Upgrade your account to a Community Supporter account and remove most of the site ads.

Reply to thread

Message: <blockquote data-quote="nothing to see here" data-source="post: 3148091" data-attributes="member: 16432">Sorry to bore you hereRaven raises a good point -- many households have one member who answers the phone far more frequently than anyone else...fortunately there are ways around that.  Most professional research companies ensure a degree of randomness within a household as well.  For instance a common, easy to implement trick is to ask for the adult member of the household who's birthday falls closest to (random month) 1st, to participate in the survey as well.  And keep in mind that -- for most consumer purposes 'household's' aren't the units of measurement that people are interested in*.  Household's don't buy game products, individual consumers do.  As long as the person answering the phone is a randomly selected repreentative of game-buyers in general...your research is rocking and rolling.On that point, it is also worth noting that the whole sample-size = confidence intreval notion does not apply in it's entirety here.  The math behind statistics only works 'cleanly' assuming a completely random sample of a given population.  Complete 'randomness' is next to impossible for any population, but for wide scale research (i.e. political voter intention research) you can get pretty darn close.This is not the case here, however.  Since the 1000 people sampled in WOTC's market research sample were nothing more than a (prbably random) selection of the 20,000 (or fewer)households that responded to the inital post card survey, then an important staistical rule regarding 'probability' of survey accureacy was broken -- namely that the population being sampled was not random, but self-selecting.  This does not invalidate the research, however.  The leap of faith comes from your confidence that the 20,000 gamers contacted by the post cards are representative of everyone who plays D&D.  Charle's Ryan's point is very accureate...using this or other fcustomer feedback mechanisms to create a sample-able sub population of the overall buying public is quite common as well as, usually, very accureate.  Quite frankly, it is hard to to demand random sampling of the entire buying public in almost any industry (though the growth of the virtual marketplace will make that easier as time goes on).  For one, you have to do some arcane statistics-fu and with regard to weighting your sample so that it corresponds, demographically, to the actual buying public.That's probably more about modern research than you want to know.  Sorry.  It's not exactly shooting fish in a barrell, but neither is it all Smoke and Mirrors either.*one exception remains Nielsen and other TV ratings systems where what is watched on TV is very much a communal 'household' choice.</blockquote>

Verification