Importance of 16 bit goes up with the amount of post-processing of the image. More layers mean the error due to truncation will multiply (banding being one of the typical outcomes.) So if one is doing bare minimum changes after image capture and does not need to recover details in the shadows, may be 8 bit will suffice. Otherwise I would do most of my processing in Lightroom supplanted with in PS Elements when necessary, if those were my only options.
Once all the changes are done, flattening and converting to 8 bit would have very little perceptive change from the 16 bit post-processed image. Printers print at 8 bits, as far as I know. However, there is data manipulation done after sending the file and before the nozzles fire up (dpi, printer profile conversions, RGB to CMYK etc) for which sending a 16 bit file may be beneficial, if the printer drivers can take it, as opposed to 8 bit. Probably not by much though.
:Niranjan .
P.S. Very nice article.....thanks for sharing, Eric.