We spend a great deal of time on the phone and via email coaching customers regarding which conjoint method and software they should use for different research situations. Indeed, this decision can be a source of frustration when people new to conjoint analysis are compelled to act quickly to purchase the right software for an immediate project. We've written white papers on the subject, we've posted recommendations on our Online Forum, and over the years, we've come to the opinion that the choice is largely driven by a few key factors: interviewing mode (i.e. computer, paper), number of attributes, maximum number of levels in the largest attribute, sample size, and whether price is an attribute. Given that conclusion, we investigated whether an interactive tool, posted on our website, could make good, real-time recommendations. To build such a tool, we naturally turned to conjoint analysis.
We started by creating a traditional full-profile conjoint plan (using our CVA software), reflecting different combinations of the five project-related factors. For each research situation (conjoint profile), we rated the likelihood that we'd recommend each of the conjoint methods. We estimated a set of “utilities” associated with the project factors for each of the conjoint methods so that we could predict the likelihood that we'd recommend each conjoint method under all possible project specifications on the five factors. We then evaluated the model, to see if we agreed with its recommendations. The first version was generally good, but not good enough. To improve the recommendations, we made some modest manual adjustments to the utilities and added some interaction effects to account for issues not captured in our original main-effects plan. This second version was definitely better.
Our new website features the resulting interactive project advisor. The interface uses five drop-down menus (based on the five factors outlined above), and produces bar charts showing a “usability” score for each of the main conjoint methods. The higher the value, the more confidence you may have in employing that technique. Some study specifications lead to more than one method receiving a high usability score; while others lead to all methods receiving relatively low recommendations.
Of course, choosing a conjoint method for a project involves other factors than just the five that compose this decision tool. Therefore, a tool such as this cannot guarantee an optimal recommendation in all situations. Even so, we think the benefits to customers just becoming acquainted with the multiple flavors of conjoint methods outweigh the drawbacks. It will point customers in the right direction so that they can more efficiently research and evaluate promising conjoint approaches.
Click the following button to view the interactive tool on our website.