用样本均值代替总体均值，自由度会减1，所以分母是n-1。
严格的推导如下：
首先，我们先看看方差的计算公式
Var
(
X
)
=
∑
i
=
1
n
(
X
i
−
μ
)
2
n
\operatorname{Var}(X)=\frac{\sum_{i=1}^{n}\left(X_{i}-\mu\right)^{2}}{n}
Var(X)=n∑i=1n(Xi−μ)2
其中μ
是这个总体的真实均值。但是往往μ是未知的，所以我们用样本均值
X
ˉ
\bar{X}
Xˉ来代替μ，也就是
S
∗
=
∑
i
=
1
n
(
X
i
−
X
ˉ
)
2
n
S_{*}=\frac{\sum_{i=1}^{n}\left(X_{i}-\bar{X}\right)^{2}}{n}
S∗=n∑i=1n(Xi−Xˉ)2
那么这个S∗
是正确的估计吗？我们可以计算S∗
的期望来对比一下Var(X)
。
E
(
S
∗
)
=
E
(
∑
i
=
1
n
(
X
i
−
X
ˉ
)
2
n
)
=
E
(
∑
i
=
1
n
(
X
i
−
μ
+
μ
−
X
ˉ
)
2
n
)
=
E
(
∑
i
=
1
n
(
X
i
−
μ
)
2
n
+
∑
i
=
1
n
2
(
X
i
−
μ
)
(
μ
−
X
ˉ
)
n
+
∑
i
=
1
n
(
μ
−
X
ˉ
)
2
n
)
=
Var
(
X
)
+
E
(
∑
i
=
1
n
2
(
X
i
−
μ
)
(
μ
−
X
ˉ
)
n
+
∑
i
=
1
n
(
μ
−
X
ˉ
)
2
n
)
=
Var
(
X
)
+
E
(
−
2
(
X
ˉ
−
μ
)
2
+
(
X
ˉ
−
μ
)
2
)
=
Var
(
X
)
−
E
(
(
X
ˉ
−
μ
)
2
)
=
Var
(
X
)
−
Var
(
X
ˉ
)
=
Var
(
X
)
−
1
n
Var
(
X
)
=
n
−
1
n
Var
(
X
)
\begin{aligned} \mathbb{E}\left(S_{*}\right) &=\mathbb{E}\left(\frac{\sum_{i=1}^{n}\left(X_{i}-\bar{X}\right)^{2}}{n}\right) \\ &=\mathbb{E}\left(\frac{\sum_{i=1}^{n}\left(X_{i}-\mu+\mu-\bar{X}\right)^{2}}{n}\right) \\ &=\mathbb{E}\left(\frac{\sum_{i=1}^{n}\left(X_{i}-\mu\right)^{2}}{n}+\frac{\sum_{i=1}^{n} 2\left(X_{i}-\mu\right)(\mu-\bar{X})}{n}+\frac{\sum_{i=1}^{n}(\mu-\bar{X})^{2}}{n}\right) \\ &=\operatorname{Var}(X)+\mathbb{E}\left(\frac{\sum_{i=1}^{n} 2\left(X_{i}-\mu\right)(\mu-\bar{X})}{n}+\frac{\sum_{i=1}^{n}(\mu-\bar{X})^{2}}{n}\right) \\ &=\operatorname{Var}(X)+\mathbb{E}\left(-2(\bar{X}-\mu)^{2}+(\bar{X}-\mu)^{2}\right) \\ &=\operatorname{Var}(X)-\mathbb{E}\left((\bar{X}-\mu)^{2}\right) \\ &=\operatorname{Var}(X)-\operatorname{Var}(\bar{X}) \\ &=\operatorname{Var}(X)-\frac{1}{n} \operatorname{Var}(X) \\ &=\frac{n-1}{n} \operatorname{Var}(X) \end{aligned}
E(S∗)=E(n∑i=1n(Xi−Xˉ)2)=E(n∑i=1n(Xi−μ+μ−Xˉ)2)=E(n∑i=1n(Xi−μ)2+n∑i=1n2(Xi−μ)(μ−Xˉ)+n∑i=1n(μ−Xˉ)2)=Var(X)+E(n∑i=1n2(Xi−μ)(μ−Xˉ)+n∑i=1n(μ−Xˉ)2)=Var(X)+E(−2(Xˉ−μ)2+(Xˉ−μ)2)=Var(X)−E((Xˉ−μ)2)=Var(X)−Var(Xˉ)=Var(X)−n1Var(X)=nn−1Var(X)
所以
Var
(
X
)
=
n
n
−
1
E
(
S
∗
)
=
∑
i
=
1
n
(
X
i
−
X
ˉ
)
n
−
1
\operatorname{Var}(X)=\frac{n}{n-1} \mathbb{E}\left(S_{*}\right)=\frac{\sum_{i=1}^{n}\left(X_{i}-\bar{X}\right)}{n-1}
Var(X)=n−1nE(S∗)=n−1∑i=1n(Xi−Xˉ)
3299
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
