UNIT 6 CORRELATION COEFFICIENT Correlation

Coefficient

Structure
6.1 Introduction
Objectives
6.2 Concept and Definition of Correlation
6.3 Types of Correlation
6.4 Scatter Diagram
6.5 Coefficient of Correlation
Assumptions for Correlation Coefficient
6.6 Properties of Correlation Coefficient
6.7 Short-cut Method for the Calculation of Correlation Coefficient
6.8 Correlation Coefficient in Case of Bivariate Frequency Distribution
6.9 Summary
6.10 Solutions / Answers

6.1 INTRODUCTION
In Block 1, you have studied the various measures such as measures of central
tendency, measures of dispersion, moments, skewness and kurtosis which
analyse variables separately. But in many situations we are interested in
analysing two variables together to study the relationship between them. In
this unit, you will learn about the correlation, which studies the linear
relationship between the two or more variables. You would be able to
calculate correlation coefficient in different situations with its properties. Thus
before starting this unit you are advised to go through the arithmetic mean and
variance that would be helpful in understanding the concept of correlation.
In Section 6.2, the concept of correlation is discussed with examples, that
describes the situations, where there would be need of correlation study.
Section 6.3 describes the types of correlation. Scatter diagrams which give an
idea about the existence of correlation between two variables is explained in
Section 6.4. Definition of correlation coefficient and its calculation procedure
are discussed in Section 6.5. In this unit, some problems are given which
illustrate the computation of the correlation coefficient in different situations
as well as by different methods. Some properties of correlation coefficient
with their proof are also given. In Section 6.6 the properties of the correlation
coefficient are described whereas the shortcut method for the calculation of
the correlation coefficient is explained in Section 6.7. In Section 6.8 the
method of calculation of correlation coefficient in case of bivariate frequency
distribution is explored.
Objectives
After reading this unit, you would be able to
 describe the concept of correlation;
 explore the types of correlation;
 describe the scatter diagram; 25
Correlation for Bivariate Data  interpret the correlation from scatter diagram;
 define correlation coefficient;
 describe the properties of correlation coefficient; and
 calculate the correlation coefficient.

6.2 CONCEPT AND DEFINITION OF

CORRELATION
In many practical applications, we might come across the situation where
observations are available on two or more variables. The following examples
will illustrate the situations clearly:
1. Heights and weights of persons of a certain group;
2. Sales revenue and advertising expenditure in business; and
3. Time spent on study and marks obtained by students in exam.
If data are available for two variables, say X and Y, it is called bivariate
distribution.
Let us consider the example of sales revenue and expenditure on advertising in
business. A natural question arises in mind that is there any connection
between sales revenue and expenditure on advertising? Does sales revenue
increase or decrease as expenditure on advertising increases or decreases?
If we see the example of time spent on study and marks obtained by students,
a natural question appears whether marks increase or decrease as time spent on
study increase or decrease.
In all these situations, we try to find out relation between two variables and
correlation answers the question, if there is any relationship between one
variable and another.
When two variables are related in such a way that change in the value of one
variable affects the value of another variable, then variables are said to be
correlated or there is correlation between these two variables.
Now, let us solve a little exercise.
E1) What do you mean by Correlation?

