MOMENTS

Moments are a very useful statistical tool as they enable us to describe a set of data for its
central tendency, dispersion, skewness and kurtosis. Before discussing their usefulness, let us
define moments and look into their computational aspects. A moment may be defined as the
mean of the deviations of a variate from some value raised to a certain power.

Calculation of Moments
To calculate moments, the following steps are involved:
(i) The mean of the data is calculated, if moments from mean are to be computed. Else, we
consider the given value, A, about which moments are required to be calculated.
(ii) The deviations of each of the values from the mean (or A, whichever relevant) are taken
(iii) The deviations are raised to a certain power, r.
(iv) The powered deviations are added up and the total is divided by n, the number of values
in the data.
The moment calculated by using a given power r is referred to as the rth moment. To
illustrate, when the deviations are raised to power 3, the calculated moment is termed as the
third moment, and when the power taken is 4, the corresponding moment is called the fourth
moment. It may be mentioned that although moments can be calculated by raising deviations
to any power, we are concerned with only the first four moments.

Deviations Taken from Mean

For a given set of data, when moments are calculated by measuring deviations from its mean,
the moments are said to be moments about mean or central moments. They are represented by
the Greek letter μ. Thus,
2 4
Σ ( X−X ) Σ ( X −X ) Σ ( X− X ) Σ ( X− X )
μ1= , μ 2= , μ 3= , and μ 4 =
n n n n
Here μ1, μ2, μ3 and μ4 are, respectively, the first, the second, the third and the fourth central
moments. Further, n refers to the number of values in the data.

Similarly, in case of frequency distributions we have,

Σ f ( X− X ) Σ f ( X− X )2 Σ f ( X− X )3
μ1= , μ2= , μ3 = ,
Σf Σf Σf
4
Σ f ( X −X )
and μ4 =
Σf
It may be noted here that
1. Since the sum of deviations from arithmetic mean is equal to zero, so that Σ ( X− X )=0 ,
the first central moment, μ1, is always equal to zero, and
2. The second moment, μ2, is essentially equal to the variance, the squared standard
deviation.
Deviations Taken from Arbitrary Origin, A
Moments can be calculated by measuring deviations from an arbitrary point, A. such
moments are called moments about A (for example, moments about 6 if A = 6, moments
about 10 if A = 10, and so on) and are symbolized by the Greek letter ν (nu). Thus, with d =
X – A, we have
2 3 4
Σd Σd Σd Σd
ν1 = , ν2= , ν 3= , and ν 4 =
n n n n
For frequency distributions, when deviations are taken (d = X – A):
2 3 4
Σ fd Σf d Σf d Σf d
ν1 = , ν 2= , ν 3= , and ν 4 =
Σf Σf Σf Σf
Further, in case step-deviations are taken [where d’ = (X – A)/C], we have
' '2
Σf d Σf d 2
ν1 = ×C , ν 2= ×C ,
Σf Σf
Σ f d'3 3 Σ f d'4 4
ν3 = ×C , and ν 4= ×C
Σf Σf

It may be noted that since A can be assigned any value, innumerable four-moment value sets
can be derived for a given set of data. However, a significant value that A may be assigned is
zero. When A = 0, the calculated moments are known as moments about origin (or moments
about zero). The first moment about zero (or origin) is equal to mean.

Example 1 For the following data, calculate

(a) Central moments
(b) Moments about 6, and
(c) Moments about origin
9, 4, 6, 9, 11, 13, 8, 4

The calculation of moments is given below. The inputs are shown in Table 1.

Table 1 Calculation of Moments

Deviations from mean, X =8 Deviations from A = 6
X 2 3 4
( X −X ) ( X −X ) ( X −X ) ( X −X ) d d2 d3 d4
9 1 1 1 1 3 9 27 81
4 −4 16 −64 256 −2 4 −8 16
6 −2 4 −8 16 0 0 0 0
9 1 1 1 1 3 9 27 81
11 3 9 27 81 5 25 125 625
13 5 25 125 625 7 49 343 2,401
8 0 0 0 0 2 4 8 16
4 −4 16 −64 256 −2 4 −8 16
64 0 72 18 1,236 16 104 514 3,236
 Σ X 64
X= = =8
n 8
(a) Central moments:
Σ(X −X ) 0
μ1= = =0
n 8
μ2=Σ ¿ ¿
μ3=Σ ¿ ¿
μ4 =Σ ¿¿
(b) Moments about 6:
Σ d 16
ν1 = = =2
n 8
2
Σ d 104
ν 2= = =13
n 8
3
Σ d 514
ν3 = = =64.25
n 8
4
Σ d 3,236
ν 4= = =404.5
n 8
(c) Moments about origin:
They imply that A = 0. In such a situation, d would be same as X. Accordingly,
Σ X 64
ν1 = = =8
n 8
2 2 2 2
Σ X 9 +4 + …+ 4 584
ν 2= = = =73
n 8 8
3 3 3 3
Σ X 9 + 4 +…+ 4 5,842
ν3 = = = =730.25
n 8 8
Σ X 4 9 4 +4 4 + …+4 4 62,228
ν4= = = =7,778.5
n 8 8
Notice that the first moment about mean is zero while the first moment about zero is equal to
mean.

