Characteristics of distribution

Chapter 5- Moments ,Skewness and Kurtosis

Definition of Moments in Statistics

In statistics, a moment is a quantitative measure that describes the shape and

characteristics of a distribution. Moments provide information about the
location, dispersion, symmetry, and peakedness of a set of data.

Moments are a set of statistical parameters to measure a distribution

.Four moments are commonly used .
∑ 𝑓(𝑥−𝑥̅ )
1st moment :Mean (Describe central value), 𝜇1 =
𝑛
∑ 𝑓(𝑥−𝑥̅ )2
2nd moment :Variance (Describes dispersion) ,𝜇2 =
𝑛
∑ 𝑓(𝑥−𝑥̅ )3
3rd Moment :Skewness (describe asymmetry) ,𝜇3 =
𝑛
th ∑ 𝑓(𝑥−𝑥̅ )4
4 moment:Kurtosis (describe peakness), 𝜇4 =
𝑛

Skewness :
TL;DR: Skewness and kurtosis are two statistical measures that describe
the shape of a data distribution, going beyond the mean to reveal how
your data is actually behaving. Skewness indicates whether the data
leans to one side, whereas kurtosis indicates how peaked or flat the
distribution is. Together, they are necessary tools for any real analysis.
Data rarely behaves as cleanly as we expect. According to the National
Center for Biotechnology Information, only 5.5% of real-world datasets
are close to being normally distributed. The rest lean, spike, or spread in
ways that a simple average won't capture. When a distribution isn't
symmetric, or its tails behave unexpectedly, concluding standard
measures alone can quietly lead you in the wrong direction.

That's where shape-based measures step in. This tutorial covers

skewness and kurtosis, two foundational statistical concepts. We'll walk
you through what they are, their types, formulas, and real-world
examples so that you can explain skewness and kurtosis confidently in
any analytical context.
What is Skewness?

Skewness is used to determine how asymmetrical a distribution is. It

tells you whether your data leans toward one side of the mean or the
other.

The mean, median, and mode are all equal in a perfectly normal
distribution. The curve is symmetrical on both sides. However, most
real-world data isn’t perfectly balanced. The values are concentrated at
one end, so the tail is pulled towards the other end. Skewness measures
that "pull".

• A skewness value of zero indicates a perfectly symmetric

distribution
• Positive values point to a right-leaning tail
• Negative values point to a left-leaning tail
The further the value moves from zero, the more asymmetric your data
is.

Skewness is important because it directly affects the interpretation of

the mean. In a skewed distribution, the mean is pulled towards the tail,
and it no longer accurately represents a typical value. Early awareness of
the meaning helps you select appropriate statistical methods and avoid
drawing conclusions unsupported by the data.

What is Kurtosis?

As skewness tells you of the direction of the lean in your data, kurtosis
meaning tells you: "how sharp or how flat the peak of your distribution is,
and how much the weight lies in the tails."

To be more precise, kurtosis is used to understand how many extreme

values you have, relative to a normal distribution. A high-peaked, sharp-
tailed distribution has very different behavior from a broad and flat one,
even if both have the same mean and standard deviation. That
difference is captured by kurtosis.
The kurtosis value of a normal distribution is 3 and is used as the
reference point. Some analysts use excess kurtosis, obtained by
subtracting 3 from the raw value, so that the scores of the normal
distribution are zero, making comparisons easier to interpret.

• When kurtosis is high, you are more likely to find extreme values
in your data
• When it is low, the distribution has lighter tails, meaning fewer
extreme values
This makes kurtosis particularly useful in areas such as finance and risk
analysis, where knowing the likelihood of extreme outcomes is not only
helpful but necessary.

Did you know? Data is typically considered approximately normal only

when skewness and excess kurtosis both fall between -1 and +1.
(Source: PMC, Descriptive Statistics and Normality Tests)

Types of Skewness

There are three types, and the distinction is straightforward.

1. Positive Skewness (Right-Skewed)

In a positively skewed (right-skewed) distribution, most values are

concentrated on the left side, while the tail extends toward the right.
Because of the long right tail, the mean is typically greater than the
median, and the median is greater than the mode.

