You Â might want to bookmark this article for future reference. Â It draws on a very useful post I hacked from ProfessorÂ Thomas Lumley on the Stats Chat blog:
The 3% â€˜margin of errorâ€™ usually quoted for poll is actually the â€˜maximum margin of errorâ€™, and is an overestimate for minor parties. On the other hand, it also assumes simple random sampling and so tends to beÂ an underestimate for major parties.
Lower and upper â€˜margin of errorâ€™ limits for a sample of size 1000 and the observed percentage,Â assuming that complications in sampling inflate the variance by a factor of 2, whichÂ empirically is about right for National. Read more »