International Journal of Social Economics

Empirical study in finite correlation coefficient in two phase estimation

M. Khoshnevisan, F. Kaymarm, H.P. Singh, R. Singh, F. Smarandache,
Article information:
To cite this document:
M. Khoshnevisan, F. Kaymarm, H.P. Singh, R. Singh, F. Smarandache, (2004) "Empirical study in finite
correlation coefficient in two phase estimation", International Journal of Social Economics, Vol. 31 Issue:
10, pp.890-902, https://doi.org/10.1108/03068290410555381
Permanent link to this document:
Downloaded by Professor Florentin Smarandache At 08:14 27 April 2018 (PT)

https://doi.org/10.1108/03068290410555381
Downloaded on: 27 April 2018, At: 08:14 (PT)
References: this document contains references to 18 other documents.
To copy this document: permissions@emeraldinsight.com
The fulltext of this document has been downloaded 1046 times since 2006*
Users who downloaded this article also downloaded:
(2011),"A new space-time correlation coefficient and its comparison with Moran's Index
on evaluation", Management Decision, Vol. 49 Iss 9 pp. 1426-1443 <a href="https://
doi.org/10.1108/00251741111173925">https://doi.org/10.1108/00251741111173925</a>
(1973),"A WEIGHTED RANK CORRELATION COEFFICIENT FOR THE COMPARISON OF RELEVANCE
JUDGEMENTS", Journal of Documentation, Vol. 29 Iss 4 pp. 380-389 <a href="https://doi.org/10.1108/
eb026564">https://doi.org/10.1108/eb026564</a>

Access to this document was granted through an Emerald subscription provided by

For Authors
If you would like to write for this, or any other Emerald publication, then please use our Emerald for
Authors service information about how to choose which publication to write for and submission guidelines
are available for all. Please visit www.emeraldinsight.com/authors for more information.
About Emerald www.emeraldinsight.com
Emerald is a global publisher linking research and practice to the benefit of society. The company
manages a portfolio of more than 290 journals and over 2,350 books and book series volumes, as well as
providing an extensive range of online products and additional customer resources and services.
Emerald is both COUNTER 4 and TRANSFER compliant. The organization is a partner of the Committee
on Publication Ethics (COPE) and also works with Portico and the LOCKSS initiative for digital archive
preservation.

*Related content and download information correct at time of download.

 The Emerald Research Register for this journal is available at The current issue and full text archive of this journal is available at
www.emeraldinsight.com/researchregister www.emeraldinsight.com/0306-8293.htm

IJSE
31,10 Empirical study in finite
correlation coefficient in two
phase estimation
890
M. Khoshnevisan
School of Accounting and Finance, Griffith University, Australia
F. Kaymarm
Department of Mechanical Engineering, Massachusetts Institute of Technology,
USA; currently at Sharif University, Tehran, Iran
Downloaded by Professor Florentin Smarandache At 08:14 27 April 2018 (PT)

H.P. Singh and R. Singh

Department of Mathematics and Statistics, Vikram University, India
F. Smarandache
Department of Mathematics, University of New Mexico, Gallup, USA
Keywords Correlation analysis, Variance, Estimation
Abstract This paper proposes a class of estimators for population correlation coefficient when
information about the population mean and population variance of one of the variables is not
available but information about these parameters of another variable (auxiliary) is available, in two
phase sampling and analyzes its properties. Optimum estimator in the class is identified with its
variance formula. The estimators of the class involve unknown constants whose optimum values
depend on unknown population parameters. In earlier research it has been shown that when these
population parameters are replaced by their consistent estimates the resulting class of estimators
has the same asymptotic variance as that of optimum estimator. An empirical study is carried out
to demonstrate the performance of the constructed estimators.

