Measures of Dispersion

Dispersion:
The meaning of the dispersion is scatteredness. Dispersion is the degree of the scatter or
variation of the variables about a central value. It helps us in studying the important
characteristic of the distribution.
A measure of variation is designed to state the extent to which the individual measures
differ on an average from the mean.
Example:
Let the scores of two batches each of size 4, be as follows:

Batch I 49 50 50 51
Batch II 0 0 100 100

The average score for each batch is 50 the average respondents the first set of scores very
well but it is hardly typical of the second set. This is because the values in the first set
bunch around the average but the values in the second set are widely scattered. Thus
although the two sets have the same mean, they are markedly different in their variability
or dispersion. This means that although the two sets of data are quite different in nature,
the measure of location has failed to bring out this difference. The need for a measure of
dispersion in addition to a measure of location is thus obvious.
Properties of a good measure of dispersion:
A good measure of dispersion should possess the following properties:
i. It should be simple to understand.
ii. It should be easy to compute.
iii. It should be rigidly defined.
iv. It should be based on all observations of the distribution.
v. It should be amenable to further algebraic treatment.
vi. It should have sampling stability.
vii. It should not be unduly affected by extreme observations.
Different measures of Dispersion:
There are several methods of measuring dispersion. These measures can be divided into
two groups:

1. Absolute measure
2. Relative measure
Measures of Dispersion

Absolute measure:
Absolute measures of variation are expressed in the same statistical unit in which the
original data are given such as rupees, kilograms, tones etc. These values may be used to
compare the variation in two or more than two distributions provided the variables are
expressed in the same units and have almost the same average value.
Following are the absolute measures of variation or dispersion
i. Range
ii. Quartile Deviation
iii. Mean Deviation
iv. Standard Deviation
Relative Measures of Variation:
A measure of relative variation is the ratio of a measure of absolute variation to an
average. Relative measures may also be used to compare the relative accuracy of data.
Following are the relative measures of variation:
i. Co-efficient of range
ii. Co-efficient of quartile deviation
iii. Co-efficient of mean deviation
iv. Co-efficient of standard deviation
v. Co-efficient of variation
Range:
Range is the simplest method of studying variation. It is defined as the difference
between the value of the smallest observation and the value of the largest observation
included in the distribution. Symbolically,
R  L-S
Where, L  Largest value and S  Smallest value
The relative measures corresponding to range, called the co-efficient of range, is obtained
by applying the following formula
LS
Co-efficient of Range =
LS
In a frequency distribution, range is calculated by taking the difference between the lower
limit of the lower class and the upper limit of the highest class.
Example:
The following are the points of shares of a company from Monday to Saturday:
Day Prices (Tk.)
Monday 200
Tuesday 210
Wednesday 208
Thursday 160
Friday 220
Saturday 250
Calculate range and co-efficient of range.

2
Measures of Dispersion

Solution:
We know that, Range R  L - S
Here, L  250 and S  160
Range = 250-160 = Tk. 90
L  S 250  160
Co-efficient of Range = = = 0.219
L  S 250  160
Example:
Calculate coefficient of range and range from the following data:
Profits (Tk. lakhs) No. of Companies.
10-20 8
20-30 10
30-40 12
40-50 8
50-60 4
Solution:
In a frequency distribution, range is calculated by taking the difference between the lower
limit of the lower class and the upper limit of the highest class.
Range = L - S = 60-10 =50
L  S 60  10 50
Co-efficient of Range = = = = 0.714
L  S 60  10 70
Example:
The following are the wages of 8 workers of a factory. Find the range of variation and
also compute the co-efficient of range.
Wages in Tk.
1400 1450 1520 1380 1485 1495 1575 1440
Solution:
Range  L - S
Where, L  Largest value and S  Smallest value
Hence, L  1575 and S  1380
Range  1575-1380 =Tk.195
L  S 1575  1380 195
Co-efficient of Range = = = = 0.065
L  S 15755  1380 2955
Example:
The following are the marks of 80 students of a class. Find the range of variation of
marks and also compute co-efficient of range.
Marks 0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80
No. of
4 12 20 18 15 8 2 1
Students

