QUANTITATIVE ANALYSIS FOR BUSINESS

360 DECISIONS -|
between diagrams &graphs?
38. Which are thetwo differences
Oct. 89,
Oct. 98 &Apr. 2000 CHAPTER 4
39. What are the utilities of diagrams?
Oct. 91
graphs? MEASURES OF DISPERSION AND SKEWNESS
40. Compare diagrams & Ap. 92
41. What are the advantages of diagrammatic presentation of data? The various measures off central tendency give us one single figure,
the entire data. But the averages alone cannct adequately
that represents
Nov. 92 describe a set of observations, unless all the observations are the same. It
necessary to describe the variability or dispersion of the observations. It
42. State the generalrules for constructing diagrams? Ap. 93
is
ia alwavs found, that in two or more distributions, the central value may be
43. What do youmean bya bar diagram? What are it's special the same., but still there can be wide disparities in the formation of the
merits? distribution. Measures of variation help us in studying the important
characteristic of a distribution, that is the extent to which the itens vary
Nov. 93
from one another and from the central value.
44. What is a bar diagram? Nov. 94 Kafka defines measures of variation as, "the measurement of the
Scatterness of the mass of fiqures in a series about an average, is called
45. When an ogive curve is given, how do you locate median? Nov. 94 measure of variation or dispersion". So it is ciear from the explanation that
dispersion is the spread or scatter of item values from a measure of central
46. Distinguish between diagrams andgraphs? Ap. 97
tendency.
47. What are less than and more than ogives? Ap. 96 The significance of measures of variation can be illustrated with the
help of the following example.
48. What is meant by ogive? What are its uses? Oct. 98
Series A Series B Series C

100 100 1

100 105 489

100 102 2

100 103 3

100 90 5
Total |500 500 500

Mean 100 100 100

values differ
arithmetic mean is the same in all cases, but the
widely tromHere,onetheanother. In seriesA, each and veryitemis perfecty represented
by the arithmetic mean, or in other words,, none of the items of series Adeviates
Irom the In series B, the

tems arithmetic
asingleVary,
itembut
isthe
mean and hence there is no dispersion.
variation isbyverytheSmall
repreesented compared
arithmetic
seriesC. in series C, not
meantoand theitems vary widely
Irom one B.In measuring greatercompared
muchamount
to series
another.
only and not t in
In otherwords,
dispersion, We are
in series C, interested is the
dispersionin of variation

the direction of variation.

 FOR BUSINESS MEASURES OF DISPERSION AND SKEWNESS 363
QUANTITATIVE ANALYSIS DECISIONS - |
362
6 It should have sampling stability, and
Variation
Objectives of Measuring Itshould not be affected by extreme values.
Measures of variation are neededI for four basic purposes. 7.

Absolute and Relative Measures ofVariation

an average
To measure exactly the reliability of
1. Measures of dispersion may be either
Measures of variation pOint out as to how far an averans. Absolute, or
representative of the mass. Whenthe dispersion is small, the average is a 1.
represents the individualvotSa
typical value, in the sense that it closely 2.
Relative
estimate of the averane i
it is reliable in the sense that, it IS agood Absolute measures of disperSIon are expressed in the same
corresponding universe. ieical unit in which the orginal data are given, such as rupees, kilograms,
two
On the other hand, when dispersion is large, the average is not so tonnes, etc. These variables may be used to compare the variation in
and are
typical and unless the sample is very large, the average may be
ouite distributions provided the variables are expressed in the same units
range, mean
unreliable. of the same average size. or eXample, range, inter quartile
To serve as a basis for the control of variability deviation, standarddeviation, etc. are absolute measures of dispersion.
2.
If the two sets of data are expressed in different units, Such as
In the matters of health, variations in body temperature, to measure
quintals of sugar versus tonnes of sugar cane, or if the averages size is very
the inequality of income, inequality of wealth, etc., measure of dispersion is different, such as manager's salary versus worker's wages, in such cases,
Used to find out the degree of variation and therefore, it serves as a basis for
the control of variability. relative measures of dispersion should be used. Examples are, coefficient
variabililty of mean deviation, coefficient of quartile deviation, coefficient of variation,
3. Tocompare two or more series with regard to their etc., are relative measures of dispersion.
Measures of dispersion enable a comparison to be made of twoof Ameasure of relative dispersion, is the ratio of a measure of absolute
more series, with regard to their variability. The study of variation may also dispersion to an average. Someasures of variation are usually measured as
be lookedupon asa means of determining uniformity or consistency. Ahign an average of deviations, about some central value. Arithmetic mean is
degree of variation would mean little uniformity or consistency, whereas, a used in calculating mean deviation and standard deviation. Measures of
low degree of variation would mean great unifornity or consistency. dispersion is thus a type of average, and therefore it is also known as a
4 To facilitate the use ofother statistical measures second order average.
correlation
Many powerful analytical tools in statistics, such as are nelative measure of dispersion is also called a coefficient ot
etc.,
analysis, the testing of hypothesis and analysis of time series,measureof dispersion, because "coefficient means a pure number that is independent
of the unit offmeasurement". It is also to be remembered that while computing
based on measures of variation in some way or other. Therefore
variation is used in the calculationof other statistical measures. the relative dispersion, the average used as a base should be the same one,
from which the
Properties of a Good Measure ofVariation possible absolute deviations are measured.
Methods of StudyingVariation
Agood measure ol dispersion should possess as far as
the following properties. The following are theimportant methods of studying variation.
1.
1
It should be simple to
understand. The Range
2 2
It should be easy to The Quartile Deviation
3 It should be well compute. 3
The Mean Deviation
defined. distribution 4
4. The Standard Deviation, and
It should be
based on each andIeveryitem of the 5
5. It should be treatment. The Lorenz Cunve.
capable off1further algebraic
 364 QUANTITATIVE ANALYSIS FOR BUSINESS DECISIONS MEASURES OF DISPERSION AND SKEWNESS 365