1. Introduction
Consider a finite population U ¼ {1; 2; . . .; i; . . .; N }: Let y and x be the study and
auxiliary variables taking values yi and xi, respectively, for the ith unit. The correlation
coefficient between y and x is defined by
ryx ¼ S yx =ðS y S x Þ ð1:1Þ
where
X
N
S yx ¼ ðN 2 1Þ21  i 2 XÞ;
ð yi 2 YÞðx 
i¼1

X
N
S 2x ¼ ðN 2 1Þ21  2;
ðxi 2 XÞ
International Journal of Social i¼1
Economics
Vol. 31 No. 10, 2004
pp. 890-902 X
N
q Emerald Group Publishing Limited S 2y ¼ ðN 2 1Þ21  2;
ð yi 2 YÞ
0306-8293
DOI 10.1108/03068290410555381 i¼1
 X
N Finite correlation
X ¼ N 21 xi ;
i¼1
coefficient
X
N
Y ¼ N 21 yi :
i¼1
891
Based on a simple random sample of size n drawn without replacement, (xi, yi),
i ¼ 1; 2; . . .; n; the usual estimator of ryx is the corresponding sample correlation
coefficient:
r ¼ syx =ðsx sy Þ ð1:2Þ
Downloaded by Professor Florentin Smarandache At 08:14 27 April 2018 (PT)

where
X
n
syx ¼ ðn 2 1Þ21 ð yi 2 y Þðxi 2 x Þ;
i¼1

X
n
s2x ¼ ðn 2 1Þ21 ðxi 2 x Þ2
i¼1

X
n
s2y ¼ ðn 2 1Þ21 ð yi 2 y Þ2 ;
i¼1

X
n
y ¼ n 21 yi ;
i¼1

X
n
x ¼ n 21 xi :
i¼1

The problem of estimating ryx has been earlier taken up by various authors including
Gupta and Singh (1989), Gupta et al. (1978, 1979), Koop (1970), Rana (1989), Singh et al.
(1996) and Wakimoto (1971), in different situations. Srivastava and Jhajj (1986) have
further considered the problem of estimating ryx in the situations where the
information on auxiliary variable x for all units in the population is available. In such
situations, they have suggested a class of estimators for ryx which utilizes the known
values of the population mean X and the population variance S 2x of the auxiliary
variable x.
In this paper, using two-phase sampling mechanism, a class of estimators for ryx in
the presence of the available knowledge (Z and S 2z ) on second auxiliary variable z
is considered, when the population mean X and population variance S 2x of the main
auxiliary variable x are not known.

2. The suggested class of estimators

In many situations of practical importance, it may happen that no information is
available on the population mean X and population variance S 2x ; we seek to estimate
 IJSE the population correlation coefficient ryx from a sample “s” obtained through a
31,10 two-phase selection. Allowing simple random sampling without replacement scheme
in each phase, the two-phase sampling scheme will be as follows:
(1) The first phase sample s* ðs* , U Þ of fixed size n1, is drawn to observe only x
in order to furnish a good estimates of X and S 2x :
(2) Given s*, the second-phase sample s ðs , s* Þ of fixed size n is drawn to observe
892 y only.
Let
X
x ¼ ð1=nÞ xi ;
i[s
Downloaded by Professor Florentin Smarandache At 08:14 27 April 2018 (PT)

X
y ¼ ð1=nÞ yi ;
i[s

X
x * ¼ ð1=n1 Þ xi ;
i[s*

X
s2x ¼ ðn 2 1Þ21 ðxi 2 x Þ2 ;
i[s

X
s*x2 ¼ ðn1 2 1Þ21 ðxi 2 x * Þ2 :
i[s*

We write u ¼ x =x* ; v ¼ s2x =s*x2 : Whatever be the sample chosen let (u, v) assume
values in a bounded closed convex subset, R, of the two-dimensional real space
containing the point (1, 1). Let h (u, v) be a function of u and v such that:
hð1; 1Þ ¼ 1 ð2:1Þ
and such that it satisfies the following conditions:
(1) The function h(u, v) is continuous and bounded in R.
(2) The first and second partial derivatives of h(u, v) exist and are continuous and
bounded in R.
Now one may consider the class of estimators of ryx, defined by:
r^hd ¼ rhðu; vÞ ð2:2Þ
which is double sampling version of the class of estimators
r~t ¼ rf ðu* ; v* Þ
 
 v* ¼ s2 =S 2 and X;
Suggested by Srivastava and Jhajj (1986), where u* ¼ x =X;  S2
x x x
are known.
Sometimes even if the population mean X and population variance S 2x of x are not
known, information on a cheaply ascertainable variable z, closely related to x but
 compared to x remotely related to y, is available on all units of the population. This Finite correlation
type of situation has been briefly discussed by, among others, Chand (1975) and coefficient
Kiregyera (1980, 1984).
Following Chand (1975) one may define a chain ratio-type estimator for ryx as
    *2  2 !
x * Z sx Sz
r^1d ¼ r ð2:3Þ 893
x z* sx2
s*z 2

