PROBABILITY DISTRIBUTIONS

PROBABILITY DISTRIBUTIONS

• Probability distribution:Shows how the probability of an event is

distributed over all possible outcomes.

Discrete: Continuous:
The possible outcomes are The possible outcomes are unlimited within
limited and can be counted as a range of decimal values.
whole numbers.

E.g.: If we roll a pair of dice, the E.g. If the variable in question is body weight,
possible totals are 2, 3, 4, 5, 6, 7, we cannot list all possibilities as exact numbers.
8, 9, 10, 11, 12. If we say a person is 90 kilos, we mean that the
weight is between 89.5 and 90.5 kilos.

2
 The Relative Frequency Distribution of a discrete variable is an
example of a Discrete Probability Distribution

X P(X)
0 .18
Discrete 1 .39 Probability
random 2 .24 that X
variable takes a
represents 3 .14
given value
“number of 4 .04 e.g. P(0) = 0.18
children”. 5 .01
Total 1.00

This is a relative frequency distribution from a survey conducted to determine

the number of children present in each household.

3
 The plot of the Discrete Probability Distribution

0.6 -
P(X) Probability associated with
0.5 - each value of X

0.4 -

0.3 - Plotted points are not joined

as in between values of X are
0.2 - not meaningful (X is discrete).

0.1 -

! ! ! ! !
0 1 2 3 4 5 Number of children

4
The Binomial Distribution
- a discrete probability distribution
Some trials have only two outcomes as in H or T in the toss of a coin
or defective and non-defective items in quality control.
They are Binomial.
Binomial calculations can determine for example the probability of
detecting 2 defectives if 25 items are inspected.
Notation: P(An event occurs, in a single trial) = p;
X = number of times the event occurs in n trials.
Ex 1: Toss a coin three times. P(Head) = p = 0.5 and n = 3 All 8 outcomes
The possible outcomes are: are equally
HHH, HHT, HTH, THH, HTT, THT, TTH and TTT. likely, ie. each
Let X = number of heads. X can be 0, 1, 2 or 3. has a chance
The Binomial Probability Distribution of X: of 1/8.
P(X = 0) = 1/8 Or P(HHH)
P(X = 1) = 3/8 = P(HHT)
P(X = 2) = 3/8 = 0.5x0.5x0.5
P(X = 3) = 1/8 = 1/8
8/8 5
Ex 2:
Consider a game of chance in which a die is rolled 10 times and let X denote
the number of ‘6’ results.
p = 1/6 = 0.1667
x n
What is the probability that 5 times out of 10 trials the outcome is a ‘6’ result?

Using Excel: Enter, = BINOMDIST(x, n, p, FALSE)” to get P(X = 5)

ie. Enter, = BINOMDIST(5,10,0.1667, FALSE)

to get the result 0.013

Entering TRUE instead of FALSE in the above function gives the cumulative
probability of up to x successes, ie. P(X  5)

6
Ex 3: A sign on the petrol pumps of a chain of petrol stations encourages
customers to have their oil checked, claiming that one out of every four cars
should have its oil topped up.
P(the next car entering the station should top up
n its oil) = p = 0.25
a) Of the next 10 cars entering a petrol station, what is the probability that
exactly 3 of them should have their oil topped up?
k
P(X = 3) = P(X  3) - P(X  2) = 0.776 - 0.526 = 0.25

b) What is the probability that at least half of the next 10 cars entering a petrol
station should have their oil topped up?
Its complementary event is that 0-4 of the
next 10 cars should have their oil topped up.
Using the complement rule and the cumulative binomial probability
table (, p = 0.25 and k = 4)

P(X  5) = 1 - P(X  4) = 1 - 0.922 = 0.078

7
 Cumulative Binomial Table (an extract)

n=10 p
k 0,01 0,05 0,1 0,15 0,2 0,25 0,3 0,35 0,4 0,5
0 0,904 0,599 0,349 0,197 0,107 0,056 0,028 0,013 0,006 0,001
1 0,996 0,914 0,736 0,544 0,376 0,244 0,149 0,086 0,046 0,011
2 1,000 0,988 0,930 0,820 0,678 0,526 0,383 0,262 0,167 0,055
3 1,000 0,999 0,987 0,950 0,879 0,776 0,650 0,514 0,382 0,172
4 1,000 1,000 0,998 0,990 0,967 0,922 0,850 0,751 0,633 0,377
5 1,000 1,000 1,000 0,999 0,994 0,980 0,953 0,905 0,834 0,623
6 1,000 1,000 1,000 1,000 0,999 0,996 0,989 0,974 0,945 0,828
7 1,000 1,000 1,000 1,000 1,000 1,000 0,998 0,995 0,988 0,945
8 1,000 1,000 1,000 1,000 1,000 1,000 1,000 0,999 0,998 0,989
9 1,000 1,000 1,000 1,000 1,000 1,000 1,000 1,000 1,000 0,999

