Correlation Analysis CHAPTER

z1 INTRODUCTION
7
We know that the area A of a circle of
the larger
radius will always have a
larger
radius r is given by A P
area than a circle This gives us that a
with functional|relationship between A
exists a and r. Now consider the two smaller radius. Thus circle
with a
weight offa given group of people. We know that in most of the variables h and w, the there 31
and higher weights as compared to
persons of cases the height 6
have short fat person who may haye a highershorter height. But we also come taller people will
the weight of person from his weight as compared to a lean tallacross, the sarme
time, a determine at
height person. Thus we
the statement.though
not
can about his weight, i.e., we cannot put we could try to
guess "A tall
inweighrthan ashort person'" , as an exact rule to be true in allperson
make a
cases. is more likely toreasonable
be heavier
we shall study the relationship of the type:
Inthis chapter
and age or price and
orheighr relationship. Special
demand or wage and price index,relation etc.
between height and weight
Such a relationship is
-3( statistical methods have been developed to
statistical relationship between the two variables from a bivariate data. discover the existence of
called
thebivariate data are quantitative, we use the term When both variables in
Correlation Analysis describe the
to
designedtofindout if the statistical relationship between the two variables exists or not. methods But in a
bivariate data having both the variables as quantitative variables, the methods of
value of one variable from the known value of the other variable form a part of Regression estimating the likely
Analysis.
Croxton and Cowden defined the Correlation as
sThe relationship of quantitative nature. The appropriate statistical tool for discovering and
mensuring the relationship and expressing it in brief formula is known as correlation'".
According to the statistician A.M. Tuttle,
uCorrelation is an analysis of the covariation between two or more variables."
12 7.2 CONCEPT OF CORRELATION
Correlation is a statistical tool which studies the relationship between two variables, and
Correlation Analysis involves various methods and techniques used for studying and measuring
orrecta

Sinterd the extent of the relationship between two variables.

items but not between the
The correlation expresses rates between the groups of
functional.
individual items. The relationship between the two variables is not
procedure by which we can determine the
In other words, correlation analysis is astatistical variables. In this chapter we shall restrict
2
more
egree of association or relationship between two oramount of correlation in a sample (of data) is
The
Ouselves to the case of two variables only. generally, denoted by r or by p.
correlation, which is,
uiasured by the sample coefficient of
13 COEFFICIENT OF CORRELATION interrelation variation and the coefficient
tendency towards the two variables are
eiattonmay bedefined as a i.e., the degree to which
gives the
of a measure of such a tendency, calledthe coefficient of correlation. It
correlationis is
interrelated measured by a coefficient which is
degree of correlation.
7.1
 COVARIANCE
given
quantitative 5
relationship
7.population in frequency. distribution. example,For DISTRIBUTION asBIVARIATE7.4 7.3.1 pvariable
results as In
ariables
the terested y) the ,y) positive a
Cov. Then set Before other thisInoccursf, 2. 1. (m) There
() 5. 4. Properties 3. 2.corresponding1. TheDefinition.
greater
or in
of involved. Frequency. The The heights scale.
2 said Ifr= and
The It
(X, the and to The Correlation iscoefficient p.
n we between variable. BivariateUnivariate are 1s
Y) variables section,
know we times, dailyheights a correlation
correlation to a negative or
covariance units study the 0,
=
of two relative be The
of variables whether In Let petrol the then measure of
) i=1 the the then independent. coefficient of
observations The other (,, Distribution. and students types
Distribution. there is
coefficient th,
correlationrelationship
-)(-) correlationand X tw o shall be f y), usedweights
between knowledgewörds, is ofmeasure is of change
variables ar e there coet f ic ie nt perfect th e
Y. called i= by distributions.
of
Let said dealing 1, a of nocloseness of
lies
be we
exists 2, Distribution a
These of or in
X analysis
the th e scooter the class correlation correlation between
the
and given
to wish
of 3, een
betwsmaller
(x+ corresponding can be a
withfrequency .... students association isa beween
some orare other
- Y by we correlated be such to owner marks the
pure positive fa
he two
) is established know bivariate m; j = involving -l tchange
(denoted
y; introduce
the a sort distributions between number beween if fitin two variable variables
relationship of
- ordered if
of 1,2, and a obtained and+
r
y)+....+.(, of th e . . , the class an d the = a variables in
values
(i.e., population.
functional the In two the and
a pair
n, tw o is tw o an d
rell at1,ive such
by the and tochange mileage in
be by in i.e .,
pairs concept measured predict ( a, discrete
a or not it x,
other also that
Cov. of enables a thewhich variables issense.
y)bivariate school. more affect e d -1SrSl. isy a a
the in covered perfect greater
variable change
)
- (X, relationship one and studentsthere
variables generaly
two of and us
population
bivariate j=l
i=1
variables. is
by variable such onl y is by and and
in
(y, Y). covariance change
variables control to distribution. pair by of a th us known
-) It means analyse i, it. is a change negative denotedi one
is the Methodn
in lsical
between called class. one
correlalion, by one
defined ofthe affects variables
of
X lavariable if r
If r=-1 orr variable
between events).
correlation. the = origin resulty
and N= the Bivariate
as bivariatechange a twhoe and are
I awe
re the
Ol The oal
t
L6.(2,9),(3, C S EP IV NG
RULE e
Hete
Sesiun. WhtHe
ior
n.
e IEP
.
I5,44),
120 Calangje
2. lje6+9+ Callumpcleul1.
. PIL ofa
j-lx6+2
=6+18 + ED the a
EKY hic alsO reefdormuula eITor tor es
 13 chances
to -(2) (3) variates 1x
=04 variables
X
more
easier
calculations
)
(X, two
10
+
17
=41-407)
have
respectively.
Cov. the
the ofobservations
+9x
12+8x3
ofobservations
the
of
),. 30;= =40.
may
and
the
makes
get
to (10, 5. 41-5x4
411. 180 240
complicated y, n
by 17, =+9+10 6 6
respectively.
which XY
and
x
products
IM ofpairs (9,
3), 74.
7x
+
=
10
6x5+ 10
pairs 37). 50 +37 25+30+40+
45+37+34
|following(50,
STEP
ytyitmt
formula
is error. Y, following +81= 153
(8,
covariance X, y,. + 34),
the in 12), 7
+6++ +
of 17 8 24
series a
below
an
expectations and 2x
n of obtained
the (7, -'ry)
x + = 10 the (40,
20+
6 6
44+43+
of
chances
(i=1
i=1 x, y; 5), + 5 84
5 3 7+ of 37).
andY of
measure'', 2 the i=1
difference
Obtain of
covariance
difference (6, + + + 15+
give
calculation E(Y), sums i=l 12
4+3+ x 30 (30,
covariance
the
sum 8), + 4 +
X
of
noW
Covariancethe
absolute Wereduces
also
which
n
E(X)
(XY)
the
(XY)
are
the (5, 5 +6 40
7), +2 +8 x3 +28
n
=-Exy 45), =y
the
means the (25,
the CalculateCalculate the
(4, 1 7+x9+
the
are
Calcformul
anis of
7u5l1atioocnauITence
for error.
Covariance an
of: isformula
The =
E
ni-1
-
E (Y), RULE WORKING
DivideCalculate
6), =Ex,++9+618
(3,
Here
x6+2
+
+ Y) n

