Skritter has so much data, I think you (or someone) should use it to find out something about learning. I can remember there was a thread about this...
One thing I would really like to know is the difference between those people who use mnemonics and those who don't...best would be something like this in the end: People who use mnemonics learn 1.x times more characters in the same time with the same retention rate than people who don't.
Not sure if it's possible like this but there should be at least some information hidden in the data.

There's a lot of interesting data to analyze and not a lot of time. The great thing about built-up data like this is that it only gets better the later we wait. The bad thing is that because the dataset is so large, it gets less convenient to download and work with. It's something we want to do, to look for interesting patterns to share, like how well mnemonics work, how does length of definition affect difficulty, is traditional or simplified easier, what tones are hardest, etc.

Well here's some data for you. If I manage to remember the mnemonic, it REALLY helps me remember the character. Unless it's a crap mnemonic. (Sometimes happens)