BEHAVIORAL GENETICS OR FLAWED STATISTICS – MEASURING THE ROBUSTNESS OF ACE MODELS IN CONTEXT DEPENDENT CTD STUDIES

Open Access
Author:
Lordan, Daniel J
Area of Honors:
Political Science
Degree:
Bachelor of Arts
Document Type:
Thesis
Thesis Supervisors:
  • Peter Hatemi, Thesis Supervisor
  • Michael Berkman, Honors Advisor
Keywords:
  • Behavioral genetics
  • likert-type
  • methodology
  • classical twin design
  • ace model
  • genes and environment
Abstract:
Behavioral genetic analyses have provided evidence that a majority of human traits and behaviors are at least partially the result of genetic influences. This research has recently crossed into the study of political orientations. The vast majority of political behavior research has relied on two forms of measurement, Likert-type ideology questions and a Wilson Patterson Inventory of social and political attitudes. This is important for two reasons. First, extant research has shown heritability results may differ by type of measurement. Due to the only recent introduction of behavioral genetics to political science, it remains unknown how sensitive the heritability of political orientations is to measurement. This leaves open the questions of the robustness and reliability of the genetic influences on political orientations and whether other types of survey instruments would result in different estimates of genetic and environmental influences. Of equal importance, reliance on only these two types of measures provides limited coverage of the context-dependent nature of political thinking. That is, asking questions about single issues in a vacuum may not accurately portray the state-dependent nature of how humans make decisions or the relative importance of one issue to another. Here I seek to address this lacuna by testing different types of survey measures on social and political attitudes and comparing the estimates of the variation that they attribute to genetic and environmental influences. The findings both validate previous studies but also raise new questions. There was just as much variation observed within question types as there were across the content specific nature of the questions. This supports the idea that the wording of individual questions matters as much as the type of question asked and opens the door for future testing using varied survey methods.