Well, I carefully factor in the mean, mode, appearance of the dots, the octane of fuel I'm running, the phase of the moon, and the current valuation of AAPL and come up with a number. Which I then ignore and go with what feels right.
Basically, I look at both of mean and mode and then look at the distribution of the dots and try to figure out what the data are really telling me. Most of the time, I use the modal correction. The exception to this is when the mode is 0 and the mean is still >1. I'll shift a bit in the same direction as the mean in hopes of reducing the mean while keeping the mode at 0. In a perfect world, only the mode would matter, because after all, that's the usual correction. However, the data are noisy enough that it takes a little fudging.
Tomorrow I'm going to do a writeup on how to use the MAFtool in 785. It provides 60% as good of results for 20% of the work.