Analysis of Two Group Comparisons: A Practical Guide
In health science research, comparing two groups on a continuous outcome variable is one of the most common statistical analyses performed. One must choose between two families of tests: parametric and nonparametric. Parametric methods depend upon more stringent assumptions, and violating these assumptions can result in incorrect conclusions. The dominant two-sample approach appearing in Medical Problems of Performing Artists seems to be the Student t-test, which is a parametric test. Despite being the standard choice, this test is not always appropriate. Common textbook prescriptions regarding its use are often misleading and defective in some respects. This article revisits the basic and methodologically important problems in comparing two groups and seeks to provide a practical guide to researchers.