The moments of a set of values are calculated by taking deviations from X or some
value A, raising them to powers 1, 2, 3 and 4 and then dividing their respective totals by
n. This gives four moments. Measuring deviations from X yields central moments while
others are called moments about (arbitrary value) A. If A = 0, the moments are termed
moments about origin. The two sets of moments are, expectedly, interrelated.

Conversion of Moments about Arbitrary Origin to Central Moments

We have seen that moments can be calculated about any given value in the same way as the
moments about mean. It is easy to visualise that when moments are calculated about mean
and an arbitrary point for a given set of values, they are expected to be related to each other in
some way. In fact, the moments about an arbitrary origin can be used to obtain central
moments. This is done by using the following formulae:
μ1=ν 1−ν1
 2
μ2=ν 2−ν 1
μ3=ν 3−3 ν 2 ν 1 +2 ν 13
μ4 =ν 4−4 ν 3 ν 1 +6 ν 2 ν12−3 ν 14

Conversion of Central Moments to Moments About an Arbitrary Origin

The given central moments for a set of data can be used to calculate the moments about an
arbitrary point. However, to be able to do so, the arithmetic mean of the data should also be
known along with the arbitrary point about which moments are to be calculated. With d
defined as equal to X – A, the required moments are obtained using the following
relationships:
ν1 =μ 1+ d
2
ν 2=μ2+ d
ν3 =μ3 +3 μ 2 d + d3
ν 4=μ 4 +4 μ 3 d +6 μ2 d 2 +d 4

Thus, if X > A, then d is positive and if X < A then d is negative. Further, if A = 0, then the
moments so obtained would be the moments about origin.

Example 2 (a) Using the moments about 6 calculated in Example 1, obtain central moments.
Verify that the values are the same as obtained earlier.
(b) From the central moments in (a), compute moments about (i) 6, and (ii) 10.

(a) The first four moments about 6 as calculated earlier are 2, 13, 64.25 and 404.5. Using
these values, the central moments are calculated here:
μ1=ν 1−ν1 =2−2=0
2 2
μ2=ν 2−ν 1 =13−2 =9
3 3
μ3=ν 3−3 ν 2 ν 1 +2 ν 1 =64.25−3 ×13 ×2+2 ×2 =2.25
2 4 2 4
μ4 =ν 4−4 ν 3 ν 1 +6 ν 2 ν1 −3 ν 1 =404.5−4 ×64.25 × 2+ 6 ×13 ×2 −3 × 2 =154.5
(b) Moments about 0 are 0, 9, 2.25, and 154.5, while mean = 8.
(i) For A = 6, d = 8 – 6 = 2. Thus,
ν1 =μ 1+ d=0+ 2=2
2 2
ν 2=μ2+ d =9+2 =13
3 3
ν3 =μ3 +3 μ 2 d + d =2.25+3 ×9 × 2+ 2 =64.25
2 4 2 4
ν 4=μ 4 +4 μ 3 d +6 μ2 d +d =154.5+4 × 2.25 ×2+6 × 9× 2 + 2 =404.5
(ii) For A = 10, d = 8 – 6 = −2. Accordingly,
ν1 =μ 1+ d=0−2=−2
ν 2=μ2+ d 2=9+ (−2 )2=13
3 3
ν3 =μ3 +3 μ 2 d + d =2.25+3 ×9 × (−2 )+ (−2 ) =−59.75
2 4 2 4
ν 4=μ 4 +4 μ 3 d +6 μ2 d +d =154.5+4 × 2.25 × (−2 ) +6 × 9× (−2 ) + (−2 ) =368.5

