Is that last 25% actually 24% (not sure on the current theory myself)?

Yes, I meant to say 24. I'll edit to correct the typo.

There is clearly really powerful data showing that white melee attacks suffer a 4.8% crit reduction. However, the only tests I see in this thread with any real yellow attack data are from Antiarc, and they show a fairly normal crit rate on his SS attacks.

Has anyone done any analysis on just yellow attacks? It changes the spreadsheet values fairly substantially if yellows are immune to this crit suppression, and based on glancing blow mechanics I can't see a good reason to assume that yellows and whites work the same way in this regard.

Testing done on the PTR:
With 4.92% chance to crit, I saw a 2% crit rate on auto-attacks. Either the 4.8% crit depression has been reduced to somewhere around 3%, or rogues and warriors are affected differently by this mechanic(which seems rather unlikely). Any rogues care to re-test this on the PTR?

You could also repeat that experiement on live with using your warrior.

Yep, I plan to. I also realized it could be caused by the low crit rate since just doing a typical DPS test in full gear showed about a 5% crit drop across all white and yellow attacks after 2 hours or so. Perhaps the crit depression won't reduce you below a minimum value.

As part of my spell crit depression testing last week, I also did some tests on the normal crit rate. 4.8% looked very solid without extensive testing. In fact, one of my 45 minute tests on spell crit, I also happened to have just a smidgen below 4.8% (tooltip read 4.80% in fact, theoretical was 4.796%, 107 crit, 230 agility) and I observed 0 crits on white attacks or specials. My notes have 595 successful wound poison applications for that run so I'd guess that was about 1500 normal attacks without a crit. I also tested just special attacks and 4.8% also seemed correct. I do not have the combat logs handy as I didn't bother to fully quantify that part of the testing given that I did not expect any crits anyway.

Edit: Also I just noticed in your picture that your weapon is broken. Did the weapon start broken or did it break during the testing? And does that weapon happen to have stats on it that enhances the crit rate.

It was a tanking weapon, no crit, also no axe spec or any talents that would influence crit in any way, and both 1h axe/unarmed skill are maxed. It broke during testing when I had a few hundred hits to go until 10k and didn't want to repair. Since rogue tests on the PTR have re-confirmed the 4.8% reduction and that it's possible to reduce your crit chance all the way to 0%, the only 2 explanations I can think of that remain are:

A) Warriors only receive a ~3% reduction in crit chance.
B) DWing causes a 4.8% reduction while single wielding(one handers at least, and perhaps 2hers) only receive a ~3% reduction.

Edit: Tested a few thousand hits dual wielding tomahawks with a 0/0/0 spec and no other gear. This rules out B, and it must just be the warriors and rogues have different crit penalties, weird. 276 crits out of 14250 swings is a 1.94% crit rate, which is 2.98% below the expected value of 4.92%.

What would be more helpful than just screenshots of recount, would be combat logs themselves. These might provide insight into the issue. For example. we could see if the crits are more prominent at certain times or see whether there were unexpected debuffs on the boss that you did not notice.

The debuff thing is correct, the crit frequency changing isn't really an issue, the sample size is large enough that error from that sort of luck is tiny. In fact, doing a normal approximation confidence interval on this data shows the following hit table for Rallik while Dual Wielding:

miss -> 27.6 ± 0.6%
glance -> 23.6 ± 0.6%
dodge -> 7.1 ± 0.3 %
crit -> 1.8 ± 0.2%
Hit -> whatever make this 1.

This is all assuming that additional debuffs weren't added to the dummy at any time.

Rallik, can you redo both sets of data and post wws parses? Don't wear any gear except the weapon.

Wow Web Stats

Done with 2xDalaran Sword/4.92% crit/and 0 talents.

So to sound dumb, Does this mean if i wanted to theoretically crit on every attack i would need to be exp/hit capped and have 104.8% Crit. Is this what this information is telling us... or am i miss interpreting somewhere?

