Ladder representation is often used as an indicator of class or spec strength. If a spec is highly represented on the ladder, we generally consider that spec to be doing well. What representation fails to account for, however, is how many people are actually playing a given spec in PvP. For example, if a mediocre spec is extremely popular and many people play it in PvP, we would expect to see higher representation at the top of the ladder. However, this is not because the spec is strong, but rather because so many people play it. This is further complicated by the fact that when a spec is strong, more people are likely to play it, overinflating the ladder representation and making a spec seem stronger than it is. Similarly, weaker specs may be avoided in PvP, underinflating the representation and making it seem weaker than it actually is.
The reason representation is used as a proxy for spec strength is mostly convenience. Blizzard do not have an easily accessible dataset which describes how many people of a given spec are playing arena. I've recently sampled a large subset (\(\approx\) 40,000 characters) of the WoW population on US and EU realms, which allows us to actually look at the representation of classes on both high rating and in the general population.
Suppose that 10% of the top of the 2v2 ladder are playing a given spec, but only 5% of everyone who plays 2v2 are playing the same spec. This would mean that we see this spec twice as often at the top of the ladder as we would expect if all classes were equally strong. Contrariwise, if we only saw the spec at the top of ladder 2.5% of the time, while 5% of rated PvP players were playing the spec, we'd only be seeing the spec half as often at the top of the ladder as we would expect if all specs were equally strong.
Assuming that the players of each spec are equally skilled, the ratio between the top ladder representation and overall representation indexes how strong the spec is. From the above cases we'd have 10/5 = 2 in the first example (i.e. the spec is seen twice as often at the top compared to in the general ladder), and 2.5/5 = 0.5 in the second example (i.e. the spec is seen half as often at the top compared to in the general ladder).
At the very top of our melee list, we find Windwalker Monks. Monks have been the least popular class for a long time, so it's not terribly surprising that Windwalker Monks are hugely overrepresented given their high representation at high rating. What is slightly more surprising is how far ahead of Arms Warriors they are. This is because Arms is an incredibly popular spec, and so while Arms is very strong at the minute, it is also played by a huge number of people. Arms Warriors are not doing that much better than what we would expect from their popularity alone whereas Windwalker Monks appear to be stronger by quite a margin.
Following Arms Warriors we have Feral Druids. Feral Druids have an entirely average representation at high rating, but they are a relatively uncommon spec to play in 2v2, so we see them more often than we would expect at high rating. Interestingly, this data suggests that Retribution Paladins are nowhere near as powerful as the PvP forums would suggest. They may have the power to burst people down, but they are actually underrepresented at high rating given how much they are played in the general 2v2 population. Subtlety Rogues are in a bit of an odd spot since they have had such a powerful nerf recently. This data tracks all historical data (i.e. also data prior to the sub nerf) so there is a fair bit of uncertainty associated with this number. Sub rogues have been falling in representation at the top of the ladder for several weeks, so it is possible (perhaps likely) that Sub Rogue strength in 2v2 is lower than what is noted here.
In what might be the early stages of a comeback story, Havoc Demon Hunters have increased their representation quite significantly since I did the previous State of the Ladder (from 1.1% to 1.5%). This may in part be due to new Conduits opening up for Night Fae. That being said, Havoc DHs are by no means a top tier spec in 2v2, but the trend will be a welcome one for many players. Enhancement Shamans continue to underperform in 2v2, being among the weaker 2v2 melee specs. At the very bottom of the list, we find some specs that desperately attention from the dev team. Both DPS specs available to Death Knights appear to be virtually unplayable in high-rated 2v2. Fury Warriors and Outlaw Rogues are virtually absent from the arena (very low representation at the top of the ladder and in general). Outlaw Rogues and Fury Warriors at least have an alternative viable spec, whereas Death Knights do not have any viable alternative to play in 2v2. Not shown are the data for Survival Hunter and Assassination Rogue as the data for these specs was very noisy due to low PvP participation. The little data we do have suggests that Survival Hunter might not be in such a terrible state despite being an unpopular spec.
Moving on to ranged DPS, we find at the top Elemental Shamans, Fire Mages, and Shadow Priests. Elemental Shamans have modest representation at the top of the ladder (2.88%), but are played by a relatively small number of people (occupying 1.87% of the general ladder). Fire Mages are doing very well, and although fairly popular, they are slightly overrepresented at the top of the ladder, giving them a spec strength value of just over 1. Shadow Priests are among the most popular ranged specs, but also do fairly well at the top of the ladder. Balance Druids appear to played by a fair number of players for what is often considered a niche spec. Interestingly, despite being the beneficiaries of a great deal of constructive criticism regarding their convoke one-shots, the data do not suggest that they are overtuned in 2v2. Balance Druids may not be particularly fun to play against, but they are not overrepresented given their popularity. If we compare Elemental Shamans and Balance Druids, we can see that over twice as many people play Balance Druids as Elemental Shamans, so we would expect to see twice as many Balance Druids at 2k+ if they were as strong as Elemental Shamans. The fact that we don't explains why Elemental Shamans are considered a much stronger spec than Balance Druids in 2v2.
Affliction Warlocks have increased slightly in representation at the top of the ladder since the last State of the Ladder on the 23rd of December (from 1.53% to 1.69%). However, they remain an underperforming spec, and the strongest 2v2 spec available to Warlocks. The poorest performing 2v2 specs are Frost and Arcane Mage, Destruction Warlock, and Marksmanship and Beast Mastery Hunters. Unfortunately, after the significant nerf to MM Hunters just prior to the start of the season, it seems that this spec really isn't very viable in 2v2 for the general player. I'd like to stress that these numbers do not mean that solid players can't make these specs work, but rather that people in general aren't able to do so. Not shown are the data for Demonology Warlock (again, due to high noise from low participation).
The biggest surprise on the healer list is the emergence of Holy Paladins as the kings of the hill. Holy Paladins are both very popular on the general ladder and well represented at the top. They are closely followed by Discipline Priests and Restoration Shaman, which are both well-performing healer specs in 2v2. It is worth nothing that these specs have very high spec strength values (≥ 2) because healers are in general very well represented at the top of the ladder. This is a well-known phenomenon: at lower rating, you see a lot of double DPS comps, but as you move up through the ratings, healer/DPS teams start to dominate. The consequence of this is that healers are overly represented at the top of the ladder compared to the rest and so are given a high representation index (what I have called "spec strength"). This is also the reason why we should be careful in comparing representation indices between healers and DPS. Restoration Druids are lagging behind quite significantly in 2v2, being firmly a tier below the top three healers. This is followed by Mistweaver Monks which are both unpopular and poorly represented at the top of the ladder. It will come as no surprise that Mistweaver is the worst healing spec by some margin in 2v2.