I fully agree with /dmc regarding surveys. I wish to add that surveys are only as good as the sampling method. If the sampling is skewed then results are either skewed (how much is the question) or completely bogus. Volunteers for a survey tend to be bored (unemployed, retired, disabled, stuck at home) or have a burning opinion that probably does not reflect the general opinion of the target group. The type of statistical test is also important. If the incorrect statistical method is used then results are not accurate. The field of statistics is one that a little knowledge is truly disasterous. Sophomore = wise fool, which all too often describes those who conclude drastic "new" information from misapplied statistics, poor research design, or poorly controlled sampling.
I wonder at the stats AristoCraft and LGB provided. LGB seems to claim a smaller market share than purchases indicate in Oregon and California at least. The term "market" could be interpreted very differently. Market can reflect actual purchasers of their products, purchasers of their companies product in the USA, or projections of potential market. Nothing is as simple as we would like. >bagrs annual meeting in which reveal the answers to some basic >questions of five different large-scale scale makers (Aristo, >Hartland, LGB, MTH and USA). the results were completely >inconclusive, but what blew my mind was their individual projections >of the size of the market. >Harland, MTH and USA didn't answer the question; Aristo said more >than 100,000 and LGB said less than 50,000. Gary - Brushing off graduate school statistics concepts instead of running trains over Terror Trestle in Eugene, Oregon http://www.angelfire.com/or/trainguy http://community.webshots.com/user/raltzenthor >a bunch of people answering questions off a web site or off a mailing >list (even one as august as this) is still just a "straw poll" and >has no real meaning. >if one of the manufacturers wants to commission a real survey, i'd be >happy to provide the names of a half-dozen research firms that could >do the job. >/dmc
