Abstract：This article sets out what can (and what cannot) be inferred from the unrepresentative samples of the Chinese population that are the basis for most survey research conducted in mainland China. Clearly, a sample of Chinese interviewed in a locality or localities selected for convenience does not reliably represent a larger population of interest. This article shows that, in principle, there is no reason to think of local samples as "case studies". While data from local samples cannot be relied on for descriptions of a larger population along any single dimension, this unrepresentativeness does not necessarily affect generalizability of findings about relationships between variables.
