Gorgon Zee
Hero
I cannot tell if you are intentionally trying to misunderstand what flaw aramis erak was pointing out ...
They made the statement, which I think you are agreeing with, that the number of events at a con reflects the size of staff support more than the size of player base. I just don't believe that because Gen Con numbers for D&D 5E, for example, absolutely do not support that. But actually, I don't want to focus on that -- I'm happy you have found a new data source, thank you!
http://www.seankreynolds.com/rpgfiles/gaming/WotCMarketResearchSummary.html
This is a survey only on US population between the ages of 12 and 35. The study states that they were biasing it towards their target audience, so *by design* this study is meant to look at people more likely to play D&D than the general population. Their methodology is very good, though, so I am comfortable assuming their results reflect 12-35 year olds in the US for 1999.
The survey gives information on D&D vs. other games by dollars spent:
Total RPG spending by age:
12-17: $297
18-24: $850
25-25: $2,213
Total D&D spending by age:
12-17: $164
18-24: $443
25-35: $1,642
So, if you assume people within an age group spend money on games proportional to the amount you play, D&D has between 50%-70% of the play time. I might argue that people are likely to spend more on D&D than other games (with notable exceptions!) so 55% (for the younger age groups) seems a better figure to go with. That also represents 66% of the survey, so even if we just weight by population, we'd end up with a figure of 60%.12-17: $297
18-24: $850
25-25: $2,213
Total D&D spending by age:
12-17: $164
18-24: $443
25-35: $1,642
The information on "which games have you played last month" give that D&D is played by at least once in the month by 66% of people, and other games are played a cumulative 132% of the time. What we don't know is how much these groups overlap. At one extreme, we could assume every single D&D player plays only D&D -- complete disconnect. In that case D&D has 66% of the play time and the other players play different non-D&D games, but only comprise 33% of players. The other extreme is that people are completely mixed; all players play games with the probabilities as given. In that case D&D would have (66 / 198)% == 33% of the play time.
Both those scenarios are extreme. So the answer lies somewhere in the middle, between 33% and 66%. 50% would be the midpoint, which is pretty close to the 50-60% number we established by looking at dollar spending. However the lack of "other" category is worrying, as we know that games not in this list were played. What percentage do these small-fraction games hold cumulatively? Anything over 0% further reduces the estimations of D&D's share.
In summary, the data support a maximum D&D share of 66%, a minimum D&D share of 33% and a likely share of 55%.
TLDR This data supports the conclusion that people saying that "virtually every game" or "90% of all games" were D&D in the early 2000's are wrong. A figure of just above 50% seems more correct.