Abstract :
In a recent paper I proposed a system for qualitative probabilistic reasoning, based on argumentation, and proved its correctness with respect to probability theory. This system was flawed. In particular, it failed to take proper account of d-separation, and so can give erroneous results in certain cases. This paper identifies some of the problems caused by this flaw, examines their extent, and then fixes the flaw. While the main thrust of the paper is to overcome this flaw, the discussion of the problems caused by the flaw exposes some general issues in qualitative probabilistic reasoning.
