MODULE 4

Module IV: Correlation & Regression Analysis

1. Correlation

Definition:

Correlation measures the degree and direction of relationship between two variables.

Example:

• Income ↑ → Expenditure ↑ (positive relation)

• Price ↑ → Demand ↓ (negative relation)

2. Types of Correlation

(A) Based on Direction

1. Positive Correlation

• Both variables move in same direction

• Example: Income & savings

2. Negative Correlation

• Variables move in opposite direction

• Example: Price & demand

(B) Based on Degree

• Perfect correlation (r = +1 or −1)

• High correlation

• Low correlation

• Zero correlation (r = 0)

(C) Based on Form

 • Linear correlation (straight line)

• Non-linear correlation (curved)

3. Methods of Studying Correlation

1. Scatter Diagram

• Graphical method

• Points plotted on graph

Pattern shows type of correlation

2. Karl Pearson’s Coefficient of Correlation

Measures strength of linear relationship

Formula:

∑(𝑋 − 𝑋ˉ)(𝑌 − 𝑌ˉ)

𝑟=
√∑(𝑋 − 𝑋ˉ)2 ∑(𝑌 − 𝑌ˉ)2

Interpretation of r:

Value of r Meaning

+1 Perfect positive

−1 Perfect negative

0 No correlation

3. Spearman’s Rank Correlation

Used when data is in ranks

Formula:
 6∑𝑑 2
𝑟 =1−
𝑛(𝑛2 − 1)

Where:

• 𝑑= difference in ranks

• 𝑛= number of observations

4. Properties of Correlation

• Value lies between −1 and +1

• Independent of origin and scale

• Shows direction and strength

5. Limitations of Correlation

• Does NOT imply causation

• Only measures linear relationship

• Affected by extreme values

6. Regression Analysis

Definition:

Regression analysis studies the functional relationship between variables and helps in
prediction.

Used to estimate one variable based on another

7. Types of Regression

1. Simple Regression

• One independent variable

2. Multiple Regression

• More than one independent variable

8. Regression Lines

There are two regression lines:

1. Regression of Y on X

2. Regression of X on Y

Equations:

Regression of Y on X

𝑌 − 𝑌ˉ = 𝑏𝑦𝑥 (𝑋 − 𝑋ˉ)

Regression of X on Y

𝑋 − 𝑋ˉ = 𝑏𝑥𝑦 (𝑌 − 𝑌ˉ)

9. Regression Coefficients
Cov(𝑋, 𝑌)
𝑏𝑦𝑥 =
𝜎𝑋2
Cov(𝑋, 𝑌)
𝑏𝑥𝑦 =
𝜎𝑌2

10. Relationship Between Correlation and Regression

𝑟 = √𝑏𝑥𝑦 ⋅ 𝑏𝑦𝑥
 Sign of r depends on regression coefficients

11. Properties of Regression

• Two regression lines intersect at mean point

• Regression coefficients are independent of origin but not scale

• Useful for prediction

12. Difference Between Correlation and Regression

Correlation Regression

Measures relationship Predicts relationship

Symmetrical Asymmetrical

No cause-effect Shows dependency

Single value (r) Two equations

13. Applications in Business

• Sales forecasting

• Demand estimation

• Cost analysis

• Market research

Module IV – Additional Theory (Advanced Concepts)

1. Scatter Diagram (Detailed Understanding)

A scatter diagram is the simplest way to study correlation.

Interpretation:
 • Points close to a straight line → High correlation

• Points widely scattered → Low correlation

• Upward trend → Positive correlation

• Downward trend → Negative correlation

No formula required — purely visual method

2. Probable Error (P.E.) of Correlation

Used to test reliability of correlation coefficient

Formula:

1 − 𝑟2
𝑃. 𝐸. = 0.6745 ×
√𝑛

Interpretation:

• If 𝑟 < 𝑃. 𝐸.→ Not significant

• If 𝑟 > 6 × 𝑃. 𝐸.→ Highly significant

Frequently asked theory/numerical question

3. Coefficient of Determination (r²)

𝑟 2 = (correlation coefficient)2

Meaning:

• Shows percentage of variation explained

Example:

• If 𝑟 = 0.8, then 𝑟 2 = 0.64

• → 64% variation explained

4. Coefficient of Non-Determination

1 − 𝑟2

• Shows unexplained variation

5. Concurrent Deviations Method

• Used when exact data not available

• Based on direction of change (+/−)

Simple but less accurate method

6. Assumptions of Correlation

• Relationship is linear

• Variables are quantitative

• Data is homogeneous

• No extreme outliers

7. Limitations of Correlation (Advanced)

• Cannot establish cause-effect relationship

• Misleading in case of spurious correlation

• Only measures linear relationships

• Affected by extreme values

8. Spurious Correlation

• False or meaningless relationship

Example:
 • Ice cream sales & drowning cases
(Both increase in summer but not directly related)