8
The Poisson Distribution
- another discrete probability distribution

The Poisson distribution is concerned with the number of “successes”

that occur in a given time interval or in a specified region, if these
events of interest, “successes” occur randomly, independently of one
another, and rarely.
Eg. The number of traffic accidents per week, the number of typos on a page,
the number of cigarette-stubs in a rubbish bin, the number of people joining a
queue, etc.
Say X = number of “successes” in a given time interval or region;
µ = the average number of “successes” in the given time interval
or region.
The Poisson function calculates the probability of X successes in a given time
interval if the average number of successes in that time interval is µ.

For example, it can calculate the probability of 3 people joining a queue in the
next hour, if the average number of people that join a queue in an hour is 6.

9
Ex 4: (Selvanathan p.216, ex.5.66 (2000 edition – p.177, ex. 5.44)
The numbers of accidents that occur on an assembly line have a Poisson
distribution, with an average of three accidents per week.
µ
Find the probability that at least three accidents will occur a week.
Its complement is at most 2.

Using the complement rule and the cumulative Poisson probability table
(Selvanathan, Table 2 of Appendix C, µ = 3 and k = 2)

P ( X  3 )  1  P ( X  2 )  1  0.423  0.577

Using Excel: Enter, =POISSON(x, µ, TRUE)

ie. To obtain P(X 2) enter, =POISSON(2, 3, TRUE)

= 0.423

10
Ex 5: (Selvanathan p.216, ex.5.102 (2000 edition – p.199, ex. 5.76))
The scheduling manager for an electricity supply company knows that there are
an average of 12 emergency calls regarding power failures per month.
µ
a) Find the probability that the company will receive at least 12 emergency
calls during a specified month. Its complement is at most 11.

11
P( X  12)  1  P( X  11)  1   p ( x)  1  0.462  0.538
x 0

b) Suppose that the company can handle a maximum of three emergency

calls per day. What is the probability that there will be more emergency
calls than the company can handle on a given day? (Assume that a month
consists of 30 days.) µ = 12 / 30 = 0.4 per day. X>3

3
P( X  3)  1  P( X  3)  1   p( x)  1  0.999  0.001
x 0

11
 An extract from the Poisson Tables


x 0.50 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00
0 0.607 0.368 0.135 0.050 0.018 0.007 0.002 0.001 0.000 0.000
1 0.910 0.736 0.406 0.199 0.092 0.040 0.017 0.007 0.003 0.001
2 0.986 0.920 0.677 0.423 0.238 0.125 0.062 0.030 0.014 0.006
3 0.998 0.981 0.857 0.647 0.433 0.265 0.151 0.082 0.042 0.021
4 1.000 0.996 0.947 0.815 0.629 0.440 0.285 0.173 0.100 0.055
5 0.999 0.983 0.916 0.785 0.616 0.446 0.301 0.191 0.116
6 1.000 0.995 0.966 0.889 0.762 0.606 0.450 0.313 0.207
7 0.999 0.988 0.949 0.867 0.744 0.599 0.453 0.324
8 1.000 0.996 0.979 0.932 0.847 0.729 0.593 0.456
9 0.999 0.992 0.968 0.916 0.830 0.717 0.587
10 1.000 0.997 0.986 0.957 0.901 0.816 0.706
11 0.999 0.995 0.980 0.947 0.888 0.803
12 1.000 0.998 0.991 0.973 0.936 0.876
13 0.999 0.996 0.987 0.966 0.926
14 1.000 0.999 0.994 0.983 0.959

12
PROBABILITY DISTRIBUTION OF A CONTINUOUS
RANDOM VARIABLE

The relative frequency distribution of a continuous

random variable is its probability distribution

Assembly time Relative

(seconds) Frequency
aX<b P (a  X < b)
50 - (< 60) .05
Continuous 60 - (< 70) 0.12
data 70 - (< 80) 0.67 Total
80 - (<90) 0.13 relative
90 - (<100) 0.03
frequency
Total 1.00 is 1.

