Select a normality test. Anderson-Darling is appropriate in most situations.
- Anderson-Darling: This test is usually more effective than the other two tests for the detection of nonnormality in the tails of the distribution.
- Ryan-Joiner: This test has similar power to Anderson-Darling for the detection of nonnormality.
- Kolmogorov-Smirnov: This test is less sensitive to small deviations from a normal distribution.
For more information, go to Test for normality and click "Types of normality tests".