Μη παραμετρικοί έλεγχοι για τη σύγκριση δύο ανεξάρτητων πληθυσμών
Non-parametric tests for comparing two independent populations
View/
Keywords
Μη παραμετρική στατιστική συμπερασματολογία ; P-value ; Μέθοδοι προσομοίωσης Monte-Carlo ; Ισχύς του στατιστικού ελέγχου ; Έλεγχος υποθέσεωνAbstract
Statistical hypothesis testing is one of the fundamental areas of statistical analysis, as it allows decision-making based on sample data. This study focuses on comparing two independent populations using non-parametric methods. Initially, a brief introduction to statistical hypothesis testing is presented, including the formulation of null and alternative hypotheses, the use of test statistics, the selection of appropriate significance levels, the p-value, and the power of a test. Subsequently, the study analyzes the main non-parametric tests for comparing two independent populations. Particular emphasis is placed on categorizing these tests based on possible differences between the two populations, which may be due solely to location parameters, solely to scale parameters, or to both categories of parameters. Next, the effectiveness of well-known non-parametric tests for comparing two independent populations is investigated through Monte Carlo simulation. Popular tests such as the Mann-Whitney test, Mood test, Lepage test, and others are examined and evaluated in terms of their power across different data scenarios. The study includes a comparative evaluation based on simulated data from various distributions, where the corresponding parameters are varied. Finally, the study applies non-parametric tests to real-world data. This application highlights the usefulness of non-parametric methods in cases where the assumptions of parametric tests are not met, offering reliable and interpretable results. Overall, this research provides a comprehensive comparative approach to various statistical methods for comparing two independent populations, emphasizing the advantages and limitations of each method. The study’s findings offer valuable guidelines for selecting the most appropriate statistical method, depending on the nature of the data and the conditions of the analysis problem.
Postgraduate Studies Programme
Εφαρμοσμένη ΣτατιστικήDepartment
Σχολή Χρηματοοικονομικής και Στατιστικής. Τμήμα Στατιστικής και Ασφαλιστικής ΕπιστήμηςNumber of pages
81Language
GreekCollections
Except where otherwise noted, this item's license is described as
Αναφορά Δημιουργού-Μη Εμπορική Χρήση-Όχι Παράγωγα Έργα 3.0 Ελλάδα
Αναφορά Δημιουργού-Μη Εμπορική Χρήση-Όχι Παράγωγα Έργα 3.0 Ελλάδα