Key findings

  • -0.0003 vs 0.0006average annual improvement rate after the pandemic, females versus males, a difference significant at less than the one-percent level
  • 0.0009 per yearfemales’ average improvement rate from 2010-2019, which was 0.0005 per year faster than that of males
  • 0.2%less improvement per year for the eleventh-ranked US female swimmer than for the top-ranked female in the post-COVID era, from the female, ranking and era interaction
  • 100regressions in each of two sets with synthetic height and long COVID variables, which did not have significant explanatory power

Abstract

The purpose of this study is to measure how performances of elite US swimmers have fared since 2021, when the effects of the COVID-19 pandemic started to subside in the United States. Findings suggest females’ rate of improvement has slowed relative to that of male swimmers since the pandemic. This change is an about face from improvement rates in the years before the pandemic, with the largest share of the reduction occurring among the lower ranked set of elite female swimmers. Attempts are made to explain these relatively lower improvement rates with swimmers’ ages, differences in heights, which might leverage the ability to use new racing techniques, and exposure to symptoms related to long COVID. These covariates contain little in the way of explanatory power, leaving females’ slower rates of improvement in the current era a mystery that might be solved with future paths of research that focus on equity to innovations of the sport or effects specific to strokes or events.

Introduction

Like nearly all else, the sporting world’s routines were altered dramatically by the COVID-19 pandemic in the spring of 2020. In the United States, the National Basketball Association’s season was postponed and finished later in the year in Orlando, while Major League Baseball played an abbreviated schedule, with games occurring in nearly empty stadiums. In the United Kingdom, the Premier League’s season was postponed in March 2020 and finished in June. The Wimbledon tennis tournament and the Open Championship were not conducted in 2020, the first time since 1945 in which neither event was conducted.

Similarly, the 2020 Olympic Games were scheduled to start in Tokyo in late July of that year, but the International Olympic Committee (IOC) decided to postpone the Games, which eventually took place a year later. McFall and Whitehead1 surveyed US swimming coaches in 2021 to show that the pandemic created problems with the training schedules of elite swimmers who were optimizing training schedules around making their country’s Olympic team and shrunk dramatically access to training spaces, which in turn affected athletes’ opportunities to train as they wished in preparation for either the US Olympic Trials or the Olympics. The result is that less than 10% of swimmers ranked in the top 20 of the 2021 Olympic Trials performed better than their forecasted performance. These findings on reduced performance capacity are similar to findings on the effects the pandemic had on academic performance by Kofoed et al.2 and Mervosh3 and on mortality reductions in Chen and Krieger4.

This paper follows McFall and Whitehead1 and explores elite US swimming in the five years following the pandemic to determine how performances of this cohort changed since the interruptions of the pandemic. To perform this analysis, I collected data on the top 16 performers for the years 1999-2026 in the 13 individual swimming events that have been conducted at every Olympic Games during this century.i I define three eras of swimming, based on exogenous factors that impacted the sport, that roughly adhere to the decades of the century and compare how improvements have changed across these eras.

The headline result of this analysis is that since the pandemic females’ rates of improvement have fallen behind that of male swimmers, a turnaround from the decade of swimming that took place prior to the pandemic. Females’ average improvement rate, by ranking, was 0.0009 per year (0.09%) from 2010-2019, which is 0.0005 per year faster than the same rate for males, a difference significant at the 1% level. In contrast, males are improving at an average rate that is 0.0009 per year faster than females after the pandemic, a difference significant at less than the 1% level.

Trying to find explanations as to why this turnabout has happened is the chief concern of this paper, and three possible explanations are entertained. The first is the age composition of swimmers in the dataset. The second is the height differences between male and female swimmers, which might be leading to disparate impact across the sexes of leveraging fully new swimming techniques. Third, the possibility that males were less prone to be affected by long COVID symptoms than females. The data for swimmers’ ages are observed directly in the data, whereas I use synthetic data on how height and post-pandemic health might explain improvement differences.