6.3 TYPES OF CORRELATION

According to the direction of change in variables there are two types of

correlation
1. Positive Correlation
2. Negative Correlation
1. Positive Correlation
Correlation between two variables is said to be positive if the values of the
variables deviate in the same direction i.e. if the values of one variable
increase (or decrease) then the values of other variable also increase (or
26
decrease). Some examples of positive correlation are correlation between
1. Heights and weights of group of persons; Correlation
2. House hold income and expenditure; Coefficient
3. Amount of rainfall and yield of crops; and
4. Expenditure on advertising and sales revenue.
In the last example, it is observed that as the expenditure on advertising
increases, sales revenue also increases. Thus, the change is in the same
direction. Hence the correlation is positive.
In remaining three examples, usually value of the second variable increases
(or decreases) as the value of the first variable increases (or decreases).
2. Negative Correlation
Correlation between two variables is said to be negative if the values of
variables deviate in opposite direction i.e. if the values of one variable increase
(or decrease) then the values of other variable decrease (or increase). Some
examples of negative correlations are correlation between
1. Volume and pressure of perfect gas;
2. Price and demand of goods;
3. Literacy and poverty in a country; and
4. Time spent on watching TV and marks obtained by students in
examination.
In the first example pressure decreases as the volume increases or pressure
increases as the volume decreases. Thus the change is in opposite direction.
Therefore, the correlation between volume and pressure is negative.
In remaining three examples also, values of the second variable change in the
opposite direction of the change in the values of first variable.
Now, let us solve a little exercise.

E2) Explore some examples of positive and negative correlations.

6.4 SCATTER DIAGRAM

Scatter diagram is a statistical tool for determining the potentiality of
correlation between dependent variable and independent variable. Scatter
diagram does not tell about exact relationship between two variables but it
indicates whether they are correlated or not.
Let ( x i , yi ) ; (i  1,2,..., n) be the bivariate distribution. If the values of the
dependent variable Y are plotted against corresponding values of the
independent variable X in the XY plane, such diagram of dots is called scatter
diagram or dot diagram. It is to be noted that scatter diagram is not suitable for
large number of observations.
6.4.1 Interpretation from Scatter Diagram
In the scatter diagram
1. If dots are in the shape of a line and line rises from left bottom to the right
top (Fig.1), then correlation is said to be perfect positive.
27
Correlation for Bivariate Data

Fig. 1: Scatter diagram for perfect positive correlation

2. If dots in the scatter diagram are in the shape of a line and line moves from
left top to right bottom (Fig. 2), then correlation is perfect negative.

Fig. 2: Scatter diagram for perfect negative correlation

3. If dots show some trend and trend is upward rising from left bottom to
right top (Fig.3) correlation is positive.

28 Fig. 3: Scatter diagram for positive correlation

 Correlation
4. If dots show some trend and trend is downward from left top to the right Coefficient
bottom (Fig.4) correlation is said to be negative.

Fig. 4: Scatter diagram for negative correlation

5. If dots of scatter diagram do not show any trend (Fig. 5) there is no
correlation between the variables.

Fig. 5: Scatter diagram for uncorrelated data

6.5 COEFFICIENT OF CORRELATION

Scatter diagram tells us whether variables are correlated or not. But it does not
indicate the extent of which they are correlated. Coefficient of correlation
gives the exact idea of the extent of which they are correlated.

Coefficient of correlation measures the intensity or degree of linear

relationship between two variables. It was given by British Biometrician Karl
Pearson (1867-1936). 29
Correlation for Bivariate Data Note: Linear relationships can be expressed in such a way that the
independent variable is multiplied by the slope coefficient, added by
a constant, which determines the dependent variable. If Y is a dependent
variable, X is an independent variable, b is a slope coefficient and a is
constant then linear relationship is expressed as Y  a  bX .
In fact linear relationship is the relationship between dependent and
independent variables of direct proportionality. When these variables
plotted on a graph give a straight line.

If X and Y are two random variables then correlation coefficient between X

and Y is denoted by r and defined as
Cov (x , y )
r  Corr( x , y)  …(1)
V ( x ) V ( y)
Corr ( x , y) is indication of correlation coefficient between two variables X and
Y.
Where, Cov( x , y) the covariance between X and Y which is defined as:
1 n
Cov(x, y)   ( x i  x)( yi  y)
n i 1
and V (x ) the variance of X, is defined as:
1 n
V( x )   (x i  x ) 2 ,
n i 1
n
Note: In the above expression i 1
denotes the sum of the values for i =
n
1 to n; For example x
i 1
i means sum of values of X for i = 1 to n. If n =
2
2 i.e.  x i which is equivalent to x 1  x 2 . If limits are not written the
i 1

summation expression i.e.  x , which indicates the sum of all values

of X. You may find the discussed formulae without limits in many
1 n 1
books. We can also write  ( x i  x )(y i  y) as  (x  x )(y  y)
N i1 N
and both have same meaning.

