With cluster sampling, the population is split into separate groups, called clusters, and simple random sample of clusters is selected from the population.
Whereas in stratified sampling each stratum tends to be more homogeneous, this is not the case for cluster sampling. Stratified sampling also differs in that a sample is taken from every stratum.
For cluster sampling, only some of the clusters are taken. In the case of single stage cluster sampling, the entire cluster is sampled. For multistage cluster sampling, random samples are selected within the chosen clusters in one or more stages.
For the same sample size, while stratified sampling may reduce sampling error, cluster sampling increases it.
On the other hand cluster sampling saves cost. The process reduces the cost per respondent especially if travel costs between clusters is high, so for the same cost, it may result in lower error.
Ideally, variation within strata should be small (homogenous), while variation within clusters should be a large, though this usually not within our control.
In practice the population is divided into clusters (for instance different towns), the clusters are grouped into strata, and a cluster sample is then taken from every stratum.
To see how this works, consider the urban India household panel which used to be the largest panel of its kind in the world. Set-up by Hindustan Lever, the panel was configured by splitting all Indian cities and towns (clusters) into groups based on size and geographical location. A selection of about 20 of these clusters was made covering small, medium and large urban centres, across north, south, east and west. The selected towns/cities were then split into blocks. These blocks were stratified according to variables like household income and household size, which strongly influence consumption behaviours. The panel was formed by randomly selecting homes from the chosen urban blocks such that all strata were adequately represented. This was the third and final stage of sampling. (First cities, then blocks and finally homes).
To conduct a national survey of individuals or households, in China, US or the UK, the fundamental approach would not be dissimilar to that described above.
Note: To find content on MarketingMind type the acronym ‘MM’ followed by your query into the search bar. For example, if you enter ‘mm consumer analytics’ into Chrome’s search bar, relevant pages from MarketingMind will appear in Google’s result pages.