#4 A relatively small sample size, yes, on the order of 1000 per country, could be indicative IF the samples were balanced in each country to match. So, if each sample had the same number of, say, middle class professional married males aged 35-50, females ditto, unmarried blue-collar males aged 20-34, females ditto but including pink-collar workers, etc, then the results could be considered valid. There is a lot of statistical science in putting together a sample pool -- the art lies in constructing it to provide the bias you want (in the unlikely event that bias is desired -- but of course biased results lead to faulty conclusions, which everyone knows is a Bad Thing). |