| A large body of scholarly work on 'retrospective voting' finds evidence that citizens tend to reward and punish incumbent politicians at the ballot box for changes in policy outcomes. This rational pattern of citizen behavior, which is the subject of a vast body of research, represents the best available empirical defence of representative democracy. By voting retrospectively, citizens both signal their preferences to elites and ensure that elected officials pursue desired policy outcomes. I contribute to this literature by incorporating a pervasive trait of human cognition, the tendency for individuals to respond more strongly to negative stimuli than to positive stimuli, into models of retrospective voting. I offer a thorough and comprehensive consideration of the largely overlooked 'negativity effect' in retrospective voting in U.S. elections. I apply a range of quantitative methods to a diverse set of both aggregate-level observational data and individual-level experimental data to demonstrate empirical support for novel theoretical claims about the conditional nature of the negativity bias. My research finds consistent evidence of a negativity effect in the aggregate-level relationship between economic conditions and support for both presidential and gubernatorial incumbents and further establishes that the severity of the negativity effect in this relationship is related to the quality of prevailing economic conditions. I also employ experimental methods to: replicate the findings from the aggregate-level data, test for a negativity effect in individuals' responses to novel information about policy change, and to demonstrate how the negativity effect in retrospective voting affects policy-making on divisive political issues such as the expansion of minority rights. This project speaks to important questions about the impact of public preferences on elite decision making through a detailed analysis of the characteristics of the signals citizens send to elites. |
