Different Approaches using the Normal and the Exponential Distribution in the Evaluation of the Customer Satisfaction

Keywords

customer satisfaction, normal distribution, exponential distribution, multilevel models

Abstract

The Customer Satisfaction is generally evaluated using the data collected with questionnaires. The data are organized on an ordinal scale and, for this reason, it is convenient to transform them in pseudo-interval support. The psychometric methods used for this transformation generally hypothesize that the latent variable has a normal distribution. Sometimes, particularly when the frequencies are concentrated on the left extreme or on the right extreme of the distribution, this assumption brings to preposterous results. In these cases the use of other types of distribution, as, for example the exponential distribution, is preferable. In this paper we show how the results of a survey can change using the normal distribution, the exponential distribution or the two distributions alternatively. We use, in fact, the results coming from the different transformations, to apply a multilevel model.

(top)

(back to issue)