= < <
2. $f X N(,
2
), it can be shown that %(X) & and '(X) &
2
. This implies that
there is a famil! of normal curves, and each member of this famil! is completel!
determined b! the values of and
2
. (oreover, ever! normal curve is s!mmetric
about its mean (), and its spread is determined b! its variance (
2
).
). *eardless of the particular values of the parameters and
2
, ever! normal
distribution has the followin properties+sometimes called the %mpirical *ule,
-..) percent of all the observations lie within one standard deviation of the mean
/0.0 percent of all the observations lie within two standard deviations of the mean
//.1 percent of all the observations lie within three standard deviations of the mean
2. $t seems obvious that computin probabilities associated with a random variable
havin a normal pdf miht be difficult. $ndeed computin interals of the form
( )
d"
2
2
2
b
1
P(a X b)
2
a
x
e
=
b. 3ortunatel!, selected values associated with the 5tandard Normal Distribution
are iven in Table %.
i. 3or e"ample, if 7 N(0, 1), we can use Table % to find 9(7 1.2.) to be .2)6-
: .0 & ./)6-.
ii. 3ind the 9(7 ; 1.12)
9(7 ; 1.12) & .0 < 9(6 = 7 = 1.12) & .0 > .)12/ & .1211
iii. 3ind the value of 8 such that 9(7 ; 8) & .1162
?e see@ the value of 8 such that the area to its riht is .1162. 5ince this
probabilit! is reater than 6.0, we @now that 8 is to the left of 8ero. 3rom Table
%, we find 8 & >.12. Note that the area between >.12 and 6 is .2162, and we add
this to the area to the riht of 8ero to et .2162 : .0 & .1162.
2. Throuh an appropriate transformation, the standard normal distribution can be
used to compute probabilities associated with other, nonstandard, normal
distributions.
a. $f X N(,
2
), then 7 & (X > )/ N(6, 1). That is an! random variable that
has a normal distribution can be standardi8ed b! subtractin its mean from it
and dividin b! its standard deviation. Thus, an! probabilit! involvin X can
be e"pressed as a probabilit! involvin a standard normal random variable, 7.
This ver! powerful result enables us to use Table % to compute probabilities
associated with normall! distributed random variables.
b. 3or e"ample, Telcom 5atellite provides communication services to businesses
in the Ahicao area. Aompan! officials have learned that the averae satellite
transmission is 106 seconds, with a standard deviation of 10 seconds. These
times appear to be normall! distributed. To properl! estimate customer demand
for its services and establish a rate structure which will ma"imi8e corporate
profits, Telcom must determine how li@el! some calls are to happen. 3or
e"ample, what is the probabilit! that a call will last between 120 and 100
secondsB
?e have assumed that X, the lenth of satellite transmissions are normall!
distributed with mean 106 seconds and standard deviation 10 seconds. ?e want
to compute 9(120 = X = 100).
9(120 = X = 100) & 9C(120 < 106)D10 = 7 = (100>106)D10E
& 9(>.)) = 7 = .)))
& 2F.12/)
& .20.-
). There are %"cel functions that can be used in place of Table % to compute
probabilities associated with a normal or standard normal distribution. 4ecause of
their ease of use, $ definitel! prefer usin these %"cel functions.
a. &Normdist(",,,1) & 9(X = ") if X N(,
2
). Thus,
9(120 = X = 100) &Normdist(100,106,10,1)>Normdist(120,106,10,1)
b. &Normsdist(8) & 9(7 = 8)
c. &Norminv(p,,) returns the p
th
percentile of the N(,
2
) distribution. Thus, to
find the lenth of call that represents the /6
th
percentile, we would use
&Norminv(./,106,10).
d. Gi@ewise, &Normsinv(p) returns the p
th
percentile of the N(0, 1) distribution.