I think you divide distance between the two supports by the deflection, so it works the same for 28, 26, 24 " etc
Obviously the further apart the supports the greater deflection and thus the easier it is to measure it and there is less error.
It should be easy enough to check with a long arrow, do the test at different distances and see how the results tally (If you do it, please let us know how it works out)
Mind like Badger says, it's all relative anyway, as long as your method is consistent, the figures will be meaningful for your tests.
Del