Inference for the Ratio of Variances: How Robust are These Procedures? 

A discussion of the effect of violations of the normality assumption on confidence intervals for the ratio of variances. The effects of different violations of the normality assumption are investigated through simulation. The quick summary: These procedures are very sensitive to violations of the normality assumption, and often perform very poorly when the normality assumption is violated.
This video is very similar in content and results to my video that investigates the effect of violations of the normality assumption on inference procedures for a single variance.