On The Estimation of Power and Sample Size in Test of Independence
No Thumbnail Available
Date
2010
Journal Title
Journal ISSN
Volume Title
Publisher
Asian Network for Scientific Information, Pakistan
Abstract
In this study, power and sample size estimations in the context of test of independent between categorical variables were examined. The required sample size in an experiment is a function of the alternative hypothesis, the size of type I error and the variability of the population. Power of a test is the probability of rejecting the a false null hypothesis and it depends on the effect size, type I error and sample size. A priori power analysis is determination of minimum sample size to obtain a required power while post-hoc power analysis is calculating power of a test. A test with small effect size requires large sample size to achieve a power of 80% or more while effect size of medium or large size needs small sample size to achieve that. Test with small degrees of freedom will attain higher power than the same test with larger degrees of freedom.
Description
Keywords
Effect size, Contingency Table, Power, Sample size, Type I error
Citation
Asian Journal of Mathematics and Statistics.