MAT2001–Statistics for Engineers

CORRELATION ANALYSIS

Concept of Correlation :

Correlation refers to the relationship between two or more variables. Simple correlation
studies the relationship between two variables. Correlation analysis attempts to
determine the degree of relationship between variables.

If the change in one variable is accompanied by the corresponding change in the other
variable, then we say that there is correlation between the given two variables. In other
words, these two variables are said to be correlated variables.

Types of Correlation

(a) Positive Correlation :-

If the increase change in one variable is accompanied by the corresponding increase
change in other variable; or if the decrease change is one variable is accompanied by the
corresponding decrease change in other variable, then we say that these is Positive
Correlation between the two variables. In this case, the two variables more in the same
direction.

Example : Correlation between (i) sales and expenditure on advertisement; (ii) Income
and Expenditure of Households; (iii) Production and Labour (iv) Heights and weights of
students etc., are the cases of positive correlation.

(b) Negative Correlation : If the increase change in one variable is accompanied by the
corresponding decreasing change in other variable; or if the decrease change in one
variable is accompanied by the corresponding increase change in other variable, then we
say that there is Negative Correlation In this case the two variables more in the opposite
direction.

Examples : Correlation between (i) Price and Demand of commodities; (ii) Sales and
Prices of Commodities; (iii) Volume and pressure of a perfect gas; (iv) sales of sweaters
and day temperature etc., are the cases of negative correlation.

Dr.Mokesh Rayalu,M.Sc,Ph.D.,
 MAT2001–Statistics for Engineers

(.C) Perfect Correlation: If a change in one variable is accompanied by a corresponding

proportional change or constant change in other variable then we say that there is perfect
correlation between the two variables. If these changes are in the same direction then we
say that there is perfect positive correlation between two variables. If these changes are
in the opposite direction, then we say that there is perfect negative correlation between
the two variables.

Example (1) : Perfect positive correlation between X and Y :

(i)
X 2 4 5 6 8 9
Y=3X 6 12 15 18 24 27
(ii)
X 1 2 3 4 5
Y=2+5X 7 12 17 22 37

Example (2) : Perfect negative correlation between X and Y :

(i)
X 1 2 5 8 9
Y=-2X -2 -4 -10 -16 -18
(ii)
X 1 2 3 4 5
Y=5 - 8X -3 -11 -19 -27 -35

(d) Simple, Partial and Multiple Correlation

The correlation between only two variables is called ‘simple correlation’. The concern
study is known as ‘simple correlation analysis’. In the study of correlation, when more
than two variables are involved then correlation may of either partial or multiple
correlation. Partial correlation is the correlation between any two variables in a group of
more than two variables, where, the linear effects of other variables on then have been
eliminated from both variables separately.

Dr.Mokesh Rayalu,M.Sc,Ph.D.,
 MAT2001–Statistics for Engineers

When we consider three or more variables at a time, multiple correlation reveals the joint
effect of a group of variables on a specified variable, which is not included in that group.

METHODS FOR STUDYING SIMPLE CORRELATION

The following are different methods for studying correlation between two variables,
which are frequently used in practice :
1. Scatter diagram
2. Karl Pearson’s coefficient of correlation (r) (read ‘’ as row)
3. Spearman’s rank correlation coefficient ()
4. Kendall’s coefficient of concurrent deviations (r) (Read ‘’ as
Tou)
Scatter Diagram

Scatter diagram is the simplest way of graphic representation of a bivariate data, where
the given set of ‘n’ pairs of observations on two variables X and Y say
(X1, Y2), (X2, Y2) …(Xn, Yn) may be plotted as dots by considering X-values on X-axis
and Y-values on Y-axis. By scatter diagram, we can get some idea about the correlation
between X and Y.

In a scatter diagram, it the points are closer and show either upward or downward trend
then there is high degree of correlation. Other wise, if the points are scattered and do not
show any trend then there is no correlation or low degree of correlation between
variables.

Scatter diagram shows the direction of correlation but it can not measure the degree of
correlation.

Dr.Mokesh Rayalu,M.Sc,Ph.D.,
 MAT2001–Statistics for Engineers

The above scatter diagrams give some idea about the various types of correlation.

