Reason for Using Frequency and Percentage to Analyse Categorical Data

Abstract

This article examines why frequency and percentage remain the most suitable descriptive statistical tools for analysing categorical data, irrespective of the number of response categories or the degree of measurement involved. Categorical data, whether nominal or ordinal, do not carry the numerical properties needed for statistics such as the mean or the standard deviation, which makes frequency counts and their percentage equivalents the most meaningful way to summarise how respondents are distributed across categories. Drawing on the theory of levels of measurement and on established practice in descriptive survey research, the article argues that frequency and percentage allow a researcher to present data honestly, to compare groups of unequal size, and to communicate findings to readers who may not have a statistical background. The article also considers how a benchmark value can be applied once percentages have been computed, to help a researcher decide whether a response should be treated as significant to the study. The article concludes that frequency and percentage should remain the default choice for any categorical variable, and it recommends that researchers confirm the level of measurement of a variable before choosing a statistical tool to analyse it.