Question-order effects in market research: An information processing approach
This research examines the effect of answering a specific attribute rating question on a later overall brand evaluation in a marketing survey. The mechanisms which underlie such effects for respondents varying in product category knowledge are described using an information processing perspective. Two studies were conducted, including a laboratory experiment about an unfamiliar bicycle brand, and a telephone survey about real brands of running shoes. Results from both experiments suggest that low knowledge respondents are unable or not motivated to retrieve attribute information from memory to compute an overall brand evaluation. Therefore, low knowledge respondents are likely to rely on attribute information made accessible by prior questions in making a brand evaluation, resulting in carryover effects. The impact of earlier attribute questions on overall brand evaluations is more complex for moderate and high knowledge respondents, who showed both carryover and backfire effects. The results of Experiment 1 suggest that moderate and high knowledge subjects discount accessible information with positive implications for the brand, resulting in a backfire effect, but use accessible information with negative implications, resulting in a carryover effect. The results of Experiment 2 suggest that prior questions have little effect on high knowledge respondents' evaluations of familiar brands, but do affect their evaluations of unfamiliar brands. When evaluating unfamiliar brands, high knowledge respondents appear to discount attribute information which is low in diagnosticity, resulting in a backfire affect, but use attribute information which is high in diagnosticity, resulting in a carryover effect. This pattern of results can be explained using categorization theory. The findings suggest that brand attitudes may be multi-dimensional or non-existent in many cases. Implications for questionnaire design and attitude measurement are discussed.