Infoskripsi arrow Reference arrow Sampling Techniques: Cluster Sampling




   


Sampling Techniques: Cluster Sampling
Cluster Sampling In cluster sampling the unit of sampling is not the individual but rather a naturally occurring group of individuals. Cluster sampling is used when it is more feasible or convenient to select groups of individuals than it is to select individuals from a defined population. This situation occurs when it is either impractical or impossible to obtain a list of all members of the accessible population. Suppose, for example, that one's defined population consists of all residents over the age of eighteen in a particular city. Simple random sampling or systematic sampling could be used if an up-to-date, complete census of all the city's individuals and their ages is available. If not, then cluster sampling is advisable. The city might be divided into areas containing 16 square blocks. Each area would be listed and numbered, and the areas to be sampled would be drawn at random. All individuals who meet the age requirement in each sample area would be studied, excepting those who cannot be reached or who are uncooperative. Thus, the unit of sampling is a 16-square-block area rather than the individual citizen. Multistage cluster sampling is a variant of cluster sampling. Once the square block areas have been randomly selected, the researcher can further reduce the sample size by only studying a random sample in each square-block area. For example, the researcher might list the addresses of all houses in the area and then study the residents of ten randomly selected houses in each 16-block area included in the sample. In essence, multistage cluster sampling consists of two or more cycles of listing and sampling. Several sampling stages or cycles may be carried out in order to arrive at the subjects to be ultimately included in the sample. Cluster sampling is sometimes used in educational research with the classroom as the unit of sampling. Suppose that one wishes to administer a questionnaire to a random sample of 300 pupils in a population defined as all sixth graders in four school districts. Let us say there are a total of 1250 sixth graders in 50 classrooms, with an average of 25 pupils in each classroom. One approach is to draw a simple random sample using a census list of all 1250 pupils. In duster sampling, though, one would draw a random sample of 12 classrooms from a census list of all 50 classrooms. Then one would administer the questionnaire to every pupil in each of the 12 classrooms. The main advantage of cluster sampling is that it saves time and money. The use of this sampling technique enables one to confine questionnaire administration to 12 of the 50 classrooms. If simple random sampling were used one might have to arrange for access to all 50 classrooms, even though in some of these classrooms the researcher might have selected only one student for the sample. Disadvantages of cluster sampling are (1) that it is less accurate than simple random sampling because in simple random sampling there is only one sampling error while in multistage sampling there is a sampling error at each stage, and (2) one cannot use the conventional formulas for computing statistics on one's research data. Also, the statistics are less sensitive to population differences. Nevertheless, these disadvantages must be weighed against the considerable savings in time and money that can result from using cluster sampling. Besides the sampling techniques described above, which are sufficient for most educational research, a variety of more sophisticated techniques are intended primarily for use in large-scale survey research. See the annotated references section of this chapter for sources that deal with these procedures.