Jury-Contestant Bipartite Competition Network: Identifying Biased Scores and Their Impact on Network Structure Inference

TL;DR

This paper models judge-contestant scoring as a bipartite network to detect biases and assess their effects on the inferred structure, highlighting the importance of bias detection in competitions.

Contribution

It introduces a network-based method to identify biased scores and evaluate their impact on competition structure inference, demonstrated on a real-world example.

Findings

A single biased score can significantly distort network structure inference.

No significant racial bias was found among judges in the analyzed competition.

Bias detection is crucial for fair assessment in judged competitions.

Abstract

A common form of competition is one where judges grade contestants' performances which are then compiled to determine the final ranking of the contestants. Unlike in another common form of competition where two contestants play a head-to-head match to produce a winner as in football or basketball, the objectivity of judges are prone to be questioned, potentially undermining the public's trust in the fairness of the competition. In this work we show, by modeling the judge--contestant competition as a weighted bipartite network, how we can identify biased scores and measure their impact on our inference of the network structure. Analyzing the prestigious International Chopin Piano Competition of 2015 with a well-publicized scoring controversy as an example, we show that even a single statistically uncharacteristic score can be enough to gravely distort our inference of the community structure, demonstrating the importance of detecting and eliminating biases. In the process we also find that there does not exist a significant system-wide bias of the judges based on the the race of the contestants.