Originally Posted by
wphamilton
This is the part that makes direct comparisons difficult. It is natural to choose a bike for a particular type of ride, and for particular types of weather conditions. Or at least be more likely to, resulting in more cold-weather commutes (for example) on one bike than on the other. The nature of the ride, as well as the bike itself, affects the speed. So unless you're choosing the bikes randomly I think that you need more granularity in your data or else normalize somehow.
Your point is valid if comparing the Marin to the Felt, the Marin is a SS winter commuter and the Felt... is not. It is true that this comparison would be more scientific with a randomize to select the bike for the day. So, not all bike comparisons are valid. However, the Mondonico and Felt have the same conditions requirements (dry). Therefore, the comparison of the Felt and Mondonico is more interesting.
I also consider the top 20% across all bikes mildly interesting, since this only compares top performances when conditions are best. All six bikes have been commuter bikes, and this segment is on my commuting route. I make it a point to ride all six bike throughout the summer months - personal goal to ride every bike each month when there is no snow.
Look specifically at the Kona and how the average speed changes from the "total" to the "top 20%". Since this is the Kona is the work-horse bike and sees some tough conditions, removing the bad weather days makes the biggest difference in the statistics. This shows that the Kona is a faster bike than the total average speed would indicate, but it is still not a pure road bike.
IMHO it's interesting, but not much more than interesting.