Calculate43).
(20,
above ) =(XY) 9), 18
=6+ (x, 44),
Solution.
Here
V and Fwhere and
Out
carTy (X, E(X), II.STEP I. STEP 1.Example(2, Solution.
IV.
STEP 6 1 Cov. Example
2.
I.STEP 2y, Lxy=
=
The Cov. Cov. 6), (15,
Corte the where
E X (4,
Y.and
and
Y.
of
 are and
alys I
ew n leIvfter
ttohcorner
° Sant
devianon inare
Correlation
Positive
Perfect
corresponding the on
valueother in
the
ete
height X results,bill
deviate proportional xy
thethe handcorrelation.
any
in of The y its 7.1.
Fig.
increase
values Y. variable
Statistical(x,30)
O,-
40) a direct.
average andX and
variable left
variates
=6 electricity
Comprehensive
-6030 -25 -6060
nAlso an
the
or variables other and
lower
perfect
l.e.,in an positive the
increase corresponding the two
235. direction, on of of of
Covariance results twovalues valuesconsumption correlation.frompositive
the7.1).
=- nearly.
=-39.17 corresponding
be correlation.
Y
40 60 to the or a straight
(See runs in Fig.
-6-3 same X said the the the the is
3 5 -3 +0-60 say if
inverse on example,
between man or changes
linedirect (see
of
the
variable
is
corelation
positive
in in
decrease)
increase) approximately a
linear. a
by perfect
:Computation the a the correlatio
on
perfectly followedCorrelation
commodity correlation
25 235
in or proportional
deviate
a or For ofearnings exactly andthe
- O;-)_
(x,-F) in one of negative correlation. 7.1). be called
4 43 45 37 34 37 30 average,
of theexamnple
(or(or
Correlation linear
60-
-
3 variables value
increase
decrease are Correlation
lie be is tosaid 1.
positive
CORRELATION Fig. variable
Table Y,
say be the and to Positive is
perfectly =
y) then(seesavings points
y)= an the a inverse is it equal y)
*-30 variable
an to
the corresponding
of (x;, said
correlation then (x,
-15 -10-5 0 10 20 ( (y,-
) - 7.6.1 two
Correlation
Positive on in ischildInverse said demand Correlation
Linear
7.6.3 linear 7.6.5
Perfect
Correlation
one perfect
decrease points
line, plotted
is the orp
theresults, if or Linear correlation is corner,
Perfect
=
Y) of
othergrowingiscorrelation
or
i.e.,
and
direction,
negative straight
ploted be
the in
deviation correlationif a 1
words,
(X, OF Negative to between the is r=
value ya
variable the price said correlation.
linear
Perfectly the or hand it case
Cov. TYPES a all then Direct then
a
E the Or in a
decrease
of
in
The opposite
of
The examples
average, the about
is correlation Whenthe
variables the theright
other
this
15 20 25 30 40 S0 Here
If one
variable weight If then7.1)
Fig. other, direction,
7.6.2 If In
7.6
near 7.6.4 line, 7.6.6If upper In
of the
 75 points
are
rising
of corner
xbetween
degree
Pertect
Negative
Gorrelatlon hand Correlation
X
Correlation high correlation
left
Fig.
7.2. a upper Negative
the be
Negative
andwill 7.6.
Fig.
Fig.
7.4. fromnegative Degree
Degree
bandthere
band
narrow
then of Low
High degree
narrow
corner,
linehand thethenegative
words,
in a
in a high
the then Jallhand in X
right changes 7.3). ;fall a
which ; be
direction,
perfect n right
Correlation.
lower 3,.., Fig. n will
proportional =-1, Correlation Correlation
in
or 2, upper(see 2,..,there
the
correlation XCorrelation I, i=1,
opposite y)
in then Correla
inverse = they
to
Negative (x, L and
corner y), y), corner,
equal p to x Negative
ie.,
is Positive Positive
(;,cornerbetween(x,, Positiv
Perfect theare and
linear
perfec1ly
hand
left if
7.2).
y
r=-1, 7.3.
Fig.
Degree
points points
hand
Degree 7.4).
Degree handcorrelation Degree
or upper
Inverse Fig. variates cor case
el ation.
yand
xbetween this
In plotted plotted
right
Fig.
fronM(see otherX High left Low
767 is borner
a relation High
theJower
High
positive If
the lower
(see
CoI
It uns In wo Y
0
If
76.8 Anw 7.6.9 the y
to and
 point eachvariables.The two
Draw Correlation
Curvilinear 7.6.12
independent
On axes, between
between arevariables Correlation
paper, 7.6.11 because
No and T.0.10
whose of Solution.scatter Example Notes. said
the a thin
e If
paper X,
abscissa data, graphical Y productionthe the to then the If negative The
the
draw by diagrams 3. 2. 1.
Scatter correlation be figures Low
there way
If More
Whatcoefficient. tplotted
wo independent. (see and there plotted points
the 10 20 30 the
40 50 60 correlation Degree
on two of 70
wil
the representation Diagram. when isdots the Y-Axis
variablespoints Fi g . 7.5
axes points a 10 variables points is are
correspond
diagram scatter are plotted 7.7 ).
the widely 20 of correlation no scattered andPositive
at is rice is Here
rightcalled coefficient points a lie beetween 7.6
It
diagram? Rain 30 bivariate
and said e
r= arScattered show or
is is scattered,
a of said much Negative
Xangles athe a Fall 40 rainfall to 0.
and pair scattergraphical lie in be The the low a
of 50 to
be away
the ofone data Giv e onCentimetres di st rib
inution between all variatesdegree
correlation then a Fi7.g8. Correlation
y
axis in 60 from
inatevalues diagram.
curvicms.linear variables over
th e the straight
a 70
80
90 in staatistically the the he Xof
responding squaremethod procedure x tpositive
X, correlation (asee more and line. and
is r line, vil age or twOgraph
correspond Y of y y
Y.
whichrepresenting region =
+
Fig. less
The showing the is
and lbetweenof 7.a8). form Comprehensive
data in to drawing greater a
turn X r=-1. curvilinear For Curve, CorrelationNo
and correlation X-Axis + example
ted will the is Fig.
7.7.
the a the correlation. then Melhod
Statistical
variables
Scatter the the
ouend the
in magnitude correlation
correlation
this variablebets ween aiayu
w 0 po. is
 Positive,
3. ()2. 1. 6. 5. 2. 1. Hil
Negative,
Perfect () scattercorrelation,
wat Define
correlationDefine ()
Distinguish ()
correlation Write a (c) (a) rig ht, (vi) insurance (i) (ida
)Thy
e ()
drawing While a ) (v) () State Define
Positive Linear, PerfectPerfectUnemployment
then ProductionDemand Age The The the ofrelatprio ernts hipgivethe a
gative diagram, is scatter scatter
short the s
corelation between between
(ii) negative it temperature. ofsale ofamount colour of
anature
non-linear andclearly positive is applicants
for term
Non-linear note
tion. the diagram diagram Negative inferredscatter of woollenthe crops.correlation.
goods of
ofExplain
correlation
following
rainfall
saree
the Fi4.g9.
? two between scatter on
two correlation,
correction, index pig
correlation,
positive,
(vi) (ii)No presence
perfect
Explainof
correlation and for
correlated and variables
(üi) diagram
that if iron time.and garments and variables.
between
depict correlation Durgapur.
and and their life and
draw there the
the lineardlagram. tne soot yieldthe oihes
x purchasing prices intelligence
implications and the four is... all and the
and and contents two
yariables.IODeisio
no
scatter y.
points under and EXERCISE-7.1
concept
correlation. different
ANSWERS positive perfect Draw of
the (positive, X
diagrams appearpower in
normal of We
positive
negative.
(vi) (iü) of (i) correction
(iv) sketches d) (b)
of (No premium the of
and can
positive types linear to of girl positive
x Simple negative
perfect for and form the interpret
of of who
and y common insurance. and
and perfect scatter uncorrelated positive a wears
ive
negatnegative Non-linear scatter straight or
negative
no the
positive diagram1s it.
correlation) 4.10.
Fig.
diagramcorrelation
to line man way
correlation.
correlation. correlation
correlation, correlation going sctter
positive(i) for
different show downward
with of
Show cater
perfect followingthe examples.
by from
degrees
means negative left 11
to he
of of
 7.8 Comprehensive Statistical Methe,