Ofthese, first four are mathematical methods and the last is a Discrete Series
one. Accordingto the syllabus
Calculating Range for
first four methods only.
requirements, we have to studygraphithec For calculting
range for discreteseries, frequency column isignored.
values of the Xvariable only. This is explained
calculationis based on the
1. RANGE The tollowing example.
with the help of the
78 96 100 150 230 240 261
It is the simplest measurement of dispersion. The 64
the largest item and the smallest item is known as range. difference
between X 43
105 91 82 61 32 70 88 67
f 100
It is defined as "the difference between the two extreme items of tha
distribution". Inother words, range is the difference between the largest and is based on the X values only. In
As explained earlier, calCulation smallest item is 43
smallest items of the distribution. is 261, and the
this range the largest item
R = L-S = L- S
Range
L =Largest item = 261 - 43 = 218
S = Smallest item L - S
Coefficient of Range =

L +S
The relative measure corresponding to range is called the coefficient 261 - 43
of range, which is obtained by applying the formula.
261 + 43
Coeficient of Aange L+ S 218
a. Calculating Range for Individual Observations 304
= 0.7171
rom the given values, the difference between the largest value and
the smallest value willgive the value of range. It is explained with the help of
the following example. C. Calculating Range for Continuous Series
Problem range is calculated by taking the
In the Continuous series. the lowest class and upper limit of the
Find out the range and coefficient of range from the folowing Uiierence between the lower limit of the the following example.
of
Values of X- 42, 47, 54,64, 72, 94, 96, 45. gnest class. This method is explained with the help
Here L= 96 and
Problem
the following data.
S= 42 balculate Range and coefficient of range from
So R= L-S =96 - 42 Number ofWorkers
Wages (in Rs)
= 54 8
10-20
Coefficient of Range = 20-30
10
Subsituting the values in the eguation, we get 30-40 12
Coefficient ot Range =96 - 42 54 = 0.3913
40-50
8
96 + 42 138
4
50-60
 QUANTITATIVE ANALYSIS FOR BUSINESS MEASURES OF DISPERSION AND SKEWNESS
366 DECISIONS -| 367

lowest class = 10 and upper limit of the highest Range is used in industry for the statistical quality control of the
Here lower limit of the
class = 60
manufactured products by the inspection at regular intervals. For
example, when a machine is turning out large number of a particular
60- 10 = 50
So range = L- S= component, a small sample of four or five is taken at half an interval
L - S examine whetherthe process remains constant within the limits
Coeficient of Range = of error.
L +S
EINTERQUARTILE RANGE ANDTHE QUARTILE DEVIATION
Substituting the values in the equation,we get 2. THE
60- 10 Interquartilerange includes the middle 50 percent of the distribution.
of the observations at the lower end and another quarter
Coefficient of Range = 60+10 That is one quarter
at the upper end of the distribution are excluded in
of the observations
50
= 0.7143 computing the interquartile range.
70 In other words, interquartile range represents, the difference between
Merits of Range the third quartile and the first quartile. Symbolicaly.
1. Range is simple to understand. where Q.D. Quartile deviation
Q.D.
2 Range is easy to calculate.
3 Ifone is interested in getting a quick rather than a very acCurate Quartile deviation gives the amount by which the two quartiles differ
picture of variability one may compute range. from the median. In a symmetrical distribution, the two quartiles Q, and Q,
Limitations of Range will be egquidistant from the median. But nearly all distributions of social
1. Range is not based on each and every item of the distribution. Science series are asymmetrical. In an asymmetrical distribution Q,and Q,
are not equidistant from the median. As a result, an asymmetrical distribution
2 Range iS affected by extreme values in the series, because it takes indludes only approximately 50 percent of the observations.
into account only the largest and smallest items.
When quartile deviation is very small, it describes high uniformity or
3. Range cannot be calculated for open-ended classes. Smallvarition of the central 50 per cent items and a high quartile deviation
4. The valuethe of means that the variation among the central items is large.
Range is highly affected by fluctuations of sampling.increases,
range is never stable. As the size of the sample Quartile deviation is an absolute measure of dispersion. The relative
range tends to increase though not in the same ratio. ure Corresponding to this measure is called the coefficient or quartile
5.
Calculation of rrange does not take into account the Composition
of deviation and is calculated as follows,
the series. The calculation is done using the smallest andlargest Q, -Q,
items of the series. So the value of range is considerably changed, Coefficient of Quartile Deviation.=
if any item is added or deleted on either side of extreme valu. Q, + Q,
Uses of Range Coefficient
Ofvariation in of iquartile deviation can be used to Compare the degree
1
Range is prices of
shares
a. different distributions.
and stock, useful in
gold, studying
the variations in the
Computation of Quartile Deviation for rIndividual Observations
2 In weather jewellery etC. difference For computing quartile deviation for individual observations, it is

