The Role of Conditioning Information in Reports of Subjective Phenomena

Reports of subjective phenomena are inherently subject to some uncertainty on the part of the respondent. The existence of uncertainty is obvious when the reports are of subjective expectations about prospective outcomes. In other cases, it is not so clear. Sometimes respondents are asked to make hypothetical choices, in which case uncertainty remains unless all relevant elements of the hypothetical scenario are specified. More generally, whenever concepts are loosely defined, be they prospective outcomes, hypothetical scenarios, or other relevant aspects of the question, uncertainty about the appropriate response will remain. This paper considers the interaction between this uncertainty and the text and context of the survey question. Three types of questions are discussed -- questions eliciting unconditional expectations, conditional expectations, and hypothetical choices. In each case, it is difficult to convey to respondents the types of information on which they should and should not condition. When respondents use information in a way that differs from the researcher s assumptions, the researcher may infer that respondents are deficient in some way (e.g., do not provide coherent probability assessments or do not have consistent preference orderings).