Policy Expectations and Institutional Arrangements in the American States

现有的选民行为研究是基于这样的前提，即选民是否以问题为导向。她应该投票给个人立场或竞选纲领最接近自己的候选人。在一个民主制度下，一个单独的民选官员制定所有的政策。这样一种议题表决的观点可能是有道理的。但在一个有多个决策者的民主制度中--比如行政机关和立法机关以及三权分立--选民想象一个候选人一旦上台就会实现她所说的一切是没有意义的，如果选民像唐斯（1957）所说的那样以问题为导向。他们的目标应该是选出那些可能制定选民所喜欢的公共政策的官员。议题投票的基础是选民对政府在每位候选人当选后将采取的政策的期望。在总统和州长选举中，由于行政机关相对于立法机关的机构权力和分裂的政府，与每个候选人相关的这些政策期望可能与候选人的个人立场不同。此外，选民在决定支持哪位候选人时，在多大程度上使用政策预期或候选人立场，也可能受到分裂政府的前景和选民对行政分支权力的看法的影响。在许多情况下，个人立场与选民立场最接近的候选人并不是能够提出与选民偏好最接近的政策的候选人。PI在2001年将举行州长选举的两个州（新泽西州和弗吉尼亚州）对1000名登记选民进行了电话调查。调查选举将在选举前两周举行将向受访者提出一系列问题，要求他们将自己的立场放在一个问题量表上。州长候选人的立场州立法机构的立场。以及州政府在每位州长候选人领导下的地位。调查还将询问受访者他们认为哪个政党将在州议会中占多数，他们认为州长与立法机构相比有多强大，以及他们打算在州州长选举中投票给哪个候选人。调查还将收集受访者的政党认同，社会经济特征，政治信息水平，对政治的关注以及其他相关数据。然后，PI将衡量政策预期与候选人立场对选民选择候选人的影响。在这两个州，他们预计会发现，一旦考虑到选民的政策预期，选民比研究人员以前发现的更注重问题。此外，候选人的立场对选民的投票决定几乎没有影响，除非行政部门可能对政策制定有很大的控制权，或者不太可能有任何控制权。在后一种情况下，选民将投一个表达的候选人的个人立场，他们最喜欢的候选人充分知道，候选人将不会有什么影响力的政策。本项目的结果将有助于研究投票行为和州政府。对于文学矿石投票行为。PI将提供第一个真正的问题投票测试的基础上，选民的偏好是定义在预期的政策结果，而不是候选人的立场的理论。通过这项研究;研究者希望能提供经验性的证据，以解决有关分裂政府和定向投票模式的争议。对于州政府的研究，PI将提供一个新的数据来源，关于政府行政分支在两个不同州的相对影响力。这些数据也可以与PI以前在俄亥俄州和德克萨斯州政府的工作相结合。调查数据将提供给其他研究人员和公众。这项研究有望大大增进我们对这一主题的理解。

Existing studies of voter behavior are based on the premise that if a voter is issue-oriented. she should vote for the candidate whose personal position or campaign platform is closest to her own preferences. In a democratic system in which a lone elected official makes all policies. such a view of issue voting may make sense. But in a democratic system with multiple policymakers - such as an executive and a legislature and a separation of powers, it makes little sense for voters to imagine that a candidate will accomplish everything she says she will once in office.If voters are issue-oriented, as Downs (1957) argued. their goal should be to elect officials who will likely produce public policies that the voter prefers. The foundations of issue-voting are the expectations voters hold about the policies the government will adopt given the election of each of the candidates. In presidential and gubernatorial elections, these policy expectations associated with each candidate may differ from the candidate's personal position due to the institutional power of the executive relative to the legislature and to divided government. Furthermore, the extent to which voters use policy expectations or candidate position when deciding which candidate to support may also be influenced by the prospects of divided government and the voters perception of the power of the executive branch. In many cases the candidate whose personal position is closest to the voter's own will not be the candidate who would produce policies closest to the voter's preferences.The PI's conduct a telephone survey of 1000 registered voters in each of two states (New- Jersey- and Virginia) that, will have gubernatorial elections in 2001. The survey. which will take place the two weeks preceding the election. will present respondents with a series of questions asking them to place on an issue scale their own position. the positions of the gubernatorial candidates. the position of the state legislature. and the position of the state government under each gubernatorial candidate. Respondents will also be asked which party they believe will have a majority in the state legislature, how powerful they believe the governor is compared to the legislature, and for which candidate they intend to vote in their state's gubernatorial election.The survey will also collect data on respondents' party identification, socioeconomic characteristics, level of political information, attention to politics, and other relevant, data. The PI's will then measure the impact of policy expectations versus candidate positions on voters' choice of candidates. Across both states, they expect to find that once voters' policy- expectations are taken into account, voters are more issue-oriented than researchers have previously found. Also, candidate positions will play little role in the vote decisions made by voters unless the executive is likely to have tremendous control over policymaking or is unlikely to have any control. In the latter case, voters will cast an expressive vote for the candidate whose personal position they most prefer knowing full well that the candidate will have little influence over policy.Results from this project will contribute to research on voting behavior and state government. For the literature ore voting behavior. The PI's will provide the first true test of issue voting based on the theory that voter's preferences are defined over expected policy- outcomes rather than the positions of the candidates. With this research; the investigators' hope to provide empirical evidence relevant to controversies concerning both divided government and directional voting models. For studies of state government, the PI's will provide a new source of data on the relative influence of the executive branch of government across two different states. These data can also be combined with previous work by the PI's on Ohio and Texas state government. Data from the survey will be made available to other researchers and to the general public. The research promises to enhace substantially our understanding of the topic.