arXiv subjects

Evan T. R. Rosenman

Publications and source records attributed to Evan T. R. Rosenman.

10 recordsLinked to original sources

arXiv2026

Statistical Assessments of Representational Reforms: A Case Study from Los Angeles

Electoral reforms -- including changes to district boundaries, electoral rules, and the size of elected bodies -- are a recurrent feature of American state and municipal politics. The representational impact of such reforms is challenging to assess before they are implemented, and even well-intentioned civic reformers are often surprised by the unintended consequences of changes to the political system. We develop a statistical framework for the prospective evaluation of electoral reforms by integrating several modes of analysis: descriptive statistics of the voter file, probabilistic race and ethnicity imputation, ecological inference, and redistricting simulations via ensembles. Together, these tools provide a unified assessment of how proposed reforms affect whose voices are ultimately heard in government. We apply this framework to reforms currently being considered for the Los Angeles City Council, including increasing the number of single-member districts, adopting ranked-choice voting, and introducing multimember districts elected by proportional ranked choice voting. We find that expanding the Council alone is unlikely to substantially improve representation for underrepresented communities. In contrast, eliminating low-turnout primary elections and adopting multimember districts produce larger improvements in representational equity. Our analyses were shared with L.A.'s Charter Reform Commission via public testimony in late 2025. Although motivated by Los Angeles, the framework is broadly applicable to evaluating prospective electoral reforms across American jurisdictions.

stat.AP↗

arXiv2026

Adaptive Experimental Design Using Shrinkage Estimators

In multi-armed trials, adaptive designs are a popular way to increase estimation efficiency or identify optimal treatments. Several recent papers have proposed adaptive variants of the classical Neyman allocation to assign treatments in sequential trials, with the goal of minimizing the error of a Horvitz-Thompson-style estimator. However, this approach may be inefficient, because it fails to borrow information across the treatment arms. In this paper, we consider adaptivity in a sequential trial with K active treatments and a control, and suggest the use of Stein-like shrinkage estimators to obtain the final causal estimates. These estimators share information across arms, yielding provable reductions in expected squared error loss relative to estimating each causal effect in isolation. Moreover, for each of our candidate shrinkers, the risk is the expectation of ratios of Gaussian quadratic forms, and can be computed efficiently via numerical integration. Hence, we suggest a simple algorithm for sequential adaptivity: assign treatments to each new arrival by choosing the arm that will minimize the estimated shrinker loss. Through simulations, we demonstrate that this approach can yield meaningful reductions in estimation error, especially in the low signal-to-noise regime. We also characterize how our adaptive algorithm assigns treatments differently than would a sequential Neyman allocation, and suggest a method for constructing shorter confidence intervals at the trial's conclusion.

stat.ME↗

arXiv2026

Bayesian Uncertainty Quantification for Ranked Choice Voting Polls

Ranked choice voting (RCV) is a popular alternative voting method in which voters are asked to list their favored candidates in preference order, rather than vote for a single candidate. When these ballots are tabulated, candidates are successively eliminated, and their votes are reallocated to each voter's next-preferred choice. The process continues until a candidate commands a majority of the active ballots and is declared the winner. As RCV gains wider adoption, the method poses novel challenges for pollsters. Unlike plurality elections, the event that a candidate wins cannot be expressed in terms of a single population parameter. Hence, the basic concept of a margin-of-error is not straightforward to define. Moreover, a candidate's ability to win may depend on both their support across the ballot and the order in which other candidates are eliminated. Existing measures of sampling uncertainty for polls of RCV elections do not clearly quantify these path-dependent outcomes. Here, we propose a simple, Bayesian framework to quantify uncertainty in polls of RCV elections. We cast the problem as one of estimating win probabilities for each leading candidate, and leverage a simple conjugacy relationship to estimate these probabilities conditional on the poll results. We include applied analyses involving two prominent ranked choice voting elections: the 2021 New York City Democratic mayoral primary, in which Eric Adams narrowly defeated Kathryn Garcia in the final round; and the 2022 special election to Alaska's U.S. House seat, in which Mary Peltola was elected despite not being a Condorcet winner. Using the cast vote records from both elections, we demonstrate some challenges of traditional frequentist uncertainty quantification in RCV polls. We also demonstrate the utility of our approach using a poll of the NYC primary obtained from the polling firm Data for Progress.

