As contributors have indicated, capacitance will be easier than resistance. If you must use resistance, I suggest a non-polarizing electrode, like silver/silver chloride, using AC stimuli that integrate to zero, and not DC, would help you avoid the effects of electrode potentials and would give you your best results-- but you still have many of the problems described above concerning contaminants and nonlinearities (and movement artifacts, if you're not careful).
For capacitance approaches on stuff like this, I like using the capacitive sensor as the capacitor (or one capacitor) in some sort of monostable vibrator like a one-shot or appropriate 555 timer circuit to generate a square wave. If you need a DC voltage, you can use a frequency-to-voltage conversion technique (maybe even just a low-pass filter, if you have things set up so your capacitance is linear with your duty cycle), or you can pass it to the CCP or timer interrupt of a microcontroller to measure the frequency or pulse width.