Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I have a question about part of a proof of a Lemma in a book (Casella's Statistical Inference) I'm reading. This it how it goes.

Let $X_1, \cdots ,X_n$ are a random sample from a population and let $g(x)$ be a function such that $\mathbb{E}g(X_1)$ and $\text{Var}\,g(X_1)$ exist. Then $$ \text{Var}\,\left(\sum_{i=1}^{n}g(X_i)\right)=n\left(\text{Var}\,g(X_1)\right).$$

So this is how I proceeded to to prove it.
Since the $X_i's$ are independent, we have that
$$ \begin {align*} \text{Var}\,\left(\sum_{i=1}^{n}g(X_i)\right)&= \text{Var}\,g(X_1)+\cdots +\text{Var}\,g(X_n)\\ &= n\text{Var}\, g(X_1). \end {align*}$$ where the last equality holds because the $X_i's$ are identically distributed. Can I do this? I'm asking this because the proof in the book started by using the definition of the variance and somewhere along the lines involved the covariance matrix.

share|cite|improve this question
Yes, you can do this. Let $Y_i = g(X_i)$. Then $Y_1,\ldots,Y_n$ are i.i.d. random variables with ${\rm Var}(Y_1) < \infty$ (hence also the mean exists). It is an elementary (and very useful) result that, in this case, ${\rm Var}(Y_1 + \cdots + Y_n) = {\rm Var}(Y_1) + \cdots + {\rm Var}(Y_n)$. – Shai Covo Jun 24 '11 at 2:49
Thanks. That's refreshing to know...) – Nana Jun 24 '11 at 2:56
up vote 4 down vote accepted

Shai has answered Nana's question, but in the interest of this question being "officially" answered let's prove the elementary result Shai cites; namely, that if $Y_1, \ldots, Y_n$ are independent random variables with finite mean and variance then $\newcommand{\Var}{\mathrm{Var}}\Var(Y_1 + \cdots + Y_n) = \Var(Y_1) + \cdots + \Var(Y_n)$.

First, let's prove it in the $n=2$ case. If $Y_1$ and $Y_2$ are independent then we know that $E[Y_1 Y_2] = E[Y_1] E[Y_2]$. By a basic property of the variance,

$$\Var(Y_1 + Y_2) = E[(Y_1 + Y_2)^2] - (E[Y_1 + Y_2])^2 = E[Y_1 + 2Y_1Y_2 + Y_2^2] - (E[Y_1] + E[Y_2])^2$$ $$= E[Y_1^2] + 2E[Y_1Y_2] + E[Y_2^2] - E[Y_1]^2 -2E[Y_1]E[Y_2] - E[Y_2]^2$$ $$= E[Y_1^2] - E[Y_1]^2 + E[Y_2^2] - E[Y_2]^2 = \Var(Y_1) + \Var(Y_2).$$

Then, applying the result for the $n=2$ case successively to the general case we have $$\Var(Y_1 + \cdots + Y_n) = \Var(Y_1) + \Var(Y_2 + \cdots + Y_n) = \Var(Y_1) + \Var(Y_2) + \Var(Y_3 + \cdots + Y_n)$$ $$= \cdots = \Var(Y_1) + \cdots + \Var(Y_n).$$

share|cite|improve this answer
Thanks Mike.... – Nana Sep 7 '11 at 0:48

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.