The whole point was that you were basing your assumptions based on a very small select data set which is not typical at all. I was just trying to point out that extremes can be found in either way and 90+% market penetration is not a figment of imagination but can be found in most stats.
Actually, the hotel may be a small select data but it offers data from many nationalities and that's important. Greeks, Americans, Russians etc (the Americans almost exclusively use Macs)... Afterall someone doesn't need to be able to pay for a room in a 5 stars hotel to be able to afford an iphone or a macbook.
Also, don't forget that I mention simple guests who use our wi-fi for surfing, no the proffesionals who come for meetings and conferences.
My point is that if you remove companies, universities, colleges and their office computers then we have a large percentage of Apples in the rest.
I would argue that all of these datasets are relevent as long as one takes them with the appropriate grain of salt. Everyone I talk to feels that Linux usage has been increasing significantly over the last couple of years, but there are still no credible numbers to support that perception, at least not sufficiently credible to alter the allocation of development funding.
These datasets won't be the answer, but they all might provide clues for where the answer can be found.
I would argue that all of these datasets are relevent as long as one takes them with the appropriate grain of salt.
Yes, the main reason why I pulled numbers from my university was that I believe the "what OS do people use" is an even worse question than "which party do people support", as in that it all depends on which place you pick. Pick some annual Republican meet and do a gallup on which party they support and I'm fairly sure you can draw the conclusion Democrats don't exist or are a statistical error.