Stated choice experiments with three-level factorial designs

Abstract

Stated choice experiments have emerged as a powerful evaluative technique for uncovering critical factors within research contexts. In scenarios where resource constraints exist, these experiments offer an invaluable approach to maximizing information acquisition. This paper presents an innovative strategy for crafting effective stated choice experiments through the utilization of a blocked fractional factorial design (BFFD). This methodology is particularly adept at developing symmetric attributes for choice experiments, catering to selection sets of varying sizes 3, 9, and notably, 27.The process begins with the construction of selection sets of sizes 3 and 9, with resolutions spanning three, four, and five. Through a systematic approach, we reveal the alias structures and potential confounding effects inherent in these designs, thus elucidating their implications on estimation outcomes. A meticulous calculation of the A-efficiencies of the proposed designs reaffirms the remarkable efficiency of the 27-selection sets with size 3 designs, outperforming alternative approaches in terms of data precision. To illustrate the practicality of this novel approach, a real-world example is provided, showcasing its superiority over other methodologies. The results not only underscore the enhanced efficiency achieved by BFFD-designed experiments but also emphasize the broader applicability of such an approach in optimizing information-rich choice experiments. As researchers continue to navigate resource limitations, our findings highlight the significance of employing innovative design strategies to extract maximum insights from limited data resources.

Keywords

• Stated choice experiment
• Confounding
• Aliasing