If every rater ranks all the s attributes under considcration, with rank I being low and rank s high, the mean rank assigned to an attribute may be a reasonable measure of its importance.However, this first procedure has three major drawbacks:I) asking for only the top few k preferences can leave many attributes with low aggregate rank scores, resulting in crowdingat the bottom of the IPA grid; this effcct will increase with greater numbers of attributes; 2) this procedure omits the reported orders of prefercnce listed by each rater, 3) comparisonwith perfiormance mean values would be difficult in both quadrant and diagonal approaches to IPA.