A researcher wishes to use the $z$-test to compare the means of two groups. Which requirement is not necessary for the $z$-test?

The population variances must be known

The samples must be independent

The data must be measured at least at the interval scale

The sample sizes must be less than $10$