So I read this about representative samples and I think what I want is to accurately portray the views held by "Europeans" on a certain topic. How many people need to be included in my representative sample?
I'm thinking of research like this, portrayed in the media like this.
The stack exchange question you linked answers this for you. Representativeness is not really related to sample size.
You can have a sample size of 50 million but only sample people in the UK but that won't be representative of all of Europe.
You could ask 1000 people from every country but this won't be representative because (assuming that different countries have different views) the survey has an equal number of participants from Germany and Ireland even though Germany has a far larger population.
On the other hand, you could ask one person for every 10 million in each EU country, this would balance the survey according to populations and it would be representative according to how nationality influences view. This would only survey about 30 people so the small sample size will mean that you cannot be confident in your results. The lack of confidence comes from random variation of views within the population of each nation, this is a source of error but it isn't biased in favour of one view or another so it is still representative.
