Page 39 - Agricultural Census 2000 NWFP Vol II Part 1

sub-division (g), then the
/\ r2 qi (Nk - N'k - N"k) Yh'2ikl' · R.V.for the district with g = 1 to
yh'2 = L [ L . L u sub-divisions in the district is
k I"' ph'2jk LI /\
2
I ( RVYW ) Y w
9 9
Qj N"k g
yh'2jkl"
+ I I RVYW =
k I" ph'2j

then, R.V. for the sub-division
and
estimate of aggregates of the
attribute is obtained by taking
LI /\
square of differences between 2
:I ( RVXW ) X w
9
estimates for successive pairs g 9
of half strata divided by the
RVXV\/ =
sub-division estimates as
LI /\
follows:- ( :I Xw·)2
9
I\ I\ g
I\ I\ L' yh'1 -Yh'2 R.V. estimates for provinces
Var.(Y) = [C.V.(Y)]2 = I[ ]2 and country levels have also
h' I\ been worked out on the same
y
pattern as at the district level,
Similarly, R.V. for the sub-
division estimate of units having (2) Non-Sampling Errors
the attribute is :-
The errors other than those due
I\ I\
to sampling are called non-
I\ I\ L' xh'1 - xh'2 sampling errors. These
Var.(X) = [C.V.(X)]2 = I [ ]2 include errors in sampling
h' I\
frame, selection of the
X
sample, enumeration, incorrect
Per cent Sampling,_ Error, i.e.
response due to biases and
ignorance, errors in recording
(C.V.) = /R.V. x 100
answers, processing errors, etc.

b) Estimates at District, Province
Interestingly, sampling and non-
and Country Levels
sampling errors are negatively
related. In other words, ah
If the R.V. for the aggregate of
increase in sampling error
characteristic (W) for. sub-
would cause a decrease in the
division (g) is denoted by non-sampling error and vice
RVYW and Yw is the estimate versa. Normally, an increase in
9
9
for the aggregate (Y) sampling error is due to
of characteristic (W) for decrease In sample size.

( XXXVII )