77
METHODS TOof STUDY
The CORRELATIONvariates r and y can be measured he
he
intensity correlation between IWo
methods.
()Scatter or Dot Diagram MethodCorreationor Covariance Method
fol owing
(i) Karl Perason's Coefficient of
(ii) Spearman's Rank Correlation Method
(iv) Two-way Frequency Table Method
() Concurrent Deviation Method METHOD
DDIAGRAM correlationbetweentha
two
yariablVaesriablesy
the
7.8 SCATTER DIAGRAM OR DOT methodofshowing
the
Let
distribution. choOSing a
of
the valuessuitable x
Scatter diagram is a graphical bivariate
a coordinate plane by consideration scale, so andyi
. , nbe
Then th at
and y. Let (r,y) i =1, 2, 3,and y-axisin a
be plotted along the I-axis
data of boththe
(series) under
variates
point
measures the range of the therecorrespondsa called a scatter
coordinate
or a dot
inthe plane.
diagram or a
cordot responding
to every ordered pair (x, y) t pointsso obrainedis
The diagram of dotsmay or degree andthe
type of Scattediragrdagram
correlation. From
the two
relationship between shown
am,
indicateboth the
The scatter diagram thoughroughideaabout scattered diagrams
as variables. The
by
we can form a fairly
good,
correlation are
depicted by means of figures1.1.
diferent types of Diagramooee
7.6. and 7.7.
7.2, 7.3, 7.4, 7.5, Disadvantages of Scatter
Advantages and
7.8.1 form a rough idea of the
Advantages comprehensible and enables us
y
to
. hnto nature of
readily x and
1. It is two variables
relationship between the observations.
affeced by extreme
2. It is not size ofextreme items.
influenced by the
3. It is not
Disadvantages
number of observations is very large.
not a suitable method if the
1. It is methol
to obtain an approximate estimating line or line of best fit by this
2. It enables us exact magnitude cannot be known
measure of correlation where the
3. It is only a rough
COEFFICIENT OF CORRELATION
7.9 KARL PEARSON'S
following mathematical
Karl Pearson (1857 - 1936) was a great statistician. He gave the variables. If X
formula for measuring the magitude of linear correlation coefficient between two
and Yare twovariables, then the correlation coefficient p (X, Y) between them is given by:
FIRST FORM
Cov. (X, Y)
p (X, ) =
SECOND FORM Var. (X )xVar. (Y )