Sheppard’s Corrections of Moments for Grouping Error

When data are grouped in to frequency distributions, the values lose their individual identity.
For computing mean, standard deviation, and moments etc it is assumed that the frequencies
are concentrated at the mid-points of the class intervals. But this assumption need not be
correct because mid-values of some of the classes may exceed their respective mean values
while for some others the reverse may be true. As a consequence of the assumption of the
mid-point concentration, the calculation of moments from a grouped frequency distribution
introduces errors arising out of grouping.
When arithmetic mean of a grouped data involving continuous variable and symmetrical
distribution is computed, the grouping errors tend to balance, leaving the result approximately
to what would be obtained using ungrouped data. A similar balancing occurs when the third
moment is computed because the deviations retain their signs when cubed. On the other hand,
no such balancing occurs when computing second and fourth moments because negative as
well as positive deviations are transformed into positive values when they are squared or
raised to fourth power. Accordingly, the values of these moments tend to be slightly larger
than their values computed from the raw scores. Thus, the assumption of concentration of
frequencies at mid-points imparts bias to the calculation of moments of the even order.
Sheppard articulated a set of corrections for doing away with this upward bias in the
moments. According to these corrections,
1 2
^μ2=μ 2− i
12
1 2 7 4
^μ4 =μ4 − μ2 i + i
2 240
where ^μ2 and ^μ4 the corrected second and fourth moments respectively and i is the class-
interval size.
Sheppard’s corrections are applicable only when the following conditions are satisfied.
1. The distribution should involve a continuous variable.
2. It should have finite range.
3. The distribution tapers off to horizontal axis such that there is a high-order contact at
the terminals of its range.
4. The distribution involves a high value of n.
Thus, a distribution has to be symmetrical, or closely so, for Sheppard’s corrections to work
effectively. To the distributions that are skewed or J-shaped, where there is no high order
contact on both the ends, Sheppard’s corrections should not be applied because they would
operate in wrong direction and may over-correct.

The following points should be noted.

(a) The Sheppard’s corrections aim at removing definite bias and do not make up for
general inaccuracies introduced by having too large groups. Also, they do not remove
the bias in its entirety but bring the results closer to reality.
(b) The corrections do not justify their use if the original data consist of observations that
represent rather approximate measurements of the items.

Example 3 For the frequency distribution given below,

(a) Calculate moments about 115.
(b) Obtain central moments.
 (c) Apply Sheppard’s corrections to the central moments.

Class interval Frequency

70 – 80 12
80 – 90 18
90 – 100 35
100 – 110 42
110 – 120 50
120 – 130 45
130 – 140 20
140 – 150 8
Total 230

The calculations are shown here.

(a) Moments about 115: The calculations for obtaining moments are given in Table 2, where
step-deviations are taken with A = 115 and C = 10.

Table 2 Calculation of Moments

Class 2
X f d' fd ' f d'
3 4
' '
fd fd
interval
70 – 80 75 12 −4 −48 192 −768 3,072
80 – 90 85 18 −3 −54 162 −486 1,458
90 – 100 95 35 −2 −70 140 −280 560
100 – 110 105 42 −1 −42 42 −42 42
110 – 120 115 50 0 0 0 0 0
120 – 130 125 45 1 45 45 45 45
130 – 140 135 20 2 40 80 160 320
140 – 150 145 8 3 24 72 216 648
Total 230 −105 733 1,155 6,145

Now,
'
Σf d −105
ν1 = ×C= × 10=−4.57
Σf 230
Σ f d'2 733
ν 2= ×C 2= ×10 2=318.70
Σf 230
'3
Σf d 3 −1,155 3
ν3 = ×C = × 10 =−5,021.74
Σf 230
Σ f d'4 4 6,145 4
ν4= ×C = ×10 =2,67,173.91
Σf 230
(b) Central moments are shown calculated here:
μ1=ν 1−ν1 =−4.57−(−4.57 )=0
μ2=ν 2−ν 12=318.70−(−4.57 )2=297.82
3 3
μ3=ν 3−3 ν 2 ν 1 +2 ν 1 =−5,021.74−3 × 318.70× (−4.57 )+2 × (−4.57 ) =−843.25
2 4
μ4 =ν 4−4 ν 3 ν 1 +6 ν 2 ν1 −3 ν 1
2 4
¿ 2,67,173.91−4 × (−5,021.74 )(−4.57 ) +6 × 318.70× (−4.57 ) −3 × (−4.57 ) =2,14,004.10

(c) Sheppard’s corrections applied to the second and fourth moments are shown here:
1 1
^μ2=μ 2− i 2=297.82− ×102=289.48
12 12
1 2 7 4 1 2 7 4
^μ4 =μ4 − μ2 i + i =2,14,004.10− ×297.82 ×10 + ×10 =1,99,405
2 240 2 240

Significance of Moments
The first four moments are very useful in a study of the different aspects of the formation of a
frequency distribution, namely central tendency, dispersion, skewness and kurtosis. As
already indicated, we have first moment about origin equal to arithmetic mean and the second
moment about mean equal to variance. These moments help us, thus, to have an idea about
the central tendency and the scatter of observations about the central tendency. The moments
are profitably employed to study also the shape of the distribution in terms of the skewness
and kurtosis. The skewness and kurtosis are studied with the help of beta and gamma
coefficients. The gamma coefficients are referred to as g-statistics.

