Selection-Dependent Differences in Youth Elite Basketball Players’ Relative Age, Maturation-Related Characteristics, and Motor Performance

ABSTRACT

Purpose: The aim of the present study was to investigate the influence of players’ relative age, maturation-related characteristics, and motor performance tests on two different stages of talent selection (regional and national level) in youth elite basketball. Methods: Relative age, maturation-related variables (height, weight, maturity offset, maturity timing) and motor performance (Countermovement jump, Standing long jump, Modified agility T-test, Linear sprint 20 m) of N = 68 male youth basketball players (M_age = 14.39 ± 0.28 years) were assessed during the initial selection tournament for the German U15 national team. Pre-selection biases regarding relative age and maturity timing on the regional level were investigated utilizing one-sample t-tests. Differences in relative age, maturation-related characteristics and motor performance between players selected (n = 27) and non-selected (n = 41) for the youth national team were examined via independent samples’ t-tests and logistic regression analyses. Results: Strong pre-selection biases toward early-born and early-maturing players were confirmed on the regional level. Significant advantages in height and weight and higher values in maturity offset and maturity timing were found for selected players. Among the motor variables, only Countermovement jump performance was significantly better in selected players. When controlling for relative age and biological maturation regression models including motor performance variables did not significantly discriminate players’ selection status. Conclusion: Coaches working in national, but also preceding selection stages (regional and club level) should raise their awareness to relative age and biological maturation when evaluating players’ potential and current (motor) performance.

KEYWORDS:

• Biological maturation
• diagnostics
• somatic age
• talent predictors
• team sports

A primary objective of many competitive youth basketball programs is the identification, development and selection of talented young players. In prospective talent research, talented athletes are those who have the utmost likelihood of future success in their respective sports (Höner et al., 2023, p. 551). The process of identifying, developing, and selecting such talented players is, however, inherently complex and presents significant challenges for decision-makers in clubs or national federations (e.g., coaches or scouts; Bar-Eli et al., 2023). There are multiple factors (e.g., anthropometric, physiological, psychological, sociological, technical, tactical) that contribute toward and confound performance in young athletes that are important to consider when evaluating players’ potential (Baker et al., 2022). In addition to subjective performance evaluations through game observations (Lath et al., 2021; Roberts et al., 2019), objective diagnostics assessing physical abilities and motor skills are used in team sports to inform such complex selection decisions (Höner et al., 2017; Johnston et al., 2018). With respect to basketball, however, there is limited evidence for the relevance of such tests in predicting future success (Johnston et al., 2022). Nevertheless, test of fitness and technical skills are widely used for selection purposes in youth basketball (Gál-Pottyondy et al., 2021b; Mancha-Triguero et al., 2019; Morrison et al., 2022; Ziv & Lidor, 2009), and players are often selected based on their performances on such tests.

Important identification and selection processes of talented young basketball players commonly take place in early to mid adolescence (Paulauskas & Radu, 2019; Trunić & Mladenović, 2014), when individual differences in physical and functional aptitude add to the challenges of evaluating players’ current ability and future potential. In this important stage of development, athletic performance may be influenced by relative age and biological maturation (Arede, Ferreira, et al., 2019). Accordingly, these factors need to be considered in talent selection to ensure a fair and ultimately effective selection process, preventing highly talented players from being overlooked or deselected due to younger relative age or delayed biological maturation (Cumming et al., 2017).

The Relative Age Effect (RAE) relates to a predominant number of early-born players (i.e., born closer to the age cohort’s annual cut-off date) in selected groups and is associated with advantages of early-born players in current performance (e.g., Wattie et al., 2015). Relative age is determined by the players date of birth and the cut-off date for age group membership. In most age groups relative age is limited to a single year though may be as broad as two to three years in less popular sports or in low-density populations (rural). The RAE is present from early childhood and generally maintained within youth sports through early to late adolescence (Johnson et al., 2017; Kelly et al., 2024). It is a well-established phenomenon in talent research, particularly in the context of team sports such as basketball (e.g., de la Rubia Riaza et al., 2020; Gonçalves & Carvalho, 2021; Kalén et al., 2021; Kelly et al., 2021; López de Subijana & Lorenzo, 2018; Rubajczyk et al., 2017). Initial explanations of the RAE focused upon age-associated differences in anthropometric, maturational and physical fitness, however, recent evidence suggests that this phenomenon is more likely to result from developmental differences in psychosocial, behavioral, and neuromotor attributes which are more likely to be present in early childhood and explain the presence of equivalent RAE biases in non-physical domains (Radl & Valdés, 2023).

