Page 41 - 1990(3) Cesus Of Agriculture, provience report-Punjab

LIMITATION OF CENSUS DATA

1. The results of sample census/ . take into account not only the
survey are . subjected to two point estimates but should also·
types of errors viz., ( i) the take into account the sampling
sampling errors and (ii) the error associated with the es-
non~sampling errors. The timates. Thusj the diffe~ence in
sampling errors are not •~rrors' various items of 1990 Census may
but . the probable difference not be taken as the real dif-
between the population mean and ference in absolute terms as both
the sample mean. Non-sampling the estimates are subject to
errors , are due to biases sampling variability. '
introduced by inefficient
implementation of sampling plan. .The successive census~s taken
In case two sample surveys, after a perioq of 10 years are
having . the same d~sign and same bound to reflect not only the
size are . conducted at · th~ same diff~rences due to passage of
time with equal effi6iency on a time but also.have an i~pact of
given population, it. is not any change/ improvement . made in
necessary that the two will give the census methodology~ The . cen-
identical · results. - The sus results are also affected by
difference will be due • to socio/political conditions
"sampling errors". While one can prevailing at the time of
measure the · extent of sampling conducting a c~nsus;
er·rors, the magnitude of . non-
sampling errors is difficult to 4. An Agrictiltural Census primarily
measure. portrays utilization .of :agricul-: .
tural resources and provides base
2. In spite of adequate training of for structural information for
_, enumerators, and elaborate ar--' formulating long term
rangements for supervision, to- agricultural policies. It also
tal elimination of different ~provides base to st~dy relation-
sources of errors .. is not pos- ship between different factors of
sible in an operation so exten- production by fQrm size and
sive and wide spread as that of tenurial arrangements. The ag-
a census. Starting with the in- gregates · of certain ' _ cparacte-
accuracies on the part of ristics arrived at during the
enumerators and f ci.ul ty compre- course of stud~ing various fac-
hension or misapprehensions of tors affecting · the structure of
the respondents, errors in- agriculture are not of much sig-
troduced during processing and nificance in the census but it is
then during publication of the reiationship between various
results, all add up towards the factors which . is of paramount
non-sampling errors. The total impprtanbe. The aggregative type
extent of non~sampling errors of data on land and crops can , .
~arinot be precisely measured best be had from the · existing
though an attempt has been made elaborate system of current
to have an idea about the .same statistics.
through a Po~t Enumeration Stir-
Vey carried out immediately 5. There are valid reasons for the
after the census. The results of census figures being d'ifferent
Post Enumera.tion Survey have from the · current agricultural
been included in the report. ·statistics. lt is ·part;.ly be-
cause of conceptual difference
3. While comparing the data of the and partly .on • account of dif-
two sample censuses, . one - has to ference -in •data · collection and ·

(xxxix)

I