by Brina Aug. 1, 2020
There’s a narrative I often encounter about the evolution of the national women’s gymnastics program in the United States. In the early days, there were a very small number of gymnastics clubs that could get you to the Olympics, creating a sort of "club of clubs." Gym hopping was rampant: adolescent girls regularly moved across the country to train, and coaches lived in fear that their top athletes would leave for someone with a bigger name.
But as the national team camps and the developmental program grew, so too did the number of top clubs. Camp served as a training opportunity for coaches as well as for athletes, and new coaches learned what it took to train a national team athlete.
So when I came across a list of former national team athletes and their clubs, I saw an opportunity to test the truth of this narrative. The data ranges from 1982 to 2019.
I found that the general trends hold up: national team athletes have become less concentrated at top gyms, and they tend to change clubs less frequently. But when we look more closely at the timing of these trends, we don’t see a sea change in the 2000s accompanying the onset of the current semi-centralized system. The trends are true, but the explanation isn’t.
Gymnasts on the National Team now come from more clubs.
I first calculated the average number of athletes on the national team per club in each year. If the number of athletes per club is high, then top gymnasts are clearly congregating at the same gym.The results are in the plot below.
I also measured the concentration of gymnasts at top clubs using the Herfindahl-Hirschman Index (HHI). This metric comes from economics, where it’s used to measure industry concentration: you calculate the market share of each firm in an industry, square it, and add up the results for each firm. Similarly, I calculated the share of national team athletes at each club in each year, squared those numbers, and added them up. The metric ranges from 0 to 1, with higher values indicating more concentration of gymnasts at a small number of clubs.
The HHI is a nice metric because it’s very sensitive to extremely large, dominant clubs. Consider two cases: 1) five clubs each send two athletes to the national team, and 2) one club sends six athletes to the national team, and four other clubs each send one athlete to the national team. In both cases, there are ten athletes and five clubs, giving us an average of two gymnast per club. However, in the second case, one club is much more dominant than the others. The HHI will reflect this fact: the first situation has an HHI of just 0.2, while the second situation has an HHI of 0.4.
The results are in the plot below. As we can see, the overall trend still exists: gymnasts were much more concentrated at top clubs in the early years of US gymnastics. In 1985, the year with the highest HHI on record, 8 of the 28 gymnasts on the national team came from SCATS, the gym of child abuser Don Peters.
However, we see that the HHI has remained relatively stable since around 2000, if not earlier. In other words, it seems as though the era of extremely high concentration of top gymnasts ended well before national team training camps and the developmental program became institutionalized.
Taken together, these two metrics confirm that the national team now represents a wider range of clubs than it did in the last — but they don’t prove that the USAG pipeline for younger gymnasts is the cause.
More clubs started breaking in before the camp system began.
The narrative gives us a specific explanation for this fall in concentration: the developmental program has supported new coaches, enabling new clubs to bring athletes to the elite ranks. To test this, I looked at the number of clubs who sent an athlete to the national team for the first time in each year, from 1992 onward.1
It’s simply not the case that an unprecedented number of new clubs have sent athletes to the national team since the onset of the developmental pipeline in the early 2000s. In fact, 1997 and 1998 were peak years for clubs to have their first athlete make the national team, with clubs like WOGA, Gym Max,Bart Connor Gymnastics, Gym Cats, and more reaching that milestone in that period.
What happened the late 1990s? Well, for one thing, the flood of young girls who signed up at their local gymnastics clubs after seeing Mary Lou Retton atop the podium in 1984 were becoming old enough to make the national team. My best guess is that the increasing variety of clubs at the highest level has as much to do with the rising popularity of the sport as it does with institutional factors within USAG. There was simply a greater chance that a girl with world-class potential would sign up at her local gym -- even if that gym wasn't already nationally renowned. Of course, this is just a hypothesis: I can’t say for sure if this explains the trends in the graph above. If anyone has historical data on gymnastics participation in the United States going back to the 1980s, send it my way — I’d love to test it out.
The plot shows that a healthy number of clubs are still having their first athlete make the national team — it’s not like the doors have closed. For example, World Champions, First State, ENA Paramus, IGN, and many more have all had their first athlete on the national team in recent years. It’s just not true that drastically more clubs have achieved this goal since the early 2000s.
Club hopping is less common now than it was in the past.
Finally, I counted the percent of national team athletes who changed clubs in each year. Specifically, I counted a gymnast as changing clubs if she represented a different club on the national team in the previous year.
I should note that this only measures club hopping at the highest level — that is, club hopping amongst gymnasts who have already been on the national team. The trend is more often described as young gymnasts who change clubs as soon as their potential becomes clear, hoping for a better shot at “making it.” It also counts gymnasts as club hopping when their club changes names, or when they follow their same coach to a new club. Nevertheless, the trends are informative:
It’s clear that club hopping now is less common than it used to be. In any given year, a smaller share of national team athletes are training a new gym that they weren’t at in the previous year. The correlation between the year and the percent of gymnast who changed clubs is -0.48, and it’s statistically significant at alpha = 0.01.
This trend might be related to the falling concentration of athletes at a small number of gyms. Anecdotally, much of the club hopping of the 1990s and early 2000s occurred when a promising athlete felt she wasn’t getting enough attention due the presence of other top elites in her training group. For example, Kerri Strugg temporarily left the Karolyis after feeling overlooked, and the Rybackis famously played Vanessa Atler and Jamie Dantzscher off each other.
Of course, club hopping isn’t necessarily a bad thing — athletes should feel empowered to change coaches when a training situation isn’t working for them. But if the fall in club hopping is a sign that more athletes feel like they can achieve their goals at the local gym where they grew up, so much the better.
The general narrative is true: national team athletes are no longer so concentrated at a few top clubs, and they’re less likely to hop around during their tenure on the national team. The era of the "club of clubs" is gone. But the common explanation for these trends doesn’t hold up. The data simply doesn’t support the claim that these changes occurred due to the rise of the national team training camps or the developmental pipeline that became institutionalized in the 2000s.
So what does explain the trends? My best guess is that it has to do with the rising widespread popularity of the sport, but I can’t be sure. If you’ve got another idea, drop me a comment or let me know on Twitter!
1These results might contain some errors because my data only goes back to 1982 — it’s possible that some clubs had athletes competing internationally before 1982 but not in the decade between 1982 and 1991. My method would count those as “new” clubs on the scene.
Tags: Fun with Other Data