I'm a Senior Engineer on the Photoshop team.
Photoshop's codebase is 20 years old, but constantly cleaned and refactored. Our codebase is C++, and pretty portable. That has nothing to do with not porting to Linux.
I doubt that the research is well published, since marketing research is usually perfomed by well paid research companies (and they want to keep getting paid). I see the research done every year or two, and the results barely change with respect to Linux users.
Linux still lacks standards for color management or fonts, and just barely has standards for printing. Things like tablet support are more than a little hacked, and drivers are still a nightmare. And that's just the problems I can see from occasional use, I'm sure there are more. And where is the standard UI toolkit (currently you can pick from 6 or more bad choices based on the mistakes of X Windows)?
Stabilizing the OS and adopting standards would make Linux more attractive to ports (and to write anything more complicated than command line apps). It would also be far more attractive when code can run on more than one distribution of the OS without major effort.
Yes, Linux devs without experience on other platforms might find the facts hard to swallow. But Linux is not a single OS, but a kernel used in many fragmented OSes with few standards.
And, AGAIN, the primary reason for no Photoshop on Linux: there is no market. Linux users are still not willing to pay for commercial software. You have to solve that problem before you'll get serious commerical applications.
Solving the standards problems would make Linux more attractive to developers, and then you might get more users. Right now, the sets of Linux users and developers are overlapping a little too much (when app tutorials start with "download these 8 packages and build them", you've got a problem).