where the population mean Z and population variance S 2z of second auxiliary variable z
are known, and
X
z* ¼ ð1=n1 Þ zi ;
Downloaded by Professor Florentin Smarandache At 08:14 27 April 2018 (PT)

i[s*

X
s*z 2 ¼ ðn1 2 1Þ21 ðzi 2 z* Þ2
i[s*

are the sample mean and sample variance of z based on preliminary large sample s* of
size n1 ð. nÞ:
The estimator r^1d in equation (2.3) may be generalized as
 a1  a2  a3 *2 !a4
x s2x z* sz
r^2d ¼ r ð2:4Þ
x * s*x2 Z S 2z

where ai 0 s ði ¼ 1; 2; 3; 4Þ are suitably chosen constants.

Many other generalization of r^1d is possible. We have, therefore, considered a more
general class of ryx, from which a number of estimators can be generated.
The proposed generalized estimators for population correlation coefficient ryx, is
defined by

r^td ¼ rtðu; v; w; aÞ ð2:5Þ

 a ¼ s*2 =S 2 and t(u, v, w, a) is a function of (u, v, w, a) such that
where w ¼ z* =Z; z z

tð1; 1; 1; 1Þ ¼ 1 ð2:6Þ

Satisfying the following conditions:

(1) Whatever be the samples (s* and s) chosen, let (u, v, w, a) assume values in a
closed convex subset S, of the four-dimensional real space containing the point
P¼ (1, 1, 1, 1).
(2) In S, the function t(u, v, w, a) is continuous and bounded.
(3) The first and second order partial derivatives of t(u, v, w, a) exist and are
continuous and bounded in S.
To find the bias and variance of r^td we write:
 IJSE s2y ¼ S 2y ð1 þ e1 Þ;
31,10
 þ e1 Þ;
x ¼ Xð1

 þ e* Þ;
x * ¼ Xð1 1
894
s2x ¼ S 2x ð1 þ e2 Þ

s*x2 ¼ S 2x ð1 þ e*2 Þ;

 þ e* Þ;
z* ¼ Zð1
Downloaded by Professor Florentin Smarandache At 08:14 27 April 2018 (PT)

s*z 2 ¼ S 2z ð1 þ e*4 Þ;

syx ¼ S yx ð1 þ es Þ
such that
Eðe0 Þ ¼ Eðe1 Þ ¼ Eðe2 Þ ¼ Eðe5 Þ ¼ 0
and

E e*i ¼ 0 ; i ¼ 1; 2; 3; 4;
and ignoring the finite population correction terms, we write to the first degree of
approximation:
   
E e20 ¼ ðd400 2 1Þ=n; E e21 ¼ C 2x =n; E e*12 ¼ C 2x =n1 ; E e22 ¼ ðd040 2 1Þ=n;

  
E e*22 ¼ ðd040 2 1Þ=n1 ; E e*32 ¼ C 2z =n1 ; E e*42 ¼ ðd004 2 1Þ=n1 ;

 n  o. 
E e25 ¼ d220 =r2yx 2 1 n; Eðe0 e1 Þ ¼ d210 C x =n; E e0 e*1 ¼ d210 C x =n1 ;

 
Eðe0 e2 Þ ¼ ðd220 2 1Þ=n; E e0 e*2 ¼ ðd220 2 1Þ=n1 ; E e0 e*3 ¼ d201 C z =n1 ;

E e0 e*4 ¼ ðd202 2 1Þ=n1 ; Eðe0 e5 Þ ¼ {ðd310 =ryx Þ 2 1}=n;

 
E e1 e*1 ¼ C 2x =n1 ; Eðe1 e2 Þ ¼ d030 C x =n; E e1 e*2 ¼ d030 C x =n1 ;
 
E e1 e*3 ¼ rxz C x C z =n1 ; E e1 e*4 ¼ d012 C x =n1 ; Eðe1 e5 Þ ¼ ðd120 C x =ryx Þ=n;
  
E e*1 e2 ¼ d030 C x =n1 ; E e*1 e*2 ¼ d030 C x =n1 ; E e*1 e*3 ¼ rxz C x C z =n1 ;
  
