A sample is biased when it is collected in such a way that not every member of the population it intends to sample has an equal chance of being included. A sample is representative (= unbiased) when the frequency of members in the sample is the same as that of the population it is sampling.
For example, a telephone survey sample is found to include 60% women and 40% men, when the population ratio is known to be 52% women and 48% men. This sample is biased, oversampling women and undersampling men. As the population and sample patterns are known, this bias can be corrected by reweighting the sample to fit it to the population pattern, multiplying males by 48/40, and females by 52/60.
Case study research sites are usually selected in a biased manner. Given the strong tendency to select case study research sites based on their accessibility, interesting locales, and prior work by other researchers, samples composed of existing case study research sites tend to include these biases.
Geographic bias is the tendency to select research sites for studies in a way that precludes or reduces selection in some geographic areas. For example, there is a known temperate zone and wealthy nation biases in the selection of field research sites for ecological studies (Martin et al. 2012).
Martin, L.J., Blossey, B. and Ellis, E., 2012. Mapping where ecologists work: biases in the global distribution of terrestrial ecological observations. Frontiers in Ecology and the Environment 10 (4), 195–201. [download pdf]