Analysis of 0 to 10-point response scales using factorial methods: a new perspective

This paper analyses the implications of using survey response scales with 10 or more categories which are often analysed using factorial techniques such as principal components analysis. Although it is the type of metric that determines the choice of this method of analysis, these scales, despite being considered metric, lack some metric properties, such that a score of 10 in a satisfaction survey, for example, does not represent twice the satisfaction of a score of 5. Their ordinal nature also means they cannot be treated as purely categorical. Thus, having both metric and categorical properties, they allow the use of different factor analysis techniques. The second objective is to apply different factorial techniques to a given test item in order to detect potential differences in the analytical information obtained by each method. Finally, we will discuss their relative advantages and disadvantages for use with this type of survey question.