E e*1 e*4 ¼ d012 C x =n1 ; E e*1 e5 ¼ ðd120 C x =ryx Þ=n1 ; Finite correlation
  
coefficient
E e2 e*2 ¼ ðd040 2 1Þ=n1 ; E e2 e*3 ¼ d021 C z =n1 ; E e2 e*4 ¼ ðd022 2 1Þ=n1 ;

Eðe2 e5 Þ ¼ {ðd130 =ryx Þ 2 1}=n; E e*2 e*3 ¼ d021 C z =n1 ;
895
 
E e*2 e*4 ¼ ðd022 2 1Þ=n1 ; E e*2 e5 ¼ {ðd130 =ryx Þ 2 1}=n1 ;
 
E e*3 e*4 ¼ d003 C z =n1 ; E e*3 e5 ¼ ðd111 C z =ryx Þ=n1 ;
Downloaded by Professor Florentin Smarandache At 08:14 27 April 2018 (PT)


E e*4 e5 ¼ {ðd112 =ryx Þ 2 1}=n1 :
where
  X
N
p=2 q=2 m=2  p ðxi 2 XÞ
 q ðzi 2 ZÞ
 m;
dpqm ¼ mpqm = m200 m020 m002 ; mpqm ¼ ð1=N Þ ð yi 2 YÞ
i¼1

( p, q, m) being non-negative integers.

To find the expectation and variance of r^td ; we expand t(u, v, w, a) about the point
P ¼ ð1; 1; 1; 1Þ in a second-order Taylor’s series, express this value and the value of r in
terms of e’s. Expanding in powers of e’s and retaining terms up to second power, we
have:

Eðr^td Þ ¼ ryx þ oðn 21 Þ ð2:7Þ

which shows that the bias of r^td is of the order n 2 1 and so up to order n 2 1, mean
square error and the variance of r^td are same.
Expanding ðr^td 2 ryx Þ2 ; retaining terms up to second power in e’s, taking
expectation and using the above expected values, we obtain the variance of r^td to the
first degree of approximation, as:
 h
Varðr^td Þ ¼ VarðrÞ þ r2yx =n C 2x t 21 ðPÞ þ ðd040 2 1Þt22 ðPÞ 2 At 1 ðPÞ
i  h
2 Bt2 ðPÞ þ 2d030 C x t 1 ðPÞt 2 ðPÞ 2 r2yx =n1 C 2x t21 ðPÞ þ ðd040 2 1Þt 22 ðPÞ
ð2:8Þ
2 C 2z t23 ðPÞ 2 ðd004 2 1Þt 24 ðPÞ 2 At 1 ðPÞ 2 Bt2 ðPÞ þ Dt 3 ðPÞ þ Ft 4 ðPÞ
i
þ 2d030 C x t 1 ðPÞt 2 ðPÞ 2 2d003 C z t3 ðPÞt4 ðPÞ

where t1(P), t2(P), t3(P) and t4(P), respectively, denote the first partial derivatives of
t(u, v, w, a) respect to u, v, w and a, respectively, at the point P ¼ ð1; 1; 1; 1Þ; with
 h  i
VarðrÞ ¼ r2yx =n d220 =r2yx þ ð1=4Þðd040 þ d400 þ 2d220 Þ 2 {ðd130 þ d310 Þ=ryx } ð2:9Þ
 IJSE A ¼ {d210 þ d030 2 2ðd120 =ryx Þ}C x ;
31,10
B ¼ {d220 þ d040 2 2ðd130 =ryx Þ};

D ¼ {d201 þ d021 2 2ðd111 =ryx Þ}C z ;

896
F ¼ {d202 þ d022 2 2ðd112 =ryx Þ}
Any parametric function t(u, v, w, a) satisfying equation (2.6) and the conditions 1 and 2
can generate an estimator of the class equation (2.5).
The variance of r^td at equation (2.6) is minimized for:
Downloaded by Professor Florentin Smarandache At 08:14 27 April 2018 (PT)

½Aðd040 2 1Þ 2 Bd030 C x 
t1 ðPÞ ¼  ¼ aðsayÞ;
2C 2x d040 2 d2030 2 1
 
BC 2x 2 Ad030 C x
t2 ðPÞ ¼ 2  ¼ bðsayÞ;
2C x d040 2 d2030 2 1
ð2:10Þ
½Dðd004 2 1Þ 2 F d003 C z 
t3 ðPÞ ¼  ¼ gðsayÞ;
2C 2z d004 2 d2030 2 1
 
