In many marketing research projects, conducting a census study or study of the entire universe will be impractical on account of limitations of time and money. Hence, sampling becomes inevitable.
Sampling is used to collect primary data when the sources of data are far too many to be exhaustively handled. In view of the importance of sampling in the data collection process, we are discussing the subject at length. Obviously, a sample is only a portion of the universe or population. The success of sampling depends on the extent to which the characteristics of the sample truly represent those of the universe.
According to Yule, a well known statistician, the object of sampling is to get the maximum information about the parent population with the minimum effort. Often, people who are not familiar with the scientific basis of sampling have an impression that data collected through sampling is less reliable than data generated by exhaustively covering the entire population. This impression is erroneous. If properly done, sampling produces representative data on the entire population.
Advantage of sampling:
Sampling saves cost and time. It enables collection of information that is â€˜okâ€™ for the given purpose at a lesser cost and time. It enables better supervision of the information gathering task and better presentation of the data. It also helps ensure the required degree of precision. Before we go into further details relating to sampling, it is essential to understand certain terminologies that are commonly used in relation to statistical sampling.
Terms commonly used in sampling:
Population or universe: In sampling, the term â€˜populationâ€™ or â€˜universeâ€™ means the totality of all elements that are relevant to the scope of the problem under study. All of them posses the characteristics under investigation and hence all of them are sources of relevant data as far as the given research issue is concerned. Any population is specific to the study on hand and the characteristics under investigation. For example, if a study is undertaken to find out the fertilizers buying habits of small farmers owning less than half a hectare of land will collectively constitute the population for the study.
When the researcher defines the characteristics under investigation, it automatically delimits the population. Population, in the sampling context, need not necessarily mean living persons; it may denote non-living objects like companies, houses or events. â€˜Populationâ€™ is also referred to as â€˜universeâ€™.
Sample: Sample is a part of the population or universe.
Parameter: It is a value or a statement that explains true characteristics of an entire population. It is the value that would be obtained if the population were covered in the measurement. In the example of small farmers cited above, suppose the entire population is surveyed and it is found that the quantity of fertilizers purchased by them is between 2 to 4 bags per year, the statement that the small farmers in Meerut buy 2 to 4 bags of fertilizers per annum, is a parameter.
Statistic: It is the value of a characteristics obtained from a sample of the population. It merely provides an estimate of what would be the populationâ€™s parameter with regard to the specified characteristic. Normally, the statistic will vary somewhat from the parameter due to errors associated with sampling. Suppose in the above cited example, instead of a survey of the entire population, 10 percent sample of the farmers are surveyed and it is found that the quantity of fertilizers purchased by them falls in the range of 3 to 4 bags, this value is a statistic.
Precision: It is the degree of closeness between a statistic and a parameter. Precision is the direct outcome of two qualities of a sample: (i) its representative ness, and (ii) its stability. Representative ness denotes to what extent the sample is close to the population. The sampling method is supposed to ensure that the sample is a fairly accurate cross section of the population. Stability of the sample is related to the sample size and proportion of the population included. Representative ness and stability of a sample together decide the precision of the sample.
Developing the sample design
A sample design has the following components:
Choosing the sample unit (who are to be surveyed)
Choosing the sample size (how many to be surveyed)
Choosing the sample procedure (how to ensure those who are to be into grated are included in the sample)
Choosing the sample media (how to reach the respondents in the sample- through mail interview, personal interview, or telephone interview)
Different methods can be employed to select the sample units. These methods, termed as sampling methods, fall under two broad categories:
Probability/ random sampling methods
Non profitability sampling methods
Profitability/ Random sampling:
In probability sampling methods, the sample units are selected at random. The term random should not be understood as haphazard or arbitrary. Instead, it must be understood as selecting the units â€˜free of biasâ€™. When samples are chosen in an arbitrary manner, they are full of but often they are unaware that they are biased. Random sampling follows a precisely specified system where there is no scope for any biased selection of the sample units. Randomness ensures that the selection of the units takes place by sheer chance. It means that every member of the population has an equal chance of being selected.