Inducing Strategic Bias: and its implications for Choice Modelling design

It has been suggested that the nature of the task within a multi-attribute multi-alternative choice experiment may be sufficiently complex to make it difficult for individuals to develop response strategies to strategically bias their answers. This experiment tested that hypothesis by setting experimental conditions that provide incentives for strategic bias. By changing design parameters one can investigate whether the strategic bias can be reduced. The answer is effectively no: under most circumstances, respondents could find a strategy that achieved significant bias in inferred preferences. The circumstances where this did not occur (involving ranking alternatives, rather than selecting a single preferred alternative) the inferred preferences reflected neither the intended bias, nor their original preferences, making the answers useless to both respondent and researcher.