ENH: stats: add F-test of equality of variances

This enhancement adds the F-test of equality of variances to SciPy. The F-test evaluates the null hypothesis that two normal populations have the same variance. Despite its sensitivity to the non-normality of distributions, this statistical test is widely used and efficient for comparing two or more approximately normally distributed data samples.

There is an associated issue with this enhancement (ENH: Add F-test of equality of variances to scipy.stats · Issue #19646 · scipy/scipy · GitHub). During my research, I found some people searching for ways to perform an F-test in Python. Given that SciPy already includes Levene’s and Bartlett’s tests, adding the F-test could provide additional value.

Do you think the F-test is needed?

GH Discussion here: ENH: stats: add F-test of equality of variances by gnc1999 · Pull Request #20810 · scipy/scipy · GitHub).

I think it could have a positive impact on scipy’s publicity to have the F-test of variance, as it might appear near the top of search engine’s results when people search for “f test in python” or “f test in scipy”.

That is SEO and there are ways to do that otherwise than adding the support for the function.

See the discussion in the linked issue otherwise. The most important part for us is the maintainability and what this test would bring to our current offering. (We have more powerful tests.)