Predicting the winner of the US 2024 elections using trust analytics

TL;DR

This paper introduces Trust Analytics, a social media-based tool that predicts US 2024 election outcomes by analyzing public trust and distrust in candidates, offering an alternative to traditional polling methods.

Contribution

It presents a novel approach using social media trust metrics to forecast election results, diverging from emotion-based sentiment analysis.

Findings

Trust ratio predicted Trump as the winner.

Trust and distrust levels showed week-to-week fluctuations.

The model's predictions aligned with election outcomes.

Abstract

A number of models and techniques has been proposed for predicting the outcomes of presidential elections. Some of them use information on the socio-economical status of a country, others focus on candidates' popularity measures in news media. We employ a computational social science approach, utilising public reactions in social media to real-life events that involve presidential candidates. Contrary to the popular approach, we do not analyse public emotions but ethotic references to the character of politicians which allows us to analyse how much they are (dis-)trusted by the general public, hence the name of the tool we developed: Trust Analytics (TrustAn). Similarly to major news media's polls, we observe a tight race between Harris and Trump with week to week changes in the level of trust and distrust towards the two candidates. Using the ratio between the level of trust and distrust towards them and changes of this metric in time, we predict Donald Trump as the winner of the US 2024 elections.