Lies, damn lies and statistics

I'd really rather not shamelessly swipe someone else's very clever words for the title of this entry, but it's really perfect. On so many dimensions.

So, I've been doing some research. I don't like doing research, by the way. I've been trying to find some actual evidence that stuff on video is "better" than stuff not on video (printed text, still images, audio...). I got really excited when I bumbled into this statistic that I had heard of before:

people tend to remember 20 percent of what they hear, 30 percent of what they see, 50 percent of what they hear and see

Nice! Well, having been recently bit by the "due diligence" bug, I did a bit more snooping, and found this:

research indicates that we remember 20 percent of what we read, 30 percent of what we hear, 40 percent of what we see, 50 percent of what we say, 60 percent of what we do, and 90 percent of what we see, hear, say and do

Ok, so which is it? Or, more appropriately, exactly whose research?

I found only one actual citation to any published research, and that was a citation from the author of the paper itself. I found one that claimed the study was done by 3M, but no supporting evidence. I found another site that was more than willing to charge me for a business intelligence paper that seemed to be claiming that the entire thing was nothing but an urban legend. Nope, nothing on Snopes about that.