3
Measures of Dispersion

Solution:
Range  L - S
Here, L=80, S=0
Range = 80-0
= 80
L  S 80  0
Co-efficient of Range = = =1
L  S 80  0
Merits:
i. It is the simplest measure of dispersion.
ii. It is easy to calculate and easy to understand.
iii. It gives us a quick idea of the variability of the observations involving
least amount of time and calculations.
Limitations:
i. Range is not based on each and every observation of the distribution.
ii. Range cannot be computed in case of open-end distributions.
iii. Range cannot tell us extreme observations. For example, observe the
following series:
Series A 6 46 46 46 46 46 46 46
Series B 6 6 6 6 46 46 46 46
Series C 6 10 15 25 30 32 40 46

In all the series range is the same (i.e. 46-6 = 40) but it doesn’t mean that the
distributions are alike. Range is, therefore, more unreliable as a guide to the
variation of the values within a distribution.
Uses of Range:
Despite serious limitations range is useful in the following cases:
 Quality control
The object of quality control is to keep a check on the quality of the product without
100% inspection.
 Fluctuations in the share prices
Range is useful in studying the variations in the price of stocks and shares and other
commodities etc. They are very sensitive to price changes from one period to another. For
example, by computing range we can get an idea about the range of variation of, say,
gold prices. If the minimum price for 10 gm during 1989-90 was Tk. 3010 and the
maximum price Tk. 3350 this at once tell us about the range of variation
i.e. 3350-3010 = 340.

4
Measures of Dispersion

 Weather forecasts
The meteorological department does make use of the range in determining the difference
between the minimum temperature and maximum temperature. This information is of
great concern to the general public because they know as to within what limits the
temperature is likely to vary on a particular day.
Mean deviation:
Mean deviation is obtained by calculating the absolute deviations of each observation
from mean (or median or mode) and then averaging these deviations by taking their
arithmetic mean.
Let X1 , X 2 ,...,X n be n observations of a variable with mean ( x ), median ( M e ) and Mode
( M o ) then mean deviation is defined by:
For Ungrouped Data:
1
M .D x  =
n
 xx
1
M .D( Me ) =
n
 x  Me
1
M .D( Mo ) =
n
 x  Mo
For grouped Data:
M .D x  = ∑f | ̅|

M .D( Me ) = ∑ f | |

M .D( Mo ) = ∑ f | |

Where, N= ∑ f
Co-efficient of Mean deviation (C.M.D):
Co-efficient of mean deviation is the ratio of the mean deviation measured from certain
measure of central location to the corresponding measure of central location and is
defined as follows:
M .D x
C.M .D x  =
x
M .D( M e )
C.M .D M e  =
Me
M .D( M o )
C.M .D M o  =
Mo

5
Measures of Dispersion

Example:
Calculate the mean deviation from
(a) Arithmetic mean
(b) Mode
(c) Median
In respect of the marks obtained by nine students given below and show that the mean
deviation from median is minimum.
Marks
7 4 10 9 15 12 7 9 7
(Out of 25 )
Solution:
Calculation of mean deviation from mean, median and mode:
We know that,
1
M .D x  =  x  x
n
1
M .D( Me ) =
n
 x  Me
1
M .D( Mo ) =
n
 x  Mo
Now,

Mean x =
x 80 i
=
= 8.89
n 9
For calculating median the items have to be arranged

Marks
4 7 7 7 9 10 10 12 15
(Out of 25 )

n 1 9 1
Median = size of th item = = 5th item
2 2
Here, Size of 5th item = 9.
Hence, median=9.
Mode = 7 (since 7 is repeated the maximum number of items i.e. 3)
Calculation of Mean Deviation:

Marks Deviations from Deviations from Deviations from

x Mean = x i  x Median = xi  Me Mode = xi  Mo

7 1.89 2 0
4 4.89 5 3
10 1.11 1 3

6
Measures of Dispersion

9 0.11 0 2
15 6.11 6 8
12 3.11 3 5
7 1.89 2 0
9 0.11 0 2
7 1.89 2 0

Total  x  x = 21.11
i  x M
i e = 21  x M
i o = 23

Now we have,

1 1
M .D x  =
n
 x  x =  21.11 = 2.34
9
1 1
M .D( Me ) =  x  M e =  21 = 2.33
n 9
1 1
M .D( Mo ) =  x  M o =  23 = 2.56
n 9
From these calculations it is clear that the mean deviation is least from median.
For grouped Data:

M .D x  = ∑f | ̅|

M .D( Me ) = ∑ f | |

M .D( Mo ) = ∑ f | |
Where, N= ∑ f
Example:
Calculate mean deviation (taking deviations from mean) from the following data:
x 2 4 6 8 10
f 1 4 6 4 1
Solution:
We know that,
M .D x  = ∑f | ̅|
Now,

̅=

7
Measures of Dispersion

By using calculator we get,  fx = 96 and N= ∑ f = 16.

96
x= = =6
16
Calculation of mean deviation
Deviations from
x f f xx
Mean = x  x
2 1 4 4
4 4 2 8
6 6 0 0
8 4 2 8
10 1 4 4
Total f x  x = 24

 M .D x  = ∑f | ̅|
1
=  24
16
= 1.5
Example:
Calculate mean deviation (from mean) from the following data:
Size of the item 3-4 4-5 5-6 6-7 7-8 8-9 9-10
Frequency 3 7 22 60 85 32 8
Solution:
We know that,
M .D x  = ∑f | ̅|
Calculation of mean deviation from mean
Size of Frequency Mid point Deviations from f xx
the item f x Mean = x  x
3-4 3 3.5 3.59 10.77
4-5 7 4.5 2.59 18.13
5-6 22 5.5 1.59 34.98
6-7 60 6.5 0.59 35.4
7-8 85 7.5 0.41 34.85
8-9 32 8.5 1.41 45.12
9-10 8 9.5 2.41 19.28
Total N= ∑ f = 217 f xi  x
=198.53

8
Measures of Dispersion

Now we have,

1538 .5
x= = = 7.09
217
 M .D x  = ∑f | ̅|
1
=  198.53 = 0.9148 = 0.915
217
Example:
Age distribution of hundred life insurance policy holders is as follows:
Age as on nearest birthday Number
17-20 9
20-26 16
26-36 12
36-41 26
41-51 14
51-56 12
56-61 6
61-71 5
Calculate mean deviation from median age.
Solution:
We know that,
M .D( Me ) = ∑ f | |
Calculation of mean deviation
Age as on Number c. f Mid point Deviations from f x  Me
nearest birthday f x Median = x  M e
17-20 9 9 18.5 19.75 177.75
20-26 16 25 23 15.25 244.00
26-36 12 37 31 7.25 87.00
36-41 26 63 38.5 0.25 6.50
41-51 14 77 46 7.75 108.50
51-56 12 89 53.5 15.25 183.00
56-61 6 95 58.5 20.25 121.50
61-71 5 100 66 27.75 138.75
Total f i f x  Me
=100 =1067

9
Measures of Dispersion

We know that,
N
 p.c. f
Median = L  2 i
f
N 100
Here, = = 50
2 2
Here median class is 36-41.

 L  36 , p.c. f  37 , f  26 , i  5
So we have,
50  37
Median = 36   5 = 38.25
26
1
 M .D( Me ) = ∑ f | |=  1067 = 10.67
100
Merits:
i. It is easy to understand.
ii. It is relatively easy to calculate.
iii. It takes all the observations into account.
iv. It is less affected by the extreme values.
Demerits:
i. It is not amenable to further algebraic treatment.
ii. It cannot be calculated if the extreme classes of the frequency distribution
are open.
iii. It is less stable than standard deviation.
Uses:
Because of its simplicity in meaning and computation it is specially effective in reports
presented to the general public or to groups not familiar with statistical methods. This
measure is useful for small samples with no elaborately analysis required. Incidentally it
may be mentioned that the National Bureau of Economic Research has found in its work
on forecasting business cycles, that the mean deviation in the most practical measure of
variation to use for this purpose.