Moments provide a useful mechanism to study different characteristics of a

distribution. The first moment about zero gives arithmetic mean; the second central
moment is equal to the variance; while two beta coefficients β 1 and β2, and the g-
statistics based on these coefficients provide an insight into the skewness and kurtosis
of a distribution.

Moments and Skewness

The third central moment is employed as a measure of skewness. If a distribution is
symmetrical, positive and negative deviations balance and cancel out in moments of odd
order, so that the third moment is equal to zero. Any balance of the histogram to the right or
left will be reflected in the third moment which will be positive or negative respectively. The
third moment about mean is thus a measure of absolute skewness in the distribution.

A relative measure of skewness is given by beta-one (β1) is defined as under:

2
μ3
β 1= 3
μ2

It is essentially an index of skewness. Its value is equal to zero for a symmetrical distribution
and the greater its value the higher the extent of skewness in the distribution. It may be
observed that this coefficient cannot assume a negative value because neither the numerator
nor the denominator can be a negative quantity.

The other relative measure of skewness is gamma-one ( g1 ), defined as follows:

μ μ
g1= 3 3 ∨ 33
√ μ2 σ
Evidently, it is equal to square root of β1. However, unlike β1 it takes negative as well as
positive values. Its sign is determined from the sign of μ3. If g1 is negative, the skewness in
the distribution is negative, and if g1 is positive, the skewness is also positive. Like for β 1, if
g1 is zero, it indicates a symmetrical distribution.

Moments and Kurtosis

Moments are used also in studying the peakedness of a distribution. The fourth moment about
mean is used an absolute measure of kurtosis. However, one cannot say whether the
distribution is leptokurtic, mesokurtic or platykurtic by looking at the fourth moment, μ 4. To
be able to say so, μ4 must be considered in relation to second moment of the data.
Nevertheless, it can be shown why μ 4 measures kurtosis. We know that in a peaked
distribution, there are relatively large numbers of items which deviate widely from the mean
whereas, on the other hand, in a flat-topped distribution such large deviations are few. Large
deviations from mean tend to have bearing on values of the second and fourth moments of a
distribution. The deviations are squared in computing the second moment whereas they are
raised to fourth power for calculating the fourth moment. The latter process enhances the
importance of the deviations in the calculations more than the former. For example, in a
given set of data, there be a small deviation of 3 and a large deviation of 12 and these are
squared up, these would result in 9 and 144. That is to say, the second is 16 times greater than
the first one. But when they are raised to the fourth power, the result is 81 and 20,736, which
means that the second is more than 250 times the first. Thus, the presence of such large
deviations in the data in addition to other smaller deviations would affect the fourth moment
more than it would affect the second. Evidently, large deviations are greatly magnified by
raising them to the fourth power and if there is an excess of large deviations, the distribution
will be leptokurtic. On the other hand, the more flat-topped the distribution, the lower is the
number of large deviations about mean and, consequently, the less high is the fourth moment
compared to the second moment.

A measure involving the fourth and the second moments, for relative kurtosis in a distribution
is β2. We have,
μ
β 2= 42
μ2
For a mesokurtic distribution, β2 = 3, while for a leptokurtic distribution β2 > 3 and for a
platykurtic distribution β2 < 3. Obviously, the higher the value of β2, the more peaked is the
curve of the distribution and the smaller its value the more flat-topped is the curve.

The gamma measure, g2, based on the β2 coefficient is defined as follows:

 g2 = β2 – 3

If g2 = 0, it indicates a mesokurtic distribution, while if g2 > 0, the distribution is leptokurtic

and if g2 < 0, it is platykurtic.

Example 4 Using the central moments calculated in Example 3, calculate the beta and the
gamma coefficients. Comment on the results. For your calculations, use the corrected values of
the moments.

The four moments obtained earlier are 0, 289.48, −843.25 and 1,99,405. On the basis of
these, we have

Beta coefficients:
μ 32 (−843.25 )2
β 1= 3 = 3
=0.029
μ2 ( 289.48 )
μ4 1,99,405
β 2= 2 = =2.380
μ 2 ( 289.48 )2
Gamma coefficients:
μ −843.25
g1= 33 = =−0.171
σ √ ( 289.48 )3
μ4 1,99,405
g2= 2 −3= −2.380=−0.620
μ2 ( 289.48 )2
On the basis of the results obtained, it may be observed that the given distribution is slightly
negatively skewed and it is platykurtic since β2 < 3 (alternately, g2 < 0)

Example 5 For a distribution, the first four moments about 20 are 8; 212; 1,510 and 28,214
respectively. Calculate moments about 30 for this distribution.

