A researcher wants to compare the variances of two different teaching methods using a two-sample $F$-test. Which condition must be satisfied before using this test?

The two samples must be dependent.

Both samples must be randomly selected from normal populations.

The populations must have equal means.

The sample sizes must be equal.