X(;-I) (;-)
i=1
p X, ) =
n

where , y are means of Xand Y i=1

(i=1
p(X, ) is also series respectively.
written as r or r or simply p.
Corelation
Analysis

THIR
F DORI

We
knowthat 79
n Z(y,- y)=Eyy,)
E-)(y}- y) = Xr,y, -(Ex (Ey)
these results in the form
Substituting
and
(II) of p (X, Y) we
get
p (X, Y) = Lxy, -(x)(2y)

n
(3)
saves a lot of computational
This
formula
androundingoff
labour. It reduces the error due
to
Example 4. Calculate
the coefficient of
correlation
(,2),(2, 4),(3, 8), (4, 7), (5, 10), (6, 5), (7, 14), (E,for16),
the (9, 2), (10,data:
following
20).
computation
= 10,
Solution. Here n
1+2 + 3 +4 + 5+6+7+ 8+9+ 10 = 55
-2+4 + 8 +7 t 10 + 5 + 14 + 16 +21-20 =88
-1+4 +9 + 16 + 25 + 36 + 49 + 64 + 81 +10 =
385 = 1114
= 4 + l6 + 64 +- 49 + 100 + 25 + 196 + 256 + 4 + 400
Sx, y,= 2 + 8 +24 + 28 + 50 +-30 + 98 + 128 + 18 +200 =586.

Now
(Ex) (2y,)
586 - 55x88
10
=
p (X, ) (2 y,)' 55x 55 88x88
n 10 1114 10

586 484 102 102 102

= 0.61 nearly.
J385302.5) (1 114-7745) J(82.5) (3395) 2808.75 167.35
The above example shows that the computation is not simple. We give below another method
to make the computation simpler.
7.9.1 Direct Method

If Xand Yare two variates having their means + and ùrespectively. then
E (dr x dy)
p (X, Y) =

where dr =x, - I, dy =y, - ù, dr² =(,- and dy² =(y,- ùP.

It can also be written as :
E (dr xdy)
no, Xo, and
where n is the are standard deviation of X
series,
of observations in Kor Y , 0,
numnber
Yseries respectively.
 RULE where
WORKING 7.10
from
Now =r 71 70 69 68 67 66 65 Solution. cm):yn) ofbrothers
Heights
(isisters Heights
(in ofthe StepIV.
Step StepV. SteIII.p StepI . Step I.The The
Example
cm)following is
VI. coefticient the following
=0
dx2 Let (): number
(x-68) dx
= .
y')Edr dy
dxL
-1 -2 -3
the data: Find deviations
The where colu mnfrom
Take Denote
Apply Square Calculatei
3 2
Mul tiply of formula
heights the above n the these and ù the correlone ation of
following the wrihecaded
E Table coefficientmethod
is
ofthe deviations observations. can
dx' 68)(x- dr= of 67 65 x
respective
deviations te andseries also
4 1 0 1 4 9 andnumber by
x32V28 = Computation : y= brothers it ù by is be
20
28
illustrated
is y dr=X-
formula dr ina the ofthcaleofcaulnd axted
65+66+67
67+68+66 68 66 of series. of
/Edr'Edy²x and deduced
be correlation observations dxdy E and column
69 72 72 69 66 68 67 denoted to writobservations
e i. and xthe
7.5 5 of 66
69 by calculate and dy
themheaded
Take otyher thby
e
Edxdy from
Correlation ++68+ by 67
x).67 =
69 between the or=
r the series
folserieslowing above
2 following in write (Edx)
(Edy)
+72 under ns in
dy
= 0 3 3 0
0
-3 -1 -2
69) (y
- dy
=
Coefficient.o
+72+70+71
+69

