It's not just that MR is a poor representation. It's that it is a self-selected survey with numbers that are far too small to generate anything resembling meaningful information.
It's like going to a Green Day concert and asking 5 people if they plan to buy a T-shirt. You're not going to get a reliable figure on the total revenue of T-shirt sales as a result.
As to the business size, if you go to the source of the stats (
http://www.publishers.org/main/Press...sinOctober.htm), you will see they claim that all US book revenues they track were $725 for October 2009. Out of that, $18 million was ebooks. That's around 2.5%, which may be a bit low but really is not that far out of line of other estimates. I haven't seen anyone claim ebooks have gone over 5% of total revenues yet, for example. There is still a lot more hype than reality when it comes to ebooks right now.
Looks like they publish aggregate annual figures in early March. That may have a bigger picture.
The AAP publishes a list of companies that report to them:
http://www.publishers.org/main/Membership/member_02.htm Note that it doesn't include organizations like Amazon, which sells PD books.
Also keep in mind that ebooks tend to be less expensive than paper. So that's going to affect revenues.