Suppose you are really bored one afternoon and decide to toss two coins a large number of times, say 10,000 times, recording the number of heads after each pair of coins is tossed. We can treat the tosses as binomial experiments with and . Let denote the number of heads obtained on the toss. Then, we can expect after performing these experiments that around 2,500 of the 's will be 0, around 5,000 of the 's will be 1, and around 2,500 of the 's will be 2. Now suppose we wish to find the average of the 's, that is, the average number of heads per experiment. Based on the number of times we expect to observe each of the possible values, 0,1,2, the average we can expect to see would be,

This quantity is referred to as the **expected value** of the random
variable , the number of heads when two coins are tossed. Note that the
expected value represents the average we would expect to observe if an
experiment is repeated a large number of times, just as the probability of
an event is the proportion of times we would expect the event to occur if
we repeated the experiment a large number of times. It is no
coincidence that the expected value in this case coincides with

where is the p.m.f of .

**Definition**. The **expected value** of a discrete random variable
with p.m.f. is defined to be

The expected value is also commonly referred to as the

- If , then

- More generally, if

then

- Let denote the indicator function for the set . That is,
if and if . Then

- If and are independent, then

The expected value of a r.v. describes the center of the possible values of
the r.v. Note that it is a weighted average of those values in which the
weights are given by the p.m.f. A related quantity, called the
*variance*, describes the variability of those values about this
center. Let be a discrete r.v. with expected value . Then the
*variance* of is defined by:

The variance of a r.v. is commonly represented by . The square is because the unit of measure of the variance is the square of the unit of measure of the r.v.

Note that the properties of expectation imply:

This is usually the easiest way to obtain the variance.

The square root of the variance, called the *standard deviation*
and denoted by , represents a measure of distance between a r.v. and
its mean. In particular, since , then

and so,

Also, if and only if with probability 1. If the s.d. of a r.v. is small, then the r.v. tends to be close to its mean, but if the s.d. is large, then the r.v. tends to be farther from its mean. This is made more precise by Chebychev's inequality.

**Chebychev's inequality**: if is a random variable with mean
and variance , then for any positive constant ,

In particular,

**Example**. Suppose that a company receives a shipment of 50 new
PC's, 4 of which are defective. Suppose that your cost center is given 5 of
these PC's, assumed to be randomly selected from the shipment. Find the
expected value and s.d. of the number of defective PC's received by the cost
center.

**solution**. Let denote the number of defective PC's. We must first
obtain the p.m.f. of . Note that the sample space of is .

Note the these probabilities sum to 1. The expected value and variance can be obtained most easily from the following table.

0 | 0.64696 | 0 | 0 | 0 |

1 | 0.30808 | 0.30808 | 1 | 0.30808 |

2 | 0.04299 | 0.08598 | 4 | 0.17196 |

3 | 0.00195 | 0.00585 | 9 | 0.01755 |

4 | 0.00002 | 0.00008 | 16 | 0.00032 |

sum | 1 | 0.4 | 0.49791 |

So, , , , .

2013-12-17