I've gone back to the Yarmey 2000 study which examined the accuracy of people's estimation of durations, and have used their data to work "backwards", take an estimation and work out the average real duration it represents. I've also worked out the 95% confidence interval (basically, the range of real time durations that a given estimation represents). Meaning, if we have someone say "it was about 5 minutes", then the actual duration will fall, 95% of the time, between some minimum and maximum range of times. While I think the average and maximum values are probably reasonably good, working out the minimum part of the range has a few issues that I can't address due to not having the raw data and only the summary they report. However, this isn't a publication for scientific purposes and the values should be good enough to serve our purposes.
I've included the plot here, including the equations one can use to calculate the minimum (dotted series) average (solid series) and maximum (dashed series). You just enter the duration in seconds as X in the equations, and you get the "real duration" times back in seconds.
Using the above equations, here's a table that covers estimations from 1 minute up to 60 minutes that people can, if they wish, use to guide their interpretations.
The ranges present the min, average, and maximum associated with a stated estimate in minutes (which is usually what we're working with). The minimum and maximums are not absolutes, there is a 5% chance that the real duration is outside this range (2.5% chance to be shorter, and 2.5% chance to be longer). That means, you can always claim that it is not impossible for a duration to be outside this range if you need to argue it for your theory, but it also means you cannot rationally deny that your possible value is improbable. Given the ranges are clearly skewed (the difference between the minimum and average is much smaller than the difference between the average and the maximum), the mode (the most common value) will be located between the minimum and average value. I don't have enough information to work out what those would be though.
Enjoy

- Jeff
Leave a comment: