I found this on another message board and I thought it was fun so I thought I'd post it over here. Somebody is conducting an experiment on kanji lookup methods:

http://www.ish.ci.i.u-tokyo.ac.jp/~julian/TEST/
It doesn't matter how much or how little experience you have with kanji; the estimated time is 30 minutes, but it only took me 11 minutes.

[ファブリス]

Interesting.

search time: 11min 29s

I was already used to SKIP, as well as the radical search in JWPce text editor.

The vertical/horizontal/other method was really inefficient for me.

SKIP is my preferred method. It's not perfect but quite fast. Not much faster than radical search though. But radical search may be slower in a real dictionary, I have yet to see and try that.

It took me about the same, 11 minutes. I also prefer SKIP, which is what I normally use, but for me on this test I was faster using the radical search. I do think that radicals would be slower in a physical dictionary, since you don't have the quick culling that you get on this test.

It took me around 13 min. I always used radical search since when I first wanted to look up kanji, that was the method that was easiest (didn't know how to count strokes, etc.) and it works pretty well. I had no experience with the other search methods. For me in the test the SKIP method was fastest, but I'd have to try it out a bit more to see if it really works better for me than radical lookup.

Search time 7min 11s (search time 9min 26s) for me. First time using SKIP and the one with the Japanese name. I rarely look up kanji, but when I do I usually just copy and paste... So when I do run into a kanji written on paper that I don't understand, it takes a long time looking it up by stroke count using my electronic dictionary.

It's an interesting research, but the results doesn't mean much. The methodology seems to fail in several aspects: I think there's not enough kanjis to look up; you can't try the interface before using it; the search methods are tried one right after the other; ...

But the idea is great! I hope someone would make a better test (even if it takes more time to participate), even including other methods (for instance, I never used stroke count alone, but sometimes I use it in conjunction with radical lookup).