4

Measures of Central Tendency

In the previous chapters, data collection and presentation of data

were discussed. Even after the data have been classified and tabulated
one often finds t0o much details for many uses that may be made of the
information available. We, therefore, frequently need further analysis of
the tabulated data. One of the powerful tools of analysis is to calculate
a single averagevalue that represents the entire mass of data. The word
average is very commonly used in day-to-day conversation. For example,
we often talk of average work, average income, average age of employees,
etc. An 'average' thus is a single value which is considered as the most
representative or typical value for a given set of data. Such a value is
neither the smallest nor the largest value, but is a number whose value is
Somewhere in the middle of the group. For this reason an average is
frequently referred to as a measure of central tendency or central value.
Measures of central tendency show the tendency of some central value
around which data tends to cluster.
Objectives of Averaging
There are two main objectives of the study of averages :
() To get one single value that describes the characteristie of the
entire data. Measures of central value, by condensing the mass of data in
one single value, enable us to get an idea of the entire data. Thus one
value can represent thousands, lakhs and even millions of values. For
example, it is impossible to remember the individual incomes of millions
of earning people of India and even if one could do it there is hardly any
MS. But if the average income is obtained, we get one single value that
Tepresents the entirepopulation. Such a figure would throw light on the
Standard of living of an average Indian.
(i) To facilitate comparison. Measures of central value, by reducing
dhe nass of data in one single figure, enable comparisons to be made.
Omparison can be made either at a point of time or over a period of
me. For example, the figure of average sales.for December may be
COmpared with the sales figures of previous pnths or with the sales figure
of another competitive firm.
 BUSINESS STATISTICS MEASURES OF CENTRAL TENDENCY C-9945
100 101
A. ARITHMETIC MEAN
Characteristics of a Good Average The most popular and widely used measure for
Since an average is a síngle value representing a group of values, it data by one value is what most laymen call an representing the entire
'average' what the statisticians
isdesirable that sucha value satisfies the folHowing properties : call the arithmetic mean. Its value is obtained by and adding together all the
() II should be casy to understand. Since statistical methods are observations and by dividing this total by the number of
designed to simplify complexity, (it is desirable that an average be such Calculation of Arithmetic Mean--Ungrouped Data observations.
that can be readily understood; otherwise, its use is bound to be very For ungrouped data, arithmetic mean may be computed by applying
limited. any
of the following methods :
(i)It should be simple to compute. Not only an average should be (i) Direct method,
easy tounderstand but also it should be simple to compute so that it can
be used widely. However, though case of computation is desirable, it (ii) Short-cut method.
should not be sought at the expense of other advantages, i.e., if in the Direct Method
interest of greater accuracy, use of a more difficult average is desirable The arithmetic mean, often simply referred to as mean, is the total of the
one should prefer that.
(ii) tshould be bascd on all the observations. The average should values of a set of observations divided by their total number of observations.
depend upon each ánd every ob servation so that if any of the observation Thus, if X,, X,.., X, represent the values of items or observations, the
is dropped average itself is altered. arithmetic mean denoted byX is defined as :
(iv) I: should be rigidBy defined. An average saould e properly
defined so that it has one and only one interpretation. It should prefer I + X , +.+XE i=1
ably be defined by an algebraic formula so that if different people compute N
the average from the same figurcs they all get the same answer (barring
arithmetical mistakes). If the subscripts are dropped, the formula becomes :
() Lshould be capable of further algebraic treatment. We shoutd
prefer to have an average that could be used for further statistical com
putations,:For eKample, if we are given separately the figures of average 6 I tmay be pointed out that in keeping with standard statistical practice, the
income and Humber of employees of two or more factories we should be
able to compute the combined average. symbolX will represent throughout this text the arithmetic mean of aset of
(vi) It should have sampling stability. We should prefer to get a value observations.
Illustration 1. The monthly income (in rupees) of 10 employees working in a firm is as
which has what the statisticians call 'sampling stability'. This means that
if we pick 10 different groups of college students, and compute the average
follows
4446 4475 4492 4572 4516 4468 4489
values. 4487 4493 4502
of each group, we should expect to get approximately the same Find the average monthly income.
It does not mean, however, that there can be no difference in the value of Solution. Let income be denoted by X.
different samples. There may be some diference but those averages in EX =
4487+4493+4502+4446+4475+4492+4572+4516+4468+4489 = 44,940
which this difference, technically called sampling fluctuation, is less are LX 44940
considered better than those in which this difference is more. X= = 4494
(i) It shold not be unduly afected by the presence of extreme values. 10
Hence the average monthly income is Rs.
4494.
Although each and every observation should influence the value of the also be calculated by
average, none of the observations should influence it unduly. If oae or two Short-cut Method. The arithmetic mean* can
very small or very large observations unduly affect the average, i.e., either which case the formula shall be :
taking deviations from any arbitrary point in
increase its value or reduce its value, the average cannot be really typical
of the entire set of data. In other words, extremes may distort the averag X=A+ N
and reduce its usefulness.
*This formula is derived as follows :
The following are the important measures of central tendency which Let d = X-A
X=A + d
are generally used in business : Taking summation of both sides
and dividing by N, we get
A. Arithmetic mean, B. Median,
C. Mode, D. Geometric mnean, and N
E. Harmonic mean. N letter mu) and the sample mean
donated by 4 (4 is the Greek
The population mean is
by X.
 BUSINESS STATISTICS MEASURES OF CENTRAL TENDENCY 103
102
within the income group 4000-4500 is 50 we cannot say as to
where d= (X- A) how many
and A= Arbitrary point (or assumed mean). persons out of 50 are getting 4001, 4002, 4003, etc. We, therefore, make an
assumption while calculating arithmetic mean that the frequencies within each
It should be noted that any value can be taken as arbitrary point and the class are distributed uniformly or evenly over the range of the class-interval,
method.
answer would be the same as obtained by the direct
ie., there will be as many observations below the mid-point as above it. Unless
llustration 2. Calculate average monthly income by the short-cut method from data of
llustration I taking deviations from 4460 as the arbitrary point. such an assumption is made the value of mean cannot be computed.
CALCULATION OF AVERAGE INCOME Illustration 3. The following are the figures of profits earned by 1,400 companies during
Solution. 1999-2000.
X (X-4460)
(Rs.) Profits No. of Profius No. of
(Rs. Lakhs) Companies (Rs. Lakhs) Companies
4487 +27 200-400 500 1,000-1,200 I00
4493 +33
400-600 300 1,200-1,400 80
4502 +42
600-800 280 1,400-1,600 20
4446 -14
44%5 +15 800-1000 120
4492 +32
4572 +112 Calculate the average profits for all the companies.
4516 +56 Solution. CALCULATION OF AVERAGE PROFITS
4468 + 8
4489 +29
Profits Mid-points No. of
(Rs. Lakhs) X Companies
Ed = +340
340 ies, 200-400 300 500 1,50,000
= 4460 + = 4460 + 34 = Rs. 4494.
N 10 400-600 500 300 1,50,000
One may find that the short-cut method takes more timne as compared to 600-800 700 280 1,96,000
direct method. However, this is true only for ungrouped data. In case of grouped 800-1,000 900 120 1,08,000
data, considerable saving in time is possible by adopting the short-cut method. 1,000-1,200 1100 100 1,10,000
1300 80 1,04,000
Calculation of Arithmetic Mean Grouped Data 1,200-1.400
1,400-1,600 1500 20 30,000
For grouped data, arithmetic mean may be computed by applying any of
the following methods : N= 1,400 EX= 8,48,000
(i) Direct method, ZfX 8,48,000
(ii) Short-cut method. X= = 605.71
1,400
(i) Direct Method. When direct method is used Thus, the average profit is Rs. 605.71 lakhs.
N
(ii) Short-cut Method*, When short-cut method is used, the
where X= mid-point of various classes.
(direct method)
f= the frequency of each class. N
N= the total frequency. X-4
d = X=A+ id
Note. For computing mean in the case of grouped data the mid-points Now
of the various classes are taken as representative of that particular class. Substituting the value of X in the direct method, we get
The reason is that when the data are grouped, the exact frequency with
which each value of the variable occurs in the distribution in unknown. Ef4+id) Ef4 ESA
N
We only know the limits within which a certain number of frequencies
occur. For example, when we say that the number of persons Hence
(.EfaN)
N
 BUSINESS STATISTICS
MEASURES OF CENTRAL TENDENCY
104 105
(6) The mean of 200 obervations was 50. Later on, it
following formula is applicd : two observatioDs were wrongly
correct mean,
read as 92 and 8 instead of 192 was iscovered that
and88. Pind out the
N
Solutlon.
Here R=50 and N=200
X-A
where d s ZX-200x 5010,000
Isize of the cqual class interval.
Less Incorrect observations 100
und
Illustration 4. Calculate the average profit by the short-cut method from the data of 9,900
illustration 3. Add Correct observations 280
Solution. CALCULATION OF AVERAGE PROFITS Correct total 10,180
Mid points (X-900)/200 Correct meana 10, 180-509
Profius 200
(Rs. Lakhs) d fd
|
Mathematical Properties of Arithmetic Mean
200 400 300 500 -3 -1,500 The important mathematical properties of arithmetic mean are :
400 600 S00 300 -2 600 1. The algebraic sum of the deviations of all the observ ations from
600 - 800 700 280 --1 280 arithmetic mean is always zero, i.e., E(X-X)=0. This shall be clear
800 - 1.000 900 120 0
from the following example :
1,000 -- 1,200 1100 100 + + 100
X (X-3)
10 -20
1,200 -- 1,400 1300 80 +2 + 160 20 -10
30
1,400 - 1,600 1500 20 +3 + 60 40 +10
50 +20
N= 1,400 Efd = - 2,060 EX=150 ZX-)-0
2,060
X=4+2xi =900 1,400
x 200 = 900 - 294.29 = Rs. 605.71 Here R= EX =30. When the sum of the deviations from
Hence the average profit is Rs. 605.71 lakhs. the actual mean, i.e., 30, is taken it comes out to be zero. It is because
of this property that the mean is characterised as a point of balance, i.e.,
Correcting Incorrect Values the sum of the positive deviations from mean is equal to the sum of the
It sometimes happens that due to an oversight or mistake in copying, negative deviations from mean.
certain wrong values are taken while calculating the mean. The problem is how 2. The sum of the squared deviations of all the observations from
to find out the correct mean. The process is very simple. From X deduct arithmetic mean is minimum, that is, less than the squared deviations of
wrong observations and add correct observations and then divide the correct all the observations from any other value than the mean. The following
EXbythe number of observations and the result so obtained will give the value example would clarify the point :
of the correct mean. (X-X
X-)
llustration 5. (a) The average weekly wage for a group of 25 persons working in a factory -2 4
was calculated to be Rs. 378.40. It was later discovercd that one figure was misread as 160 1nstead
of the correct valuc Rs. 200. Calculate the corTect average wage. 0
1
Solution. EX = NX = 25 x 378.4 = 9460 5 +1
6 +2
160
Less Incorrect figure EX-20 E(X-X=0 E(X-X=10
9300
Add Correct figure case.
