What is it about?
Despite the academic value of factor analysis (FA) on Likert scale data, its statistical legality has been relentlessly questioned by professional statisticians. This article reviews relevant literature and proposes a statistically appropriate method of getting most out of FA on Likert scale.
Featured Image
Photo by Ag PIC on Unsplash
Why is it important?
The review shows that a larger sample size is an important consideration in improving the appropriateness of FA on LS as it improves the solution in terms of normality, communalities and loadings. Further, a 7-9 point or greater scale to account for normality, an alpha level of 0.01 or 0.005, polychoric correlation instead of Pearson’s are reported to improve the statistical appropriateness of the test of Likert scale. Moreover, using non-parametric alternatives like CATPCA to testify the results of FA greatly increased the overall value and validity of the test.
Perspectives
Likert scale was first introduced by Likert (1932) as a measure of attitude or opinion on an odd-numbered response set with options including ‘strongly approve’, ‘somewhat approve’, ‘no idea’, ‘somewhat disapprove’ and ‘strongly disapprove’. The scale was later used in many diverse variations in academic studies and business research including measurements of happiness, intelligence, completeness, excellence, dullness, superiority, priority, importance and so on (Clason &Dormody, 1994; Vogt, 1999).
Dr. Sana-ur-Rehman .
NFC- Institute of Engineering and Technology, Multan, Pakistan
Read the Original
This page is a summary of: Manualizing Factor Analysis of Likert Scale Data, Journal of Management Sciences, October 2020, Geist Science, Iqra University,
DOI: 10.20547/jms.2014.2007204.
You can read the full text:
Contributors
The following have contributed to this page