Or alternatively does that mean, the highest ammount of crit i can get is 95.2%... but with this new information we now know that those 4.8% remaining will not be glancings but actual normal non-critting hits?

24% of your attacks will always be glancing even with 104.8% crit. I also believe it was tested to show that 4.8% of your crits are force-converted into hits in a later topic.

This is correct. If you're hit and Expertise capped, at best, you'll see an observed crit rate of 71.2%, if you have gear bringing you to 76% crit. Crit conversion and Glancing Blows cannot be geared around.

Since people are commenting about unexpected buffs appearing on boss dummies, buffs from totems and the such that will contaminate the test data i would like to point out there is a second set of lesser known dummies in Org if you are unaware of these.

They are located in the Valley of Honor. Head towards the PvP building but before you go in the door follow the wall around to the rear of the building. There is a boss dummy and 2 lvl 80 dummies that hopefully no one is using. Hopefully this allows you to better control your experiments environments.

I do not know of any likewise dummies on the alliance, if someone would be able to follow this up, it may benefit us all.

I don't know if it's the same on all realms but on my realm the dummies just inside SI:7 in Stormwind seem to be used far less than the ones in Ironforge, so good for testing without nearby players adding in random effects.

The 4.8% is already factored into the Crit capping in the spreadsheets.

To further expand: It means the opposite. There's a forced conversion imposed so that 4,8% of your white attacks are converted from crits to hits; in other words, regardless of how high your crit rating is, at least 4,8% of your white attacks can never crit. To this, add Glancing Blows.

In this light, I believe you should be able to picture this roughly drawn-up Crit Cap scenario:

You're hitting a boss mob and you're feeling pretty good about yourself. Why? Because you're exactly where you're supposed to be: Rightfully positioned behind the boss so he can forget about blocking or parrying any of your attacks. Additionally, you *just* got that last piece of Expertise gear to reach the soft Exp cap, so there's no dodging your attacks either. Also, you've gone the extra mile and managed to get 722 hit rating, thus reaching the hard hit cap; you simply can't miss.

Now, with all of this, your attacks can't be dodged, blocked, parried and they can't miss either. On top of that, you've somehow managed to up your crit rating, so that your crit chance is 100%, no less. (arbitrary, granted, but you get my point). In fact, now you're absolutely certain that all of your white attacks will crit, right? Wrong. Because you still have Glancing Blows and the Forced Crit Conversion biting you in the ass, reducing your effective crit chance to 71,2% (100%-24%(Glancing Blows)-4,8%(Forced Crit Conversion).

Has anyone testet if the forced conversion does also apply to style attacks? Would be interesting for tha Arms Warriors in regard of Overpower.

I believe tests have been carried out with a 54,78%(if I recall correctly) Crit Chance combined with the added 50% crit chance from Overpower; the test gave overwhelming evidence to Overpower not being affected by the Forced Crit Conversion.

Just to come back to this, I disagree with both points. It doesn't increases your leeway on crit, but nor does it decrease it. It's a conversion, so it effectively reduces your crit chance by 4.8% while also reducing the end point of the range by 4.8%, thus having no net effect. The maximum percentage of your hits that can be crits is affected by it, but your 'wasted' crit is not.

Observe the hit table:
Miss
Glancing etc (zero from behind and exp capped)
Crit -4.8 conversion
Hit (zero when crit capped)
4.8 guaranteed hit

Note the 4.8 and the -4.8, cancelling out their effect on the size of the available crit zone.

I'm unclear as to how the crit conversion affects yellow attacks - does it not affect them at all? Or is it merely removable from the end of the table unlike white attacks?

This has not been as extensively studied due to it being a lot more work to come up with a bajillion yellow attacks than it is to come up with a lot of white attacks (though I suppose strictly speaking Tiny Abom might help us here), but the current theory is that there is crit reduction, that it's about 4.8%, but that it doesn't stop you from getting all the way to 100% crit.