To proceed, in the next section, I provide background information about elite swimming in this century to provide context for the post-COVID time period. Next, I describe the methods and data I employ to conduct my analysis. In the fourth section, I present the findings of my analysis, which shows that females’ performances at lower ranking levels are behind similarly situated performers from previous years. I attempt to explain improvement patterns with swimmers’ age and random variables on athletes’ heights and incidence of long COVID. I offer concluding remarks in the last section.

Background on Elite Swimming in the 21st Century

Like nearly all sports, swimming has evolved dramatically in the 21st century, as technological innovations and knowledge collection have spurred training, technique, and equipment changes that have helped alter swimmers’ improvement paths. The first decade of the century saw techniques used by swimmers change dramatically, with the widespread adoption of using more fully the advantages of swimming underwater whenever possible being the most important change.

McFall, Griffith, and Rotthoff5 discusses the implications of underwater swimming, showing that performances in backstroke and butterfly, the two strokes in which this technique was employed most fully, improved more rapidly compared to other strokes. The first swimmers to employ underwater kicking after starts and turns competed in the early 1980s. One of the first American swimmers to train for the use of underwater swimming was the David Berkoff, 1988 Olympic silver medalist in the 100 backstroke. He altered his training considerably from the status quo, concentrating on developing leg strength and increasing his anaerobic capacity over the then more common method of dense aerobic training. As McFall, Griffith, and Rotthoff5 attest, after he set a world records in the event in 1988, several swimmers mimicked his racing and training techniques, which had him dolphin kicking underwater for the first 40 meters of the opening half of the 100-meter race and half of the remaining distance on the second length of the race.

In 1998, the governing body of aquatic sports, FINA (now World Aquatics), outlawed swimming underwater for more than 15 meters after starts and turns in the backstroke, butterfly, and freestyle races, but the use of the technique did not stop with the rule change. By the end of the 1990s, American swimmers like Misty Hymen and Natalie Coughlin were training with the intent of maximizing the returns from swimming underwater for relatively long distances during races. However, the universal adoption of dolphin kicking as a race strategy and training for the use of dolphin kicking was not seen until well into the 2000s, as Mortensen6 discusses.

Additionally, late in the first decade of the 21st century, swimsuit manufacturers started to introduce new competitive suits that aided swimmers relatively more than suits worn previously. These suits not only were made with materials that provide less friction to water than shaved human skin, the most advanced of the suits, the LZR Racer also improved swimmers’ buoyancy, thereby lowering the cost of the act of swimming. These suits were introduced in 2008 and banned in 2010. As McFall, Griffith, and Rotthoff5 discuss improvement rates for the two years in which these suits could be worn outpaced rates swimmers achieved in all other years in this century. In fact, at the time of this article’s publication, two world records set in 2009 still exist on the men’s side of the record board.

After 2010, the swimsuits worn by all swimmers started to resemble more those worn in 2005 than in 2009, and improvements came from the application of technological innovations like low-cost video recording and data collection methods, both of which enhanced every aspect of swimming performance.

The next era, which I define as occurring between 2010 and 2019, is centered around the cost reductions of information gathering and sharing. To this end, consider first the video recording. In 2000, the video-sharing website YouTube was still five years from launching, as discussed in Rana,7 and Apple’s iPhone 4, the first iPhone with a video camera, was ten years from hitting the market, as discussed in Ekuwem.8 The wide-spread dissemination of these innovations made it more possible for swimmers to see themselves and elite swimmers in action. Now, it is second nature for swimmers to watch their races or practices to detect opportunities for improvement while also watching elite swimmers’ techniques to mimic. This new feedback loop created a better environment for improvement for swimmers who could use visual information to their advantage.