This will be done by first converting the given moments into central moments and there from
to the required moments.
Step 1: Conversion into central moments
μ1=ν 1−ν1 = 8 – 8 = 0

μ2=ν 2−ν 12 = 212 – 82 = 148

3
μ3=ν 3−3 ν 2 ν 1 +2 ν 1 = 1,510 – 3 × 212 × 8 + 2 × 83 = –2,554
μ4 =ν 4−4 ν 3 ν 1 +6 ν 2 ν12−3 ν 14
= 28,214 – 4 × 1,510 × 8 + 6 × 212 × 82 – 3 × 84 = 49,014
Step 2: Conversion of central moments into moments about A = 30
Here, X =A + ν 1 = 20 + 8 = 28. Thus, d= X – A=28−30=−2. Accordingly,
ν1 =μ 1+ d = 0 – 2 = – 2
2
ν 2=μ2+ d = 148 + (– 2)2 = 152
3
ν3 =μ3 +3 μ 2 d + d = – 2554 + 3 × 148 × (– 2) + (– 2)3 = –3,450
 2 4
ν 4=μ 4 +4 μ 3 d +6 μ2 d +d
= 49,014 + 4 × (–2,554) × (–2) + 6 × 148 × (– 2)2 + (– 2)4 = 73,014
Example 6 The first four moments of a distribution are given to be equal to 12; 268; 2,985
and 6,456 respectively. The mean of this distribution is 12. Calculate coefficient of variation of
this distribution. Also, examine kurtosis of this distribution.

We first convert the given moments into central moments.

μ1=ν 1−ν1 = 12 – 12 = 0
2
μ2=ν 2−ν 1 = 268 – 122 = 124
3
μ3=ν 3−3 ν 2 ν 1 +2 ν 1 = 2,985 – 3 × 268 × 12 + 2 × 123 = –3,207
μ4 =ν 4−4 ν 3 ν 1 +6 ν 2 ν12−3 ν 14 = 6,456 – 4 × 2,985 × 12 + 6 × 268 × 12 2 – 3 × 124 =
32,520
Now, Standard deviation ¿ √ μ2=√ 124=11.14
Coefficient of variation,
σ 11.14
CV = × 100= × 100=92.83 %
X 12
For kurtosis,
μ 32,520
β 2= 42 = 2
=2.11
μ2 124
Since β2 < 3, the distribution is platykurtic.

Example 7 For a normal distribution, the first moment about 12 is 28, and the fourth
moment about 40 is equal to 1,875. Find the coefficient of variation of the distribution.

We know, X =A + ν 1
Thus, here X =28+12=40
Now, since fourth moment about 40 is given, it is the value of fourth central moment.
Further,
μ4
β 2= 2
μ2
With the value of β2 = 3 for a normal distribution, we have
1875
3= 2
μ2
2
Or, μ2 =1875 /3=625
Thus, µ2 or variance = 25, and standard deviation = 5.
σ 5
Co−efficient of variation= ×100= ×100=12.5 %
X 40

Example 8 Examine if the following calculations are in respect of a normal distribution.

Σf = 100, Σfx = 0 Σfx2 = 1,500 Σfx3 = 11,000 Σfx4 = 67,500
Using the information given, we have

μ1=
∑ fx =0
∑f
μ2=
∑ fx2 = 1,500 =15
∑ f 100
μ3 =
∑ 3
fx 11,000
= =110
∑ f 100
μ4 =
∑ 4
fx 67,500
= =675
∑ f 100
These are central moments since the first moment is equal to zero. Now,
μ4 675
β 2= 2 = 2 =3
μ 2 15
The distribution is mesokurtic but is not symmetric because µ 3 ≠ 0. A normal distribution is
both symmetric and mesokurtic. Hence the given calculations are not in respect of a normal
distribution.

Example 9 The calculation of moments of a set of 50 values yielded the following results:
Mean = 16, standard deviation = 7, and β1 = 0.6196 with a positive µ3.
It was later on discovered that one observation 12 was recorded as 22. Find the corrected
values of the first three moments.

Given, standard deviation = 7. Thus, variance or µ2 = 72 = 49.

With β1 = 0.6196, we have
2
μ3
0.6196= 3
49
or μ3 =0.6196 × 493
2

∴ μ 3=270
But, since one observation was recorded incorrectly, it follows that the moments calculated
are no longer central moments. Instead, they are moments about A = 16.
Let us first find corrected mean.
ΣX = 50 ×16 = 800
Corrected ΣX = 800 + 12 – 22 = 790
Corrected X =790/50=15.8
Now, before we calculate moments about 15.8, we first correct moments about 16 calculated
above for mistake made.
Incorrect ∑ x
2
Incorrect ν 2=
n
Incorrect ∑ x
2
49=
50
Incorrect Σx = 49 × 50 = 2,450
2

Corrected Σx2 = 2,450 – (22 – 16)2 + (12 – 16)2

 = 2,450 – 36 + 16 = 2,430

2430
Corrected ν 2= =48.6
50
Similarly,
Incorrect ∑ x
3
Incorrect ν 3=
n
Incorrect ∑ x
3
270=
50
Incorrect ∑x = 270 × 50 = 13,500
3

