Quote:
Originally Posted by spiderplant
Back to stats school for you! 5,100 is a huge sample, regardless of the population size.
Those were opinion polls, asking people how they plan to vote. They could lie or later change their minds. But BARB report on what people actually did watch.
|
Indeed, political polling often has a weighted “how likely are you to vote?” and has “don’t know”.
A closer comparison to BARB would be a UK General Election exit poll which, with the exception of 1992, is usually pretty decent considering it’s essentially 660 individual predictions and has to try to account for regional variations.