Author: Allison Goldstein
For marathon runners—or really, runners of almost any distance—negative splitting is commonly considered the ideal pacing strategy. (“Negative splitting” means running the second half of a race faster than the first half.) The main draw is that running the race this way feels better both mentally and physically: if you speed up later in the race, it means you’ve conserved enough energy to continue feeling strong (rather than “hanging on for dear life”), and you’ll also pass people who are slowing down, which is a great mental boost.
However, as highly as many runners regard negative splitting, there isn’t much data out there to indicate if it will actually result in a faster race time—and if it does, who tends to succeed at it.
A look at marathon world records
It’s a common assumption that negative splitting is the “best” pacing strategy for running distances a mile and longer. However, a study recently published in the European Journal of Sports Science (EJSS) tested that assumption and found that even pacing might be the best strategy for running a marathon. 
In their article, researchers at University of the Basque Country UPV/EHU and University of Huelva examined men’s marathon world records set between 1967 and 2018. They divided all of the marathons into 5km segments and then calculated the relative and average speeds of each segment. They found that the pacing strategies employed by world record-setters in 1967–1988 differed from strategies employed in 1988–2018: the earlier runners (whom the researchers termed “classic athletes”) actually started their races faster and lost speed around 25km (a phenomenon called “positive splitting”), whereas the later runners (“contemporaneous athletes”) demonstrated more consistent pacing across all segments of the race.
These findings don’t necessarily indicate that negative splitting is a bad strategy . . . only that it wasn’t the strategy most commonly used by world record marathoners—or at least male ones. Researchers at the Institute of Sport and Exercise Science, University of Worcester looked at results from the 2009 IAAF Women’s World Marathon Championships and found that the top 25% of the field ran the first 10km (6.2 miles) of the race at a pace that was relatively slower than their average, and then increased their speed above average after 35km (21.7 miles).  Moreover, a quick look at Paula Radcliffe’s 2003 marathon world  record shows that she negative split by 39 seconds. However, Radcliffe set her record in a race that included men at the start, whereas marathons today separate the men’s and women’s starts (with the women starting first so that they are not aided—or blocked—by men). For these women’s-only races, Mary Keitany holds the world record, and unlike Radcliffe, she set her record with a positive split of over 3 minutes (1:06:54 for her first half and 1:10:07 for her second). 
What about non-elites?
Despite Keitany’s example, studies show that women—at least non-elite women—are better at even pacing and less likely to positively split than men. One study that looked at data from 14 US marathons in 2011  found that regardless of age, finishing time, and even running experience, women were more likely to maintain their running pace throughout a marathon and less likely to slow down in the second half. The authors of this study speculated that this may have to do with differences in women’s physiology and decision-making, while another paper that specifically studied runners of the 2013 Houston Marathon  found that men’s overconfidence led to the discrepancy in pacing strategies.
More recent data from the 2017 Berlin marathon  showed similar pacing results in terms of gender: women stayed closer to their average pace than men throughout the race, and more men ran large positive splits than women. However, these results also revealed another factor that correlated with pacing: relative speed. Boston-qualifying runners (i.e., those who are fast for their age) tended to slow down during segments of the race by only 6–8% of their average pace; non-Boston-qualifying runners, on the other hand, slowed down by 11–15%. While actual speed probably doesn’t have anything to do with one’s ability to keep a steady pace, being faster than one’s peers may be an indication of having more experience, which could explain the ability to pace evenly (or even negative split).
Another potential indicator of experience—and therefore the ability to pace evenly—is age. A study of athletes who ran the New York City (NYC) Marathon between 2006–2016  found that older runners paced more evenly than younger ones. (Like the Berlin results, these results also showed a correlation between speed and pacing strategy, with faster runners—who may have more experience—pacing more evenly than slower runners.)