C 2z F 2 Dd003 C z
t4 ðPÞ ¼ 2  ¼ dðsayÞ;
2C z d004 2 d2003 2 1

Thus the resulting (minimum) variance of r^td is given by:

  " #
1 1 2 A 2 {ðA=C x Þd030 2 B}2
min:Varðr^td Þ ¼ VarðrÞ 2 2 r þ 
n n1 yx 4C 2x 4 d040 2 d2030 2 1
" # ð2:11Þ
  D2 {ðD=C z Þd003 2 F}2
2 r2yx =n1 þ 
4C 2z 4 d004 2 d2003 2 1

It is observed from equation (2.11) that if optimum values of the parameters

given by (2.10) are used, the variance of the estimator r^td is always less than that of r
as the last two terms on the right hand sides of (2.11) are non-negative.
Two simple functions t(u, v, w, a) satisfying the required conditions are:
tðu; v; w; aÞ ¼ 1 þ a1 ðu 2 1Þ þ a2 ðv 2 1Þ þ a3 ðw 2 1Þ þ a4 ða 2 1Þ
tðu; v; w; aÞ ¼ u a1 v a2 w a3 a a4
and for both these functions t 1 ðPÞ ¼ a1 ; t2 ðPÞ ¼ a2 ; t 3 ðPÞ ¼ a3 and t4 ðPÞ ¼ a4 : Thus,
one should use optimum values of a1, a2, a3 and a4 in r^td to get the minimum variance.
 It is to be noted that the estimated r^td attained the minimum variance only when the Finite correlation
optimum values of the constants ai (i¼ 1, 2, 3, 4), which are functions of unknown coefficient
population parameters, are known. To use such estimators in practice, one has to use
some guessed values of population parameters obtained either through past experience
or through a pilot sample survey. It may be further noted that even if the values of the
constants used in the estimator are not exactly equal to their optimum values as given
by equation (2.8) but are close enough, the resulting estimator will be better than the 897
conventional estimator, as illustrated by Das and Tripathi (1978, Section 3).
If no information on second auxiliary variable z is used, then the estimator r^td
reduces to r^hd defined in equation (2.2). Taking z ; 1 in equation (2.8), we get the
variance of r^hd to the first degree of approximation, as:
 
1 2h 2 2
Downloaded by Professor Florentin Smarandache At 08:14 27 April 2018 (PT)

1
Varðrhd Þ ¼ VarðrÞ þ
^ 2 r C h ð1; 1Þ þ ðd040 2 1Þh22 ð1; 1Þ
n n1 yx x 1
ð2:12Þ
i
2 Ah1 ð1; 1Þ 2 Bh2 ð1; 1Þ þ 2d030 C x h1 ð1; 1Þh2 ð1; 1Þ

which is minimized for,

½Aðd040 2 1Þ 2 Bd030 C x 
h1 ð1; 1Þ ¼ ;
2C 2x ðd040 2 d2030 2 1Þ
  ð2:13Þ
BC 2x 2 Ad030 C x
h2 ð1; 1Þ ¼
2C 2x ðd040 2 d2030 2 1Þ

Thus the minimum variance of r^hd is given by:

  " #
1 1 2 A 2 {ðA=C x Þd030 2 B}2
min:Varðr^hd Þ ¼ VarðrÞ 2 2 r þ  ð2:14Þ
n n1 yx 4C 2x 4 d040 2 d2030 2 1

It follows from equations (2.11) and (2.14) that

" #
  D 2 {ðD=C Þd 2 F}2
2 z 003
min:Varðr^td Þ 2 min:Varðr^hd Þ ¼ ryx =n1 þ  ð2:15Þ
4C 2z 4 d004 2 d2003 2 1

which is always positive. Thus, the proposed estimator r^td is always better than r^hd :

3. A wider class of estimators

In this section, we consider a class of estimators of ryx wider than equation (2.5) given
by:

r^gd ¼ gðr; u; v; w; aÞ ð3:1Þ

where g(r, u, v, w, a) is a function of r, u, v, w, a and such that

  
IJSE gðr; 1; 1; 1; 1Þ ¼ r^td and
›gð·Þ
¼1
31,10 ›r ðr;1;1;1Þ

Proceeding as in Section 2, it can easily be shown, to the first order of approximation,

that the minimum variance of r^gd is same as that of r^td given in equation (2.11).
It is to be noted that the difference-type estimator
898
r d ¼ r þ a1 ðu 2 1Þ þ a2 ðv 2 1Þ þ a3 ðw 2 1Þ þ a4 ða 2 1Þ;
is a particular case of r^gd ; but it is not the member of r^td in equation (2.5).

