It is very much contextual and I use various methods in different situations. Two of my favorites though are used to develop accurate estimates when there is very little factual information by using very basic statistics. They are effective types of "SWAG" estimates.
The first is the 3-point estimate. If I don't know what a product or service costs, but I can find premium and budget quotes for the same type of thing, then I can effectively bound and bias the solution. For the most likely value decide if it is closer to the high or low end of the range, pick a best guess value, and use {low + 4(most likely) + high)/6 and it's probably not too far off.
The second is if you have a lot of opinions, but little data. It is discussed in the fascinating book, "The Wisdom of Crowds". If many people guess at how many beans are in a jar, and you can average all those guesses, the average of them all has a very high probability of being closest to the actual number. This can be done with project estimate guesses essentially like multi-voting except instead of picking the option with the most votes, average all the votes.
Community Champion
