Correlation Analysis

Correlation
 Correlation is used to measure and describe
a relationship between two variables.
 Measure of correlation called correlation
coefficient which tells about the degree
and direction of correlation.
 Correlation analysis measures the closeness
of the relationship between variables.
 Ex- Husband & wife’s age, sales of a
company and expenditure on advertisement
Describing relationships:
An example…
 Correlation & Causation
 Correlation Causation
 Causation Correlation

 Correlation may be coincidental especially

in small samples.
 The relationship between variables may be
caused by some third variable.
 Both the variables may be influencing each
other so that neither can be designated as
the cause and other as the effect.
 Types of Correlation
 Positive and negative correlation-
 Depends upon the direction of change of
the variables.
 If both the variables are varying in same
direction called positive correlation.
 X 2 4 6 8 10 OR X 50 40 30 20 10
 Y 13 579 Y 24 21 19 18 14
 Negative Correlation

 The variables are varying in opposite

directions.
 X 2 4 6 8 10 OR X 50 40 30 20 10
 Y 9 7 5 31 Y 24 26 28 30 32

Simple/Partial/Multiple
Correlation-
 Distinction between three depends on the
number of variables studied.
 When only two variables are studied then simple.
 When three or more variables studied
simultaneously then multiple.
 Recognize more than two variables but consider
only two variables to be influencing each other
and keeping other variables as constant, then
partial.
Linear/ Non-Linear Relationship
 Depends upon the constancy of the ratio of the
change between the variables.
 If the amount of change in one variable tends to
bear constant ratio to the amount of change in
other variable then it is said to be linear.
 X 10 20 30 40 50
 Y 70 140 210 280 350
 If the amount of change in one variable does not
bear a constant ratio to the amount of change in
other variable then it is said to be non-linear.
Methods of Correlation
 Scatter Diagram Method
 Simplest device for ascertaining whether
two variables are related is to prepare a
dot chart.
 Greater the scatter of the plotted points,
lesser the relationship between variables.

Positive Negative No Correlation

 Scatter Plot
 What is the relationship between level of
education and lifetime earnings?

Education Level and Lifetime Earnings

X (Education) Y (Income)
8 3.4
7 4.4 5
6 2.5 4
(Criterion Variable)
Lifetime Earnings

5 2.1 3
4 1.6
2
3 1.5
2 1.2 1
1 1 0
0 2 4 6 8 10
Education (Predictor Variable)
Merits/ Demerits of
Scatter Diagram

 Useful for gaining a visual impression of

the relationship.
 Cant establish the exact degree of
correlation between variables, so more
quantitative description is needed
 Gives rough indication of nature and
strength of relationship between variables.
Karl Pearson’s Coefficient of
correlation
 Measure of linear correlation
 Widely used method
 Pearsonian Correlation Coefficient is


denoted by ‘r’.
 The value of r lies between –1 and +1.

1  r  1
 Pearson’s r
 Definitional formula:
degree to which X and Y vary together
r
degree to which X and Y vary separately

r
COVXY
COV XY 
 ( X  X )(Y  Y )
(sx )(sy ) n

Computational formula:
n( XY )  ( X )( Y )
r
( n X  ( X ) )( n Y  ( Y ) )
2 2 2 2
An Example: Correlation
X Education Y Income XY X2 Y2
8 3.4 27.2 64 11.56
7 4.4 30.8 49 19.36
6 2.5 15 36 6.25
5 2.1 10.5 25 4.41
4 1.6 6.4 16 2.56
3 1.5 4.5 9 2.25
2 1.2 2.4 4 1.44
1 1 1 1 1
36 17.7 97.8 204 48.83

 X  36
 Y  17.7 n( XY )  ( X )( Y )
r
 XY  97.8 ( n X 2  ( X ) 2 )( n Y 2  ( Y ) 2 )
 X 2  204
 Y 2  48.83
n8
An Example: Correlation

 X  36
 Y  17.7
 XY  97.8
 X 2  204
 Y 2  48.83
n8
 An Example: Correlation
 Researchers who measure reaction time for human
participants often observe a relationship between the
reaction time scores and the number of errors that the
participants commit. This relationship is known as the
speed-accuracy tradeoff. The following data are from a
reaction time study where the researcher recorded the
average reaction time (milliseconds) and the total number of
errors for each individual in a sample of 8 participants.
Calculate the correlation coefficient.
Speed Accuracy Tradeoff

