Rabat - As the US’ 2024 presidential election approaches, political analysts and journalists are diving deep into the data, sifting through polls and early-voting statistics to forecast who Vice President Kamala Harris and former President Donald Trump will emerge victorious.
However, a team of researchers at New York University (NYU) led by Anasse Bari, a professor of computer science, has taken an unorthodox approach to understanding voter sentiment. The group is examining online searches related to yard signs and campaign merchandise associated with the two candidates.
Since Harris declared her candidacy on July 21, she has dominated searches for yard signs, garnering 76% of online interest compared to Trump's 24%.
This data, compiled by the Predictive Analytics and AI Research Lab at NYU's Courant Institute of Mathematical Sciences, suggests that yard sign searches may reflect potential voter enthusiasm and support.
Shifting trends in merchandise searches
While yard signs indicate a level of grassroots support, the search trends for candidate-related attire tell a different story. From July to mid-October, searches for "Trump hat" outpaced those for "Harris hat" by a ratio of 2 to 1.
Interestingly, during the early weeks of Harris's campaign, searches for “Harris shirts” actually led “Trump shirts” by approximately 15%. However, this trend flipped dramatically; by September 1, “Trump shirts” searches surged ahead by 129%.
“If someone is searching for their candidate’s yard sign, shirt, hat, flag, or bumper sticker, there is a strong likelihood that they will vote for them,” explained Anasse Bari on the NYU website. “While it’s uncertain how predictive these behaviors are of election results, search data that supplement traditional election-prediction methods may enhance the accuracy of political forecasts.”
This innovative approach mirrors findings from the fall of 2020, when the same NYU team discovered that online searches for yard signs had indicated Biden’s support over Trump by a substantial margin.
This historical context adds weight to the current findings and suggests that yard signs and merchandise may serve as modern indicators of political engagement.
Recent dynamics and external influences
Recent months have seen fluctuations in yard sign searches, particularly following Trump's controversial conviction for falsifying business records in late May.
During this time, interest in Trump yard signs surged, peaking during the week of his conviction and illustrating how external events can impact voter sentiment.
Moreover, a notable social media influence emerged when Elon Musk encouraged Trump supporters to showcase their yard signs in Democratic neighborhoods via an October 17 tweet.
This call to action may have contributed to a spike in searches for Trump yard signs, marking the first time this cycle that searches for Trump signs surpassed those for Harris, despite Harris's earlier lead.
“In the realm of political analysis, understanding online behavior can illuminate public sentiment,” Bari stated in an interview on CNN. “Our research provides an additional layer of insight that can complement traditional polling methods.”
Swing State searches highlight voter sentiment
The study also analyzed data from key swing states—Arizona, Georgia, Michigan, Nevada, North Carolina, Pennsylvania, and Wisconsin. In these battlegrounds, searches for “Trump hat” comprised at least 60% of total searches, while “Harris hat” searches never exceeded 40%.
Interestingly, Harris's campaign had notable success with her "Harris Walz hat," which accounted for at least 84% of searches related to hats in these states, overshadowing “Trump Vance hats,” which garnered less than 16%.
This disparity in search interest not only highlights voter preferences but also the potential significance of merchandise in shaping campaign dynamics.
Methodology and implications for future research
The NYU researchers relied on publicly available Google search data, focusing on specific terms such as “Harris Yard Sign” and “Trump-Vance Yard Sign” to develop a comprehensive index for each candidate.
Their analysis included assessing sentiment through natural language processing algorithms, capturing the tone of discussions surrounding the candidates on social media platforms.
Bari emphasized the novelty of their approach. “We are exploring innovative ways to measure public opinion beyond traditional metrics like polls and campaign contributions,” he said. “The behavior we see online reflects real interest and potential support for the candidates.”
Looking ahead, Bari envisions a future where such search data can enhance traditional polling methods, providing deeper insights into voter sentiment.
“As we move closer to the elections, it’s crucial to consider how online searches can serve as a gauge of enthusiasm and support,” he concluded. “This research not only sheds light on current trends but could also inform strategies for upcoming campaigns.”