13
P (a  X < b) The Plot of a Continuous Probability Distribution
0.7 - X
Relative Frequency
0.6 -
on the vertical axis

0.5 -

0.4 -

0.3 -

0.2 -

X Assembly
0.1 - X Time (s)
X X
! X ! ! ! ! ! ! X ! X
40 50 60 70 80 90 100 110
f(x) Probability Density Function

0.07 -
Relative Frequency X
Class Interval
0.06 -
on the vertical axis
Area = Rel. Freq. x Class Int.
0.05 - Class Int.
Area = Relative Frequency
0.04 -
Since the total relative
0.03 - frequency is 1, the total
area is 1.
0.02 -

X Assembly
0.01 - X Time (s)
X X
! X ! ! ! ! ! ! X ! X
40 50 60 70 80 90 100 110
Ex 6: Say X : Assembly time and n =1000.

5 class intervals 10 class intervals

20 class intervals 30 class intervals

16
Finally, when the class intervals are quite 50 class intervals
narrow, the relative frequency histogram
can be replaced with a smooth curve.

The area of each rectangle is

proportionate to the corresponding relative
frequency.

As the area under this smooth curve is 1,

the area between two values a and b a b
measures the probability that X takes a
value between a and b,
i.e. P (a  X  b).

Note: P (X = a) = 0, i.e. the probability that X (or in general any continuous

random variable) assumes a particular value is always zero.
P (a  X  b) = P (a < X < b), i.e. it does not matter whether
the interval is closed or open.
17
 NORMAL DISTRIBUTION
• It is the most important continuous probability distribution for three
reasons:
i. It describes the distributions of numerous variables observed
in practice (like e.g. test results, heights of pupils) fairly well.
ii. It approximates many other probability distributions (e.g.
binomial).
iii. It is used intensively in inferential statistics.

• The curve of its density

function is called normal f (X)
curve. It is
– bell-shaped,
– symmetrical around the
mean (μ),
– tailed indefinitely into both μ
X
directions,
– with a total area of one. Note: The formula of the normal density
function can be found in Selv. p.184.
18
 • The normal distribution has two parameters:
μ (mean) and σ (standard deviation)

It specifies the central location. It determines the spread.

σ1

σ 1 < σ2
σ2

μ1 < μ2

Since μ can take on any real number and σ can take on any positive real
number, there are infinitely many different normal distributions.
One of them, the so called standard normal distribution has μ = 0 and σ = 1.

19
• All normal random variables (say X with μ and σ parameters) can be
transformed to the standard normal variable (denoted as Z) by the
following formula:
X 
Z

If we substitute a certain value for X, say x, the corresponding
value of Z equals the distance from x to μ, measured in σ.
Z = 2 means x is two standard deviations away from the mean.

Using this transformation formula and the standard normal table we can
compute probabilities for any normal distribution.

It tabulates P (0 < Z < z) for 0  z  3.09.

Since the standard normal distribution is
symmetrical around zero, there is no need
to tabulate negative z values.

20
 An extract from the Normal table

Z 0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07

0.0 0.0000 0.0040 0.0080 0.0120 0.0160 0.0199 0.0239 0.0279
0.1 0.0398 0.0438 0.0478 0.0517 0.0557 0.0596 0.0636 0.0675
0.2 0.0793 0.0832 0.0871 0.0910 0.0948 0.0987 0.1026 0.1064
0.3 0.1179 0.1217 0.1255 0.1293 0.1331 0.1368 0.1406 0.1443
0.4 0.1554 0.1591 0.1628 0.1664 0.1700 0.1736 0.1772 0.1808
0.5 0.1915 0.1950 0.1985 0.2019 0.2054 0.2088 0.2123 0.2157
0.6 0.2257 0.2291 0.2324 0.2357 0.2389 0.2422 0.2454 0.2486
0.7 0.2580 0.2611 0.2642 0.2673 0.2704 0.2734 0.2764 0.2794
0.8 0.2881 0.2910 0.2939 0.2967 0.2995 0.3023 0.3051 0.3078
0.9 0.3159 0.3186 0.3212 0.3238 0.3264 0.3289 0.3315 0.3340
1.0 0.3413 0.3438 0.3461 0.3485 0.3508 0.3531 0.3554 0.3577
1.1 0.3643 0.3665 0.3686 0.3708 0.3729 0.3749 0.3770 0.3790
1.2 0.3849 0.3869 0.3888 0.3907 0.3925 0.3944 0.3962 0.3980
1.3 0.4032 0.4049 0.4066 0.4082 0.4099 0.4115 0.4131 0.4147
1.4 0.4192 0.4207 0.4222 0.4236 0.4251 0.4265 0.4279 0.4292
1.5 0.4332 0.4345 0.4357 0.4370 0.4382 0.4394 0.4406 0.4418
1.6 0.4452 0.4463 0.4474 0.4484 0.4495 0.4505 0.4515 0.4525
1.7 0.4554 0.4564 0.4573 0.4582 0.4591 0.4599 0.4608 0.4616
1.8 0.4641 0.4649 0.4656 0.4664 0.4671 0.4678 0.4686 0.4693
1.9 0.4713 0.4719 0.4726 0.4732 0.4738 0.4744 0.4750 0.4756
2.0 0.4772 0.4778 0.4783 0.4788 0.4793 0.4798 0.4803 0.4808
2.1 0.4821 0.4826 0.4830 0.4834 0.4838 0.4842 0.4846 0.4850

