Further Muddying the Waters? A Comment on Bell et al's 2021 Definition of Youth Sport Specialization

Bell et al¹  acknowledged that their proposal is preliminary and “under construction.” Nevertheless, we trust we can take their carefully chosen words literally, particularly given that they aimed to present a definition.

Their proposal comprises 4 components that consist of 21 elements, as outlined in the Table. Operational definitions are lacking for 20 elements. For example, exactly how does one determine whether or not (1) participation in a sport is “intentional”; (2) participation is “focused”; (3) practices are “organized”; (4) training is “structured”; (5) practice, training, and/or competitions are “regular”; (6) a potential restriction of opportunities and/or time for an activity has been causally evoked by one's main sport participation and not by any other factor (eg, lacking or diminishing interest, lack or expiring of activity programs, mutual restrictions of available opportunities and/or time among activities other than one's main sport); and (7) an athlete's motive to limit, end, or generally forgo other sports participation has been to enable focused single-sport participation and not any other motive (eg, lacking or diminishing interest, lack or expiring of programs for other sports, time demands of academics, other extracurricular activities, community engagement, family, or friends)? Furthermore, (8) relative to what baseline value across what time period does one determine the potential limitation of an activity?

Relatedly, the dichotomous nature of each element would require cutoff values. However, the relevant evidence needed to empirically substantiate cutoffs is not available. This is perhaps one of the reasons for the plethora of ad hoc definitions of youth sport specialization in the literature⁴  and for Bell et al's consent-based—rather than evidence-based—approach.

The term “may” renders statements empirically meaningless.³  Meeting all or some of the elements in components 2.1–2.3 (Table) may or may not define an athlete as specialized.

Components and elements are connected by “AND,” “OR,” or “AND/OR” relationships. The formal logic of the AND relation between criteria 1d and 1e excludes cases in which either only other sports OR only other activities are restricted. The OR relationships among the elements in criterion 2.2 exclude cases in which athletes limited AND then ended other sport participation. The OR relationships among the elements in criterion 2.3 exclude cases in which both opportunities AND time for an activity are limited.

Bell et al's suggestion allows a wide range of participation patterns to be defined as specialized. The “AND,” “OR,” and “AND/OR” relationships of their 4 components and 21 elements define at least 4371 different combinations as specialization (meeting criterion 1 plus 0–9 elements in criterion 2, such that 2.1 can be fulfilled in 7 different ways, 2.2 in 3 ways, and 2.3 in 72 ways). The construction “such as” in criterion 2.3 implies further relevant activities, resulting in an even higher number of specialized combinations. Overall, the multiple relationships among the elements defy the Popperian imperative²  that scientific statements should be formulated as simply as possible. This further impedes empirical testing of youth sport specialization.

Bell et al's¹  proposal fails to meet the basic requirements of a scientific definition by (1) using several empirically meaningless criteria, (2) basing their definition on undefined key terms, (3) failing to operationally define variables, and (4) failing to provide cutoff values for dichotomized criteria. Taken together, under Bell et al's proposal, athletes can flexibly be categorized as specialized or nonspecialized, making specialization subject to opinion. In its present form, Bell et al's approach is therefore not suitable for guiding empirical research.

Bell et al,¹  along with other scholars,^5–7  have suggested defining specialization as a dichotomous variable distinguishing specialized from nonspecialized participation patterns. However, this use is imprecise. One's participation pattern is generally defined by several continuous parametric variables, including starting age and amounts of coach-led practice and peer-led play in both one's main sport and other sports through different ages.^8,9  These continuous variables provide a more accurate description of athletes' participation patterns. At the same time, they enable us to determine in which aspects, and how closely, one's participation pattern corresponds to facets of a sport specialization construct. Researchers aim to describe, explain, and predict relationships between participation variables and outcomes, that is, the extent to which individual differences in participation variables predict individual differences in outcomes. Both participation variables and outcomes can be quantified.

Furthermore, participation patterns may yield short-term and long-term positive and negative outcomes, to different magnitudes and at different probabilities, including benefits (eg, performance, enjoyment, prestige, and financial income), costs (eg, opportunity costs, coaching, and facilities), and risks (eg, injury and burnout). Additionally, the values of outcomes may vary (eg, gold at the Olympic Games versus state championships or a contusion versus a bone fracture). Investigators should thus seek to explore 3 main questions for different participation patterns:

1. Which short-term and long-term positive and negative outcomes do the participation variables yield, to what magnitude, and at what probability?

2. What material and immaterial value does each positive and negative outcome have?

3. What is the eventual ratio of the summed value of all positive outcomes relative to the summed value of all negative outcomes accumulated throughout one's athletic career?

Such research is not possible when comparing only 2 dichotomized participation constructs, namely, specialization versus nonspecialization.¹  However, it is possible to quantify the different continuous participation and outcome variables and examine their relationships. This has been shown in multiple studies of performance outcomes^9,10  and is presumably possible for health-related outcomes, too. To empirically evaluate the relationships among participation and outcome variables, a specialization construct and the labeling of athletes as specialized or nonspecialized is neither necessary nor productive.