In this section, we first point out that the approximated number of degrees of freedom, ˆ γ , in (3) is
overestimated. By the fact that (n2−1)S2
2/σ 2
2 is chi-square distributed with n2−1 degrees of freedom,
we have
σ2
2
(n2 − 1)S2
2
∼ Inv-χ2
n2−1, (4)
where Inv-χ2
n2−1 is the inverse-chi-square distribution with n2 − 1 degrees of freedom. By (4), it is
easy to see that for any n2 > 5,
E
1
S2
2
=
n2 − 1
(n2 − 3)σ 2
2
, (5)
E
1
(S2
2 )2
=
(n2 − 1)2
(n2 − 3)(n2 − 5)σ 4
2
. (6)
Further, we have