The sum of the squared deviations is equal to 10 in the above
200
squared
If the deviations are taken from any other value, the sum of the square
Total 9500
deviations would be greater than 10. This is known as the least
.. Correct LX = 9500
9500
Hence corect average = = 380.
25
 106 BUSINESS STATISTICS
MHASURES OF CENTRAL TENDENCy
107
property of the arithmetic ean and becomes the basis for defining the concept Solution. Let N, represent
N,+N, -100. percentage of men and N,
percentage of women so that
of standard deviation.
3. If we have the arithmetic mcan and number of observations of
We are given X2 30
more than two related groups, we can compute combincd average oftwvo or
these 32 (mean agc of
group of men)
groups by applyingthefollowing formula: X, 25 (mean age of
group of women)
N+ N,
N, +N, N, (32) + N, (25)
30 =
Combined mean of the two groups. 100
= Arithmetic mean of the first group. 3000
= Arthmetic mean of the second group.
32N,+(100-N )25
32N, + 2500 - 25N, = 3000 or
N, =71.43
N = Number of observations in the first group. N, = |00 71.43 = 28.57
N, - Number of observations in the second group. Hence he perccntage of men and women is
respectively 71.43 and 23 57
The following example will illustrate the application of the above Merits and Limitations of Arithmetic Mean
formula. The arithmetic mean is the most popular average in practice. it is due to
ltustration 6. There are two branches of a company
the fact that it possesses first six out of seven characteristics of a good average
employing 100 and 80 employecs
respectively. if arithmetic means of the monthly and no other average possesses such a large
4570 and Rs. 6750 respectively, find the arithmetic salaries paid by two branches are Rs.
mean of the salaries of the enmployces However, arithmetic mean is unduly
number of characteristics.
affected by the
of the company as a whole.
Solution. We should compute the combined mean. The formula is values. Also in open-end frequency distribution it is presence of extreme
dificult to compute
N,X, +NX mean without making assumption regarding the size of the class-interval
of the open-end classes. The arithmetic mean is usually neither the most
N, + N
Given
N, 100, X 4570, N, - 80, X, = 67S0 commonly occurring value nor the middle value in a distribution and in
extremely asymmetrical distribution, it is not a good measure of central
(100 x 4570) + (80 x 6750) S97000
100+ 80 $538.89 tendeney.
180
If we have to find out the combined mean of three related groups, the Weighted Arithmetic Mean
above formula can be extended as follows : One of the limitations of the arithmetic mean discussed above is that it
NX+ N,X, + N,X gives equal inportance to all the observations. But there are cases where
the relative importance of the different observatioins is not the same. When
N,+ N, +N, this is so, we compute weighted arithmetic mean. The terms 'weight'
Illustration 6. (a) The mean of marks in Statistics of 100 students of a class was 72. The
mcan of marks of boys was 75, while their number was 70. lind out the nean marks of' girls in the stands for the relative importance of the different observations. The formula
ciass
(MBA, Osmania Univ, 1996) for computing weighted arithmetic mean is :
Solution. We are given N 100, A 72, N,. ie., mean marks of boys 75, N, number XWX
of boys70 We have to ind out the mean murks of girls, i.e., X,. X = XW
N,X, + N,X, where X represents the weightedarithmetic mean
N, +N, X=The variable.
70(75) +30X.
W= Weights attached to the var1able X.
72
70+ 30 An important problem that arises while using weighted mean is
7200 S2 50+30 X regarding selection of weights. Weights may be either actual or arbitrary.
L.e., estimated. Needless to say, if actual-weights are available, nothing
1950
65 like this. However, in the absence of actual-weights, arbitrary or imag1nary
Hence mcan narks of girls in the class 65.
0
weights may be used. The use of arbitrary weights may lead to someif
tound that
IMustration 6, (b) The mcan age of a combincd group of men and women is 30 yeats. I| erTror, but this is better than no weights at all. In practice, it is
Jmcan age of the geoup of men is 32 and that of the group of wonen s 25, Nnd phenomenain view, the eror
Weights are intelligently assigned keeping the overlooked.
out the percenmuge
lnvolved will be so small that it can be easily
men and vomcn n he group
 BUSINESs STATISTICS MEASURES OF CENTRAL TENDENCY
109
eaual parts. If there are even number of
Weighted mean is specially useful in problems relating to the construction
of index numbers and standardised birth and death rates. actual value exactly in the middle of theobservations in a series, there is no
series and as
llustration 7. Acontractor employs three types of workers--male, female and children. indeterminate. In this case the median is arbitrarily taken to such the median is
be halfway between
Toa maie worker he pays Rs. 100 per day, to a female worker Rs. 75 per day and toa child worker the two middle observations. For example, if there are 10
Rs 3$ per day. What is the average wage per day paid by the contractor? series. the mediarn position is 5.5, that is observations in a
the median value is halfway between
Solution. The avrage wage is not the simple arithmetic mean. ie 00+ 3+35 = Rs. 70 the value of the observations that are Sth and 6th in order of magn1tude. Thus
3 when Nis odd, the median is in an actual value with the remainder of thesernes
per day. If we assume that the number of male. female and child workers is the same, this answer in two equal parts on either side of it. ifN is even, then the median 1s a derived
would be comect For exaple, if we take 10 workers in cach case then the average wage would be
ffgure, i.e., half the sum of the two values.
(10x 100) +(10x 75)+ (10x 35) 1000+ 750 + 350
10+ 10+ 10
= Rs. 70 Calculation of Median--Ungrouped Data
However, the number of male, female and child workerS employed is generally different. If Arrange the data in ascending or descending order of magnitude. (Both
we know how many workers of each type are employed by the contractor in question, nothing arrangements would give the same answer.)
Ike this. However, in the absence of this we take assumed weights. Let us assume that the number
of male. female and child workers employed is 20, 15 and 5. respectively. The average wage would Apply the formula : Median =Size of 2
th observation.
be the weighted mean calculated as foilows : Illustration 8. From the following data of wages of 7 workers. compute the median
#age per day (Rs.) No. of workers wage
WX Wages (in Rs.) 2600 2650 2580 2690 2660 2606 264
I00 20
ie Solution : CALCULATION OF MEDIAN
2000
75 1125 S. No. Wages arranged in
35 175 ascending order
LW= 40 SWX = 3300 2580
2 2600
3300
= 82.50 3 2606
40
4 2640
Hence the average wage per day paid by the contractor is Rs. 82.50 to all types of 5 2650
workers.
6 2660
B. MEDIAN
7 2690
The median is the measure of central tendency which appears in the
"m1ddle"' of an ordered sequence of values. That is. half oftheobservations Median= Size of
N+1
th observation = 7+l = 4th observation.
2 2
in aset of data are lower than it and half of the observations are greater Value of 4th observation is 2640. Hence median wages = Rs. 2640.
than it.
the above illustration the number of observations was odd and. therefore, it w2s possible
As distinct from the arithmetic mean which is calculated from the to determine the value of 4th observation. When the number of observations is even.
for
of
value of every observation in the series, the median is what is called a example, if in the above case the number of observations are 8 the median would be the value
positional average. The term 'position' refers to the place of a value in a 8+1
4.5th observation. For finding out the value of 4.5th observation we shall
take the
series. The place of the median in a series is such that an equal number of 2 = be
average of 4th and 5th observations. Hence the median shall
observations lie on either side of it. For example, if the income of five 2640 + 2650
persons is Rs. 7000, 7200, 7500, 7600, 7800, then the median income = 2645
2
would be Rs. 7500. Changing any or both of the first two values with any
other numbers with value of 7500 or less and/or changing any of the last Caleulation of MedianGrouped Data N
two values to any other values with values of 7500 and more, would not value of median lies. Use o
affect the value of the median which would remain 7500. In contrast. in Determine the particular class in which the
N+1
case of arithmetic mean the change in value of single observation would locate the median class and not 2 because in the use of grouped data it is
cause the value of the mean to be changed. Median is thus the central two equal parts.
value of the distribution or the value that divides the distribution into two NI2which divides the area of the curve into
Lo
MEASURIS OF CENTRAIL TENDENCY 111
As I10 BUSINESS STATrISTICS
Solution : CALCULATION OF MEDIAN
Apply thc following fomula for determining the exact value of median: Profits (Rs. Crore) No. of companles ) cf.
tMedian
13 Lower limit of median class, i.e., the class in which the
middle observation in the distribution lies.
10
20- 3 0
30
40

20
40
S0

4
12
24
36
40
76

4
20 96
p.e,f. = Preceding cumulative frequency to the median class. S0 112
f= Frequency of the median class. 60 70 120
The class-interval of the median class. 70 80 5 125
llustration 9. () ,500 workcrs arc working in an industrial cstablishent. Thcir agc is 125
Med, Sizc ofth obscrvation 62.5th obscrvation.
class1ficd as follows : 2
Mcdian lics in the cluss 30--40).