stat.AP↗

arXiv2023

Empirical Bayes Double Shrinkage for Combining Biased and Unbiased Causal Estimates

Motivated by the proliferation of observational datasets and the need to integrate non-randomized evidence with randomized controlled trials, causal inference researchers have recently proposed several new methodologies for combining biased and unbiased estimators. We contribute to this growing literature by developing a new class of estimators for the data-combination problem: double-shrinkage estimators. Double-shrinkers first compute a data-driven convex combination of the the biased and unbiased estimators, and then apply a final, Stein-like shrinkage toward zero. Such estimators do not require hyperparameter tuning, and are targeted at multidimensional causal estimands, such as vectors of conditional average treatment effects (CATEs). We derive several workable versions of double-shrinkage estimators and propose a method for constructing valid Empirical Bayes confidence intervals. We also demonstrate the utility of our estimators using simulations on data from the Women's Health Initiative.

stat.ME↗

arXiv2022

Comment: The Essential Role of Policy Evaluation for the 2020 Census Disclosure Avoidance System

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.

stat.AP↗

arXiv2022

Addressing Census data problems in race imputation via fully Bayesian Improved Surname Geocoding and name supplements

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.

stat.ML↗

arXiv2022

Race and ethnicity data for first, middle, and last names

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.

stat.OT↗

arXiv2022

Designing Experiments Toward Shrinkage Estimation

We consider how increasingly available observational data can be used to improve the design of randomized controlled trials (RCTs). We seek to design a prospective RCT, with the intent of using an Empirical Bayes estimator to shrink the causal estimates from our trial toward causal estimates obtained from an observational study. We ask: how might we design the experiment to better complement the observational study in this setting? We propose using an estimator that shrinks each component of the RCT causal estimator toward its observational counterpart by a factor proportional to its variance. First, we show that the risk of this estimator can be computed efficiently via numerical integration. We then propose algorithms for determining the best allocation of units to strata (the best "design"). We consider three options: Neyman allocation; a "naive" design assuming no unmeasured confounding in the observational study; and a "defensive" design accounting for the imperfect parameter estimates we would obtain from the observational study with unmeasured confounding. We also incorporate results from sensitivity analysis to establish guardrails on the designs, so that our experiment could be reasonably analyzed with and without shrinkage. We demonstrate the superiority of these experimental designs with a simulation study involving causal inference on a rare, binary outcome.

stat.ME↗

arXiv2021

Recalibration of Predictive Models as Approximate Probabilistic Updates

The output of predictive models is routinely recalibrated by reconciling low-level predictions with known derived quantities defined at higher levels of aggregation. For example, models predicting turnout probabilities at the individual level in U.S. elections can be adjusted so that their aggregation matches the observed vote totals in each state, thus producing better calibrated predictions. In this research note, we provide theoretical grounding for one of the most commonly used recalibration strategies, known colloquially as the "logit shift." Typically cast as a heuristic optimization problem (whereby an adjustment is found such that it minimizes the difference between aggregated predictions and the target totals), we show that the logit shift in fact offers a fast and accurate approximation to a principled, but often computationally impractical adjustment strategy: computing the posterior prediction probabilities, conditional on the target totals. After deriving analytical bounds on the quality of the approximation, we illustrate the accuracy of the approach using Monte Carlo simulations. The simulations also confirm analytical results regarding scenarios in which users of the simple logit shift can expect it to perform best -- namely, when the aggregated targets are comprised of many individual predictions, and when the distribution of true probabilities is symmetric and tight around 0.5.

stat.ME↗

arXiv2021

Robust Designs for Prospective Randomized Trials Surveying Sensitive Topics

We consider the problem of designing a prospective randomized trial in which the outcome data will be self-reported, and will involve sensitive topics. Our interest is in misreporting behavior, and how respondents' tendency to under- or overreport a binary outcome might affect the power of the experiment. We model the problem by assuming each individual in our study is a member of one "reporting class": a truth-teller, underreporter, overreporter, or false-teller. We show that the joint distribution of reporting classes and "response classes" (characterizing individuals' response to the treatment) will exactly define the bias and variance of the causal estimate in our experiment. Then, we propose a novel procedure for deriving sample sizes under the worst-case power corresponding to a given level of misreporting. Our problem is motivated by prior experience implementing a randomized controlled trial of a sexual violence prevention program among adolescent girls in Nairobi, Kenya.

stat.ME↗