4. Optimum values and their estimates

Downloaded by Professor Florentin Smarandache At 08:14 27 April 2018 (PT)

The optimum values t 1 ðPÞ ¼ a; t 2 ðPÞ ¼ b; t3 ðPÞ ¼ g and t 4 ðPÞ ¼ d given at equation
(2.10) involve unknown population parameters. When these optimum values are
substituted in equation (2.5), it no longer remains an estimator since it involves
unknown (a, b, g, d), which are functions of unknown population parameters, say,
dpqm ðp; q; m ¼ 0; 1; 2; 3; 4Þ; Cx, Cz and ryx itself. Hence, it is advisable to replace them
by their consistent estimates from sample values. Let ða^; b^; g^; d^Þ be consistent
estimators of t1(P), t2(P ), t3(P) and t4(P ), respectively, where
^ ^ 2 1Þ 2 B^ d^030 C^ x 
^t1 ðPÞ ¼ a^ ¼ ½Aðd040  ;
2 2
2C^ x d^040 2 d^030 2 1
h 2 i
^ d^030 C^ x
B^ C^ x 2 A
^t2 ðPÞ ¼ b^ ¼  ;
2 2
2C^ x d^040 2 d^030 2 1
ð4:1Þ
^ ^ 2 1Þ 2 F^ d^003 C^ z 
^t3 ðPÞ ¼ g^ ¼ ½Dðd004
  ;
2 2
2C^ d^004 2 d^ 2 1
z 003

h 2 i
C^ z F^ 2 D^ d^003 C^ z
^t4 ðPÞ ¼ d^ ¼  ;
2 2
2C^ z d^004 2 d^003 2 1

with
^ ¼ ½d^210 þ d^030 2 2ðd^120 =rÞC^ x ;
A

B^ ¼ ½d^220 þ d^040 2 2ðd^130 =rÞ;

D^ ¼ ½d^201 þ d^021 2 2ðd^111 =rÞC^ z ;

F^ ¼ ½d^202 þ d^022 2 2ðd^112 =rÞ;

 C^ x ¼ sx =x; Finite correlation
coefficient
C^ z ¼ sz =z;
 
p=2 q=2 m=2
d^pqm ¼ m^pqm = m^200 m^020 m^002
899
Xn
m^pqm ¼ ð1=nÞ ð yi 2 y Þp ðxi 2 x Þq ðzi 2 zÞm
i¼1

Xn
z ¼ ð1=nÞ zi ;
Downloaded by Professor Florentin Smarandache At 08:14 27 April 2018 (PT)

i¼1

X
n
s2x ¼ ðn 2 1Þ21 ðxi 2 x Þ2 ;
i¼1

X
n
x ¼ ð1=nÞ xi ;
i¼1

r ¼ syx =ðsy sx Þ;

X
n
s2y ¼ ðn 2 1Þ21 ð yi 2 y Þ2 ;
i¼1

X
n
s2z ¼ ðn 2 1Þ21 ðxi 2 zÞ2 :
i¼1

We then replace (a, b, g, d) by ða^; b^; g^; d^Þ in the optimum r^td resulting in the estimator
r^*td say, which is defined by:

r^*td ¼ rt* ðu; v; w; a; a^; b^; g^; d^Þ; ð4:2Þ

where the function t*(U ), U ¼ ðu; v; w; a; a^; b^; g^; d^Þ is derived from the function t(u,
v, w, a) given at equation (2.5) by replacing the unknown constants involved in it
by the consistent estimates of optimum values. The condition (2.6) will then imply
that

t* ðP* Þ ¼ 1 ð4:3Þ

where P* ¼ ð1; 1; 1; 1; a; b; g; dÞ:

We further assume that
  
IJSE ›t* ðU Þ ›t* ðU Þ
t1 ðP* Þ ¼ ¼ a; t 2 ðP* Þ ¼ ¼b
31,10 ›u U ¼P * ›v U ¼P *

 
›t* ðU Þ ›t* ðU Þ
t3 ðP* Þ ¼ ¼ g; t 4 ðP* Þ ¼ ¼d
900 ›w U ¼P * ›a U ¼P *
ð4:4Þ
 
›t* ðU Þ ›t* ðU Þ
t5 ðP* Þ ¼ ¼ 0; t6 ðP* Þ ¼ ¼0
›a^ U ¼P * ›b^ U ¼P *
Downloaded by Professor Florentin Smarandache At 08:14 27 April 2018 (PT)

 
›t* ðU Þ ›t* ðU Þ
t7 ðP* Þ ¼ ¼ 0; t8 ðP* Þ ¼ ¼0
›g^ U ¼P * ›d^ U ¼P *

Expanding t*(U) about P* ¼ ð1; 1; 1; 1; a; b; g; dÞ; in Taylor’s series, we have:


r^*td ¼ r t* ðP* Þ þ ðu 2 1Þt *1 ðP* Þ þ ðv 2 1Þt*2 ðP* Þ þ ðw 2 1Þt*3 ðP* Þ

þða 2 1Þt *4 ðP* Þ þ ða^ 2 aÞt*5 ðP* Þ þ ðb^ 2 bÞt*6 ðP* Þ þ ðg^ 2 gÞt *7 ðP* Þ ð4:5Þ

þðd^ 2 dÞt*8 ðP* Þ þ second order terms

Using equation (4.4) in (4.5) we have:

r^*td ¼ r½1 þ ðu 2 1Þa þ ðv 2 1Þb þ ðw 2 1Þg þ ða 2 1Þd þ second order terms ð4:6Þ
Expressing equation (4.6) in term of e’s squaring and retaining terms of e’s up to second
degree, we have:
 2
 * 2 1  
r^td 2 ryx ¼ r2yx ð2e5 2 e0 2 e2 Þ þ a e1 2 e*1 þ b e2 2 e*2 þ ge*3 þ de*4 ð4:7Þ
2
Taking expectation of both sides in equation (4.7), we get the variance of r^*td to the first
degree of approximation, as:
 "
 #
 1 1 2 A 2 {ðA=C x Þd030 2 B}2
Var r^*td ¼ VarðrÞ 2 2 r þ 
n n1 yx 4C 2x 4 d040 2 d2030 2 1
" # ð4:8Þ
  D2 {ðD=C z Þd003 2 F} 2
þ r2yx =n1 þ 
4C 2z 4 d004 2 d2003 2 1

which is same as equation (2.11), we thus have established the following result.
 Result 4.1. If optimum values of constants in equation (2.10) are replaced by Finite correlation
their consistent estimators and conditions (4.3) and (4.4) hold good, the resulting coefficient
estimator r^*td has the same variance to the first degree of approximation, as that of
optimum r^td :
Remark 4.1. It may be easily examined that some special cases:
^ ^
(1) r^*td1 ¼ ru a^v b w g^a d ;
 901
(2) r^*td2 ¼ r{1 þ a^ðu 2 1Þ þ g^ðw 2 1Þ} {1 2 b^ðv 2 1Þ 2 d^ða 2 1Þ}
(3) r^*td3 ¼ r½1 þ a^ðu 2 1Þ þ b^ðu 2 1Þ þ g^ðw 2 1Þ þ d^ða 2 1Þ
(4) r^*td4 ¼ r½1 2 a^ðu 2 1Þ 2 b^ðu 2 1Þ 2 g^ðw 2 1Þ 2 d^ða 2 1Þ21
of r^*td satisfy the conditions (4.3) and (4.4) and attain the variance (4.8).
Remark 4.2. The efficiencies of the estimators discussed in this paper can be
Downloaded by Professor Florentin Smarandache At 08:14 27 April 2018 (PT)

compared for fixed cost, following the procedure given in Sukhatme et al. (1984).