Age (yrs. ) No. of workers Age (yrs. ) No. workers NI2- p.c.f.
18-22 120 38-42
Med, L+
184
22-26 125 42-46 162 1.30, NI2 62.5, p.c.f. 40, f 36, i= 10
26--3(0 280 46--50 86 62.5- 40
30 34 260 50--54 Med, 30 + 10 30 + 6.25 = 36.25.
75 36
34 38 155 5458 53 Hence 50% of the companics have profits upto Rs. 36.25 crores and the remaining S0% of
Calculatc the nedian age. the companics havc profits more than Rs. 36.25 crores.
Solution : CALCULATION OF MEDIAN AGE Merits and Limitations of Median
Age group cf. The median is superior to arithmetic mean in certain respects. For example,
18-22 120
it is especially useful in case of open-enddistribution and also it is not influenced
120
22-26 125 245
by the presence of extreme values. In fact when extreme values are present in
26 30 280 525 the data, the median is a more satisfactory measure of central tendency than
30--34 260 785 the mean.
34-38
38-42
155
184
940 The sum of the deviations of observations from median (ignoring signs) is
1,124
42--46 162 1,286
minimunm. In other words, the absolute deviation of observations from the
46-50 86 I,372 median is less than from any other value in the distribution. For example, the
50
54
-54
58
75 I,447 median of items 4, 6, 8, 10 and 12 is 8. The deviations from 8 ignoring signs
53 1.500 are 4, 2, 0, 2, 4 and the total is 12. This total will be smaller than the one
1,500 obtained if deviations are taken from any other value. Thus, if deviations are
Median Sizc of th observation 750th obscrvation,. taken from 7 the deviations ignoring signs would be 3, 1, 1, 3, 5 and the total
2
Hence median lics in the class 30.--34. is 13. In an estimation situation, if one is interseted in minimising the absolute
Median + NI2-p.J, i3(0 + 750-525 4 30+3.46 m33.46
amount of error and the sign of the error is not particularly important, then the
median is preferable to arithmetic mean.
260
Hencc the median agc of the workers is 33.46 ycars. However, since median is a positional average, its value is not determined
(0) Calculute thc mcdian from the following data pertaining to the profits (in crore Rs.) of
n6 companies
by each and every observation. Also median is not capable of algebraic
treatment. For example, median cannot be used for determining the combined
Profits (Rs. crore)
less than 10
No. of companies median of two or more groups. Also the median is less reliable average than the
less than 20
4
mean for estimation purposes since it is more affected by sampling fluctuations.
less than 30 40
Furthermore, the median tends to be rather unstable value if the number of
less than 40 76. observations is small.
less than 50 96
less than 60 112
Related Positional Measures or Quantities
less than 70 120 Besides median, there are other measures which divide a series into
less than 80 125 two equal parts. Important amongst these are quartiles, deciles and
(MBA, MD Univ., 1994) percentiles. These quartiles, deciles and percentiles are all special cases of
 BUSINESS STATISTICS MEASURES OF CENTRAL TENDENCY 113
112
llustration 10. The profits earncd by 100 companies during 1998-99 are given below
quantit ies. Quartiles are those values of the variate which divide the total
frequency into four equal parts, deciles divide the total frequency in 10 Profits No. of Profits Na. of
equal parts and the percentiles divide the total frequency (Rs. lakhs) companies (Rs. lakhs) companies
parts. Just as one point divides a series into two parts, in 100 equal 20 --30 4
three points would 60 70
divide it into four parts, S9 points into 10 parts and 99 points into 100 30 - 40 70 80 10
parts. Consequently there are only 3 quartiles, 9 deciles and 99 percentiles 40 50 80 - 90 8
for a series. The quartiles are denoted by symbol 0, deciles by D and 50 -- 60 30 90- 100
percentiles by P. The subscripts 1, 2, 3, etc., beneath , D, and would
refer to the particular value that we want to Calculate ,, median, D, and P and interpert the values. (MBA, D.U., 1999)
compute.
denote first quartile, , second quartile, , third quartile, Thus , would
D, first decile, Solution. CALCULATION OF 2,, 2,. D, AND Po
D, eighth decile, P first percentile, etc.
Profits cf.
Graphically any set of these partition values serves to divide the (Rs. lakhs)
area of the frequency curve or histogram into equal parts. If vertical lines 4 4
20 30
are drawn at each quartile, for example, the area of the histogram will be
divided by these lines into four equal parts. The 9 deciles divide the
30 40 12
of the histogram or frequency curve into 10 equal parts and the 99 area per
40 -
50
50
60
18
30
30
60
centiles divide the area into 100 equal parts. 60 70 15 75
In economics and business, quartiles are more widely used than 70 80 10 85
deciles and percentiles. Tbe quartiles are the points on the X-scale that 80 90 8 93
divide the distribution into four equal parts. Obviously there are three 90- 100 7 100
quartiles, the second coinciding with the median. More precisely stated, 100
the lower quartile, 2, is that point on the X-scale such that one-fourth
of the total frequency is less than , and 9, = Size of N/4th observation = =25th observation.
The upper quartile, 3, is that point on thethree-fourths is.greater than ,.
X-scale such that
of the total frequency is below 2, and one-fourth is above it. three-fourths Hence , lies in the class 40 50.
The deciles and percentiles are important in 2, =L +
N/4- p.c.f 25-12
x 10 = 40 + 7.22 = 47.22
educational statistics concerning grades, rates,
psychological and f
x i= 40 +
18
of use in economics and business in personnelscores and ranks ; they are
department, productivity 25 per cent of the companies earn an annualprofit of Rs. 47.22 lakhs or less.
ratings and other situations. 2N 200
Computation of Quartiles, Deciles, Percentiles, etc. Median or , = Size ofth observation = = 50th observati0n. 0,
4 4
The procedure for computing quartiles, deciles, etc., is the same as lies in the class 5060.
for median. 2N /4-p.c.f 50-30
For grouped data, the following formulae are used for quartiles, 2, =L+ X i= 50 +
30
x 10 = 50 + 6.67 = 56.67
deciles and percentiles ; 50 per cent of the companies earn an annual profit of Rs. 56.67 lakhs or
jN less.
4 -p.c.f. 4N
=L+ -Xi D, = Size of th observation = 40th observation
for j=1, 2, 3 0
kN D, lies in the class 50-60.
10- P c .
D,=L+ Xi D, = L +
4N/10- p.c.f. x i = 50 +
40-30
for k=1, 2,..,9 30
x 10= 50 +3.33 = 53.33.
IN
100 -P.cf. Thus 40 per cent of the companies eam an annual profit of Rs. 53.33
lakhs or less.
Pi=L+
for l=1, 2,...,99 P80 = Size of
80N 80 x 100
where the symbols have their usual meanings and -th observation = =80th observation
interpretation. 100 100
Pao lies in the class 7080.
80 N/100- p.c.J. 80-75
PL+ x /= 70 +
10
x 10= 70 + 5 = 75
 BUSINESS STATISTICS MEASURES OF CENTRAL TENDENCY 115
114
earn an annual proft Illustration 12. You are given the net profits carned by some companies. Estimate
This means that 80 per cent of the companies carn an annual graphically the percentage of companies getting profits between Rs. 25,000 and Rs. 75,000.
of Rs. 7S lakhs or Jess and 20 per cent of the companies Profits (in Rs.) No. of companies Profits (in Rs.) No. of companies
profit of more than Rs. 75 lakhs.
10,000--20,000 15 60,000-70,000 22
Determination of Median, Quartiles, etc., Graphically 20,000--30,000 35 70,000--80,000 2
Median can be determined graphically by applying any of the 30,000--40,000 47 80,000-90,000
40,000--50,000 90,000-1,00,000 8
following two methods : by more
68
1. Draw two ogives-one by 'less than' method and other 50,000-60,900 32
than' mnethod. From the point wherc both these curves intersect each Solution. Finding percentage from the given data :
other, draw a perpendicular on the X-axis. The point, where this per Profits less than No. of companies Percentage
pendicular touches the X-axis gives us the value of median.
2. Draw only one ogive by less than' method. Take the varjable Rs. 20,000 15 6.0
20.0
on the X-axis and frequency on the Y-axis. Determine the median value
30,000 50
40.000 97 38.8
N 50,000 165 66.0
by the formula : median =Size of 2 -th item. Locate this value on the 60,000 197 78.8
70,000 219 87.6
Y-axis and from it draw a perpendicular on the cumulative frequency 80,000 231 92.4
curve. From the point where it meets the ogive draw another perpen 90,000 242 96.8
dicular on the X-axis and the point where it mects the X-axis is the 1,00,000 250 100.0
median.
Plotting the data on the graph paper :
The other partition values like quartiles, deciles and percentiles can
also be determined graphically. PERCENTAGE OF COMPANIES EARNING PROFITS 8ETWEEN
Illastration 11. Using the´ data of illustration 10, determine graphically the 100
RS. 25.000-RS. 75,000
values of Q,. s, Deo and Pso r92
Solution. Draw the ogive by the less than' method as shown in the graph. |80
PERCENT
To determine different quartiles, horizontal lines (bro7en) are drawn from the
cxample if we want to determine the valne of
cumulative AGE OF 60
median. a hoetai line can. rOr be drawn from the cumulative frequency value of 0-50 COS
40
to the less than curve and then extending a vertical line to the horizontal xis. In a 92-13
similar manner other values can be determined as shown in the graph. Therefore, 79 20
Q=47:22, e,-5667, D,=53 33 and Pp=75. This may be noted down here that 13
these graphical values are same as obtained by the formulac.
10,000 30,000 50,000 70,000 90,000
20,000 40,000 60,000 80,000 100,000
FCRUEMQLANTCIVYE
100 100 Profts (ln Rs.)
The graph shows clearly that the percentage of companies earning profits less than Rs.
