I also downloaded the data for the Eugene marathon for years 2008, 2009, 2010 and 2011 just to get more data ;)
As Nelly asked before, I'll tell you that when I say 'fade' I mean the percentage by which your marathon time went up due to extra time in the second half.
So if you ran a 5 hour marathon, and you took 2 hours in the first half and 3 in the second half you would have
fade_pct = (5hrs - (2 * 2hrs)) / 5 hrs = 1/5 = 20%
You will notice two lines on the plots..these show a best-fit line matching the average fade for men (red) and women (green). The slopes of the line are noted as well in the X-axis label text.
I have cut off the data at 5 hours..slower than that and you are getting into walking, as well as some people that were clearly not faded but injured (calf strain, GI problems, ??).
You can see that for both men and women the percentage spread of fades increases with running time. ...e.g. the really fast runners are less vulnerable to the problem. Remember that these numbers are already percentage based on the finishing time so the increases in fade are a double whammy to the finishing times.
To return to my original hypothesis about women and men's marathon pacing: with these plots, you can clearly see that the women have:
1) Less overall percentage fade. At a 4 hour finishing time the women have 2-3% less fade. This is 6 minutes!
2) Less increase in percentage fade vs total running time...by about .5-1% per hour. So as we are all out there the men fade more AND their fade increases about 1% per hour more than for the women.
There is a lot of individual variation but I think these plots show the effect pretty clearly.