Constructive Processes as a Source of Context Effects in Survey Research: Explorations in Self-Generated Validity
Within the last several years, there has been an increasing realization that both bias and accuracy in responses to survey measures of many types (e.g., reports of behavior, expressions of attitude and preferences) are outcomes of fundamental cognitive and affective processes (e.g., Bradburn, Rips, & Shevell, 1987; Isen, 1989; Schwarz, 1990; Tourangeau & Rasinski, 1988).1 This concern with cognition parallels the study of context and framing effects in judgment and decision making (e.g., Hogarth, 1982, 1987; Kahneman, Slovic, & Tversky, 1982) and the general interest in response construction processes within cognition and social cognition (e.g., Barsalou, 1987; Higgins, 1989b; Petty & Cacioppo, 1986b; Srull & Wyer, 1989; T. D. Wilson, Dunn, Kraft, & Lisle, 1989). Research in “contingent decision processes,” as exemplified by J. W. Payne (1982), Feldman and Lindell (1990), and Biehal and Chakravarti (1983), is part of the same theme.
KeywordsContrast Effect Purchase Intention Category Representation Similarity Judgment Constructive Process
Unable to display preview. Download preview PDF.
- 1.The references included here and subsequently are intended to be illustrative rather than exhaustive.Google Scholar
- 2.This presentation will, of necessity, be highly abridged. Readers are urged to consult Feldman and Lynch (1988).Google Scholar
- 3.Measures of memory organization, rating accuracy, etc., were also obtained but are not of immediate relevance. Also not discussed are the effects of delay on recall and recall-judgment correlations. Recall was scored in terms of “favorability of incident recalled,” with a favorable behavior = 1, unfavorable =-1, and neutral =0. These were summed for each subject, producing a “memory favorability” score.Google Scholar
- 4.A complete report is available from the author on request.Google Scholar
- 5.Order of rating/ranking was manipulated but did not affect the results. Other hypotheses considered in the complete article are not discussed here; readers are urged to contact the author for the complete study.Google Scholar