Year: 2015
Type: article
Abstract: The normal distribution is the foundation of many statistical analysis techniques. These so called ‘parametric methods’ use the parameters of the distribution (mean, standard deviation) as part of the calculations. When data is analysed using parametric statistics, certain conditions should be met to apply those statistics correctly. One of these conditions is that data are normally distributed, and some suggest that this should be determined early in any analysis [1, 2]. But others suggest it is unnecessary [3], as normality is not an important assumption [4] and many parametric tests are ‘robust’ and can deal with non-normal data distributions [3]. Yet, readers of research papers seek assurances that the data analysis is appropriate [5]. In spite of authors discussing their need, Thode [6] described approximately 400 methods to test for normality. So, many options are available to researchers which range from informal plotting through to formal hypothesis testing which tests the null hypothesis of ‘the variable being examined follows a normal distribution’ [1]. So, a P value below a given significance level, suggests departure from normality. Researchers need to be aware of these techniques, so that they can determine their analysis options, and to make sure the data contains no surprises. The purpose of this paper is to outline some of the techniques that can be used.
DOI: 10.3233/ppr-150069
Author: C. Gissane
Institution: St Mary's University Twickenham London
Language: English
Subfield: Statistics and Probability
Field: Mathematics
Domain: Physical Sciences
Open Access status: bronze
OpenAlex ID: W2285247342