• In this distribution, Mean > Median > Mode

 Figure: Positive Skewness

2. Negative Skewness (Left-Skewed)

In a negatively skewed, or left-skewed, distribution, most data points are

concentrated toward the right side, while the tail extends toward the left.
Because of this longer left tail, the mean is typically less than the
median, and the median is less than the mode.

• In this distribution, Mode > Median > Mean

Figure: Negative Skewness

3. Zero Skewness (Symmetric)

A perfectly balanced distribution on either side gives a zero skewness.

Mean, median, and mode are all equal, and neither tail is heavier than the
other. This is the normal distribution in its ideal form.
Types of Kurtosis

Kurtosis has three types, based on the value.

1. Leptokurtic (Positive Kurtosis)

A leptokurtic distribution is one with a kurtosis greater than 3.

Leptokurtic (Positive Kurtosis)

It has a tall, sharp peak and thick tails; that is, data is extremely
concentrated around the mean; however, when extreme values occur,
they may be important. There is a higher likelihood of outliers in this type
than in the normal distribution.

2. Platykurtic (Negative Kurtosis)

Platykurtic distributions have a kurtosis value below 3.

 Platykurtic (Negative Kurtosis)

The peak is flatter and wider, and the tails are thinner. Here, the data are
more widely distributed, with fewer extreme values. The distribution is
nearly stretched compared to a normal curve. A good example of
platykurtic behavior is a uniform distribution.

3. Mesokurtic (Kurtosis = 3)

Mesokurtic Kurtosis

This is the baseline, the normal distribution itself. The kurtosis of a

mesokurtic distribution is exactly 3, with balanced peaks and tails, which
is taken as the reference point for comparing leptokurtic and platykurtic
distributions.

Formula for Skewness and Kurtosis

Having understood the meaning of skewness and kurtosis in principle,

the next step is to learn how to calculate them. The formula for
skewness and kurtosis is a direct result of the concepts; both of them
represent the shape of a distribution in the form of a number that you
can calculate, compare, and operate on.

Skewness Formula
Pearson's skewness coefficient is the most commonly employed, and it
exists in two forms:

• Pearson's First Coefficient of Skewness = (Mean -

Mode)/Standard deviation
• Pearson's Second Coefficient of Skewness = 3(Mean -
Median)/Standard deviation
When the mode is unclear or unstable, as with continuous data, the
second coefficient is usually used. The two formulas are used to
determine the distance between the mean and the center of the
distribution, normalized by the standard deviation, to make the outcome
similar across data sets.

Interpreting skewness values:

• -0.5 to 0.5 → Approximately symmetric

• -1 to -0.5 or 0.5 to 1 → Moderately skewed
• Less than -1 or greater than 1 → Highly skewed

Kurtosis Formula

The kurtosis formula quantifies how steep the peak is and the weight of
the tails when compared to a normal distribution:

K = [Σ(X - X̄)⁴ / n] / s⁴

Where:

• X = each data point

• Xˉ= mean of the dataset
• n = number of data points
• s = standard deviation
This gives you the raw kurtosis value, where 3 is the baseline for a
normal distribution. Practically, excess kurtosis has often been used by
many analysts, which is computed as:
 • If K=3 , the frequency curve is Mesokurtic
• If K>3 ,the frequency curve is leptokertic
• If K< 3 , the frequency curve is plytokurtic
This is just a simple re-centering of the scale so that a normal
distribution assigns a score of zero, making it easy to quickly determine
whether the distribution has heavier or lighter tails than normal.

• When excess kurtosis is positive, tails are heavier,

• When it is negative, they are lighter.

