Saul Dobney, Carlos Ochoa and Melanie Revilla

The main goal of this research is to study the impact on the answers and data quality of making conjoint questions more realistic by introducing some randomised noise into the descriptions of the conjoint levels or by simulating the way an e-commerce website displays products. Conjoint analysis is an advanced market research technique commonly used to estimate preference share for products and services with different attributes and levels. A common criticism of it is in regard to the repetitive nature of the questions. In order to study this, an experiment was implemented in Spain using 1,600 respondents from the opt-in online panel Netquest. The respondents were randomly assigned to one of the following four conditions: classic conjoint design without noise (control group); classic conjoint design with some random textual and numerical noise added to the attribute level descriptions; conjoint simulating e-commerce display of products but no noise; and conjoint simulating e-commerce display and some random textual and numerical noise. The four groups were compared in terms of data quality, survey evaluation and substantive results. The results show a directional but not statistically significant improvement of quality of estimations. In terms of survey evaluation, even if the improvements are not systematic, there is a clear tendency for an improved evaluation when an e-commerce layout is used, but not when random noise is used. Substantive results are not affected.

Content for MRS Members only. Sign in or become a member today.

Get the latest MRS news

Our newsletters cover the latest MRS events, policy updates and research news.