But what about a weekly times training run, not races. Or parkrun?

parkrun has a great philosophy, and that is to provide an organised and timed free weekly training run where they can all over the country (OK it is the UK, so 4 countries, and it is spreading overseas too).

Parkrun results shouldn't be included on this website:
- It is not a race! It is a timed training run... though some people will use to run at race speed
- It is not run on roads - so even if it was a race it is not a road race! Some courses are cross country, though most are on paved traffic free wide paths - comparable to a road
- It is not a race, but is pretty close to one.

I am considering 3 things here:
The Parkrun Averages - similar to what I added up for the other races
Which are 'fast' Parkrun courses, and can I use them to identify why a course is fast
Is there a fast month for runners

Parkrun is good for this analysis because the same routes are run every week, they are all the same measured distance and the mix of runners each week is pretty similar (indeed some runners run their local one every week). If you look at enough results from the same course it will remove inconsistencies from the results compared to one off events such as the weather (wind, hot cold etc), or an unusual mix of runners (such as events that are part of club championships where large number of club runners run).

I am only analysing these runs up to certain limit this is because, like the other race distances, the numbers converge to a steady average and there is little benefit to adding more and more records to my sums. I am assuming that all Parkrun have the same make up of runner types - the same percent of fast, medium and slow runners, and pretty much the same proportions of club runners, and men and women.