Predictor Variables of Primary Presidential Partisan Elections In the U.S.
Abstract
General elections have been able to be predicted with relatively high accuracy due to a party split between the final candidates. However, this is not the case in a primary election as each candidate is usually of the same party. The only factors which have been found to relatively correlate with primary results are polling results and endorsements for each candidate. Current models which factor polling and endorsements typically predict primary elections with only 57% accuracy, while current models which only factor polling predict primary elections with about 43% accuracy (Silver, 2016). This research sought to find other variables which may be used to predict primary presidential elections with greater accuracy than these current models. A sample of voting-age participants was surveyed and asked about various demographics, and they were required to vote for a primary candidate from the 2016 presidential election from either the Democratic or Republican Party. Their demographic data was compared to the demographic data of the candidates which they voted for, and a chi-squared goodness-of-fit test was used to determine whether there was a correlation between the demographics of voters and the demographics of the candidates which they vote for. The results showed that voters were most likely to vote for candidates with close proximity in age and of a similar race, while other demographics did not provide statistically significant results.
Predictor Variables of Primary Presidential Partisan Elections In the U.S.
Founders Hall 251 C
General elections have been able to be predicted with relatively high accuracy due to a party split between the final candidates. However, this is not the case in a primary election as each candidate is usually of the same party. The only factors which have been found to relatively correlate with primary results are polling results and endorsements for each candidate. Current models which factor polling and endorsements typically predict primary elections with only 57% accuracy, while current models which only factor polling predict primary elections with about 43% accuracy (Silver, 2016). This research sought to find other variables which may be used to predict primary presidential elections with greater accuracy than these current models. A sample of voting-age participants was surveyed and asked about various demographics, and they were required to vote for a primary candidate from the 2016 presidential election from either the Democratic or Republican Party. Their demographic data was compared to the demographic data of the candidates which they voted for, and a chi-squared goodness-of-fit test was used to determine whether there was a correlation between the demographics of voters and the demographics of the candidates which they vote for. The results showed that voters were most likely to vote for candidates with close proximity in age and of a similar race, while other demographics did not provide statistically significant results.
