draws many plans in parallel, the SMC algorithm can efficiently explore the relevant space of redistricting plans better than the existing Markov chain Monte Carlo (MCMC) algorithms that generate plans sequentially. Our algorithm can simultaneously incorporate several constraints commonly imposed in real-world redistricting problems, including equal population, compactness, and preservation of administrative boundaries. We validate the accuracy of the proposed algorithm by using a small map where all redistricting plans can be enumerated. We then apply the SMC algorithm to evaluate the partisan implications of several maps submitted by relevant parties in a recent high-profile redistricting case in the state of Pennsylvania. We find that the proposed algorithm converges faster and with fewer samples than a comparable MCMC algorithm. Open-source software is available for implementing the proposed methodology., Comment: 19 pages, 7 figures, plus appendices; revised validation section, discussion, and appendices

As granular data about elections and voters become available, redistricting simulation methods are playing an increasingly important role when legislatures adopt redistricting plans and courts determine their legality. These simulation methods are designed to yield a representative sample of all redistricting plans that satisfy statutory guidelines and requirements such as contiguity, population parity, and compactness. A proposed redistricting plan can be considered gerrymandered if it constitutes an outlier relative to this sample according to partisan fairness metrics. Despite their growing use, an insufficient effort has been made to empirically validate the accuracy of the simulation methods. We apply a recently developed computational method that can efficiently enumerate all possible redistricting plans and yield an independent uniform sample from this population. We show that this algorithm scales to a state with a couple of hundred geographical units. Finally, we empirically examine how existing simulation methods perform on realistic validation data sets., Comment: 32 pages, 14 figures

Using the concept of principal stratification from the causal inference literature, we introduce a new notion of fairness, called principal fairness, for human and algorithmic decision-making. The key idea is that one should not discriminate among individuals who would be similarly affected by the decision. Unlike the existing statistical definitions of fairness, principal fairness explicitly accounts for the fact that individuals can be impacted by the decision. Furthermore, we explain how principal fairness differs from the existing causality-based fairness criteria. In contrast to the counterfactual fairness criteria, for example, principal fairness considers the effects of decision in question rather than those of protected attributes of interest. We briefly discuss how to approach empirical evaluation and policy learning problems under the proposed principal fairness criterion.

In recent years, fully automated content analysis based on probabilistic topic models has become popular among social scientists because of their scalability. The unsupervised nature of the models makes them suitable for exploring topics in a corpus without prior knowledge. However, researchers find that these models often fail to measure specific concepts of substantive interest by inadvertently creating multiple topics with similar content and combining distinct themes into a single topic. In this paper, we empirically demonstrate that providing a small number of keywords can substantially enhance the measurement performance of topic models. An important advantage of the proposed keyword assisted topic model (keyATM) is that the specification of keywords requires researchers to label topics prior to fitting a model to the data. This contrasts with a widespread practice of post-hoc topic interpretation and adjustments that compromises the objectivity of empirical findings. In our application, we find that keyATM provides more interpretable results, has better document classification performance, and is less sensitive to the number of topics than the standard topic models. Finally, we show that keyATM can also incorporate covariates and model time trends. An open-source software package is available for implementing the proposed methodology.

Many causal processes have spatial and temporal dimensions. Yet the classic causal inference framework is not directly applicable when the treatment and outcome variables are generated by spatio-temporal point processes. We extend the potential outcomes framework to these settings by formulating the treatment point process as a stochastic intervention. Our causal estimands include the expected number of outcome events in a specified area under a particular stochastic treatment assignment strategy. Our methodology allows for arbitrary patterns of spatial spillover and temporal carryover effects. Using martingale theory, we show that the proposed estimator is consistent and asymptotically normal as the number of time periods increases. We propose a sensitivity analysis for the possible existence of unmeasured confounders, and extend it to the Hajek estimator. Simulation studies are conducted to examine the estimators' finite sample performance. Finally, we illustrate the proposed methods by estimating the effects of American airstrikes on insurgent violence in Iraq from February 2007 to July 2008. Our analysis suggests that increasing the average number of daily airstrikes for up to one month may result in more insurgent attacks. We also find some evidence that airstrikes can displace attacks from Baghdad to new locations up to 400 kilometers away

In this paper, we evaluate Apache Spark for a data-intensive machine learning problem. Our use case focuses on policy diffusion detection across the state legislatures in the United States over time. Previous work on policy diffusion has been unable to make an all-pairs comparison between bills due to computational intensity. As a substitute, scholars have studied single topic areas. We provide an implementation of this analysis workflow as a distributed text processing pipeline with Spark dataframes and Scala application programming interface. We discuss the challenges and strategies of unstructured data processing, data formats for storage and efficient access, and graph processing at scale.

This commentary has two goals. We first critically review the deconfounder method and point out its advantages and limitations. We then briefly consider three possible ways to address some of the limitations of the deconfounder method.

The increasing availability of individual-level data has led to numerous applications of individualized (or personalized) treatment rules (ITRs). Policy makers often wish to empirically evaluate ITRs and compare their relative performance before implementing them in a target population. We propose a new evaluation metric, the population average prescriptive effect (PAPE). The PAPE compares the performance of ITR with that of non-individualized treatment rule, which randomly treats the same proportion of units. Averaging the PAPE over a range of budget constraints yields our second evaluation metric, the area under the prescriptive effect curve (AUPEC). The AUPEC represents an overall performance measure for evaluation, like the area under the receiver and operating characteristic curve (AUROC) does for classification, and is a generalization of the QINI coefficient utilized in uplift modeling. We use Neyman's repeated sampling framework to estimate the PAPE and AUPEC and derive their exact finite-sample variances based on random sampling of units and random assignment of treatment. We extend our methodology to a common setting, in which the same experimental data is used to both estimate and evaluate ITRs. In this case, our variance calculation incorporates the additional uncertainty due to random splits of data used for cross-validation. The proposed evaluation metrics can be estimated without requiring modeling assumptions, asymptotic approximation, or resampling methods. As a result, it is applicable to any ITR including those based on complex machine learning algorithms. The open-source software package is available for implementing the proposed methodology., Comment: Accepted at JASA