Divergence from Normalcy –
Skewness and Kurtosis
Understanding deviations in
the normal probability curve
Presented By - Nakul Yadav
The Normal Probability Curve (NPC) is a symmetrical, bell-
Normal Probability shaped curve that represents the distribution of many naturally
occurring variables (e.g., height, IQ, test scores).
Curve Properties of Normal Curve:
• Symmetrical about the mean (Mean = Median = Mode)
• Unimodal (one peak)
• Total area under the curve = 1
• 68-95-99.7 Rule:
• 68% within 1σ
• 95% within 2σ
• 99.7% within 3σ
Divergence from Normalcy
Divergence from normalcy refers to how much a data
distribution deviates from the ideal normal (bell-shaped)
curve. In real-world data, distributions are rarely perfectly
normal. They may show asymmetry or differ in peakedness or
flatness. These deviations are measured using two key
concepts — Skewness (asymmetry of the curve) and Kurtosis
(degree of peakedness or flatness). Together, they help
describe the shape and behavior of a distribution compared
to the normal curve.
Skewness measures the degree and direction of asymmetry in a
Skewness – frequency distribution.
Measure of A perfectly normal distribution is symmetrical, but real data often leans
to one side.
Asymmetry • Indicates whether the tail of the curve is longer on one side.
• Describes departure from symmetry.
• Helps understand whether mean, median, and mode are equal or
differ.
According to Croxton and Cowden:
“A distribution is said to be skewed when it departs from symmetry, that
is, when the values are unequally distributed about the mean.”
Type of Skewness
1. Positive Skewness (Right Skewed Distribution):
A distribution is said to be positively skewed when the tail of the curve extends more towards the right-hand
side of the graph, that is, toward the higher values. In such distributions, most of the observations lie on the
left side of the mean, and a few extremely large values pull the mean towards the right
As a result, the mean is greater than the median, and the
median is greater than the mode (Mean > Median >
Mode)
.
This type of distribution is common when a dataset has a
few exceptionally high values
For example, income distributions, where most people
earn average or low incomes, but a few individuals earn
extremely high salaries, pulling the mean upward.
2. Negative Skewness (Left Skewed Distribution):
A distribution is said to be negatively skewed when the tail of the curve extends more towards the left-
hand side, i.e., toward the lower values. Here, most of the data values lie on the right side of the mean,
while a few very small values pull the mean towards the left.
Hence, the mean is less than the median, and the median
is less than the mode (Mean < Median < Mode).
This type of distribution occurs when most of the scores
are relatively high but a few very low values shift the
balance
For example, exam marks where most students perform
well, but a few very low scorers pull the mean downward.
3. Zero Skewness (Symmetrical or Normal Distribution):
When a distribution is perfectly symmetrical, it is said to have zero skewness. In this
case, the data values are evenly distributed on both sides of the mean, and the mean,
median, and mode coincide (Mean = Median = Mode).
The curve has a perfect bell-shaped form with identical tails on both sides.
Karl Pearson’s Coefficient of Skewness
The Karl Pearson’s Coefficient of Skewness is one of the most widely used and simplest methods to
measure the degree of asymmetry in a frequency distribution. It was developed by Karl Pearson, a
pioneer in modern statistics. This coefficient provides a numerical value that indicates both the
direction and extent of skewness in a dataset.
Formula:
OR
Interpretation of Skewness (Karl Pearson’s Coefficient):
• If Sk = 0, the distribution is symmetrical (no skewness).
• If Sk > 0, the distribution is positively skewed (tail extends to the right).
• If Sk < 0, the distribution is negatively skewed (tail extends to the left).
The greater the absolute value of Sk, the more asymmetric the distribution is.
Kurtosis – Measure of
Peakedness or Flatness
Kurtosis is a statistical measure that describes the shape of a
frequency distribution curve, particularly the degree of
peakedness or flatness in comparison to a normal distribution
curve.
It tells us how concentrated or spread out the data values are
around the mean, and whether the distribution produces
heavy tails (more extreme values) or light tails (fewer
extremes).
According to Garrett (1969):
“Kurtosis indicates the degree to which the distribution
is more or less peaked than the normal curve.”
Types of Kurtosis
1. Mesokurtic Distribution (β₂ = 3)
2. Leptokurtic Distribution (β₂ > 3)
3. Platykurtic Distribution (β₂ < 3)
1. Mesokurtic Distribution (β₂ = 3)
Mesokurtic distribution is the standard or normal distribution that serves
as a benchmark for comparison. In this type, the data values are
moderately concentrated around the mean, and the tails of the
distribution are neither too heavy nor too light. The normal curve is the
perfect example of a mesokurtic distribution.
2. Leptokurtic Distribution (β₂ > 3)
A Leptokurtic distribution is more peaked than the normal curve.
In this case, a large number of values lie close to the mean, resulting in a
tall and narrow peak with thicker tails.
Leptokurtic distributions indicate less variability near the mean but greater
probability of outliers.
For example, scores of students in a competitive exam where most perform
similarly well but a few perform exceptionally high or low.
3. Platykurtic Distribution (β₂ < 3)
A Platykurtic distribution is flatter than the normal
curve. Here, the data are more evenly spread out, and
the curve has a lower and wider peak with thinner
tails. This means there is greater variability in the
dataset and fewer extreme values.
Platykurtic distributions often appear in
heterogeneous groups, such as personality test scores
where responses vary widely.
This shape indicates that the data do not cluster much
around the mean and are more dispersed.
Moment Coefficient of Kurtosis
Kurtosis is calculated using moments — specifically, the second and
fourth central moments of a distribution.
OR
To compare a distribution to the normal distribution (which has
kurtosis = 3), we use:
Excess Kurtosis=β2−3
• If Excess Kurtosis = 0, the distribution is mesokurtic (normal).
Excess Kurtosis
• If Excess Kurtosis > 0, it is leptokurtic (more peaked).
• If Excess Kurtosis < 0, it is platykurtic (flatter).
ANY
QUESTIONS?
THANK YOU