Statistical Methods for Multidimensional Scaling and Non- Linear MIMIC Models in Political Science

点击翻译按钮获取中文摘要

Dimensional thinking is a mainstay of political discussion. Journalists identify political candidates as on the left or on the right of the ideological spectrum. Voters describe politicians as close to them or far away from them. Politicians take sides, defend positions, move towards some groups, and move away from others. As political events unfold, politicians, voters, and issues are often described by using these kinds of geometric and spatial terms. Not surprisingly, psychologists have found that most people, not just sophisticated political analysts, often conceptualize and describe the social and political world in spatial and geometric terms. This provides a key to understanding voting behavior, public opinion, and many other forms of political behavior, but to be truly useful, we must have a better understanding of what people mean when they say that one politician is close to another or that they feel closer to one issue position than another. This project aims to develop a better understanding of how people think about politics by developing statistical methods for uncovering the dimensional structure of people's choices, opinions, and judgments. For example, the project will develop statistical methods and computer programs for relating peoples' rankings of political candidates to their underlying opinions and beliefs about the characteristics of these candidates. Preliminary research has shown that peoples' choices can be explained partly by a remarkably small number of underlying dimensions. The methods to be developed in this research will enable a clearer interpretation of why this small number of dimensions explains so much. These methods should make it possible to get a clearer understanding of the major concerns which motivate voters and other people who make political choices. In addition, since similar methods have been widely applied to understanding economic choices, it seems likely that the methods developed in this project will be useful in areas outside political science.