9. Regression (Deep Concept)

Regression shows:

• Dependence of one variable on another

• Used for forecasting and estimation

10. Regression Lines (Important Theory)

• Two lines:

o Y on X

o X on Y

Key Points:

• Both lines intersect at mean point (𝑋ˉ, 𝑌ˉ)

• If correlation is perfect → both lines coincide

• If correlation is zero → lines are perpendicular

11. Regression Coefficients (Properties)

• Both coefficients have same sign

• If one > 1 → other < 1

• Independent of origin but not scale

12. Angle Between Regression Lines

• Smaller angle → higher correlation

• Larger angle → lower correlation

Concept-based question
 13. Difference Between r and r²

r r²

Measures direction & strength Measures proportion explained

Can be negative Always positive

Range: -1 to +1 Range: 0 to 1

14. Uses of Correlation & Regression

• Business forecasting

• Demand estimation

• Price analysis

• Financial planning

• Economic research

15. Practical Interpretation

High correlation:

• Strong relationship

Low correlation:

• Weak relationship

Zero correlation:

• No relationship

Module IV – Formula Sheet

1. Karl Pearson’s Correlation Coefficient

Direct Formula

∑(𝑋 − 𝑋ˉ)(𝑌 − 𝑌ˉ)

𝑟=
√∑(𝑋 − 𝑋ˉ)2 ⋅ ∑(𝑌 − 𝑌ˉ)2

Shortcut Formula
𝑁∑𝑋𝑌 − (∑𝑋)(∑𝑌)
𝑟=
√[𝑁∑𝑋 2 − (∑𝑋)2 ][𝑁∑𝑌 2 − (∑𝑌)2 ]

2. Spearman’s Rank Correlation

6∑𝑑 2
𝑟 =1−
𝑛(𝑛2 − 1)

Where:

• 𝑑= difference in ranks

• 𝑛= number of observations

3. Probable Error (P.E.)

1 − 𝑟2
𝑃. 𝐸. = 0.6745 ×
√𝑛

4. Coefficient of Determination

𝑟2

5. Coefficient of Non-Determination

1 − 𝑟2

6. Regression Equations
Regression of Y on X

𝑌 − 𝑌ˉ = 𝑏𝑦𝑥 (𝑋 − 𝑋ˉ)

Regression of X on Y

𝑋 − 𝑋ˉ = 𝑏𝑥𝑦 (𝑌 − 𝑌ˉ)

7. Regression Coefficients
𝜎𝑌
𝑏𝑦𝑥 = 𝑟 ⋅
𝜎𝑋

𝜎𝑋
𝑏𝑥𝑦 = 𝑟 ⋅
𝜎𝑌

8. Relationship Between r and Regression Coefficients

𝑟 = √𝑏𝑥𝑦 ⋅ 𝑏𝑦𝑥

9. Covariance (Basic Idea)

∑(𝑋 − 𝑋ˉ)(𝑌 − 𝑌ˉ)

Cov(𝑋, 𝑌) =
𝑁
Solved Sums – Module IV

1. Karl Pearson’s Correlation

Q1. Find correlation coefficient

Data:

X12345

Y 2 4 6 8 10
Step 1: Observe relationship

Y = 2X → perfectly linear

So,

𝑟 = +1

Answer: +1 (Perfect Positive Correlation)

2. Shortcut Formula (Numerical)

Q2. Find r

X123

Y235

Step 1: Calculate values

∑𝑋 = 6, ∑𝑌 = 10
∑𝑋𝑌 = (1 × 2) + (2 × 3) + (3 × 5) = 2 + 6 + 15 = 23
∑𝑋 2 = 14, ∑𝑌 2 = 38

Step 2: Apply formula

3(23) − (6)(10)
𝑟=
√[3(14) − 62 ][3(38) − 102 ]
69 − 60 9 9
𝑟= = = ≈ 0.98
√[42 − 36][114 − 100] √6 × 14 √84

Answer: 0.98 (High Positive Correlation)

3. Spearman’s Rank Correlation

Q3. Find rank correlation

X 10 20 30 40

Y 15 25 35 45

Step 1: Assign ranks

Both increasing → same ranks

X Rank Y Rank d

10 1 15 1 0

20 2 25 2 0

30 3 35 3 0

40 4 45 4 0

∑𝑑 2 = 0

Step 2: Formula
6(0)
𝑟 =1− =1
4(16 − 1)

Answer: 1 (Perfect Positive Correlation)

4. Probable Error

Q4. Find P.E.

