Can response latencies be used to detect survey satisficing on cognitively demanding questions?

According to Krosnick's influential account, survey satisficing occurs when a respondent decides to use a lower level of cognitive effort in order to provide a satisfactory but less accurate answer than would have been produced if a greater amount of effort had been expended on the task. Satisficing theory has rapidly become the dominant framework in survey methodology for assessing response quality, with an increasing number of studies seeking to understand the causes and consequences of the decision to take cognitive shortcuts. However, the utility of commonly used empirical indicators of satisficing for assessing the accuracy and completeness of response data is open to question, because their prevalence is likely to be related to a range of factors, in addition to a respondent' s decision to shortcut. In this paper, we use response latencies to assess whether Don' t Know and rounded responses take, on average, less time for respondents to produce, as should be expected if these response styles reflect respondent decisions to minimize cognitive costs. Our analyses reveal, however, that both types of response were associated with significantly longer response latencies. This suggests that caution must be exercised in the selection of empirical indicators of survey satisficing behavior.