I’ve been wanting To Write This Post For Quite Some Time. Whenever there is an article or an interview posted about the Boston Marathon results review, I get emails and comments about runners worrying that they will get flagged.
This will be an attempt to walk you through the process I use to prioritize the Boston Marathon reviews.
For the past 2 years, I have reviewed Boston results looking for those that cheated in order to run Boston. This includes course cutting, bub swapping, and using “bib mules” to run a qualifying time for them. The main assumption in how the review is prioritized is that most of the runners that cheat to enter Boston do so because they are unable to run a time fast enough to qualify legitimately.
The review is ongoing, and is kicking into high gear. I have been focusing on identifying those that have cheated in their 2017 Boston Marathon qualifiers. There are a few cases where I have emailed qualifying marathons on questionable runners, and I am following up on those cases.
But, since we are outside of the busy marathon season, this gives me an opportunity to finish the 2016 Boston Marathon review. This is still important to do because I have found that many of the runners I identify have cheated multiple times. By identifying cheating in prior races, it enables me to find other instances of cheating as well to prevent future infractions.
2016 Boston Marathon Review –
The first step in the review begins when bib #s are assigned. Once the bib #s are assigned, it enabled me to determine the qualifying time submitted by each runner. By looking up qualifier cutoffs for each corral, I was able to estimate qualifying time for each bib.
What this allowed me to do was to quickly identify the qualifying marathon for many of the runners. I have a database of historical marathon times which I can match up to the correct runner and estimated qualifying time. This saves me quite a bit of manual work in looking up qualifiers.
This chart shows Actual Boston times plotted against bib #s. Once I received the download of Boston results, I was able to quickly plot this out.
I’ve reviewed about 1/3 of the results that I plan to look at before it’s time to move on to Boston ’17 results. The red dots represent where cheating was confirmed. The green dots represent where runners were cleared, or there was not enough information to make a determination. The black dots still need to be reviewed. The runners noted in blue are not being reviewed. Their results are within the expected range.
Ideally, if I had unlimited time and resources, I would review more runners – or all runners so that I could better validate the prioritization.
A couple notes regarding the data. There is a band of runners starting at the 15500 range of bibs. These runners received entries without qualifying. They were mostly military or service exemptions. These entrants stand out immediately when looking at the graph.
Also, the charity entries start with bib #26000. You can visualize this clearly by looking at the top chart.
Refining The Process
Instead of just looking at the qualifying times to predict the runners’ Boston times, we went further.
With the input of others we developed predicted Boston times.
- Course Conditions – Average times were slower in 2016 (It was hotter). By starting with the average times looking at ranges of bib #s, we adjusted the finish time predictions.
- Qualifying Course – I utilized course rankings from findmymarathon.com to further adjust predicted finish time. Last year I found that I was reviewing a large # of runners from the downhill marathons (St. George, Revel Canyon, etc.). I made adjustments to the predicted finish times based on the difficulty of the qualifying race.
- Runner Demographic Data – One of the people that helped me with the data determined that if you chart gender separately, there is a different profile to the data. There was a larger variance between Boston Time vs. Qualifying Time for men than there was for women. This was accounted for in the predicted times. For Boston ’17 I will also incorporate age group data to further refine the predictions.
To support the blog, please consider a small contribution.
|One Time Contribution|