For the past bit I've been collecting data on affixes. I started with a very simple, but very obvious, premise: all affixes are not "created equal." That is to say that all affixes do not have the same chance to be randomized. This thought struck me because I was chatting with a friend and he had absolutely no clue what Missile Dampening was. Then I got to realizing that I really don't see that affix that much. A little after I started this Blizzard announced their three affix categories and I adjusted my spreadsheet to reflect that. It was at that point that some patterns became clear.
As you can tell by the sheet almost all of the collection was done in Act 1, some was done in Act 2 all was Inferno and 90%+ of it was me playing solo. There was a simple reason for this - I knew for a fact that no mobs in Acts 1/2 had any restrictions on what affixes they could have, unlike the Morlu in Act 4, for example and I didn't want that tainting the data. Why? I thought you'd never ask. The procedure was simple. I'd run around doing a fairly normal Act 1 farming run. When I'd bump into a champion or rare pack I'd take a screenshot and quickly tab out to make sure the screenshot was usable (nothing was obscuring the affixes, I actually had the mob targetted at the moment I took the screenshot, etc.) and then I'd move into the spreadsheet and quickly note the mob type (Returned, Retching Cadaver, etc.) and whether it was a Champion or Rare. When I was finished the run I would go back and match the screenshots to the lines in the spreadsheet (this acted as a small error checking step) and Xd off their affixes.
The first sheet is a small summary of the data contained on the 2nd sheet. Feel free to peruse that if you like, but it's kinda boring stuff.
There is a total count for the number of champion/rare packs I faced. There is a count and a percentage breakdown for how many champions and rares I faced. Oddly enough it hovered at the 5 champions to 4 rares ratio for almost the whole time I was collecting data. This may be slightly skewed because the two Festering Woods dungeons are always rares. However, it won't be tremendously skewed because my average run was 15+ champions/rares and the odds tells us that it was highly likely that one or the other would have been a rare to begin with so you're looking at, typically, one "extra" rare (or one "fewer" champion) per farming session.
Then there are the affixes. Each affix has a count and a rate. The rate is representative of how likely it is for a champion OR rare to have that affix. It is calculated as:
(affix count / mob count) * 100
Now this works great for most affixes. But as you'll notice there's another column with three blue and three yellow cells. These are the lovely affixes that are either champion-only (Avenger, Fire Chains, Health Link) or rare-only (Horde, Invulnerable Minions, Missile Dampening). The colored boxes represent the rate at which these affixes occur within their given elite type. This is calculated as:
(affix count / <champion or rare count>) * 100
For example, Missile Dampening has a 1.09% chance to occur among any champion OR rare. However, it has a 2.43% chance to occur among ONLY rares. We want this data because, frankly, it's more important when analyzing the patterns to understand the chance for Missile Dampening to occur on a rare than it is against the entire sample since 55% of the sample can never actually generate that affix. However, it still is interesting to know what your overall chance of bumping into a Missile Dampening mob is. That is why both numbers are present.
All affixes are NOT created equal. However, within their categories they ARE, in fact, mostly equal.
Looking at the Aggressive category it is pretty obvious that there is a 20-25% chance to bump into any of those affixes. This makes perfect sense as Aggressive affixes should have the highest representation due to the fact that they are not limited in any manner. The only affix which seems to be an outlier is Frozen, coming in a decent bit below the next lowest affix in the category.
Taking a gander at the Strong CC category it is pretty clear that there is a 17-19% chance to run into a mob with one of those affixes. This is logical because you can only ever get one of them at a time. Once you get Nightmarish you cannot get Knockback or Vortex. However, the fact that there are only three affixes in this category means that restriction doesn't truly lower their representation THAT much.
Moving on to the Defensive category we see that there seems to be a 12-15% chance to find a mob that has one of those affixes. Again, this follows what logic would dictate because you can only ever get one affix from this category on a mob. There is a larger selection here (seven affixes for rares, six affixes for champions), therefore it would follow that it would have a larger impact on their rarity than in the Strong CC category. However, we see that both Illusionist and even moreso Missile Dampening are rarer than their counterparts in the category. Illusionist came up only 8.75% of the time and Missile Dampening only 2.43% of the time. It is very safe to say that Missile Dampening does not have the same chance to appear on a rare that other affixes do, and even Illusionist seems to be about 50-60 occurences short of where it should be to be "even" with the rest of the affixes in the category.
Here's a bit of the mad scientist in me. If I had to guess, I would say that all the affixes (perhaps with the exception of Illusionist and Missile Dampening) all have the same basic chance to randomize on a champion/elite. I would guess that this rate is probably right around 18%. I would further hypothesize that the general variations we see among the categories are due almost exclusively due to the limitations imposed by the categories and not by the base chance for an affix to be present on a mob. Missile Dampening and Illusionist seem to be the only obvious exceptions to this hypothesis because they deviate far enough from the rest of their category that their lack of representation cannot be explained away in that manner. In this guessing game it would be highly likely that they simply had a smaller base chance to appear.
To translate this a bit into layman's terms, what I am saying is that I believe the system works such that if Blizzard removed, say, Extra Health from the Defensive category and left it in the restriction-free Aggressive category that it would end up with similar representation as the other affixes in the Aggressive category without any other tinkering to rates. I am sure that with a little messing around with the spreadsheet I could probably ascertain if this is true or not. Perhaps that will be my next step.
Blizzard confirmed that Invulnerable Minions is being removed entirely in 1.0.4. This means that post-1.0.4 I will be starting with a new spreadsheet as, obviously, I cannot simply ignore the Invulnerable Minions occurences in this spreadsheet as that would lead to some mobs only having 3 of 4 affixes.