Why may Bartlett's variance test reject for groups with equal variances but nonnormal distributions?
StatisticsLevel: IntermediateDistribution checking
Why may Bartlett's variance test reject for groups with equal variances but nonnormal distributions?
Hints
- A variance test also has distribution assumptions.
- Check tail sensitivity.
Answer
Its calibration is sensitive to nonnormality. A robust alternative such as Levene or Fligner may be more appropriate for the sampling assumptions.
Source and licence: SciPy — distribution diagnostics and robust tests · BSD-3-Clause · answer: Quantoros editorial