Corrected ∑x3 = 13,500 – (22 – 16)3 + (12 – 16)3

= 13,500 – 216 + 64 = 13,348
13,348
Corrected ν 3= =266.96
50
Now, we have
X =A + ν 1
or ν 1=X + A
= 15.8 – 16 = –0.2
The corrected values of the first three moments are calculated here:
μ1=ν 1−ν1 = (–0.2) – (–0.2) = 0

μ2=ν 2−ν 12 = 48.6 – (–0.2)2 = 48.56

3
μ3=ν 3−3 ν 2 ν 1 +2 ν 1 = 266.96 – 3×48.6 × (–0.2) + 2 × (–0.2)3 = 296.104

EXERCISES
1. Calculate (a) central moments, (b) moments about A = 6, and (c) moments about origin for the
following data:
9, 4, 6, 9, 11, 13, 8, 4
2. Calculate four central moments for the following distribution:
Class
10 – 20 20 – 30 30 – 40 40 – 50 50 – 60 60 – 70
Interval:
Frequency: 8 12 15 8 5 2
3. Given, n = 10, ∑(X – 10) = – 40, ∑(X – 10) = 2,870 and ∑(X – 10) = – 480. Calculate the
2 3

following:
(a) Arithmetic mean
(b) Variance
(c) Third moment about 10, and
(d) A measure of relative skewness based on moments
4. For the following distribution, calculate first four (i) moments about 45, and (ii) central
moments. Also, comment on the shape of the distribution using gamma statistics.
Class
10 – 20 20 – 30 30 – 40 40 – 50 50 – 60 60 – 70 70 – 80
Interval:
Frequency: 2 5 12 20 8 2 1
5. Given the following distribution:
 Class
10 – 20 20 – 30 30 – 40 40 – 50 50 – 60 60 – 70 70 – 80
Interval:
Frequency: 2 6 12 20 12 6 2
(a) Calculate mean, median and mode and hence find the value of Karl Pearson’s coefficient of
skewness.
(b) Calculate the first four central moments and the beta coefficient of skewness.
(c) Comment on the results.
6. The first four moments of a distribution are 0, 5, 0, and 75 respectively. Comment on the shape
of the distribution.
7. The first four moments of a distribution are 2, 20, 40, and 500 respectively. Comment on the
shape of the distribution.
8. The first four moments of a distribution about 12 are: –8, 210, –42 and 5216 respectively. You
are required to
(a) Calculate moments about origin.
(b) Calculate the two gamma coefficients and comment upon the shape of the distribution.
9. The first four moments of a distribution are calculated as 6; 235; 1,248; and 24,680. You are
required to examine the skewness and kurtosis of the distribution.
10. From the following information about two distributions, find which is more skewed?
Distribution Second Moment Third Moment
A 16 –15.7
B 40 25.8
11. For a mesokurtic distribution, the fourth central moment is 768. Obtain its standard deviation.
12. The first four moments of a distribution about the value 4 of the variable are –1.5, 17, ––30 and
108. Its mean is given to be 2.5. You are required to
(a) Calculate the central moments and moments about origin.
(b) Determine the coefficient of variation and variance of the distribution.
(c) Examine the shape of the distribution.
13. A distribution has standard deviation equal to 5. For what values of the fourth central moment
will this distribution be (a) mesokurtic, (b) platykurtic, and (c) leptokurtic?
14. For a mesokurtic distribution, coefficient of variation = 40% and arithmetic mean = 40. Find the
value of its fourth central moment.
15. If variance = 42, then what values of µ4 would make a distribution (i) mesokurtic, (ii) platykurtic,
and (iii) leptokurtic?
16. For a set of 20 observations, mean and variance were calculated as 9 and 18 respectively. Later,
it was discovered that one observation 12 was misread as 21. Rectify the error and obtain the
corrected values of mean and variance. Also, calculate the values of the first and second
moments about origin.
17. The first two moments about 40 for a set of 25 values were calculated as equal to 65 and 2,985
respectively. Test if the calculations are consistent.
18. For a mesokurtic distribution, it is known that the first moment about 32 is 28 while the fourth
moment about 60 is 62,208. Determine the values of mean and coefficient of variation.
19. For a normal distribution, the first moment about 8 is 22 and the fourth moment about 30 is 243.
Find the coefficient of variation of the distribution.
20. The following are the calculations for a frequency distribution:
∑f = 520; ∑f(X – 56) = 0; ∑f(X – 56)2 = 143,000; ∑f(X – 56)3 = –28,750
∑f(X – 56) = 78,650,000
4
Class interval width = 10
You are required to calculate the four central moments and use Sheppard’s corrections.
 ANSWERS TO EXERCISES
1. (a) 0, 9, 2.25, 154.5 (b) 2, 13, 64.25, 404.5 (c) 10. Distribution A
8, 73, 730.25, 7778.5 11. 4
2. 0; 179.4; 951; 80,354 12. (a) 0; 14.75; 39.75; 142.3125; 2.5; 21; 166;
3. (a) 6 (b) 271 (c) −48 (d) g1 = 0.733 1,132 (b) 183.3%; 21 (c) g1 = 0.702 g2 = −2.35
4. (i) −2.6; 150; −1,100; 75,000 (ii0 0; 143.24; 13. (a) 1,875 (b) <1,875 (c) >1,875
34.848; 69,506.9; g1 = 0.0203 g2 = 0.388 14. 1,96,608
5. (a) 45, 45, 45; 0 (b) 0; 180; 0; 90,000; 15. (i) 5,292 (ii) <5,292 (c) >5,292
β1 = 0 16. 8.55; 11.05; 8.55; 84.15
6. Symmetrical, mesokurtic 17. No, μ2 < 0
7. β1 = 1, β2 = 2.39 18. 60; 20%
8. (a) 4; 162; 5,790; 150,080 (b) g1 = 2.25 19. 10%
g2 = 0.388 20. 0; 275; −55.3; 1,51,250 Corrected: μ2 = 266.7
9. Low positive skewness, Platykurtic μ4 = 137,791.7
 CHAPTER SUMMARY