5. Empirical study
To illustrate the performance of various estimators of population correlation
coefficient, we consider the data given in Murthy (1967, p. 226). The variates are:
y ¼ output; x ¼ number of workers; z ¼ fixed capital; N ¼ 80; n ¼ 10; n1 ¼ 25;

X ¼ 283:875; Y ¼ 5182:638; Z ¼ 1126; C x ¼ 0:9430; C y ¼ 0:3520;

C z ¼ 0:7460; d003 ¼ 1:030; d004 ¼ 2:8664; d021 ¼ 1:1859; d022 ¼ 3:1522;

d030 ¼ 1:295; d040 ¼ 3:65; d102 ¼ 0:7491; d120 ¼ 0:9145; d111 ¼ 0:8234;

d130 ¼ 2:8525; d112 ¼ 2:5454; d210 ¼ 0:5475; d220 ¼ 2:3377; d201 ¼ 0:4546;

d202 ¼ 2:2208; d300 ¼ 0:1301; d400 ¼ 2:2667; ryx ¼ 0:9136; rxz ¼ 0:9859;

ryz ¼ 0:9413:

The percent relative efficiencies (PREs) of r^1d ; r^hd ; r^td with respect to conventional
estimator r have been computed and compiled in Table I.
Table I clearly shows that the proposed estimator r^td (or r^*td ) is more efficient than r
and r^hd :

Table I.
Estimator r r^hd r^td (or r^*td )
The PRE’s of different
PRE(., r) 100 129.147 305.441 estimators of ryx
 IJSE References
31,10 Chand, L. (1975), “Some ratio-type estimators based on two or more auxiliary variables”,
PhD dissertation, Iowa State University, Ames, Iowa.
Das, A.K. and Tripathi, T.P. (1978), “Use of auxiliary information in estimating the finite
population variance”, Sankhya, Ser. C, 40, pp. 139-48.
Gupta, J.P. and Singh, R. (1989), “Usual correlation coefficient in PPSWR sampling”, Journal of
902 Indian Statistical Association, Vol. 27, pp. 13-16.
Gupta, J.P., Singh, R. and Lal, B. (1978), “On the estimation of the finite population correlation
coefficient-I”, Sankhya C, Vol. 40, pp. 38-59.
Gupta, J.P., Singh, R. and Lal, B. (1979), “On the estimation of the finite population correlation
coefficient-II”, Sankhya C, Vol. 41, pp. 1-39.
Kiregyera, B. (1980), “A chain-ratio type estimators in finite population, double sampling using
Downloaded by Professor Florentin Smarandache At 08:14 27 April 2018 (PT)

two auxiliary variables”, Metrika, Vol. 27, pp. 217-23.

Kiregyera, B. (1984), “Regression type estimators using two auxiliary variables and the model of
double sampling from finite populations”, Metrika, Vol. 31, pp. 215-26.
Koop, J.C. (1970), “Estimation of correlation for a finite Universe”, Metrika, Vol. 15, pp. 105-9.
Murthy, M.N. (1967), Sampling Theory and Methods, Statistical Publishing Society, Calcutta,
India.
Rana, R.S. (1989), “Concise estimator of bias and variance of the finite population correlation
coefficient”, Jour. Ind. Soc. Agr. Stat., Vol. 41 No. 1, pp. 69-76.
Singh, S., Mangat, N.S. and Gupta, J.P. (1996), “Improved estimator of finite population
correlation coefficient”, Jour. Ind. Soc. Agr. Stat., Vol. 48 No. 2, pp. 141-9.
Srivastava, S.K. and Jhajj, H.S. (1986), “On the estimation of finite population correlation
coefficient”, Jour. Ind. Soc. Agr. Stat., Vol. 38 No. 1, pp. 82-91.
Sukhatme, P.V., Sukhatme, B.V., Sukhatme, S. and Asok, C. (1984), Sampling Theory of Surveys
with Applications, Indian Society of Agricultural Statistics, New Delhi.
Wakimoto, K. (1971), “Stratified random sampling (III): estimation of the correlation coefficient”,
Ann. Inst. Statist. Math., Vol. 23, pp. 339-55.

Further reading
Singh, R.K. (1982), “On estimating ratio and product of population parameters”, Cal. Stat. Assoc.
Bull., Vol. 32, pp. 47-56.
Srivastava, S.K. (1967), “An estimator using auxiliary information in sample surveys”, Cal. Stat.
Assoc. Bull., Vol. 16, pp. 121-32.
Srivastava, S.K. and Jhajj, H.S. (1983), “A class of estimators of the population mean using
multi-auxiliary information”, Cal. Stat. Assoc. Bull., Vol. 32, pp. 47-56.
Srivenkataremann, T. and Tracy, D.S. (1989), “Two-phase sampling for selection with
probability proportional to size in sample surveys”, Biometrika, Vol. 76, pp. 818-21.