Identification of unacceptable attribute levels for preference measurement: An empirical comparison of different methods

Eliminating unacceptable attribute levels when measuring consumer preferences represents an important problem, both academically and managerially, since including unacceptable attribute levels in preference measurement may cause distorted parameter estimates and result in inaccurate estimation of market shares, for example. This research tests different direct and stimulus evaluation task methods empirically in a hypothetical and a realistic situation using strong and weak wording as well as an isolated and a non-isolated presentation to determine their ability to eliminate unacceptable attribute levels. For all analysed research objects, in an isolated versus non-isolated presentation of attribute levels, participants consider an attribute level unacceptable more frequently when they must evaluate it directly compared with when they assess a stimulus and reject unacceptable attribute levels presented in isolation rather than non-isolation. Furthermore, respondents reject a particular attribute level more often if the description of the decision context employs weak rather than strong wording.