Types of sampling design
The most common sampling designs are:
- Simple random sampling (SRS): The sample is randomly selected from a comprehensive listing of elements.
- Stratified survey with SRS: Elements are divided into two or more mutually exclusive geographic areas (called strata, singular is stratum) and, within each stratum, a sample of elements is randomly selected.
- One-stage cluster survey: Primary sampling units (PSUs) are selected from a sampling frame, and all elements within each PSU are sampled.
- Multi-stage cluster survey: Primary sampling units (PSUs) are selected from a sampling frame, and a sample of elements is selected from within each PSU. Where the selected elements are households or schools, there may be a third level of selection, namely, of individuals (as elements) from within the households or schools.
- Stratified multi-stage cluster survey: Elements are divided into two or more mutually exclusive geographic areas (strata) and, within each stratum, a two- or three-stage cluster survey selection of elements is implemented.
Population-based micronutrient surveys are often based on a multi-stage cluster design, usually stratified. In large-scale surveys, a multi-stage cluster design is a much more cost-effective method of data collection than simple random sampling (SRS). It also works well in situations where reliable nationwide household listings may not be available. The first stage of sampling, the selection of geographical units (primary sampling units, or PSUs), allows the development of lists of households, sometimes including a listing of individual household members, within the selected PSUs. In the second and third stages, samples of households are then randomly selected from those lists and, depending on the survey design, individuals may then be selected from within the identified households.
The survey design will affect how data analysis is done. More information on this can be found in Module 15, Data processing and analysis.
This manual focuses on multi-stage cluster and stratified multi-stage cluster designs. An SRS design may be relevant for assessing micronutrient status in relatively small, defined areas, for example, a refugee camp, therefore some information on sample size calculation using SRS is briefly included in Module 5, Sample size.
Box 4.1. Definitions related to sampling design
- Sample: the subset of elements drawn from the frame (the survey sample).
- Element: the basic unit that represents whatever is being sampled and from which survey data are to be gathered. For health-related surveys, this is frequently the individual or the household.
- Primary sampling units (PSU): the sampling units that are selected in the first stage of a multi-stage sample. The PSUs may be an area with clearly defined, non-overlapping geographic boundaries, for example villages in rural areas and wards or blocks in urban areas.
- Frame (sampling frame): a listing of PSUs, or of elements within a PSU, from which the sample is drawn.
- Strata: Groups into which elements are divided for stratified sampling (for example regions: North, South, Central).