Free Elections in the Free State:Ensemble Analysis of Redistricting in New Hampshire

Atticus McWhorter, Daryl DeFord · 2025

Read Free Elections in the Free State:Ensemble Analysis of Redistricting in New Hampshire by Atticus McWhorter, Daryl DeFord (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2509-07328.txt.

Introduction

New Hampshire's legislative maps drew lawsuits after the 2020 census, with claims about fairness and gerrymandering. Atticus McWhorter and Daryl DeFord answer with an ensemble analysis in a 2025 research paper. They encode the state's redistricting rules, then generate a large collection of alternative maps by computer. The enacted maps can then be judged against that neutral baseline.

Their comparison shows how the chosen districts trade off geographic goals against partisan outcomes. The analysis also tests non-partisan justifications offered for the enacted plan. A further finding concerns the election data itself: how results are selected and combined changes measures of partisan symmetry. The work gives courts and the public a firmer mathematical footing for judging whether a map is an outlier.

Catalog record

CREATORAtticus McWhorter, Daryl DeFord
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE32 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2509-07328

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Free Elections in the Free State:Ensemble Analysis of Redistricting in New Hampshire Atticus McWhorter, Daryl DeFord Abstract The process of legislative redistricting in New Hampshire, along with many other states across the country, was particularly contentious during the 2020 census cycle. In this paper we present an ensemble analysis of the enacted districts to provide mathematical context for claims made about these maps in litigation. Operationalizing the New Hampshire redistricting rules and algorithmically generating a large collection of districting plans allows us to construct a baseline for expected behavior of districting plans in the state and evaluate non-partisan justifications and geographic tradeoffs between districting criteria and partisan outcomes. In addition, our results demonstrate the impact of selection and aggregation of election data for analyzing partisan symmetry measures. keywords Computational Redistricting, Markov Chain Monte Carlo, Ensemble Analysis, Spanning Trees, New Hampshire 1 Introduction While computational methods for optimizing political redistricting plans for multiple criteria have been been applied since at least the 1960’s Becker and Solomon (2020) , recent advances in data availability and computational power have created both new tools and new challenges for analyzing enacted and proposed maps. In this paper we apply one of these modern techniques, the ensemble method Clelland et al. (2020) , to create and analyze a large collection of districting plans for electoral districts in the State of New Hampshire. This approach allows us to formulate a baseline of reasonable values across a wide variety of metrics for each type of district, while directly incorporating the state’s political geography and legislative constraints. Individual numerical measurements of compactness have long been applied to detect partisan gerrymandering, following the intuition that achieving improper partisan aims would require distorting and perturbing the boundaries of districts away from regular or smooth shapes. Examples of shape-based measurements include the isoperimetric ratio, known as the Polsby-Popper score in this domain, or the Reock score, which measures the ratio between a district and its circumscribing circle. These metrics have recently been criticized Barnes and Solomon (2021) ; Assaf Bar-Natan and Schutzman (2020) ; Duchin and Tenner (2024) as insufficient to detect modern gerrymanders. North Carolina’s Congressional maps in the previous cycle provide an example of this issue, as a particularly non-compact map was struck down as unconstitutional Harper et al. (2023a) and replaced with a map that had similar partisan properties but much better performance on the shape-based measures Harper et al. (2023b) ; Herschlag et al. (2020) . Similarly, many measurements have been proposed for directly evaluating the partisan fairness of a districting plan, without necessarily taking into account the shape of the districts. These include measures of partisan symmetry Grofman (1983) ; Katz et al. (2020) , as well as the efficiency gap Bernstein and Duchin (2017) ; Stephanopoulos and McGhee (2014) ; Veomett (2020) , declination Campisi et al. (2019) ; Warrington (2018) , partisan dislocation Eubank and Rodden (2020) ; DeFord et al. (2022) and other recent geographic extensions Barton and Eguia (To Appear) ; Wise (To Appear) , and the GEO metric Campisi et al. (2022) . These measures often imply specific and sometimes contradictory normative definitions of fairness with ideal values but relying on a single measurement of this sort to define a fair plan can sometimes lead to unintuitive results DeFord et al. (2023) ; DeFord and Veomett (To Appear) . Even simple measures of proportionality can fail to be achievable, even if it is not required or endorsed by the courts. For example, in several elections in Massachusetts with approximately a 30% Republican vote share an outcome that does not elect any Republican representatives is required by a single member district system due to the homogeneity of the partisan distribution Duchin et al. (2018) . That said, recent analysis of the “Freedom to Vote Test” with a proportional baseline Duchin and Schoenbach (2022) demonstrates possible advantages of this measure. The ensemble method attempts to address this problem by explicitly taking into account a specific state’s political geography in the formation of districting plans. By creating a large collection of maps sampled from among the enormous collection of possibilities it is possible to provide context for the values of any metric applied to a specific map under consideration. For example, if the value of a metric applied to an enacted map lies in the extreme tail of the distribution of values over a neutral ensemble that provides evidence about the intentions of the line drawer Herschlag et al. (2020) . Over the past decade, these ensemble methods have become increasingly relevant in court cases and reform efforts about political redistricting. State courts across the country have heard testimony about ensembles and many of the current Supreme Court justices has blessed their use in various applications. For example, in the Rucho v. Common Cause dissent, Justice Kagan argues that the creation of ensembles would prove beneficial toward determining whether a map is constitutional Kagan (2019) , and as Clelland et al. (2021) note, the methods have been used widely. Ensemble methods have been widely successful both in analyzing the nation-level representative bodies and state legislatures Caldera et al. (2020) ; Carter et al. (2020) ; DeFord and Duchin (2019) ; Dobbs et al. (2023) ; Herschlag et al. (2020) ; Herschlag et al. (2017) ; Zhao et al. (2022) . Especially in light of Rucho v. Common Cause, which ruled that partisan redistricting is not a justiciable issue, state-level courts have come to play an increasingly powerful role in the process of evaluating legislative maps, leading to increased attention. In the recent court challenge to the New Hampshire’s State Senate and Executive Council districts, no ensemble analysis was used to support evaluations of quantitative measures of symmetry, compactness, or fairness applied to the enacted plans Brown et al. (2022) . The plaintiffs do argue that misshapen districts were formed to pack together left-leaning cities to increase Republican representation in the remainder of the state. Additionally, the lawsuit does not provide a complete analysis of the impacts of the volatility of the electorate on the fairness metrics. As we demonstrate in Figure 1 , in a single year voters may overwhelmingly support a Democratic candidate for President and a Republican candidate for Governor. And as such, the choice of election data has large ramifications on the election outcome when used to analyze districting plans. To analyze this issue and provide additional content for the evaluation of the plans, we apply quantitative measures to an ensemble of millions of plausible State Senate and Executive Council districting plans. These ensembles are overlaid with data from statewide elections that occurred during the previous census cycle, allowing us to perform a detailed analysis of the currently enacted plan. We then proceed to address the claims made in the lawsuit and other related evaluations of the potential maps for the state. 2 Methods This section describes the technical choices that generated the ensemble. Most computations were done on an Apple M2 CPU, with memory-intense tasks being preformed on Dartmouth’s d oob cluster, a 96-core 3.6GHz AMD Epyc system. The project uses the 0.2.17 release of GerryChain. 2.1 Data The data for this project was taken from the Metric Geometry and Gerrymandering

Read the complete text