Measuring the Degree of Ambiguity about Probability: Experimental Evidence