Submitted by jetay1 on Sat, 03/04/2017 - 22:39
I've started working on the Step 1 stats sheet. Hopefully they start to get easier after a while b/c I found the first question quite tough.
I think I got there after persisting for while and putting it down and coming back to it later, but I have a couple of questions (I appreciate you're not step examiners, but any advice will be gratefully received):
1. For proving $E(X)=\dfrac{1}{pq}-1$, I used the formula for $E(X)$ from the start of the question and then worked out $P(X\geqslant n)$ for $n=1,2,3,4$. Once I saw that it went $1+(p+q)+(p^2+q^2)+(p^3+q^3)$, I just assumed that the pattern would continue, and then the required formula followed easily. My question is: would it be ok to just guess that the pattern continues or should I be proving it?
2. To show $E(X)\geqslant3$, I differentiated wrt $p$ and found a minimum at $p=\frac{1}{2}$. I used a first derivative test to verify it was a minimum and then $E(X)=3$ at $p=\frac{1}{2}$, hence $E(X)\geqslant3$. I feel like there's often a bit of guesswork involved in deciding exactly what things are allowed in a proof, especially one that starts 'hence...'. Do you think this approach would be acceptable? Does anyone reading this have a more elegant way of showing $E(X)\geqslant3$?
1. You should definitely be
1. You should definitely be proving it, there's no reason for the pattern to continue. However, in this case, it's sort of obvious why $\mathbb{P}(X \geq n) = p^{n-1) + q^{n-1}$ for $n\geq 3$. I would justify by saying that there can't have been a mummy and daddy penguin amongst the first $n-1$ boxes so it follows, but avoid saying "because the pattern continues". I'd suspect that for this particular question, you wouldn't lose any marks for just writing down $p^{n-1} + q^{n-1}$ because of how straightforward it is, but I'd caution against this in general. Justify your answers.
2. Your approach is fine. Other ways include:
(i) $pq = p(1-p) = \frac{1}{4} - \left(p-\frac{1}{2}\right)^2 \leq \frac{1}{4}$ so $\frac{1}{pq} \geq 4.$
(ii) Use the AM-GM inequality that says that $\sqrt{pq} \leq \frac{p+q}{2}$ but $p+q = 1$ so $pq \leq 4$ so $\frac{1}{pq} \geq 4$
(iii) Write $\frac{1}{pq} = \frac{(p+q)^2}{pq}$ and then work with that.
(iv) there are loads of methods, but yours is fine tbh. (although I would personally justify that you don't need to check the endpoints (p=0 or p=1) because $p,q \neq 0$ from the STEM, although you wouldn't lose any marks, extra justification won't hurt - remember that the derivative only tells you that there's a minimum in (0,1) [p=0 or p=1 might be global minimums])
Hmm, looks like the forum
Hmm, looks like the forum doesn't have the amsmaths package, anyway the broken LaTeX there should read $P(X \geq n) = p^{n-1} + q^{n-1}$
Thanks very much.
Thanks very much.
I think your broken latex code is caused by a round bracket where there should be a curly bracket.
No problem. Good spot.
No problem.
Good spot.