At the same time costs of watching swimming fell, data collection methods were improving, thereby creating environments within sport for more extensive testing of training methods and techniques. In the simplest form, data observed by a coach on an aspect of a swimmer’s performance could be entered into a spreadsheet, and these data could be used to explain the good and bad of a swimmer’s performance. More complexly, tracking systems can monitor swimmers’ velocities at various junctures of a race or a stroke cycle or heart rate so that coaches can understand levels of strain imparted from workouts. Innovators in this space include Ken Ono, who has worked with Emory University’s and the University of Virginia’s swimming teams, as was detailed in Kamran,9 and Russell Mark, who consults with national teams on ways to use data to monitor technique and was discussed in Reid.10 Of course, these innovations might not impact all swimmers for various reasons, but this study is about elite swimmers, thus it seems safe to assume the likelihood of this set of swimmers being impacted in some way by the data revolution seems rather large. Further discussion of these changes can be found in Douglass et al,11, Law et al,12 and Goodwin et al.13

The decade of information applications, which I characterize as 2010-2019 in this study, ended suddenly in the spring of 2020, when the full scope of the COVID-19 pandemic became known. At that time, massive segments of the United States were shut down, including facilities at which swimmers trained. These closures kept swimmers from executing their normal training routines for the bulk of 2020, as discussed in McFall and Whitehead.1 The 2020 Olympics swimming competition was held in front of nearly an empty stadium in the summer of 2021 in Tokyo, a symbolic beginning-to-the-end to the restrictions that elite swimmers faced due to the pandemic.

The main question of this study is how well elite swimmers have responded to a more normal world of training options since the pandemic ended. In one sense, one might expect swimmers’ improvements to slow relative to other years because they need time to make up for lost months of training. Unbeknownst to us, many swimmers could have fallen ill with COVID and suffered setbacks from the sickness. Indeed, Race13 discusses the struggles Hungarian superstar Kristof Milak has endured since he contracted COVID-19. On the other hand, it is possible the pandemic created a churn effect within the sport, expediting the exodus of older swimmers while new, younger athletes take their place. It is possible a younger generation might bounce back from the pandemic-related constraints of training to improve more rapidly than previous cohorts of swimmers.

Materials and Methods

I collected from the World Aquatics website the top 16 US performers in the 26 individual swimming events that have been conducted at every Olympic Games this century. (Thus, the males’ 800-meter freestyle and the females’ 1500-meter freestyle races are not part of this dataset.) The resulting repeated cross-section of swimming performances contains 27 years (1999-2019, 2021-2025) of top-16 rankings in 26 events, for a total of 10816 observations. The performances for the 2020 season are excluded from the final dataset because swimmers were unable to perform at a meet that season in peak condition due to the pandemic. Swimmers who tied with one another were randomly assigned the higher ranking and the immediately lower ranking. Finally, the 16 best performers were selected because this number reflects the sport’s custom of granting spots in the consolation and championship final heats or to semifinal heats to the best 16 performers from the preliminary heats. Thus, I am assuming that the 16 swimmers for each year/race combination had an occasion to swim that race as fast as possible in that year.

The data from World Aquatics contain information on the swimmer’s name, age, and the official time, presented in minutes, seconds, and hundredths of seconds, of their performance. I converted the presented performance time to the number of seconds. Table 1 shows important summary statistics for each of the 26 events.

Table 1. Summary Statistics of Top 16 U.S. Performers, by Event (1999-2025)

FemalesMen
EventObservationsMean Seconds (sd)MinMaxObservationsMean Seconds (sd)MinMax
50 freestyle41625.18 (0.49)23.9126.2541622.27 (0.44)21.0423.26
100 freestyle41654.73 (0.98)52.0458.8941649.02 (0.84)46.8150.98
200 freestyle416118.78 (1.87)113.61122.73416107.94 (1.73)102.96111.78
400 freestyle416249.82 (3.81)236.46257.30416230.47 (3.12)222.78237.59
800/1500 freestyle416513.55 (8.06)484.12526.95416917.79 (14.69)870.67950.97
100 backstroke41660.99 (1.53)57.1364.1041654.40 (1.15)51.8557.16
200 backstroke416131.36 (3.01)123.35137.22416118.62 (2.33)111.92123.88
100 breaststroke41668.36 (1.56)64.1372.0041661.19 (1.33)58.1464.26
200 breaststroke416147.84 (3.10)136.74153.75416132.80 (2.86)126.54138.95
100 butterfly41658.90 (1.19)54.6061.7641652.62 (1.13)49.4554.91
200 butterfly416130.26 (2.18)123.84135.15416117.85 (2.13)111.51122.15
200 individual medley416133.40 (2.73)125.87139.08416120.49 (2.768)114.00126.35
400 individual medley416282.87 (4.89)271.12294.19416257.78 (5.06)243.84268.40

