By Dr. Samir Abu Rumman
This article was published in Arabic before Election Day on November 5, 2024.
There is no doubt that American public opinion polls have a long history, dating back to the 1930s, of accurately predicting the results of US presidential elections and identifying the winner of the race to the White House through hundreds of polls conducted by various organizations. During these decades of successes, there have been some limited failures, such as in 1948, when nearly all polls predicted a victory for Republican candidate Thomas Dewey over Democratic candidate Harry Truman. The polls at that time relied on traditional methods and samples that were not fully representative of voters, resulting in a surprising outcome with Truman’s victory. A similar situation occurred in 2016, when most polls predicted a win for Hillary Clinton over Donald Trump, only for the latter to emerge victorious.
Nevertheless, the role of opinion polls has remained essential in understanding public opinion trends and predicting election outcomes. They have seen significant advancements in accuracy, sample representation, and the adoption of advanced analytical methods! In this past particular presidential election, polls reached a level of forecasting that, given scientific standards, the scope, value, accuracy, and nature of the polls, we still cannot know the winner! The figures in articles and writings that compile results from various polling centers, such as the article “I think the pollsters will win this election” by Dr. Claire Durand, a sociology professor at the University of Montreal and former president of the World Association for Public Opinion Research (WAPOR), discuss the accuracy of polls in the 2024 US presidential election. The author notes that polls do not reveal a clear winner due to the close results, with support for each candidate ranging between 48% and 52%. She suggests that if the polls are incorrect, it might be due to last-minute shifts in voter preferences, although she considers this unlikely, pointing to advancements in data collection and analysis that have improved result accuracy.
Given this reality, pollsters must acknowledge that reaching a decisive result may not always be possible in certain cases. It is essential to focus on the differences that do not definitively close the gap, taking into account the known margin of statistical error, which could favor one of the candidates. Therefore, the forecasts of polls regarding the 2024 US presidential election represent a critical test, requiring the public, researchers, and opinion analysts to exercise humility. We must admit that we cannot answer the question that preoccupies the world today: who will win this electoral battle, which has reached an unprecedented level of intense competition? It is an opportunity to practically and compellingly illustrate to people the realities of sampling errors and significant non-sampling errors!
This article was published in Arabic before Election Day on November 5, 2024. It has been translated and edited using human supervised AI with the English translation appearing exclusively on Generation1.ca.
Samir Abu-Rumman | Princeton Politics is a visiting research scholar at Schar School of Policy and Government, George Mason University in Washington DC, with extensive years of experience in research, education, and development in different countries. He is the supervisor of “World of Opinions” in Kuwait, Jordan, and the US, has led and supervised different regional and global research projects for organizations such as the World Values Survey and Arab Barometer for Princeton University and also last participated as a keynote panelist in our fall 2023 virtual insights career fair and case competition.
