You could compare election opinion polls to penalty shoot-outs at a World Cup final: there’s huge pressure to get it right and we remember the big misses most of all.
We are often presented with surveys that claim to show how we all think on a certain subject. But what sample size is big enough to make the findings credible?
What can we draw from the results of the same-sex marriage postal survey if only four out of five eligible Australians took part? Scott Sisson and Peter Baker crunch the numbers.