Performance of the traditional pooled variance t-test against the bootstrap procedure of difference between sample means

• Main Authors: Teh, Sin Yin, Md Yusof, Zahayu, Yaacob, Che Rohani, Othman, Abdul Rahman
• Format: Article
• Language: English
• Published: Universiti Putra Malaysia 2010
• Subjects: QA Mathematics
• Online Access: http://repo.uum.edu.my/19401/1/MJMS%204%201%202010%2085-94.pdf; http://repo.uum.edu.my/19401/; http://einspem.upm.edu.my/journal/fullpaper/vol4no1/6.%20Teh%20Sin%20Yin.pdf

In this study, two methods of comparing the means of two samples were conducted.The first method used the traditional pooled variance t-test while the second method used the bootstrap method of comparing means.In the second method, pseudo sampling distributions were generated for the normal and non-normal distributions.For the non-normal we used the 23χ and a g=.5, h=.5 distribution. Group sizes {5, 15} and {15, 25} with equal and unequal variances were generated and for each method Type I error rates were evaluated.We found that about 20% and 33% of the study conditions using the pooled variance t-test were robust when group sizes are {5, 15} and {15, 25}, respectively.Whereas, about 33% and 47% of the study conditions in bootstrapped procedure were robust when group sizes are {5, 15} and {15, 25}, respectively.