"Using statistic to prove how healthy a piece of software is, can be very misleading." 

True, but these statistics are not looking at the health of the actual software packages, they are looking at the vitality of the associated marketplaces to give an indicator as to the current trends associated with the software package usage. Again, when looking at several relevant sources of numbers and statistics in cohesion there does seem to be some indicators that Poser is not the flag ship it used to be in the industry. I have heard a lot of opinion based defence denying this is true, but in reality opinion is an even less reliable indicator than statistics based determinations.

We all know that statistics can be used in ways to apparently skew facts like 95% fat free, which really means the product is %5 fat. But any marketplace, especially web based ones, that say a decline in traffic is a non issue is a little in denial of what is happening in their own market sector. 

A weather man can take all available data and run various simulations using multi million dollar equipment and from that evaluate that there is a 90% chance that the wil be 2-3 inches of rain tomorrow or you can just ask your neighbour "Ya reckon we'll get some rain tomorrow, Jim?" I guess both could vary differently in their degree of accuracy but which do you think most people would trust more for both long term reliability and a higher accuracy rate in their forecasting.