Math 143 C/E, Spring 2001
IPS Reading Questions
Chapter 4, Section 4
The expected value is not the outcome you
should expect for a random variable from a single
iteration of the random phenomenon it describes.
That is, you should not expect a payoff of $0.50
on one Tri-State Pick 3 lottery ticket; in fact, the
payoff (to a single ticket-holder) is always either
nothing at all or $500.
Rather, if you keep a running total of the random
variable's value over the course of many iterations
of the same phenomenon (many lottery tickets purchased,
for instance) and divide by the number of iterations,
you should get something close to the expected value.
This means that the state should go into this venture
expecting that the total money paid out to lottery winners
divided by the number of players should be about
$0.50 per player. That's not too bad if players are
charged $1.00 per ticket.
A similar interpretation of the expected value of a random variable can be applied to both Examples 4.20 and 4.21.
You could roll the dice many times, each time keeping track of the roll (the total of pips). After, say, 1000 rolls, you could calculate the average = (total of all rolls)/(number of rolls) and demonstrate that this average was close to 7. You could continue on to 10000 rolls and show that the average was now (probably) closer to 7. Fifty-thousand rolls would generally yield an average even closer to 7.
In the last paragraph on p. 331, they say that most people tend to think a truly random set of 10 coin tosses will not contain any runs of heads (nor of tails) longer than two when, in fact, the probability that such a run will occur for a fair coin is greater than 0.8. In the next paragraph, they point out how people look at streaks in athletics (streaks of made free-throws, streaks of consecutive games where a batter gets a hit) as an indication that the next opportunity (the next free-throw attempt, plate-appearances in the next softball game, etc.) is somehow influenced by or dependent upon recent performance (the hot hand theory), when actually one would expect such streaks even when each opportunity is treated as independent of previous ones.
Since the random variables X and Y are independent (the height of a randomly-chosen male should not affect the height of a randomly-chosen female), Z shows the distribution of the combined heights of one randomly-chosen male and one randomly-chosen female. Nevertheless, people do not choose their spouses randomly, and hence this distribution may not be appropriate for the combined heights of spouses.