Solved Problem

Find the coefficient of Kurtosis

Class 20-30 30-40 40-50 50-60 60-70 70-80 80-90
interval
F 5 14 20 25 17 11 8
 Ans:
Class F X u=
𝑥−𝐴
f𝑢2 f𝑢2
ℎ
interval
20-30 5 25 -3 45 405
30-40 14 35 -2 56 224
40-50 20 45 -1 20 20
50-60 25 55 →A 0 0 0
60-70 17 65 1 17 17
70-80 11 75 2 44 176
80-90 8 85 3 72 648
100 254 1450

Coefficient of Kurtosis =

∑ 𝑓(𝑥−𝑥̅ )4 ∑ 𝑓𝑢4 𝑥−𝐴

𝜇4 = = × ℎ4 ,Where u= u=
𝑛 𝑛 ℎ

1490
𝜇4 = × 10000 =149000
100
∑ 𝑓𝑢2 254
𝜇2 = × ℎ2 = ×100=254
𝑛 100
𝜇4 149000
Coefficient of kurtosis ,K= = =2.31
𝜇2 2 (254)2

K=2.31 < 3 ,so the frequency curve is playtokurtic .

Q.5.35(Page-126):Find Pearson’s (a)first and (b)second coefficients of
Skewness for the following distribution – Page-73-table-3.7,
Table 1
X 462 480 498 516 534 552 570 588 606 624
F 98 75 56 42 30 21 15 11 6 2

Table 2
Weight(lb) 118- 127- 136- 145-153 154-162 163-171 172-180
126 135 144
f 3 5 9 12 5 4 2

Solution: Solution of Table 2

Given grouped data:

Weight (lb) Frequency

118–126 3

127–135 5

136–144 9

145–153 12

154–162 5

163–171 4

172–180 2

Calculations
Weight Frequency Cumilative Class fX X-𝑋̅ (X − 𝑋̅ )2 f(X − 𝑋̅)2
lb f frequency mark(X)
118–126 3 3 122 366 -24.975 623.7506 1871.2519
127–135 5 8 131 655 -15.975 255.2006 1276.0031
135.5–144.5 9 17 140 1260 -6.975 48.6506 437.8556
144.5–153.5 12 29→ 149 1788 2.025 4.1006 49.2075
154–162 5 34 158 790 11.025 121.5506 607.7531
163–171 4 38 167 668 20.025 401.0006 1604.0025
172-180 2 40 176 352 29.025 842.4506 1684.9013
40 5879 7530.975
Mean = Σfx/N = 5879/40 = 146.975 lb

Median class = 145–153

Using the grouped-data median formula:

𝑁
−𝑐
2
Median=L+ ×ℎ
𝑓

where

• L=144.5 (lower class boundary)

• N=40
• C=17 (cumulative frequency before median class)
• f=12
• h=9
40
−17
Median=144.5+ 2 40 ×9

Standard deviation = √[(Σfx²/N) − Mean²] = √(871597/40 − 146.975²) = 13.72 lb

Karl Pearson’s Second Coefficient of Skewness:

Sk = 3(Mean − Median)/SD = 3(146.975 − 146.75)/13.72 = 0.049 ≈ 0.05

Interpretation
The coefficient is positive, indicating the distribution is slightly positively (right) skewed and is nearly
symmetric.

Find First Four Moments (Central Moments) – Grouped Data

Data taken from the given grouped frequency distribution of weights of 40 male college
[Link] first four moment of the following data .

Weight (lb) Frequency (f) Midpoint (x)

118–126 3 122

127–135 5 131

136–144 9 140

145–153 12 149

154–162 5 158

163–171 4 167

172–180 2 176

Total frequency, N = 40
Step 1: Arithmetic Mean
Σfx = 5879

Mean (x̄) = Σfx / N = 5879 / 40 = 146.975 lb

Step 2: First Four Central Moments

Let d = x − x̄

Moment Formula Value

First (μ₁) Σf(x−x̄) / N 0

Second (μ₂) Σf(x−x̄)² / N 188.2744

Third (μ₃) Σf(x−x̄)³ / N 441.2956

Fourth (μ₄) Σf(x−x̄)⁴ / N 91,271.4985

Final Answers
μ₁ = 0

μ₂ = 188.2744

μ₃ = 441.2956

μ₄ = 91,271.4985

Interpretation
μ₁ = 0 : First central moment is always zero.

μ₂ = 188.2744 : Variance of the distribution.

μ₃ = 441.2956 : Positive value indicates slight positive skewness.

μ₄ = 91,271.4985 : Used to measure kurtosis (peakedness and tail heaviness).