69and x
that
68
the
example.
heights
series,

x
or or
r deviaby
it
yX0,ng, dxdy E r,under the =
the columns y from

ti
dy
o respectively.
x-series by
y.
steps: xZdy'(2dyj? formula
Comprehens
of 72 69 the y-series -tthhee of
69)(y
- dy' = =68,and sisters G, coefficient
ù.
2 69. - and o,
columnheaded
observations and X
dy' 9 4 of
= by brothers
write
32 72 70 are headed
of by
then y, the correlation d in it
and byand under
dy 0 3 6 drdy 69 71 sisters
standard
dxdy. d.
z
 236
=druy
51 -). dxdy 104 20 0
15
=ty-y
dxdy 12 2
=
ddy
=
0.95.erin
o
21 35.
followino
dnn: (y
=
dy
25|
(Y-
= 169 25 0 6H
|Edy'=
E358
16 dy 60 and d
8 (Y
-12 16 =
16 0 =0dy' 17 33
(x-x) (x-7)
0.9=
86.
dy
4
the 0
7
dy
=
25)(Y- -5 =dy
= between = 1-3 0 8 10
6
13
-12;
dx (Y-
dy
=
12)
-3-4 -2 0 -1 1 2 4 3 dy
L 13 25
dr
Let
E
236
239.33
60 coefficient
J60x60
5757 25.
ù2Y
108
5 9
10 12 11 13 14 16
= 33 35
Y 9 8 15 125
5
y 2 20 25 358
V160x
n
correlation
4
12
60 9
20 =
Y 13)2
(X- 160 236
-=5; dx² (X
-5)2
= 16 9 4 4 9 16 = JEdr'x
dy =
dx?
==13, dr²=
10 dx 64 16 0 16 64
45 9
0 Edx
L
dy Calculate
the
3
EX 65 5 dy²
12
Analeatoysis 2 8 FHere
solution.=dx 5)
(X-
-4 -3 -2 1 0 1 2 3 4
=
dx
5 X
Here dx
=
13)
(X
-8 -4 0 4 8
0=
dr
dx
2dy x
dr'
E r=
Example
7. 2
9 Solution. V
1 2 3 6
Now X Y 17
4 8 9 5 9 13 21
ti
eunde
t
aded
btu
fy
correlai
the
standag Sand
sisten 71
üsical
o1
nyn
s 69
nen
 the 38
dxdy 6 0 1 12 12 2
46 =0.949. below: 4
= 10.
6 2 9
dxdy given7
dxdy 1 48.497
Comprehensive
Statistical
hesthet
3. Z
data 2 5 respectively
dy? 49
7 84
=8
46
9 1 1 16 16 1
4
data:13 12 dy? 25 9 4 9 36 dy? V2352
V28x84the 13
E 46 for 8 Coefficent
following 9).and Coefficient
Computation
of 28|
correlation
10 dx?
(Y- = 6 9 9 32
11 9 9 dy? 46 and 4 4 1 1 9 4
1
=
4 4
Correlation
the (Y-ø)
Table:
Correlation E 5
from /zdr?xEdy²
of 4
coefficient series
8 be
5y-
10 Y 0
coefficient9 = = 57 1
dy Y- -1-3 -6 dy =9. Edxdy 8 -2 3 -1 -1 4 -4 1
5 2 3 y of = 1
= E dy
and Computation
dy
10)
Pearson's12 x
correlation
8 for -5
(X- r= mean
correlation: 7 x -2 -1 -3 2 2
(X-X)=X 0
Karl
3 = 2 1 3
the 9 X-
-1 -2
= X:
Variable
Independent Table:dx
assumed
-3 dx
CalculateY:
Variable
Dependent
8 2 0 1 3
Calculate = Z
dx
= 10,
= of the
12 14 dx Coefficient Y 7 12 8 8 10 13 5 10
Let 63 70 9. Let
8. 14 8 6 9 11 12 3 7
lè Solution.
Y Here
X= Example Solution.
X: Y:
.. Total
X 12 9 8 10 11 13 7
70 X 3 5 4 6 8 2 7
 743 or of y-series under
Take
x series. Then
difficult
formula for
mean 4. it
foldata.lowing y-series.
of y dr=I- write
becomes
following mean for
assumed 9.
mean and the 3
for
assumed i.e, -b for mean
-0.96. y as assumedy dy², yand 2 4 assumed
andthe (Zdyf series a, =
fromdy
39.24
37.75 when apply (Ede)
(Edy) the i.e., and xbetween
(Ldey
(dy) x as (Edy)
(2dr)
of isb of one) series
b, the
convenientwe
calculation one) from d dxdy.
columns
(1
8 cases
-b,
y
middle
middle xin series column correlation
example.
be
2x1 (Zdr' dy
=
observations -E(dxdy)
4and
these (dxdy) (preferably under x-series
dxdy 8 not the r-series, y 0
is andIn
the in the
dy. it under
(preferablyobservations
following
38
r integers. E and write ofcoefficient
calculating
big for
of
are y.
meanor b the dx and it 1 mean
y not termof by write formula
the
Apply the
and Tsy x
observations
assumed
of a the
deviations headed
dy2 by the assumed
for
x are
term
any
of and
and xy
illustrated
-2
Calculate
hmethod
4
dxdy
MetShort
19.od 2-cutseries
y
deviations
anyand columns
dx2
table:
AnaColryeslaitson direct
the or the
RULE
WORKING
series
Take Take EindFind -3
Let
be
2 following
oofnumber
means is is
a=I-a,
X
of 10.
method 9
Theterms meanAsSumed
above STEPL.II.STEP IL.
STEPIV.
STEP
EXample Solution.
