[Paper Review] Pot and ladle: seat allocation and seat bias under the Jefferson-D'Hondt method

Jarosław Flis, Wojciech Słomczyński+1 more|arXiv (Cornell University)|May 21, 2018

Electoral Systems and Political Participation47 references3 citations

TL;DR

This paper introduces a novel formula to estimate national seat shares under the Jefferson-D'Hondt method in multi-district elections, based solely on national vote shares and electoral parameters. It reveals how seat bias depends on the number of parties and districts, provides highly accurate estimates across eight EU countries, and generalizes the efficiency gap for gerrymandering analysis.

ABSTRACT

We propose a simple and new formula for estimating national seat shares in multidistrict elections employing the Jefferson-D'Hondt method for seat allocation solely on the basis of the national vote shares and fixed parameters of the electoral system. The proposed formula clarifies the relationship between seat bias and both the number of parties and the number of districts. We discuss how our formula differs from the simple generalization of the single-district asymptotic seat bias formulae and what assumptions must hold for our method to give exact results. We further demonstrate that despite minor violations of those assumptions, the formula provides a good estimate of actual seat allocations for all EU countries that have a national party system and employ the Jefferson-D'Hondt seat allocation method in multi-district parliamentary elections, i.e., Croatia, the Czech Republic, Finland, Luxembourg, the Netherlands, Poland, Portugal, and Spain. Moreover, we show that the formula constitutes a generalization and extension of the McGhee-Stephanopoulos efficiency gap test for political gerrymandering. Finally, we present a number of additional applications of the formula for the evaluation of political strategies.

Motivation & Objective

• To develop a simple, accurate formula for estimating national seat shares in multi-district elections using only national vote shares and system parameters.
• To clarify the relationship between seat bias and the number of parties and districts under the Jefferson-D'Hondt method.
• To assess the conditions under which the formula yields exact results and evaluate its robustness under minor assumption violations.
• To demonstrate the formula’s applicability and accuracy across real-world multi-district systems in EU countries.
• To extend the efficiency gap framework for gerrymandering detection by generalizing it to multi-district, multi-party systems.

Proposed method

• The authors derive a closed-form analytical formula that estimates seat shares based on national vote shares and fixed electoral parameters such as district number and threshold.
• The formula is grounded in the mathematical structure of the Jefferson-D'Hondt method and accounts for the non-linear allocation process across multiple districts.
• It generalizes single-district asymptotic seat bias formulae by incorporating inter-district variation and the number of competing parties.
• The method assumes uniform party vote distribution across districts and fixed district magnitude, though it remains robust under mild deviations.
• The formula is validated by comparing predicted seat shares against actual outcomes in eight EU countries using historical election data.
• The approach is further applied to analyze political strategy, such as vote distribution and districting effects, by simulating seat allocation under different scenarios.

Experimental results

Research questions

• RQ1 How does seat bias in the Jefferson-D'Hondt method vary with the number of political parties and the number of electoral districts?
• RQ2 To what extent can seat shares be accurately predicted from national vote shares alone, without district-level vote data?
• RQ3 How does the proposed formula differ from single-district asymptotic seat bias approximations?
• RQ4 Under what conditions does the formula yield exact results, and how robust is it to violations of these assumptions?
• RQ5 Can the formula be extended to generalize the efficiency gap test for detecting gerrymandering in multi-district, multi-party systems?

Key findings

• The formula provides highly accurate estimates of actual seat allocations across eight EU countries—Croatia, the Czech Republic, Finland, Luxembourg, the Netherlands, Poland, Portugal, and Spain—despite minor violations of ideal assumptions.
• Seat bias increases with the number of parties and is amplified in systems with more districts, highlighting structural distortions in the Jefferson-D'Hondt method.
• The formula generalizes the McGhee-Stephanopoulos efficiency gap by extending its applicability from single-district to multi-district, multi-party electoral systems.
• The method outperforms simple generalizations of single-district asymptotic formulae by accounting for inter-district variation and the number of competing parties.
• The formula enables practical evaluation of political strategies, such as optimal vote distribution across districts to maximize seat share.
• The results confirm that the Jefferson-D'Hondt method systematically favors larger parties, and the formula quantifies this bias with precision.