Since I am most concerned with elite swimmers’ rates of improvement, I calculate the variable percent improve for each year/ranking combination possible in the dataset. Percent improve is calculated by differencing the time of performance in year t+1 from the time of performance in year t and dividing this difference by the reference time from year t, for the same ranked performers across consecutive years. Thus, when percent improve is positive, the similarly ranked swimmer from the more recent year was faster than their counterpart from the prior year. As an example, in 2013, the 12th ranked 800-meter freestyler among female swimmers in the United States was Leah Smith, who clocked a time of 8 minutes 34 seconds and 34 hundredths of a second (514.34 seconds). In 2013, the 12th ranked swimmer in the same event was Sierra Schmidt, whose time was 8:34.58 (514.58 seconds). Across these two years, the value of percent improve is -0.00047.

Table 2 presents five different summary statistics of percent improve for both sexes in all three eras. The row for all performances is the average of perfect improve for all 16 performers, while the other rows show the average for various, specific rankings.

Table 2. Average Annual Improvement Rates, by National Ranking, for Elite U.S. Swimmers (1999-2025)

FemalesMen
RankingPre-2010 (N, sd)2010-20192021-2025Pre-20102010-20192021-2025
Overall Average0.0029 (2080, 0.0080)0.0009 (2080, 0.0073)-0.0003 (1040, 0.0070)0.0032 (2080, 0.0029)0.0004 (130, 0.0073)0.0006 (1040, 0.0053)
1st Ranked0.0032 (130, 0.0131)0.0007 (130, 0.0109)0.0015 (65, 0.0082)0.0038 (130, 0.0089)-0.0007 (130, 0.0101)0.0009 (65, 0.0072)
5th Ranked0.0026 (130, 0.0075)0.0009 (130, 0.0008)0.0007 (65, 0.0081)0.0038 (130, 0.0081)-0.0001 (130, 0.0081)0.0007 (65, 0.0050)
8th Ranked0.0029 (130, 0.0070)0.0011 (130, 0.0006)-0.0004 (65, 0.0065)0.0032, (130, 0.0075)0.0006 (130, 0.0070)0.0006 (65, 0.0049)
16th Ranked0.0011 (130, 0.0061)0.0013 (124, 0.0061)-0.0019 (65, 0.0060)0.0026 (130, 0.0063)0.0001 (130, 0.0059)-0.0001 (65, 0.0056)

The inclusion of age in the World Aquatics data allows for the analysis of the potential impact that an age cohort has on improvement, which is an update from McFall and Whitehead.1 However, understanding the way improvement is linked to the evolution of the sport in this century begs for the gathering of other correlates. In particular, because females improved more rapidly than males in the second era I define, height differences were possibly linked to these performance differences. Changes to techniques, for instance, could have been more easily utilized by females than males. To explore this possibility, I created two random variables for height based on two different distributions, one from NHANES14 measurements of a general population of US females and the other from Rogers15 observations of 2012 Olympic swimmers’ heights. For females, the standard population distribution has a mean of 162.5 cm and a standard deviation of 9.99 cm, while 172.9 cm and 6.97 cm characterizes the Olympic distribution. The same values for males are 175.8 cm and 9.52 cm, and 185.7 cm and 7.68 cm, respectively. With these properties, I created the random variables height and Olympic_height, and female_height and female_Olympic_height, all distributed normal, by interacting the random height variables with a female-specific indicator variable.

