SP RESEARCH: SCALES OR CHOICES: WHICH ARE BEST?
暂无分享,去创建一个
LT Buses commissioned a small study to investigate whether any differences are obtained from stated preference research that use response scales compared to those which collect choice data. In this research, three response scales were investigated: binary choices, 7-point scaled responses and 101-point scaled responses. The 101-point scaled responses were obtained using a pseudo-continuous slider, whereas the binary and 7-point scaled responses were obtained as discrete choice responses. To determine which was best, the research examined respondents' reported preferences, their behavioural choice differences and differences in the model results. The data were analysed using binary logit, logit analysis using a linear transformation to a probability scale and regression analysis using a linear strength of preference function. The study found that respondents clearly preferred the slider presentation. The model results, however, indicated that the models based on the slider data were less efficient than those obtained from the binary and 7-point data. (A) For the covering abstract see ITRD E107985.