| CONTENTS |
|---|
| Week 1: In this blog post, I explore the impact of a poll’s sponsor on its prediction. Specifically, I’m interested in seeing if the results of unsponsored polls differ from those sponsored by news/media outlets like Reuters or the New York Times. |
Week 2: In this week’s post, I explore economic predictors of a candidate’s success in an election. I focus on whether foreign exchange rates and the strength of the U.S. Dollar can be useful in predicting outcomes.
Week 3: In this week’s blog post, I focus on the role of polls in U.S. presidential elections. I investigate whether close polls correspond to higher voter turnout and whether landslide predictions contribute to lower voter turnout. Then I explore the impications of adjusting for these expected turnout rates in models to achieve more accurate predictions.
Week 4: In this week’s post, I explore the effect of incumbency on candidate success and strategy in presidential campaigns since 1948. In particular, I look to see if certain states vote for incumbent candidates more often than others, regardless of partisanship, and evaluate whether campaigns should take that into account when allocating funds/resources.
Week 6: In this week’s post, I look at why candidates run ads and investigate differences between ad tones in incumbent and challenger candidates, and successful and unsuccessful candidates (based on electoral college outcome).
Week 7: In this week’s post, I explore Bayesian models and propose adjusting polling predictions based on early election returns. This is an exercise I hope to repeat (in the spirit of continuous Bayesian updating) once the data on both recent polls and early voting returns are more complete.