 Skewness and kurtosis are both related to shape of the frequency curve.
 Skewness means lack of symmetry, which implies that the mean, median and mode
are unequal in such a case. Skewness is positive when its longer tail is to the right and
negative when it is on the left.
 There are three measures of skewness, given by Karl Pearson which is based on
averages and standard deviation; by Bowley which uses median and quartiles; and by
Kelly, based on median and the tenth and ninetieth percentiles.
 Kurtosis refers to relative height of the frequency curve. Distributions can be
mesokurtic, leptokurtic and platykurtic on this basis.
 Five-point summary, comprising lowest and highest values, median and two quartiles
is often used to learn about a given distribution. This information is represented by a
box and whisker plot.
 Moments provide a useful method to study various characteristics of a set of data.
Moments calculated about mean are called central moments. There can also be
moments about any given value A. When A = 0, moments calculated are called
moments about origin.
 It is possible to convert moments about A in to central moments and vice-versa.
 The first moment about zero is equal to mean and the first moment about mean is
equal to zero.
 The second central moment is the variance of the distribution.
 The beta and g-statistics are calculated with central moments and are used to learn
about skewness and kurtosis. β1 and g1 are measures of skewness while β2 and g2
measure kurtosis.
 If the third central moment is positive, the skewness is positive while if it is negative,
the skewness is negative. For a symmetrical distribution, it is equal to zero. β1 is
always positive but greater its value, more the skewness.
 If β2 = 3, the distribution is mesokurtic; if β 2 is less than 3, the distribution is
platykurtic while if it is greater than 3, the distribution in question is leptokurtic.
 MULTIPLE CHOICE QUESTIONS
1. For a leptokurtic distribution, the standard deviation is equal to 5. What should be the
value of its fourth central moment?
(a) 1875
(b) greater than 1875
(c) no more than 1875
(d) at least 1875

2. Mark the wrong statement:

(a) For a bell-shaped symmetrical distribution, the lower and upper quartiles are
equidistant from the median.
(b) In a positively skewed distribution, mode exceeds median.
(c) A positively skewed distribution indicates the presence of extremely low
values in it.
(d) Lack of symmetry in a distribution is seen when mean, median and mode
differ from each other.

3. Mark the wrong statement:

(a) A flat-topped curve is called platy-kurtic curve.
(b) The percentile coefficient of kurtosis is the ratio of semi-inter-quartile range to
inter-percentile range.
(c) Sheppard’s correction for fourth central moment stipulates: (corrected)
1 2 7 4
^μ4 =μ4 + μ2 i − i
2 240
(d) For a mesokurtic curve, the percentile coefficient of kurtosis = 0.263.

4. Choose the correct formula:

(a) µ4 = ν4 – 4 ν3ν1 + 6ν2ν12 – 3ν14
(b) µ4 = ν4 + 4 ν3 ν1 – 6 ν2ν12 + 3 ν14
(c) µ4 = ν4 + 6 ν3 ν1 – 4ν3ν1 – 3 ν14
(d) µ4 = ν4 – 4 ν3 ν1 – 6 ν2ν12 + 3 ν14

5. A symmetrical distribution of n values has mean equal to X and standard deviation

equal to σ. If the smallest and the largest of the values are excluded, what will be the
effect on its coefficient of variation?
(a) It will remain unchanged.
(b) It will decrease.
(c) It will increase.
(d) Nothing can be stated for sure.
6. A distribution has µ4 = 30,614 and µ2 = 81. Which of the following statements is true
about it?
(a) It is a platykurtic distribution.
(b) Its standard deviation is 9.
(c) It is a leptokurtic distribution with variance equal to 6561.
(d) Kurtosis cannot be judged from the given information.

