Voter Turnouts Govern Key Electoral Statistics

TL;DR

This paper demonstrates that voter turnout data can accurately predict key electoral statistics and reveals universal patterns in election margins across different scales and contexts.

Contribution

It introduces a random voting model to analytically derive vote distributions and validates these predictions with extensive Indian election data, uncovering scale-invariant behaviors.

Findings

Strong correlation between turnout and electoral statistics

Universal distribution patterns in margins of victory

Scale-invariant behavior in scaled margins

Abstract

Elections, the cornerstone of democratic societies, are usually regarded as unpredictable due to the complex interactions that shape them at different levels. In this work, we show that voter turnouts contain crucial information that can be leveraged to predict several key electoral statistics with remarkable accuracy. Using the recently proposed random voting model, we analytically derive the scaled distributions of votes secured by winners, runner-ups, and margins of victory, and demonstrating their strong correlation with turnout distributions. By analyzing Indian election data -- spanning multiple decades and electoral scales -- we validate these predictions empirically across all scales, from large parliamentary constituencies to polling booths. Further, we uncover a surprising scale-invariant behavior in the distributions of scaled margins of victory, a characteristic signature of Indian elections. Finally, we demonstrate a robust universality in the distribution of the scaled margin-to-turnout ratios.