Page 42 - Agricultural Census 2000 NWFP Vol II Part 1

LIMITATIONS OF CENSUS DATA

1. Like faces and · watches, the that these may not loom large in
estimates of sample censuses the census estimates.
are destined to differ with each
other. For example, if one 3. While, comparing census data
conducts two sample censuses with previous sample censuses,
having the same design, same one has to take into account not
size and at the same time with only the estimates but also
equal efficiency on a given the associated sampling errors
population, it 1s not necessary and changes / improvements
that the two will give identical incorporated in the census
estimates. The difference will be methodology. Further, the
due tG "sampling and non- successive censuses taken after
sampling errors" which are bound a period of ten years are bound
to occur. Therefore, the estimates to reflect the differences not only
presented 1n this report are, due to passage of time and
inevitably subjected to sampling legacy of the Government
as well as non-sampling errors policies but also on account of
as.already discussed in detail. the socio-political conditions
prevailing at the time of
2. The complete elimination of non- conducting a census. · The
sampling errors is not possible in constitution of new tehsils,
an operation so extensive and districts and divisions also
wide spread as that of a census, renders the comparison of
despite the adequate training historical data at these levels
of enumerators and elaborate meaningless without proper
arrangements for superv1s1on . . adjustment for administrative
Starting with the inaccuracies boundaries.
and faulty comprehension on
the part of enumerators, 4. An agricultural ·census primarily
misapprehensions of the portrays utilization of agricultural
respondents, errors introduced resources and provides base
during processing and then for structural information for
during publication of results, all formulating long term agricultural

add up towards the non-sampling policies. It also provides base
errors. The existence of such to study relationships between
errors might not be out rightly different 'factors of production by
ruled out in the census estimates. farm size and by tenure.
However, every effort was made The aggregates of certain
to keep strict check on all characteristics arrived at during
possible non-sampling errors, so the course of studying various

(XL)