It's always interesting to see "common knowledge" quantified in studies. I've had a few of the same observations as you, although I have a slightly smaller sample size to work from.