Pakistan Science Abstracts
Article details & metrics
No Detail Found!!
Testing difference between dependent correlations when there is a common variable: A Monte Carlo simulation study.
Author(s):
1. Mehmet Mendes: Biometry and Genetics Department, Canakkale Onsekiz Mart University, Turkey
2. Handan Camdeviren: Biostatistics Department, School of Medicine, Mersin University, Turkey
Abstract:
A simulation study was conducted to compare Williams’ modified (1970) t test, Dunn and Clark’s (1969) z test, and Steiger’s original modification of Dunn and Clark’s z test in terms of Type I error rates and test power. Simulation results indicated that both Type I error rates and test power depend on sample size, magnitudes of the dependent-independent correlations and effect size. For the sample sizes of 10 and 20 and for dependent-independent correlations of 0.1, all of the empirical alphas for the three tests were under 5%. The dependent-independent correlations were 0.7 and sample sizes were 10, Type I error rates were very high (13.6%) for WM test while the other tests had close values to the predetermined alpha level. When the sample size increased to 50, Type I error rates increased in all tests.
Page(s): 251-256
DOI: DOI not available
Published: Journal: Pakistan Journal of Statistics, Volume: 22, Issue: 3, Year: 2006
Keywords:
Keywords are not available for this article.
References:
References are not available for this document.
Citations
Citations are not available for this document.
0

Citations

0

Downloads

5

Views