75,000 is 92 and the percentage of companies earming profits less than Rs. 25,000 is 13. Thus the
70
percentage of companies making prafits between Rs. 25,000 and Rs. 75,000 is (92-13) = 79.
jo 60
CDIAM j0-50 C.MODE
40 Daoo-49 Mode is defined as that value which occurs the maximum number of
I-025$ times, i.e., having the maximum frequency. For example, if we take the values
l20 of six different observations as S, 8, 10,8,5, 8, mode will be 8 as it has
occurred maximum number of times, i.e., 3 times. Graphicalty, it 1s the value
20 30 40 3 (60 1080 93 00 on the X-axis below the peak, or highest point, of the frequency curve as can
! be seen from the following diagram.
47 221 55 67 75
53 33
PROFITS (Rs LAKHS )
MODE
 BUSINESS STATISTICS MEASURES OF CENTRAL TENDENCY 117
16
fashion
This interpretation ofthe statistical mode is analogous to that of the maximum frequency. For example, if the data is :
current
rnode. A person dressing with current styles is "in the mode'". But a Size of shoes 6 9 10
fashion can be a poor description of what most persons are wearíng because No. of persons: 10 20 25 40 22 13
of the variety of styles worn by the general public. In statistics, the mode only the modal size is 8' since it appears maximum numtber of timnes in the
teils us which singie value occurs most often; it may, therefore, represent a data,
rnajority of the total population.
Caleulation of Mode
It is possible that a distribution may be bimodal. This happens when there
may be two or more values of equal or nearly equal occurence as can be seen Deterrnining the precise value of the mode of a frequency distributior.
from the following diagram : is by no means an elernentary calculation. Essentially it involes fitting
mathematically some appropríate type of frequency curve to the grouped
data and the determination of the value on the X-axis below the peak of the
curve. However, there are several elementary methods of estimating the
mode. These methods have been discussed for ungrouped and grouped
data.
Calculation of Mode-Ungrouped Data
For determining mode count the number of observations the various
values repeat thernselves, and the value which occurs the maximum number
of times is the modal.value.
The presence of more than one mode has a special significance in statístícal Ilustration 13. The following figures relate to the preferences with regard o size of
analysis,for it indicates potential trouble. It ís usually dangerous to compare screen (in ínches) of T:V. sets of 30 persons selected at random from a locality. Find the
bimodal populations or to draw conclusions about them because they usually modal size of the T.V. screen.
12 20 2 24 29
arise when there is some non-homogeneous factor present ín the population. 20 12 20 29 24
If the collected data produce a bimodal distribution, the data themselves 24 20 12 20 24
should be questioned. Quite often such a condition is caused by the taking of 29
24
24 24 20 24
too small a sarnple: the difficulty can be remedied by increasing the sample 24
20
20
24
29
24
24
12
24
size. In instances where a distribution is bímodal and nothing can be done to
change it, the mode is obviously eliminated as a measure of central tendency. Solution. CALCULATION OF MODAL SIZE
There are many situations in whích arithmetic mnean and median fail to
Size of inches Tally Frequency
reveal the true characteristic of data. For example, when we talk of most
common wage, most common income, most common height, most common 12
size of shoe or ready-made garments, we have in mínd mode and not the 20 8
arithmetic mean or medían discussed earlier. The mean does not always provide 24
13
an accurate reflectíon of thedata due to the presence of extreme valucs. Median
29
may als0 prove to be quite unrepresentative of the data owing to an uneven 4
distribution of the series. For example, the values in the lower half a of Total 30
distribution range from, say, Rs. 10to 100, while the same number of items in Since size 24 occurs the maximum number of times, therefore, the modal size of TV.
the upper half of the series range from Rs. 100 to Rs. 6,000 with most of screen is 24 inches.
them
near the hígher limit. In sucha distribution the median value Rs. 100 willprovide Calculation of Mode--Grouped Data
little indícation of the true nature of the data.
Both these shortcomings may be overcome by the use of mode. Mode In the case of grouped data the following formula is used for
refers to that valuc which occurs most frequently in a distribution. calculating mode:
is the easiest to compute since it is the value Mode
corresponding to the Mo = L +
A, + A, ...(0)
 MEASURES OF CENTRAL TENDENCY 119
118 BUSINESS STATISTICS
where L= Lower limit of the modal class. Locating Mode Grapkically
A, = The difference between the frequency of the modal class In a frequency distribution the value of mode can also be determined
and the frequency of the pre-modal class, i.e., preceding grapbically. The steps in calculation are :
class. 1. Draw a histogram of the given data.
A, = The difference between the frequency of the modal class bar,
2. Drw two lines diagonally on the inside of the modal classof the
and the frequency of the post-modal class, i.e., starting from each upper corner of the bar to the upper corner
succeeding class. adjacent bar.
i = The size of the modal class. 3. Draw a perpendicular line from the intersection of thevalue. two
Another form of this formula is : diagonal lines to the X-axis (horizontal scale) which gives us modal
IIlustration 15. The daily profits io rupees of 100 shops are given as follows:
Mo= L + Profits (in Rs.) No. of shops
Profits (in Rs.) No. of shops 20
0-100 12 300 4C0
Lower limit of the modal class. 400-500 17
where L= 100-200 18 6
S = Frequency of the modal class. 200-300 27
500-600
S,= Frequency of the class preceding the modal class. Draw 1he histogram and thence find
the modal value. Check this value
by
J,= Frequency of the class succeeding the modal class. irect calculation.
While applying the above formula for calculating mode, it is necessary to Solution,
see that the class intervals are uniform throughout. If they are unequal they Direct calculation:
should first be made equal on the assumption that the frequencies are equally Mode lies in the class 200300.
distributed throughout the class, otherwise we will get misleading results. Mode=L+ A1-xi-200+ 9 4 7
9 x 100-256 25
A distribution having only one mode is called unimodal. If it contains A1tAs
more than one mode, it is called bimodal or multimodal. In the latter case the
value of mode cannot be determined by the above formula and hence mode is 28
LLOCATING MODE GRAPHICALLY
ill-defined when there is more than one value of mode.
Where mode is ill-defined, its value may be ascertained by the following 24
approximate formula* based upon the relationship between mean, median
and mode. 20
Mode =3 Median -2 Mean
Illustration 14. The following data relate to the sales of 100companies :
...(iii)
OFSNO.HO216PS CNPLORADSEsL CMPLOASTLs 17
Sales No. of companies Sales No. of companies
(Rs. lakhs) (Rs. lakhs)
Below 60 2 66--68 10
60-62 68--70 3
62-64 25 70-72 2 6
64--66 30
Calculate the value of modal sales.
Solution. Since the maximum frequency 30 is in the class 64--66, therefore, 64--66 is
the modal class.
100 200 300 400O 500 60o
Mode L+ MODE Re256
...(iii) PROF1TS (IN Rs
L= 64, A,= (30 - 25) = 5, 4,- (30-10) = 20, i = 2
5 10 From the above diagram, the modal value is also 256. Hence by
Mode = 64 +
5+20
x 2 = 64 +
25
= 64.4
both the, methods we get the same value of mode.
Hence modal sales are Rs. 64.4 lakhs.
*Sce page 120.
 BUSINESS STATISTICS MEASURES OF CENTRAL TENDENCY 121
120
Mode can also be dtermined form frequency polygon in which case If we know any of the two values out of the three., we can compute the
perpendicular is drawn on the base from the apex of the polygon and the point third from these relationships. The following example will illustrate this point
Illustration 16. In a moderately asymmetrical distribution the Mode and Mean are 32
where it meets the base gives the modal value. and 35.4 respectively. Calculate the Median.
However, graphic nmethod of determining mode can be used only where Solution. Mode = 3 Median -2 Mean
there is one class containing the highest frequency. If two or more classes Mode = 32.1, Mean = 35.4
Substituting the values
have the same highest frequency, mode cannot be determined graphically. 32.1 = 3 Median -2 (35.4) or 3 Med 102.9 or Med. = 343
Merits and Limitations of Mode
D.GEOMETRIC MEAN
Like median, the mode is not affected by extreme values and its value can
be obtained in open-end distributions without ascertaining the class limits. Mode In business and economic problems,very often we are faced with questions
can be easily used to describe qualitative phenomenon. For example, when we pertaining to percentage rates of change over time. Neither the mean. the median
want to compare the consumer preferences for different types of products, nor mode is the appropriate average to use in these instances. For example.
say, soap, toothpastes, et., or different media of advertisingwe should compare consider the following figures of sale ofa company :
Year: 1997 1998 1999 2000
the modal preferences. In such distirubtions where there is an outstanding
Sales (million tonnes) : 20.2 22.5 23.9 28.0
large frequency, mode happens to be meaningful as an average.
However, mode is not a rigidly defined measure as there are several formulae Suppose we want to find out the average percentage rate of change per
for calculating the mode, all of which usually give somewhat different answers. year in sales. To answer this question we must specify what we mean by the
Also the value of mode cannot always be computed, such as, in case ofbimodal "Average percentage rate of change per year'. The most generally useful
distributions. interpretation of this term is the constant percentage rate of change which if
Relationship among Mean, Median and Mode applied each year would take us from the first to the last figure. Hence in the
A distribution in which the values of mean, median and mode coincide above illustration we would be interested in that constant yearly percentage
is known as symmetrical distribution, Conversely stated, when the välues rate of change which would be required to move from 20.2 million tonnes of
of mean, median and mode are not equal the distribution is known as sales in 1997 to 28.0 million tonnes in 2000. None of the previously discussed
asymmetrical -or skewed. In moderately skewed or asymmetrical averages provides the correct answer to this question. The correct answer can
distributions a very important relationship exists among mean, median and be obtained through the use of the geometric mean or, what amounts to the
mode. In such distributions the distance between the mean and the median same thing, through the use of the familiar compound interest formula. In the
is approximately one-thrid of the distance between the mean and mode as discussion which follows, the geometric mean is defined, and the relationship
will be clear from the following diagram between this average and compound interest calculations is indicated.