Finally, an explanatory variable is needed to describe the way exposure to the COVID-19 virus might have correlated to swimming performance in the years following the pandemic. Of course, no data exists that identify specific instances in which the elite swimmers in the dataset might have been exposed, so I turned to the conclusion of Maglietta,16 a meta-analysis that summarizes how long COVID symptoms differ across sexes. The conclusion of the study is that females who had contracted COVID-19 were 1.54 times more likely than males to suffer from symptoms related to long COVID, with an assumed likelihood of long COVID symptoms for females at 0.49 and males at 0.37. Based on these likelihoods, I created a random binomial variable, random_COVID, to represent the chance elite swimmers had to endure a bout with long COVID. As before with, I interact this variable with the female indicator to create a female-specific measure.

To measure the impact time, age, height, and long COVID exposure had on percent improve, measured in ranking, sex, and time dimensions. I start by using data observed in the World Aquatics rankings, namely time, age, and sex. I use equation 1 to analyze the correlation between improvement and observed correlates.

(1) Percent improverst = constant + β1*rank + β2*age + β3*sq_age + β4*female + β5*era2 + β6*era3 + Γ(rank*era interactions) + Ω(age*era interactions) + Δ(sq_age*era interactions) + Φ*female*rank + γ*female_age + η*sq_female_age + Θ(female*era interactions) + Ν(female*rank*era interactions) + Τ(female*age*era interactions) + Υ(female*sq_age*era interactions) + ε, where ε ~ N(0, σ2).ii

Results and Analysis

First, note that the average improvement of male swimmers is the same in the years preceding and following the pandemic. Before the pandemic, the average annual improvement rate was 0.0004, which is only 0.0002 per year less than the average post-pandemic rate. This difference is statistically insignificant. (t-test = 0.54, 3129 df) The same cannot be said of females, though. For them, the average improvement in era 2 was 0.001, a rate that is 0.0012 faster than the post-pandemic average. This difference is statistically significant at less than the 1% level. (t-test = 4.52, 3113 df)

Additionally, the improvement rates across the sexes are different from one another, pre- and post-pandemic. In era 2, the female average improvement rate is 0.0005 greater than the male improvement rate, a difference that is statistically different from zero at the 2% level. (t-test, 2.36, 4,158df) A glance at the way rates of improvement unfold across rankings in this era shows that the top-rated females improved more rapidly than did similarly ranked males during this era. Salient stories about swimming would reinforce this observation, as the likes of Katie Ledecky, Simone Manuel, and Lilly King stepped to the fore during this decade and rewrote the American record book. In comparison, of the great American swimmers of this decade, only Ryan Lochte and Ryan Murphy stand out as pushing already existing performance boundaries of the sport. Michael Phelps, as great as he was in the second era, was past his legendary performances of the first era.

Just as quickly, though, females’ performances slowed relative to males’ performances in the years after COVID, starting with 2021. As shown in Table 2, females’ average overall rate of improvement was -0.0003, while males’ rate was 0.0006. The two-tailed t-test for a difference between these rates is statistically significant at less than the one-percent level. (t-test = 3.11, 2078 df)

The regression results from equation 1 shown in Table 3 start to explain the post-COVID difference, as rates of improvements are inversely correlated with the numeric ranks of female swimmers. The coefficient estimate on the variable female_rank_era3, the interaction of the female indicator, ranking, and the era3 indicator, is -0.00023, an estimate that is significant at less than the 1% level. To provide context, this estimate means the average rate of improvement of the eleventh-ranked swimmer in the US is, in the post-COVID era, about 0.2% less per year than the top-ranked female.