Karl Pearson’s Coefficient of Correlation

Karl Pearson (1867-1936), a British Biometrician suggested a measure of correlation

between two variables in year 1896, which is known as Karl Pearson’s coefficient of
correlation. It is useful for measuring the degree of linear relationship between the two
variables X and Y. It is usually denoted by rXY or r. It is also sometimes called ‘product
movement coefficient of correlation. It is defined as the ratio of covariance between x
and y say Cov (X,Y) to the product of the standard deviations of X and Y, say  X  Y

Cov( XY )
i.e rXY 
σXσY

Consider a set of ‘n’ pairs of observations (X1, Y1), (X2, Y2), … (Xn, Yn) on two variables
X and Y. Then we have, Covariance between X and Y

Cov( XY ) 
 ( X  X ) (Y  Y ) or
  XY


 ( X )(Y )
n  n 
1
=  XY 
 X Y 

n  n 
 X  X   X 2 2
2

Standard Deviation of X : = σ X = or   X  
n  n 

=
 X
 X 2  
1
2


n n 


Standard Deviation of Y : = σ Y
 Y  Y  2
 Y 2 2
 Y  
= or 
n  n 

Dr.Mokesh Rayalu,M.Sc,Ph.D.,
 MAT2001–Statistics for Engineers

=
 Y
 Y 2  
1
2


n n 


Where, X 
X , Y 
Y and n= Number of pairs of observations by substituting
n n
these expansions, we get the following formula :

 ( X  X ) (Y  Y )  XY  n X Y )
 X  X    Y  Y    X  nX  Y  nY 
rXY = (OR) rXY =
2 2 2 2 2 2

 X  Y 
 XY  n
rXY =
  X   
 X  
2
 Y  2


2

n  
 Y 
n
2


  

Properties of Coefficient of Correlation

1.The value of Karl Pearson’s Correlation coefficient lies between -1 and +1.
i.e., -1  rXY  1
In other words, the limits of correlation coefficient are  1.

If rxy = + 1, then we can say that there is perfect positive correlation between X and Y.

If rXY = -1. Then we can say that there is perfect negative correlation between X and Y.

If rXY is greater than 0.5. then we can say that there is a high degree of positive
correlation between X and Y.

If rXY is greater than 0.5 with negative sign then we can say that there is high degree of
negative correlation between X and Y.

Similarly we can say about less degree of positive correlation and less degree of negative
correlation based on the value of rxy (i.e., r  0.5 and r  0.5 with negative sign).
Dr.Mokesh Rayalu,M.Sc,Ph.D.,
 MAT2001–Statistics for Engineers

1. Correlation coefficient is independent of change of origin and scale.

2. If rxy=0 then the given two variables X and Y are called uncorrelated variables.

3. Two independent variables are uncorrelated. But uncorrelated variables need be

independent. In other words, if rxy= 0, the variables X and Y need not be
independent. In this case, there may be nonlinear relationship between X and Y.

SOLVED PROBLEMS

Problem Find the coefficient of correlation between the ages of brothers (X) and ages of
sisters (Y) from the following data :

X 23 27 28 28 29 30 31 33 35 36
Y 18 20 22 27 21 29 27 29 28 29

Solution : Given,

X Y X2 Y2 XY
23 18 529 324 414
27 20 729 400 540
28 22 784 484 616
28 27 784 729 756
29 21 841 441 609
30 29 900 841 870
31 27 961 729 837
33 29 1089 841 957
35 28 1225 784 980
36 29 1296 841 1044
300 250 9138 6414 7623
Total

From the table we have,

X = 300, Y = 250, X2 = 9138, Y2 = 6414, XY = 7623 and n = 10

Dr.Mokesh Rayalu,M.Sc,Ph.D.,
 MAT2001–Statistics for Engineers

 X  Y 
 XY  n
rXY =
  X   
 X  
2
 Y  2


2

n  
 Y 
n
2


  

7623 
300 250 
10 7623  7500
= =
 300   6414 - 250 2 
2 9138  9000  6414 - 6250 
9138    
 10   10 

123 123 123

= = = = 0.8176
(138)(164 ) 22632 150.4394
The correlation coefficient between ages of brothers and Ages of sisters is
0.8176, which shows high degree of positive correlation between them.

