Meaning and Types of Correlation

Correlation
It is a statistical tool that measures the quantitative relationship between different variables. It studies
the degree & intensity of the relationship between two variables.
 Meaning and Types of Correlation
Price Demand Hours Marks
10 2 1 20
12 1 2 40
8 4 3 60
6 8 4 80
Negatively Related Positively Related

Correlation
It is a statistical tool that measures the quantitative relationship between different variables. It studies
the degree & intensity of the relationship between two variables.
 Types of Correlation

Positive and Linear and Simple and

Negative Curvilinear Multiple

Positive Correlation Negative Correlation

When 2 variables move in When 2 variables move in the
the same direction opposite direction
Example: Example:
X Y X Y
10 15 10 36
20 22 20 29
30 29 30 22
40 36 40 15
 Types of Correlation

Positive and Linear and Simple and

Negative Curvilinear Multiple

Linear Correlation Curvilinear Correlation

When ratio of change between 2 When ratio of change between 2
variables is constant variables is not constant
Example: Example:
X Y X Y X Y
50 40 50 46 50 10
55 42 55 44 55 12
60 44 60 42 60 17
65 46 65 40 65 30
 Types of Correlation

Positive and Linear and Simple and

Negative Curvilinear Multiple

Linear Correlation Curvilinear Correlation

When ratio of change between 2 When ratio of change between 2
variables is constant variables is not constant
Example: Example:
X Y X Y X Y
50 40 50 46 50 10
55 42 55 44 55 12
60 44 60 42 60 17
65 46 65 40 65 30
Curvilinear Correlation
Linear Positive Correlation Linear Negative Correlation (depicted by quadratic
(depicted by positively sloped (depicted by negatively graph, parabola, hyperbola
straight line graph) sloped straight line graph) etc.)
 Types of Correlation

Positive and Linear and Simple and

Negative Curvilinear Multiple

Simple Correlation Multiple Correlation

When relationship between When relationship between 3 or more than 3
only 2 variables is studied variables is studied simultaneously
Example: Example:
Price Demand Price Demand Income

10 50 5 10 60

20 40 4 20 70

30 30 3 30 80

40 20 2 40 90
 Degrees of Correlation

Perfect Zero Limited Degree

Correlation Correlation of Correlation

When two variables change When there is no relation When an increase(or decrease)
in exactly equal proportion between two variables in one variable is not accompanied
by an equal proportionate increase
(or decrease) in other variable
Perfect Positive

Perfect Negative
 Degrees of Correlation
Degrees of Correlation Positive Negative
Perfect Correlation +1 −1
Very High Correlation Between + 0.75 and + 1 Between − 0.75 and − 1
Moderate Degree Between + 0.25 and + 0.75 Between − 0.25 and − 0.75
Low Degree Between 0 and + 0.25 Between 0 and − 0.25
Zero 0 0

-1 - 0.75 - 0.25 0 + 0.25 + 0.75 +1

Perfect Zero Perfect
Negative Correlation Positive
High High
Negative Positive
Moderate Moderate
Negative Positive
Low Low
Negative Positive
 Correlation and Causation
Weight Hours Put to Study

12 3
14 4
16 5
Weight and hours put to study change
18 6 in exactly equal proportion
20 7

Perfect Positive ↑ study hours  ↑ weight  Fatter student

Correlation
↓ study hours  ↓ weight  Thinner student

Do you think this

will always be NO!!
true??

does not
Correlation Cause and Effect Relation
always
Correlation Cause and Effect Relation
 Methods of Measuring Correlation

Karl Pearson’s
Coefficient of Correlation

Spearman’s Rank
Correlation Coefficient

Scattered Diagram
 Methods of Measuring Correlation

It is a mathematical measure of correlation and denoted by ‘r’

r=  xy
Karl Pearson’s N x y
Coefficient of Correlation
r = Coefficient of correlation
x = X −X
y=Y−Y
 x = Standard deviation of X series
Spearman’s Rank y
= Standard deviation of Y series
Correlation Coefficient
N = Number of Observations

Example: Calculate Karl Pearson’s correlation coefficient.

Particulars X Y
N 10 10
Scattered Diagram X 25 27
 2 4

Sum of product of deviation of X and Y series is 25

r = 25 = 0.31
Solution: 10  2  4
 Karl Pearson’s Coefficient of Correlation
Actual Mean Method
Step 1: Calculate arithmetic mean of X and Y series
Step 2: Obtain deviation of X & Y series from X & Y respectively& denote it as x & y respectively
Step 3: Square the deviations of X and Y series & obtain their sum as  x2 &  y2 respectively
Step 4: Multiply deviations and obtain their sum as  xy
Step 5: Calculate Karl Pearson’s Coefficient of Correlation
r =  xy
 x2   y 2 X =  X = 105 = 15
N 7
Example:
X Y x x2 y y2 xy Y = Y = 84 = 12
N 7
5 2 − 10 100 − 10 100 100
7 4 −8 64 −8 64 64 r = 406 = 406 = 0.99
398 × 416 406.9
10 7 −5 25 −5 25 25
15 12 0 0 0 0 0
19 15 4 16 3 9 12 High degree of positive
22 19 7 49 7 49 49 correlation
27 25 12 144 13 169 156
X = 105 Y = 84 x2 = 398 y2 = 416 xy = 406
 Karl Pearson’s Coefficient of Correlation
Direct Method
Step 1: Square each value of X and Y series & obtain their sum as X2 and Y2 respectively
Step 2: Multiply each value of X series with corresponding value of Y series & obtain sum as XY
Step 3: Calculate Karl Pearson’s Coefficient of Correlation
r = N  XY −  X Y
− (  X )  N Y 2 − ( Y )
2 2 2
N X

