Skip to main content

Sampling method

(go to Outline)

Have the survey managers clearly defined the sampling universe? That is, have they described the population from which the sample was chosen? This must be clearly defined in order to know: a) who is eligible for inclusion in the survey sample, and b) to what population the results of the survey can be generalized.

Was the sampling frame out of date and if so, was it adjusted in any way? The sampling frame is the list of units from which the sampling was done. In cluster surveys, there may be a sampling frame consisting of a list of villages from which 30 are selected as clusters. If the list is old, some villages may be missing or other villages may have incorrect populations. This could introduce sampling bias.

Is the sampling method described in enough detail? Could you, as the reader, spot mistakes and potential sources of sampling bias? The description of the sampling, and all other aspects of the survey methods, should be detailed enough so someone else could repeat the survey relying only on the survey report.

Just stating in a survey report "We selected a random sample of households in the refugee camp" is insufficient. Exactly how was this done? Walk through the camp and stop at any house they liked? (This is an unacceptable non-random convenience sample.) List all the households, assign each a number, and select them using a random number table? (This is excellent simple random sampling) (see section on Sampling methods)