Page 38 - Agricultural Census 2000 NWFP Vol II Part 1

In general terms the quality of items, their sampling errors are
the census data may be adjudged on brought on comparable terms
the criterion of (1) internal by expressing them as
consistencies and reasonableness; percentages of relevant sample
and (2) by comparison with other estimates. The resultant
corresponding series. The estimates measure is designated as co-
presented in this report were checked efficient of variation which can
for quality for technical terms as well be worked out directly from the
as for general terms. census data. For the
Agricultural Census 2000 co-
(1) Sampling Error efficients of variation have been
worked out for a number of
Sampling error is a measure to important items and given in ·
determine the precIsIon / Appendix-Ill for enlighting the
consistency of the sample data users about accuracy of
results, i.e. the closeness of the concerned estimate(s). The
sample results to those formula used for estimating
obtained through complete , relative variance (R.V.) is
count conducted with exactly described in the succeeding
similar arrangements of paragraphs.
training, enumeration,
supervision, processing, etc. In a) Relative Variance (R.V.)
other words, sampling error is
Estimates at Sub-division Level
not 'error', as the word error
portrays, but is the probable At the sub-division level, R.V.
difference between the
can be estimated simply by first
population mean and the
raising the sample observations . .
sample mean. Suppose, if a
of each patwar circle to stratum
number of estimates for an
level using the overall raising
item are developed from all factor for each holding but
possible samples of same size
excluding all National Certainty
drawn with simirar procedure
Holdings (NCHs), and then
from a population,, their mean
summing up to obtain an
would be equal to the mean
estimate for each half stratum
based upon complete count as under:-
and their dispersion about that
mean, measured in terms of
A r1 qi (Nk - N'k - N;'k) Yh'1ikl'"
variance, is called sampling
yh'1 = L [ L L
variance. The square root of
k I"' ph'1jk
sampling variance is known as
sampling error.
Qj N"k yh'1jkl"
+ I I ]
Further, to compare precision of
k I" ph'1j ·
sample estimates of different

( XXXVI)