This article examines the validity of an objective measure of partisan ambivalence. More generally, it draws attention to the idea that measurement is an active step taken by the researcher and therefore should be subjected to empirical examination. It is also argued that treating a variable at a higher level of measurement than warranted will cause a valid measure to cease to be so, and that the appropriate level of a variable is determined not by the underlying concept or the measurement procedure, but by the researcher. I demonstrate the measure examined can distinguish between various attitudinal states, but only at the nominal level. This is largely driven by individuals who offer no responses to open-ended questions, who are termed indifferent. Furthermore, different coding schemes result in different conclusions. For example, the finding that indifferent individuals are less likely to rely on partisanship than ambivalent individuals when evaluating candidates is obscured when treating the measure as interval. The findings suggest the measure should be coded to account for indifferent individuals, and that even well developed measures of clearly defined concepts need to be subjected to empirical examination.