Unit 1: Descriptive Statistics

Detailed Notes — Mean, Median, Mode and Beyond

1. Mean (Arithmetic Average)

The mean is the sum of all observations divided by the total number of observations. It
is the most widely used measure of central tendency.

1.1 Mean for Ungrouped Data

Formula:
Mean (M) = Sum of all observations (SX) / Total number (N)
Where: M = Mean, SX = Sum of all observations, N = Total number of observations

Example: Find the mean of 16, 14, 12, 18, 21, 22, 13, 15, 16, 18
M = (16 + 14 + 12 + 18 + 21 + 22 + 13 + 15 + 16 + 18) / 10 = 165 / 10 = 16.5

1.2 Mean for Grouped Data

Method 1: Direct Method
Used when data is in a frequency distribution with class intervals.
Formula:
Mean = Sum(xi * fi) / Sum(fi)
Where: xi = midpoint of each class = (Upper Limit + Lower Limit) / 2, fi = frequency of
that class

Steps:
• Find the midpoint xi for each class interval.
• Multiply each midpoint by its frequency to get xi x fi.
• Sum all xi x fi values and divide by the total frequency.

Example: Class intervals 0-10, 10-20, 20-30, 30-40, 40-50 with frequencies 9, 13, 8,
15, 10
Class Interval Frequency (fi) Midpoint (xi) xi x fi
0-10 9 5 45
10-20 13 15 195
20-30 8 25 200
30-40 15 35 525
 40-50 10 45 450
Total 55 1415

Mean = 1415 / 55 = 25.73

Method 2: Assumed Mean Method (Short-cut)

Used when midpoints are large, to simplify calculations.
Formula:
Mean = a + Sum(di * fi) / Sum(fi)
Where: a = assumed mean (usually the midpoint of the class with the highest
frequency), di = xi minus a

Steps:
• Calculate midpoints xi for each class.
• Choose an assumed mean 'a' (pick a convenient central midpoint).
• Calculate deviation di = xi minus a for each class.
• Multiply each di by its frequency fi, then apply the formula.

Method 3: Step Deviation Method

A further simplification when all class widths are equal.
Formula:
M = AM + (Sum(f * x') / N) x i
Where: AM = Assumed Mean, x' = (midpoint minus AM) / class width, i = class width, N
= total frequency

2. Median (Middle Value)

The median is the middle value of an ordered dataset. Half the observations lie above it
and half below. It is a measure of position, not magnitude.

2.1 Median for Ungrouped Data

Steps:
• Arrange data in ascending or descending order.
• Count the number of observations (n).

If n is odd:
Median = ((n + 1) / 2)th observation
If n is even:
Median = [(n/2)th observation + (n/2 + 1)th observation] / 2
Example (Odd n): Data: 12, 2, 16, 8, 14, 10, 6
Sorted: 2, 6, 8, 10, 12, 14, 16 | n = 7 (odd)
Median = (7+1)/2 = 4th value = 10

Example (Even n): Data: 7, 9, 3, 4, 11, 1, 8, 6, 1, 4

Sorted: 1, 1, 3, 4, 4, 6, 7, 8, 9, 11 | n = 10 (even)
Median = (5th + 6th)/2 = (4+6)/2 = 5

2.2 Median for Grouped Data

Formula:
Mdn = l + [(N/2 - F) / f] x i
Where: l = lower limit of the median class, N/2 = half the total frequency, F = cumulative
frequency of the class before the median class, f = frequency of the median class, i =
class width

How to find the median class: Compute N/2, then look at the cumulative frequency
column. The first class whose cumulative frequency is greater than or equal to N/2 is
the median class.

Example:
Class Frequency (f) Cumulative Frequency
(CF)
9.5-14.5 3 3
14.5-19.5 6 9
19.5-24.5 8 17
24.5-29.5 11 28
29.5-34.5 8 36
34.5-39.5 4 40

N = 40, N/2 = 20. First CF >= 20 is 28, which belongs to class 24.5-29.5 (the median
class).
Mdn = 24.5 + [(20-17)/11] x 5 = 24.5 + 1.36 = 25.86

3. Mode (Most Frequent Value)

The mode is the observation that occurs most frequently. A distribution can be unimodal
(one mode), bimodal (two modes), or multimodal (more than two modes).
3.1 Mode for Ungrouped Data
Simply find the value that appears the greatest number of times.