Similarly,
V ( y) the variance of Y is defined by
1 n
V ( y)   ( y i  y) 2
n i 1
where, n is number of paired observations.
Then, the correlation coefficient “r” may be defined as:

30
 1 n Correlation
 ( x i  x )( yi  y)
n i1 Coefficient
r  Corr( x , y)  … (2)
1 n  1 n 
  ( x i  x ) 2   ( y i  y) 2 
 n i1  n i1 
Karl Pearson’s correlation coefficient r is also called product moment
correlation coefficient. Expression in equation (2) can be simplified in various
forms. Some of them are
n

 (x
i 1
i  x )( y i  y)
r … (3)
 n  n 
  ( x i  x ) 2   ( y i  y) 2 
 i1  i1 
or
1 n
 x i yi  x y
n i 1
r … (4)
 n 2  n 2 
  x i   y i 

i1
 x 2  i 1  y2 
 n  n 
  
or
n

x y
i 1
i i  nx y
r … (5)
n 2 2 
n
2 2
 i x  n x  y i  ny 
 i1  i1 
or
n n n
n  x i y i   x i  yi
i 1 i 1 i 1
r … (6)
2 2
 n
2    2
n
 n   n

  i
n x    i   i  
x  n y   y i  
 i1  i1   i1 i 1  

6.5.1 Assumptions for Correlation Coefficient

1. Assumption of Linearity
Variables being used to know correlation coefficient must be linearly
related. You can see the linearity of the variables through scatter diagram.
2. Assumption of Normality
Both variables under study should follow Normal distribution. They
should not be skewed in either the positive or the negative direction.
3. Assumption of Cause and Effect Relationship
There should be cause and effect relationship between both variables, for
example, Heights and Weights of children, Demand and Supply of goods,
etc. When there is no cause and effect relationship between variables then
correlation coefficient should be zero. If it is non zero then correlation is
termed as chance correlation or spurious correlation. For example,
correlation coefficient between: 31
Correlation for Bivariate Data
1. Weight and income of a person over periods of time; and
2. Rainfall and literacy in a state over periods of time.
Now, let us solve a little exercise.

E3) Define correlation coefficient.

6.6 PROPERTIES OF CORRELATION

COEFFICIENT
Property 1: Correlation coefficient lies between -1 and +1.
Description: Whenever we calculate the correlation coefficient by any one of
the formulae given in the Section 6.5 its value always lies between -1 and +1.
Proof: Consider
2
1  x  x   y  y 
    0
n   x    y 
(Since square quantity is always greater than or equal to zero)
2 2
1  x  x  1  y  y  1  x  x  y  y  
       2    0
n   x  n   y  n   x   y  
2
 2x  y 2Cov (x , y)
 2 2 0
x  y xy
1
Since
n
x  x2  Variance of X  σ2x,
1 2 1
Similarly, y  y  σ 2y and x  x y  y  Cov (x, y)
n n

Therefore, after putting the values

 1  1  2r  0
 2  2r  0
1 r  0 … (7)
If we take positive sign in equation (7) then
r  1
It shows that the correlation coefficient will always be greater than or equal
to-1.
If we take negative sign in equation (7) then
1 r  0
r 1
It shows that correlation coefficient will always be less than or equal to +1.
Thus

32  1  r  1
If r  1 , the correlation is perfect positive and if r  1 correlation is perfect Correlation
negative. Coefficient

