The paper argues that a blind reliance on the use and results of predictive models led to the major pollsters never knowing that Hillary Rodham Clinton would lose the 2016 Presidental election. The reason is that history is not necessarily a good predictor of the future and that predictive models employed by the pollsters assumed the existence of a steady-state which was not present at the time. The paper introduces predictive models by providing a short history, followed by explaining how predictive models work. The paperlists the fundamental assumptions of predictive models, highlighting the advantages and disadvantages. The paper then analyzes in some detail why various predictive models incorrectly predicted that Hillary Clinton would win the 2016 presidential election, when in fact Donald Trump was elected President of the United States. The thesis is that when a predictive model experiences an exogenous shock or a superseding intervening cause, the dependence on data before the shock occurred is unwarranted. In fact, it is argued that all data before the exogenous shock or superseding intervening cause should have beenignored, and only the data that appeared after the shock should have been used in making predictions. The paper concludes that the predictive models used by the pollsters during the 2016 Presidential election were incapable of recognizing an exogenous shock or a superseding intervening cause and that human intervention was needed to correct for the limitations of the predictive models that were employed.
Buresh, D. L., & Pavone, T. (2019). Why No One Knew that Hillary Clinton Would Lose the 2016 Election. Retrieved from https://touroscholar.touro.edu/tuw_pubs/15
Article published here: https://www.ivyunion.org/index.php/ajpsr/article/view/1248/451
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 International License.