Chapter seven: MEASURES OF

DISPERSION

03/13/25 1
 MEASURES OF DISPERSION

Measures of central tendency (mean, median, mode, GM and HM) do

not provide all information about the observations contained in a
data set that how the individual observations are scattered around
the central value. It is possible with the help of measures of
dispersion.

A single value which measure that how the individual observations of

a data set are scattered/dispersed around the central value, is called
measure of dispersion. Measures of dispersion are classified as
“Absolute measures” and “Relative measures” of dispersion.

03/13/25 2 2
 MEASURES OF DISPERSION

03/13/25 3 3
 CONT’S
• Absolute measure: A type of dispersion which can be
expressed in the same unit of measurement in which
the original series/data set/ distribution is given, is
called “Absolute measure” of dispersion. For
example, Range, Quartile deviation, Mean deviation,
Standard deviation etc. Similarly, a type of
dispersion which is independent of unit of
measurement is called “Relative measure” of
dispersion. For example, coefficient of range,
coefficient of quartile deviation, coefficient of mean
deviation, coefficient of variation etc.
03/13/25 4
 COMMONLY USED MEASURES OF DISPRSION

Absolute Measures of Dispersions

 Range
 Inter quartile range
 Semi inter quartile range or Quartile Deviation (QD)
 Mean deviation
 Variance and
 Standard deviation
Relative Measures of Dispersions
 Coefficient of Range
 Coefficient of Inter quartile range
 Coefficient Semi inter quartile range
 Coefficient Mean deviation
 Coefficient of Variation
03/13/25 5 5
Range and Its Coefficient
Range is an absolute measure of dispersion. “It is the difference between
maximum (Xm) and minimum (X0) values of a data set”. Mathematically, range
is defined as:
Range = Xm – X0
Its relative measure is called coefficient of range and can be defined as:
Xm  X0
Coefficient of Range =
For Example: The followingXdata
m + indicate
X0 the amount of fill (in ml) of 5
different bottle by a soft drink company. The data is 12.5, 12.3, 12, 13, 12.8
So, Range = 13-12 = 1 ml.
Coefficient of Range = (13-12)/(13+12) = 1/25.
Note: Generally, range is a rough measure of dispersion and not preferred to be
used for the same purpose.
03/13/25 6 6
Inter Quartile Range and Quartile Deviation

Inter Quartile Range (IQR) is an absolute measure of dispersion. “It is the

difference between upper quartile (Q3) and lower quartile (Q1) of a data set”.
Mathematically, IQR is defined as:
IQR = Q3 – Q1
Quartile Deviation (Semi inter quartile range): It is half of the inter quartile
range. It is also called quartile deviation (QD) and is expressed as:
SIQR = QD = (Q3 – Q1)/2
A relative measure of IQR and SIQR is called coefficient of IQR and
coefficient of SIQR (coefficient of QD), respectively and can be expressed as:
Coefficient IQR = (Q3 – Q1)/ (Q3 + Q1)

Coefficient of SIQR =Coefficient of QD = (Q3 – Q1)/ (Q3 + Q1)

03/13/25 7 7
 Example: The following data shows the frequency distribution of the salary of
50 employees of a firm. Calculate the following:
1. Inter quartile range and its coefficient
2. Semi inter quartile range and its coefficient

Salary (000) Class boundary f cf To compute the required quantities,

first we will find the group/class in
5-9 4.5-9.5 5 5 which it will lie to occur.
10-14 9.5-14.5 8 13 For Q1: n/4 = 50/4 = 12.5. It indicate
15-19 14.5-19.5 20 33 the Q1 lies in the group (9.5-14.5).
20-24 19.5-24.5 10 43 For Q3: 3n/4 = 3*50/4 = 37.5. It
25-29 24.4-29.5 4 47 indicate the Q3 lies in the group
(19.5-24.5).
30-34 29.5-34.5 3 50
h
50 Q3 = l  (3n / 4  c);
h n
f
Q1 = l + (  c)
f 4 put l = 19.5, h 5, f 10, c 33
put l 9.5, h = 5, f = 8, c = 5
5 Q3 = 19.5+5/10(37.5-33) = 21.75
Q1 = 9.5 + (12.5  5)
8
Q1 = 9.5 + 4.69 = 14.19

03/13/25 8 8
 1. IQR = Q3 – Q1

= 21.75 - 14.19 = Rs. 7.56

Coefficient of IQR = (Q3 – Q1)/(Q3 + Q1)

= (21.75 - 14.19 )/(21.75 + 14.19 )

= 7.56/35.94 = 0.21
2. SIQR = QD = (Q3 – Q1)/2 = (21.75 - 14.19 )/2

= 7.56/2 = Rs. 3.78

Coefficient of SIQR = (Q3 – Q1)/(Q3 + Q1)
= (21.75 - 14.19 )/(21.75 + 14.19 )
= 7.56/35.94 = 0.21
03/13/25 9 9
MEAN DEVIATION AND ITS COEFFICIENT

