Leveraging Brand Google Search Trends to Forecast Election Results

• In Event: Campaigns and Elections

Abstract

This paper presents a novel approach to forecasting election results by leveraging brand-related Google Search trends as a proxy for political orientation. Building on Brand Identity Theory, which posits that brand preferences can serve as extensions of individual identities, and supported by findings from social network analyses, the study examines the predictive power of search interest in particular brands of cars, news outlets, and clothing lines.

The data, comprising Google Trends from 2016 to 2020, socioeconomic factors from the American Community Survey, and 2020 US election outcomes, is analyzed using decision tree models. These models identify significant correlations between brand preferences and political leanings. The study's findings suggest that search patterns extend beyond mere consumer behavior, offering insights into electoral behavior and opening a new avenue for understanding the intricate relationships between consumer choices and political dispositions.

The research illustrates the potential of integrating digital trace data with traditional data sources to enhance the accuracy and depth of political forecasting methods.

Author

• Joy Youn Soo Kim, Columbia University