Page 22 - 1990(3) Cesus Of Agriculture, provience report-Punjab
P. 22
cluster sampling were located in such an activity to dispel un-
chunks which were easy to founded fears from the mind of
enumerate. The use of cluster the respondents.
sampling facilitated enumeration
work particularly in 6ase of m) A farmer could report the area
mouzas having scattered popula- coveted by chemical fertilizets
tion. and plant protection measures but
the quantity ,involved dould not
h) The time required for selecting be ascertained precisely.
a stratified sample in ·the field
was much more than the time re- n) Study of variance indicated that
quired for clustEsr sampling. a two stage sample with selection _
of - mouzas in the first stage and ·
i) In stratified sampling, there of househofds in the second stage
could be no control on the num- would be desirable. But it was
ber of households to be enume- found that adopting a three stage
rated in each mouza as the dis- design by selection of patwar
tribution of households by cate- circles in the first stage,
gories, would not be known in mouzas in the second -stage and
advance at mouza_ level. In case households in the third stage
of cluster sampling, the minimum would be preferable in view of
number of households to be the lower cost and better con-
enumerated per mouza could be trol over the number ·of the
fixed even on the basis of enumerators to be employed.
roughly estimated households.
o) The impact of large variation in
j) Within the sample mouzas cluster the size of mouzas on the extent
sampling could be resorted to of sampling variance could be
only for those households which reduced by selecting pat war
had less than 25 acres of land circles and mouzas with robabi-
and households reporting 25 lity proportionate , to size.
acres or more should be
enumerated completely to control p) The within mouza component of the
variability. variance in a three stage design
was usually the major component
k) Of the total hou seholds select-ed of overall vari~nce and was more
for enumeration, nearly 30% were significant in cluster sampling
found to be without land. A than in systematic sampling. But
short version of the main ques- cluster sampling was considered
tionnaire, excluding questions to be more suitable for
on land and crops, could be used controlling the non-sampling
for landless househo1ds. · errors.
1) The enumerators did not find it q) It was confirmed that enumera-
difficult to enlist the co- tion of large holdings with cer-
operation of the respondents for tainty reduci~d the within mouza
pretest work because of their sampling variance appreciably.
contacts with the farming
community but some of the · r) It was observed that computer
farmers misunderstood purp-ose of edit of census data was neces-
the pretest and considered it a sary even though a comprehen-
step towards a new programme of sive manual edit was carried out.
agrarian reforms, taxation At the same time manual edit was
proposals or compulsory pro- -found quite helpful as it took
curement of foodgrains. A - care of the enumeration and other
publicity programme would be errors as well as 9ases of poor
very essential before launching l~gibility which facilitated the
. (XX)
