A central issue for evaluating the quality of democratic governance is the extent to which the policy choices of public officials reflect the wishes of citizens. Elections are viewed as the most common institution through which citizens affect these policy choices, and four conditions must be met if elections are to facilitate public influence over policy making: Opposing candidates for office must offer voters differing issue positions Voters must perceive the issue positions of candidates Voters must cast their ballots on the basis of these perceived issue positions Winning candidates must vote in accordance with their pre-election issue positions
A fair summarization of the research on this topic would conclude that there is a great deal of support for the first condition within the United States, and that there is at least qualified support for the second and third conditions. Comparatively little attention has been paid to the fourth condition. We know very little about the extent to which the policy decisions of elected officials are related to or governed by their pre-election policy statements as candidates. This investigation answers three questions about the relationship between the pre-election policy positions of legislative candidates and their post-election policy choices, focusing on the U.S. Congress and legislatures in twelve selected states. 1. How often are the pre-election policy positions of candidates consistent with their post-election policy choices? 2. How useful are policy positions for predicting the policy choices of legislators? 3. How consistent is the policy position - policy choice connection, and can we explain any systematic variation in this relationship?
The policy positions of candidates are taken from two sources: the National Political Awareness Test (a pre-election survey of candidate issue positions) and policy statements made on candidate campaign websites. The policy choices made by legislators are taken from roll call votes on bills that are directly related to candidate policy positions taken on the NPAT or candidate websites. For each vote on each bill, then, the researchers can assess whether the vote was consistent or inconsistent with a legislator's pre-election policy position. Using these measures, the first question can be answered by calculating the proportion of the time roll call votes are consistent with policy positions. The second question can be answered by placing the measures of policy positions within well-respected multi-equation, multivariate models of roll call voting behavior. If the measures of policy positions are significant in these models, policy positions provide useful information to voters. Finally, the proposal offers a theoretical framework for explaining the conditions under which legislators will be more likely to make policy choices consistent with their policy positions. The utility of this framework will be evaluated using multi-equation, multivariate models that employ the framework to predict a binary variable identifying roll call votes that are consistent and inconsistent with policy positions. Broader Value: Evaluating the strength of the connection between the policy positions and policy choices of legislators has obvious implications for democratic theory and the quality of democratic governance in the United States. In addition, the connection between policy positions and policy choice is relevant to general issues of trust in government in the United States. A lack of scholarly attention has not deterred citizens from drawing their own conclusions on this topic: many citizens believe that the policy choices of legislators are not guided by their statements as candidates. Moreover, there is substantial evidence that public skepticism regarding candidates' policy statements affects overall levels of trust and cynicism among citizens. If efforts at civic education are to counteract this cynicism, we need to know more about the relationship between the policy positions of candidates and the policy choices they make as legislators.
