When I do this demo in class (see here for details or here for the brief version), I phrase it as “the percentage of countries in the United Nations that are in Africa.” This seems less ambiguous than Kahneman and Tversky’s phrasing (although, I admit, I haven’t done any experiment to check). It indeed works in the classroom setting, although with smaller effects than reported by Kahneman and Tversky (see page 89 of the linked article above).
That book is indeed a great one and I have used many ideas from it in teaching an undergraduate probability class myself. I’m a grad student in applied math, so I may not see you in many of the same conferences, etc., so LW appears to be as good a place as any to say thanks. The Bayesian Data Analysis book is also quite good.