In a statistical population we can divide the population into finite distinct and identifiable units which are called as sampling units. The smallest units which the population can be divided are known as elements. We can create groups of elements and call them clusters.
Clusters are constructed in such a way that within the clusters the sampling units are heterogeneous but they are homogeneous among the clusters. The clusters can be constructed with equal or unequal sizes.
Cluster sampling is unbiased when the clusters are of same size. When the cluster sizes are different the probability proportion to the size is to be used such a way that the larger clusters have greater probability than the smaller one.
There are two cluster sampling methods
1) One Stage Sampling:
All elements of the cluster are used for sampling then it is called one stage sampling.
2) Two Stage Sampling:
If we select the random sample from the one stage sampling then this is called the two stage sampling.
• Divide the population into finite clusters using an well defined rule
• Each cluster would be treated as a sampling unit.
• Select the cluster
• Collect/enumerate all the details you are looking for from the selected cluster.
• The cluster sampling is use when there is no reliable list or information is present.
• List is present but enumerating the required information is difficult due to location and distance.
• The cost and time of collecting the information by way of interviews, where one has to spend long time at place or travel a lot.
• It is time and cost efficient especially involving large geological areas.
• Easy to handle bigger samples by breaking it into clusters and with better precision.
• One need to collect more information
• Sampling error is more compared to the other sampling methods
• Say you live in a large city and to list out each person in the city is a tedious task. We can select small areas in the city as clusters and person in them as elements. Sampling from few area is then enough for a cluster sampling process.
• Cluster sampling can be used as a tool to estimate mortalities during natural calamities like earth quakes, floods and famines.