oversampling

views updated

oversampling A departure from simple random sampling, in which the same sampling fraction is applied to all cases in the population being studied, so that an invariant one in 50 or one in 200 (for example) are selected for interview. Oversampling is the selective application of a higher sampling fraction to rare sub-groups of particular interest in the population studied, so as to ensure that the final sample includes a sufficient number of these rare cases to permit separate analysis. For example, ethnic minority groups are often oversampled, as compared with the sampling fraction for the majority ethnic group, to enable researchers to analyse the results for each group separately. When summary results from the complete survey are reported, sampling weights are used to restore the original importance of sub-groups in the total population.