XPost: rec.arts.sf.written, rec.arts.sf.science   
   From: bap@shrdlu.com   
      
   On 24/01/14 00:21, Your Name wrote:   
   > In article <1xxsbf7bzxt8w.1244utm9kehz9.dlg@40tude.net>, Brian M. Scott   
   >  wrote:   
   >> On Fri, 24 Jan 2014 09:27:46 +1300, Your Name   
   >>  wrote in   
   >>  in   
   >> rec.arts.sf.written,rec.arts.sf.science,rec.arts.sf.misc:   
   >>   
   >> [...]   
   >>   
   >>> The problem is "surveys" are never accurate because they   
   >>> simply ask far too few people [...]   
   >>   
   >> On the contrary, sample size is rarely a problem.  The   
   >> problem is getting a simple random sample.   
   >   
   > Few surveys are actually random. Idiot companies like Neilsen pick and   
   > choose who can and can't be surveyed in the pretense of obtaining a   
   > "statistically equivalent" ratio of people, which in itself means it's   
   > never going to be random nor accurate.   
      
   You clearly know very little about statistics in theory or practice.   
   Neilsen's methods are externally audited and have to be based on sound   
   statistical techniques.   
      
   >   
   > Sample size is also a big problem. There's no point asking 50 people   
   > out of a population of 50 million ... the results are useless, and the   
   > way such results are reported is even worse.   
      
   Of course. Which is why organisations like Neilsen include sample sizes   
   in every report they produce. The figures that get reported in the press   
   are not the whole story.   
      
     The survey is ONLY ever   
   > going to be accurate for those 50 people. Anything else supposedly   
   > "proven" is pure gueswork and manipulation, and often (especially in   
   > marketing) statistically manipulated to "prove" wahtever the people   
   > paying for it want it to "prove".   
      
   That's why in the advertising industry the statistical methods used get   
   approved by a cross-section of the industry, some wanting higher numbers   
   and some lower. In the UK there is a Joint Industry Committee for each   
   advertising medium. Other regions have their own equivalents.   
      
   (If you Google for Joint Industry Committee you may well come across my CV.)   
      
   --   
   bap@shrdlu.com   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)