Okay, this is what you want to here (hopefully).
The bit depth refers to the value range each sample can be. For 16 bits, the samples range from 0 to 64k, or 2^16. Since audio is based on the sine wave, values would more accurately be described as ranging from -32k to |PLS|32k. The values represent the given value of the wave form for that sample - the "loudness" of that sample.
The sampling frequency refers to how many samples per second (expressed in Hertz, Hz). Convention says that the sampling frequency must be at least twice the highest frequency of the sample to represent the waveform accurately.
All this being said, I'll give you an analogy of what you are attempting to do. What you are doing is analogous to converting a GIF picture (256 colors) to a JPG (millions of colors) picture. You're only going to get 256 colors out of the millions because the original picture only had 256 colors. It's not a perfect analogy, but it gets the point across. You can't get BETTER quality than the original sample. You just can't.
Changing the sampling rate of a digital waveform would do one of two things, depending on the criteria. Reading the same amount of samples in a shorter time would transpose the song into a higher "key" (although it's not strictly true, because the increment may not be an actual music key), in a proportionately shorter time - like playing a 33rpm record at 45rpm. If you tried cramming more samples in the same time frame, you end up with the analogy mentioned in the above paragraph.
So this is the bottom line: digital waveform re-sampling from lower to higher bit rate = waste of time. (Higher to lower has its advantages, though. You save lots of disk space - with the tradeoff that the resulting waveform is described less accurately than before - but you might not be able to notice the difference.) As for changing the sampling rate, I *did* find one use for it. Some of my (former) band's stuff was mastered with uncalibrated equipment (if I could find the guy, I'd kill him!), and the speed, when played back on a cassette player (it was that long ago, okay!), it sounded like it was playing at the wrong speed - way too fast! So I digitized it, then changed the sampling rate to "stretch it back out" to what it should have been (relying upon my ears to tell me when the music was being played back at the correct pitch - not very scientific, huh?)
Bill
Oh, I forgot to mention - for all A to D conversions (analog to digital), use the highest sampling rate and bit depth you can handle for the highest quality. Just remember that the higher the quality, the more disk space you consume - and, if you sample above 16 bits or 44kHz, you will have compatibility problems with sound cards that cannot process those files. Unless you use MJ! (I actually haven't tried that yet, though!)