Hi, the author is suggesting that we need to minimize the impact of the referenced inferential errors that juries can make. each of these possible errors occurs at the individual level- these are mistakes of reasoning that a person commits when they are biased or swayed inappropriately by evidence that does not suggest what one would take it to suggest.
However, if for whatever reason these errors disappear at the group level, then juries would not be subject to inferential error at all. The authors argument implicitly relies on the idea that these errors that an individual makes are also relevant to the jury as a whole. If, as B tells us, this shift in the frame of reference is not valid, than the authors argument makes little sense.
As for the wrong answers, they just don't give us any reason to doubt the author.