Biological maturation refers to progress toward the mature stature and can be characterized by the stage of maturation at the time of observation (i.e., maturity status), the age at which certain maturational events such as the age of peak height velocity (APHV) occur (i.e., maturity timing) and the rate at which it progresses (i.e., maturity tempo; Malina et al., 2019). Differences in biological maturity status within same age peers can vary up to six years during adolescence (Malina et al., 2004), which may lead coaches or scouts to preferentially select early maturing players, for example, due to maturity associated advantages in anthropometry or physical performance (Johnston et al., 2018; Malina et al., 2019). Selection biases associated with maturity emerge at pubertal onset, vary relative to the demands of the sport, and typically increase with age. Maturity and relative age selection biases have the potential to play decisive roles in the processes of talent identification and selection, yet exist and partly operate independent of one another. Emerging evidence in soccer suggests that differences in maturity rather than relative age may have the greatest impact upon player selection during adolescence (Sweeney et al., 2022; Towlson et al., 2021).

Selection biases associated with biological maturation were observed in youth basketball. More specifically, Te Wierike et al. (2015) found that 13- to 16-year-old male players from a Dutch basketball academy experienced their APHV at an earlier age compared to 14-year-old males in the general population. Further, several studies showed that selected male youth players were taller, heavier and experienced their APHV at an earlier age when compared to their non-selected peers (Baxter-Jones et al., 2020; Guimarães, Baxter-Jones, et al., 2019; Ramos et al., 2019; Torres-Unda et al., 2013). Similarly, Arede, Ferreira, et al. (2019) demonstrated higher chances to be selected for the Portuguese U16 national team for earlier in comparison to late maturing boys.

A selection bias in male basketball for players that mature early makes sense; especially as parameters such as height contribute more to players’ performance than in other team sports (e.g., soccer; Pino-Ortega et al., 2021). Biological maturation may impact several performance indicators in youth basketball. For instance, differences in anthropometric and physical indicators were found comparing youth athletes with respect to their maturity timing (i.e., early, on time or late) in favor of early maturing athletes (for a review see Albaladejo-Saura et al., 2021). Specifically for basketball, anthropometric measurements indicated that early maturing youth players were taller and heavier compared to their later maturing counterparts (Fragoso et al., 2021; González et al., 2022; Gryko, 2021; Guimarães, Ramos, et al., 2019; Jakovljevic et al., 2016; Torres-Unda et al., 2013). Moreover, comparisons of physical (motor) performance indicators revealed advantages for early maturing players in agility (Guimarães, Ramos, et al., 2019), linear sprint (Arede, Ferreira, et al., 2019; Gryko, 2021; Guimarães et al., 2023; Guimarães, Ramos, et al., 2019; Torres-Unda et al., 2013) and jumping abilities (Arede, Ferreira, et al., 2019; González et al., 2022; Gryko, 2021; Guimarães et al., 2023; Torres-Unda et al., 2013). Not least for this reason, it is not surprising that the importance of assessing and accommodating for individual differences in biological maturation when conducting physical performance tests for talent selection purposes was highlighted in recent systematic reviews (Albaladejo-Saura et al., 2021; Gál-Pottyondy et al., 2021b). In this context, there are increased efforts in sport science to examine the relevance of such performance indicators when controlling for players’ biological maturation (Charbonnet et al., 2022; Gryko, 2021).

The aim of the present study was to investigate the influence of players’ relative age, maturation-related characteristics, and motor performance tests upon talent selection in elite youth basketball across two selection stages for youth national team trialists. First, the pre-selection bias regarding relative age and biological maturity timing was analyzed (i.e., regional level). This objective was pursued as players nominated for such a tournament on national level usually went through several selection procedures in their respective clubs and regional federations prior to their participation in the selection tournament. As early-born and early-matured players have traditionally been favored in male sports, the following directed hypothesis was tested:

H1:

There is a pre-selection bias toward earlier born and earlier maturing players.

Second, differences in relative age, maturation-related variables (i.e., height, weight, maturity offset, maturity timing) and motor performance (i.e., Countermovement jump, Standing long jump, Modified agility T-test, Linear sprint 20 m) were assessed between players selected and not selected at the tournament (i.e., national level). As advantages in the considered variables might align with advantages for selection, the following directed hypotheses were tested:

H2a:

Selected players are advanced compared to their non-selected counterparts with respect to relative age, maturation-related variables, and motor performance.

Further, as current literature emphasizes the consideration of relative age and biological maturation when analyzing motor diagnostics (e.g., Charbonnet et al., 2022), selection-dependent differences in motor performance under consideration of players’ relative age and biological maturity status were examined. As selected players may exhibit motor performance advantages, and in line with the former hypotheses, the following hypotheses were tested:

H2b:

The results of the motor diagnostics predict players’ selection status (i.e., selected or non-selected) when controlling for relative age and biological maturity status (i.e., maturity offset), with selected players outperforming non-selected players.

