It eliminates the variable entirely.

It reduces the experimental error by accounting for the nuisance variable.

It increases the sample size.

It changes the independent variable.