Problem : Find the correlation coefficient between X and Y using the following data;

X = 30, Y = 40, X2 = 650, Y2 = 460 XY = 508 and n = 25

 X  Y 
 XY  n
rXY =
  X   
 X  
2
 Y  2


2

n  
 Y 
n
2


  

508 
30 40 
25 460
rXY = = = 0.9329


30 2  460  40  2  493.0963
 650   
 25   25 
rXY = 0.9329

Dr.Mokesh Rayalu,M.Sc,Ph.D.,
 MAT2001–Statistics for Engineers

We say that there is high degree of positive correlation between X and Y.

Short cut Method by Using Change of Origin

Define U=X-A and V=X-B.

Then, we can reduce the original data consists of larger values on X and Y
into deviations data with smaller values in U and V.

The coefficient of correlation is given by

U V 
UV  n
rXY = rUV =
  U   
 U  
2
V  2


2

n  
 V 
n
2


  

Problem : Calculate correlation coefficient between the ages of brothers (X) and
ages of sisters (Y) from the following data.

X 23 27 28 28 29 30 31 33 35 36
Y 18 20 22 27 21 29 27 29 28 29

Solution : Given n=10, Let A = 28, and B = 29

X Y U=X-A V=Y-B U2 V2 UV
23 18 -5 -11 25 121 55
27 20 -1 -9 1 81 9
28 22 0 -7 0 49 0
28 = A 27 0 -2 0 4 0
29 21 1 -8 1 64 -8
Dr.Mokesh Rayalu,M.Sc,Ph.D.,
 MAT2001–Statistics for Engineers

30 29=B 2 0 4 0 0
31 27 3 -2 9 4 -6
33 29 5 0 25 0 0
35 28 7 -1 49 1 -7
36 29 8 0 64 0 0

20 -40 178 324 43

From table

 U=20,  V=-40,  U2=178,  V2=324,  UV=43, and n=10

( U ) ( V )
UV - n
rXY  ruv 

 U -
2 U 2

  V -
2  V2

 n   n 

(20) (-40)
43 
rXY  ruv  10
 (20)  
2
(40) 2 
178 - 324 - 
 10   10 
43  80 123 123
 = =  0.8176
(178 - 40) (324 - 160) (138) (164) 150.4394

This shows that there is high degree of positive correlation between brothers and sisters.

Problem: Calculate the Coefficient of Correlation between X and Y from the

following data

X-Series Y-Series

Dr.Mokesh Rayalu,M.Sc,Ph.D.,
 MAT2001–Statistics for Engineers

No. of pairs of observations 15 15

Arithmetic Mean 25 18
Standard Deviation 3.0111 3.033
Sum of squares of deviations from 136 138
mean

Sum of products of deviations of X 122

and Y from their respective means

Solution : Given that n=15, X  25, Y  18,  X  3.01,  Y  3.03

 ( X  X ) 2   U 2  136,  (Y  Y ) 2   V 2  138, and  ( X  X ) (Y - Y)   UV  122
 ( X  X ) (Y - Y)  UV
 ( X  X )  (Y  Y ) 
rXY  
2 2
U V 
2 2

122 122 122

rXY   
(136 ) (138) 18768 136.9964

rXY = 0.8905

Hence, there is high degree of positive correlation between X and Y

series.

Problem : From the data given below, find the correlation coefficient. xy=130,
x =90, y2 = 640and n=10, Here x and y are deviations from arithmetic means.
2

Solution : Given,
xy = (X - X ) (Y - Y ) = 130
x2 = (X - X )2 = 90
y2 = (Y - Y )2 = 640, n = 10

Dr.Mokesh Rayalu,M.Sc,Ph.D.,
 MAT2001–Statistics for Engineers

Hence,
 (X - X) (Y - Y ) 130 130 130
 ( X  X )  (Y  Y ) 
rXY  =    0.5417
2 2
(90) (640) 57600 240

rXY = 0.5417

Hence, there is high degree of positive correlation between X and Y,

Problem : Calculate correlation coefficient from the following results.