Methods

Setting and sample

The present study was conducted at the U15 national selection tournament of the German Basketball Federation (Deutscher Basketball Bund, DBB) in October 2021. This annual event is the very first stage of talent selection at the national level in Germany. At this tournament, players compete in regional selection teams and are evaluated and selected by the youth national team coaches for the U15 youth national team. Before the start of the competition, players pass a standardized physical performance test battery (see Measures and Procedures for further details on the test battery). Until players are selected for the regional squads, they go through several selection procedures on regional level. In such selection processes on the regional level prior to the tournament, qualified regional basketball coaches select players mainly based on their subjective evaluations during practice sessions and competitions. Both selection stages, the selection on the regional level before the tournament (H1) and the selection on the national level at the tournament (H2a and H2b) are considered within this study.

The study sample consisted of N = 68 male youth basketball players (M_age = 14.39 ± 0.28 years) with n = 27 (M_age = 14.39 ± 0.30) being selected and n = 41 (M_age = 14.40 ± 0.28) being not selected for further talent development purposes by the federation. Players’ selection status was determined via an online search (Deutscher Basketball Bund, 2021).

Written informed consent for the collection and scientific use of the data was provided by all participants’ legal guardian/next of kin. The research was reviewed and approved by the Ethics Committee of the Faculty of Economics and Social Sciences at the University of Tübingen. Further, the implementation of the study was approved by the German Basketball Federation (Deutscher Basketball Bund, DBB).

Measures and procedures

Anthropometric measurements

Anthropometric measurements were carried out with players wearing standardized practice uniforms (i.e., individually numbered jerseys and matching shorts), but no shoes. In line with Leyhr et al. (2020), weight was measured to the nearest 0.1 kg with a calibrated scale (seca 813 electronic flat scale; seca GmbH & Co. KG, Hamburg, Germany). Height and sitting height were obtained to the nearest 0.1 cm with a fixed stadiometer (seca 213 portable stadiometer; seca GmbH & Co. KG, Hamburg, Germany). To assess the height, players were instructed to stand in straight posture with feet together and arms relaxed. To determine the sitting height, players were asked to sit on a table with their trunks upright and their backs against the measuring device. In all of these measurements, the players’ head was aligned with the Frankfurt horizontal plane (Pearson & Grace, 2012).

Relative age and biological maturation

Players’ relative age was registered by calculating the day of birth within a calendar year (e.g., Leyhr et al., 2018). For example, a player born on January 1^st would have a relative age of 1, while a player born on July 1^st would have a relative age of 182.

The anthropometric data were used to determine players’ biological maturity status. Players’ maturity offset (MO) from their APHV was estimated according to the following formula (Mirwald et al., 2002):

$\mathit{MO}\left(\mathit{years}\right)=-9.236+(0.0002708\times \left(\mathit{leg}\mathit{length}\times \mathit{sitting}\mathit{height}\right))+\left(-0.001663\times \left(\mathit{CA}\times \mathit{leg}\mathit{length}\right)\right)+\left(0.007216\times \left(\mathit{CA}\times \mathit{sitting}\mathit{height}\right)\right)\mathit{\hspace{0.17em}}+\left(0.02292\times \left(\mathit{weight}\mathit{by}\mathit{height}\mathit{ratio}\times 100\right)\right)$

where leg length was calculated by subtracting sitting height from body height.

Players’ biological maturity timing (BMT) was determined by the difference between somatic age (SA) and chronological age (CA):

$\mathit{BMT}\left(\mathit{years}\right)=\mathit{SA}-\mathit{CA}$

where SA was calculated by adding the average individual APHV for boys (i.e., 13.8 years; Malina et al., 2004) to MO:

$\mathit{SA}\left(\mathit{years}\right)=\mathit{MO}+13.8$

Motor diagnostics

After the anthropometric measurements and before starting the test procedure, players went through a standardized 15-minute warmup routine. A standardized physical performance test battery was used, of which four individual tests were analyzed in this study (see the test manual for further details; Deutscher Basketball Bund, 2022). Players performed each test twice with sufficient time to recover between trials and the best trial being scored. In the jumping tests (i.e., Countermovement jump, Standing long jump), higher values reflected better performances. The time-based speed tests (i.e., Modified agility T-test, linear sprint) were negatively coded with lower values reflecting better performances.

Countermovement jump (CMJ)

The players were instructed to jump as high as possible from a standing position after initially squatting down to a comfortable depth. The hands had to be placed on the hips at all phases of the exercise. The jump height was determined to the nearest 0.1 cm with an Optojump system (Microgate, Bolzano, Italy).

Standing long jump (SLJ)

The players were instructed to jump as far as possible from a standing position behind the starting line. They were allowed to make a single initial movement (also with swinging arms) before jumping. The distance was measured with a tape measure from the starting line to the heel of the nearest foot to the nearest 1 cm.