V.STEP
Now the the The y:
the
(i) (ii) dr is
wheren have
and We
 for drdy 48 16 30
Comprehensive
Methe,
dxdyy following 6 0 0
dxdy -25 9 3 mean
8
EV 20.1 0 the
Statistical Z
V84-0
x assumed apply 484
84 the
dy' 300 300
+7 perfe
(y-
4 =
dy² for 14464 36 16 36 66
CoefCorficreentlation
25 9 16 9 0 25
Z
0-0
256 yand
16 the
be
14)2. Coefficent
0 4
=
shall 7
106+
7 we and
4) 0 between
x 14 -
(y so positiv
=dy -3 -4 -3 =dy 14 anddy²= dy'
5 (Edy)' correl atio n Correlation 16 4 0 4 9 25 59 means.
uncorrelate d.
5
E x
of
=
(Edy)
(Zdr) N 14),
values assumed dy)²
(E =1, is
y
5
10 -
-12
22 808 808 and
the(y
coefficient for =dy ofComputdyation -8-6 -4
of 0 9 - dy) x
Computation
9 4 1 4 0 2 = between
dy from (E
56
(Edr)? re of
-(dxdy)
E are
mean(r-5)2,
4 8
E dr)
deviations 808
V652864
assumed mean(E
(r2) = =3
dr
E dr)'
(2 correlatio
dx? y0’rand -2 -10 2
dr 25 16 9 4 =
d? Table:dx 4 -drdy
assumed
Table: the - 3 5 V404
x1616
the=X-5,
L Calculate
2
the
n
E 808
14 be taken
for of
coeficient
= X-2
dx -5 -4 -3 -2 -1 =- = 15 =7
n correlation
dx r 11. 2 Let :dx 2 8 14 16 20 have
Let
Example Solution.
2 we the
y. Sinceof Thus
Now of =7 formula
-3 -2 -1 0 2 values 4 5 7
10
7.14 X
3 1
n
 7.15 y
cocfficient (-23), xbetween 122. for
nd
dxdy
48
=
respectively
of 24 0 4 3 0 10 9 24 dxdy =
m
values27 33. ds -
2
correlationmeans 122.
of
Then 70 respective
138
pairs 26 30 respectively. dy' 16 70-0 136,
=
F)
following
28 25
Coetficent
4 0 4 9 36 =
dy
2 [480]xdy'-2d)vI240 ofcoefficient are
their
means
(-
18.=y +)
124
27 dy?
from (r- =0.89.
the Correlation 36 16 4 9 9
25 9
16 dr'=
18. seriesrespective 136.996
for 20 27 and the E
correlation 23 Z
=0.515.
computeisyof andy information
: 122
be 0 that their 138,
y = (y-
dy 27)
-1 = n 25=I 138
V136x 0.89.
212626 and of 4 0 -2 0 -2 3 dy (E
dx)
dy) 93.166
/8680 and x
Computation- E x,E
data of from =
following-y? 122
25 deviations
of x 48 iscorrelation
for table. following deviation
15.
E(
25 means
followingdx 23)
(r- 0 (
d)? isseries
=
= 0-6
4 -2 B3
-3 5 = -dxdy
E 48 values
the assumed
3 4 dx (2 the F)
Table: -
E 70
24x
= the x of products
of series.
squares - of
Calculate
19 27 of of given
15
n=
136, (yE(r-) coefficien
AnalCoysirselation the
the have n=8
dr
1/2 48 0.515.
r= From Mean pairs
Arithmetic yand are
Example Solu0-tio2n.7).
12. 17 23 Let We 23 27 25 26 27 25 30 33 13. the the of
Number
We
Example ofSumofSumseries Solution.
=
(r-)
the
Hence
n=8 Now Hence E
Now
r=
h= 17 19 21 26 20 28 26 27 y.and
() (ii) (ii)X (iv)
 3 2 4
0012dxdy
=
6 35 18 -2
Englis 17. dxdy ots 21 0
28 10
20 respectively.
48 -
in
marks 21 24
128
4- 10
128
Statistical J90x6An the 25 23 18)?
(y- 1 25 36 =
between =
dy² 1 dy² 4x210
120 18 36 16 4 9 0
Comprehensive
X=X 12 17
and Coefficent E
n.items
L.e.s =0.5 120×120
90x coefficient
0.25
18 122 35010 =0.575
of
E=
90r=0.5,
Ey=
number mean, (Y-)
E(X-X) 11 18 be
y
and dy
=
18)
(y-
Correlation 6 4
-2 -3 0
1 7 - 1=2
5
6
dy N348.4N12
18.665x
210.83
121.2
,8,arithmetic =90. =
64n 10.= correlati22on 18 X
E
ofvalues
the 0.64x90x
25
X)' 120
120x
=120,
6,
find mean.
-e=(X
120x120
64n
90x
25 15
the
Computation12 22 16 15 18
of 18 23 24 17 N
Edy)/
N
below, from
arithmetic
Pearson 13
's 1622 for 121.2
deviations means
given 18)2
(r- 350 (Edx
0.25
= = 25 =
dx? =
(Edr)?
n Karlstudents. Table: 64 49 25 49 16 49 36 49 9 4 dy' --
Edxdy N
data from assumed
are
Calculate10 12 2 Ldr'. 122-0.8
the deviations
y
and 10
From :
x
by English
Marks
in
Hindi:
Let
the
= 18x--8
dx oso-5 -7-6 4
=
14.
Here 15.obtained 7 7 4 7 3 2 dx r=
are Z
le y
Solution. Examplein Solution.
and Agan1 Marks Here
Now Also Hindi
x 10 25 13 25 22 11 12 25 21 20
where Or
and
Corelaton
Anal

