First, we suspect that this method is far more appropriate for circumstances where a strong case can be made that rational decision-making will generally prevail. Because the nature of the task requires subjects to "decompose" their judgments into the component parts, we believe this decomposition requires a greater degree of rationality. More emotional decisions, those involving beauty, physical attractiveness, taste, and the like, may not work as well. Note that we do not yet have data in support of this assumption, but it is a significant concern that we have.

Second, as currently formulated, the underlying attributes must be binary characteristics. While it would certainly be possible to get the mathematics and statistics to work out by using dummy variables to capture the multiple categories for a single attribute, we are concerned with the practical aspects of collecting both the rating and the ranking data under these circumstances, and the effect that these logistical problems would have on the validity of the data. We suspect that using a dummy variable elaboration will be more feasible if the attribute's levels are truly nominal (e.g., black, white, or green) as opposed to ordinal (e.g., hot, warm, or cold).

Despite these limitations one speculative and the other probably a function of the state of development we believe that the approach has substantial value. By greatly simplifying the task of collecting conjoint-related data, we have made it possible to obtain data with greater external validity at a lower cost than traditional methods.