Modified agility T-test (MAT)

A linear sprint over 5.0 m was followed by a lateral shuffle to the left over 2.5 m, then a lateral shuffle to the right over 5.0 m, and again a lateral shuffle to the left over 2.5 m, before finishing with a linear backpedal over 5.0 m (see Figure 1).

Figure 1. Outline of the Modified agility T-test (MAT).

The order of movements is indicated in lower-case letters (a-e).

The players started each run on their own from a line marked 50 cm behind the starting gate in high start position. They were instructed to complete the pattern as quickly as possible and to cross the finish line in full speed. They also had to touch the top of a cone (at least 50 cm high) at any change of direction besides constantly facing forward and keeping their legs uncrossed. Execution time was measured to the nearest 0.01 s using a light barrier system (photoelectric sensor LRS-4050-103, Contrinex AG, Givisiez, Switzerland).

Linear sprint (LS)

The players performed 20 m linear sprints. They started each run on their own from a line marked 50 cm behind the starting gate in high start position. They were instructed to complete the distance as quickly as possible and to cross the finish line in full speed. Execution times were measured to the nearest 0.01 s using a light barrier system (TCi System, Brower Timing Systems, Draper, UT, USA).

Statistical analysis

Data analyses were performed using IBM SPSS Version 27.0 (IBM Corporation, Armonk, NY, USA). The significance level was set to p < .05. One-tailed tests were utilized in this study based on the directed hypotheses (H1, H2a, H2b).

One-sample t-tests were used to analyze the pre-selection biases regarding relative age and biological maturity timing (H1). Specifically, mean values for relative age (i.e., day of birth within a year) and biological maturity timing (i.e., difference between SA and CA) were compared to conventional values for the general population. Regarding RA, an equal distribution was assumed with regard to birth rates within the year. So, July 1^st (i.e., day 182 within a year) served as the expected mean birth date of the population (M_RA = 182). With respect to BMT, it was presumed that the average population is biologically as advanced as chronologically, hence somatic age is equal to chronological age (M_BMT = 0). Therefore, the reference value for the one-sample t-test was set to M_RA = 182, and M_BMT = 0, respectively. Additionally, Cohen’s d (including 90% confidence intervals) was calculated as effect size. According to Cohen (1992) the magnitude of effect sizes can be classified as small (.20 ≤ d ≤ .49), medium (.50 ≤ d ≤ .79) or large (d ≥ .80).

To compare differences between selected and non-selected players with respect to relative age, maturation-related variables, and motor performance (H2a), t-tests for independent variables were performed and Cohen’s d was calculated as effect size. Further, logistic regression analyses were performed to analyze whether the motor diagnostics discriminate players’ selection status when controlling for relative age and biological maturity status (H2b). Selection status (non-selected = 0, selected = 1) was chosen as the binary criterion variable in the four different regression models. Each model included one test result of the motor diagnostic (i.e., CMJ, SLJ, MAT or LS) as the predictor variable, whereas relative age and maturity offset served as covariates in all models. Due to the limited sample size and the expected problems regarding the robustness of the results, the use of an overall model containing covariates as well as all four motor performance variables was considered not appropriate and, thus, neglected. The overall model fit was analyzed with the likelihood ratio chi-squared test and Nagelkerke’s R². Additionally, the odds ratios (ORs) e^b and their 90% confidence intervals (CIs) were computed to estimate a player’s relative chances to get selected, depending on the considered predictor and controlling for his biological maturity status. To enable comparisons between the predictors, the ORs were additionally adjusted to the standard deviations of all players (Höner & Votteler, 2016). The resulting (e^b)^SD displays the OR for being selected by a standard deviation increase within the respective predictor. Because of its negative coding, the adjusted OR for the time-based speed tests (i.e., MAT and LS) was inverted by 1/(e^b)^SD.

To determine the size of a potentially detectable population effect, sensitivity was calculated by post hoc power analyses using G*Power version 3.1.9.7 (α = 0.05, 1 – β = 0.80, one-tailed). The analyses determined the sensitivity for discovering small effects for hypothesis 1 (i.e., Cohen’s d > 0.2) and at least medium effects for hypothesis 2a (i.e., d > 0.5). With respect to hypothesis 2b, analyses were sensitive enough to detect ORs of at least e^b = 1.9.

Results

Pre-selection bias on regional level (objective 1)

Selection biases were detected for both, relative age and biological maturity timing on the regional level (H1). Descriptive statistics indicate that players were born on day M = 103.38 ± 103.73 within the year (i.e., May 10^th; 70.6% born within the first half, 29.4% within the second half of the calendar year) and were estimated to be biologically M = 0.82 ± 0.73 years older than chronologically (see Table 1). The mean relative age of the sample was significantly older than the expected relative age of the general population with moderate effect size (t(67) = 4.10, p < .001, d = 0.50). Further, the mean biological maturity timing of the players was significantly advanced compared to the expected timing in the average population with large effect size (t(67) = 9.19, p < .001, d = 1.11). The pre-selection biases for relative age and biological maturity timing are illustrated in Figure 2.

