On the Experimental Robustness of the Allais Paradox