DOI: 10.5176/2251-1865_CBP16.56
Authors: R Glynn Owens and J R Oyede
Abstract: Perhaps one of the first things learned by students of statistics is that it is not possible to generalize from a small sample or single individual. An implied corollary is that generalization is best from a suitably large and representative sample. The present paper argues the case for improved generalisability, under certain circumstances, from small samples than large ones. The key issues underlying the argument are the impracticability of generalizing from a group to an individual, and the failure to distinguish statistical and scientific generalization. It is concluded that, where practical, a small sample experimental approach may be scientifically preferable to a large sample statistical one.
Keywords: Small-N; refutability; randomised controlled trials; scientific generalisation;experimental design
Updating...