Perhaps a bit off topic, but: I don't think your market research observation is all that true. Of course, we should all take survey based results carefully. But (having formerly worked in market research) from what I've seen, representativeness is an issue those folks are VERY well aware of and do try to address. And who are you saying should be aware? The consumer (in which case a bad survey wouldn't affect, except if they stop making Choc. PB based on the survey and your a Choc. PB lover) or the corporate client of the market research firm (well ok, but at worst they're contributing $$ to the economy)? Which brings up (off topic) an appearant disdain towards statistics in commerce, from some "real" statisticians. IMO, more statistician should embrace it and try to weasel/attract more "true" statisticians into business. Surely, this would help attract more people into stats and hence more funding and hence, more big fat endowed chairs in the Stats dept. at XYZ university and hence higher salaries for all of us who read sci.stat.xxx! If I'm just knocking down a straw-man, well, I'm glad I'm wrong.
-Iyue "Carl W." <[EMAIL PROTECTED]> wrote in message news:<[EMAIL PROTECTED]>... > "Gary Klass" <[EMAIL PROTECTED]> wrote in message > [EMAIL PROTECTED]">news:[EMAIL PROTECTED]... > > http://lilt.ilstu.edu/gmklass/COW/ > > Does this prove that TV makes you bad at math or that if you're bad at math > you tend to watch more TV? > > This is a bit like the fact that market researchers only glean information > from people who are willing to take part in market research... Consider: > > What if a propensity to say 'Yes' to 'Do you want to take part in a survey?' > would also make you like (say) Chocoloate Peanut Butter more? > > Given that a small number of people actually take part in market research, > The Fictional Peanut Butter Corporation is going to start manufacturing > Chocolate Peanut Butter based on their survey, and then find no (or not > many) takers in the 'real world'. > > Beware statistics, and more importantly, anything based on market research > or a survey. > > Okay, rant over, > Carl ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at http://jse.stat.ncsu.edu/ =================================================================
