Finding the chances on $n$-th try without replacement

341 Views Asked by At

In Finan's probability book 13.19

A box contains 100 computer mice of which 95 are defective. One mouse is taken from the box at a time (without replacement) until a non-defective mouse is found. Let $X$ be the number of mouses you have to take out in order to find one that is not defective. Find $P(X = n)$.

I thought the way to do this is

${{95!(100-n+1)!}\over{100!(95-n+1)!}}{5\over100-n+1}$

and indeed such is the way it seems from

Repeating something with (1/n)th chance of success n times

see answer from copper.hat there,

but in the book the answer is

$P(X=n)= {95\over100}{94\over99}{93\over98}...{95-n+2\over100-n+2}{5\over100-n+1}.$

Why is it $-n+2$?

2

There are 2 best solutions below

9
On BEST ANSWER

When you have two positive integers $k<m$, then $$\frac{m!}{(m-k)!}=\frac{m\cdot(m-1)\cdot(m-2)\cdot\ldots\cdot (m-k+1)\cdot \color{red}{(m-k)\cdot(m-k-1)\cdot(m-k-2)\cdot\ldots\cdot2\cdot 1}}{\color{red}{(m-k)\cdot(m-k-1)\cdot(m-k-2)\cdot\ldots\cdot 2\cdot 1}} =m\cdot(m-1)\cdot(m-2)\cdot\ldots\cdot (m-k+1).$$

Taking $m=95$ and $k=n-1$, we get $${{95!}\over{(95-n+1)!}}=95\cdot 94\cdot 93\cdot\ldots\cdot (95-n+2).$$

Similarly (and taking the multiplicative inverse of both sides) we have $${{(100-n+1)!}\over{100!}}={1\over 100\cdot 99\cdot 98\cdot\ldots\cdot (100-n+2)}.$$

So indeed $$P(X=n)= {95\over100}{94\over99}{93\over98}...{95-n+2\over100-n+2}{5\over100-n+1}={{95!(100-n+1)!}\over{100!(95-n+1)!}}{5\over100-n+1}.$$


Note that the above works for $m>k>0$, but in some way also for $k=0$. If $k=0$, then we simply get $\frac{m!}{m!}=1$. As you can see, in the case $n=1$ (so $k=n-1=0$) the left fractions in the answer of the book simply disappear (just like a sum iterating from $0$ up to $-1$ is equal to $0$, products of nothing are equal to $1$), while your left fraction turns into $1$, so your answer agrees with the one of the book for all $n\geq 1$.

3
On

Take $n=3$ for example. We have to pull two defective mice, then a non-defective one. The chance of that is $\frac {95}{100} \frac {94}{99}\frac 5{98}$ You can see the decreasing numerators end at $95-3+2=94$ and the denominators at $100-3+1=98$

For the general case, we need to have $n-1$ numbers counting down starting at $95$. That is $n-2$ increments, so the last one is $95-n+2$