Page 98 - DSOC404_METHODOLOGY_OF_SOCIAL_RESEARCH_ENGLISH

Methodology of Social Research

notes religion, social class, gender etc.) of the universe. in other words, on the basis of common attributes
or aspects, by dividing the units of universe in various sub-samples or classes, first an attempt is
made to bring homogeneity in the universe. Considering the attributes of the universe and keeping
in mind that problem of research, researcher may easily decide that on what basis and in how many
categories, division must be made. After dividing the universe in categories in such a manner, with
the help of any appropriate method of selecting random sample, samples are selected in appropriate
number from each category. While selecting type sample from each category, it must be kept in mind
that as far as possible, only that much units are selected from each category, in which ratio total units
of category are there in the universe. for e.g. if in a universe, teachers are 120, engineers 10, doctors
30 and lawyers 50 and if we have to take out 10 percent sample, then we will select 12 teachers,1
engineer, 3 doctors and 5 lawyers as sample with the help of any one method of selecting random
sample. In this way, in stratified sampling help is taken of method of selecting random sample also,
which is why it is often called stratified random sampling.
Kinds of Stratified Sampling—stratified samples are of three types: - a. Proportionate—in it from
each category units are selected in the same ratio in which ratio total units in the category are with
the universe. b. Disproportionate—in it equal numbers of units are chosen from each category. No
matter how many numbers of units are there in any category of universe. this means that if in various
categories, units are not in equal numbers, then, in sample their number will be disproportionate. c.
Stratified Weighted Sampling—it is a mixed form of the above given two types. in it equal numbers
of units are selected from each category but afterwards, units of categories of more numbers are
provided weight and their impact is increased. this weight is increased in the same ratio in which
ratio units of the category are there in the universe.
Merits of Stratified Sampling—in such kind of sampling, following attributes may be mentioned:
(a) in it units of each category of universe get a place in sample and there is no possibility of
negligence of any important category. though in random sampling there is equal possibility
of each unit being selected, still some times some important category or unit is left out.
Stratified sampling prevents this possibility.
(b) in it if various categories are divided carefully, then from various categories small number
of units may also represent the universe. in random sampling attribute of representation is
achieved only when units are selected in sufficient numbers.
(c) in it there is a facility of discarding any unit and selecting some other units in its place, if need
be. if in initially chosen sample any person or representative is such that it is not possible
to establish contact with him, then in its place from the same category, some other person
or unit may be selected. With such kind of change there is no obstacle in sample remaining
representative.
(d) Category division can be done on geographical basis also. By categorising from the point of
view of area, time and money is saved and it is easy to contact the units. in random sampling
no such control can be there and selected units may be wide-spread.
Defects of Stratified Sampling—despite of above-mentioned merits, stratified sampling has some
defects also-
(a) if division of categories is not done properly there may be false inclination or bias created
in the sample. in sample selected in that manner, units of any category may be too many or
too less in number and in this case, sample does not remain representative.
(b) if there is too much difference in size of various categories, then selecting units in equal ratio
will be difficult and in this manner, if sample is not proportionate, it cannot be representative
also.
(c) if selection of units of category is done on disproportionate basis, then afterwards, use of
weightage has to be done. in this task, researcher’s false inclination or bias may leave their
impact.

92 loVely professional uniVersity