About the Data

To assess who was being excluded by running industry standards, I realized I needed to seek out a dataset that was not necessarily representative of the universe of road racers, but one that was more likely to include the runners who are traditionally excluded from the community and research around it. An added challenge is that road race results typically only collect age, gender and location, limiting the amount of demographic data available. Unless Athletes with Disabilities (AWD) are competing in a handcycle or wheelchair category, they are not designated as AWD in race results. Only a limited number of races provide gender options for registration beyond male and female, and the world governing body for the sport, World Athletics, is actively trying to exclude runners who don’t fit specific criteria for gender. Finally, race time limits push slower runners to self-select out of the data by setting finish times that discourage them from registering for races.

The New York Road Runners dataset addressed a number of those issues. Because New York City’s population is more racially and ethnically diverse than most of the country, the road racing community is also more diverse. The organization provides programs geared toward encouraging youth, senior citizens and low-income runners to participate. It allows runners to register for races in a non-binary category. It also partners with Achilles International, a nonprofit that serves athletes with disabilities, for the largest Achilles race in the United States each year, as well as providing AWD registration options for all races. NYRR is large enough and conducts enough races each year at varying distances to provide a robust data set in just the few post-2020 years. Finally, in late 2022, the organization lengthened the time limits for many of its races, and by the end of 2025, all the races included in this dataset had significantly longer time limits than pre-2020.

I built a series of Python scripts to scrape the race results data from the New York Road Runners results website, compiled them and calculated a pace field for each finish to allow for comparisons. I scraped the results of each 5K, 4-mile and 10K race in 2022 through 2025, a total of 459,812 records. I also used the age field to set an age group for each record to align with industry standards.

While each record has a pace field, I analyzed only the finishes at a given distance when comparing actual results to industry standards for specific distances, so only 5K finishes are used to compare the field to Couch to 5K training and only 10K finishes are used to compare the field to the corrals seeded by a 10K-equivalent pace.

One limitation of this data set is there is no information on how experienced the runners are, and training gains compound over time. Thus beginner runners who are the ones Couch to 5K is aimed at are likely to be slower than the field as a whole, but there is no way to measure that with this data set. Additional survey research in conjunction with a race organization to ask runners registering for races how much running experience they have would allow for improved data to better answer this question.

All the data is available for download on my GitHub respository for this project, and I invite other researchers and runners with an interest in this data to use the data and continue this analysis to explore further questions around the road racing community.