New Approaches in Frequency Table Construction for Continuous Symmetrical Data

Abstract

In computing statistical measures from a frequency table, the midpoint represents the observations in each class. The assumption that the observations are distributed uniformly and concentrated around the midpoint is not always valid. In this article, two new methods of constructing a frequency table, using the mean and median of the classes, are compared with the existing method, which uses the midpoint. The simulation study using symmetrical positive data shows that the method which used the classes’ mean performed well compared with the other methods. Freedman and Diaconis (FD) rule is the recommended rule when choosing the number of classes.