http://controlfreak.net/2006/12/18/statistical-significance/ a survivor's guide Mon, 30 Jan 2012 18:57:38 +0000 hourly 1 http://wordpress.org/?v=3.3.1 http://controlfreak.net/2006/12/18/statistical-significance/comment-page-1/#comment-61079 Maggie Sat, 20 Nov 2010 01:58:17 +0000 http://controlfreak.net/2006/12/18/statistical-significance/#comment-61079 But this completely ignores the bias due to the fact that (a) people who feel strongly about the issue or who have time to reply or believe in the concept of polling will actually reply and (b) polling inevitably reaches only certain areas or groups of people. Personally I hate it when I see polls quote a "3 percent margin of error". That may be statistically correct according to Poisson statistics, but it is still complete B.S. to use that poll for policy making. All it means is that they polled 1000 people--who could be all from a Republican district, or all living in suburbia, or all in some way NOT representative of the population as a whole. But this completely ignores the bias due to the fact that (a) people who feel strongly about the issue or who have time to reply or believe in the concept of polling will actually reply and (b) polling inevitably reaches only certain areas or groups of people. Personally I hate it when I see polls quote a “3 percent margin of error”. That may be statistically correct according to Poisson statistics, but it is still complete B.S. to use that poll for policy making. All it means is that they polled 1000 people–who could be all from a Republican district, or all living in suburbia, or all in some way NOT representative of the population as a whole.

]]>