Given: 𝑟 = 0.8, 𝑛 = 16

1 − 𝑟2
𝑃. 𝐸. = 0.6745 ×
√𝑛
1 − 0.64 0.36
= 0.6745 × = 0.6745 × = 0.6745 × 0.09 ≈ 0.061
4 4
 Answer: 0.061

5. Regression Coefficient

Q5. Find 𝒃𝒚𝒙

Given:
𝑟 = 0.6, 𝜎𝑋 = 2, 𝜎𝑌 = 3
𝜎𝑌 3
𝑏𝑦𝑥 = 𝑟 × = 0.6 × = 0.6 × 1.5 = 0.9
𝜎𝑋 2

Answer: 0.9

6. Regression Equation

Q6. Find regression line of Y on X

Given:
𝑋ˉ = 10, 𝑌ˉ = 20, 𝑏𝑦𝑥 = 2

Step 1: Formula

𝑌 − 𝑌ˉ = 𝑏𝑦𝑥 (𝑋 − 𝑋ˉ)

Step 2: Substitute

𝑌 − 20 = 2(𝑋 − 10)
𝑌 − 20 = 2𝑋 − 20
𝑌 = 2𝑋

Answer: 𝑌 = 2𝑋
 7. Find Correlation from Regression Coefficients

Q7. Given:

𝑏𝑥𝑦 = 0.5, 𝑏𝑦𝑥 = 0.8

𝑟 = √0.5 × 0.8 = √0.4 ≈ 0.63

Answer: 0.63

8. Coefficient of Determination

Q8. If 𝒓 = 𝟎. 𝟕, find 𝒓𝟐

𝑟 2 = (0.7)2 = 0.49

49% variation explained

Answer: 0.49

MCQs – Module IV (with Answers)

Basics of Correlation

1. Correlation measures:
A. Average
B. Relationship between variables
C. Frequency
D. Dispersion
Answer: B

2. Correlation coefficient (r) lies between:

A. 0 and 1
B. −1 and +1
C. 0 and ∞
D. −∞ and +∞
Answer: B

3. If r = +1, it indicates:
A. No correlation
B. Perfect positive correlation
C. Perfect negative correlation
D. Weak correlation
Answer: B

4. If r = 0, it means:
A. Perfect correlation
B. No correlation
C. Negative correlation
D. High correlation
Answer: B

Types of Correlation

5. When variables move in opposite direction:

A. Positive correlation
B. Negative correlation
C. Zero correlation
D. Linear correlation
Answer: B

6. Income and expenditure usually have:

A. Negative correlation
B. Positive correlation
C. No correlation
D. Zero correlation
Answer: B
 Methods of Correlation

7. Scatter diagram is a:
A. Mathematical method
B. Graphical method
C. Statistical test
D. Algebraic method
Answer: B

8. Spearman’s rank correlation is used for:

A. Raw data
B. Grouped data
C. Ranked data
D. Continuous data
Answer: C

Karl Pearson

9. Karl Pearson’s coefficient measures:

A. Non-linear relationship
B. Linear relationship
C. Random data
D. Frequency
Answer: B

10. Correlation does NOT imply:

A. Relationship
B. Association
C. Causation
D. Direction
Answer: C

Regression
11. Regression is used for:
A. Description
B. Prediction
C. Classification
D. Tabulation
Answer: B

12. Number of regression lines:

A. 1
B. 2
C. 3
D. 4
Answer: B

13. Regression lines intersect at:

A. Origin
B. Median
C. Mean point
D. Mode
Answer: C

Regression Coefficients

14. Regression coefficients are:

A. Always negative
B. Always positive
C. Same sign
D. Opposite sign
Answer: C

15. If one regression coefficient is greater than 1:

A. Other is also >1
B. Other is <1
C. Both are equal
D. None
Answer: B

Relationship

16. Relationship between r and regression coefficients:

A. r = b₁ + b₂
B. r = √(b₁ × b₂)
C. r = b₁ − b₂
D. r = b₁ / b₂
Answer: B

Probable Error

17. Probable error is used to:

A. Calculate mean
B. Test reliability of r
C. Find median
D. Find mode
Answer: B

Determination

18. r² represents:
A. Mean
B. Variance
C. Explained variation
D. Total frequency
Answer: C

Concepts

19. Spurious correlation means:

A. Strong relation
B. False relation
C. Negative relation
D. Perfect relation
Answer: B

20. Correlation is independent of:

A. Scale and origin
B. Mean
C. Variables
D. Data
Answer: A

Numerical-Type

21. If r = 0.6, then r² =

A. 0.12
B. 0.36
C. 0.6
D. 1
Answer: B

22. If r = −1, relationship is:

A. Perfect positive
B. Perfect negative
C. Zero
D. Weak
Answer: B

Application-Based

23. Regression helps in:

A. Drawing graphs
B. Forecasting
C. Classification
D. Collection
Answer: B

24. Which is graphical method of correlation?

A. Regression
B. Scatter diagram
C. Mean
D. Median
Answer: B

25. High correlation means:

A. Weak relationship
B. Strong relationship
C. No relation
D. Zero relation
Answer: B