Example: Data: 34, 46, 45, 39, 43, 22, 27, 37, 46, 35, 34, 39, 40, 30, 30, 41, 37, 46, 39,
29, 34, 39, 35, 43, 30
39 appears 4 times (most frequent). Mode = 39

3.2 Mode for Grouped Data

Formula:
Mode = l + [(f1 - f0) / (2f1 - f0 - f2)] x h
Where: l = lower limit of the modal class (class with highest frequency), f1 = frequency
of the modal class, f0 = frequency of the class before the modal class, f2 = frequency of
the class after the modal class, h = class width

Example: Family sizes: 1-3 (7 families), 3-5 (8), 5-7 (2), 7-9 (2), 9-11 (1)
Modal class = 3-5 (highest frequency = 8). l = 3, f1 = 8, f0 = 7, f2 = 2, h = 2
Mode = 3 + [(8-7)/(2x8-7-2)] x 2 = 3 + (1/7) x 2 = 3 + 0.286 = 3.286

Geometric Relationship (moderately skewed distributions):

Mode is approximately equal to 3 x Median minus 2 x Mean

4. Mean Deviation
Mean deviation measures the average of the absolute deviations of each observation
from a central value (usually the mean). Absolute value means we ignore negative
signs.

General Formula:
Mean Deviation = Sum|X - X_mean| / N

4.1 Mean Deviation for Ungrouped Data

MD = Sum|xi - mu| / n
Steps: (1) Find the mean. (2) Calculate |xi minus mean| for each observation. (3) Sum
all absolute deviations. (4) Divide by n.

Example: Data: 5, 3, 7, 8, 4, 9
Mean = 36/6 = 6
xi xi - mean |xi - mean|
5 -1 1
 3 -3 3
7 +1 1
8 +2 2
4 -2 2
9 +3 3
Total 12

MD = 12/6 = 2

4.2 Mean Deviation for Grouped Data

MD = Sum(fi * |xi - mu|) / Sum(fi)
Where xi is the midpoint of each class, and mu is the mean of the grouped data.

5. Variance
Variance is the average of the squared deviations from the mean. It is denoted as sigma
squared. Squaring ensures all deviations are positive and gives more weight to larger
deviations.

5.1 Variance for Ungrouped Data

Population Variance:
sigma^2 = Sum(xi - mu)^2 / N
Sample Variance:
s^2 = Sum(xi - x_mean)^2 / (n - 1)
Note: Population variance divides by N; sample variance divides by (n-1) to correct for
bias (called Bessel's correction).

Example: Die roll outcomes {1, 2, 3, 4, 5, 6}, Mean = 3.5

sigma^2 = (6.25 + 2.25 + 0.25 + 0.25 + 2.25 + 6.25) / 6 = 17.5 / 6 = 2.917

5.2 Variance for Grouped Data

Population Variance:
sigma^2 = Sum(f * (m - x_mean)^2) / n
Sample Variance:
s^2 = Sum(f * (m - x_mean)^2) / (n - 1)
Shortcut Formula (Sample):
s^2 = [Sum(f * x^2) - (Sum(f * x))^2 / n] / (n - 1)
Where m = midpoint of each class, f = frequency, n = total frequency.

6. Standard Deviation
Standard deviation is the square root of variance. It tells us how spread out the data is
from the mean, in the same units as the original data.

Population SD:
sigma = Square root of [Sum(Xi - mu)^2 / N]
Sample SD:
s = Square root of [Sum(xi - x_mean)^2 / (n - 1)]

Key relationship: Standard Deviation = Square Root of Variance

Example (Ungrouped): Gold coins: 4, 2, 5, 8, 6. Mean = 5.

Sum(xi - x_mean)^2 = 1 + 9 + 0 + 9 + 1 = 20
Sample SD = Square root of (20/4) = Square root of 5 = 2.236

7. Quartile Deviation
Quartile Deviation (QD), also called the Semi-Interquartile Range, is half the difference
between the upper quartile (Q3) and lower quartile (Q1). It represents the spread of the
middle 50% of the data.

Formula:
QD = (Q3 - Q1) / 2

7.1 Quartiles for Ungrouped Data

For n observations arranged in ascending order:
Q1 = ((n+1)/4)th item
Q2 = ((n+1)/2)th item (this is the Median)
Q3 = (3(n+1)/4)th item

Example: Data: 2, 5, 7, 7, 8, 8, 10, 10, 14, 15, 17, 18, 24, 27, 28, 48 (n=16)
Median Q2 = (10+14)/2 = 12
Lower half: 2, 5, 7, 7, 8, 8, 10, 10 --> Q1 = (7+8)/2 = 7.5
Upper half: 14, 15, 17, 18, 24, 27, 28, 48 --> Q3 = (18+24)/2 = 21
QD = (21 - 7.5) / 2 = 6.75

7.2 Quartiles for Grouped Data

Formula:
Qr = l1 + [r(N/4) - c] / f x (l2 - l1)
Where: r = 1 for Q1, r = 3 for Q3, l1 and l2 = lower and upper limits of the quartile class,
f = frequency of the quartile class, c = cumulative frequency of the class before the
quartile class, N = total observations

To find Q1: Compute N/4 and identify the class where cumulative frequency first
exceeds N/4.
To find Q3: Compute 3N/4 and identify the class where cumulative frequency first
exceeds 3N/4.

