It might be an idea to compare different agility builds with the same base wpf, i.e. the attack speed at 100 wpf and 6 agility, then 9 agility, then 12, etc.
Maybe the increase of speed with agility is based on the amount of wpf. If you do the same test on the page before with more wpf maybe the increase of speed with the increase of agility is higher.
i was thiking something like this too... maybe 3 and 27 agi is the same with 1 wpf but has a difference with 150 wpf.
If this were the case, it would be evident in the plot of time per attack verses weapon proficiency. The fact that the data was uncontrolled for agility but remains linear is sufficient proof to me.
IMHO to have a better picture you need to gather data using swings or overhead - one that takes most of the time to complete. Cause playing str build I don't see a problem in making fast stabs but swings go really slow compared to agi build.
Every test I have preformed indicates that there is a negligible difference in time per attack between overhead, thrust, and slash animations of a weapon. The most likely reasons some attacks feel faster is that the different animations move through different positions relative to the combatants. For example, players can turn while slashing to increase or decrease the rate at which the weapon will advance. It should be noted that turning does not change the duration of the attack animation, which is what I am reporting on, but can change the point in the animation sequence that a hit will be registered. If agility affects the rate at which a player can turn, either directly or indirectly through athletics, it might explain the experiences of some of the posters.
I wonder if an identical test with a 2-handed weapon would be comperable or not to polearm, thus clarifying if animation in fact changes weapon speed.
I have data for the time per attack for different weapon speeds at 1 proficiency for each weapon type:
The data shows that time per attack for one handed, two handed, and two handed polearm weapons have a colinear relationship with weapon speed. One handed polearms are slower for a given weapon speed, even those without the "penalty with shield" modifier. For example, the time per attack for my 94 speed masterwork awlpike with a shield is equivalent to the time per attack for a speed 74 one handed, two handed, or two handed polearm weapon.
Time per attack for a speed 94 one handed polearm weapon with 1 proficiency = -0.0195 seconds * (weapon speed) + 3.257 seconds = 1.424 seconds
Time per attack for a speed X one handed weapon with 1 proficiency = -0.0124 seconds * X + 2.3368 seconds = 1.424 seconds
X = 73.6
Also, from a limited test of time per attack verses weapon proficiency for one handed and one handed polearm weapons, it appears that 100 proficiency for other weapon types yields the same ~6.5% reduction in time per attack.
I want to remind everyone that the data presented here is the duration of the attack animation which does not necessarily correspond to the time it takes for the attack to land an effective hit. I think the most important findings are that the relative reduction in time per attack for increasing proficiency is constant for all weapons (~6.5% per 100 proficiency), the absolute reduction is greater for slower weapons than for faster weapons (6.5% of 2 seconds is greater than 6.5% of 1 second,) and that agility does not directly impact the time per attack.