Standard deviation:
Standard deviation may be defined as the positive square root of the arithmetic mean of
the squares of deviations of given observations from their arithmetic mean.
For ungrouped data
Let x1 , x2 ,..., xn denote n values of a variable x . The standard deviation denoted by S x is
defined as

10
Measures of Dispersion

 x 2  x 2 
= Sx 
 ( x  x) 2

=       =
 x


2
2


 x 
n  n   n 
 n     

Where, x 
x .
n
Example:
Find the standard deviation from the weekly wages of ten workers working in a factory:
Workers wages workers wages
A 320 F 340
B 310 G 325
C 315 H 321
D 322 I 320
E 326 J 331
Solution:
We know that,
 x 2  x 2 
=      
 n 
 n   
By using calculator we get,
 x 2  1043912,  x =3230 and n  10 .
So that,

1043912  3230  
2

Sx      = 7.886 = 7.89
 10
  10   
For grouped data:
If x1 , x2 ,..., xn occurs with frequency f1 , f 2 ,..., f n respectively, the standard deviation is
defined as

 f  x  x
2

  fx   fx  
2 2
  fx 2 2

= Sx  =   

  =  
 x 
N 

N  N     N 
Where N   f and x is the arithmetic mean of the distribution.

11
Measures of Dispersion

Example:
An analysis of production rejects resulted in the following figures:
No. of rejects per operator No. of operators
21-25 5
26-30 15
31-35 28
36-40 42
41-45 15
46-50 12
51-55 3
Solution:
We know that

  fx   fx  
2 2

    
 N  N  
 
Where, N   f .
Calculation of Standard Deviation
No. of rejects No. of operators Mid point
x2 fx fx 2
per operator f x
21-25 5 23 529 115 2645
26-30 15 28 784 420 11760
31-35 28 33 1089 924 30492
36-40 42 38 1444 1596 60648
41-45 15 43 1849 645 27735
46-50 12 48 2304 576 27648
51-55 3 53 2809 159 8427
Total 120 10808 4435 169355
From the data by using calculator we get,
 fx  4435 , N   f =120 and  fx 2
 169355

169355  4435  2 
 =    
 120  120  
= 6.7359
= 6.74

12
Measures of Dispersion

Merits and Limitations of Standard Deviation:

Merits:
i. The standard deviation is the best measure of variation because of its
mathematical characteristics. It is based on every item the distribution.
ii. It is possible to calculate the combined standard deviation of two or more
groups. This is not possible with any other measure.
iii. For comparing the variability of two or more distributions coefficient of
variation is considered to be most appropriate and this measure is based on
mean and standard deviation.
iv. Standard deviation is most prominently used in further statistical work.
For example, in comparing skewness, correlation etc., and use in made of
standard deviation.
Limitation:
i. As compared to other measures it is difficult to compute.
ii. It gives more weight to extreme values and less to those which are near the
mean.

Co-efficient of Standard Deviation:

Co-efficient of standard deviation is defined by
S .D  s
C.S .D   =
Mean x x

Co-efficient of Variation:
The relative measure of dispersion based upon standard deviation is called co-efficient of
standard deviation. The co-efficient of standard deviation multiplied by 100 gives the co-
efficient of variation.
Thus
Standard deviation
Co-efficient of variation (C.V) = 100
Mean

=  100
x
Where,  and x are both measured in the same units.
Note:
Co-efficient variation is more useful when the two distributions are entirely different and
the units of measurement are also different. Co-efficient of variation being a pure number
is independent of the units of measurement and thus is suitable for comparing the
variability, homogeneity and uniformly of two or more distributions. The series having

13
Measures of Dispersion

greater C.V is said to be more variable than the other and the series having lesser C.V is
said to be more consistent (or homogenous) than the other. Co-efficient of variation is,
however unreliable if x is near to zero.

