How Should Very Large Likert Datasets Be Analyzed?

Abstract

Very large Likert databases, consisting of millions of cells, pose a special analytical problem because of the processing limitations of most desk top computers running Microsoft Excel. This paper shows that while there is validity to both the ordinal and the interval approaches to Likert analysis, the Marascuilo Procedure, an ordinal method based upon summations, proportions and Chi Square is superior in its ability to detect statistically significant intergrouprelationships. An illustration of its usage is given. It is the procedure of choice which operates well within desktop and more powerful computer systems. Commercially available add-ins are available.