ornecessary
der. that the data should be arrangeddin
forecasts also, range is the or descending
between the used to determine to varyon
likely t the ascending
a particular day.
minimum and maximum temperature The Q, - Q,
iormula used for Quartile Deviation = 2
 ANALYSIS FOR BUSINESS MEASURES OF DISPERSION AND SKEWNESS
368
QUANTITATIVE DECISIONS -| 369

Cimilarlv, the formula to calculate coefficient of quartile deviation

Interquartile range = 0, -Q,
is,
=
Coefficient of quartile deviation 0, +Q, Coefficient of Q.D. =

The calculations are explained and made Clear with the hel. Substituting the values in the equation, we get
following examples. 40 - 15
Coefficient of Q.D. = 25 = 0.4545
PROBLEM 40 +15 55
Runs scroed by a batsman in acricket match - 20, 28, 40, 12 20 4e PROBLEM

SOLUTION Using the data given below, calculate

As the first step,data should be arranged in the ascending order. a Interquartile range
that is 12, 15, 20, 28, 30, 40, 50. Quartile deviation and

Now we have to calculate Q,and Qa C. Coefficient of Quartile deviation

Q, is calculated using the formula term 8,10, 14, 22, 26, 28, 30, 39, 44, 59, 64
N= the number of observations SOLUTION
The data is given in the ascending order. So in order to calculate Q,.
=2nd term = 15 runs
The formula is term = term
Similarly Q, can be calculated using the formula.
= 3rd term = 14
= term
Similarly Q,can be calculated using the equation
= 3x2 = 6h term = 40 runs term

Now,quartile deviation can be calculated using the equation 3 x3 9h term = 44

Q.D.
2
Inerquartle range can be calculated using the equation Q,-a,
Substituting the values, we get 44 - 14=30
Substituting the values in the equation, we ger
40 - 15
Iherefore, interquartile range = 30
Q.D.
2
Quartile deviation is calculated using the formula, Q.D=
44-14
25
12.5 runs
C Substituting the values, We get Q.D, =
2
30
2
= 15
2
CoeffiQcient of quartile deviation can be calculated using the equation,
 BUSINESS MEASURES OF DISPERSION AND
370
QUANTITATIVE ANALYSIS FOR DECISIONS -| SKEWNESS 371

equation, we aet 43 + 1 = 11" term = 20

Substituting the values in thne 4
44 - 14
Coeficient of Q.D. = 44 + 14 20
So Q, = Rs.
30 =0.517
58 Similarly Q, is calculated using the equation 3 term

b. Computation of
Quartile Deviation for Discrete Series.
= 3x11 =33d term = 40
The formula to be used is the same i.e.,
Coefficient of Quartile Deviation = So Q, = Rs. 40
Q.D. =
2
Q,+Q, NowQ.D. can be calculated using the equation
The following examples clearly explains the calculation
Q.D.
PROBLEM 2
Substituting the values in the equation, we get
Compute Quartile Deviation and Coefficient of Quartile Deviation 40- 20 20
= Rs. 10
using the following data. Q.D.
2
Coefficient of quartile deviation can be calculated using the equation,
Wages in Rs. 10 20 30 40 50 60
i.e.,
Number of Workers 4 7 15 8 7 2

Substituting the values in the equation, we get

SOLUTION
40 20 20
Q, and Q, calculation for discrete series is using the cumulatve 40 + 20 60
frequency, that is explained below. = 0.33

Wages in Rs. Number of Workers C.f. PROBLEM

4
10
20
11 From the following data, calculate
7

30 15
26
b Interquartile range
40
34 Semiiinterquartile range (Quartile deviation), and
C.
50 7
41

43
Coeficient of Quartile deviation of the marks of students
60 2
Marks 40 45 52 53 54 55 56 57 68 70

th Number of
Q,is calculated using the
equation Q, =
term
Students 10 12 14 10 11 12 11

Where N= frequency total

Substituting the values in the equation, we ge
 ANALYSIS FOR BUSINESS MEASURES OF
372
QUANTITATIVE DECISIONS -| DISPERSION AND SKEWNESS 373

SOLUTION Computation of Quartile Deviation for the Continuous Series

c. For computing quartile deviation for the continuOus series, the
f
Marks C.f same. The method is explained with the help of sone
10 formulas are the
40 10
examples.
45 12

