A0500
Title: Some new insights into measures of location and adopting voting technique into measures of variability
Authors: Kayode Ayinde - Northwest Missouri State University, Maryville, Missouri, USA (United States) [presenting]
Abstract: Both measures of location and variability in data sets have become increasingly popular and relevant in almost all fields. There arises a question of how agreeable or acceptable each measure of location is to the individual subjects, as this may not be the most representative. Some new insights into measures of locations are considered by incorporating the existing measures into the five and three summary statistics of a data set. The voting technique is also adopted as a measure of variability to address the challenge of acceptance by most subjects. Monte Carlo simulation studies of symmetric and skewed data sets with and without outlier(s) were conducted on the measures, and real-life data sets were applied. Results reveal the consistent minimization of the mean absolute deviation by the median and mean squared deviations by the arithmetic mean and that the challenge of non-existence and non-uniqueness of mode can be overcome with the voting technique as any of the averages can emerge as the most representative average. Furthermore, the harmonic mean of the five summary statistics is identified as the best average, especially with data sets that have outliers in the right direction in both simulation and real-life application studies.