RELATIONSHIP OF ARITHMETIC MEAN MEDIAN AND MODE Geometric mean is defined as the Nth root of the product of Nobservations
DIVIDES AREA of a given data. If there are two observations, we take the square root: if there
UNDER PEAK IN HALVES are three observations, the cube root; and so on, symbolically.
OF CURVE CENTRE OF
GRAVITY G.M. = x, xX, xX,x..x XN
where X,, X,, X, .., X, refer to the various observations of the data.
MoX When the number of observations is three or more the task of
Me
multiplying the number and of extracting the root beconmes quite difficult.
Karl Pearson has expressed this approximate relationship as follows
: To simplify calculations logarithms are used. Geometric mean is then
Mean - Median = (Mean - Mode) calculated as follows:
log X, +log X, +....+ log XE £ log X
Mode = 3 Median-2 Mean log G.M. =
N N
and Median : 2 Mean + Mode Elog N
3 G.M. = antilog N
 122 BUSINESS STATISTICS MEASURES OF CENTRAL TENDENCY 123
Calculation of Geometric Mean observations, situations which we usually come across in social and economic
In ungrouped data geometric mean is calculated with the help of the fields.
following formula : The following examples illustrate the use of geometric mean.
G.M. - AL.ig
N
Illustration 17. Compared to the previous year the overhead expenses went up by 32% in
1999; they increased by 4% in the next year and by S0% in the following8 year. Calculate the
In grouped data for calculating geometricmean first we will find the average rate of increase in the overhead expenses over the three years.
Solution. In average ratios and percentages, geometric mean is more approprnate Appiying
midpoints and then apply the following formula : geometric mean here :
% Rise Expenses at the end of the year log X
G.M. = A.L. taking preceding year as I00
N
32 132 2.1206
where X= midpoint. 2.1461
40 140
2.1761
Compound Interest Formula S0 150
The compound interest formula is expressed as follows: E log X = 6.4428
P,= P, (1 +r)
where P = amount accumulated at the end of n periods ; G.M. = A.L.
(E log X - A.L.
6.4428
A.L. 2.1476 140.5.
N 3
Pa = Original principal; Averagc rate of increase in overhcad expenses
140.5 100 = 40.5%.
r= Rate of interest expressed as a decimal: and IIlustration 18. The annual rates of growth of output of a factory in 5
ycars are 5.0. 75,
n= Number of compound periods. of output per annum for the
2.5, 5.0 and 10.0 respectively. What is the compound rate of growth
It follows from the above formula that period ?
Solution. CALCULATING COMPOUND RATE OF GROWTH
r= Annual rate of Output relatives at the log X
growth end of the year
If interest is compounded at different rates in each time period, and if 2.0212
5.0 105.0
these successive rates are denoted by r,, r,, .. ,r, then the amount accumulated 107.5 2.0314
7.5
at the end of n periods with an original principal of P, is 102.5 2.0107
2.5
P= P, (1 +r) (1 +r) (1 +r). 5.0 105.0 2.0212
Applications of Geometric Mean 10.0 I10.0 2.04|4
Geometric mean is specially useful in the following cases : £ log X= 10.1259
1. The geometric mean is used to find the average per cent increase in
sales, production, population or other economic or business data. For example, (E log X (10.1259 - A.L. 2.0252 105.9.
G.M. = A.L. - A.L.
5
from 1997 to 1999 prices increased by 5%, 10% and 18% respectively. The N
$.99%
annum for the period is 105.9-100
average annual increase is not 11% as given by the arithmetic average but |The compound rate of growth of output per
2,00,000 and sold t0 years later
10.9% as obtained by the geometric mean. This average is also useful in Illustration 19. A piece of property was purchased for Rs.
return on the original tnvestment
measuring the growth ofpopulation, because population increases in geometric for Rs. 23,26,000. What is the average annual rate of
Solution. 2.00,000 X'° 3,26,000
progression.
2.Geometric mean is theoretically considered to be the best average in the 3,26,000
= 1.63
construction of index number.* It makes index numbers satisfy the time reversal 2,00,000
test and gives equal weights to equal ratio of change. Log l.63 0.2122
-0.0212
3. It is an average which is most suitable when large weights have to Log X 10
be given to small values of observations and small weights to large
values of X= A.l.. (0.0212) = I.05 or 105%. t0-ycar
retum of 105-100 = 5 per cent over the
* See chapter on Index Numbers. Hence the investment yielded a mcan rate
period.
 124 BUSINESS STATISTICS MEASURES OF CENTRAL TENDENCy 125

Combined Geometric Mean reciprocal of the individual observations. Thus by dcfinition

Just as we have talked of combined arithmetic mean, in a similar manner N
H.M,=
we can also talk of combined geometric mean. the geometric mean of N
observations is 6 and these N observations are divided into two sets first
containing N, and second containing N, observations having G, and G, as the When the number of observations is large the computation of har
respective geometric means, then monic mean in the above manner becomes tedious. To simplify caBcula
tions we obtain reciprocals of the various observations and appiy the
Log G = N, Log G+N, Log G, formulae
N, +N, For ungrouped data, H.M.=
N
Thus if the geometric mean of 5 observations is 20 and of another 10
observations is 35.28, the combined geometric mean shall be (4)
N
S Log 20 + 10 Log 35.28 (5x1.3010) +(10 ×1.5475) For grouped data, H.M.=
Log G = 5+10 15
6.505 + 15.475 21.98 Illustration 21. (i) Calculate ha_monic mean of numbers 10, 20, 25, 40, 50.
= 1.465 (it) Calculate harmonic mean from the following frequeacy distribution :
15 15 40-50
X:0-10 .1020 20-30 30-40
G = A.L. 1.465 = 29.17. f: 8 15 20 4 3
Illustration 20. Three groups of observations contain 8, 7, and 5
geometric means are 8.52, 10.12 and 7.75 respectively. Find the observations. Their Solution. (i) CALCULATION OF HARMONIC MEAN
observations in the single group formed by pooling the three groups. geometric mean of the 20 1|X
Solution. Log G =
N, Log Gi+ N, Log G, +N, Log G
10 0100
N, + N + N 0"050
20
8 Log 8.52 +7 Log 10.12 +5 Log7.75 25 0040
40 0-025
8+7+5 0-020
50
(8x.9304) +(7 x 1.0052) +(5x .8893) E1/X=0-235
20
7.4432 +7.0364 + 4.4465 18.9261
20 20
= 0.9463 -H.M.= 3-2128
G= A.L. 0.9463 =8.837. ()
Hence the combined geometric mean of the 20 observations taken
together is 8.837. (ii) CALCULATION OF HARMONICMEAN
Merits and Limitations of Geometric Mean Y f fx l|X
Variable
Geometric mean is highly useful in averaging ratios and percentages and I'600
in determining rates of increase and decrease. It is also capable 8
0-10
of algebraic 10-20 15 1 1-000
manipulation. For example, if the geometric mean of two or more series and 20-30 25 20 0:800
0114
their numbers of observations are known, a combined geometric 3040 35 4
0 057
mean can 40-50 45 3
easily be calculated.
However, compared to arithmetic mean, this average is more difficult N=50 ) 3-531
compute and interpret. Also geometric mean cannot be computed when thereto
are both negative and positive values in a series or more $0
zero value.
observations are having H.M.=
N 13"96.

E. HARMONIC MEAN ()
The harmonic mean is based on the reciprocal of
averaged. It is defined as the reciprocal of the arithmetic the numbers
mean of the
 126 BUSINESS STATISTICS MEASURES OF CENTRAL TENDENCY 127
Illustration 23. A toy factory has assigned a group of 4 workers to complete
Applications of Harmonic Mean an given
order.of
are below1,400
: toys of a certain type. The productive rates of the four workers
The barmonic mean is restricted in its field of application.* It is Workers Produetive rates
useful for computing the average rate of increase of profits or average 4 mioutes per toy
speed at which a journey has been performed or the average price at B 6 minutes per toy
which an article has been sold. The rate usually indicates the relation 10 minutes per toy
between two different types of measuring units that can be expressed D 15 minutes per toy
reciprocally. For example, if a man walked 20 km. in 5 hours, the rate of
his walking speed can be expressed as follows : Find the average minutes per oy by the group of workers.
20 km. Solution, If we assume that each of the four workers is assigned the same
=4 km. per hour 1,400
5 hours number of toys (constant value) to meet the orders, or 4 350 toys per worker,
where the unit of the first term is a km. and the unit of the second term the aritbmetic mean would give the correct answer.
is an hour. Or reciprocally, 4+6+10415
5 hours 1 4 t P -8minutes per toy
hour per km.. Verification
20 km. 4
Time required by A to complete 350 toys=350 x 4=1,400 minutes
where the unit of the first term is an hour and the unit of the second term Time required by B to complete 350 toys =350 x 62,100 minutes
is akilometre. Time required by C to complete 350 toys=350 x 10=3,500 minutes
Jllustration 22. In a certain factory a unit of work is completed by A in Time required by D to complete 350 toys=350 x 155,250 minutes
4 minutes, by B in 5 minutes, by Cin 6 minutes, by D io 10 minutes and by Ein
12 minutes. 12,250 minutes.
(a) What is the average number of units of work completed per minute ? In 12,250 minutes 1,400 toys will be completed.
(6) At this rate how many units will they complete in a six-hour day ?
Hence, in completiog one toy time taken will be
Solution. (a) The average number of units per mioute(MBA. M.D. Univ.,1994) 12,250
calculating the harmonic mean. will be obtained by
1,400 -= 8minutes.
CALCULATION OF HARMONIC MEAN G However, if we assume that each worker works the same amouat of time but
produces ifferent number of toys, barmonic mean would be more appropriate. This
assumption is more true in practiçe (people workipg same anount of time but haviog
1|X ifferent output).