Example:
X Y X2 Y2 XY
r= 6  621 − 60  72
5 18 25 324 90 6  676 − (60)2  6  1050 − (72)2
7 16 49 256 112
8 13 64 169 104 r= − 594 = − 594 = − 0.83
21.35  33.41 713.3
12 16 144 256 192
13 6 169 36 78
15 3 225 9 45
X = 60 Y = 72 X2 = 676 Y2 = 1050 XY = 621 High degree of
negative correlation
 Karl Pearson’s Coefficient of Correlation
Assumed Mean Method or Short-Cut Method
Step 1: Take assumed mean for X & Y series to obtain deviations as d x & d y respectively. Obtain sum as  dx
& d y
Step 2: Square the deviations of X and Y series & obtain their sum as  d x2&  d y2 respectively
Step 3: Multiply deviations and obtain their sum as  d xd y

Step 4: Calculate Karl Pearson’s Correlation Coefficient

( d x )  ( d y )
 d xd y −
r= N
2  
2
2−  y 
d 
2 −  x 
d  

 x
d    y
d 

Example: N N 136 − 9×16

2 2 r= 6

X Y dx 


d x  dy  d y  d xd y 9
 
2 16 
 
2
31 55 −4 16 −4

16

16 101−    194 −  
6 6
34 57 −1 1 −2 4 2 112
r= = 0.97
35 = A 59 = A 0 0 0 0 0 9.35 × 12.3
37 64 2 4 5 25 10
39 66 4 16 7 49 28
High degree of positive
43 69 8 64 10 100 80
correlation
= = 101 = 16 = 194 =
9 136
 dx  d x2 d y  d y2  d xd y
 Karl Pearson’s Coefficient of Correlation
Step-Deviation Method
• Same as Assumed Mean Method
• Value of deviations is further reduced by dividing them by a common factor
Calculation of Karl Pearson’s Correlation Coefficient
 d 'x     d ' y 
  

 d 'x d ' y −


r = N
2 2
 d 'x   d ' y 
   

 d 'x 2 −  N    d ' y 2 −  N 

X Y dx d 'x d 'x 2 dy d 'y d ' y2 d 'x d ' y 

4 × 4 

150 340 −30 −3 9 −30 −3 9 9 44 − 
r = 8
160 350 −20 −2 4 −20 −2 4 4 44 − 16  44 − 16
8 8
170 360 − 10 −1 1 −10 −1 1 1 42
r = = 1
6.48 × 6.48
180 = A 370 = A 0 0 0 0 0 0 0

190 380 10 1 1 10 1 1 1
Perfect Positive
200 390 20 2 4 20 2 4 4
correlation
210 400 30 3 9 30 3 9 9
220 410 40 4 16 40 4 16 16
 d 'x  d 'x 2  d 'y  d ' y  d 'x d ' y
2

=4 = 44 =4 = 44 = 44
 Properties of Correlation Coefficient

Free from any units, i.e. it is a pure number

A negative correlation coefficient indicates inverse relationship

between two variables and vice-versa

If its value is zero, then there is no linear relation between two

variables

Its value lies between −1 and + 1

Independent of its origin & any change in scale of graph

 Merits and Demerits- Karl Pearson’s Correlation Coefficient
Merits De-Merits

✓ Most important and an ideal method ✓ Affected by presence of extreme values

✓ Helps in assessing type and magnitude of linear ✓ Involves tedious & time consuming calculation
relationship between two variables
✓ Fails to study non-linear relationship
✓ Helps in measuring exact correlation between two
variables
 Methods of Measuring Correlation

Karl Pearson’s
Coefficient of Correlation

Different values of 2 variables are plotted on graph

Spearman’s Rank
Correlation Coefficient Set of points obtained is called scattered diagram

Analysis of scatter diagram gives fair idea of the degree of

correlation between two variables

A graphical measure of estimating direction, magnitude &

Scattered Diagram
degree of correlation existing between two variables
 Degrees of Correlation- Scattered Diagram
Different Degrees

Case 1 Case 2 Case 3 Case 4 Case 5

r=+1 r=−1 0<r<1 −1<r<0 r=0

Perfect Perfect High Degree of High Degree of No

Positive Negative Positive Correlation Negative Correlation
Correlation Correlation Correlation
 Degrees of Correlation- Scattered Diagram
Different Degrees

Case 1 Case 2 Case 3 Case 4 Case 5

r=+1 r=−1 0<r<1 −1<r<0 r=0

Line of Best Fit is the line that

passes in such a way that half
of the scattered points lie on
either sides of the line.
 Degrees of Correlation- Scattered Diagram
Example 1: Explain the relation between price and quantity supplied for the following data through a scattered
diagram.
Price Quantity Supplied
10 25
20 50
30 75
40 100
50 125
60 150

The points obtained on scattered

diagram lie on a positive sloped
straight line
r=+1
 Degrees of Correlation- Scattered Diagram
Example 2: Explain the relation between X and Y variables for the following data with the help of scattered
diagram.
X Y
8 6
7 8
6 10
5 12
4 14

The points obtained on scattered

diagram lie on a negative sloped
straight line
r = −1
 Merits and Demerits- Scattered Diagram
Merits De-Merits

✓ Easy to draw & understand ✓ Presents only a rough estimation of correlation

✓ Does not involve tedious & difficult calculations ✓ Fails to reveal anything about magnitude &
degree of correlation
✓ Not affected by presence of extreme
values ✓ Fails to ascertain correlation between more than
two variables
✓ Reveals type of correlation between two variables
merely at a glance ✓ Fails to reveal the direction of causation