Hypotheses about the variance
Suppose we have a sample with 
variance and we assume that the true value of the sample variance is 
, then we can test the following hypotheses: H0 - ratio and the alternative hypothesis HA:
.
In practice, variance is often a measure of risk; therefore, while testing the hypothesis of the equality of variances, we evaluate how accurate is the risk degree is.
The verification of this hypothesis is performed with the help of the VarianceTest function. This function can also check hypotheses on the equality of variances in several samples. For example, if one sample's variance is 
and the other's -
, then the main hypothesis will be an equality check, 
. Mathematica conducts a series of tests to determine the probability of the hypothesis's truth.
|
Test's name |
Type of test |
Description |
|---|---|---|
|
"BrownForsythe" |
Robust |
A robust Levene test |
|
"Conover" |
Symmetry |
This is based on squared ranks of the sample data |
|
"FisherRatio" |
Normality |
This is based on |
|
"Levene" |
robust... |
