I’ve seen a lot of previous test compare the stat variables at 20, 40, 60, and 80. Based on the info given to use by SE there is a problem with the ranges used for testing so far. A stat going from 20 to 40 should a much larger impact on the stats than an increase from 40-60. As testing continues from now on the ranges of test will have to increase as you apply more points to each stat.
Every point between 20 and 40 increases the affect that the bonus point has on the stat. But between 40 and 60 every 2 points affect the stat. There are a total of 75 increase factors between 1 and 127 bonus points. The ranges of 1-40 have 40 offer 40 increases of a stat, 20 factors from 41-80, and only 15 from 81-127.
I propose that testing happens in new ranges based on how many times the stats has the time to increase inside the ranges of your testing.
The new ranges for testing progression of stats should be in terms of 15 chances to increase. For instance, to test the impact of INT on magic damage I purpose tests at
INT: 15, 30, 50, 80, 127.
If these ranges are not enough variance to test, then factors of 5.
INT: 5, 10, 15, 20, 25, 30, 35, 40, 50, 60, 70, 80, 95, 110, 125.
At each of these points that damage increase should be constant. If not then we have to look into the possibility that the amount of increase between each factor is not linear in affect. The new ranges would help us test that likelihood.
If we assume that these factors are linear in effect, then your stats will only get 1.33% increase per point between 1-40; .66% per point between 41-80; .44% increase per point between 81-127; and .33% > 127.