NUANCES OF IDENTIFICATION FOR NORMAL DISTRIBUTION

Abstract

Statistical analysis of sample data is an effective tool for researching trends in economic processes and their critical conditions. The techniques in statistical analysis that are widely used in practice are based on the assumption that the sample data being considered follows a normal distribution. In the article the author reveals that the application of the popular K. Pearson criterion of agreement in such problems to confirm normality distributions of sample data can lead to false conclusions, in cases where the original general population is distributed according to the normal law, and the criterion indicates a low probability of implementing the normality hypothesis. The author proposes a numerical procedure for studying the nuances of identifying the normality in sample data; it uses a novel technique that is based on reference statistical series which correspond to samples of a certain size with the given, fixed estimates of the expected value and standard deviation. The author presents a numerical modeling method and the results of studying the characteristics of sample data that affect the errors in the identification of the normality of the sampled populations. The performed numerical experiments allowed us to obtain statistical data for investigating the reliability of the identification of the sampled distributions. The author presented recommendations that can help to avoid errors in identifying normality.