Thanks for pointing this out.

For those keeping score: It turns out that my original results file had some duplicate entries somehow. I must have messed up somewhere in the script I guess. The new results do not have any duplicates, so it is a much more accurate representation of my data.