Property 2: Correlation coefficient is independent of change of origin and

scale.
Description: Correlation coefficient is independent of change of origin and
scale, which means that if a quantity is subtracted and divided by another
Xa
quantity (greater than zero) from original variables , i.e. U  and
h
Yb
V then correlation coefficient between new variables U and V is
k
same as correlation coefficient between X and Y, i.e. Corr( x , y)  Corr(u , v) .
x a yb
Proof: Suppose u  and v  then
h k
x  a  h u and x  a  h u … (8)
and y  a  k v and y  a  h v … (9)
where, a, b, h and k are constants such that a > 0, b > 0, h > 0 and k > 0.
We have to prove Corr( x , y)  Corr (u , v) i.e. there is no change in correlation
when origin and scale are changed.
1
Cov(x, y) 
n
 x  x y  y 
1

n
 a  h u  a  h u b  k v  b  b v 
1
 hk  u  u v  v  ,
n
Cov ( x , y)  hkCov ( u , v )
and
1 2
V( x ) 
n
 x  x 
1
  a  hu  a  h u 2
n
1
 h2  u  u 2
n
V( x )  h 2 V( u )
Similarly,
V( y)  k 2V(v)
Cov(x, y)
Corr( x, y) 
V ( x ) V ( y)
hkCov ( u , v)
Corr ( x , y) 
h 2 V(u)k 2 V(v) 33
Correlation for Bivariate Data Cov(u, v)
Corr( x, y) 
V ( u ) V ( v)
Cor r (x , y )  Corr ( u , v)
i.e. correlation coefficient between X and Y is same as correlation coefficient
between U and V Thus, correlation coefficient is independent of change of
origin and scale.
Property 3: If X and Y are two independent variables then correlation
coefficient between X and Y is zero, i.e. Corr ( x , y)  0 .
Proof: Covariance between X and Y is defined by
1
Cov(x, y) 
n
 x  x y  y
1

n
 xy  y x  x y  x y
1 1 1 1

n
 xy  x  y  y  x  x y 1
n n n
1

n
 xy  x y  y x  x y
1

n
 xy  x y
 x y x y
1
(if variables are independent then,
n
 xy  x y ). Therefore,
Cov ( x, y)  0
Thus, correlation is
Cov(x, y) 0
Corr(x, y)   0
V ( x ) V ( y) V( x)V( y)
As correlation measures the degree of linear relationship, different values of
coefficient of correlation can be interpreted as below:
Value of correlation coefficient Correlation is
+1 Perfect Positive Correlation
-1 Perfect Negative Correlation
0 There is no Correlation
0 - 0.25 Weak Positive Correlation
0.75 - (+1) Strong Positive Correlation
−0.25 - 0 Weak Negative Correlation
−0.75 - (−1) Strong Negative Correlation

Let us discuss some problems of calculation of correlation coefficient.

34
Example 1: Find the correlation coefficient between advertisement Correlation
expenditure and profit for the following data: Coefficient

Advertisement expenditure 30 44 45 43 34 44

Profit 56 55 60 64 62 63

Solution: To find out the correlation coefficient between advertisement

expenditure and profit, we have Karl Pearson’s formula in many forms [(2),
(3), (4), (5) and (6)] and any of them can be used. All these forms provide the
same result. Let us take the form of equation (3) to solve our problem which
is
n

 (x
i 1
i  x )( y i  y )
r  Corr( x , y) 
 n  n 
  (x i  x ) 2   ( y i  y) 2 
 i 1  i 1 
Steps for calculation are as follow:
1. In columns 1 and 2, we take the values of variables X
and Y respectively.
2. Find sum of the variables X and Y i.e.
6 6

 x i  240 and
i 1
y
i 1
i  360

3. Calculate arithmetic means of X and Y as

n 6

x
i 1
i x
i 1
i
240
x    40
n 6 6
n 6

 yi
i 1
y
i 1
i
360
and y    60
n 6 6
4. In column 3, we take deviations of each observations of X from mean of
X, i.e. 30 − 40 = −10, 44 − 40 = 4 and so on other values of the column
can be obtained.
5. Similarly column 5 is prepared for variable Y i.e.
56 – 60 = −4, 55 − 60 = −5
and so on.
6. Column 4 is the square of column 3 and column 6 is the square of column
5.
7. Column 7 is the product of column 3 and column 5.
8. Sum of each column is obtained and written at the end of column.

