Today I was looking at warcraft logs rankings of retribution paladins to see if my spec is actually a good one. The first ranking I was linked to shows ret paladins in 22nd place of 24 specs. Ouch. Then I realized that this was only showing Mythic difficulty, so I changed it to Heroic difficulty because that is what I actually raid. Now I am showing up as 2nd place instead of 22nd. There is some small correlation between the two data sets, but it is pretty random. When I switch to Normal or Looking for Raid I similar giant swings.
A lot of this is going to be due to people's perceptions of how specs are performing. If the absolute best players all decide that one spec is a couple percent higher than another, then that best spec will end up overrepresented amongst the top players. This will inflate its numbers, and even though all things being equal one spec might do 95% of the damage of another, once you account for the shift of top players towards the better spec it suddenly appears as though the number is 85%.
You also have problems with some players doctoring logs to make themselves better, some groups organizing raids just to give certain specs great parses, and specific sets and legendaries making perfectly optimized characters of specific specs amazing without actually affecting the average people playing those specs. People represented in mythic raids are going to represent not the average but the extreme top end and that means that highly variable specs will be overepresented.
In other words, much as people really want to use warcraftlogs rankings to complain about the shape their own spec is in I don't think you can use it effectively that way.
I figure that I am consistently top two in my pickup group Heroic raids and guild raids. I don't get cut from groups because of my spec. Given that, I ought not to worry about it. Far better to work on being better than to look at online statistics and fuss over my place in them.