Sampling is a sensible option, not a proven method for developing sporting talent

The truth lies in the middle: Youth talent programs should resist both exclusive early specialisation and one-size-fits-all sampling approaches

Now before you collect your pitchforks and torches to chase me down, hear me out first.

The claim that young athletes should sample several sports before specialising has intuitive appeal. It promises broader movement experience, less repetitive load and a more forgiving route to eventual excellence. And let's admit, that is something we all want to believe is true. And in part this is suppote by the broader research liteature: early exclusive single-sport involvement is likely not required for elite success in most of the settings studied.

BUT that is not the same as proving that adding sports causes future talent. So, if you and your organisation have been considering encouraging sampling as a viable option for developing talent, this post is for you.

It is my understanding, from the literature, my discussions with coaches, and my discussion with experts in the area (including research methodology), that in the context of the relaionship between sampling and athletic talent, we should be wary of turning a reasonable developmental relationship into a universal causal prescription.

This distinction is important because programmes are often asked to make expensive changes such as redesigning calendars, delaying selection, funding multi-sport access (which in some countries comes with a significant cost to parents), or even restrict sport-specific practice on the assumption that sampling itself produces better talent and ultimately better senior athletes. I know what you are thinking. Sampling's importance may be much broader than in the context of athletic talent alone, I know. But this blog is about the link between sampling and talent. It is inspired by my endless conversations with coaches, including at this week's 'The Coaches Site Live' in Edmonton, Alberta, Canada where I gave the keynote address, and had plenty of excellent conversatuobns with hockey coaches of all coaching levels.

Here is what it comes down to: evidence is stronger against the claim that early specialisation is a prerequisite than it is for the claim that sampling is the active ingredient of talent, but we often conflate the two, resulting in the erroneous interpretation that sampling causes athletic talent. While we dig into the literature later on in this blog, a systematic review of 22 performance studies found no prospective studies, inconsistent definitions, and concluded that youth specialisation was not necessary for elite success. However, this review did not establish that sampling caused later performance (Kliethermes et al., 2020).

A look at some of the evidence

The most substantial evidence in this area compared participation histories of higher- and lower-performing athletes. Across 71 reports and thousands of athletes, higher-performing juniors tended to have started their main sport earlier and accumulated more main-sport practice. Among adult world-class athletes, the pattern was reversed: they had, on average, started later, had more coach-led practice in other sports, less early main-sport practice and slower early progress than national-class peers (Barth et al., 2022). A related review argues that youth pathways can therefore look different depending on whether the outcome is junior selection or adult world-class performance (Güllich et al., 2023). It is important however to consider these reviews in light of their methdogolical artefacts, those interested in this discussion can find more about it here.

Regardless, there seems to be some useful evidence against the notion of “specialise early or fall behind.” Additionally, it seems feasible to infer from the liteature that junior results should not be used as a proxy for adult potential. Yet, we should also acknowledge that this evidence remains largely evidence about associations based retrospective histories and comparisons between athletes who have already arrived at different performance levels. Even when these studies reveal consistent patterns, it does not tell us what would have happened had the same athlete followed a different pathway.

Sampling is likely a marker and not a cause

I learned a lot about causation from Prof Franco Impellizzeri at the University of Technology Sydney, but I cannot claim, by any reasonable measure, expertise in this area. Nonetheless, here is my layman's interpretation, which you can take or leave.

I think there are several credible alternatives can generate the same observed link between sampling and later success, which are often dismissed in favour of the causal explanation for the link between sampling and athletic talent. I think one of the main reasons why that happens, is because of our confirmation bias and a degree of ecological validity in the argument. That means that 1) I think we would love to believe that sampling early matters because most of us in sport have seen the potential first-hand detrimental effects of early specialisation such as burn-out, overuse injury, drop-out from sports etc., 2) the argument makes logical sense. If you sample more activities, you build a broader repertoire of movements, which you can then rely on later when you choose to specialise.

Nonetheless, we should consider these potential alternative mechanisms, which will ultimately reduce the confidence we can have in the causal explanation to be the only, or even most probable, explanation for this phenomenon.