35
Correlation for Bivariate Data
To find out the correlation coefficient by above formula, we require the values
n n n
of  (x i  x )( yi  y) ,  (x i  x ) 2 and  (y i  y)2 which are obtained by the
i 1 i 1 i 1
following table:

x y (x  x) (x  x)2 (y  y ) (y  y)2 (x  x)(y  y)

30 56 −10 100 −4 16 40
44 55 4 16 −5 25 −20
45 60 5 25 0 0 0
43 64 3 9 4 16 12
34 62 −6 36 2 4 −12
44 63 4 16 3 9 12
6 6
6 6 6

x i
y i  (x i  x)  (x i  x) 2  (y i  y)
7

 ( yi  y)2  (x
i 1
i  x)( y i  y)
i 1 i 1 i 1 i 1 i 1

 240  360 0  202 0  70  32

6 6 6
Taking the values of  ( x i  x )( yi  y ) ,  (x i  x ) 2 and  (y i  y)2 from
i 1 i 1 i 1
the table and substituting in the above formula we have the correlation
coefficient
6

 (x
i 1
i  x)( yi  y)
r  Corr(x, y) 
6 2 
6
2
 ( x i  x)  ( yi  y) 
 i 1  i 1 
32 32 32
r  Corr(x , y)     0.27
202  70 14140 118.91
Hence, the correlation coefficient between expenditure on advertisement and
profit is 0.27. This indicates that the correlation between expenditure on
advertisement and profit is positive and we can say that as expenditure on
advertisement increases (or decreases) profit increases (or decreases). Since it
lies between 0.25 and 0.5 it can be considered as week positive correlation
coefficient.
Example 2: Calculate Karl Pearson’s coefficient of correlation between price
and demand for the following data.
Price 17 18 19 20 22 24 26 28 30
Demand 40 38 35 30 28 25 22 21 20

Solution: In Example 1, we used formula given in equation (3) in which

deviations were taken from mean. When means of x and y are whole number,
deviations from mean makes calculation easy. Since, in Example 1, means x
and y were whole number we preferred formula given in equation (3). When
means are not whole numbers calculation by formula given in equation (3)
becomes cumbersome and we prefer any formula given in equation (4) or (5)
or (6). Since here means of x and y are not whole number, so we are
36 preferring formula (6)
 n n n
Correlation
n  x i yi   x i  yi Coefficient
i 1 i 1 i 1
r
2 2
 n 2  n   n 2  n  
n  x i    x i    i
n y    i 
y 
 i 1  i 1   i 1  i 1  

Let us denote price by the variable X and demand by variable Y.

To find the correlation coefficient between price i.e.X and demand Y using
n n n
formula given in equation (6), we need to calculate,  x i ,  yi , x y , i i
i 1 i 1 i 1
n n

 x i2 and
i 1
y
i 1
2
i which are being obtained in the following table:

x y x2 y2 xy
17 40 289 1600 680
18 38 324 1444 684
19 35 361 1225 665
20 30 400 900 600
22 28 484 784 616
24 25 576 625 600
26 22 676 484 572
28 21 784 441 588
30 20 900 400 600
 x  204  y  259 x 2
 4794 y 2
 7903  xy  5605
n n n
n  x i yi   x i  yi
i 1 i 1 i 1
r
2 2
 n 2  n   n 2  n  
n  x i    x i  n  y i    y i  
 i 1  i 1   i 1  i 1  

r  Corr(x, y)
(9  5605)  (204)(259)

(9  4794)  (204  204)(9  7903)  (259  259)
50445  52836
r  Corr( x, y) 
(43146  41616)  (71127  67081)
 2391
r  Corr ( x , y) 
1530  4046
 2391
r  Corr ( x , y ) 
2488 .0474
r  Corr ( x , y)   0.96

37
 n
Correlation for Bivariate Data
Note: We can use  x instead of  x i . Second expression indicates sum
i 1

over x i for i =1 to n. On the other hand first expression (  x ) indicates sum

over all values of X. In the current example we are using summation sign
( ) without limit.
Now, let us solve the following exercises.

