t-Test, Paired Samples

Before describing a paired samples t-test, it is helpful to offer a brief reminder of the t-test itself. The t-test statistical procedure is used when a researcher wants to know whether two group means are different and whether that difference can be shown to be statistically significant. Statistical significance indicates that the outcome of the statistical procedure is rare enough to confirm that the differences observed would not have occurred solely by chance. For communication research (and many other social sciences), a one in 20 chance is usually considered sufficiently rare. Thus, the alpha (or significance) level for most t-tests is set at .05.

The paired samples t-test is one of three main types of t-tests, the other two being the one-sample t-test and the independent samples t-test. Determining which t-test to use depends on the nature of the data itself and the hypothesis or research question that is posed. In short, what is the nature of the two means one wishes to compare? If the means are derived from matched pairs (or matched subjects) or if they are derived from a repeated measures design (similar to “before and after” scores), then the paired samples t-test is the correct choice.

This entry offers a definition of the paired samples t-test, a detailed explanation of the types of study design that may require the use of this test, the assumptions associated with this statistical procedure, as well as a case example to illustrate an appropriate use of the test.