19.4. Estimation by Random Sampling 805
divided byn:
Var
Sn
n
D
1
n
2
VarŒSnç (Square Multiple Rule for Variance (19.9))
D
1
n^2
Var
"n
X
iD 1
Gi
(def ofSn)
D
1
n^2
Xn
iD 1
VarŒGiç (pairwise independent additivity)
D
1
n^2
n^2 D
^2
n
: (19.23)
This is enough to apply Chebyshev’s Theorem and conclude:
Pr
ˇˇ
ˇ
ˇ
Sn
n