One final factor that appears to be correlated with pacing strategy is nationality. In a separate analysis of the same 2006–2016 NYC Marathon data, researchers found that Ethiopians and Kenyans demonstrated more even pacing strategies than other nationalities, while Americans were most likely to put in an “end spurt,” i.e., accelerating at the very end of the race.  The researchers concluded that the influence of nationality on pacing strategy might have to do with performance level of a given nationality (presumably implying that Kenyans and Ethiopians are better runners and therefore more capable of pacing evenly than runners from other nations), as well as running culture and common tactics of a given nation.
Pacing strategies across running events
So age, gender, relative speed, and nationality all correlate to some degree with the ability to pace evenly, which, according to the study in EJSS, might be the optimal pacing strategy for a marathon. But does this hold true across other non-sprint running events?
Evidence points to “no.” In a study examining optimal 5k running performance, researchers found that runners (all female, n = 11) achieved their best time by running the first 1.63km (~1 mile) 3-6% faster than their average race pace.  A study of 24 male runners found that 10k pacing  depended upon running ability, with faster runners starting out faster than their average pace, and slower runners actually pacing more evenly at the start.
Suffice to say, more research remains to be done. In fact, the authors who analyzed men’s marathon world records are working on a follow-up study comparing male and female marathon world records. It will be interesting to see if they find gender-based differences and, if they do, what they are able to conclude about these differences. In the meantime, the debate over the “best” pacing strategy will continue.
 Díaz, J. J., Fernández-Ozcorta, E. J., & Santos-Concejero, J. (2018). The influence of pacing strategy on marathon world records. European Journal of Sport Science. http://doi.org/10.1080/17461391.2018.1450899
 Renfree, A. & St Clair Gibson, A. (2013). Influence of different performance levels on pacing strategy during the women’s world championship marathon race. International Journal of Sports Physiology and Performance, 8(3), 279–285. https://doi.org/10.1123/ijspp.8.3.279
 Radcliffe Runs 2:15:25 in London! (13 April 2003). Retrieved from https://www.iaaf.org/news/news/radcliffe-runs-21525-in-london
 Doyle, M. (23 April 2017). Mary Keitany shatters the women’s-only world record at the London Marathon. Canadian Running. Retrieved from https://runningmagazine.ca/mary-keitany-shatters-womens-world-record-london-marathon/
 Deaner, R. O., Carter, R. E., Joyner, M. J., & Hunger, S. K. (2015). Men are more likely than women to slow in the marathon. Medicine & Science in Sports & Exercise, 47(3), 607–616. http://doi.org/10.1249/MSS.0000000000000432
 Hubble, C. & Zhao, J. Y. Hubble. (2015). Gender differences in marathon pacing and performance prediction. Retrieved from http://dx.doi.org/10.2139/ssrn.2516575
 Snyth, B. (2017). The Berlin Marathon, 2017: An analysis of participation and performance at the Berlin Marathon 2017. Retrieved from https://medium.com/running-with-data/the-berlin-marathon-2017-8d5f7aad1241
 Nikolaidis, P. T. & Knechtle, B. (2017). Effect of age and performance on pacing of marathon runners. Open Access Journal of Sports Medicine, 8, 171–180. https://doi.org/10.2147/OAJSM.S141649
 Aschmann, A., Knechtle. B., Onywera, V., & Nikolaidis, P. T. (2018). Pacing strategies in the New York City Marathon – Does nationality of finishers matter? Asian Journal of Sports Medicine, 9(2), e57596. http://doi.org/10.5812/asjsm.57596
 Gosztyla, A. E., Edwards, D. G., Quinn, T. J., & Kenefick, R. W. (2006). The impact of different pacing strategies on five-kilometer running time trial performance. The Journal of Strength & Conditioning Research, 20(4), 882–886. http://doi.org/10.1519/R-19275.1
 Lima-Silva, A. E., Bertuzzi, R. C. M., Pires, F. O., Barros, R. V., Gagliardi, J. F., Hammond, J., Kiss, M. A., & Bishop, D. J. (2009). Effect of performance level on pacing strategy during a 10-km running race. European Journal of Applied Physiology, 108(5), 1045-1053. https://doi.org/10.1007/s00421-009-1300-6