52 14 36 PROBLEM

53 9 46 Compute quartile deviation and coefficient of fquartile deviation from

54 10 55 the following data.
55 11 66
56 12 78 Marks of Students 10-20 20-3030-40 40-5050-60 60-7070-80
57 11 89 Number of Students 12 19 5 10 6 6
68 o 98
SOLUTION
70 8 106
X f C.f

10-20 12 12
Q, is calculated using the equation,
20-30 19 31
Q, = term, where N = frequency total
30-40 5 36
Substituting the values in the equation,we get 40-50 10 46
106 + J th
50-60 55
26.75hterm
60-70 6 61

52 Marks 70-80 6 67
Similarly Q, can be calculated using the equation
Q, class can be calculated using the equation term
O, = 3 term
Where N -0 frequency total
= 3x 26.75 = 80.25th term = 57 Marks 4
So N
Q, = 4
=6.75h term
Interquartile range is calculated using the equation,
a, -Q, =57 - 52 = 5 marks
b
Semi interquartile range or Quartile deviation isScalculated
usingthe Remember that it is only
equation.
57 - 52
Now Q, class is 20 - 30
) 2 . 5 Marks ,can be calculated using the formula
C formula
Coefficient of Quartile deviation is calculated using the Q, = L+ N
, -Q,57 - 52 - c.f X0
0.0458 4
, +Q, 57 +52 109
 BUSINESS MEASURES OF
QUANTITATIVE ANALYSIS FOR DECISIONS -| DISPERSION AND SKEWNESS 375
374

c.f = cumulative frequency of the preceeding class,,i.e., the Where L= Lower limit of the Q, class
class preceeding the quartle class. PROBLEM
f = simple frequency of the quartile class Calculate the
class interval of the quartile class. interquartile range
Substituting the values in the equation we get.
a.

16.75-12 semi interquartile range (quartile deviation), and

20 + x 10
Q, = 19 C.
coefficient of quartile deviation
20+ 2.5 = 22.5 marks 0-10| 10-20 20-30 30-40 40-50 50-60 60-70
=
Marks
th 6
term i.e., 3 x 16.75 = 50.25h term, So Number of Students 16 22 30 24 12
Now Q, class is 3x
0, lies in 50-60 class. (the corresponding cumulative frequency falls against SOLUTION

this class). Marks (X) f C.f

Q, can be calculated using the equation. 0-10 8 8

10-20 16 24
- c.f
20-30 22 46
Q, =L+3 X0
f 30-40 30 76
Substituting the values in the equation,we get 40-50 24 100
50.25-46 50-60 12 112
Q, =50+ 9
x 10
60-70 6 118
=50 + 4.722
= 54.722 marks Q, class is term

a. Quartile deviation can now be calculated using the equation - 29.5th term
Q.D. = , -Q, 54.72-22.5 Inerefore 20-30 is the Q. class. Q. is calculated using the formula
2 2 N
-c.f
32.22
= 16.11 marks =L+ f
xi

b the equaion Substituting the values in the equation, we

Coefficient of quartile deviation can be calculated using ge
29.5- 24
,-Q, = 20 + x10
, Substituting the values, we get
9,+Q, = 20 + 2.5 = 22.5
marks
54.72 -22.5
54.72+225
32.22
77.22
0.4172 Similarly03 class is 3x term
=3 x 29.5 = 88. 5th term.
 e 0 |ANTITATIVE ANALYSIS FOR BUSINESS DECISIONS
MEASURES OF DISPERSION AND SKEVWNESS 377

So Q, class is 40-50. Q, can be calculated using the equation; measuring dispersion. An ideal method shouid be based on each and
everyitem.
argument against quartile deviation is that, it is not ameasure
2. Another as it does notshowthe scatter around an averages, rather
x0 of dispersion
f is a distance
on a scale. This means that quartile deviation is not itself
Substituting the values in the equation, we get measured from an average. But it is a positional average. Hence, some
economists and statisticians are of the viewthat quartile deviation is a
measure of partition rather than a measure of dispersion.
88.5 -76
x 10
a, =40 + 24 treatment.
oOuartile deiviation is not amenable to further algebraic both ends.
hecause it ignores the first and last 25 percent items on
= 40 + 5.21
values of central items.
AQuartile deviation is.basically dependent on the affected adversely.
= 45.21 marks
If these values are irregular, then the results are
fluctuations.
a Interquartile range =Q, -Q, 5. Quartile deviation is highly affected by sampling
the distribution of items
= 45.21 - 22.5 6. Lastly quartile deviation is not affected by
the quartiles. That
= 22.71 marks
between Q, and Q, or by the distribution outside
be the same for two dissimilar
Semi interquartile range or Quartile deviation means, the values of the quartiles may
b
series.
45.21 - 22.5 DEVIATION
=11.36 marks 3. THE AVERAGE DEVIATION OR MEAN
2 2
Range and
methods of dispersion discussed above namely
TWo strict sense of the
Coefficient of Quartile Deviation Quartile Deviation are not measures of dispersion in the
C

