The Role of Normality on One Sample and Two sample T-Tests

Abstract

The present paper is an attempt to bring to the notice of the researchers, the role of normality inusing a t-test and show them empirically the effect of violation of normality on the inferences drawn on population mean, based on a t-test. The main discussion is on power of the test and how power decreases due to violation of normality. The data considered has been generated in accordance with the objectives and discussion has been provided accordingly. One can use this paper for a sample data which falls under similar situations. We also discuss the advantages of trimming in non-normal situations and how this can improve the power of the t-test. This paper is mainly written to caution the researchers on using a t-test under non-normal environments and introduce trimming in t-test.