N=10,  X=140,  Y=150,  (X-10)2 =180,  (Y-15)2 =215,

 (X-10) (Y-15) =60

Solution: Let U = (X-10) and V=(Y-15),

From the given data we find

 U =  (X-10) = X-10 N = 140 – (10 X 10) = 40
 V =  (Y-10) = X-15 N = 150 – (15X10) = 0
 U2 =  (X-10)2 = 180; V2 =  (Y-15)2 = 215
UV =  (X-10) (Y-15) = 60

U V 
UV  n
rUV 
  U   
 U  
V 
2 2


2

n  
 V 
n
2


  

60 
40 0
10 460 60
r = 


40 2  215  0 2  20 X 215 65.5244
180   
 10   10 
=0.915
Dr.Mokesh Rayalu,M.Sc,Ph.D.,
 MAT2001–Statistics for Engineers

Problem : A computer while calculating the correlation coefficient between two

variables X and Y from 30 pairs of observations obtained the following
results.
n = 30, X = 120, X2 = 600, Y = 90, Y2 = 250, XY = 356.

It was however, later discovered at the time of checking that two pairs of
observations were not correctly copied. They were taken as (8,10) and (12,
77) while the correct values were (8,12) and (10,8). Obtain the correct value
of correlation coefficient between X and Y.

Solution: Given, n = 30, X = 120, X2 = 600, Y = 90, Y2 = 250, XY = 356

Incorrect Values : Correct Values :

X Y X2 Y2 XY X Y X2 Y2 XY
8 10 64 100 80 8 12 64 144 96
12 7 144 49 84 10 8 100 64 80
Total 20 17 208 149 164 Total 18 20 164 208 176

Now, we have,
Corrected  X = 120 – 20 + 18 = 118
Corrected  Y = 90 – 17 + 20 = 93
Corrected  X = 600 – 208 + 164 = 556
2

Corrected  Y = 250 – 149 + 208 = 309

2

Corrected XY = 356 – 164 + 176 = 368

Corrected correlation coefficient is given by

Dr.Mokesh Rayalu,M.Sc,Ph.D.,
 MAT2001–Statistics for Engineers

 X  Y 
 XY  n
Corrected rXY 
  X   
 X  
2
 Y  2


2

n 
 Y 
n
2


 

368 
118 93
= 30
 118   
2
93 
2

556   309  
 30   30 

368  365.8 2.2

 rXY 
556 - 464.1333 309 - 288.3 91.8667 20.7 

2.2 2.2
rXY   = 0.504
1901 .6407 43.6078

PROBABLE ERROR CORRELATION COEFFICIENT

Probable error of correlation coefficient `r’, usually denoted by P.E.(r), is regarded as a

measure of testing the significance of an observed value of Karl Pearson’s correlation
coefficient based on a random sample drawn from a bivariate normal population.
It is defined as
1  r 2 
P.E(r) = 0.6745  
 n 
Where n is number of pairs of observations in a sample.

1  r 2 
Remark : The expression   is known the standard error (S.E.) of sample
 2 
correlation coefficient. The details of this concept will be discussed is the later chapter.
The probable error of correlation coefficient determines the limits [r  P.E. (r)] with in
which the population correlation coefficient may be expected to lie.

Interpretation of P.E. (r) :

Dr.Mokesh Rayalu,M.Sc,Ph.D.,
 MAT2001–Statistics for Engineers

The interpretation of the significance of correlation coefficient may be drawn as follows:

i.r< P.E. (r)  No evidence of significant correlation between two variables
ii.r>6 P.E.(r)  Certainly, significant correlation exists between the two variables.
iii.P.E.(r)  r  6 P.E.(r)  Definitely exists insignificant correlation between the two
variables.

Problem: Compute the Karl Pearson’s correlation coefficient between exports

and imports from the following data. Also calculate probable error of correlations
coefficient.

Year 1994 1995 1996 1997 1998 1999 2000

Exports (Rs.in Hundred Crores) 42 44 58 55 89 98 66
Imports (Rs.in Hundred Crores) 56 49 53 58 65 76 58

Solution :

Exports X Imports Y X2 Y2 XY
42 56 1764 3136 2352
44 49 1936 2401 2156
58 53 3364 2809 3074
55 58 3025 3364 3190
89 65 7921 4225 5785
98 76 8604 5776 7448
66 58 4356 3364 3828
Total 452 415 31970 25075 27833

From the table, we have

X=452, Y=415, X2 = 31970, Y2=25075 and XY =27833, n=7

Dr.Mokesh Rayalu,M.Sc,Ph.D.,
 MAT2001–Statistics for Engineers

 X  Y 
 XY  n
r 
  X   
 X  
2
 Y  2


2

n 
 Y 
n
2


 

