374 CHAPTER 6 Inferential Statistics
x 1 ,x 2 ,...,xnof this random variable. Associated with this sample are
Thesample mean: this is defined by setting
x =
x 1 +x 2 +···+xn
n
.
The basic reason for considering the sample mean is the follow-
ing. Suppose that we have taken the samplesx 1 , x 2 , ..., xnfrom
a population whose mean is μ. Would we expect that x ≈ μ?
Fortunately, the answer is in agreement with our intuition; that
we really do expect that the sample mean to approximate the
theoretical (or population) mean. The reason, simply is that if we
form the random variable
X=
X 1 +X 2 ···+Xn
n
,
then it is clear that E(X) = μ. (Indeed, we already noted this
fact back on page 324.) That is to say, when we take nindepen-
dent samples from a population, then we “expect” to get back the
theoretical meanμ. Another way to state this is to say thatxis
anunbiased estimate of the population meanμ.
Next, notice that sinceX 1 , X 2 ,···, Xnare independent, we have
that
Var(X) = Var
(X
1 +X 2 +···+Xn
n
)
=
1
n^2
Var(X 1 +X 2 +···+Xn)
=
1
n^2
(Var(X 1 ) + Var(X 2 ) +···+ Var(Xn))
=
σ^2
n