Figure 2. Pre-selection bias on regional level for relative age and maturity timing.

A positive d reflects a bias toward early-born (relative age) and early-maturing (maturity timing) players.

Table 1. Descriptive overview of all assessed variables and inferential statistics for group comparisons according to selection status.

Variables	Total (N = 68)	Selected (n = 27)	Non-selected (n = 41)	Group comparison
		M ± SD		t(df)	d
Age
Chronological age (years)	14.39 ± 0.28	14.39 ± 0.30	14.40 ± 0.28
Relative age (day of birth within a year)	130.38 ± 103.73	132.41 ± 109.16	129.05 ± 101.35	-0.13 (66)	-.03
Anthropometry
Height (cm)	183.34 ± 9.12	186.90 ± 8.45	180.99 ± 8.86	2.74 (66)	.68**
Weight (kg)	68.78 ± 11.86	72.18 ± 10.52	66.54 ± 12.27	1.96 (66)	.49*
Biological maturation
Maturity offset (years from APHV)	1.41 ± 0.76	1.60 ± 0.61	1.28 ± 0.83	1.76 (66)	.44*
Maturity timing (somatic age-chronological age)	0.82 ± 0.73	1.02 ± 0.57	0.68 ± 0.80	2.03 (65.61)	.47*
Motor performance
Countermovement jump (cm)	36.04 ± 5.54	37.66 ± 6.12	34.98 ± 4.92	1.99 (66)	.49*
Standing long jump (cm)	227.71 ± 20.94	231.00 ± 26.17	225.54 ± 16.65	1.05 (66)	.26
Modified agility T-test (s)^#	5.78 ± 0.34	5.79 ± 0.45	5.77 ± 0.24	0.19 (35.88)	.05
Linear sprint 20 m (s)^#	3.22 ± 0.16	3.24 ± 0.18	3.21 ± 0.15	0.66 (66)	.16

**p < .01; *p < .05.

^#Time-based tests were negatively coded with lower values reflecting better performances.

Selection-dependent differences on national level (objective 2)

The descriptive and inferential statistics for all study variables separated by selection status are displayed in Table 1. Regarding the second objective, no significant difference between groups was found with respect to relative age (t(66) = –0.13, p = .45). In contrast, significant differences between selected and non-selected players were detected in height (t(66) = 2.74, p < .01, d = 0.68) and weight (t(66) = 1.96, p < .05, d = 0.49) with large and small effect sizes, respectively (H2a). Further, differences between the groups were revealed regarding maturity offset (t(66) = 1.76, p < .05, d = 0.44) and maturity timing (t(65.61) = 2.03, p < .05, d = 0.47) with small effect sizes.

Moreover, descriptive statistics indicated that selected players outperformed their non-selected counterparts in the jumping tests (i.e., CMJ, SJL), while non-selected players were slightly better in the speed tests (i.e., MAT, LS). However, significant differences between groups with small effect size were only demonstrated for CMJ (t(66) = 1.99, p < .05, d = 0.49). The other motor diagnostics were not sensitive enough to discriminate players’ selection status (each p > .15).

The results of the logistic regression analyses (H2b) are presented in Table 2. The overall models did not significantly discriminate players’ selection status when controlling for relative age and maturity offset. Here, the overall model including CMJ performance just failed significance to discriminate between selected and non-selected players (model 1: χ²(3) = 7.63, p = .05, Nagelkerke’s R² = .14). Nevertheless, the one-tailed analyses of the single regression coefficients within the model including CMJ as predictor variable revealed a certain explanatory power of the CMJ variable (β = 0.10, p < .05). The discriminatory power for the models including SLJ (model 2: χ²(3) = 4.62, p = .20, Nagelkerke’s R² = .09), MAT (model 3: χ²(3) = 3.62, p = .31, Nagelkerke’s R² = .07), or LS (model 4: χ²(3) = 3.91, p = .27, Nagelkerke’s R² = .08) failed significance by far.

Table 2. Results of the logistic regression analysis for the prediction of selection status based on results in motor diagnostics controlled for biological maturity status and relative age.

