This is the second of a series of six articles describing how to choose the most appropriate statistical method to test for relationships between two variables. When the response variable is continuous and the explanatory variable is dichotomous, the t-test is often used. The article indicates when this is really appropriate and when the test should be modified or replaced by another.