Test statistics for the comparison of means for two samples that include both paired and independent observations

Derrick, Ben; Russ, Bethan; Toher, Deirdre; White, Paul

Abstract

©2017 JMASM, Inc. Standard approaches for analyzing the difference in two means, where partially overlapping samples are present, are less than desirable. Here are introduced two test statistics, making reference to the t-distribution. It is shown that these test statistics are Type I error robust, and more powerful than standard tests.

• Journal Article Type: Article
• Acceptance Date: Dec 22, 2016
• Online Publication Date: Jun 2, 2017
• Publication Date: Jun 2, 2017
• Deposit Date: Jan 25, 2017
• Publicly Available Date: Jun 3, 2018
• Journal: Journal of Modern Applied Statistical Methods
• Print ISSN: 1538-9472
• Peer Reviewed: Peer Reviewed
• Volume: 16
• Issue: 1
• Pages: 137-157
• DOI: https://doi.org/10.22237/jmasm/1493597280
• Keywords: statistics, t-test, partially overlapping
• Public URL: https://uwe-repository.worktribe.com/output/886845
• Publisher URL: http://dx.doi.org/10.22237/jmasm/1493597280
• Additional Information: Additional Information : Material in this article is copyrighted and initially published by JMASM," Derrick, B., Russ, B., Toher, D. and White, P. (2017) Test statistics for the comparison of means for two samples which include both paired observations and independent observations. Journal of Modern Applied Statistical Methods, 16 (1). pp. 137-157. ISSN 1538-9472 [In Press] Available from: http://eprints.uwe.ac.uk/30815
• Contract Date: Jan 25, 2017