4 4x 60
0:250
0:200
H.M,- 1
356 minutes per toy
0167
10 0-100
Time required to complete 1,400 toys= 1,400 x48 -=9,600 minutes
12 0-083
7
E1|X=08 9,600
6Vertficatton, Each worker works for 4
-=2,400mioutes
N
H.M.= =625
08 Toys produced by Ain 2400 minutes-600
()
Hence the average number of units completed per 2400
minute=625. 6 =400
The average units per minute=016 2400
ln 6hours, Le, 360 minuters, total number of unils
C 99 99
10 240
by all the five workers combincd. produced will be 360 x 0.16x5= 288
2400
D . 160
15
"The harnouic mean is a measure of ceotral tendency for data expressed as
rates, such as kms. per hour, tonoes per day, kms. per litre, furlopgs per fortnight Total -1,400
ctc.
 128 BUSINESS STATISTICS MEASURES OF CENTRAL TENDENCY 129
Merits and Linitations of Harmonic Mean The process of computing the progressive average shall be clear with
The harmonic mean, like the arithmetic mean and geometric mean,. the help of the following examples.
iscomputed from all observations. It is useful in special cases for averag
ing ratcs.
ilurajon 24. Calculate progressive averaze from the foilowing data
Aear Sale of steel Year Sale of steet
(in m. tonnes) fin m toneS)
However, harmonic mean cannot be computed when there are both
positiveIt and
value. negative observations or one or more observations have zero
also gives largest weight to smallest observations and as such is.
1994
1995
12
14
1998
1999
25
22
not a good representation of a statistical series. In dealing with business
problems, harmonic mean is rarely uscd.
1996
1997
15
18
2000 30
Solution. CALCULATING PROGRESSIVE AVERAGE
Relationship among the Averages Year Sales Progressi ve Years Progresstve
In any distribution when the original observations differ in size, the (ln m. lonnes) (otals included averGre
values of A.M., G.M. and H.M. would also differ and will be in the: 1994 2 12 12 00
following order : 1995 14 26 2 1300
A.M.>G.M.>H.M. 1996 15 41 3 13 67
1997 18 59 4 475
i.e., arithmetic mean js greater than geometric mean
is grcater than harmonic mean. The cquality signs and geomctric mean
hold only if all tho
1998
1999
25
22
84 16 80
17.67
106 6
observations X,, X,..... XN are identical. 2000 30 136 1943
Progressive Average The progressive avarage makes it clear that the above conpany s stead1ly progress1ng
ycar aftcr ycar.
This average is also based upon the arithmetic mcan. Tho
features of this averagc aro : important Which Average to use?
(i) It is a cumulative average. In the
We have explained di fferent methods of computing the various types
previous figures are added and no previous çalculation of this average alI of averages and also their distinctive features. At this pint the reader can
CAse of moving average.* figuro is
left as is done in the question "which of these averages should I use'"? Or "which of these is
The progressive average for the first year would
remain the sume; the progressive average for the scecond yoar is cqual the best average to be used'"?
a+b to It must be clearly understood that no single average can be regarded
2for thethird year 4+b+c for the fourth year a4b+c+d ; and
SO on.
4 as best for all purposes. The following two considerations should be kept
in mind in the selection of an average
(i) The average value can be obtaincd for all
1. The type of data available. Are they badly skewed (avoid the mean ).
average, on the other hand, caonot the years. The moving gappy around the middle (avoid the median), or unequal in class interval
be çomputcd for all tho
longer the period of moving averago,
which the moving average cannot the greater tho number ofyears. The
years for (avoid the mode)?
be computed.
This average is generally used 2. The concept of the typical value required by the problem. Within
the framework of descriptive statistics the main requirement is to know
during the
of a business. For
cach successive yearexamplo,be the figures of sules,carly years of the working what each average means and then select one that fulfils the purpose on
entire previous period inmayorder compared with the profits or production
respective of
igures for the hand. Is a composite average of all absolute or relat1ve values needed
to find out how a busine8s is growing. (arithmetic mean or geometric mean)? Or, is a middle value needed (median ).
or the most common value (mode)?
On occasions, it may even be advisable to work out more than one
"for detaile see chap er on average and present them, although, to be sure, this procedure creates an
Busincss Porecusting and Timo Series Analysis, added burden for the reader as well as for the statistician. But the added
burden is preferable to the use of a single average that may prov1de an
incomplete deseription. To use it alone is like look1ng through akeyhole:
the part of the room you can see cannot give afullideaof the whole room.
 130 BUSINESS STATISTICS MEASURES OF CENTRAL TENDENCY 131
Arithmetic Mean
In the following cases arithmetic mean should not be used : GENERALLIMITATIONS OFANAVERAGE
1. In highly-skewed distributions. 1. Since an average is asingle value representing agroup of values, it must
2. In distributions with open-end intervals. be properly interpreted,otherwise, there is every possibility of jumping to
3. When the distribution is wrong conclusion. This can be best illustrated with the help of a story. A
small or large at irregular points. unevenly spread, concentration being person had to cross a river from one bank to another. He was not aware of the
4. When an average rate of growth or change over a depth of the river, so he enquired from another man who told him that the
is required. period of time
5. When the observations form a geomctric average depth of water is 160 cms. The man was 175 cms and he thought that
4, 8, 16, etc. progression, i.e., 1, 2, he can very easily cross the river because all the time he would be above the
6. When averaging rates (.e., spced, fluctuations water level. So he started. In the beginning the level of water was very shallow
articles, etc.). in the prices of
but as he reached the middle, the water was 500 cms deep and he lost his life.
7. When there are very large
arithmetic mean would be seriously and very small values of observations
misleading on account of undue
The man was drowned because he had a misconception that average depth
ence of extreme values. influ means uniform depth throughout. But it is not so. An average represents a
Leaving aside the above specific group of values and lies somewhere in between the two extreme values.
geometric menn or harmonic mean is more cases where either median, mode,
appropriate, in 2. An average may give us a value that does not exist in the data. For
should apply as a rule of thumb
and widely uscd average in practicc. the arithmetic mean-the other cases we
most popular 800
example, the arithmetic mean of 100, 300, 250, 50,100 is 5
= 160, a value
Median,. The median is
grouped distributions, cspecially generally the best average in open-end
that does not exist in the data.
one gets a Jor reverse J çurve, where if plotted as a frequency curve 3. Attimes an average may give absurd results. For example. if we are
or income distribution. In such for example,highin case of price distributi on calculating average size of afamily we may get avalue 4.8 .But this is impossible
cause the mean to be higher or cases very or very low values would
In such instances, the median lower than the most 'common"
may be more representative to valucs.
as persons cannot be in fractions. However, we should remember that it is an
describing the mass of data. use in average value representing the entire group of families.
Mode. Generally speaking, the significance of mode 4. Measures of central value fail to give us any idea about the formation of
that it can be used to describe qualitative lies in the fact
problems involving the data. The mode can be used jin the series,Two or more series may have the same centralvalue but may differ
ments are not possible. expression of preference where
Thus the preferred type of packagequantitative measure widely in composition. For example, observe the following two series.
a number of alternative designs design among
compare consumer preferences would be thc modal design. If we want to Series A: 150 170 190 210 280
kinds of advertising, we can for different kinds of products, or diflerent
different groups of people butcompare the modal preferences expressed by Series B: 300 500 20 78 102
Mode is a particularly uscful we cannot calculate the median or mean.
average for discrete
people wearing a series, e.g., number of In both series, average X = 200.
etc. The mode isgiven size of shoes, or
best suited where number of children per household,
5. We must renmember that an average is a measure of central tendency.
frequency. there is an outstandingly large
Hence unless the data showa clear-cut concentration of observationsan average
Geometric Mean.
ing index numbers, rates ofThe geometric mean is typically used is averag
change, ratios and other may not be meaningful at all. This evidently precludes the use of any average to
sets of data
in percentage form. It is
Statistics in index numberparticularly important in Economics andexpressed Business
typify a bimodal or a U-shaped or a J-shaped distribution.
construction.
Harmonic Mean. MISCELLANEOUS ILLUSTRATIONS
values of a variable are Harmonic mean is useful in problcms in which
vari able, i.e., rates, time, çompared with a constant quantity of anothcr lustration 25. Calculate mean, median and mode for the following data pertaining to
tities purchascd or sold perdistance covered within certain time and quan
unit, etc.
mar7s in Statistics out of 140 marks for 80 students in a class
Marks more than 0 20 40 60 l00 120
No. of Students 76 50 28 3
(MBA, MD niv., /994)
 BUSINESS STATISTICS MEASURES OF CENTRAL TENDENCY 133
132
MODE
Solution. CALCULATION OF MEAN, MEDIAN AND 600
(X-70)/20 Med. = Size of th observation = 2
= 300th observatíon
Marks No. of Students m.p.
X fd CJ. Median lies in the class 450 - 525
-3 -12
0-20
20-40 26
10
30 -2 -52
4
30
Med = L+ N/2-pe£xi
40-60 22 50 -1 -22 52 300 236
60-80 10 70 0 62 = 450 + x 75 = 450 + 23.2 = Rs. 473.2
207
80-100 90 +1 +9 71
110 +2 +12 OGIVE BY LESS THAN METHOD
100-120 6 77
600
120-140 3 130 +3 +9 80
N= 80 Efd = -56 500
Mean :} =A+2
N
x = 70 - 80
2 0 = 70 - 14 = 56
80
Median : Med = Size of th observation = =40th observation
2 300
Median lies in the class 40-60
150
Med = L+
NI2- p. c. f x /= 40 +
40-20
x 20 = 40+ 9.09 = 49.09
22
375 450 525 600 625 750 825
Mode : Since the highest frequency is 20, mode lies in the class 20-40. 473.2 WAGES (Rs.)
22
Mo = L+ x /= 20 + x 20 = 20 + 16.92 = 36.92 The median value as shown in the graph above is also Rs. 473.2.
22+ 4 0, and
Illustration 26. The following table gives the distribution of weekly (b)The limits of weekly wages of central 50 per cent of the workers shall be given by
600 workers of a factory: wages of
N 600
Weekly wages Frequency Weekly wages 2, - Size of th observation = =150th observation
(in Rs.) Frequency 4
(in Rs.)
