A Trade War With or Without Trump: Actual Topical Knowledge as a Moderator of Question Wording Effect on Survey Responses

Abstract

It is a platitude of communication and public opinion research that responses to survey questions to a great extent depend on the words used in those questions. This idea, however, was not always well supported in empirical studies. We argue that the inconsistent findings from prior research might stem from the fact that different groups of individuals have varying sensitivity to the influences of question wording variations. With an online survey experiment testing participants’ attitudes toward the foreign trade disputes under the Trump administration, we found that the impact of changes in question wording on inducing different responses was moderated by participants’ actual topical knowledge. Referring to the issue as “Trump’s trade war” significantly reduced its favorability compared to describing it as “the trade war” without mentioning Trump, but only among participants who were unknowledgeable about tariffs and international trade issues. The normative implications of attaching political and partisan cues to polling questions were discussed.