3
$\begingroup$

You draw two numbers, $A$ and $B$ from a set of integers $\{1,2,3,4,5,6\}$. The numbers are drawn sequentially from the set, without replacement. Find the variance $Var(3A+B)$.

I have tried working on this question for a bit but cannot seem to find a quick way to compute it - perhaps there is some insight that would allow us to skip a lot of the computation?. I do not have a lot of experience with similar questions and so I have just been trying a brute force approach so far. This quickly gets quite convoluted for me, as expanding to $Var(3A+B) = Var(3A) + Var(B) + Cov(3A,B)$ all require further sub steps. Any help greatly appreciated!

My approach so far

$\endgroup$

1 Answer 1

2
$\begingroup$

Variance and covariance can have scaling factors taken out: $$\newcommand{Var}{\operatorname{Var}}\newcommand{Cov}{\operatorname{Cov}}\Var(3A)=9\Var(A)$$ $$\Cov(3A,B)=3\Cov(A,B)$$ Thus $$\Var(3A+B)=9\Var A+\Var B+6\Cov(A,B)$$ where $6$ and not $3$ is the correct multiplier for $\Cov$ because the variance of a sum involves an $(a+b)^2$-type expansion. $A$ and $B$ are identically distributed because $(A,B)$ is a uniform random $2$-sample from the set, so the expression becomes $$\Var(3A+B)=10\Var(A)+6\Cov(A,B)$$ We now derive the explicit numbers: $$E(A)=E(B)=\frac{1+\cdots+6}6=\frac72$$ $$E(A^2)=\frac{1^2+\cdots+6^2}6=\frac{91}6$$ $$\Var(A)=\frac{91}6-\frac{49}4=\frac{35}{12}$$ $$E(AB)=\frac{(1+\cdots+6)^2-1^2-\cdots-6^2}{30}=\frac{35}3$$ $$\Cov(A,B)=\frac{35}3-\frac{49}4=-\frac7{12}$$ Finally $$\Var(3A+B)=10\cdot\frac{35}{12}-6\cdot\frac7{12}=\frac{308}{12}=\frac{77}3$$

$\endgroup$
3
  • $\begingroup$ Thanks for the answer! Could you help me better understand why A and B are identically distributed in this case? The way I have been approaching it is that B is dependent on A because B can only be drawn from the numbers left in the set after A is drawn. For example, if A=1, it is not possible for B=1 however if A, B are iid then P(A=1, B=1) > 0. Am I incorrect in how I am understanding the concept of iid? Thanks again! $\endgroup$ Commented Oct 4, 2022 at 19:21
  • 1
    $\begingroup$ @AdvayMansingka They are not independently distributed as you noticed. However they are identically distributed due to symmetry. You are drawing two elements from the set and calling one $A$ and the other $B$. The order in which $A,B$ are labelled is not important. Identical distribution just means for all $k$ in $\{1,2,3,4,5,6\}$ that $\mathsf P(A=k)=\mathsf P(B=k)$ $\endgroup$ Commented Oct 5, 2022 at 0:18
  • 1
    $\begingroup$ @GrahamKemp Thank you for the comment, this is very helpful! I think (hopefully?) I have a better understanding of the distinction between identical and independent now $\endgroup$ Commented Oct 6, 2022 at 1:02

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .