Sample representativeness expresses how closely the group studied corresponds to the whole population about which conclusions are to be drawn. The most common mistake is not a small sample but a biased selection: only the most satisfied customers fill in the questionnaire, only those with spare time respond to the survey, only visitors from a single channel end up in the test. The result then accurately describes that particular group but says nothing about the rest. In company surveys, it is therefore worth tracking who was approached and who actually responded, and comparing the composition of the sample with the composition of the customer base. A small but correctly assembled sample produces more usable conclusions than a large number of responses from a one-sidedly selected group of people. In qualitative interviews there is no generalising at all; instead, recurring patterns are sought.
See also: Customer survey, Statistical significance of a test, Data sampling.
