TY - JOUR
T1 - Choice certainty in Discrete Choice Experiments
T2 - will eye tracking provide useful measures?
AU - Uggeldahl, Kennet Christian
AU - Jacobsen, Catrine
AU - Lundhede, Thomas
AU - Olsen, Søren Bøye
PY - 2016/9/1
Y1 - 2016/9/1
N2 - In this study, we conduct a Discrete Choice Experiment (DCE) using eye tracking technology to investigate if eye movements during the completion of choice sets reveal information about respondents’ choice certainty. We hypothesise that the number of times that respondents shift their visual attention between the alternatives in a choice set reflects their stated choice certainty. Based on one of the largest samples of eye tracking data in a DCE to date, we find evidence in favor of our hypothesis. We also link eye tracking observations to model-based choice certainty through parameterization of the scale function in a random parameters logit model. We find that choices characterized by more frequent gaze shifting do indeed exhibit a higher degree of error variance, however, this effects is insignificant once response time is controlled for. Overall, findings suggest that eye tracking can provide an observable and exogenous variable indicative of choice certainty, potentially improving the handling of respondent certainty and thus the performance of the choice models in DCEs. However, in our empirical case the benefits of using eye movement data as a proxy for choice certainty in the choice model are small at best, and our results suggest that response time provides a better proxy for stated choice certainty and provides larger improvements in model performance.
AB - In this study, we conduct a Discrete Choice Experiment (DCE) using eye tracking technology to investigate if eye movements during the completion of choice sets reveal information about respondents’ choice certainty. We hypothesise that the number of times that respondents shift their visual attention between the alternatives in a choice set reflects their stated choice certainty. Based on one of the largest samples of eye tracking data in a DCE to date, we find evidence in favor of our hypothesis. We also link eye tracking observations to model-based choice certainty through parameterization of the scale function in a random parameters logit model. We find that choices characterized by more frequent gaze shifting do indeed exhibit a higher degree of error variance, however, this effects is insignificant once response time is controlled for. Overall, findings suggest that eye tracking can provide an observable and exogenous variable indicative of choice certainty, potentially improving the handling of respondent certainty and thus the performance of the choice models in DCEs. However, in our empirical case the benefits of using eye movement data as a proxy for choice certainty in the choice model are small at best, and our results suggest that response time provides a better proxy for stated choice certainty and provides larger improvements in model performance.
KW - Eye tracking
KW - Choice certainty
KW - Scale heterogeneity
KW - Response time
KW - Pictorial presentation
U2 - 10.1016/j.jocm.2016.09.002
DO - 10.1016/j.jocm.2016.09.002
M3 - Journal article
SN - 1755-5345
VL - 20
SP - 35
EP - 48
JO - Journal of Choice Modelling
JF - Journal of Choice Modelling
ER -