Who knew Trump would win?

Who knew Trump would win?

Donald Trump’s election victory in 2016 will more than likely go down as the greatest political story in the United States if not the world for the decade.  What made it such a great story was the near unanimity of polls and mainstream media outlets that Trump would lose the election. The New York Times last polling data before the election had called Florida, North Carolina, Michigan, Pennsylvania, and Wisconsin all for Hillary Clinton while each of these states ended up going for Donald Trump.  Over the days and weeks after the election, it seemed that one couldn’t swing a cat without seeing an article entitled “why the polls were wrong” or something along those lines. They all went something along the lines of trying to argue that polling methods were flawed or that some version of the Bradley Effect was at play. But there was one article that captured my attention.  The article was entitled “How our company called Trump’s climb.” It was in a Canadian newspaper called The Globe and Mail. The piece was a contribution from a woman named Erin Kelly who was the CEO of a company called Advanced Symbolics Inc. Advanced Symbolics analyzes social media accounts to make polling models for the U.S. election. It does this by analyzing trends in posts of the electorate and then building subsequent electoral models.  In marketing terms, this is called sentiment analysis or tracking social media reactions toward certain products, specifically Donald Trump and Hillary Clinton. The article goes on to analyze how Donald Trump won, by ever so slightly increasing his share of the minority vote. The article cites the example in Michigan of how Donald Trump increased his share of black voters to win the state. He famously told a predominantly African-American audience in Detroit “What do you have to lose by trying something new, like Trump?”  The message took hold. The article then goes on to say that after that day Donald Trump’s support amongst black voters only went up throughout the remainder of the campaign.  The article then says that because they used higher support for Donald Trump from African-Americans in their election prediction models their vote prediction models were almost exact in Michigan.  The article summarizes this by saying that the media’s relentless portrayal that the groups Donald Trump sought to alienate are actually the groups that pushed Donald Trump over the top. In fact, exit poll data from CNN suggests that Donald Trump outperformed Mitt Romney in every racial demographic, except white Americans, and that Hillary Clinton underperformed Barack Obama in every racial demographics, thus seeming to confirm the main thrust of the article that their method of sentiment analysis proved to be correct in the election.  The article also argued that the scores of Republican officials who either refused to endorse Donald Trump or out and out endorsed Hillary Clinton fit right into Donald Trump’s message of distrusting major institutions, especially those that were at one point run by Hillary Clinton.    

2012 U.S. General Election Map. (2012, December 10). Retrieved September 1, 2018, from http://www.cnn.com/election/2012/results/race/president/

Exit polls. (2016, November 23). Retrieved September 1, 2018, from https://edition.cnn.com/election/2016/results/exit-polls/national/president

Kelly, E. (2017, April 06). How our company called Trump's climb. Retrieved September 1, 2018, from https://www.theglobeandmail.com/report-on-business/rob-commentary/how-our-company-called-trumps-win/article32843425/

Latest Election Polls 2016. (2016, June 16). Retrieved from https://www.nytimes.com/interactive/2016/us/elections/polls.html