Let $X_1, X_2, X_3$ be three independent, identically distributed random variables each with density function $$f(x)= \begin{cases} 3x^2 & 0 \le x \le 1 \\ 0 & \text{otherwise} \end{cases} $$ Let $Y=\max\{X_1,X_2,X_3\}$. Find $P(Y\gt \frac{1}{2}).$ Answer key $\frac{511}{512}$

Now the author has not defined what identically distributed random variables are so I looked it up and found:

  • Random variables $X$ and $Y$ are identically distributed if for every set $A\in S$ $P(X\in A)=P(Y \in A)$.

Using this I used the following reasoning

  • WLOG let $Y=\max\{X_1,X_2,X_3\}=X_1$

  • because one of the $X_i$ for $i=1,2,3$ has to be the max

  • and since $P\left(X_1>\frac{1}{2}\right)=P\left(X_2>\frac{1}{2}\right)=P\left(X_3 > \frac{1}{2}\right)$

then $$P(Y >\frac{1}{2})=P(X_1>\frac{1}{2})=\int_{\frac{1}{2}}^1 3x^2 \, dx= \left.x^3 \vphantom{\frac 1 1} \right|_{\frac{1}{2}}^1=1-\frac{1}{8}=\frac{7}{8}.$$ I kind of had a suspicion of the answer being wrong since I did not use independence at all but I guess I do not understand what the question is actually trying to get me to figure out. Now after seeing the solution I have noticed that somehow they must have got $$P(Y>0.5)=\left.x^9\vphantom{\frac 1 1} \right|_{0.5}^1 = \frac{511}{512}$$ but why I do not understand.


Think of a dice and $X_1$ (or $X_2,X_3$ it doesn't matter since they are iid) as the result of one toss. Then, it is clear that the result of $X_1$ is not the same as doing $X_1$ three times and taking the maximum result !

Now, mathematicaly, you need to calculate $P(Y>1/2)=P(\max(X_i)>1/2)$. But the saying that the max of three numbers is greather than a $1/2$ (or any quantity) doesn't give you much information, only that there is at least one of the $X_i$ greater than $1/2$, whereas the opposite event gives : $1-P(Y>1/2) = P(Y<1/2) = P(X_1 < 1/2, X_2 < 1/2, X_3 < 1/2)$. And then you can use independence to calculate your probability.

  • $\begingroup$ thanks it makes sense now. $\endgroup$ – Andrew Jul 9 '16 at 1:59

The probability the maximum is $\gt 1/2$ is $1$ minus the probability the maximum is $\le 1/2$.

And the maximum is $\le 1/2$ if and only if $X_1$, $X_2$, and $X_3$ are all $\le 1/2$.

By independence the probability they are all $\le 1/2$ is $(1/2^3)^3$. This is $\frac{1}{512}$.

Remark: The fact that the $X_i$ all have the same distribution does not play a crucial role. If they had different distributions, we could still find the probability that $Y\le 1/2$ by computing $\Pr(X_1\le 1/2)\Pr(X_2\le 1/2)\Pr(X_3\le 1/2)$. However, the independence is crucial, for that is what allows us to multiply.

  • $\begingroup$ @user101388: You are welcome. $\endgroup$ – André Nicolas Jul 9 '16 at 2:01

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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