The Replay Lab.

The data / Story 02

Story 02 · The data

Who hits the wall? The data

The story rests on the finishing and half-way times of 1.43 million runners in 42 marathons in Berlin, Chicago and London. The full results are too big to post here, so these two files hold what the story uses: the result of every race, and the middle-of-the-pack comparison. Both link back to the public results they came from.

2datasets 46rows in all 1.43mfinishers behind them 0figures still being checked

The numbers in the story

And which dataset each comes from

42/42marathons where more men hit the wall than womenFrom every race, one by one
77%of men who finished Chicago 2007 hit the wall, against 58% of womenFrom every race, one by one
16%of men in the middle of the pack hit the wall, against 5% of womenFrom the middle of the pack
84%of men in the middle of the pack did not hit the wallFrom the middle of the pack

Dataset 1 of 2

42 rows · 6 columns

Every race, one by one

One row for each marathon: how many runners finished, and what share of the men and of the women hit the wall. Hitting the wall here means running the second half at least 20% slower than the first.

Source
Official results for Berlin 1999 to 2019, Chicago 2000 to 2018 and London 2014 to 2018, collected in github.com/AndrewMillerOnline/marathon-results (MIT licence)
Left out
Berlin 2003 (implausible half-way times), Chicago 2012 and 2013 (no half-way times) and London 2010 to 2013 (unreliable gender labels). We also dropped 3,575 runners with no gender or time, and 434 with impossible times.
How exact
Calculated by The Replay Lab from every finisher. Recomputed independently in R in October 2026, and every number matched.
Used for
The 42 out of 42 figure, the Berlin 2010, Chicago 2007 and London 2018 figures, and the every-race chart
Download CSVwall-race-by-race.csv
Show the whole file
race,city,year,finishers,men_hit_wall_pct,women_hit_wall_pct
Berlin 1999,Berlin,1999,19072,14.4,5.7
Berlin 2000,Berlin,2000,22827,24.3,12.5
Berlin 2001,Berlin,2001,25783,13.7,6.9
Berlin 2002,Berlin,2002,25237,11.4,5.3
Berlin 2004,Berlin,2004,28013,10.4,5.6
Berlin 2005,Berlin,2005,30370,20.1,9.9
Berlin 2006,Berlin,2006,30161,23.9,12.2
Berlin 2007,Berlin,2007,32473,14.9,7.4
Berlin 2008,Berlin,2008,35723,14.3,8.3
Berlin 2009,Berlin,2009,34952,24.4,13.3
Berlin 2010,Berlin,2010,33983,8.8,6.2
Berlin 2011,Berlin,2011,32894,17.7,10.7
Berlin 2012,Berlin,2012,34254,13.7,7.6
Berlin 2013,Berlin,2013,36459,11.6,6.5
Berlin 2014,Berlin,2014,28927,18.2,10.7
Berlin 2015,Berlin,2015,36739,15.5,8.1
Berlin 2016,Berlin,2016,35962,20.5,12.1
Berlin 2017,Berlin,2017,39150,15.3,8.0
Berlin 2018,Berlin,2018,40597,26.4,14.7
Berlin 2019,Berlin,2019,43889,15.8,9.7
Chicago 2000,Chicago,2000,27762,10.2,3.2
Chicago 2001,Chicago,2001,28616,18.7,11.2
Chicago 2002,Chicago,2002,30972,15.0,9.1
Chicago 2003,Chicago,2003,32261,26.7,15.4
Chicago 2004,Chicago,2004,32955,24.9,13.7
Chicago 2005,Chicago,2005,32842,22.8,13.4
Chicago 2006,Chicago,2006,33512,18.5,10.6
Chicago 2007,Chicago,2007,25495,76.5,57.6
Chicago 2008,Chicago,2008,31329,49.2,29.7
Chicago 2009,Chicago,2009,33676,18.9,12.4
Chicago 2010,Chicago,2010,36012,48.1,33.0
Chicago 2011,Chicago,2011,35581,38.3,23.0
Chicago 2014,Chicago,2014,40445,27.4,17.4
Chicago 2015,Chicago,2015,37406,35.4,24.3
Chicago 2016,Chicago,2016,40384,28.4,18.0
Chicago 2017,Chicago,2017,44116,44.1,31.0
Chicago 2018,Chicago,2018,44281,26.1,15.7
London 2014,London,2014,35813,34.8,22.7
London 2015,London,2015,37460,24.7,18.8
London 2016,London,2016,38893,25.6,17.5
London 2017,London,2017,39232,33.0,21.3
London 2018,London,2018,39995,51.3,27.8
racecityyearfinishersmen_hit_wall_pctwomen_hit_wall_pct
Berlin 1999Berlin19991907214.45.7
Berlin 2000Berlin20002282724.312.5
Berlin 2001Berlin20012578313.76.9
Berlin 2002Berlin20022523711.45.3
Berlin 2004Berlin20042801310.45.6
Berlin 2005Berlin20053037020.19.9
Berlin 2006Berlin20063016123.912.2
Berlin 2007Berlin20073247314.97.4
Berlin 2008Berlin20083572314.38.3
Berlin 2009Berlin20093495224.413.3
Berlin 2010Berlin2010339838.86.2
Berlin 2011Berlin20113289417.710.7
Berlin 2012Berlin20123425413.77.6
Berlin 2013Berlin20133645911.66.5
Berlin 2014Berlin20142892718.210.7
Berlin 2015Berlin20153673915.58.1
Berlin 2016Berlin20163596220.512.1
Berlin 2017Berlin20173915015.38.0
Berlin 2018Berlin20184059726.414.7
Berlin 2019Berlin20194388915.89.7
Chicago 2000Chicago20002776210.23.2
Chicago 2001Chicago20012861618.711.2
Chicago 2002Chicago20023097215.09.1
Chicago 2003Chicago20033226126.715.4
Chicago 2004Chicago20043295524.913.7
Chicago 2005Chicago20053284222.813.4
Chicago 2006Chicago20063351218.510.6
Chicago 2007Chicago20072549576.557.6
Chicago 2008Chicago20083132949.229.7
Chicago 2009Chicago20093367618.912.4
Chicago 2010Chicago20103601248.133.0
Chicago 2011Chicago20113558138.323.0
Chicago 2014Chicago20144044527.417.4
Chicago 2015Chicago20153740635.424.3
Chicago 2016Chicago20164038428.418.0
Chicago 2017Chicago20174411644.131.0
Chicago 2018Chicago20184428126.115.7
London 2014London20143581334.822.7
London 2015London20153746024.718.8
London 2016London20163889325.617.5
London 2017London20173923233.021.3
London 2018London20183999551.327.8

