1. A sample may be constructed which is considered large enough, and representative enough, of a wider population to allow general findings from the research to be likely (in a statistical sense – using inferential statistics, tests designed to allow inferences to be drawn) to reflect the wider population. This approach is used in surveys (see below).
2. In experimental approaches there may be less control over the make-up of the sample of “subjects”. However, by identifying the potential factors that may very in a population and be relevant (think of the assumptions that need to be made here), it is possible to try to allow for these factors.