5 8
7
3 4 7
8 7 6
mean for the
solution.
the assumed
Let
Table: Computation
val
of
ues of X
and Y 6 2
series be 5and 6
dX=
(X-5) (X- 5)
Correlation Coefleento respectively
dY
-2
-1
4
1
5
3
(Y-6) (Y-6
dy
dXAY
0
0 4 -3
9 -2
3 7
4 1
2 8
8 2
4 16 7
1 4
1 1
1 6 1
-3
16 3 0
2 -4 2 9
-4 -9
1
E dX=
0 2 dX = 60 16
Z dY =-3 16
LdXdY -(ZdX >dY) /N Z dy= 45
dXAY =23
(Zdx)? (2dr)?
.N

23-0 23 23
J60 0) [45- (-3)° /9] V60x44 51.38 =0.448.
Example 17.. Acomputer operator while
andyfor25 pairs of observations obtained the calculating
the coefficient between two variates
following constants: x
n= 25, x = 125, E = 650, Ey = 100, Ey = 460, Zry = 508.
It was, hOwever later discovered at the time of checking that he had copied down two pairs
6) while the correct pairs were (8, 12) and (6, 8). Obtain the correct value of the
as(6,14) and (8,
correlation coefficient.
Solution. Here Corrected I=IncorTect x- (6+ 8) + (8 +6) =125.
Corrected Ey = Incorect Zy - (14 + 6) + (12 +8) =100asate
CorrectedEr= 650 (6 + 82) + (82 + 6) =650
Corrected Ey' = 460- (14' + 6) + (122 + 8) =436
Corrected E xy = 508 (84 + 48)+ (96 + 48) =520.
Correct value of correlation coefficient is :
125 xi00
520
25
(125)² 436-( l00)²
650
25 25

520- 500 20 =
V25x36 3
V(650 625) (436- 400)
 f p a i t

Cornprehensive Statlistical o f d

Xample
m e a n

were observed 18. In two

sets of observations each,
d e v i a t i o n

o f t h

s
p e
r o d u

g
o

u
cft

a
t

r e s

Xand y with 50
variables
found
R10,
Coeffionecientvalueof correl = 6, S.D ofx = 2
S.D.ofx =3, ù Ilowever on subsequent folowing d t

o
h

f
e

Ey

With
that ation Xand y is 0.3. 0) wereinacCHrate
and
of X(= 10)betweenOne value of Y( value of correlation hence ycoef iecriijecnattion
in
remaining 49 pairs of values,andhow(r-) is he
orieial weafedeedteA n n e c t i vA
e .M

(y- y)/n
Solution. know that: r=
We
o,0,
rrespec

3roo, =X(r-)(y-) = 50 x 0.3x 3 x 2 =

90
y2x+iy Ly= ny, Lz y=
) 90
’ y- (n y))-- F(n F) +n ry=
X 10x6
90+50x
y= 90+ ni y=:xy = Wt
ihtfe
ollo

’ y = 3090. 3030
t fd
e e n s to
t yf

Corrected y =3=3090 10x6=

490
2I= n x- 10= 50×10-10= Ciies

294
2y= n y-6=50x6-6= A

B
Againo= C

+ 100) =545.0 D
’Er=n (o' +F) =50(9
=n(o' + p) 50 (4 + 36) = 2000
Similarly, Ey²
= 5350
Corrected x = 5450 (10)?
1964
Corrected E y² = 2000 - (6)²= coefficient when N= 49 is: Hint
correlation 49 x3030 490x 294
Corrected value of
NExy-(Ex) (2 y)

