According to the All Sites page, traffic is reported as the median of visits, per day, over the last two weeks (click for larger image):
And Kyle Cronin ♦ states that the median is used, so this isn't just a typo.
Median is not a very useful number, for traffic, and it can't be extrapolated to give accurate monthly or yearly rates. For example, suppose a site had daily traffic of:
9, 9, 9, 9, 9, 2, 2, 9, 9, 9, 9, 9, 2, 2
over a two-week period (M-S + M-S). This shows a typical decline in weekend traffic.
The average (mean) of these numbers is 7 visits/day. And this average can be extrapolated to give roughly 213 visits/month or 2557 visits/year -- figures that are reasonably accurate for planning and comparison purposes.
But the median of those same numbers is 9 visits/day! Trying to extrapolate this would give 274 visits/month or 3287 visits/year -- figures that don't reflect reality and are not useful.
This has already been noticed as a problem. See "How can we have a 31-day old site with >14k views, but only 138 visits/day?", for example.
Using median here is also inconsistent. The "questions/day" figure uses average, like one would expect.
So, is there a good reason to report median numbers (other than making traffic look a lot bigger than it is)?
Can we please get averge (mean) numbers instead (or also)?