Homework 7 Solutions


Chapter 16.34

Let X represent the amount of money someone will pay for their dog, and Y the amount for a cat.  We're told that E(X) = 100 and E(Y) = 120.  

a) The question asks to compute E(X-Y) = E(X) - E(Y) = 100-120= -$20

b) Var(X-Y) = Var(X) + Var(Y) because X and Y can reasonably be assumed to be independent.  (There's no reason that the cost of treating a randomly selected dog should affect the cost of treating a randomly selected cat.

Var(X-Y) = 302 + 352 = 2125
SD(X-Y) = Sqrt(2125) = $46.10

We would expect the cat to cost about $20 more, give-or-take about $46.

c) If the dog costs more than the cat, then X > Y  or, using a bit of algebra, X-Y > 0.
So we want P(X - Y > 0).  Let's define W = X-Y.  What do we knwo about W?
First, the probability distribution of W can be described fairly well by the Normal model.  Second, E(W) = -20 and SD(W) = 46.10.  So to find P(W >0) we need to first convert this standard normal variables.  If Z is a standard normal variable, then
P(W > 0) = P(Z > (0 - -20)/46.10) = P(Z > .433)

(in other words, 0 is .433 standard deviations above the mean.)

Using the normal table, P(Z > .433) = 1 - P(Z < .433) = 1 - .667 = .332


*New Problem:  (a) Flip a fair coin until the first head appears.  Let X represent the number of coinflips it takes before the first head appears. Write a probability model for X.

X has values 1, 2, 3, 4, ....   up to infinity. Of course, X is more likely to be a small number than a really big one.
Let's figure out the probabilities:  P(X = 1) = 1/2  because the only way this can happen is if heads is on the first flip.
P(X =2) = (1/2) * (1/2)  = 1/4,  because the only this can happen is if the first flip is tails AND the second is heads.  Since flips of a coin are independent, P(first is tailsAND second is heads= P(first is heads) * P(second is tails) = 1/2 * 1/2.
P(X = 3) = P(first is tails and second is tails and third is heads) = (1/2)*(1/2)*(1/2) = (1/8)

See a pattern?  P(X = 4) = (1/2)4
P(X = 5) = (1/2)5
and so on...  

We could write this as a formula:  P(X = k) = (1/2)k   where k = 1,2,3,4,.....

But for full credit, its enough to say what the pattern is and mention that it continues on this way.  (So, for example, say that it's .5 raised to whatever power x is equal to.)

 (b)  Roll a fair, six-sided die until the first 6-spot appears.  Let Y represent the number of rolls it takes.  Write a probability model for Y.

Very similar problem.  But now P(Y = 1) = (1/6).   P(Y = 2) = P(first is not a 6 AND second is a six) = (5/6) * (1/6)
P(Y = 3) = (5/6)2 *(1/6)
P(Y=4) = (5/6)3 * (1/6)

The pattern is now a little different.  Now we raise (5/6) to the power that's one less than the value of x, and multiply by (1/6).  In terms of a formula:  P(X = k) = (5/6)(k-1)*(1/6)