Think of assumptions as prerequisites. These conditions must be met in order for the t-test to be used. Before running the t-test, you would want to test the appropriate assumptions. Most of the t-tests have the same basic assumptions, but there are some slight differences based on the type of t-test being conducted.
General Assumptions
- The dependent variable is continuous (interval or ratio).
- review the ASC guide on Levels of Measurement
- attend the What's a Variable group session
- Each observation is independent of other observations.
- evaluate the study design - Does any person's score depend on someone else's?
*this important assumption is often not an issue*
- No significant outliers.
- Sage Publications resource
- The dependent variable should be approximately normally distributed.
- Laerd Statistics guide
- for the Independent Samples T-test: each group should be assessed individually
- for the Dependent Samples T-test: the differences should be assessed for normality
Test-Specific Assumptions
- Independent groups (Independent Samples T-test)
- categorical independent variable with exactly two levels (i.e. yes/no, male/female)
- participants only belong to one group or level of the IV
- Paired or related groups (Dependent Samples T-test)
- categorical independent variable with exactly two levels (i.e. pre-test/post-test)
- the same participants provide scores for both groups or levels of the IV
- Homogeneity of Variances (Independent and Dependent Samples T-tests)
- are the variances of the two groups approximately the same?
- use Levene's Test by checking the appropriate box in SPSS when running the test
