From Toxicity in Online Comments to Incivility in American News: Proceed with Caution

TL;DR

This paper evaluates the effectiveness of toxicity detection models like Perspective API in identifying incivility in American news, revealing their limitations and the need for improved methods to align with human judgments.

Contribution

It demonstrates the inadequacy of current toxicity models for news incivility detection and highlights the importance of addressing spurious correlations for accurate analysis.

Findings

Toxicity models perform poorly on news incivility detection.

Spurious correlations between words and incivility affect model accuracy.

Current models may lead to misleading conclusions about incivility.

Abstract

The ability to quantify incivility online, in news and in congressional debates, is of great interest to political scientists. Computational tools for detecting online incivility for English are now fairly accessible and potentially could be applied more broadly. We test the Jigsaw Perspective API for its ability to detect the degree of incivility on a corpus that we developed, consisting of manual annotations of civility in American news. We demonstrate that toxicity models, as exemplified by Perspective, are inadequate for the analysis of incivility in news. We carry out error analysis that points to the need to develop methods to remove spurious correlations between words often mentioned in the news, especially identity descriptors and incivility. Without such improvements, applying Perspective or similar models on news is likely to lead to wrong conclusions, that are not aligned with the human perception of incivility.