Strong players (2200+ FIDE) varying castling sides?

What I've noticed is that, at class level play (<2000) that 80% of the time, people are castled kingside.  Then about 15% people will be castled queenside, and rest won't have castled at all.

At strong levels, I've noticed that castling varies A LOT.  I see people castling queenside, when it's wide open but safe, but in that particular opening, castling kingside is always the way to go.  There are a lot more queenside castlers and non-castlers at high level chess.  

I know this because if you look at Josh Waitzkin's annotated games in Chess Master, he shows games in which the players keep thinking, "Kingside? Queenside? Or not at all!"  It's a constant thing going on in their heads.

Meanwhile club players just castle kingside most of the time...

Why is this?

Generally beginners are taught to castle as quickly as possible in order to get their king safe and it quickly becomes a habit. As they get stronger they realize that in many positions delaying or foregoing castling can be beneficial, and that castling kingside too quickly can even lead to quick destruction in many cases, and so it becomes less of an automatic response. So it's a gradual change in understanding that occurs as skill increases, not some sort of dichotomy that exists between amateur and master chess where a light suddenly goes on when you hit 2200.

Also, I think one of the characteristics of modern chess is trying to be as flexible as possible in the opening and keeping your intentions hidden. So it's often seen as a good thing to avoid setting in stone your king's address for as long as you can.

I saw Topalov (please don't blame me for spelling) castle queenside in the PIRC DEFENSE against Kasparov!?!?  

I don't believe it actually has anything to do with the class of play (though it may appear that way).

The position (both tactical+ strategic nuances) on the board at any given time determines WHEN and WHERE to castle.

Lots of criterion determine a castling decision ... some obvious ... others less so.

For instance, if you are losing => it makes sense to castle on the opposite side  of your opponent's castle (if you can pull it off) as this imbalance and the ensuing pawn-storm/throw the kitchen sink at the other king is probably the only way to fight back. Likewise, If you are ahead, you tend to crave symmetry by castling on the same side.

Of course, weaker players will probably not have that much "information" to go with.

To eatingcake's point,  the transition for most players is to go from "clueless" to play "by guidelines/general principles" and as they become stronger, they come up with exceptions to the rule that are playable.

Depends on the opening u play. When u play the sicilian, white will normally castle queenside and start a kingside attack. Black will do the opposite. It all because of the plans of the opening.