Does Age Grading Work? (Part 1 of 3)

A Deep Dive utilizing the 2017 Chicago Marathon Data

Introduction

Age grading is a method for scoring and adjusting times for running races to account for age. The intent of age grading is to level the playing field across age and to allow runners to compare their times to those of older or younger runners.

A commonly used set of age grading tables are the WMA tables maintained by Alan Jones. These tables are based on the age specific world records for a particular distance. A detailed description of the methodology is available at the runscore website.

There is some debate about the validity of age grading, and the purpose of this series of posts is to bring some data to that debate. I’ve downloaded the full results of the 2017 Chicago Marathon, and I’ve adjusted each finishing time based on the runner’s age and gender and the corresponding age grading factor from the WMA tables.

In this series of posts, I plan to present three different analyses of the age graded results. For this first post, I will focus specifically on the distributions of finishing times before and after the age grading adjustments.

Finishing Time Distributions by Age Group

The following box plots show the distributions of marathon finishing times for each age group before the age grading adjustments are applied.

As expected, there is a clear slowdown in finishing times for the older age groups. The slowdown is apparent for both the fastest runners (lower whisker or dot) and for the “average” age group runner (the dark line dividing each box represents the median runner in each age group).

Here are the age group time distributions after applying the appropriate age grading factors to all finishing times.

The age graded distributions appear to show somewhat of a downward trend with age! Is the age grading overcompensating for age?

I think it is debatable. The median times do have a downward trend after age grading, but the fastest times don’t show the same downward slope. In fact, the fastest times overall still appear to be in the younger age groups where there is little or no age grading adjustment. A notable exception is the M50–54 age group (more on that particular data point in a future post). It could be that the age grading is overcompensating, but it might also be that most talented and motivated runners are the most likely to continue competing as they get older and participation numbers fall?

Conclusion

The graphs show that age grading does reduce the time differences across age groups. The adjustment seems to come close to leveling the results for the fastest runners in each age group, but for the runners toward the middle of each age group the adjustment goes beyond leveling such that the median times are faster for older runners. Opinions will differ on whether or not this result is accurate or fair. I think it will be useful to look a couple additional views of the data before forming any strong opinions. In the next post, I’ll look more closely at the age graded results for the runners near the front of each age group.