Analyzing Marathon Results and Age Grading

Age grading allows for comparisons across different age groups in marathon results, but could there be a more effective method?

As I conclude this series, I wanted to utilize updated data, transitioning from my initial dataset that spanned races from 2010 to 2019. This broader timeframe was selected to enhance the number of results, but my findings indicated that runners have been improving their finish times, particularly in recent years.

Having refined my approach, I am now basing my analysis on 2023 data. This article will delve into the dataset, while the subsequent one will focus on updating the z-score and percentile calculations, along with an updated age grade calculator.

Overview of the Dataset

This dataset encompasses results from every marathon listed on Marathon Guide for 2023, including CIM, which was omitted in earlier years.

It contains nearly 400,000 individual finishes, with details on each runner's name, gender, age, and finish time. After cleaning the data—removing entries without age or gender—388,569 finishes across 609 races remained.

• 61 races featured over 1,000 runners
• 108 races had 500 or more participants
• 312 races included 100 or more finishers

For this analysis, I categorized each finish into age groups corresponding to the BAA qualifying standards: under 35, 35–39, 40–44, 45–49, 50–54, 55–59, 60–64, 65–69, 70–74, 75–79, and 80+.

Distribution of Finishers by Age Group

While this dataset is smaller than the previous one, it encompasses a full year's worth of results instead of three months. Does this provide a sufficient number of runners in each age group for accurate percentile and z-score calculations?

The accompanying visual illustrates the number of finishers in each age group—women are shown on the left and men on the right. Generally, there are more men than women in each category, with runners under 35 representing the largest segment; however, about 50% of runners are aged 40 and above. This highlights the significance of age grading, as a majority of participants belong to the masters age group.

From previous analyses, I noted that distributions become less reliable when fewer than 1,000 runners are involved. Fortunately, most age groups in this dataset have over 2,000 finishers, although the 75–79 age group for men and 70–74 for women hover around 600 to 700 finishers. The remaining categories—men aged 80+, women aged 75–79, and women aged 80+—may be too small for dependable conclusions, requiring cautious interpretation.

Does Timeframe Influence Results?

In earlier analyses, I focused on a three-month racing window—September through November—to reduce the influence of runners participating in multiple marathons annually. However, this limited the marathons included and could yield different results if the analysis encompassed all races from January to December.

The data reveals that finishes are not evenly distributed throughout the year. Over half of the finishes (201,000) occurred from September to November, with the latter half of the year (July to December) showing approximately 50% more finishes compared to the first half.

The visual compares finishers from two halves of the year—January to June and July to December.

Does this affect finish times? The plot shows the average finish time for each age group across four different periods: the original September to November, the first half of the year, the second half of the year, and the entire year. While minor differences exist—with September to November and January to June generally yielding slightly faster averages—the overall pattern remains stable across age groups.

This suggests that making direct comparisons between percentiles or z-scores based on the full year versus a shorter timeframe may be unwise, but consistent sample use across any timeframe should yield comparable results. The full year provides the largest sample size, likely resulting in the most reliable outcomes, particularly for older, smaller age groups.

Frequency of Multiple Marathon Participation

Including a full year's results increases the likelihood of encountering runners with multiple finishes. I was curious about how common this phenomenon is.

Accurately matching results across different races can be challenging. A straightforward method involves looking for duplicates based on name and age group, though this approach can lead to false positives and negatives. For instance, two 40-year-old men named Michael Johnson could be mistakenly identified as the same individual, while variations in name spelling or age group changes could result in missed matches.

So, how many runners participate in multiple marathons each year?

The visual displays the results grouped by name and age group, revealing approximately 336,000 unique runners in the dataset. The majority, around 300,000, raced only once, accounting for over 90% of participants. About 27,000 runners (approximately 8%) completed two marathons, while around 5,000 ran three, with participation declining significantly beyond that.

Some individuals ran an impressive number of races in 2023: - Henry Rueden, a 73–74-year-old man, finished 88 races with a median time of 8:31 and a best time of 6:09. - Angela Tortorice, a 55–56-year-old woman, completed 63 races with a median time of 7:43 and a best time of 5:41.

Their high race counts were partly achieved by participating in multi-day series events, such as the Center of the Nation series organized by Mainly Marathons, alongside some major mass participation marathons.

Are Runners with Multiple Finishes Faster?

In discussions about this data, a common assertion is that runners who compete in multiple races each year tend to be faster on average than those who run only once. This seems plausible, so I decided to investigate.

In each age group, those completing one race annually had significantly slower median times than those who finished two or three races. The difference between two and three races is minimal, with three-time finishers generally being slightly faster.

The distribution of runners across one, two, and three finishes is relatively uniform: about 90% for one race, 9% for two races, and 1% for three. However, a slight variation exists, with women skewing more towards one race. Among runners under 35, the breakdown for women is 93–6–1, compared to 91–8–1 for men. Older age groups exhibit a slight increase in multiple marathoners; for instance, among those aged 70–74, the women's breakdown is 89–9–2, while men's is 86–11–3.

Next Steps

With the 2023 data now collected and prepared, I am nearing the conclusion of this series. The upcoming article will focus on preparing data for z-scores and percentiles based on 2023 times, along with an updated age grade calculator incorporating this new information and revised age grade factors.

Following that, I plan to address the broader question of how to compare race results among different age groups and whether these alternatives offer a superior approach to the traditional age grading system.

Once this series wraps up, I have additional inquiries to explore with the new dataset: - An update on comparing 3:00 marathons for men to 3:30 marathons for women - An analysis of changes in finish times over the past two decades - A review of the BAA’s qualifying standards across age and gender

If you're interested in these topics, be sure to subscribe for email updates. I also welcome any feedback or ideas that could enhance this analysis—multiple perspectives are always valuable!

As an enthusiastic runner and data aficionado, I recently turned 40, making age group comparisons particularly relevant for me. Here’s how you can follow my journey: - Follow Running with Rock to stay updated on my training - Check out my tips on selecting a marathon training plan - Connect with me on Strava