Un indicator informaţional pentru variabilele categoriale ale unei anchete. Notă de cercetare An Informational Indicator for Categorical Variables of a Survey. Research Note Section Studies

Abstract

The statistical procedures used for data processing have to be considerate depending on measurement variables level: nominal, ordinal, interval or ratio type. Here are presented some examples for ordinal variables, where calculated mean is misleading and gives a wrong interpretation. For the categorical variables, we propose an indicator resulted from the informational entropy, as a measure of the responses distribution in a survey. Calculations are made for a Likert scale, survey data from 1991 to 2010. The indicator allows us to make a global characterization of the distributions, items ranking, dynamic comparisons and patterns identification.