Improving the question formulation in Delphi‐like surveys: Analysis of the effects of abstract language and amount of information on response behavior

Prospective surveys are a common research mode used to establish a profound decision base for R&D investments, technological innovation, and policy making. The challenge faced by participants of such surveys, particularly if conducted via impersonal survey modes (e.g., mail or online surveys), is the limited ability to clarify contextual issues. Related research in socio-psychology refers to language vagueness as abstractness, cautioning that abstract language and the way people interpret information are related. We apply Construal Level Theory and the Linguistic Category Model from linguistic sciences to the context of prospective surveys and examine the effects of abstract language and its related amount of information on panelists' response behavior. Based upon 32,000 observations, our research reveals that higher abstractness and greater amount of information reduce the clarity of the respondents' assessments of survey questions. In particular, abstract language and lengthy questions lead to less extreme assessments and a reduced response variance of panelists.