Institutional Learning and Change Initiative

The following text is drawn from Guijt and Woodhill (2002).

Random sampling

The purpose of this method is to produce a sample, without any prior knowledge or consideration of particular characteristics, that can be considered to be representative of the primary stakeholders being affected by a project intervention. From an M&E perspective, the sample is needed to guide the use of information-collection methods.

Random sampling is more often used in larger-scale M&E analyses than non-random sampling. However, random sampling is not always possible or practical if, for instance, there is not enought time to make up a complete list of the information needed. Existing census records, electoral lists, telephone books or other records should be sought out and used whenever possible - but be aware that these may not always be accurate.

Nevertheless, random sampling is not useful when dealing with a very small sample size, since it is unlikely to be representative enough and therefore not able to provide accurate conclusions about the whole population.

Source: Guijt, I. and J. Woodhill (2002). Managing for Impact in Rural Development : A guide for project M & E. Rome, Italy: International Fund for Agricultural Development http://www.ifad.org/evaluation/guide/index.htm