Test for the equality of partial correlation coefficients for several populations. Issue 6 (1st October 2018)

• Record Type: Journal Article
• Authors: Gupta, Arjun K.; Bhandary, Madhusudan; Aziz, Mohammad
• Is Part Of: Journal of statistics & management systems. Volume 21:Issue 6(2018)
• Page Start: 985
• Page End: 1001
• Publication Date: 2018-10-01
• Subjects: 62H15 -- 62H20; Likelihood ratio test -- Partial correlation coefficients -- Power of the test; Statistics -- Periodicals; Mathematical models -- Periodicals; Mathematical models; Statistics; Periodicals; 519.5
• DOI: 10.1080/09720510.2018.1478624
• Languages: English
• ISSNs: 0972-0510

Abstract

Abstract: We develop an approximate likelihood ratio test in this article and this test can be used to check if there is equality of several partial correlation coefficients or not using several independent multinormal samples. We also discuss about large sample χ 2 -test for the same problem. We compare the two tests based on their power and it is found that the approximate likelihood ratio (ALR) test consistently has more power than χ 2 -test. The ALR test has size slightly more than alpha level. We strongly recommend the ALR test for data analysis.