Comparing Political Ideology Scaling Methods: A Study of Topic Polarization in the U.S. House using W-NOMINATE and Correspondence Analysis

Abstract

Growing concern among scholars about increasing polarization in liberal democracies has led to a surge in research focused on the presence and measurements of political polarization in the United States. Due to party polarization varying significantly across political issues, and with the complexity of measurement techniques seeing an escalation, the subsequent skill threshold for studying this phenomenon has been increasing. This research explores the use of Correspondence Analysis as a potentially less complex alternative for measuring the ideology of legislators by answering two research questions. First, this study identifies trends of party polarization across broad policy domains using the standard NOMINATE method on the voting behavior of U.S. House Representatives. Next, it evaluates whether Correspondence Analysis on the positive votes of these legislators can replicate the identified trends. My results demonstrate equal polarization trends across most policy domains, with only one domain deviating significantly. However, we find that the principal dimension of the Correspondence Analysis does not capture these polarization trends effectively. Although Correspondence Analysis does show clustering in its ideological spatial dimensions, these clusters do not align with the expected party identifications. In conclusion, this research indicates that a less complex model based on the positive voting behavior of U.S. House Representatives does not accurately reflect their ideological positions within specific policy domains. The study proposes further investigation into alternative, less complex methodologies, and smaller-scale research topics, as we argue that reducing the research barrier for scholars engaged in this political phenomenon represents a necessary advancement.