scatterness around an average. Ihe tWo

Ierm because, they do not show the Deviation
the Average Deviation and the Standard
Oner measures namely
Q, +Q, help us in achieving this goal.
Substitutingthe values, we get, called the Mean Deviation. It is
ne average deviation is sometimes distribution and the median or
45.21 - 22.5 22.71 Ihe average difference between the items in a
Coefficient of Quartile deviation = the deviations
45.21 +22.5 67.71 al orthat series. Theoretically. there is an advantage taking
of items from median is
mie:iedlan, because the sum of the deviations
=0.335
minimun when signs are ignored. However in practice, the arthmetic mean is
Merits of Quartile Deviation deviation and this
morefrequently calculatingthe value of average
it isinmore commonly called mean deviation.
is the reason whyused
1 It is based on 50 percent of the observations.
series also, a. Computation of Mean deviation for Individual Observations
2
Quartile deviation can be claculated for open ended
because Q, and Q, are positional averages. Theeformula for mean deviation for individual
observationsis
3. The quartile deviation is not affected by the extreme items. computing
Limatations of Quartile Deviation M.D.
percent
1. 25
Quartile the first dependon
and last 25deviation ignores 50 percent items i.e., not methodof
percent. The
every item of the series. value of quartile deviation doesgood
So it cannot be regarded as a
 BUSINESS
QUANTITATIVE ANALYSIS FOR DECISIONS -| MEASURES OF DISPERSION AND SKEWNESS 379
378

ignoring signs
Where IDl denotes deviations trom median SOLUTION
Computation of Mean Deviation for the First Set ofValues
Steps in Calculation
X D= X - 400
1 Compute the medianor mean according to the averane
100 300
2 Take deviaions of items from the average ignoring signs and deno 200 200
these deviations by |D|
400
3 Obtain the total of these deviations i.e.. IDI 600 200
ED| where N is the number of 800 400
4 Apply the formula M.D. =
observations. N 1,100
If the deviation is small, the distribution is highly uniform, since th
term = term
more than half of the values are concentrated within a small range around The median is given by
the mean or median according to the average selected.
3rd term = 400
The relative measure corresponding to the mean deviation is called
coefficient of mean deviation and it is obtained by dividing mean deviation Mean deviation can be calculated using the formula;
by the particular average (i.e., either median or mean) used in computing
mean deviation. M.D. = substituting the values, we get
N

That is, if median is used in computing mean deviation, then M.D. = 1,100 = 220
Mean Deviation 5
Coefficient of Mean Deviation Median Computation of Mean Deviation for the Second Set ofValues
If mean is used, then, X |D|= X - 4,000
Mean Deviation
Coefficient of Mean Deviation 3,000 1,400
Mean
4,000 400
observations is
The computation of mean deviation for individual 4,200 200
made clear with the following examples.
4,400
4,600 200
PROBLEM
valuesfor 4,800 400
Calculate mean deviation using median from the giiven 5,800 1,400
two sets of observations. 4,000
100 800
200 400 600 Median ot the series can be calculated using the formula,
X 3,000 4,800 5,800 term, term
4,000 4,2004,400 4,600
4th term
4,400
 QUANTITATIVE ANALYSIS
FOR BUSINESS DECISIONS MEASURES OF DISPERSION AND SKEWNESS 381
380
Deviations = (X - 80)
Now mean deviation can be calculated using the eguatios

Mean Deviation =
ED 30 20 15 10 10 20 30 40 160

Substituting the values in the equation, we get Total = 335

4,000 =571.43 Mean Deviation = Substituting the values in the equation,we
M.D. =
N
ge
This means that average deviation of the individual items from the 335
= Rs. 30.4545, Coe. of M.D. =
30.4545
M.D. = = 0.38068
median is 220 for the first set of values and 571.43 for the 80
values. Coefficient of mean deviation can be caluclated usingsecond
the
set of
b.Computation of Mean Deviation for the Discrete Series
Mean Deviation
formula,
The formula for calculating mean deviation for adiscrete series is
Median
220 M.D.= fD0
Substituting the values for the first series, we get = 0.55 N
400
Steps
1 Calculate the median of the series.
Similarly for the second set of data, coefficient of mean deviation
57143 2. Take the deviations of the items from median ignoring signs
=0.129870 and denote them by D|.
4400
3 Multiply these deviations by the respective frequencies and
PROBLEM obtain the total £ f|D|.
4 Apply the formula, 2t|D,where Ndenotes the frequencytotal
Calculate the mean deviation from the following data giving wages
obtained by 11 workers, from median. N

PROBLEM
Wages in Rs 80 100 120240 11o70 60 80 90 65 50
na survey regarding the number of home appliances they have in
245
Meanhouseholds are shown in the following frequency
deviation about distribution. Compute
SOLUTION median.
In order to calculate median, the data has to be arranged in
the Number of Home
ascending or descending order. If arrangedIin
t the ascending
orderthevalues Appliances 1 2 4 5 6 7
will be:

50, 60, 65, 70, 80, 80, 90, 100, 110, 120,
Frequency 14 21 25 43 51 40 39 12

Median of the series is given by the

24O
formula;
SOLUTIAsONthe
term term
ollowing table.first step, we have to compute median, that is explained in the

