Nobody thought that Harry Truman would really win the presidential election of 1948. The media seemed to unanimously believe in Thomas Dewey. However, on November 4, 1948, the country woke up to shocking news: Truman did in fact win the election, with 303 electoral votes.
Many people look to the inaccuracy of the polls as a crucial reason why this election came to be seen as the greatest election upset in the history of the United States. The main poll was the Gallup poll, named for George Gallup. Gallup had introduced the quota sampling method of polling over a decade before the 1948 election, and it had been successful in the past. However, when Truman was elected president, it became clear that there were flaws in this method that was being used by the polls to choose their samples.
Quota sampling is a method in which the sample is chosen using quotas that are meant to be proportional to the electorate, with categories to represent the variety of different groups that would be voting in the election. The sample size of Gallup’s presidential election poll was around 3250, and the people who were chosen were interviewed to try to reduce bias. The Gallup poll predicted that Dewey would win the election with 50% of the vote, and other polls’ predictions did not differ very much from Gallup’s poll.
Despite efforts to minimize bias, the quota sampling method was unsuccessful at providing an accurate representation of voters and therefore an accurate prediction of the outcome of the 1948 presidential election. In addition, Gallup stopped polling in October, even though the election would not take place until November, further evidence of how confident most Americans were that Dewey would emerge victorious. This was a mistake, though, and even George Gallup Jr., the Gallup organization’s co-chairman, later admitted so, commenting, “We stopped polling a few weeks too soon.” He also added, “We had been lulled into thinking that nothing much changes in the last few weeks of the campaign,” a statement to which other competitors who had also conducted inaccurate polls and other people who had confidently spread in the media that Dewey would surely win the election could likely sympathize.
The general feeling throughout the country was that Dewey would certainly win the election, and the newspapers had to get printed quickly so they could share the election results right away. So, the Chicago Daily Tribune did not know when the issue went to press before the results were definite that the newspaper they were printing would come to be a publisher’s worst fear. A photograph of Truman holding up the newspaper, smiling, would make the headline extremely famous for years to come and serve as a reminder of the 1948 presidential election upset and the shock it brought many people because of a polling system that was not as accurate as many had thought.
I really liked how you specifically pointed towards the quota polling failures in the 1948 elections. Ultimately, no matter how many responses you get for a poll, it doesn't matter if you can't poll until election day, much less until if you don't understand who you're polling. This reminds me a lot about the election from 2016. Now, we understand that polls will inevitably have polling errors, and with the exception of states like Michigan and Wisconsin, most states polled accurately. These two cases especially point out the need for accurate polling while still understanding that polls can have margins for error.
ReplyDeleteSource: http://fivethirtyeight.com/features/the-real-story-of-2016/