Be aware that this paper has been fairly thoroughly discredited. See Richard Bornat, Saeed Dehnadi, and Simon. 2008. "Mental models, consistency and programming aptitude". In "Proceedings of the tenth conference on Australasian computing education - Volume 78 (ACE '08)", http://crpit.com/confpapers/CRPITV78Bornat.pdf

We provide our students with a simple calculator at the start of the course, so they have time to get used to it during the semester, and then allow only that calculator into the exam. We have a box of spares in the exam hall, so anything which doesn't exactly match the standard model can be easily swapped out, and students who forget theirs don't suffer. We also annotate all exam papers where a calculator is not useful as "Calculators are not permitted" so we don't have to do a compliance check in the majority of cases. Having a list of "approved" models is a pain, because model numbers change so rapidly and students' claims of equivalence are hard to check on-the-fly. Ian.