I'm certainly willing to believe the cited paper (both versions), but it does seem to contradict my own experience in teaching unlikely students how to program. I've found that a excellent litmus test is the ability of the student to follow the recipe on the back of a bag of Chocolate Chips. If the result is at all edible then I am reasonably sure that they can make the necessary leap from that kind of information to that which is needed in programming. A recipe has all of the usual things that a program does, they just look a little different. Sequence, loop, decision etc. Maybe these people would have been in the 'We can program group' of the mentioned test, I don't know---just know I've managed to teach a lot of unexpected folk the basics (often using basic for that matter :) )...