All 42 races. The three named in the story are in red.

Dataset 2 of 2

4 rows · 5 columns

The middle of the pack

All 42 races added together, for every finisher and for the 648,000 runners who finished in 3:30 to 4:30. Comparing runners with the same finishing time means speed can’t explain the difference.

Source
The same results as dataset 1
How exact
Counts of real runners, so exact for these races. Recomputed independently in R in October 2026.
Used for
The 16% against 5% figure, “three times as likely”, and the 84% of men who did not hit the wall
Download CSVwall-middle-of-pack.csv
Show the whole file
group,gender,finishers,hit_wall,hit_wall_pct
all finishers,men,950329,226284,23.8
all finishers,women,476174,81156,17.0
finished 3:30 to 4:30,men,463989,74822,16.1
finished 3:30 to 4:30,women,184070,9766,5.3
groupgenderfinishershit_wallhit_wall_pct
all finishersmen95032922628423.8
all finisherswomen4761748115617.0
finished 3:30 to 4:30men4639897482216.1
finished 3:30 to 4:30women18407097665.3

The middle-of-the-pack rows are in red.

What we are still checking

Updated 19 October 2026

Use this data

Free for students and teachers

Students

Draw your own graph

Pick three races and draw a bar chart of men and women hitting the wall. Then find the race with the smallest gap and check it against ours.

Teachers

A chi-squared test in one table

The middle-of-the-pack counts make a two-by-two table: men and women, hit the wall or not. It is ready for a chi-squared test.

Everyone

Credit us

Use the files however you like. Please credit “The Replay Lab” and link back to this page.