Dear Becky,
I would be strongly inclined to think that the variabilities in BC with frequency and placement and other parameters you see in some way or another boil down to acoustic impedance mismatches. Vibrations from your BC oscillator somehow have to travel through skin then bone to the ear, and at each interface there is the potential for more or less of the sound energy to be reflected rather than transmitted, and this is likely to be frequency dependent. You say you see unexpectedly poor thresholds mostly at lowish frequencies. That sounds like frequency dependent impedance to me. The slower the vibration, the less the tiny little bone conductor has to make much of an impression on the much larger, heavier, hard, bony head, particularly when coupled through a layer of elastic skin. If you want to get a good intuitive feel for acoustic impedance, what it is, why it is important generally and also particularly for the question you are asking, I would recommend you try to get hold of a copy of Eric Heller's very fine book "Why you hear what you hear". One of the most accessible and clear introductions to physical acoustics I have come across.
All the best,
J