Table 3. Equation 1 Regression Results- Explaining Annual Improvement Rates (standard errors in parentheses)

(1)
VARIABLESRegression Results: Explaining Improvement Rates
rank1-3.99e-05
(3.50e-05)
age0.000194
(0.000405)
sq_age1.28e-06
(8.87e-06)
female-0.00730
(0.00525)
era2-0.000110
(0.00619)
era30.00666
(0.00925)
rank_era20.000149***
(4.99e-05)
rank_era3-3.03e-05
(6.08e-05)
age_era2-0.000180
(0.000532)
age_era3-0.000604
(0.000801)
sq_age_era2-8.02e-08
(1.14e-05)
sq_age_era38.45e-06
(1.73e-05)
female_rank6.65e-05
(4.96e-05)
female_age0.000609
(0.000464)
sq_female_age-1.26e-05
(1.03e-05)
female_era20.00960
(0.00748)
female_era3-0.00333
(0.0103)
female_rank_era2-0.000123*
(7.06e-05)
female_rank_era3-0.000224***
(8.60e-05)
f_age_era2-0.000739
(0.000652)
f_age_era30.000384
(0.000886)
sq_age_female_era21.60e-05
(1.43e-05)
sq_age_female_era3-8.73e-06
(1.91e-05)
Constant-0.00131
(0.00463)
Observations10,355
R-squared0.041

Standard errors in parentheses

*** p<0.01, ** p<0.05, * p<0.1

What explains this post-COVID about face of overall improvement of females and lower-ranked females when compared to males?

Possible explanations are that the age cohort of elite female swimmers relative to elite male swimmers changed across the pandemic boundary, that training or racing techniques that have been adopted in the post-pandemic era have benefited elite males over females because of general differences across body dimensions of the sexes, and perhaps female swimmers have been more affected by COVID than their male counterparts.

The impact that age might have on performance rates is an explanatory candidate because in this century elite swimmers have more financial opportunities to help them remain in the sport longer than previous generations of swimmers. The average age of females ranked in the top four nationally has increased from 20.7 years to 21.7 years, from the 2000s to the post-COVID 2020s, and illustrates well how expanded opportunities to continue competing have altered the overall age of a competitive cohort. The very best swimmers continue to compete at elite levels beyond college, so it is very possible the most elite females are now finding new plateaus that were not as likely to be reached at the start of the century. These findings are consistent with the record-breaking performances of Ledecky, Kate Douglass, and Gretchen Walsh, all females who are beyond college yet still setting new records and improving, in the post-COVID world.

Additionally, the average age of female swimmers ranked 13th to 16th in these data is about a year less than the average age of the four best swimmers, at all eras defined in the data. It is certainly possible the lower-ranked swimmers face steeper financial pressure to drop out of the sport compared to their highly elite counterparts, and so this quality of swimmer perpetually never can hone their skills at older ages, therefore leaving annual improvement to be more random compared to the very highest ranked swimmers.

The trouble with gravitating towards using age as an explanation for reduced improvement is that the coefficient estimates that include age show no correlation between improvement and age, in the post-COVID era or otherwise.

As for the explanatory power of other covariates, obtaining heights for the swimmers in the dataset is too arduous a task given the uncertainty of the variable having explanatory power. In lieu of this collection, I created random variables from two different populations, general and Olympian, of adult heights as described above. I added to equation 1 a random representation of height based on the swimmers’ sex and interacted this variable with the female indicator variable to create the variable female_height.

Obtaining data on the swimmers’ medical history is probably even more difficult than obtaining data on their height, so I turned to the findings in Meglietta16 to model the potential disparate impact of long COVID to the sexes. To equation 1 I added covid, a random binomial based on reported infections in the meta-analysis. I interacted covid with a female indicator to obtain female_covid to equation 1.

I added to equation 1 these four synthetic variables and then ran two sets of 100 regressions, one set with the standard height distribution and the other set with the Olympic distribution. From these regression results, I averaged the coefficients for the four random variables and counted the number of occasions in the 100 regressions in which the p-value of the coefficient measure reached the 5% or the 10% thresholds. Table 4 presents the results of these calculations.