Example:
Calculate co-efficient of variation from the following data:
Profits (Rs. Cr.) 10-20 20-30 30-40 40-50 50-60
No. of Companies 8 12 20 6 4
Solution:
We know that,

C.V 

 100 and x 
 fx
x N
 fx 2  
    fx
2
  
  
 
 N  N  
Calculation of coefficient of variation
Profits (Tk. crores) Midpoint  x  No. of cost  f  x2 fx fx 2

10-20 15 8 225 120 1800

20-30 25 12 625 300 7500
30-40 35 20 1225 700 24500
40-50 45 6 2025 270 12150
50-60 55 4 3025 220 12100
Total 50 7125 1610 58050
By using calculator we get
 fx = 1610,  fx 2 = 58050 and N   f = 50. 

x
 fx  1610 = 32.2
N 50
 fx 2  
  fx
2
   58050  1610  2 
     =     = 11.14
 N   50  50  
 N   
 11.14
Co-efficient of variation  100 =  100 = 34.596 = 34.6%
x 32.2

14
Measures of Dispersion

Example:
A purchasing agent obtained samples of 60 watt bulbs from two companies. He had the
samples tested in his own laboratory for length of life with the following result:
Length of life (in hours) Company A Company B
1700-1900 10 3
1900-2100 16 40
2100-2300 20 12
2300-2500 8 3
2500-2700 6 2
a) Which company’s bulbs do you think are better?
b) If prices of both types are the same, which company’s bulbs would you buy?
Solution:
Calculation of Mean and co-efficient of variation
Length of life (in hours) Midpoint Company A Company B
1700-1900 1800 10 3
1900-2100 2000 16 40
2100-2300 2200 20 12
2300-2500 2400 8 3
2500-2700 2600 6 2
For Company A:
By using calculator we get
 fx =128800,  fx 2 =279840000 and N   f = 60
x 
 fx  128800  2146 .67 and
N 60
 fx 2  
    fx
2
    279840000  128800  2 
  =     = 236.267 = 236.27
 N    60  
 N    60

 236 .27
Co-efficient of variation  100 =  100 = 11.00 = 11%
x 2146 .67
For Company B:
By using calculator we get
 fx =124200,  fx 2
=258600000 and N   f = 60

x 
 fx  124200  2070 and
N 60
 fx 2  
    fx
2
    258600000  124200  2 
  =     = 158.429 = 158.43
 N    60  
 N    60

15
Measures of Dispersion

 158 .43
Co-efficient of variation  100 =  100 = 7.65%
x 2070

(a) Since average is higher in case of company A, hence bulbs of company A are
better.

(b) Co-efficient of variation is less for company B. Hence if prices are same we will
prefer to buy Company’s B’s bulbs.
Example:
The prices of a Tea company shares in Dhaka and Chattogram Markets during the least
ten months are recorded below:
Month Dhaka Chattogram
January 105 108
February 120 117
March 115 120
April 118 130
May 130 100
June 127 125
July 109 125
August 110 120
September 104 110
October 112 135
Determine the arithmetic mean and standard deviation of the prices of shares. In which
market are the shares prices stable?
Solution:
For determining in which market prices of shares are more stable we shall compare the
co-efficient of variation. Let prices in Dhaka and Chattogram are denoted by X and Y
respectively:
Dhaka:

X 
 X  1150  115
N 10
 X 2  X  2 
    
   =8.33

 N  N  
 8.33
C.V   100 =  100 = 7.24%
X 115

16
Measures of Dispersion

Chattogram:

X 
 X

1190
 119
N 10

 X 2  X  2 
    
   =10.089 = 10.09

 N  N  

 10.09
C.V   100 =  100 = 8.478 = 8.48%
X 119

Since the co-efficient of variation is less in Dhaka, hence the share price in the Dhaka
market shows greater stability.

17