7. A distribution has standard deviation equal to 4. What value of µ4 will make it a

leptokurtic distribution?
(a) Greater than 48.
(b) 768.
(c) Greater than 768.
(d) Less than 768.

8. What value of µ4 will make a distribution mesokurtic, if its standard deviation is 6?

(a) 1296
(b) 3888
(c) 108
(d) None of the above

9. What is the standard deviation of a distribution which has µ 4 = 12,288 and known to
be other than platykurtic or leptokurtic?
(a) It cannot be calculated.
(b) 3
(c) 192
(d) 8

10. Mark the wrong statement:

(a) In positive skewness, median > mode.
(b) Mean < Median < Mode, in case of negative skewness.
(c) A distribution with J = –0.7 is more skewed than a distribution with J = 0.4
(d) In negative skewness, most items are concentrated on the lower side (i.e.
towards the lower values of the variable).

11. Mark the wrong statement:

(a) When the lower quartile is located at a smaller distance from median than the
upper quartile is from it, the distribution has negative skewness.
(b) For a positively skewed distribution, Q3 + Q1 ≠ 2 Median.
(c) A right-tailed distribution is a positively skewed distribution.
(d) A negatively skewed distribution has longer tail to the left.

12. Mark the wrong statement:

(a) The greater the value of β2, the more peaked the distribution.
(b) The greater the value of coefficient of variation, the greater the variation in set
of data.
(c) The coefficient of variation ranges between ±100(%).
 (d) If all values in a set are equal, then each of the measures of variation is equal
to zero.

13. Mark the correct statement:

(a) A symmetrical distribution is also mesokurtic but every mesokurtic
distribution is not necessarily symmetrical.
(b) A normal distribution is mesokurtic as well as symmetrical.
(c) A positively skewed distribution is leptokurtic in nature.
(d) A skewed distribution is either leptokurtic or platykurtic.

14. Mark the correct statement:

(a) The Karl Pearson’s coefficient based on mean, median and standard deviation
can have values ranging between ±1.
(b) If for distribution A, J = – 0.7 and for distribution B, J = 0.5, then we can say
that the distribution B has higher degree of skewness since 0.5 > (– 0.7).
(c) In a given distribution, if the sum of two quartiles is less than twice the median
value, then the distribution is negatively skewed.
(d) For a distribution, the coefficient of skewness calculated by Pearson’s formula
always matches with coefficient obtained by Bowley’s formula.

15. If each of the values in a given set is increased by k, it will:

(a) Increase the Pearson’s coefficient of skewness value by k.
(b) Cause no change in Pearson’s coefficient of skewness value.
(c) Not cause a change in Bowley’s coefficient of skewness value.
(d) Reduce the Pearson’s coefficient of skewness value.

16. If each value in a given set of data is divided by 3, how will it affect the Pearson’s
coefficient of skewness?
(a) It will reduce the value of the coefficient to one-thirds.
(b) It will increase the value of the coefficient to three times.
(c) It will not affect the coefficient of skewness value.
(d) Nothing can be said about the change.

17. Mark the correct statement:

(a) Kelly’s measure of skewness is based on P10, P90, and P50.
(b) Kelly’s measure of kurtosis is based on P10, P90, and P50.
(c) Kelly’s measure of kurtosis, K = 0.2631 for a mesokurtic distribution.
(d) Kelly’s measure of kurtosis, K, ranges between –0.2631 and +0.2369.

18. Mark the wrong statement:

(a) In a symmetrical distribution, the left and right whiskers are equal.
(b) In a positively skewed distribution, the right-side whisker is longer than the
left-side whisker.
(c) Box and whisker plot is based on the smallest value, the largest value, median
and the two quartiles.
(d) Box and whisker plot is used in studying skewness and kurtosis.

19. Mark the wrong statement:

(a) The moments about A = 0 are called moments about origin.
(b) The first moment about origin is equal to arithmetic mean.
 (c) The second moment about origin is equal to variance.
(d) If the first moment about 20 is equal to –5, then mean is equal to 15.

20. Mark the wrong statement:

(a) Skewness and kurtosis both deal with shape of the frequency curve.
(b) A distribution is normal if it has µ3 = 0 and β2 = 3.
(c) Sheppard’s corrections aim to correct for the grouping error.
(d) For a given set of data, the second central moment may be more than or less
than the second moment about an arbitrary point A.