E4) Calculate coefficient of correlation between x and y for the following

data:
x 1 2 3 4 5
y 2 4 6 8 10

E5) Find the coefficient of correlation for the following ages of husband and
wife:

Husband’s age 23 27 28 29 30 31

Wife’s age 18 22 23 24 25 26

6.7 SHORT-CUT METHOD FOR THE

CALCULATION OF CORRELATION
COEFFICIENT
When values of variables are big and actual means of variables X and Y i.e. x
and y are not whole number (in Example 1 mean of X and Y i.e. x  40 and
y  60 were whole number ) then calculation of correlation coefficient by the
formula (2), (3), (4), (5) and (6) is somewhat cumbersome and we have
shortcut method in which deviations are taken from assumed mean i.e. instead
of actual means x and y , we use assumed mean, hence ( x i  x ) and
( yi  y) are replaced by x i  A x  d x and y i  A y  d y where A x and A y are
assumed means of (Assumed mean may be any value of given variable of our
choice) variables X and Y respectively. Formula for correlation coefficient
by shortcut method is
nd xd y   d x  d y
r  Corr( x , y) 
n  d 2
x 
 ( d x ) 2 n  d 2y  ( d y ) 2 
Here,
n = No. of pairs of observations,
A x = Assumed mean of X,
A y = Assumed mean of Y,

 d =  (x  A
x x ) : Sum of deviation from assumed mean Ax in X-series,
38
 d =  (x  A ) : Sum of deviation from assumed mean A in Y-series,
y y y Correlation
Coefficient
 d d   (x  A )( y  A ) : Sum of product of deviations from assumed
x y x y

means A X and A Y in x and y series respectively ,

2 2
 d =  (x  A
x x ) : Sum of squares of the deviations from assumed
mean A x , in x series and
2 2
 d =  (y  A
y y ) : Sum of squares of the deviations from assumed mean
A y in y series .
Note: Results from usual method and short-cut method are same.

Example 3: Calculate correlation coefficient from the following data by short-

cut method:
x 10 12 14 18 20
y 5 6 7 10 12

Solution: By short-cut method correlation coefficient is obtained by

n  d xd y   d x  d y
r  Corr(x , y) 
 nd 2
x  ( d x ) 2  n  d 2
y  ( d y ) 2 
2 2
d , d , d
x y x dy
,
d x
and d y
are being obtained from the following
table.
Let A x = Assumed mean of X =14 and A y = Assumed mean of Y = 7

x y dx = x-14 dx2 dy= y-7 dy2 dxdy

1 5 10-14 =-4 16 5 −7 = −2 4 8
12 6 12−14 = −2 4 6 −7 = −1 1 2
14 7 14−14 = 0 0 7−7 = 0 0 0
18 10 18−14 = 4 16 10−7 = 3 9 12
20 12 20−14 = 6 36 12−7 = 5 25 30
2
x y d x 4 d x d y
2
 d  d xd y
y
 74  72  39  52
 40 5

Putting the required values in above formula

(5  52)  (4  5)
r
(5  72)  (4  4)(5  39)  (5  5)
260  20
r
360  16195  25

39
Correlation for Bivariate Data 240 240
r   0.99
344170 241.8264
Thus, there is a very high correlation between x and y.
Now, let us solve an exercise.