Reaction Time Errors 15

184 10
213 6

Number of Errors
10
234 2
197 7
189 13 5

221 10
237 4 0
192 9 150 175 200 225 250
Reaction Time
An Example: Correlation
X X2 Y Y2 XY
184 33856 10 100 1840
213 45369 6 36 1278
234 54756 2 4 468
197 38809 7 49 1379
189 35721 13 169 2457
221 48841 10 100 2210
237 56169 4 16 948
192 36864 9 81 1728
1667 350385 61 555 12308

n( XY )  ( X )( Y )
r
( n X 2  ( X ) 2 )( n Y 2  ( Y ) 2 )
8(12308)  (1667)(61)
r
 8(350385)  (1667) 2  8(555)  (61) 2 
 0.77
 Example-
Sales revenue & profit for cement companies for
quarter July-Sept 2006-07.Find r
Company Revenue Profit after tax
(Rs. Crores) (RS. Crores)

ACC 13 2.5
Grasim 21 3.2
Industries
Guj Ambuja 10 2.6
Cements
Ultratech Cement 9 1.4
Shree Cements 3 0.8
India Cements 5 1.1

Source: Economic Times , dt. 11th October 2006.

(Ans r =0.916)
Example
 The following table gives indices of
industrial production and no. of registered
unemployed people(in Lakhs.) Calculate the
value of the correlation coefficient.
 Year 1991 92 93 94 95 96 97 98
 Index of prod. 78 89 99 60 59 79 68 61
 No. of
 Unemployed 125 137 156 112 107 136 123 108

 (Ans: r= 0.014)
 Interpreting r
 How can we describe the strength of the
relationship in a scatter plot?
– A number between -1 and +1 that indicates the
relationship between two variables.
• The sign (- or +) indicates the direction of
the relationship.
• The number indicates the strength of the
relationship.
-1 ------------ 0 ------------ +1
Perfect Relationship No Relationship Perfect Relationship

The closer to –1 or +1, the stronger the

relationship.
 When r =+1, perfect positive
relationship.
 When r =-1 ,perfect negative
relationship.
 When r=0, no relationship
 Close to +1 or –1, closer the relationship
between variables.
 Closer to 0, less close the relationship.
 The closeness of relationship is not
proportional to r.
Correlation Coefficient
Spearman’s Rank Correlation
Coefficient
 This method is useful for correlation analysis
when variables are expressed in qualitative terms
like beauty, judgment, intelligence, honesty etc.
 Spearman’s Rank correlation coefficient is defined
as
6 D 2

R  1
n( n  1)
2

 Where R:Rank Correlation coefficient

 D: difference of rank between items of two
series.
 N: no. of observations
When ranks are given-
Rank as per final grade Rank as per salary offered

A 1 1
B 2 3
C 3 2
D 4 4
E 5 6
F 6 5
G 7 9
H 8 8
I 9 10
J 10 7
When ranks are not given-
 Quotations of Index numbers of security prices of
a certain joint stock company are given. Find r-

Year Debenture Share

Price Price
1 97.8 73.2
2 99.2 85.8
3 98.8 78.9
4 98.3 75.8
5 98.4 77.2
6 96.7 87.2
7 97.1 83.8
 Equal ranks
1 3 1 3
6[ D  (m  m)  (m  m)  .....] 2

R  1 12 12
n(n  1)
2

 m: number of times whose rank are common

 Obtain rank correlation coefficient between X & Y:-

 X: 50 55 65 50 55 60 50 65 70
 Y: 110 110 115 125 140 115 130 120 115
Probable Error
 Used for testing the reliability of an
observed correlation coefficient.
1 r2
P.E  0.6745 X
n
 If r< P.E, correlation is not significant
 If r > 6P.E, correlation is definitely
significant.
 By adding and subtracting the value of
probable error from the coefficient of
correlation we get limits within coefficient
of correlation is expected to lie.

  r  P.E
Standard Error
 Standard error is defined as
1 r2
S .E 
n
 If r=0.6 and n=64, find out the probable
error , standard error of the coefficient of
correlation and determine the limits for
population r.
Coefficient of Determination
 Square of correlation coefficient.

 Indicates the extent to which variation in

one variable is explained by the variation in
other.

 It is useful because it gives the proportion

of the variance (fluctuation) of one
variable that is predictable from the other
variable.
 Coefficient of determination-
 r2 = Explained variation
Total variance

Coefficient of non-determination/ alienation-

k2 = 1-r2 = Unexplained Variation
Total Variation

0 r 2
1
If r=0.9 the r2=0.81which means that 81% of the total variation in y is due to variation in x or explained by the variation
in x.The remaining 19%(=100-81)is due to or explained by some other factors
Some Practical
Examples