Below 375 69 600-675 e, lies in the class 375 450.
375-450 167
450-525 207
675-750
750-825
24
10 , -L+ N/4-pcJ xi
525-600 65
(a) Draw an ogive for the above data and 150 69 x 75 = 375 + 36.38 = Rs. 411.38
calculated value.
thence obtain the medjan value. Check it against = 375 +
167
b) Obtain the limits of weekly wages of
central 50 per cent of the workerS. 3N 3x 600
e, = Size of th observation = 450th observation
Solution. (MBA, Delhi Univ., 1996) 4 4
CALCULATION OF MEDIAN
Weekly wages (in Rs.) 2, lies in the class 525-600.
Cum freq.
Less than 375 69
69 Q, =L+ 3N/4-p.c.J xi
450 167
525 236
207
600 443 = 525 +
450 443-s25+ 8.08 =Rs. 533.08.
65
675 508 65
58
cent workers are Rs. 41 .38 and
Hence the limits of weekly wages of central S0 per
750 566
24
" 825 S90
10 Rs. 533.08.
600
 BUSINESS STATISTICS MEASURES OF CENTRAL TENDENCcY
134,? 135
skilled, 250 semi-skilled and
150 unskilled
Illustration 27. In a factory, there are l00 yOven Illustration 29. In 500 small-scale industrial units, the retum on investment ranged from
average a unit lenght of a particular fabric is
workers. It has been observed that on an o to 30 per cent; no units sustaining any loss. 5 per cent of the units had retums ranging from 0
workers in 4 hours and by an unskilled worker
a skilled worker in 3 hours, by semi-skilled workers are expected to become skilled anda
per cent to (and including) 5per cent, and 15 per cent of the units camed retums exceeding 5 per
hours. After a training of 2 years. the semi-skilled be required after 2 yeare cent but not exceeding 10 per cent. The median rate of return was 15 per cent and the upper
unskilled workers to become semi-skilled. How much less time will quartile 20 per cent. The uppermost layer of returns exceeding 25 per cent was earned by 50
training for weaving unit length of fabric by an average worker? units.
Solution. Average time pcr worker before training is. () Present the information in the form of a frequency table as follows
(100 x 3) + (250 x 4) + (150 x 5) 2050 Exceeding 0 per cent but not exceeding 5 per cent
= 4.1 hours. 10
100 + 250 + 150 500 10 15
Now afer training the compositon of workers is as follows and so on.
Skilled-workers =100+250=350 (ii) Find the rate of return around which there is maximum concentration of the units.
Semi-skilled workers = 150
Unskilled workers = Nil Solution. (i) FREQUENCY TABLE
Average time per worker after training is : Rate of Return Industrial Units
(350 x 3) + (150 x 4) 1050 + 600
= 3.3 hours.
350 + 150 S00 Exceeding 0 but not exceeding 5 per cent 500 = 25
100
Afer 2 years 0.8 hour less would be required. 15
Note. An assumption has been made that there has been no turmover of workers. 5 10 500 ×
100
= 75
Niustration 28. A Limited Company wants to pay bonus to the members of the staff. The
bonus is to be paid as under : 10 15 250 - 100 = 150
Weekly salary (Rs.) Bonus (Rs.) 15 20 ::: 375 - 250 = 125
300 but not exceeding 400 100 500 - 375 - 50 = 75
400 20 25
500 120
500 600 140 25 30 50
600 700 160
700 800 Total = 500
180
800 900 200
900 1.000 220
(ii) For finding out the rate of return around which there is maximum concentration of the
1,000 1,100 units, we will calculate mnode.
240
Actual salary drawn by the employees is given below : Mode lies the class 10-15.
325. 378, 420, 455, 620, 660, 680, 725, 863,
832, 942, 952, 800, 1002,
763, 1,028, 1,100, 610, 625, Mode L+
382, 540, 463, S78, 723, 690.
How much the company would need to pay 125) = 25; i = 5
bonus paid per member of the staff? by way of bonus? What shall be the average L=10; A, = (150-175) = 75; A, = (150
Solution. For determining the figure of total and average (MBA, Jodhput Univ., 1995) 75
of persons in each salary group. bonus, first ascertain the number Mode = 10 + x5 = 13.75.
75+25
CALCULATION OF TOTAL AND AVERAGE
BONUS Hence the maximum concentration is around 13.75 per cent retums.
Weekly salary Frequency Bonus (Rs.)
(Rs.) Total Bonus lustration 30. The mean monthly salary paid to all employees in a company is Rs. l600.
300-400 The mean monthly salaries paid to technical and non-technical employees are Rs. 1800
400-500 3
100
120
300 and Rs. 1200 respectively. Determine the percentage of technical and non-technical employees
500-600 2 360 of the company.
600-700 140 280
6
700-800 160 960
Solution. Let percentage of technical personnel be denoted by X.
3
800-900 180 540
900-1.000
3
200 Non-technical employees would be (100 - X%.
2 600
1,000 and above 220 440
3
240 720
N= 25 N+ N2
The company would necd Rs. EX = 4,200
4,200 to pay bonus. Xa=l600, X,=1800, X; =1200
Average bonus per member = 4,200
25 = Rs. 168. Let N,=X :: N, = (100 X)
 BUSINESS STATISTICS MEASURES OF CENTRAL TENDENCY 137
136
DETERMINING AVERAGE RATE OF DEPRECIATION
Substituting the values
1800X +(100 X)1200 Diminishing value
1600 =
100
Year for a value of Rs. 100) Log X
X
160000 = 1800 X+ 120000- 1200X
or 600 X = 40000 1996 100 - 40 = 60 1.7782
1997 100- 25 = 75 1.8751
40000
X= =66.67 1.9542
600 1998 100 - 10 = 90
1.9542
Hence percentage of technical employees = 66.67 and percentage of non-technical 1999 100- 10 =90
100 - 10 = 90 1.9542
employees = 100 - 66.67 = 33.33. 2000
lustration 31. A company invests Rs. 1. lakh at 10% annual rate of interest. What wil) E log X =95159
be the total amount after 6 years if the principal is not withdrawn ?
Solution. Applying the compound interest formula : E logX
lGM.- AL AL = AL 1.9032 = 80.
Thé diminishing value being Rs. 80, the depreciation will be 100 - 80 = 20%.
P, = 1,00,000, r= 10, n=6 Illustration 34. The following is the age distribution of 1,000 person working in a large
industrial house :
Age group No. of persons Age group No. of persons
P, =1,00,00 1 100 20--25 30 45-50 105
Taking logarithms 160 50-55 70
25-30
30-35 210 55-60 60
log P = log 1,00,000 +6 log 180 60--65 40
35-40
40--45 145
= log 1,00,000 + 6 log 110-6 log 100
= 5+ (6 x 2.0414) -(6 x 2) = 5.2484 Due to continuous losses, it is desired to bring down the strength to 30% of the present
P = AL 5.2484 = 1,77,200. number according to the following scheme :
Thus the total amount after 6 years would be Rs. 1,77,200. (i) To retrench the first 15% from the lower group.
Illustration 32. In a certain factory a unit of work is (ii) To absorb the next 45% in other branches.
in 15 minutes, byC in 12 minutes and by D in 20 minutes. completed by A in 10 minutes, by B (ün To make 10% from the highest age group retire permanentily. if necessary.
(i) What is the average number of units of work completed per Calculate the age limits of the persons retained and those to be transferred to other
(i) At minute ?
departments. Also find the average age of those retained.
this rate how many units will they complete in an
Solution. The average number of units of work per minute 8-hour day ?
will be obtained by.finding out Solution. The first 15% to be retrenched are from the lower group : hence their total
the harmonic mean
15x 1,000 = 150. 30 belong to 20 - 25 age group and rest, ie., (150-30), Le..
4x 120 40 number Comes to
100
H.M.=T1,L
10 15 12 20
36 3
120 belong to 25--30 age group. The next 45% are to be absorbed in other branches. They are
40 45
Hence units per minute is the average rate. 1000 x
100 = 450. They belong to the following age groups :
1 3 Age groups No. of persons
The average units per minute is 40 A0
25--30 40
3
(b) In 8 hours, i.e., 480 minutes, the total number of units 30--35 210
by all the four workers combined. produced will be 480xx4=144 35--40 180
40
Illustration 33. A machine was purchased for Rs. 10 lakhs in 40-45 20
diminishing balance was charged @ 40% in the first year, 25% in the 1996. Depreciation on the
annum during Those to retire are 10% and belong to the highest age group. Their number comes to 1.000
period?
the next three years. What is the average depreciation second year and 10% per
charge during the whole 10
Solutlon. Since we are interested in finding 100 = 100 and their age groupS are :
mean will be the out the average rate of
most
in the rate calculation. appropriate average. The cost of machine can be ignored depreciation, geometric Age groups No, of persons
as it is immaterial 40
60-65
55-60 60
 138 BUSINESS STATISTICS MEASURES OF CENTRAL TENDENCY 139
AVERAGE AGE OF THOSE RETAINED
2-pcf
(X-47.5)/5
Med. =L+
Age groups m.p. No. of
persons
d fd |15-20
33.5 = 30 + x10
40---45 42.5 125 -1 -125
45--50 0
3.5f; = 950-10,
47.5 105
f = 1,900-20f,
50--55 52.5 70 +1 +70 f, + 20f,= 1,900 ...(i)
N= 300 Efd = -5S From Eqns. (i) and (ii)
S+ 5f, = 400
55
xi =47.5 - x5 = 47.5 - 0.92 = 46.58. f, + 20f, = 1,900
N 300 Multiplying Eqn. (i) by 4.