E6) Find correlation coefficient between the values of X and Y from the
following data by short -cut method:

x 10 20 30 40 50
y 90 85 80 60 45

6.8 Correlation Coefficient in Case of Bivariate

Frequency Distribution

In bivariate frequency distribution one variable is presented in row and

another in column and corresponding frequencies are given in cells (See
Example 4).
If we consider two variables X and Y where, the distribution of X is given in
columns and the distribution of Y is given in row. In this case we adopt the
following procedure to calculate correlation coefficient.
1. Other than the given bivariate frequency distribution make three columns
in the right of the table ( f x d x , f x d 2x and f x d x d y ) two columns in left of the
table (mid value for x and class interval of variable (x) d x ), three rows in
the bottom of the table f y d y , f y d 2y and f y d x d y  and two rows in the top of
the table (mid value for y and d y ). Where fx is the sum of all frequencies
for the given x value i.e. fx= y f xy and fy is the sum of all frequencies for
the given y values i.e. fy=  x
f xy
2. Find the mid value x and d x  x i  A x i.e. deviation from assumed mean
A x or step deviation i.e. d x  x i  A x  / h where h is such that
x i  A x  / h is a whole number.
3. Apply step (2) for variable Y also.
N
4. Find f x d x by multiplying d x by respective frequency fx and get f
i 1
x dx .
N
5. Find f y d y by multiplying d y by respective frequency fy and get  f y d y .
i 1
N
6. Find f y d 2y by multiplying d 2y by respective frequency fy and get  f y d 2y .
i 1
N
7. Find f x d 2x by multiplying d 2x by respective frequency fx and get  f x d 2x .
i 1

8. Multiply respective d x and d y for each cell frequency and put the figures
in left hand upper corner of each cell.
40
9. Find f xy d x d y by multiplying fxy with d x d y and put the figures in right hand Correlation
Coefficient
lower corner of each cell and we apply the following formula:

f x d x  f yd y
f xy d xd y 
N
r
2
 2
( f x d x )  2
( f y d y ) 2 
 f x d x   y y
f d  
 N  N 

where, N =  f x   y y .

Example 4: Calculate the correlation coefficient between ages of husbands

and ages of wives for the following bivariate frequency distribution:
Ages of Ages of Wives
Husbands 10-20 20-30 30-40 40-50 50-60 Total
15-25 6 3 - - - 9
25-35 3 16 10 - - 29
35-45 - 10 15 7 - 32
45-55 - - 7 10 4 21
55-65 - - - 4 5 9
Total 9 29 32 21 9 100

Solution: Let, d y = (y − 35)/10, where assumed mean A y  35 and y = 10.

dx = (x − 40) /10, where assumed mean Ax = 40 and h = 10.

CI 10-20 20-30 30-40 40-50 50-60

MV 15 25 35 45 55
(y)

CI MV dx -2 -1 0 +1 +2 fx fx d x fx d 2x f xy d xd y
(x)
dy
15-25 20 -2 4 2 - - - 9 -18 36 30
6 3
24 6
25-35 30 -1 2 1 0 - - 29 -29 29 22
3 16 10
6 16 0
35-45 40 0 - 0 0 0 - 32 0 0 0
10 15 7
0 0 0
45-55 50 +1 - - 0 1 2 21 21 21 18
7 10 4
0 10 8
55-65 60 +2 - - - 2 4 9 18 36 28
4 5
8 20
fy 9 29 32 21 9 N= d 2x f
100
f x dx f x xy dxd y
=-8 = 122 = 98

-18 -29 0 21 18
f ydy f y dy
= -8
f y d 2y 36 29 0 21 36
f y d 2y
= 122
30 22 0 18 28
f xyd xd y f xy d xdv
41
= 98
Correlation for Bivariate Data (8  8)
98 
r 100
2
 (8)  (8) 2 
122  122  
 100  100 

98  0.64
r = 0.802
122  0.64122  0.64

6.8 SUMMARY

In this unit, we have discussed:

1. Concept of correlation;
2. Types of correlation;
3. The scatter diagrams of different correlations;
4. Calculation of Karl Pearson’s coefficient of correlation;
5. Short-cut method of calculation of correlation coefficient;
6. Properties of correlation coefficient; and
7. Calculation of correlation coefficient for bi-variate data.