The selection and matching explanation: Children with broad athletic ability, late maturation, supportive families or access to several sports may be both more likely to sample and at the same time more likely to find a sport that suits them. The eventual sport may be a better fit, rather than the earlier sampling having created the talent. The Kliethermes study we linked before explicitly questions whether is is possible that multisport athletes who succeed were already stronger athletes, and calls for prospective work to explore this mechanism further (Kliethermes et al., 2020).

The training load, pressure and programme design explanation: “Specialisation” often bundles together several exposures: a single sport, year-round participation, high training volume, early selection, competitive pressure and reduced autonomy. If an athlete is injured, burns out or leaves sport, the cause for that drop out may in fact be the excessive load they were under and not the fact that they played only one sport. Baker, Mosher and Fraser-Thomas (2020) make this distinction directly, asking whether adverse outcomes of early specialisation might be the result of a single-sport specialisation focus or instead from the intensity and pressure that are often associated with it.

The maturation and selection bias explanation. Early-maturing and relatively older athletes often look better in adolescence, this is well-supported in the broader talent literature. This may result in their earlier selection into squads. This exposes these athletes to more specialist coaching, even if that 'early' advantage ultimately fades by adulthood. A review of participation patterns identifies that both biological maturation and relative age are mechanisms that can inflate early youth performance and thus shape entry to talent programmes. If programmes reward these short-term advantages (and if you are a coach reading this, you know many of them do), the apparent superiority of one development pathway could in part reflect whom the system selected, not what was produced by the pathway they were on.

Measurement problems. The labels used for sampling and specialisation are much less precise than suggested in most common coaching conversations. In fact, a systematic review found that a collection of the published studies used 18 different approaches to classify specialisation, many of which did not include age or training intensity, and classifications disagreed markedly across studies (Mosher et al., 2020). “Sampling” definitions have certainly created similar confusion. Sampling could refer to casual play in several sports, organised commitment to several sports, or even simply having played more than one sport at some point in childhood. A scoping review found little consideration in the literature of when or how intensely other sports were played and concluded that sampling and diversification should not be treated as interchangeable constructs (Ramsay, Mosher and Baker, 2023).

So, where do we go from here?

A range of alternative explanations for the link between sampling and athletic talent in the literature clearly mean we should conclude that the causal mechanism (sampling early causes talent) should not be the standard conclusion drawn from the liteature. However, this lack of clarity in the literature should also not become an argument for doing nothing.

It is my interpretation of the literature that the evidence does not justify a complete switch to compulsory sampling on the belief that sampling manufactures talent. However, I think that it does justify questioning systems that force broad scale early specialisation alone, equate youth rankings with adult potential, or load high-pressure training onto children because they fear a missing a developmental 'window of opportunity'.

Thus, it seems I keep arguing for the same approach, one that fits somewhere 'in the middle'. For those interested in reading more about the 'nonsensical' nature of most extremist views, see this work by Ackerman (2014)

Therefore, it is likely that athletes would benefit more from designing flexible programs which keep all pathways open long enough for athletes to develop and choose what they themselves value. At the same time we probably need better ways of distinguishing all the different practice types athletes can engage in such as sport-specific practice, free play, organised participation in other sports, and so on, alongside athletes preferences and perspectives on their engagement in these activities. But most importantly, despite discussing talent here, I think we would benefit from evaluating all athletic development programs on retention (e.g. longer retention could be a great talent strategy) and healthy adult progression, not just on who makes representative teams or wins medals at World Championships at 13.

I think my view is compatible with creating programs that encourage meaningful sport-specific development rather than just encouraging sampling or specialisation as causal predictors of talent. That means that context will always be central to this debate (this is also not new given we know different specialisation ages may be relevant for different sports).

For now, I can only include one thing from the available literature though: the research (and our common sense) tells us it is likely useful to protect young athletes from unnecessary specialisation and excessive physical and psychological loads that accompany it, but at this time it does not promise that a multi-sport sampling has better odds than other methods to consistently turn them into future champions.