undefined

I don't think most polling qualifies as statistics, without providing info like selection criteria or number of respondents. Otherwise we'll walk right into another situation like when Trump won an election while polling in near last place.

Or maybe I want something a little more significant than "we asked around the newsroom, and 35% of fox news reporters approve of the current administration". Cool, thanks, but that's not helpful or even a useful datapoint.

by verteu2 hours ago|

parent|

[-]

> I want something a little more significant than "we asked around the newsroom, and 35% of fox news reporters approve of the current administration". Cool, thanks, but that's not helpful or even a useful datapoint.

The state of the art (which confirms this ~35% approval number) is far more rigorous than that.

Start with a meta-review of historical accuracy and transparency, eg: https://www.natesilver.net/p/pollster-ratings-silver-bulleti...

Details on methodology and sample size can be Googled, eg: https://poll.qu.edu/methodology/ , https://www.washingtonpost.com/news/the-fix/wp/2015/07/24/ev... , https://www.google.com/search?q=Marquette+University+Law+Sch..., etc.

Sample size is less important than you might think, due to https://en.wikipedia.org/wiki/Binomial_proportion_confidence...

by 2 hours ago|

parent|

prev|

[-]

deleted