6.9 SOLUTIONS / ANSWERS

E1) When two variables are related in such a way that change in the value of
one variable affects the value of another variable, then variables are said
to be correlated or there is correlation between these two variables.

E2) Positive correlation: Correlation between

(i) Sales and profit
(ii) Family Income and year of education
Negative correlation: Correlation between
(i) No. of days students absent in class and score in exam
(ii) Time spent in office and time spent with family

E3) Coefficient of correlation measures the intensity or degree of linear

relationship between two variables. It is denoted by r. Formula for the
calculation of correlation coefficient is
1 n
 ( x i  x )(yi  y)
n i 1
r  Corr( x , y) 
1 n  1 n 
  ( x i  x ) 2   ( yi  y) 2 
 n i 1  n i 1 

42
E4) We have some calculation in the following table: Correlation
Coefficient

︶
2 2
x y (x  x) (x  x) (y  y) (y  y) (x  x)(y  y

1 2 -2 4 -4 16 8
2 4 -1 1 -2 4 2
3 6 0 0 0 0 0
4 8 1 1 2 4 2
5 10 2 4 4 16 8
15 30 0 10 0 40 20
15 30 0 10 0 40 20

Here  x  15
 x   x / n  15 / 5  3
and
 y  30
 y   y / n  30 / 5  6
From the calculation table, we observe that
 (x  x)2  10 ,  (y  y)2  40 and  (x  x)(y  y)  20
Substituting these values in the formula
n

 (x
i 1
i  x )( y i  y)
r
 n  n 
  ( x i  x ) 2   ( y i  y) 2 
 i1  i1 
20 20
r  Corr ( x , y )   1
10  40 20
Hence, there is perfect positive correlation between X and Y.
E5) Let us denote the husband’s age as X and wife’s age by Y
x y x2 y2 xy
23 18 529 324 414
27 22 729 484 594
28 23 784 529 644
29 24 841 576 696
30 25 900 625 750
31 26 961 676 806
2 2
 x  168  y  138  x  4744 y  3214  xy  3904
Here,
2
 x  168 ,  y  138 ,  x  4744 ,
2
 y  3214  xy  3904 43
Correlation for Bivariate Data
We use the formula
n  xy ( x )( y )
r  Corr( x , y) 
 2 2
 2 2

n  x  ( x ) n  y  ( y) 
  
(6  3904)  (168)(138)
r
(6  4744)  (168  168)(6  3214)  (138  138)
23424  23184
r
28464  2822419284  19044
23424  23184
r =1
240  240
Hence there is perfect positive correlation between X and Y.
E6) By short-cut method correlation coefficient is obtained by
n d xd y   d x  d y
r  Corr(x , y) 
n  d 2
x 
 ( d x ) 2 n  d 2y  ( d y )2 
2 2
d , d , d d , d
x y x y x and d y are being obtained through the
following table.
Let A x = assumed mean of X = 30 and A y = Assumed mean of Y = 70

x y dx=x-30 dx2 dy=y-70 dy2 dxdy

10 90 -20 400 20 400 -400
20 85 -10 100 15 225 -150
30 80 0 0 10 100 0
40 60 10 100 -10 100 -100
50 45 20 400 -25 625 -500
2 2
 x  150  y  360 d x 0 d x  1000 d y  10 d y  1450 d d
x y  1150

Putting the required values in above formula

(5  1150)  (0  10)
r
(5  1000)  (0)(5  1450)  (10  10)
 5750
r
50007250  100
 5750
r  0.96.
5979 .1304

44