Median term2 term = 123 term = 4

= 6h term = Rs.80
 QUANTITATIVE ANALYSIS FOR
BUSINESS
DECISIONS - | MEASURES OF DISPERSION AND SKEWNESS 383
382

frequency total. 10 15
Here N is the Runs Scored 5 20 25 30 35 40
Computation of Median 36
Numberof Matches 16 32 44 28 18 12 14

Number of Home Appliances (X) f

C.f
14 14 SOLUTION
1 21
Calculation of Median and Mean
35
2 25 60
43
Computation of median and mean are explained in the following ta
3 103
ble
51 154
Median can be calculated using theformula.
5 40 194
6 39 233 term = term
7 12 245
= 100.5th term
Computation of Mean Deviation
Mean beviation can be calculated using the formula,
Ef D| So, Median = 20 runs

Substitutingthevalues in the equation, we get Runs Scored (X) c.f fx

366 1.4938 = 0.3734 16 16 80
M.D. = = 1.4938 Coe. M.D.
245 4 10 32 48 320
15 36 84 540
20 44 128 880
Tnis means tB.at the variation of households from median is 1.494. Ine 25 28 156 700
calculation of mean deviation is explained in the following tabie. 30 18 174 540
Number of Home fIDI 35 12 186 420
IDI 40
Appliances Frequency 56
14 200 560
14 4
63
200 4,040
21 3
5
2 23 43
43 1
51 40
5 1 Ihe formulafor EfX
6
40
39 2
78
36
calculating Mean = N
7
12 3 366
245 Substituting the values, we get
PROBLEM match.
Find 4,040
200
the meanFollwing
are the runs in a cricket and me
deviation and coefficient scored by bastmen from
mean
dian, of mean deviation -20.2 runs
 QUANTIATIVE ANALYSIS FOR BUSINESS DECISIONS - | MEASURES OF DISPERSION AND SKEWNESS 385
384
Computation of Mean Deviation The calculation is explained with the help of the following examples.
From Median
From Mean PROBLEM
f X-20 = IDI f IDI X- 20.2 = IDI
X fIDI Calculatethe Mean Deviation from the following data.
15 240 15.2
16
5
320 10.2
243.2 Marks of Students Number of Students
10
10 32
5.2
326.4 10-20 4
180
15 36
0.2
187.2 20-30 6
20 44 8.8 30-40 10
5 140 4.8 20
25 28 134.4 40-50
10 180 9.8 50-60 10
30 18 176.4
15 180 14.8 177.6 60-70
35 12
70-80 4
20 280 19.8 277.2
40 14
200 1,520 1,531.2 SOLUTION
Computation of Median
Computation of median is explained in the following table
Computation of Mean Deviation from Median
Ef |D Marks c.f
The formula for calculating mean deviation = N
10-20 4 4
Substituting the values in the equation, we get 20-30 6 10
1,520 30-40 10 20
M.D. =7.6 runs
200 40-50 20 40
M.D. 7.6
Coefficient of M.D. -0.38 50-60 10 50
Median 20
60-70 6 56
Calculation of Mean Deviation from Mean
70-80 4 60
The formulais M.D=
N
Substitutingthe values in the equation, So th
60 .=
lerm that median40-50
30th
class can be found out using term =
We get M.D = 1,531.2 =7.656 runs
200
means is the median class.

Coefficient of Mean Deviation is caalculated using the

formula, Median can be Computed using the equation,
N
M.D Mean Deviation 7.656
=0.379 M= L+c.f.
Mean x0
20.2
Computation of Mean Deviation for the Continuous Series
Mean deviation for the series is
calculated
usingthe Substituting
M= 40 + 30- 20
the valuesin the equation, we get
X 10
formula = 2|D| xi continuous =40 + 5
20
N 45 marks
 BUSINESS MEASURES OF DISPERSION AND SKEWNESS 387
QUANTITATIVE ANALYSIS FOR DECISIONS |
386
calculated using the equation; Calculationofmean
be
Mean deviation can
Computation of mean is explained below
EfD
M.D. = N d
based on the mid points represented m d= m-35 = d' f fd'
Here the calculations are as m Marks(X) 10
of the class
intervals).
(i.e., mid points 10-20 15 -20 -2
ol 8 -16
m-A
25 -10 -1 10 -10
JD|is calculated as 20-30
30-40 35 15
Where Ais the assumed mean, selected trom the mid points oran. 40-50 45 10 25 25
interval.
other value, i = the class 50-60 55 20 2 20 40

The calculation is made clear with the help of the following example 60-70 65 30 3 18 54