Mean deviation is an arithmetic mean of absolute deviations taken from any

central value (mean, median, mode) of a data set. It is an absolute measure of
dispersion. Mathematically, it can be defined as:

MD = 
|X-X|
( for individual series)
n
MD =  f |X-X|
( for discrete/contineous series)
f
Its relative measure is called coefficient of mead deviation and can be expressed
as:
Mean Deviation
Coefficient of MD =
Arithmatic Mean
MD
Coefficient of MD =
X
03/13/25 10 10
 Example: The following data indicate the consumption (000) of 7
different households in a certain locality of Peshawar. Calculate Mean
deviation and its coefficient.
2 4 6 8 10 12 14

X (X-mean) |X-mean |
Mean =  X
56 / 7 8 2 -6 6
n 4 -4 4
and 6 -2 2

MD = 
8 0 0
|X-X|
24 / 7 3.43 10 2 2
n 12 4 4
MD 14 6 6
Coefficient of MD =
Mean 56 24

3.43 / 8 0.429

03/13/25 11
 Numerical Example (Mean)

Using the following data showing the weights (in grams) of 60 apples,
calculate the mean deviation from mean and its coefficient.

Weight 65-84 85-104 105-124 125-144 145-164 165-184 185-204 Tota

l
Number of apples 9 10 17 10 5 4 5 60

MD =  f |X  X |
........... (1)
f
where, X =  fX ............... (2)
f

Necessary computations are given below:

03/13/25 12
Weight 65-84 85-104 105-124 125-144 145-164 165-184 185-204 Total
Number of apples (f) 9 10 17 10 5 4 5 60
Mid point (X) 74.5 94.5 114.5 134.5 154.5 174.5 194.5
fx 670.5 945 1946.5 1345 772.5 698 972.5 7350
f |X X| 432 280 136 120 160 208 360 1696

So equation (2) becomes:

7350
X = 122.5
60
Using equation (1), we have:

1696
MD = 28.27
60
MD
Coefficient of MD 
X
28.27
 0.2307
122.5
03/13/25 13
 Numerical Example (Mean)

Using the following discrete frequency distribution, showing the weights (in
grams) of 60 apples, calculate the mean deviation from mean and its
coefficient.
Weight 74.5 94.5 114.5 134.5 154.5 174.5 194.5 Total

Number of apples 9 10 17 10 5 4 5 60

Try yourself by using the formula as used for grouped/continuous

frequency distribution:

03/13/25 14
VARIANCE: Variance is the arithmetic mean of the squared deviation taken
from mean. Mathematically, variance is defined as:
2

 2
=  (X  ) 2

  X 2
X 
  (for population data)
N N  N 
2

S 2
=  (X  X) 2

  X 2
X 
  (for sample data)
n n  n 
For frequency distribution (discrete or continuous), variance is expressed as:
2

S 2
=  f (X  X) 2

  fX 2
  fX 
 
f f  f 
It is important to mention that variance is always be positive number (or non-
negative). It can not be a negative value.

03/13/25 15
STANDARD DEVIATION: Standard deviation is the square root of variance
and is expressed as, mathematically.
2

SD  =
 (X  ) 2

 X 

2
 X
 (for population data)
N N  N 
 
2

S=
 (X  X ) 2

 X 
2

 X
 (for sample data)
n n  n 
 

For frequency distribution (discrete or continuous), standard deviation is

expressed as:
2

SD S = Variance 
 f (X  X ) 2

 fX 2

 fX 

f  f 
  f 

Note: Standard deviation will always be a positive number (or non-negative). It

can not be a negative value. Standard deviation is an absolute measure of
dispersion.
03/13/25 16
COEFFICIENT OF VARIATION:
Coefficient of variation, denoted by CV is a relative measure of dispersion and
used for comparing the consistency to different data sets. It is the ratio of
standard deviation to arithmetic mean and the result is expressed as percentage.
Mathematically, CV is expressed as:
SD
CV = Coefficient of Variation = 100
Mean

While comparing the two data sets (series) A and B, the series A is defined to be
consistent than B if CV of series A is less than the CV of series B i.e.

