Stratified sampling reduces the amount of sampling error.

Cluster sampling, where clusters of units are sampled and each person in each chosen cluster is questioned. This type of sampling occurs frequently in industrial or organisational research. For example, a random sample of firms might be chosen, and each person in the sales force questioned.

In extensive sampling, where clusters, say on a national basis, a multi-stage sample would be necessary. For example, the researcher might take a random sample of states in America, from each of those states chosen a sample a sample of cities and towns may be taken, from each town a sample of neighbourhoods, and from each of those a sample of households. Using the zip code this would not be too difficult.It must be noted that the error could be increased at each stage.

A probability method requires the individuals who constitute the sample to be pre selected, before the fieldwork takes place.