70-80 75 40 9 36

m-A 80-90 85 50 5 5 25
Marks f m D|= f|D| 110 154
10
12 Here A = 35
10-20 4 15
2 12
20-30 25 Mean can be calculated using the equation
10
30-40 10 35 Efd'
X A+
40-50 20 451 10 N
50-60 10 55 1 Substituting the values in the equation, we get
2 12
60-70 6 65 154
3 12 35 +
70-80 4 75 x10
68 110
60
35 + 14 = 49 marks
Here A = 45 Computation of mean devlation
The formula to calculate Mean Deviation is omputation of Mean Deviation from Mean is explained below.
M.D. N Xi, Marks (X) m
D|= m - 49 f
Substitutingthe values in the equation, we get 10-20 15 272
34 8
M.D.
68
x 10 = 11.33 marks 20-30 25 24 10 240
60 30-40 35 14 15 210
PROBLEM
Meanfrom
the 40-50 45 4 25 100
Calculate Mean Deviation and its Coefficient from 30-60 55 6 20 120
following data.
l80-90
60-70 65 16 18 288
Marks 10-20 20-3030-40 40-50 50-60 60-70
|70-80 70-80 75 234

Number of
80-90 85
26
36 5 180
8 10 15 25 20 18
Students 110 1644
 QUANTITATIVE ANALYSIS FOR BUSINESS DECISIONS -|
MEASURES OF DISPERSION AND SKEWNESS 399
388
So the |D|is calculated as m- 49. We have not
Here,mean = 49, Mean can be calculated using the equation
the class interval. because it will make the values in
dividedthe values with Efd'
fractions. X=A + XI
Mean Deviation can now be calculated using the fornula
N
Substituting the values in the eugation, we get
M.D. 20
X= 50 + X 20
Rememberthat, here for calculating |D|, we have not 100
divided theuie
Values with the class interval, hence for calculating mean derivation
= 50 + 4

not have to multiply the clasS interval, Substhtutng the values in the eouuatie = Rs.54
for Mean Deviation, we get Computation of Mean Deviation from Mean
1,644 = 14.945 marks
M.D. 3
110
Wages |D= m -54
Coetficient ofMean Deviation can be claculated using the formula;
Mean Deviation 0-20 10 44 10 440
Coefficient Mean Deviation =
Mean 20-40 30 24 16 384
14.95 =0.305 40-60 50 4 30 120
Substituting the values, we get 49 60-80 70 16 32 512
PROBLEM
80-100 90 36 12 432
Calculate Mean Deviation from Mean and Median for the following. 100 1,888
0-20 20-40 40-60 60-80 80-100
Wages (Rs.)
30 32 12 Mean Deviation from Mean can be calculated using the equation
Number ofWorkers 10 16

SOLUTION 1,388_ RS. 18.88

Substituting the values, we get M.D.
100
Computation of Mean
Computation of Median
sR86lal
Wages m d=m - 50
f
fd' Wages (Rs.) f c.f
10
-20
0-20 10 10

0-20 10 -40 -2
10
-16 20-40 16 26
20-40 30 -20 -1 16 40-60 30 56
40-60 50) 30
32 60-80 32 88
60-80 70 20 1
32
24 80-100 12 100
80-100 12
90 40 2
100
20 Median classis given by N th
term =
100
= 50" term. This

CAIrmulativO frequency value falls in tho class 40-60.