21
Ex 7:
Let X be normally distributed with μ = 100 and σ = 10.
What is the probability that X 120? In symbols: P (X 120)?

P (X 120) = P (Z 2)
0.5000 0.4772 From the standard normal table
P (0  Z 2) = 0.4772.
P (Z < 0) = P (Z > 0) = 0.5,
(using the complementary rule)
X P (Z 2) = 0.5 – 0.4772 = 0.0228.
100 120
P (X 120) = 0.0228 or 2.28%.
Z
0 2
Excel gives the cumulative normal Probability:
x 120  100 P(X  x) with the function
z  2 =NORMDIST(x, μ, σ, TRUE)
 10
ie. =NORMDIST(120, 100, 10, TRUE)
Note: Each point on the Z axis = 0.9772 ie. P(X 120) =1- 0.9772=0.0228
corresponds to the point right
above it on the X axis.
22
Ex 8:
The weekly demand at a supermarket chain for 375 gram jars of KGB freeze
dried instant coffee is approximately normal with mean 77.8 cases and
standard deviation 12.9 cases.

What is the probability that

a) Demand will be between 70 and 90 cases in a particular week?
X

P(70  X  90)  P(0.60  Z  0.95) 0.2257 0.3289

70  77.8 90  77.8
z z
12.9 12.9 X
70 77.8 90
0
Z
-0.60 0 0.95

 P(0  Z  0.60)  P(0  Z  0.95)  0.2257  0.3289  0.5546

Note: Always use a simple sketch to visualize normal probability questions.

23
b) Demand will be more than 100 cases in a particular week?

P( X  100)  P( Z  1.72)
0.5000 0.4573

100  77.8
z
12.9 X
77.8 100

 0.5000  P(0  Z  1.72)  0.5000  0.4573  0.0427 0 1.72

Z

c) Demand will be more than 60 cases in a particular week?

P( X  60)  P( Z  1.38)
0.4162 0.5000

60  77.8
z
12.9
X
60 77.8

 0.5000  P(0  Z  1.38)  0.5000  0.4162  0.9162 Z

-1.38 0
24
Ex 9: Companies are interested in the demographics of those who listen to
the radio programs they sponsor. A radio station has determined that the mean
number of male listeners phoning into a morning talk-back program is 40 with a
standard deviation of 5.657. During a particular week, 200 calls are received by
this program.

What is the probability that

a) At least 50 of these 200 callers are male?
We have to find P (X  50).

Therefore, z = (50 – 40) / 5.657  1.77, and

P( X  50)  P( Z  1.77)  0.5  P(0  Z  1.77)  0.5  0.4616  0.0384

25
 The Reverse Normal Calculation

c) There is a 30% chance that the male callers among the 200 total callers
does not exceed what number?

P (X  x) = 0.3 and x is
P( xunknown.
 X  40)  P( z  Z  0)  0.2
Using the standard normal table 0.3 0.2
‘backwards’ we get that z  0.525.

Using Excel: =NORMSINV(0.3) = -0.524 X

x 40

X  X  Z  
Z
Z implies Z 0
 i.e. there is a 30% chance that there
thus -0.525 x 5.656 + 40 = 37 will be at most 37 male callers.
26
 EMPIRICAL RULE
If a sample of measurements has a mound-shaped distribution, i.e. a more
or less symmetrical distribution with a single mode, the interval
( x  s ; x  s) contains about 68% of all measurements,

( x  2 s ; x  2 s ) contains about 95% of all measurements,

( x  3s ; x  3s) contains all or a vast majority of measurements.

Any value outside the third (or even the second) interval is an outlier.

In example 4, the distribution of dividend yields was mound-shaped, so

let us calculate the third interval for this distribution:

x  3 s  4.4  3  1.8  ( 1 to 9.8 )

We expect almost all observations to be within this interval. Therefore, if

an observation happens to be outside this interval, we can consider it
an outlier.
27