Now, I got curious and did some testing myself too. To rule any disturbance, I had myslef test them seperately:
Private Sub cmd_Click()
Static sblnMod As Boolean
Dim lngCounter As Long, lngDummy As Long
Dim sngStart As Single, sngStop As Single
If sblnMod Then
sngStart = Timer
For lngCounter = 0 To 10000000
lngDummy = lngCounter Mod 2
Next lngCounter
sngStop = Timer
Else
sngStart = Timer
For lngCounter = 0 To 10000000
lngDummy = lngCounter And 1
Next lngCounter
sngStop = Timer
End If
Print CStr(sngStop - sngStart), CStr(IIf(sblnMod, "mod", "and"

)
sblnMod = Not sblnMod
End Sub
It seems that, indeed, there's a slight difference in processing time in favor of bit testing (although I had expected more...).
I can't get around the fact that accuracy of this test probably is miserable, however, this accounts for both methods and results are constantly in favor of bit testing.
I'm not familiar with the instruction set of the processor, but I'm quite sure that there's an intrinsic bit testing instruction whereas I don't know about an intrinsic mod instruction. Even if there is one, it has to be involving more clock cycles, since it's dividing and checking remains, whereas testing one bit value should be possible in one cycle.
Greetings,
Rick