372 Chapter 9: Regression
variables and is thus itself normally distributed. Because we already know its mean, we
need only compute its variance, which is accomplished as follows:
Var(A+Bx 0 )=
∑n
i= 1
[
1
n
−c(xi−x)(x−x 0 )
] 2
Var(Yi)
=σ^2
∑n
i= 1
[
1
n^2
−c^2 (x−x 0 )^2 (xi−x)^2 − 2 c(xi−x)
(x−x 0 )
n
]
=σ^2
[
1
n
+c^2 (x−x 0 )^2
∑n
i= 1
(xi−x)^2 − 2 c(x−x 0 )
∑n
i= 1
(xi−x)
n
]
=σ^2
[
1
n
+
(x−x 0 )^2
Sxx
]
where the last equality followed from
∑n
i= 1
(xi−x)^2 =
∑n
i= 1
xi^2 −nx^2 =1/c=Sxx,
∑n
i= 1
(xi−x)= 0
Hence, we have shown that
A+Bx 0 ∼N
(
α+βx 0 ,σ^2
[
1
n
+
(x 0 −x)^2
Sxx
])
(9.4.4)
In addition, becauseA+Bx 0 is independent of
SSR/σ^2 ∼χn^2 − 2
it follows that
A+Bx 0 −(α+βx 0 )
√
1
n
+
(x 0 −x)^2
Sxx
√
SSR
n− 2
∼tn− 2 (9.4.5)
Equation 9.4.5 can now be used to obtain the following confidence interval estimator of
α+βx 0.
Confidence Interval Estimator ofα+βx 0
With 100(1−a) percent confidence,α+βx 0 will lie within
A+Bx 0 ±
√
1
n
+
(x 0 −x)^2
Sxx
√
SSR
n− 2
ta/2,n− 2