Issues That Should Not Be Overlooked in the Dominance Versus Ideal Point Controversy

Abstract

Drasgow, Chernyshenko, and Stark (2010) argue that ‘‘responses to questions requiring introspection involve a comparison process,’’ and we agree with this statement. The key question is, however, what do respondents compare themselves to? Drasgow et al. suggest that respondents always compare their self-perception to the statement’s location (an ideal point); therefore, ideal point models should always be used. We feel the answer depends on the construct being measured and on the type of items used to measure it. Sometimes the response mechanism appears to be an ideal point process. But many other times we believe that given a statement and a binary (endorse/not endorse) rating scale, an individual will endorse the item if its utility is larger than a certain threshold and will not endorse it otherwise. In this case respondents compare themselves to a threshold, which calls for the use of a dominance model. To investigate whether a threshold or an ideal point mechanism best describes the response process, intermediate items should