EXTRAPOLATING SALES DATA == BAD

I am guilty (as are other indie developers) of reading far too much into far too little data. I am also guilty (and I find many 'scientific reports' guilty of this) of assuming a causal relationship where none exists (or at least may not exist).
Take the following statements:

My top selling game is the game at the top of my website
My top selling game is the only game of mine with professional artwork
My top selling game is the one that took longest to make
My top selling game is the one I would like to play most myself
My top selling game is the one with the biggest advertising budget so far
My top selling game is a space simulation game

Now all of these are true, but do they really help me work out what makes starship tycoon my best seller? I doubt it. Maybe taken collectively they can provide some clues, but I am of the impression that this whole 'selling games' malarkey is a lot more about gut instinct than hard data and marketing statistics.
I often hear it said that startopia didn't sell well purely because there was little marketing for it. Is this really true? What other factors were involved. Did it get released at the wrong time of year? Were the system reqs too high? Was it buggy?
There are so many factors to take into account when the sales of a game are concerned that I think its difficult to draw any real conclusions, especially when you only sell maybe 20-30 copies a month anyway.

I did an afternoons work on ST yesterday because Democracy was getting on top of me, I'll eventually release an update to ST with some improved graphics etc. Maybe once Democracy is in beta and awaiting feedback.