Workshop Three, Session Two
Best-Worst Scaling and its Application to Conjoint Measurement
Presenters: Hume Winzar with Chris Baumann and Steven D'Alessandro
Best-Worst overcomes the shortcomings of ratings scales (Likert-type scales). Ratings scales are "censored" data and assume particular parametric distributions. For example, what happens when we want to rate an object more than 5 (on a 5-point scale) and we assume interval-scale properties? Much cross-cultural research is wrong: Recent research has shown that different cohorts, especially different cultural groups, interpret and respond to rating scales differently, making cross-cultural comparisons biased. Finally, while popular amongst researchers, ratings scales are an unnatural form of evaluation and response for most respondents: in our everyday lives we rarely give objects a rating out of ten, for example. Instead, we usually just decide which one is preferred. Best-Worst scaling overcomes these limitations of rating scales and offers a means of drawing ratio-scaled measures. The drawback of the B-W method is more time requirements for respondents and more data processing for researchers.
In this workshop we demonstrate the principles of the Best-Worst method and provide examples of its use in construct evaluation. We walk through a problem in marketing, with implications for management, law, and public policy: the evaluation by potential franchisees of different components of the federally mandated franchise disclosure document (Wright and Winzar, 2010). We also show how the method can be applied to other areas where evaluations, trade-offs, and choice, may be required. We offer practical demonstrations of Best-Worst scaling applied to conjoint measurement with airline brand choice and preference for students' assessment mode.
Wright, Owen and Hume Winzar (2010), "Modelling Properties of the Franchising Disclosure Document," Proceedings of the 24th Annual International Society of Franchising Conference, University of New South Wales, June 7-9. (Forthcoming)