Daniel Rowan 5 Years Ago Great article Tiffany, thanks for writing it. In general it is disappointing how little awareness there is generally with the public about dealing with (or even expecting) uncertainty in published statistics, I think in part this may be down to how often indicators of data confidence are communicated by journalists and politicians (I think never? - some sort of regulatory role for the ONS in policing statistics that may be given/manipulated at press conferences or, say, written on the side of a bus would be great!). I deal a lot with consultation data, recently I was asked by a councillor about a survey which our team was designing: "So how many responses before it's statistically significant?". I talked to him about how significance is a broad concept and how it really depends on how confident you want to be with the probabilistic assumptions you wish to make about the wider population, and how there is a relationship between confidence intervals and sample size, but that it is non-linear (for instance you cannot half a margin of error by doubling the sample size). Then I saw his eyes glaze over and just before he fell asleep he said. "Great... so how many responses?" Reply 1 Reply as... Cancel