UCLA

Recent research in political science documents the “nationalization” of U.S. state and local politics; the down-ballot results of partisan elections tend to reflect the outcomes of presidential contests. The three papers of this dissertation examine this phenomenon in greater detail. The first paper critically assesses nationalization as a top-down force. I decompose a large set of election results into candidate specific, partisan, and idiosyncratic components. While it is true election results are tied increasingly to partisanship, I find it is not true that presidential elections are most strongly tied to partisanship. In the second paper, I utilize a supervised machine learning technique to determine the extent to which gubernatorial rhetoric mirrors that of presidential candidates, finding candidates largely speak on topics germane to their own jurisdictions. Finally, the third paper uses a survey experiment to find that voters use national policy signals when choosing between candidates for state and local office. I also find voters use state and local policy signals when evaluating national candidates. The stronger the partisan signal, the greater the effect on behavior.