The question naturally arises as to why the researcher needs to base his inference about the population on only a sample from it rather than on itself. In some researches the number of members in a population are so few that it is rater desirable that all members be included in the sample, in which case, the sample is the population itself. For example, a researcher aiming at improving the existing education system, needs to study the attitude of the people at the decision making level. In this case there may be total of only, say, 40 people at that level and it would be desirable to include all of them in the study. There are occasions when, even if the population is large, a complete coverage is a necessity. For example, suppose that a descriptive study is aimed at identifying the contraceptive needs of all the family planning clinics throughout the country. If the objectives of this need identification study is to satisfy these needs from the central family planning store, then taking a sample of clinics would not serve the purpose – a complete coverage of all the clinics would be required. However, there are many circumstances in which a sample survey would be preference to a complete coverage. The reasons, in brief, are discussed below.
Usually the money and
time available for a research project are adequate to study on a fraction of
the population. In a demographic research, for example, it much cheaper and
much less time consuming to collect data from, say, only 5000 couples than from
100000 couples. Obviously, the cost and the time involved are tremendously
reduced by sampling. Indeed, with a limited budget and limited time, a research
involving a large population can have afforded only through sampling. The time
factor is very important when the information is needed urgently. At the time
of an urgent need, collection of data from all members of a huge population
requiring a long time would defeat the purpose of the research. In such a
situation sampling is the only feasible solution to the problem. In some cases,
a complete coverage may take years, in which case, part if not whole, of the research
interest may be lost by the time the results are ready.
Another merit of sampling
is that a researcher can afford employing highly trained interviewers for
collecting information from the members of the sample while the enormous
interviewing project for a complete coverage would, by necessity, required a
large staff of interviewers, a many of them will, very likely, be of lower
quality. Understandably, the highly trained interviewers. Moreover, since the
volume of administrative and managerial work is reduced significantly for a
sample compared to a complete coverage, sampling permits a more careful
supervision of the field work, data quality check, subsequent processing of
data, and so on. As a result, the sample may produce even more accurate results
than would be expected on the basis of complete coverage.
As mentioned before, the
purpose of selecting a sample from a population is to draw inference about the
population characteristics on the basis of the analysis of sample data the key
to achieving this purpose is to select a representative sample. A perfectly representative
sample actually reflects all various that exist in the population. If the
population is homogeneous, a sample of only one member will perfectly represent
the population. Thus, if all households in a community have the same income, a
sample of only one household will perfectly represent the population of incomes
of the community households, an is enough to tell the income story of all
households in the community. But most of the populations a researcher usually
encounters are not homogeneous, and as a result, samples of many members need
to be drawn. A large sample is no guarantee of representativeness, however. As mentioned
before, a number of techniques have been developed to draw samples which,
although not perfectly representative, can be used, for practical purposes, as
sufficient bases for drawing inferences about the population.
Source: Methods and Techniques of Social Research by Abu Jafar Mohammad Sufian