Page 23 - 1990(3) Cesus Of Agriculture, provience report-Punjab

work of key punching and also circles helped in equalizing the
reduced the number of faulty work load per enumerator and
returns in computer edit from selecting cluster of households
about 40% to less than 10%~ facilitated the enumeration work.

Sampling Plan This sampling plan called for
preparation of rather elaborate
Different sample designs. were frames involving great deal of ef-
evolved for different parts. of the fort and expense. The frames
country depending upon local condi- prepared were:
tions and quantum of relevant in-
formation available for selecting a i) National Certainty Holdings,
sample. Consequently, in major list comprising all govern-
parts of the country comprising ment farms · and private farms of
rural settled areas of NWFP, Punjab 250 acres or more in c.ertain
and Sindh Provinces, a three stage areas, and 150 acres or more in
weighted and stratified sample was others, according to their
used whereas a single stage frequency.
weighted sample was used in rural
settled areas of Balochistan. In ii) Mouza list, containing only
rest of the country comprising en- rural mouzas arranged by pat-
tire . urban areas and unset.tled war and kanungo circles - show-
rural and tribal areas, a single ing mouza-wise total area,
stage systematic sample was used. cultivated area and number of
Barring a few exception where households. Thls list excluded
sample was selected at district purely forest mouzas and those
level, a separate sample was urban mouzas which were
selected for each sub-division. In declared urban or formed part
the three stage sample, patwar of some urban wards.
circles _in the first stage and
mouzas in the second stage were iii) A complete list of urban wards.
selected with probability propor,-
tional to size whereas households iv) .A list of households in
in the form of roughly equal selected mouzas showing total
clusters were selected in the third area owned and area rented in.
stage with equal probability. This With the help of this list,
sample design was basically the households having 25 acres or
same as used in 1972 and 1980 more as area owned and/or area
Agricultural Censuses. However, rented in were segregated to be
minor improvements in the design treated as . Mouza Certainty
and slight changes in the size were Holdings (MCH). Another list
made to further reduce the sampling of households with 25 or more
variability without an appreciable cattle and/or buffaloes; 25 or
increase in the sample size. more camels; 50 or more sheep
and/or goats was prepared .
The choice of three stage . These households were also
sample design was based not only on treated as Mouza Certainty
sampling corisiderations bcit also in , · Holdings.
view of a number 6f administrative
considerations to control the non- The selection procedure adopted
sa-mpling errors. For example, the for various sample stages is given
selection of patwar circles in the below:
first stage m~de it possible to
know, in advance, and regulate the Each of the patwar circles and
number of enumerators required. · mouzas were given a measure of $iz~
Similarly, the selection· of two as the geometric mean of the
mouzas from each of the patwar cultivated areas of 1987-88 and

(xxi)

I