In the 2016 US presidential election, the Republican candidate Donald Trump surprised many by defeating

the Democratic candidate Hillary Clinton. In particular, even right before the election, polls were predicting

that Hillary Clinton would win the election by a comfortable margin. Why did preelection polls fail to predict

the election outcome? We analyze the polling data, taken from Hufftington post, that include the most recent

polls leading up to the election. The dataset we will be analyzing (polls2016.csv) has 1391 observations,

each representing a different poll, and includes the following variables:

Name Description id Poll ID state U.S. state where poll was fielded clinton The poll's estimated level of support for Hillary Clinton trump The poll's estimated level of support for Donald Trump undecided The poll's estimated percentage of undecided voters days_to_election Number of days before November 4, 2016. electoral votes Number of electoral votes allocated to the state where the poll was fielded (a state-level variable) sample_size The number of people surveyed in the poll We will also analyze a dataset (election2016. cav) which contains the state-by-state voteshare for each candidate collected from CNN. This data set has the following variables: Name Description state U.S. state where poll was fielded clinton The percent of votes Clinton received trump The percent of votes Trump received