	Omnibus-Tests			Variables	Logistic regression coefficients^†
Model	χ^2(3)	p	Nagelkerke’s R^2		b	Wald	p	e^b [90%–CI]	(e^b)^SD
1	7.63	0.05	0.14	Constant	-5.18	–	–	–
				Maturity offset (years from APHV)	0.72	3.33	<.05	2.06 [1.07; 3.96]	1.73
				Relative age (day of birth within a year)	0.00	0.49	.24	1.00 [1.00; 1.01]	1.00
				Countermovement jump (cm)	0.10	3.72	<.05	1.10 [1.01; 1.19]	1.70
2	4.62	0.20	0.09	Constant	-4.41	–	–	–
				Maturity offset (years from APHV)	0.68	3.19	<.05	1.97 [1.06; 3.69]	1.67
				Relative age (day of birth within a year)	0.00	0.46	.25	1.00 [1.00; 1.01]	1.00
				Standing long jump (cm)	0.01	0.99	.16	1.01 [0.99; 1.03]	1.23
3	3.62	0.31	0.07	Constant	-2.09	–	–	–
				Maturity offset (years from APHV)	0.68	3.27	<.05	1.97 [1.06; 3.64]	1.67
				Relative age (day of birth within a year)	0.00	0.47	.25	1.00 [1.00; 1.01]	1.00
				Modified agility T-test (s)^#	0.08	0.01	.46	1.08 [0.31; 3.74]	0.97
4	3.91	0.27	0.08	Constant	-4.44	–	–	–
				Maturity offset (years from APHV)	0.66	3.20	<.05	1.94 [1.06; 3.58]	1.65
				Relative age (day of birth within a year)	0.00	0.45	.25	1.00 [1.00; 1.01]	1.00
				Linear sprint 20 m (s)^#	0.88	0.30	.29	2.41 [0.17; 33.45]	0.87

N = 68. CI = confidence interval.

^#Time-based tests were negatively coded with lower values reflecting better performances. Thus, the adjusted odds ratios for these variables were inverted and displayed as (e^b)^–SD.

†One-tailed p-values were considered within the analyses of regression coefficients based on the directed hypothesis (H2b).

Discussion

The current study investigated differences between selected and non-selected elite youth basketball players with regard to relative age, maturation-related characteristics, and motor performance on different selection levels for the U15 youth national team. The highly selective sample comprised players who went through several selection processes at regional and club level before being nominated for their first selection tournament on the national level investigated within the present study. Therefore, the present study enabled to analyze the influence of players’ relative age and biological maturation on selection processes within both, the regional and national selection level: on the regional level, with respect to all nominated players for the tournament, this allowed for an overview of a potential pre-selection bias in favor of early-born or early-maturing players within the team squads prior to the tournament (objective 1). On the national level, with regard to differences between selected and non-selected players for the extended squad of the German U15 national team, this provided insights into a potential impact of relative age, biological maturation and motor performance during the event on selection procedures (objective 2). In order to provide a plain view on the relevance of the assessed motor performance variables when discriminating players’ selection status, their discriminative power was additionally analyzed when controlling for relative age and biological maturation.

Pre-selection bias on regional level (objective 1)

The pre-selection bias within the selection tournament was analyzed regarding relative age and biological maturity timing. The results confirmed early-born and early-maturing players being favored in precedent selection stages (H1). A distinct relative age effect was found in the present study with 70.6% of players born in the first half-year (average day of birth M = 103.38 ± 103.73; see Table 1 and Figure 2). Further, players in this study were biologically almost one year (0.82 ± 0.73 years; see Table 1 and Figure 2) older than chronologically which confirmed a significant, large (d = 1.11) pre-selection bias associated with biological maturity timing. It should be noted that the maturity offset equations tend to overestimate age at PHV in older and more mature athletes (Kozieł & Malina, 2018). Thus, it is likely that the true maturity selection bias is even greater than that which was observed.

The presence of relative age and maturity selection biases in the current sample are not surprising and are consistent with previous research in youth basketball (Te Wierike et al., 2015; Torres-Unda et al., 2013). Regarding relative age, Torres-Unda et al. (2013) compared a cohort of youth players in Spanish local basketball leagues with their peers in the general population and showed an overrepresentation of players born in the first half of the cohort’s birth year (i.e., 68%). With respect to biological maturity timing, Te Wierike et al. (2015) found that youth players (14.66 ± 1.09 years old) from a Dutch basketball academy experienced their peak height velocity (13.06 ± 0.77 years) significantly earlier compared to similarly aged boys of the Dutch population. That said, the findings of the present study comprising a sample of the same age group (i.e., U15) show rather similar results with players who exhibited comparable values regarding both chronological age (14.39 ± 0.28 years; see Table 1) and age at peak height velocity (12.98 ± 0.73 years; see Table 1). Both, the strong relative age effect, and maturation bias might be explained by several selection procedures at club and regional level taking place before the nomination for the investigated U15 national selection tournament. Within the talent development progress this might have repeatedly led to an increased selection of early-born and early-maturing players (Cumming et al., 2018; Yagüe et al., 2018) and amplified the magnitude of the effect continuously. Consequently, the results regarding selection-dependent differences within the tournament (i.e., objective 2) have to be considered in the light of this distinct selection bias that was discovered within the sample.

