Here's a better graph:
Since everything's normalized for flow rate here I don't think the barbs would make a difference. A change in nozzle size would, changes in finish of baseplate would, and potentially the shift from copper pieces to an aluminum top (or worse yet poly) could. The last (all copper construction) is what I am guessing causes 0.5gpm flow rate performance of LRWW to be a bit better than Dtek
The 1.5GPM point is only one where reproducibility between May and Feb testing isn't stellar; even still it's within margin of error and the old point is the one that looks a bit funny to me not the new one. Hooray for reproducible results!