CVA < CVB

03/13/25 17
 Example: The following data indicate the consumption (000) of 7 different
households in a certain locality of Peshawar. Calculate standard deviation,
variance, and coefficient of variation.
2 4 6 8 10 12 14

2 X X2 ( X  X )2
S= (X  X) 2

= X 2

  
 X 
2 4 36
n n  n 
4 16 16
2
560  56  6 36 4
     16 4
7  7  8 64 0
Variance = S 2 (4) 2 16 10 100 4

Mean =  X 56 / 7 8 12 144 16

n 14 196 36
SD 4 56 560 112
CV = 100  100 50%
Mean 8

03/13/25 18
Example: The following data represent the monthly consumptions (in thousand of
Rs.) of 60 students, selected from IM Sciences Peshawar. Calculate the average
consumption of students, variance, standard deviation and coefficient of variation
Consumption (X) 9.5 10 10.5 11 11.5 12 14 Total

Number of students (f) 9 10 17 10 5 4 5 60

fx 85.5 100 178.5 110 57.5 48 70 649.5

fx2 812.2 1000 1874.25 1210 661.25 576 980 7113.7

5 5

Mean =  fx  649.5 10.83

 f 60
2

Variance = 
fx   fx  7113.75  649.5 
2 2

     118.53
  
f f 60  60 
SD = Variance  118.53 10.89
SD 10.89
CV= 100  100 100.55%
Mean 10.83
03/13/25 19
Example: Using the following data showing the sales in thousand of Rupees
of 60 different shops in Peshawar city. Calculate the variance, standard
deviation and coefficient of variation.

Sale (000) 65-84 85-104 105-124 125-144 145-164 165-184 185-204 Total

Number of shops 9 10 17 10 5 4 5 60

TRY yourself

03/13/25 20
 Example: The following data shows the score obtained by two different
students (A & B) in five different subjects. State that which student has
shown a consistent performance, by calculating CV for each.

Student Student XA XA2 XB XB2

-A -B Mean-A = 415/6 = 69.17
60 3600 70 4900
60 70 SDA= 8.89
80 6400 75 5625
80 75
70 4900 65 4225 CVA = 12.85%
70 65
75 5625 70 4900
75 70
Mean-B = 415/6 = 69.17
55 70 55 3025 70 4900

75 65 75 5625 65 4225
SDB= 3.44
415 415 415 29175 415 28775 CVB = 4.97

Since CV of Student B is less than the CV of student A, so the performance

of Student B is consistent.
03/13/25 21
 Example: The following data shows life time of refrigerators (in months)
produced by two different companies (A & B). State that which company is
consistent as compared to other one, regarding the life time of refrigerators.

Company- XA XA2 XB XB2

Company-A B Mean-A = 415/6 = 69.17
60 3600 70 4900
60 70 SDA= 8.89
80 6400 75 5625
80 75
70 4900 65 4225 CVA = 12.85%
70 65
75 5625 70 4900
75 70
Mean-B = 415/6 = 69.17
55 70 55 3025 70 4900

75 65 75 5625 65 4225
SDB= 3.44
415 415 415 29175 415 28775 CVB = 4.97

Since CV of Company-B is less than the CV of Company-A, so the performance

of Company-B is consistent.
03/13/25 22
 SAMPLE VARIANCES

2
2  (X  X) 2

 X 

X 2

S =            (1)
n n  n 
It is biased estimator of population variance
  X 
2
2
s =  (X  X )
=
1
2
X  2         (2)
n 1 n  1 n 
 
It is called unbiased estimator of population of variance

Note: Most of the statistical packages uses equation (2) instead of equation (1) for
computing the sample variance.

03/13/25 23
 SKEWNESS

Any distribution having the bell shape curve (as mentioned belwo) is
called symmetrical distribution. In symmetrical distribution, mean,
median and mode are coincides (equal to each other). On the other
hand, any departure from symmetry is called skewness.

18 Bell shape curve

15
Number of Markets

12
9
6
3
0
0 4 8 12 16 20
Sale in thousand

03/13/25 24 24
 Skewness will either be positive or negative depending on the curve of
a distribution. If the curve is elongated towards right, skewness
will be positive, and mean > median > mode. On the other hand,
if the curve is elongated towards left, skewness is defined to be
negative and mean < median < mode.
30 30
Positive Skewness Negative Skewness
Number of Markets

Number of Markets
20 mean > median > mode 20 mean < median < mode

10 10

0 0
0 4 8 12 16 20 0 4 8 12 16 20
Sale in thousand Sale in thousand

03/13/25 25
 COEFFICIENT OF SKEWNESS

The numerical measurement of skewness is called coefficient of

skewness. The coefficient of skewness, denoted by SK, is
mathematically computed by using one of the following formulae:

3(mean  median)
SK= (Pearson's coefficient of skewness)
SD

Q3  2median  Q1
SK= (Bowley's coefficient of skewness)
Q3  Q1

03/13/25 26 26
 KURTOSIS

Kurtosis measure the flatness and peackedness of the curve of a

distribution. Kurtosis is measured by a ratio called the second
moment ration, denoted by β2. It is important to mention that the

first moment ratio denoted by β1 measures the skewness of

data. For a symmetrical distribution, β1 = 0.

On the other hand if:

β2 = 3, the distribution is called Mesokurtic (Normal)

β2 > 3, the distribution is called Leptokurtic

β < 3, the distribution is called Platykurtic

2
03/13/25 27 27
 30

25 Leptokurtic

20
Mesokurtic
15

10

5 Platykurtic

0
0 5 10 15 20 25 30

03/13/25 28