Selection-dependent differences (objective 2)

The analyses of differences among players’ selection status confirmed the expected advantages for selected players for most age- and maturation-related characteristics considered in H2a. While interestingly no significant differences between selected and non-selected players’ relative age were found, the present study’s findings for height, weight, maturity offset, and maturity timing align with former studies in youth basketball highlighting similar age- and maturation-related selection advantages (Arede, Ferreira, et al., 2019; Guimarães, Baxter-Jones, et al., 2019; Ramos et al., 2019; Torres-Unda et al., 2013). It may be that the observed RAE at the first selection stage is representative of an earlier selection bias and that during adolescence individual differences in maturation are of greater importance. As noted, RAEs are present from early childhood whereas maturity selection biases only emerge with the onset of puberty.

The highest effect sizes were discovered for the anthropometric variables. Moderate to large effects for height (d = 0.68) and nearly moderate effects for weight (d = 0.49) were identified and, thus, are slightly lower in size compared with former studies. While Guimarães, Baxter-Jones, et al. (2019) report very large advantages (i.e., height: d = 1.72; weight: d = 1.40) for U14 players selected for a regional Portuguese selection team, Ramos et al. (2019) found large effects (0.82 ≤ d ≤ 1.26) with respect to superior body size values in U14 and U16 players chosen for the first team of an elite Portuguese basketball academy. Besides the slightly different age groups, particularly the level of selection examined in these studies was different compared to the present study, which incorporated the selection for a U15 national team squad. This could have caused the slightly lower effects within an already more homogenous sample of high-performing youth basketball athletes. Nevertheless, the results further highlight the importance of height as performance indicator in basketball and reinforce the assumption that coaches are particularly attentive to such details about a player (Hoare, 2000; Pino-Ortega et al., 2021; Rogers et al., 2021).

Surprisingly, no significant effect between selected and non-selected players for relative age were found within the present study. While being aware of the mentioned limitations of comparability to the former described studies in youth basketball, those found significant, moderate differences in chronological age within the considered age groups with selected players being older than their non-selected counterparts (Guimarães, Baxter-Jones, et al., 2019). In contrast, significant differences among selection status within the present study were found for biological maturation. Advantages of small to moderate effect size were found for selected players at the national selection tournament in maturity offset (d = 0.44) and timing (d = 0.47). Once again, the effects are smaller compared to the studies described above, which revealed moderate to large effects for these variables (0.71 ≤ d ≤ 1.84).

Altogether, the effects within the present sample are lower for maturation or even not present for relative age. From a practical point of view, a reason for the latter could be that coaches of the youth national team, who make the selection decisions at the tournament, have been made aware of the problem of the relative age effect within several workshops and several tournaments before, and now explicitly consider this variable in their selections. However, it remains unclear to which extent those had an influence on the selection. Nevertheless, it is important to maintain such educational initiatives, and to extend them to the effect of biological maturation in order to achieve an even higher awareness of the topic in talent identification processes (Leyhr et al., 2023).

Another explanation for the attenuated biases lies in the large pre-selection bias associated with relative age and maturation that was found in the investigated sample. The strong shift toward early-born and early-maturing players might have emerged a homogenous study sample of relatively old, biologically advanced players. For instance, with regard to maturity timing players were on average biologically 0.82 years older when compared to their CA. Due to several selections in preceding stages just prior to the selection tournament for the youth national team, a major part of late-maturing players with potential current disadvantages in performance have already been deselected (Cumming et al., 2017; Hill et al., 2020; Votteler & Höner, 2014) and have increased the homogeneity regarding performance within the sample.

A further point to make is the advanced maturity status of all investigated players who were, on average, already well over their APHV (1.41 ± 0.76 years). Most players had already reached the post-APHV development phase which coincides with a decrease in body development (e.g., Malina et al., 2015; Wormhoudt et al., 2017). Consequently, the influence of maturity status on the selection at the tournament might have been less relevant as all players had yet reached a high degree of their full development. Thus, differences in players’ performance are more likely to be related to other performance indicators. Those might have been the far developed anthropometric characteristics such as height as an important factor for game performance in basketball (e.g., Paulauskas et al., 2018; Teramoto & Cross, 2017; Xu et al., 2022), but also other performance indicators such as motor performance (e.g., speed), technical (e.g., shooting), tactical (e.g., spacing), or psychological aspects (e.g., decision-making). For homogenous, far developed samples such indicators might have a higher impact on selection (Ramos et al., 2019; Rösch et al., 2022; Torres-Unda et al., 2013) and, thus, should be considered in future studies.

