1
On McLemore, for example, you're using 2002-2004 line drive data against his 1986-2004 career. His career line drive rate would have been about 20%, not 25%, leaving his BABIP - LD rate around 100.
Stats for anything will skew simply because of random deviation, and often over more than a single season. Bill James makes this point constantly: one YEAR's data is very seldom enough for breaks to even out.