Considering that the skin resistance varies by several orders of magnitude depending on various conditions, the discussed 20% variation is almost irrelevant. It's not even expectable that the same "test person" reproduces the same skin resistance in succeeding tests. There's surely a certain amount of parallel capacitance in the skin impedance, but I doubt that becomes effective at 50 Hz. Formally, the equivalent circuit with parallel capacitance can explain the reduced AC impedance, in so far Q2 is clearly answered in the you tube video.
I see two more interesting points:
1. As the "experiment" touches the limit values for AC current that can cause ventricular fibrillation, it should never be performed outside a safe clinic lab situation.
2. Besides possible subjective effects, the reasoning for lower AC limit values in safety standards is exactly the risk of ventricular fibrillation.