Does Answering Survey Questions Create Nonseparable Preferences? A Test of Whether Political Attitudes are Inherent or Constructed

Does answering survey questions change a person’s set of available considerations about an issue? A unique double-blind randomized survey experiment provides an answer to this question. At the beginning of a sample survey, respondents answered an open-ended question asking them to list the issues that came to mind when thinking about one of eight randomly assigned issues. Respondents answered the same question at the end of the survey, but for a different issue from the set. Between the open-ended questions, respondents answered closed-ended questions about thirteen political issues, including some questions that explicitly linked issues. Responses to the open-ended question at the beginning of the survey are no different from responses at the end of the survey. The results cast doubt on theories that political attitudes are constructed at the time of a survey and support theories that political attitudes exist prior to the survey.