Table 4. Average Coefficient Estimates and Instances of Significance of Height and COVID

Using 5% Significance as Reference Point
mean heightmean femalemean height2mean female2mean height3mean female3mean covidmean female covid
Standard Height-0.00002 (36)0.00003 (36)-0.00006 (36)0.00005 (36)-0.00001 (36)0.0005 (36)0.0001 (35)0.0006 (35)
Olympic Height0.00002 (39)-0.00004 (40)0.00003 (39)-0.00001 (41)-0.00001 (39)0.000007 (40)-0.0002 (40)0.0006 (40)
Using 10% Significance as Reference Point
mean heightmean femalemean height2mean female2mean height3mean female3mean covidmean female covid
Standard Height0.00002 (66)-0.00003 (66)0.00008 (66)-0.00001 (66)0.000006 (67)0.000006 (67)0.0006 (66)0.0006 (72)
Olympic Height0.00002 (72)-0.00004 (72)0.00003 (71)0.000004 (73)0.00001 (71)0.000008 (73)-0.0002 (72)0.0006 (68)

The only reasonable conclusion can make from these results is that these measures do not have significant explanatory power to explain the mystery of performance differences. The number of occasions in the set of regressions the p-values are measured at traditionally significant levels falls outside of the bounds of what one would expect from the repetition of such measures. For instance, the number of occasions in which the p-value of the coefficient estimate on female_covid was less than 5% was 36 in the set of regressions that used the general height distribution. (Top row of Table 4.) For context, an event presumed to have a 95% likelihood of occurrence happening 36 of 100 times is 1.7x10-57. This measure, or any of the other three long COVID random variables, simply do not suggest problems with COVID explain females’ lagging improvement rate.

Analyzing data is often as good as the available data are, and in this instance, crucial insights about the sport could be unobservable due to data collection limitations. Swimming has gone through large changes in this century, and the benefits of those changes, especially in the case of females, are massing at the very top end of the sport. Perhaps the simple explanation of the pattern of improvement observed in the post-COVID era is that there are three uniquely great performers in Ledecky, Douglass, and Walsh that are blazing paths in the sport that no one currently competing can follow. Unfortunately, a well-identified reason for this performance lag is beyond this paper.

The fact remains that it is entirely possible that changes to training strategies or racing techniques comport better to a broader range of males than females, and now rates of improvement from these changes are showing in performance data. Perhaps other information about physical differences across the sexes might explain better the divergent rates of improvement in the post-COVID world. Similarly, it is possible that exposure to the COVID-19 virus actually does explain performance differences, but the measure I use simply does not explain those differences. Finally, the slower rates of improvement could be an equity issue, because only the smallest, most elite slice of swimmer has available to him or her the benefits of cutting-edge technologies.

Conclusion

This study follows the top 16 US swimmers from 1999-2025 to measure how well swimmers have rebounded from slower than expected improvement immediately following the COVID-19 pandemic. Results show lower-ranked, female swimmers are improving at historically low rates, compared not only to their male counterparts but also to similarly situated swimmers from years preceding the pandemic. I attempted to explain this relatively low rate of improvement with the age of swimmers and with synthetic controls of height and likelihood of long-term effects from exposure to COVID-19. None of these covariates offered reasonable explanatory power.

Future research might find improvement rate differences explained by event-specific or stroke-specific effects, team-specific effects, or competition-location effects.

Declarations

Acknowledgement
Chinmay Machavaram provided assistance collecting these data.

Notes

i Another way of describing the data is that we excluded the female 1500-meter freestyle and the male 800-meter freestyle because those events were first conducted at the 2021 Olympics.

ii Lower case letters represent single coefficient estimates and upper-case letters represent vectors with dimensions that depend on the number of variables created from interactions.

