| Summary: | 碩士 === 國立政治大學 === 應用經濟與社會發展英語碩士學位學程(IMES) === 107 === The purpose of the research is to estimate the effect of Taylor Swift’s endorsement of Democratic candidate on people’s voting behavior in the U.S. Midterm election 2018. The research employs difference-in-differences design to estimate causal relationship between her endorsement and election outcome and assesses her endorsement impact. In the research model, geographical variation in Swift’s popularity, generated by search keywords on Google Trends, is used as a treatment, for the reason that the research believes that areas with higher popularity of Swift will experience higher vote share of the Democratic Party. In order to give stronger evidence to the research result, the research conducts falsification exercises, using two different keywords that captures a measure of Swift’s popularity, and other two different keywords that captures popularity of two different celerities, Justin Bieber and Oprah Winfrey. As a conclusion, we find that Swift’s endorsement does not increase the Democratic vote share, however it rather prevents people from voting for the Democratic party’s competing party. Yet the research must conclude there is weak evidence to this interpretation due to the result falsification test of Bieber, which shows the magnitude of his impact not supporting anyone is fairly larger than other estimates.
|
|---|
