A Psycholinguistic Look at Survey Question Design and Response Quality

It is universally acknowledged that the wording of a survey question can have a strong influence on the answers that respondents provide. For example, many studies have shown that vague and ambiguous terms are often interpreted idiosyncratically by respondents, and thus can increase measurement error. In addition to ambiguity, the cognitive effort required to comprehend survey questions may affect data quality in a similar way. This aspect of survey question design has received comparatively little attention to date and has rarely been examined experimentally. The present thesis suggests that applying a psycholinguistic perspective to survey question design may shed some light on the relationship between the cognitive effort required to comprehend survey questions and the quality of respondents’ answers. Theoretical and empirical evidence from psycholinguistics indicates that text (or question) comprehensibility is reduced by a set of specific text features: low-frequency words, vague or imprecise relative terms, vague or ambiguous noun phrases, complex syntax, complex logical structures, low syntactic redundancy, and bridging inferences. Three experimental studies were conducted to examine whether these seven text features indeed undermine question comprehensibility and, in turn, how question comprehensibility affects the quality of respondents’ answers. Study 1 revealed that six of the seven text features reduce question comprehensibility as indicated by significantly longer response times. Moreover, the text features were found to reduce response quality by producing more neutral (i.e., midpoint) answers. For the most part, these findings were supported by study 2, in which eye-tracking parameters were used as more direct measures of cognitive effort: respondents fixated longer on questions containing one of these text features and required more fixations to process, re-read and interpret these questions in comparison to control questions that did not include the text features. Finally, study 3 showed that respondents receiving less comprehensible questions provided lower-quality responses (as indicated by number of non-substantive responses, number of neutral responses, and over-time consistency of responses) than respondents receiving control questions that were easier to comprehend. Moreover, interaction effects of question comprehensibility with respondents’ verbal skills and their motivation to answer surveys were found. Taken together, these findings indicate that response quality is reduced if questions are difficult to comprehend and exceed the processing effort that respondents are willing or able to invest during survey responding. Hence, survey designers should try to minimize the cognitive effort required to comprehend a question by avoiding the problematic text features discussed above