27833 
452 415 
= 7
 452   
2
415  
2

31970   25075  
 7  7 

27833  26797.143
r 
31970 - 29186.286 25075 - 24603.571
1035 .8571 1035.8571
= 
2783 .7143 471.4286  1145.5665

r = 0.9042

The probable error of r is given by

1  r 2  1  (0.9042) 2   0.1824 
P.E.(r) = 0.6745   = 0.6745   = 0.0465  = 0.0465
 n   7   2.6458 

P.E.(r) = 0.0465

Also, [(6) P.E (r)] = (6) (0.0465) = 0.279.

Since r > [6.P.E (r) ], We say that r is definitely significant.

i.e., There is significant correlation between exports and imports

SPEARMAN'S RANK CORRELATION COEFFICIENT

Dr.Mokesh Rayalu,M.Sc,Ph.D.,
 MAT2001–Statistics for Engineers

Rank correlation coefficient is useful for finding correlation between any two qualitative
characteristics such as Beauty, Honesty, Intelligence etc., which can not be measured
quantitatively but can be arranged serially in order of merit or proficiency possessing the
two characteristics.

Suppose we associate the ranks to individuals or items in two series based on order of
merit, the Spearman's Rank correlation coefficient  is given by

 6 d2 
 1-  2  [Read the symbol ( as 'Rho'.]
 n(n - 1) 
Where,
d2 = Sum of squares of differences of ranks between paired items in two
series
n = Number of paired items

Remarks

Spearman's rank correlation coefficient can be used to find the correlation

between two quantitative characteristics or variables. In this case, we
associate the ranks to the observations based on their magnitudes for X and
Y series separately. Let RX and Ry be the ranks of observations on two
variables X and Y respectively for a pair. Then the Spearman's rank
correlation coefficient is given by
 6 d2 
p 1-  2 
 n(n - 1) 

Where, d2=(Rx-Ry)2=sum of squares of differences between the ranks of

variables X and Y
n = number of pairs of observations

SPEARMAN'S RANK CORRELATION COFFICIENT FOR A DATA WITH

TIED OBSERVATIONS

Dr.Mokesh Rayalu,M.Sc,Ph.D.,
 MAT2001–Statistics for Engineers

In any series, if two or more observations are having same values then the observations
are said to be tied observations. If tie occurs for two or more observations in a series, then
common ranks have to be given to the tied observations in that series; these common
ranks are the average of the ranks, which these observations would have assumed if they
were slightly different from each other and the next observation will get the rank next to
the rank already assumed.

In the case of data with tied observations, the Spearman's rank correlation coefficient is
given by

 6 Adj  d 2
 1 - 

2 
 n(n - 1) 
Where
 S13  S1   S 2 3  S 2   S 3 3  S 3 
Adj  
d 2
 d 2
 
12

12
 
 12
  ...,

    
Here,
S1 is the number of times first tied observation is repeated
S2 is the number of times second tied observation is repeated
S3 is the number of times third observation is repeated etc.

Remarks

1. The value of Spearman's rank correlation coefficient always lies between -1

and +1.
i.e., -1    + 1
2. The interpretation of the Spearman's rank correlation coefficient () is similar to the
interpretation of the Karl Pearson's correlation coefficient (r).

3. The values of  and r for a data may generally be different.

4. In the calculation of Karl Pearson's correlation coefficient, generally, we assume that

the sample observations must be drawn from a normal population.
While in the case of Spearman' rank correlation coefficient, the population may not be
specified with a probability distribution function. Thus,  may be considered as a non
parametric measure.
Dr.Mokesh Rayalu,M.Sc,Ph.D.,
 MAT2001–Statistics for Engineers

PROBLEMS ON SPEARMAN'S RANK CORRELATION COEFFICIENT

Problem : In a quantitative aptitude test, two judges rank the ten competitors in
the following order.

Competitor 1 2 3 4 5 6 7 8 9 10
Ranking of 4 5 2 7 8 1 6 9 3 10
judge I
Ranking of 8 3 9 10 6 7 2 5 1 4
judge II

Is there any concordance between the two judges ?