2
 QUANTITATIVE ANALYSIS FOR BUSINESS
390 DECISIONS -| MEASURES OF DISPERSION AND SKEWNESS 391

Median can be calculated using the equation. SOLUTION

N -c.f. The given cumulative trequency distribution is to be convertec ictc.
2 xi an ordinary frequency distribution, So the Computation of Median is expiaied
M=L+
f
below:
Substituting the values in the equation, we get
Wages (X) Number oi Workers (t) c.f
50-26 0-10 5 5
M= 40 + x 20
30 10-20 13
= 40 + 16 = Rs. 56 20-30 7 20
30-40 12 32
Computation of Mean Deviation from Median
40-50 28 60
50-60 20 80
Wages (Rs.) m |D| =m -56 f|D| 60-70 10 90
70-80 10 100
0-20 10 46 10 460
20-40 30 26 16 416 Median class is given byN th term = 100 = 50n term, that
40-60 50 6 30 180 means 40-50 is the median class.
60-80 70 14 32 448
80-100 90 34 12 408
Median can be calculated using the equation,
2
-c.f.
100 1912 M=L +
f
Mean deviation can be calculated using the equation. Substituting the values in the equation, we get
Mean Deviation = Ef|D| M= 40 +
50 - 32
X 10
28
1,912 = Rs. 19.12 = 40 +6.43
Substituting the values, we get = = Rs. 46.43
100
PROBLEM Computation of Mean Deviation from Median
omputation of Mean Deviation from median is explained in the
alculate Mean Deviation from Median from the following dalo following table.
Wages Number of Workers Wages (X) m
|D = m46.43 f|D|
Less than Rs. 80 100 0-10 41.43 5 207.15
Less than Rs. 70 90 10-20 15 31.43 8 251.44
Less than Rs. 60 20-30 25 21.43 7 150.01
Less than Rs. 50
80
30-40 35 11.43 12 137.16
60 40-50 45 1.43 28 40.04
Less than Rs. 40 32 50-60 55 171.40
Less than Rs. 30 60-70 65
8.57 20
185.7
Less than Rs. 20
20
13
70-80 75
18.57
28.57
10
10 285.7
Less than Rs. 10 5 100 1,428.6
 ANALYSIS FOR BUSINESS MEASURES OF DISPERSION AND SKEWNESS 393
392
QUANTITATIVE DECISIONS -|
N
Mean Deviation can be calculated using the equation; - c.f.
2
M= L +
EfD|
M.D. N
get Substituting the vaiues in the equation, we get
Substituting the values in the equation, we 46 40
M= 40+ X 10
M.D. 1428.0 = Rs. 14.286 20
100 40 +3
PROBLEM 43
Calculate Mean Deviation and its coefficient from the followinn dals Computation of Mean Deviation from Median
Computation of Mean Deviation from iMedian is explained in the
Class Frequency following table,
0-10 5
Class (X) m D = m - 43 f f{D|
10-20
20-30 12 0-10 5 38 5 190
30-40 15 10-20 15 28 8 224
40-50 20 20-30 25 18 12 216
50-60 14 30-40 35 15 120
60-70 12 40-50 45 20
70-80 6 50-60 55 12 14 168
(April1991 - Part B) 60-70 65 22 12 Z64
70-80 75 32 6 192
SOLUTION 1,414
92
Mean Deviation can be calculated using Median or Mean. Here ie
which is
calculation is based on Median. So Median is to be calculatedfirst. Mean Deviation can be calculated using the equation
explained in the foilowing table.
M.D. 2rD|
c.f N
Class f
0-10 5 Substituting the values in the equation, we get
13
10-20
25 M.D. 1414
20-30 12
30-40 40 92
15
60 = 15.3696
40-50 20
74
50-60
60-70
14
12
86
92
Coefficient off Mean Deviation can be calculated usingthe equation,
70-80 6 Coefficient of Mean Deviation Mean Deviation
that means Median
Median class is term = 92 46h item,
Median class is 40 - 50. Median is calculated using theformula, Substituting the values in the equation, we get
 394 QUANTITATIVE ANALYSIS FOR BUSINESS DECISIONS -|
MEASURES OF DISPERSION AND SKEWNESS 395
= 15.3696
Coefficient of Mean Deviation
The standard deviation measuressthe absolute dispersion or variability
43

distribution, i.e.. the greater the amount of dispersion or variability, the

Merits of Mean
Deviation =0.3574 ofa
bee the magnitude of the deviatons of the values from their mean.
greater will
1. It is simple to understand and easy to compute A
smallIstandard deviation means a high degree of uniformity of the
2. It is based on each and every
item of tthe
data, So change in observations as well as homogeneity of a series. Alarge standard deviation
of any item would change the vàlue of the mean deviation the value Would mean just the opposite. Thus if we have two or more comparable
3. Mean deviation is less affected by the values of extreme series withidentical or
nearly identical means, it is the distribution with the
the Standard Deviation. items than smallest standard deviation that has the most representative mean. Hence
standard deviation is extremely useful in judgingthe representativeness of
4. Since deviations are taken froma central value,comparision ahoa
formation of different distributions can easily be made. the mean.
Limitations of Mean Deviation Differences between Mean Deviation and Standard Deviation
1 The areatest drawback of this is the algebraic signs which are ignored Both these measures of dispresion are based on each and every
while taking deviations of the items. This is mathematicaly incorrect item of the distribution. But they differ in the following respects
making the method non-algebraic.
1. Algebraic signs are ignored while claculating mean deviation, whereas
2. Mean deviation is not capable of further algebraic treatment. in the calculation of standard deviation, signs are taken into account.
3. Mean deviation is rarely used in social science studies.
2. Mean deviation can be computed either from median of mean. Standard,
4. This method may not give us very accurate results. The reason is that deviation on the other hand is always computed from the arithmetic
mean deviation gives us best reslts when deviations are taken from mean, because the sum of the squares of the deviations nf items from
Median. But Median is not a satisfactory measure when the degree of arithmetic mean is the least.
variability in a series is high. If we compute mean deviation from mean
that is also not desirable, because the sum of the deviation from mean a. Computation of Standard Deviation for Individual Observations
(gnoring signs) is greater than the sum of the deviations from median For individual observations standard deviation may be computed by
(ignoring signs). If mean deviation is computed from mode, that is no applying any of the following two methods.
Considered scientific because the value of mode cannot always be 1
determined (for example for open ended classes). By taking deviations of the items from the actual mean, or
The Standard Deviation 2
By taking deviations of the items from an assumed mean.
The Standard Deviation Concept was introduced by Karl Pearsonin
1893. is by far the most
It
important and widely used measure
of studying
those defects
Computation of Standard Deviation taking deviations from Actual Mean
dispersion. It's significance lies in the fact that it is most
from which the earlier
free from propertiesof 0s applied.vwnen deviations are taken from actual mean, the following formula
a good methods suffer and satisfies of the
measure of dispersion. Deviation
Standard deviation is also known as Root Mean squared Squaredeviations N
for the reason that, it the square root of the means ofthe
from the arithmetic ismean. thesmall
Greek
wnere x = (X-X) and N=number of observations
letter o(read as sigma). Standard deviation is denoted by
Sleps in Computation
1
valculate the actual mean of the series, i.e.