A discussion is given of the use of the indirect variational method for generating the trial functions needed to compute a variational estimate of a homogeneous functional of the solution to an eigenvalue equation. It is shown that one use of the method leads to no difficulties, whereas another use gives meaningless results. In this latter instance, the method of weighted residuals can be used to generate the necessary trial functions. With the trial functions known, the variational estimate of the functional of interest follows by quadrature.