Statistical Noise or Basketball Truth?

Chasing the Title reports the following line-up plus-minuses for the Fever vs. Storm game:

Top Fever Lineups

January-Zellous-Douglas-Moore-Davenport: +10

Bevilaqua-Douglas-Catchings-Sutton-Hoffman: +6

January-Douglas-Catchings-Moore-Davenport: +4

The Fever used 12 unique lineups in this game with only four generating a positive value for the team. Two of them stood out as particularly ineffective.

January-Zellous-Catchings-Moore-Davenport: -6

January-Douglas-Catchings-Sutton-Hoffman: -4

Rearranging:

Bevilaqua-Douglas-Catchings-Sutton-Hoffman: +6
January-Douglas-Catchings-Sutton-Hoffman: -4

January-Zellous-Douglas-Moore-Davenport: +10
January-Zellous-Catchings-Moore-Davenport: -6

Essentially, this suggests that

1. Bevilaqua in with the starters was great, January in with the starters stunk.
2. Douglas in with the reserves worked best but Catchings in with the reserves was disaster.

I'm not sure I believe either of those small sample statistical artifacts, although I'm inclined to see some basketball evidence that might support a preference for Tully to stick with the starters and January with the bench.

Does the bench need Douglas' scoring more than Catchings' defense?  Possibly!