Messages and Special Cases

Reading time

Location	Cause and Solution
Normality Hypothesis Rejected	Switch to the nonparametric test from the same family, or address the cause (outlier, transformation).
Hypothesis of equal variances rejected	Choose the test that does not require it, or take it into account when interpreting the results.
Outliers Detected	Understand them before taking action. Remove them only if their cause has been identified and is unrelated to the process.
«There is no hypothesis to test for this test.»	This is normal for a nonparametric test.
The result is not statistically significant, even though the difference appears clear	Check the pairing and sample size. Go to the "Power" page.
«Add a new sample to run the Chi-square test»	The test requires at least two samples.
«Not enough assembly/disassembly cycles to calculate the limits»	Repeat the assembly and disassembly cycles.
Significant ANOVA without knowing which groups differ	Work in pairs to draw a conclusion.