Page 40 - Agricultural Census 2000 NWFP Vol II Part 1

Therefore, it is conceivable that In Agricultural Census 2000 the
the estimate of a given item best that could be done was to
with very lovv sampling error control and min1m1ze non-
might in fact be a bad estimate, sampling errors through careful
as it is amenable to have planning, elaborate training
substantial non-sampling errors. and · intensive supervision.
A "good" estimate is, therefore, Comorehensive editing of
the one in which the sum total census data for consistency
of the sampling errors plus ali and relational credibility was
other errors is very !ov✓. In a another measure adopted to
census of agriculture, one may reduce such errors.
visualize that the chances of
occurrence of · non-sar,,pling (3) Consistency and Reasonable-
errors begin with planning of the ness of Data
census . and end with the
publication of data. !n other It is evident from the foregoing
words, there are almost endless , , paragraphs that, under the
chances of. the occurrence of · circumstances, it 1s not
non-sampling errors. It is due to possible to develop a
this reason that in a census, it comprehensive measure to
pays to give more attention to adjudge the quality of the
the control of non-sampling census data. The sampling
errors than to the reduction of error is one indicator with limited
sampling errors, and quite often utility while the non-sampling
some increase in the latter is to errors are of multiple nature and
be tolerated for the sake ·of very hard to estimate. Thus, it
reducing the former. But, it is becomes necessary to consider
very difficult as well as other indicators for judging the
expensive to measure non- quality of . data. An important
sampling errors in quantitative indicator in this connection is
terms and specially to obtain a the internal consistency and
comprehensive measure of an reasonableness of the data.

the errors that might have This is an extremely useful
crept into the census d~ta measure and it requires very
at various stages. Oniy a little extra census data for its
very elaborate checking and application. The data are said to
evaluation of the census be consistent when the values
operation can yield some usable obta•ned in related but separate
measure of non-sampling items bear a logical relationship
errors, but most countries find it with one another. For instance,
difficult to spare manpower and cne may compare the cropped
funds for this type of evaluation · area · with cuitivated area or the
after having spent huge sums of existence of a tubewell on the
money on the census itself. farm with area irrigated by

( XXXVHI)