Abstract
It is widely believed that results of research are only valid when the distribution of data is relatively normal. However, most previous importance–performance analysis (IPA) studies have not considered the assessment of data normality. The aim of this study is to evaluate the importance of the normality assumption in IPA studies. Four data sets obtained from four experimental questionnaire surveys were used to demonstrate the changes in outcomes from non-normally distributed data sets to relatively normally distributed data sets by removing abnormal data. The results indicate that the distributions of attributes in four quadrants in importance–performance mapping for non-normally distributed data sets and relatively normally distributed data sets are similar when using direct IPA measurement method; however, they show differences when performing indirect IPA measurement methods. It is concluded that in order to ensure high-quality research outcomes, researchers should justify data normality in their IPA studies.
Disclosure statement
No potential conflict of interest was reported by the authors.