Mence the average age of those retained is 47 years approximately. 4f+20f, = 1,600
llustration 35. The median and mode of the following wage distribution are Rs. 33.5 and ,+20f, = 1,900
Rs. 34 respectively. Howevet, three frequencies are missing. Determine their values. -3f, = -300 or f, = 100
Wages 0-10 10-20 20--30 30-40 40-50 5060 60-70 Total Substituting the value of f, in Eqn. (1),
(in hundred Rs.) 100 + 5f, = 400
Frequencies 4 16 ? 6 4 230 5f, = 300 or f, = 60
Solution. Let the missing frequencies be fo, f, and f, corresponding to classes 20-30, Since f+f, +f, = 200
S, = 200-100-60=40
30-40 and 40-50 respectively. Since median and mode are 33.5 and 34, they lie in the
Hence the missing frequencies are
class 30-40. The frequency of this class is f,.
S, = 60, f, = 100, f, = 40.
DETERMINING MISSING VALUES IIlustration 36. Calculate the arithmetic mean and the median of the frequency
between
distribution given below. Hence calculate the mode using the empirical relation
Wages Frequency Cum. frequency the three:
(in hundred Rs. ) No. of students
Height (in cms.) No. of students Height (in cms.)
5 150-154 17
0-10 4 4 130-134
135-139 15 155-159 10
10---20 16 20 160-164
140--144 28
20-30 ? 145149 24
20 +fo
Solution. CALCULATION OF ARITHMETIC MEAN AND MEDIAN
30-40 ?
20 +f,+5 (X-147)/S
40-50 Height (in cms.) m.p. cf
20 +, +f t, X d fd
50-60 6 129.5-134.5 132 5 -3 -15
226 20
134.5-139.5 137 15 -2 -30
60-70 4 48
230 28 -28
139.5-1 44.5 142
0 72
N= 230 0144.5-149.5 147 24 0
152 17 +1 +17
149.5-1 54.5 99
S+ +f, = 230 - (4 + 16+ 6+4) = 200 154.5-159.5 157 +2 +20
+3 100
s- 200 - (, +) = 200--5, 159.5-164.5 162
N= 100 Lfd = -33
Mode = Lt 2f-fo-f, 33
x5 = 147-1.65 = 145.35
Mean : Mean ù = A+ x i= 147
100
34 = 30 +
2f-fo-(200-f% -f,) xl0 Median : Median = Size of
N 100
= 50th observatìon.
4 2
10 2f-fo - 200+ f% +f Median lies in the class 144.5-149.5.
50- 48
10/, - 10,= 4(3/, - 200) or 10/, - 10, =12f, - 800 Median = L+
NI2-p. c.f
x /= 144.5 + x§ = 144.5 + 417 = 144.917.
24
-/,-10f, =- 800 or f, +Sf, =400. ...(() Mode : Mode = 3 Median-2 Mean = (3 x
144.917) - (2x145.35) = l44.051.
140
BUSINESS STATISTCs MEASURES OF CENTRAL TENDENCY 141
Ilustration 37, Incomes of employecs in an industrial concero are
given
Tbe total ipcome of the 10 employees in the class over Rs. 2,500 is Rs. below, tiustration 38. Calculate the median and mode for the distribution of the weights of
Computethe mean income. Everyemployee belonging to the top 259% of the ear30,000, 150 students from the data given below:
iB required to pay s% of his income to workere' relief fund. 1 Estímate the ers Weight (in kg): 30-40 40-50 50-60 60-70 70-80 80-90
bution to this fund. contri- 18 15
Income (Rs.) Frequency, Income (Rs.) Frequency 37 45 27
Below 500 90 1500-2000 Frequency
80 Solutlon. CALCULATION OF MEDIAN AND MODE
500--1000 150 2000-2500 70
1000--1500 J00 2500 and over 10 Weight (kg) cf.
Solation. COMPUTATION OF MBAN 30-40 18
Income (R$.) fx 40-50 37 S5
50-60 45 100
0-500 250 90
S00-1000 22500 127
750 150 60-70 27
1000--1500 1250 100
112500
1500--2000 1750 80
125000 70-80 15 142
2000-2500 2250 70
140000
2500 and over 3000 (given) 157500 80-90 8 150
10 30000
N=500 N = 150
E/X=$87500
EfX_587500 l175. 150
S00 Median = Size of th observation = = 75th observation
2 2
Number of employees belonging to the top 25%
of the earners is 25 Median lies in the class 50-60
abd the distribution of these top i00 × 500>125
follows : earners as obvious from the above 75-55
Distribution of top 25% carners
table is as
bMedian = L + NI2-p. c.f x /= 50 + x 10 = 50 + 4.444 = 54.44
f 45
Income (Rs.)
2500 and over Frequency Mode : Since highest frequency is 45, mode lies in the class 50-60.
2000--2500 10
1500-2000 70 8
4 Mode = L+ -x /= 50 + x 10 = 50 + 3.08 = 53.08.
income45 level
persons are to be taken in the last
class, ie., 1500--2000 with 8+18
frequencies arestarting
e from 2000 and
equally below. the highest
u follows : distributed throughout theUnder
clas8, tbe
the assubpa that the
he Illustration 39. Following distribution gives the pattern of overtime work per week
80 persons have
income in the range done by 100 employees of a company. Calculate median, first quartile, and 7th decile.
45 persoDs have 2000-1500=Re. S00
income in the range Overtime hours 10-15 15-20 20-25 25-30 30-35 35-40
S00
90 x45=281-25 or 281 No. of employees : 20 35 20 6
Since we are interested in th top 45 (MBA, Kurukshetra Univ., 1997)
their salaries
The
will tange
distribution from (2000-281) toearaers i.e.,
iu the income
1719 to 2000.group 1500-2000, Solution. CALCULATION OF MEDIAN. Q, AND D,
of top 125 persons is 2000,
as follows :
Income (Rs.) M.p.
Overtime (hrs.) f cf.
f Total Income
2500 and over 10-15 11
2000--2500 15-20 20 31
1719-2000 2250 70 30000 (given)
1859+5 45 157500 20-25 35 66
83677-5 25-30 20 86
N125
Hence tbe total ipcome sfX-271177s 8 94
5% of the top 25% of 30-35
Contribution to the earners is
Rs. 271177:5
fund0-05x271177:5-Rs. 13558-88.
35-40
N= 100

6 100
N 100
Med. = Size of th observation = = 50th observation
2 2
142 BUSINESS STATISTICS MEASURES OF CENTRAL TENDENCY 143
Median lies in the class 20-25 N 675
Median : Med. = Size ofth observation = 337.5th observation.
S0-31 2 2
Med. = L + NI2-p. c.f xi= 20 + x 5 = 20 + 2.714 = 22.714 Median lies in the class 40--50.
35

N 100 Med. =L+ NI2-p eL

2, = size of th observation = = 25th observation
4 L= 40, NI2 =337.5, p.c.f. = 201, f= 180, i= 10
0, lies in the class 15-20 337.5- 201
Med, = 40 + x10
180
2,- L+ N74-p. c.yi15+ 20
x 5= 15 + 3.5 = |8.5 = 40+7.58 = 47.58
Mode : By inspection mode lies in the class 40-50.
7N 7x 100
D, = Size of th observation = = 70th observation Mode L+
10 10
D, lies in the class 25-30 L= 40, A, = 180 - 106 = 74, A, = 180 - 164 = 16, i= 10.
74
Mode = 40 + 10 40 + 8.22 = 48.22
D, = L+
7N/10- p.c.f. 70- 66 74 + 16
xS = 25 + x 5= 25 + | = 26.
20 PROBLEMS
Illustration 40. In an examination of 675 candidates the examiner supplied the following 1. What are the measures of central tendency? Why are they called measures of central
information : tendency?
Marks
2. Give a brief descripion of different measures of central tendency. Why is arithmetic
No. of Marks No. of mean so popular?
obtained candidates obtained candidates 3. (a) Is it necessarily true that being above average indicates that someone is superior?
Less than 10%
Explain. (MBA, DU. 1995)
Less than 50% 381
Less than 20%
(6) What are quartiles of a distribution? Explain their uses.
39 Less than 60% 545 4. Define arithmetic mean and median and discuss their merits and demerits as measures
Less than 30% 95 Less than 70% 631 of central tendency.
Less than 40% 201 Less than 80% 675 5. How would you account for the predominant choice of arithmetic mean as a measure
Calculate the mode and median of the percentage marks obtained.
of central tendency? Under what circumstances would it be appropriate to use mode or median?
(MBA, Delhi Univ., 2000)
6. Comment on the following :
(MBA, Rohilkhand Univ., 1997)
() Ir first and third quartiles are 20 and 40 respectively the median will be 30.
Solution. CALCULATION OF MEDIAN AND MODE (ü) if daily wages paid to men and women employed in a factory are Rs. 100 and Rs. 90.
the average wage per worker would be Rs. 90.
Marks No. of,m.p. (X-45) / 10 (i)Aman claims that his average bank balance during the year is Rs. 3700. The bank,on
(in %) candidates f X the other hand, claims that he overdrew his acocunt at least 10 times during the year and as
fd c.f. such his claim is false.
0-10 7 -28 (iv) The increase in the price of commodity x is 20%. Then the price decreased 259% and
10--20 32 15
again increased 15%. The resultant increase in the price is 10%.
-96 39 (v) The mode of a distribution cannot be less than the arithmetic mean.
20---30 56 25 -2 t (vi) lf 2,, ,. ,, be respectively the lower quartile, the median and the upper quartile of
-112 95
30-40
a distribution, then ,-2, = ,-0,
106 35 -1 -106 201 (vii) Arithmetic mean is the best measure of central tendency.
40---50 180 45
381 7. What is a statistical average? What are the desirable properties for an average to
50.--60 164 possess? Mention different types of averages and state why the arithmetic mean is the most
+] +164 545 commonly used amongst them.
60--70 86 6S +2 +172 631 8. (a) What are the essential requisites of a good measure of central tendency? Compare
70--80 44 75 +3
and contrast the commonly employed measure in terns of these requisites.
+132 675 (b) Prove that the arithmetic mean of two positive numbers a and b is at least as large as
their geometric mean.