- When σ is known, you should use the z-test. When σ is unknown (as is the case in many empirical studies), you should use the t-test.
- A good rule of thumb to interpret effect sizes with Cohen's d:
small: < .20
medium: .20 - .80
large: > .80
- The t-test is robust, which means that it is hardly affected by (small) deviations from the underlying assumptions of normality of the sampling distribution and homogeneity of variance.