That strikes me as really ... odd.

To whom is the advice addressed? If something is actually untrue, and one has determined it to be untrue, then the task of being skeptical about it is finished.

I could probably find a loophole in the preceding statement, but it couldn't possibly be what Bill James was referring to.

As for directing skepticism at [claims depending upon] things that are difficult to measure, well that seems like one step away from directing skepticism at claims depending on little evidence. Which is surely what we want to do. Again, there's a loophole, but clearly not something Bill James was trying to point out.