sample rate conversion quality issues

Continuing Phild's comment, I've heard that if you plan to drop a digital recording to 44K that it's best to record at a sample rate that is an even multiple of 44K. So in that case, recording at 44K or 88K would be better than 48K or 96K. Logically it makes sense to me as converting from 48K to 44K would require much more calculations and therefore more chances for errors.

Not that easy, Paul. It really depends on the equipment (algorythms) used during any type of conversion. With inferior equipment all benefits of recording with higher resolution would not just be lost, but the resulting 16/44 would be actually worse that if it was initially recorded this way.
I did many tests using my dCS gear as a playground and as of now my solution is - try to stay in one resolution.
There are many sides to it, but in general - less conversion is better.

Alex,

The original poster asked if a 24 bit recording converted to 44.1 with good software would sound worse than the same recording originally mastered at 16/44. If we can agree that "good" software means there won't be any truncating or artifacts introduced by the downsampling, I fail to see how the downsampled recording would not be superior in resolution to the one mastered at 16/44. In my experience, even recordings made at 16/48 and dithered to 16/44 with good gear sound (slighter) better than those mastered at 16/44.

Paul