Multi-phase sampling is a type of sampling design in which required information is collected from a large sample of units, and additional information is collected from the sub-samples of the whole sample either at the same time or a later stage. With only one sub-sample, the design is called two-phase or double sampling.
Double sampling is used in some situations where making use of an auxiliary variable is desirable. Still, the necessary information for stratification or the application of ratio or regression estimator is not available.
For example, let us consider the problem of estimating the total consumer expenditure in a particular city through a sample survey when the available information is only a list of all households in the city.
If it is decided to select a sample of households, this might involve an extensive sample, and hence the cost involved may be substantial and prohibitive.
An alternative procedure may be thought of as selecting a large preliminary sample in the first phase to collect information on characteristics such as household size, occupational status, etc. and this information may be used for stratification, selection, and estimation procedures.
In the second phase, a sample from the large preliminary sample is selected to observe the main character under study.
Multi-phase sampling should not be confused with multi-stage sampling, although the two appear to be the same.
In multi-stage sampling, different types of sampling units are sampled at different stages, while in multiphase sampling, we are concerned with the similar type of sampling units, at each phase. Still, some smaller number of units are asked for more information than others.
The usual pattern for multi-phase sample surveys is to survey in the first phase.
- Simple characteristics that can conveniently be studied on a large-scale,
- Infrequent characteristics which call for large samples or
- Summary characteristics for a preliminary study to be followed by more intensive studies on a smaller scale.
In the second phase, a sub-sample of the first phase sample is usually surveyed for a study of
- More complex characteristics related to those surveyed in the first phase,
- Details of the infrequent characteristics identified in the first phase, or
- Details of the summary characteristics studied in the first phase.
The information collected in the first phase can be used with advantage for the stratification of the sampling units and selection of appropriate subsamples. The information can also be used for the identification of target groups to be studied in the second phase.
It seems intuitively convincing that the collection of some information by the sample in the first phase and then using this information as the basis for selecting a sub-sample for further study may prove more convenient or economical.
The point in support of this is that when all information from all respondents is redundant, there is little justification to add additional burden approaching everyone for everything. This saves time and money.