Solution: Let Rx : Ranking by Judge I and Ry: Ranking by Judge II

The Spearman's rank correlation coefficient is given by

6 d2 
= 1 -  2 
 n(n - 1) 
Where, d2 =  (Rx-Ry)2, n= Number o competitors

Rx Ry d= Rx-Ry d2
4 8 -4 16
5 3 2 4
2 9 -7 49
7 10 -3 9
8 6 2 4
1 7 -6 36
6 2 4 16
9 5 4 16
3 1 2 4
10 4 6 36
Total 190

Dr.Mokesh Rayalu,M.Sc,Ph.D.,
 MAT2001–Statistics for Engineers

(6)(190)
 = 1 = 1-1.1515
10(100  1)
 = - 0.1515
We say that there is low degree of negative rank correlation between the
two judges.

Problem : Twelve recruits were subjected to selection test to ascertain their suitability
for a certain course of training. At the end of training they were given a proficiency test.
The marks scored by the recruits are recorded below :

Recruit 1 2 3 4 5 6 7 8 9 10 11 12
Selection Test 44 49 52 54 47 76 65 60 63 58 50 67
Score
Proficiency 48 55 45 60 43 80 58 50 77 46 47 65
Test Scrore

calculate rank correlation coefficient and comment on your result.

Solution: Let selection test score be a variable X and proficiency test score be a variable
Y. We associate the ranks to the scores based on their magnitudes.

The spearman's rank correlation coefficient is given by

 6 d2 
p 1 -  2 
 n(n - 1) 

Where,  d 2   Rx  Ry2 = Sum of squares of differences between the ranks of

observations on X and Y

n = Number of recruits
Given

X Y RX RY d=RX-RY d2
44 48 12 8 4 16
Dr.Mokesh Rayalu,M.Sc,Ph.D.,
 MAT2001–Statistics for Engineers

49 55 10 6 4 16
52 45 8 11 -3 9
54 60 7 4 3 9
47 43 11 12 -1 1
76 80 1 1 0 0
65 58 3 5 -2 4
60 50 5 7 -2 4
63 77 4 2 2 4
58 46 6 10 -4 16
50 47 9 9 0 0
67 65 2 3 -1 1
-- -- -- -- Total 80

From the table, we have,

d 2
= 80, n = 12

 (6) (80) 
p  1 -    1 - 0.2797  0.7203
12 (144 - 1) 

  0.7203

We say that there is high degree of positive rank correlation between the
scores of selection and proficiency tests.

Problem : Following is the data on heights and weights of ten students in a class:

Heights 140 142 140 160 150 155 160 157 140 170
(in cm)
Weights 43 45 42 50 45 52 57 48 49 53
(in cm)

Calculate rank correlation coefficient between heights and weights of students.

Solution:
Dr.Mokesh Rayalu,M.Sc,Ph.D.,
 MAT2001–Statistics for Engineers

Let height be a variable X and weight be a variable Y. Since, the data contains tied
observations, we associate average ranks to the tied observations.

The spearman's rank correlation coefficient is given by

 6(Adj  d 2 ) 
 1-  2 .
 n(n - 1 
 S13  S1   S32  S1 
Where, Adj  d 2   d  
2
  +….
 12   12 
n = Number of students

X Y RX RY d=RX-RY d2
140 43 9 9 0 0
142 45 7 7.5 -0.5 0.25
140 42 9 10 -1 1
160 50 2.5 4 -1.5 2.25
150 45 6 7.5 -1.5 2.25
155 52 5 3 2 4
160 57 2.5 1 1.5 2.25
157 48 4 6 -2 4
140 49 9 5 4 16
170 53 1 2 -1 1
-- -- -- -- Total 33.00

From the table, We have,

n = 10, d 2
 33 S1 = 3, S2 = 2, S3 = 33

  33  3   2 3  2   2 3  2  
Thus, Adj  d  33   
2
  
  12   12   12  

= 33 + 2 + 0.5 + 0.5
Dr.Mokesh Rayalu,M.Sc,Ph.D.,
 MAT2001–Statistics for Engineers

= 36

 (6) (36) 
 1-    1 - 0.2182  0.7818
10 (100 - 1) 
  0.7818
We say that there is high degree of positive rank correlation between
heights and weights of students.

Dr.Mokesh Rayalu,M.Sc,Ph.D.,