Effect of readability of political tweets on positive user engagement

Twitter is one of the essential information environments where novel information join and diffuse through the public discourse via user engagement. Although the previous work has shed light on the role of the content in information diffusion, stylistic factors such as readability are still under-explored. We investigated the effect of the readability on positive user engagement for over 80,000 political tweets collected for a period of six months. We formulated a set of experiments involving the use of regression techniques to predict user engagement. Our findings indicate that the addition of the readability related features leads to more accurate and more robust predictions. Increase in the prediction robustness means that ease-of-read have a considerable influence on the positive engagement political tweets receive.

I. Introduction

Micro-blogging platforms like Twitter play an essential role in the diffusion of information and spread of news, political agenda, and rumours. Users on Twitter share their views and other information via short messages that are no longer than 280 characters. This character limit at first may seem appealing because it can make tweets easy to read due to shortness. Nevertheless, it also can make the tweets harder to read compared to longer texts ubiquitous on the web (e.g. blog articles). It is because users may need to use acronyms, abbreviations, and social media-specific features and writing styles like hashtags, emojis, and mentions to convey more information using fewer characters.