QUOTE(Moulton @ Sat 8th May 2010, 1:34pm)
Pollsters vary their sample size based on a variety of considerations. Bigger samples are more costly to obtain, but yield better accuracy. For some purposes, a sample size as small as 30 suffices, where precision to multiple significant digits is not needed. It all depends on what you are planning to do with the statistics.
A sample size of 30 has a far greater chance of having significant problems than a sample size of 3000, and this pollster took samples closer to 30 per country than 3000 per country.