Interrupted time series analysis of clickbait on worldwide news websites, 2016-2023

TL;DR

This study analyzes the prevalence and variation of clickbait on worldwide news websites from 2016 to 2023, revealing how major events like COVID-19 and the US Election affected clickbait levels.

Contribution

It introduces a large-scale dataset of clickbait scores and applies segmented regression models to analyze temporal changes related to major news events.

Findings

Clickbait levels varied significantly during major news events.

COVID-19 and the 2020 US Election influenced clickbait prevalence.

The dataset enables large-scale analysis of clickbait trends over time.

Abstract

Clickbait is deceptive text that can manipulate web browsing, creating an information gap between a link and target page that literally baits a user into clicking. Clickbait detection continues to be well studied, but analyses of clickbait overall on the web are limited. A dataset was built consisting of 451,033,388 clickbait scores produced by a clickbait detector which analyzed links and headings on primarily English news pages from the Common Crawl. On this data, 5 segmented regression models were fit on 5 major news events and averaged clickbait scores. COVID and the 2020 US Election appeared to influence clickbait levels.