Out of statistical context, you are absolutely correct, Quoll -- no argument there. I would hope it wouldn't be taken out of statistical context. Then again, I am posting it in the poser community, aren't I? lol As to the specific use of the terms "Effective Sample Representation" and "Effective Users" -- the terms simply mean nothing more than that's the end sum as it's expressed out. Based on the statistical probability of like answers, those are the "effective sum totals". That's all. Nothing funky about them. In fact, until I read your comment, Quoll, I hadn't thought about the wording -- it's a standard term. In further breakdowns as this continues I will seek to use "estimated" if all think that would work better. Statistically speaking, 259 respondents from the base 67,000 (which was a number I pulled from a two year old press release from the former CL owners online -- questionable, but it's the only one I could find. I would really, really, really love to have a better number) is valid as a sample group. This is what's known as a trend analysis, and there is a reason that anal is the first part of analysis. Since the repondents are not required to divulge additional information which enables you to break it down more effectively, the data itself is technically incomplete. A more sweeping survey would need to be done to determine a greater number of socioeconomic factors for a more accurate curve. However, I think more to your point is that this isn't really something you can use to make serious business decisions on -- I don't recommend it beyond possibly considering support such has Cris has done, but, to be frank, I'm not certain it will make much difference in the short term. Cris, the reason statistics work for large groups of people is because, as "groups" folks tend to behave in similar patterns expressed as commonalities. in other words, people who have similar situations, when acting as groups, tend to do the same thing, even though as indiviudals they may be utterly different. Folks don't particularly like that thought, but there's a pretty sizeable and incredibly effective marketing machine out there based on it. Not to mention insurance. Note, I am absolutely not trying to "prove" anything here. I just think it's cool. I didn't do this for market research puproses -- I did it because 1- I'm crazy and love doing this stuff for fun, and 2- I'm a total fangirl who just loves this stuff. Poser, DAZ, rosity -- the works. This stuff is just plain cool to me. I might bitch about one and whine about the other, but, in the end, I love it all. To get a good ratio that becomes a valid sample, you have to achieve a sort of pairity to where the number of people polled more or less is equivalent to the number of people they represent. In this case, that number worked out to be around 259. My initial guess was actually around 300 -- as I'd posted already. Note that there are some pretty startling differences between the Unadjusted and the Curve percentages. The curve percentage reflects as best as one can that "non-aggressive" user base. The typical curve for software tends to favor the immediate version prior more heavily than the current version, decreased according to the length of time between versions. I dropped a couple of curve tests for software in there. The one I ended up using was derived from some old photoshop data. There was one derived from OS data that might also have been applicable, but I thought the curve was more bellshaped -- the OS one added about a 15% deduction across the board and assigned it all to the P3 and earlier versions. And, lastly, it is important when looking at ANY list of statistics, to keep the error percentage in mind. The one noted here is pretty darn big. The analysis of the results, well -- that's for ya'll to argue about and throw eggs at me for :) hey -- I did flag it for violence...

thou and I, my friend, can, in the most flunkey world, make, each of us, one non-flunkey, one hero, if we like: that will be two heroes to begin with. (Carlyle)