This article compares three attribute elicitation procedures commonlyapplied in marketing research—free elicitation (FE), hierarchicaldichotomization (HD), and Kelly's repertory grid—on type ofinformation generated, convergent validity, efficiency in datacollection, and consumers' reaction to the elicitation task. On mostcriteria, RG and HD were not significantly different. The maincontrasts were found between FE on the one hand and RG and HD on theother hand. FE yielded more attributes, a higher proportion of abstractattributes, and a higher level of articulation and was more timeefficient. FE was also evaluated more positively by respondents thanthe other two techniques. Despite these differences, the threeprocedures exhibited a considerable degree of convergent validity interms of the basic categories of concepts uncovered in the elicitationprocedure. Unless the marketing research technique for which theattributes are elicited requires attributes at a low level ofabstraction, the results suggest that FE is to be preferred to HD andRG.