Stratified Sampling vs Cluster Sampling
In statistics, especially when conducting surveys, it is important to obtain an unbiased sample, so the result and predictions made concerning the population are more accurate. But, in the simple random sampling, the possibility exists to select the members of the sample that is biased; in other words, it doesn’t represent the population fairly. Therefore, stratified sampling and cluster sampling are used to overcome the bias and efficiency issues of the simple random sampling.
Stratified Sampling
Stratified random sampling is a sampling method in which the population is first divided into strata (A stratum is a homogeneous subset of the population). Then a simple random sample is taken from each stratum. The results from each strata combined constitute the sample. Following are examples of possible strata in populations
• For a population of a state, male and female strata
• For people working in a city, resident and non-resident strata
• For students in a college, white, black, Hispanic, and Asian strata
• For an audience of a debate regarding theology, Protestant, Catholic, Jewish, Muslim strata
In this process, rather than taking samples at random straight from the population, the population is separated into groups using an inherent characteristic of the elements (homogeneous groups). Then random samples are taken from the group. The amount of random samples taken from each group is dependent on the number of elements within the group.
This allows sampling to be made without the sample of one group being larger than the number of samples required from that particular group. If the number of elements from a certain group is larger than the required amount, a skew in distribution may lead to erroneous interpretations.
Stratified sampling enables use of different statistical methods for each stratum, which helps in improving the efficiency and accuracy of the estimation.
Cluster Sampling
Cluster random sampling is a sampling method in which the population is first divided into clusters (A cluster is a heterogeneous subset of the population). Then a simple random sample of clusters is taken. All the members of the selected clusters together constitute the sample. This method is often used when natural groupings are obvious and available.
For examples, consider a survey for evaluating the involvement of high school students in extracurricular activities. Rather than selecting random students from the student population, selecting a class as the samples for the survey is cluster sampling. Then every member of the class is interviewed. In this case, classes are clusters of the student population.
In cluster sampling, it is the clusters that are selected at random, not the individuals. It is assumed that each cluster by itself is an unbiased representation of the population, which implies that each of the clusters is heterogeneous.
What is the difference between Stratified Sampling and Cluster Sampling?
• In stratified sampling, the population is divided into homogeneous groups called strata, using an attribute of the samples. Then members from each stratum are selected, and the number of samples taken from those strata is proportional to the presence of the strata within the population.
• In cluster sampling, the population is grouped into clusters, predominantly based on location, and then a cluster is selected at random.
• In cluster sampling, a cluster is selected at random, whereas in stratified sampling members are selected at random.
• In stratified sampling, each group used (strata) include homogenous members while, in cluster sampling, a cluster is heterogeneous.
• Stratified sampling is slower while cluster sampling is relatively faster.
• Stratified samples have less error due to factoring in the presence of each group within the population and adapting the methods to obtain a better estimation.
• Cluster sampling has inherent higher percentage of error.