idaho: there's no polling at all, but sanders should get a very tentative advantage based on his wins in states like colorado and kansas - or perhaps his projected victories in washington and oregon. i don't know a whole lot about idaho, but maybe it's being pulled between the ocean and the plains, in terms of ideology? it's little more than an educated guess, but i think it's widely agreed upon. and you can't rig a caucus...
arizona: the one released poll had 24% undecided, which suggests a huge level of uncertainty. i am not at all aware of any factors that may sway undecideds in any direction, except the ubiquity of hillary clinton. i do not believe that the kind of media that would help sanders, and probably did in michigan, has happened in arizona over the past week. however, it's well known that sanders has a lot of support in tuscon, too. the cold truth is that there simply isn't any direct data to ground any kind of prediction in. but, there is a lot of reason to think it ought to be somewhere between nevada and colorado, in terms of results. if you see a clinton win with the same kind of margins that she got in some of the southeastern states, it should set off red flags.