References

  1. McFall, T., & Whitehead, J. “Measuring the Impact of the COVID-19 Pandemic on Elite Swimming Performance,” Journal of Sports Economics, 25(5): 634-656.
  2. Kofoed, M. & Gebhart, L. & Gilmore, D. & Moschitto, R. "Zooming to Class?: Experimental Evidence on College Students' Online Learning during COVID-19," IZA Discussion Papers 14356, IZA Network @ LISER.
  3. Mervosh, S. “The Pandemic Hurt These Students the Most,” New York Times, July, 28, 2021, https://www.nytimes.com/2021/07/28/us/covid-schools-at-home-learning-study.html.
  4. Chen JT, Krieger N. “Revealing the Unequal Burden of COVID-19 by Income, Race/Ethnicity, and Household Crowding: US County Versus Zip Code Analyses,” Journal of Public Health Management and Practice. 2021 Jan/Feb; 27 Suppl 1, COVID-19 and Public Health: Looking Back, Moving Forward:S43-S56.
  5. McFall, T., Griffith, A., Rotthoff, K. “The Impact of Technology and Rule Changes on Elite Swimming Performances,” The Economics of Aquatic Sports. Springer Publishing, Cham, Switzerland. 2020.
  6. Mortensen, J.P. How the Underwater Dolphin Kick Evolved over Time and Revolutionized the Sport. Swimming World. January 20, 2023.
  7. Rana, R. “A History of YouTube- How It All Began & Its Rise,” VdoCipher. June 15, 2025. https://www.vdocipher.com/blog/history-of-youtube/
  8. Ekuwem, A. “The Evolution of the iPhone’s Camera from 2007 to 2025,” Slashgear, March 2, 2025. https://www.slashgear.com/1798246/iphone-camera-evolution-specs-differences-2007-2025/
  9. Kamran, W. “Ken Ono: A Renaissance Man,” UVA Research News. https://research.virginia.edu/news/ken-ono-renaissance-man
  10. Reid, W. “Hoo-rizons: Former UVA Swimmer is Team USA’s ‘Secret Weapon’,” USA Today, March 6, 2020. https://archive.news.virginia.edu/content/hoo-rizons-former-uva-swimmer-olympic-teams-secret-weapon/
  11. Douglass, K., Lamb, A., Lu, J. et al. “Swimming in Data. Math Intelligencer 46, 145–155 (2024).
  12. Law, Barbi & Post, Phillip & McCullagh, Penny. Video-based observation in sport: From “forgotten” to ubiquitous. Journal of Sport Psychology in Action. 9. 1-11.
  13. Goodwin ML, Harris JE, Hernández A, Gladden LB. Blood Lactate Measurements and Analysis during Exercise: A Guide for Clinicians. Journal of Diabetes Science and Technology. 2007;1(4):558-569.
  14. Dietary Reference Intakes for Energy. Chapter: Appendix L: National Survey Data from United States and Canada. https://www.nationalacademies.org/read/26818/chapter/23
  15. Rogers, Simon. Olympic Athletes by Age, Weight, and Height Visualized. The Guardian. August 7, 2012. https://www.theguardian.com/sport/datablog/interactive/2012/aug/07/olympic-athletes-age-weight-height
  16. Maglietta G, et al. Prognostic Factors for Post-COVID-19 Syndrome: A Systematic Review and Meta-Analysis. Journal of Clinical Medicine. 2022 Mar 11;11(6):1541.

Article information

Authors

  • Todd A. McFallPiedmont Economic Consulting. LLC & Department of Economics, Wake Forest University

How to cite

McFall, T. A., 2026. Attempting to explain why the USA’s elite, female swimmers are improving more slowly than males after the COVID-19 pandemic. Medical Research Archives, [online] 14(10). https://doi.org/10.18103/mra.2026.0605

Record

Journal
Medical Research Archives
Section
Research Articles
Published
31 October 2026
ISSN
2375-1924
Downloads
85

Copyright and licence

Published by the European Society of Medicine under a Creative Commons Attribution 4.0 International licence (CC BY 4.0), which permits unrestricted use, distribution and reproduction in any medium, provided the original authors and source are credited.