8. Central Moments
Moments are statistical measures representing the arithmetic means of powers of
deviations. They provide a unified framework for describing central tendency, variation,
skewness, and kurtosis.

8.1 Types of Moments

Type 1: Moments about an Arbitrary Point (A)
Ungrouped: mu'r = Sum(xi - A)^r / n
Grouped: mu'r = Sum(fi * (xi - A)^r) / N
Using step deviation shortcut: mu'r = [Sum(fi * di^r) / N] x h^r, where di = (xi - A) / h

Type 2: Moments about Origin (A = 0)

Ungrouped: mu'r = (1/n) x Sum(xi^r)
Grouped: mu'r = (1/N) x Sum(fi * xi^r)

Type 3: Central Moments (Moments about Mean)

Ungrouped: mu_r = Sum(xi - x_mean)^r / N
Grouped: mu_r = Sum(fi * (xi - x_mean)^r) / N

Key results:
• mu_0 = 1 (always)
• mu_1 = 0 (always zero for moments about mean)
• mu_2 = sigma^2 (this is the variance!)
8.2 Conversion Formulas (Arbitrary Point to Central Moments)
mu_2 = mu'_2 - (mu'_1)^2
mu_3 = mu'_3 - 3 * mu'_2 * mu'_1 + 2 * (mu'_1)^3
mu_4 = mu'_4 - 4 * mu'_3 * mu'_1 + 6 * mu'_2 * (mu'_1)^2 - 3 *
(mu'_1)^4

Tip: When the mean is a fraction, it is easier to compute moments about a convenient
integer A first, then convert using these formulas.

8.3 Karl Pearson's Beta and Gamma Coefficients

These are dimensionless (unit-free) measures derived from moments:
Beta_1 = mu_3^2 / mu_2^3 (measures skewness)
Beta_2 = mu_4 / mu_2^2 (measures kurtosis)
Gamma_1 = square root of Beta_1 (coefficient of skewness)
Gamma_2 = Beta_2 - 3 (excess kurtosis)

9. Skewness
Skewness measures the asymmetry of a distribution — whether data is spread out
more on one side of the mean.

9.1 Types of Skewness

Type Description Tail Relationship
Positive (Right- Most data on the Extends right Mean > Median >
skewed) left Mode
Negative (Left- Most data on the Extends left Mean < Median <
skewed) right Mode
Zero (Symmetric) Balanced No skew Mean = Median =
distribution Mode

9.2 Measuring Skewness

Pearson's Mode Skewness:
Skewness = (X_mean - Mode) / s
Where X_mean = mean, Mode = mode, s = standard deviation

Pearson's Median Skewness (when mode is unclear or multimodal):

 Skewness = 3 x (X_mean - Median) / s

Using Moments:
Gamma_1 = mu_3 / sigma^3
Gamma_1 > 0 means positively skewed. Gamma_1 < 0 means negatively skewed.
Gamma_1 = 0 means symmetric.

10. Kurtosis
Kurtosis measures the peakedness or flatness of a distribution compared to the normal
(bell-shaped) distribution.

10.1 Measuring Kurtosis

Beta_2 = mu_4 / mu_2^2
Gamma_2 = Beta_2 - 3 (excess kurtosis)

10.2 Three Types of Kurtosis

Type Condition Shape
Mesokurtic Beta_2 = 3 Same shape as standard normal
(Normal) (Gamma_2 = 0) distribution
Leptokurtic (Tall Beta_2 > 3 More peaked than normal; heavier tails
peak) (Gamma_2 > 0) (more outliers)
Platykurtic (Flat Beta_2 < 3 Flatter than normal; lighter tails (fewer
peak) (Gamma_2 < 0) outliers)

Memory trick: Lepto = tall and pointy (think leap). Platy = flat and wide (think platypus).
Meso = in the middle (normal).

Example:
Given: mu_2 = 2.5, mu_3 = 0.7, mu_4 = 18.75
Beta_1 = (0.7)^2 / (2.5)^3 = 0.49 / 15.625 = 0.031 (nearly symmetric)
Beta_2 = 18.75 / (2.5)^2 = 18.75 / 6.25 = 3 --> Gamma_2 = 3 - 3 = 0 --> Mesokurtic

Summary Table
Measure What It Tells You Key Formula
Mean Average value Sum(x) / n
Median Middle value (n+1)/2 th term
Mode Most frequent value Value with highest frequency
Mean Deviation Average absolute spread Sum|x - x_mean| / n
Variance (sigma^2) Average squared spread Sum(x - x_mean)^2 / n
Standard Deviation Spread in original units Square root of Variance
(sigma)
Quartile Deviation Spread of middle 50% (Q3 - Q1) / 2
Skewness Asymmetry mu_3 / sigma^3
(Gamma_1)
Kurtosis (Beta_2) Peakedness mu_4 / sigma^4

End of Notes — Unit 1, Descriptive Statistics