The main theme that is associated with the sampling is to evaluate the large study area through testing the limited representative material that would then serve to explore and create trends in the area under study. The material that is taken for testing is referred to as sample and its distribution and proportion would be such as, it covers all the properties of the discipline or the area under study (Taggart, 1945). Certain methods of sampling are developed, that are executed according to the type of sampling being carried out.
The preceding context would elaborate various types of sampling, which includes
• Simple random sampling
• Systematic sampling
• Statistical surveys
• Stratified sampling
• Cluster sampling
• Multistage sampling
• Non-probability sampling
Simple Random Sampling
The simple random sampling refers to the sample selection randomly or purely by chance, without having any preference or favor for that sample at the expense of ease of handling or in other words without biasness. In such samples there is an equal chance for each entity under study to be selected as a sample, representing the entire study area or a population. Statistically sample obtained from SRS is actually the subset of the individuals, population or the area selected for study (Yates et al., 2008). However it would be note worthy that Random sampling is different from the simple random sampling.
In simple random sampling the researchers avoid replicating the chosen area being chosen again however; logically it could be done with replication or replacement. When the sampling is unbiased the sample remains authentic for the population over a long period of time; though, in fact the maximum precision is not the guaranteed and perfect representation of particular population, so the valid conclusions are hard to formulate about that population. But still the simple random sampling is easy to carry out, uncomplicated to interpret, and simple enough; without having conventional errors in that. This technique of sampling is best suitable when the data about the population is entirely absent.
This statistical method of sampling involves the systematically ordered frames for selecting the individuals or the elements. Equal probability method is the most widely accepted and used method of systematic sampling (Black, 2004). In systematic sampling the starting point is determined by the random sampling and then the preceding sampling is predetermined, which is carried on systematically. The selection of units is done systematically from the population at regular intervals; the intervals may include, every hour, after two hours, every tenth house etc.
There is a rule of thumb in the systematic sampling that one has to choose the starting point, more preferably by random sampling and then every Kth element from that first K is taken as a sample at regular intervals. In case of random placement of variables, both random and systematic sampling is considered alike.
Statistical surveys are conducted to collect data systematically and the individuals of the samples are analytically treated to extract the information. The surveys are considered to be the best tool for considering the whole population and are commonly employed for taking out the polls of public opinion. However political polling (that is non public polling), scientific surveys, sociological and psychological surveys, marketing research and much more can be inferred under the heading of statistical surveys. The main advantage of statistical surveys is that it covers the vast topics of multiple disciplines; for exampling, the alcoholic behavior of certain age group, voting for the presidential candidate, the psychological issues of working women, income information of certain community etc.
Certain factors determines the quality and reliability of the survey, such as flexibility in communication for answering the questions, the respondents interest to participate willingly, proper exposure of the target population and the cost associated with the project under survey. Statistical surveys provide the maximum output when the participating individuals extend the unbiased cooperation to the surveyor.
The stratified sampling is also a statistical technique to evaluate the maximum data associated with the population. This technique is followed to encounter each of the subpopulation in the continuously varying population. Before sampling there is a prerequisite that the whole population is homogeneously deeded in to the subpopulations, called strata here. The stratified sampling is a bit exhaustive job as each stratum should be examined for each provided element and each individual of each stratum should be accounted for that. However the practice is reliable enough and the error chances are minimized with the less variability as compared to simple random sampling and arithmetic mean. Proportionate and optimum allocation strategies are employed for the stratified sampling. The disadvantage associated with the stratified sampling is that there should be similarity in subgroups elements and the size.
The cluster sample has incorporated a random sampling in it, in a way that the population is clustered or grouped first and then the randomly selected cluster is analyzed. The cluster that is selected has various implications that it should be true representative of the whole population and should be exclusive mutually. The clusters are then analyzed for their units and the outlying units are neglected overall, and all the units are selected for sampling, differentiated from the stratified sampling because each stratum is examined there for its units. The cluster sampling can be separated into multistage, ranging from, one or two to many stages. For more than one stages subsets of units are formed from the clusters making it easy to work with, with the minimum error level. The clusters however are PSU (primary sampling units) and in the SSU (secondary sampling unit category) cluster units are analyzed.
The standard errors are minimized by analyzing the clusters in the same way as simple random sampling (SRS). The cluster sampling is usually preferred being cost effective and easily executable.
Multistage sampling as discusses in the context of cluster sampling is the composite from of cluster sampling. In multistage sampling the clusters that are identified are subgroup into two or several divisions for the elements included in the cluster. The multistage sampling is more advantageous than the cluster sampling because it is more accurate for its results as the same sample is divided further and the survey sample in multistage sampling is easily reachable. But it still lags behind the simple random sampling as far as technicality and accuracy is concerned.
The random sampling is not included in non-probability sampling and cannot infer the generic view for a population from the sample undertaken for study. And if the generalization is made, it is to be verified from reviewing the previous studies ad literature. Although the non-probability sampling is cost effective than that of probability sampling but still it is not as reliable as that of other methods of sampling. It may include sampling for the cases such as case studies, adhoc quotas, purposive sampling and the like.
• Taggart, Handbook of Mineral Dressing, Wiley, 1945, Section 19.
• Allen, Particle Size Measurement, Chapman and Hall, 1981.
• Yates, Daniel S.; David S. Moore, Daren S. Starnes (2008). The Practice of Statistics, 3rd Ed. Freeman. ISBN 978-0-7167-7309-2.
• Ken Black (2004). Business Statistics for Contemporary Decision Making (Fourth (Wiley Student Edition for India) ed.). Wiley-India.ISBN 9788126508099.
• Abramson, J.J. and Abramson, Z.H. 1999.Survey Methods in Community Medicine: Epidemiological Research, Programme Evaluation, Clincal Trials (5th edition). London: Churchill Livingstone/Elsevier Health Sciences ISBN 0-443-06163-7
• Mellenbergh, G.J. (2008). Chapter 9: Surveys. In H.J. Adèr & G.J. Mellenbergh (Eds.) (with contributions by D.J. Hand), Advising on Research Methods: A consultant’s companion (pp. 183-209). Huizen, The Netherlands: Johannes van Kessel Publishing.
• Sarndal, C.E., Swenson, B., and Wreman, J.H., Model Assisted Survey Sampling, Springer-Verlag, New York, 1992.