N E-(EnP|[N(2y')-Ey
49xS350 (490°|49x1964-(099 Wean

90 909
49 (3030- 2940) =0.3
V90000 30 Det
J450x 200
J49 (5350 4900) 49 (1964 -1764) D
unaffected.t
Thus the original value of correlation coefficient remains
EXERCISE-7.2
brothers from the gva
1. Calculate the correlation coefficient between the height of sisters and height of
data.
Height of sisters 64 65 66 67 6 8 i69 70
(in cm) :
Height of brothers 66 67 65 68 70 68 12.
(in cm) :
[Hint. Here =67, ù=68, Ex = 0, Ey = 0, E = 28, Ly=34. Also Zxry =25.
Use r=
Exy 25
==0.81.]
rxy V28x34
 719
from between
deviations
method F
300 13.
No.
of
deaths
deviations
means
122. are Pearson's 300 1440560 840 1224 312
correlation
dits y r.
Y
eries
follewing 18 303138
and
E
600 17 3200 ofvalue =0.)
(x
respective items by
correlation
=
Ery the (80) 50
the of n=10.] Population
in
'000 ofNo. Population
deaths y'=60, find
from
their
number
L(I-7)(y-9)
Exy D 700 20 120, 8075x S0 (75)²
X
series
series
15 301136 from
of
6coefficient
30 90 40 42 72 24 x'=175000, =Exy 120
E
y
and
25
series
the
Find 6.25 =
peath
rate C
400 14 ]=0.9875.140,, 130
betweenx =0.89] 5.2=90. x =
(60) (90/n) the rate. Ey?
mcan
y
and
Vi36x138 calculate
:data then V1,75,000x60 150,
fromof
x and n'
064=.
’
death metres Population. 500 16
following
B
ù==15, ,Er=
(Ey)
x(Er) N
2.5 Also
r= cities, 3200
fobservation 122
2y= the Sq. 150 180 10060 120 80
Area
tdeviations
On and
population 90 6 2 (Z)?
JErxEy2 90 (Ey)? in 80,
of
AnalCoysicoefficient
rselaton deviStasquares the
of
tindaronprdoduct
of
60,
=Er
A.respective
2xy=Mtheir
). in
data Area
following Densitythe
=
fromA
200 10 =450, /E?xEy? =
y
E Ery
Calculatpairse meanArithmetic
the
of
ofmber

Ery
0.8,
the the r=[Hint r= o(Hint.
of Cities
density Using rfind
:Death
rate
2700
6

Exy
75,
x=
E Hint
:r=
IfN=50,
the to =:Here
Num ofSumofSum Given With
A B C D F
[Hint.are :Cities:Density r=
the We
2. 4. 5,
3.
 14. 13. 12. 11. Calculate10. 9. 8. 7.
the FindObtain observations as Y: X:
It IfN=Output correlated
following Input From Income:
[Hint.Expenditure: :
Month result. Price Price Demand :
40 35 30 25 20 15following was,computer A N=
Calculate Find ascertain From :Price price
and and and and and and(years)
groupAge the
the 12
7 8 10, theshare share
less less less less less less Karl however, 50, of of the
Oil Eu the and
correct 10 constants, Electricity E 100
Eu 2
corelation
Pearson's fol 18 the correlation
following
140,X= data by 70
than than than than than thaninformation : (B) (A) following
while X= following correlation
coefficient = = using
45 40 35 30 25 20 value while 75, £ 84 14 as
later Vn (X- (X-
Rs.Rs.: working
calculating
N= E : 18 110 E Jan. Pearson'
coefficient of the Y= data, 36 46
:
correlation discovered 30, E- 10)'
n = 10)coefficient
Y= coefficicnt information
corrected 80, 1.3.95 6.9 E 150, 78 16
EX= examine uy- =2 292 160 mean
the E 17 115 (Eu) 180; 40, ,
of X= (Eu) Ev for
correlation
coefficient values at120,
correlating (X- from 40 54Feb. Detween coeticient
the 130, 8.2 whether
280 164 Telating 17
70demand.
PeopleNo.of E bet116ween n(2v) Sy==E(Y- 10) the
time X= E 16 E the ot
300 400 360 340 270 200 were r?-()'} = March
600,
coefficient Y= in put the (Y- 15)= 180, 44 56
betweenbetween of 6.5 7.8
income coreiation thto
15)2 = 260 172
e 75 18
checking Z 140, of 16120 following 0,
E(Y-
108 8 Y= Z oil
and Stock
the X XY= 215; 13 now
12 90,
between and y(10x180- 15)2=215, 54 56 66 19
age and 1.3 4.8 pai rs
Aprexpendi
il ture
that 2 120, Z 234 2
Exchange
18shares
and Y. Y= output 16120 uv=
it the find of 10x60-40x 0
the val u es 2 Z 42 58 May o A 67 20
ees162 PlayersNo.of had 250, 2125 (X - (X - and
Statisticalve
variables the 5.5 9.6 of (40)}{10x a 266 166
playing 15 Quotations Comprehensi
120 180 180 170 150 copied E electricity 10) 10) wage b
XY= values are 62 21
60(.Y- (Y -
habit 215-(02 -
58 60June earner 254 correl 170 ated for
down 356 andthe X of 3.5 8.0 13 130 15) 15) and
can 22
of r = = cormirrient peice
their in
people two obtained Y be f 54. 62July 118
2. 2 . 11 said 10. 135 =0.915
Dairs om
trom to h
be
isand 0.939. r= 4.
0.603. 15. r=0.915. 9.
0.957. 16. r=-0.915. 10. r=0. .
0.696. n= 11. r=-0.954. 6. 081. 1.
0.05. r= 13.
r=-0.96 7. 0.89. 2.
r=0.769. 8.
10. n= 3.
r=0.9875 4.
ANSWERS J8x1530-
(48)"}
3468-(108)2 x{8 x
) 2160. Jáuty=
48x108 8x2160
=0.957.] Hint.
làke
=I dr
dy=y 69, 125
112, - 137 Calculate
=dx then 78
3468 1530,E = d 108,
Z =dy 48, 156
112 Ependiture:
coeficient
89 Ad:
108. 123 136 107 97 (Rs Seles
69 corTelation of lakh):
79 59 K
61 68 between Find
and X 67 the
Y Coation
68
71. 68 65 65
69 72 72 66 between Anaysis
22 70 68 67 67 sales
69 andd
following
data: texpenditure
he firom advertising
721