Previous authors adopted a ranking approach in their IPAresearch (Matzler et al., 2003; Mersha and Adlakha, 1992;Sampson and Showalter, 1999), but they ignored the wholenumber of attributes that compose the evaluation task, and/orthe rankings given by raters, using as their measure ofimportance the absolute frequency for each attribute reportedby all raters or the mean rank of the ordered k attributes.