Aligning with these assumptions, the analyses of selection-dependent differences in the performed motor diagnostics (H2a) gave insight into the relevance of motor performance indicators that were assessed during the tournament. In general, differences in motor performance diagnostics between selected and non-selected players were comparably low. On the one hand, both sprinting and agility did not allow a significant separation between the selection status of the players. These findings are surprising as they contradict former study results related to U14 youth basketball players that highlight the relevance of speed-related outcomes for predicting their future success (Torres-Unda et al., 2013), or found players from better-ranked teams to be faster, and more agile than players from lower-ranked teams (e.g., Ramos et al., 2020). On the other hand, besides such speed-related abilities, there are also other factors that may contribute to a higher probability of getting selected. Indeed, superior performances of selected players were found for the performed jumping ability tests. Specifically, CMJ performance (i.e., vertical jump ability) was detected to be significantly better for selected players. This is in line with former research that found CMJ power to be a relevant predictor of individual performance in U14 youth basketball players (Ramos et al., 2019). Similarly, Torres-Unda et al. (2016) observed that players who performed better were assigned to greater jump capability. This underlines the importance of jumping ability for players’ performance, and presumably also for the selecting coaches assessing the potential of the players and preparing them for participation in international competitions. Vertical jump performance displays a relevant performance indicator for success in adulthood, and this is also supported by the fact that basketball players who play at a higher level have advantages in this regard (Ziv & Lidor, 2010). Further, this ability is essential for basketball players as they perform between 40 and 50 jumps during a game while, for example, making shots on offense or rebounding on defense (Ben Abdelkrim et al., 2007; García et al., 2020).

To have a plain view into the importance of the utilized motor performance diagnostics, selection-dependent differences were also analyzed when controlling for players’ relative age and biological maturity status as potential confounders (H2b). Again, LS, MAT, and SLJ did not significantly contribute to the discrimination of players’ selection status when controlling for relative age and maturity offset and, therefore, led to similar results compared to the findings without a control for the confounders relative age and maturity offset. In fact, previous studies controlling, for example, for maturity status also found similar results before and after controlling for outcomes (Guimarães, Baxter-Jones, et al., 2019; Ramos et al., 2019). At first sight, this might support the tendencies presented in current research in this field, noting limited benefits of using maturation-corrected motor performance scores to predict future success. Charbonnet et al. (2022), for example, did not find an additional predictive benefit when controlling for biological maturation in their study with 15-year-old Swiss soccer players. While there is still only limited evidence from an empirical standpoint, the authors recommend performing corrections when conducting performance diagnostics in youth samples. Indeed, when looking at the findings of the present study regarding the model including CMJ performance, the significant discriminatory power partly disappeared when controlling for those confounders given the non-significant overall model. Previous research in youth basketball has shown that biological maturation positively impacts vertical jump performance (Guimarães et al., 2021, 2023; Guimarães, Baxter-Jones, et al., 2019; Guimarães, Ramos, et al., 2019), and, therefore, might also impact the discriminatory power of the respective tests. This may have resulted in kind of a suppression effect where biological maturity status partly covers existing (selection) effects given by the CMJ performance. This may have led to an attenuated effect of CMJ performance on the selection status. On the one hand, this finding might indicate a first empirical finding that reinforces the need for controlling for maturity-related information when evaluating the discriminatory power of motor test performances within youth basketball players. On the other hand, and despite the non-significant overall CMJ model, the consideration of the regression coefficient for CMJ revealed a one-tailed significant value. This is in line with the results of H2a indicating CMJ as a relevant predictor that might have been somehow attenuated when controlling for relative age and biological maturation. This together with the fact that this finding was restricted to this specific outcome within a small sample (i.e., pre-selected elite U15 players), it is important not to generalize the present findings without further efforts. In general, the impact of controlling motor performances for relative age and biological maturation was rather low. Thus, further studies are needed that investigate the efficiency of relative age and/or biological maturity-based correction procedures in different and larger samples comprising, for example, further age groups or performance levels.

Limitations and future directions

Maturity status in the present study was assessed by estimating somatic age via the maturity offset method (Mirwald et al., 2002), a pragmatic diagnostic frequently used in youth sports (e.g., Arede, Oliveira, et al., 2021; Fragoso et al., 2021; Lüdin et al., 2022). Compared to the expensive and time-consuming reference standard estimation by radiographs or magnetic resonance imaging of the wrist and hand bones, such pragmatic methods provide an affordable and simple assessment. However, these estimations involve measurement errors, especially when applied to samples that do not correspond with the original reference data (Fransen et al., 2021; Kozieł & Malina, 2018; Malina et al., 2012). Indeed, the present study included a sample of comparably tall youth players (i.e., 183.34 cm; see Table 1) that had already exceeded the average height of the 18- to 20-year-old male population in Germany (i.e., 181.8 cm; Statistisches Bundesamt, 2023). While this is not surprising as height comprises an essential performance factor for a basketball player, such specificities are important to consider with respect to the accuracy of the assessment methods. Here, further studies are needed to evaluate the reliability of pragmatic methods based on somatic age within such special populations. Particularly for samples with tall individuals, it seems at least questionable as to whether the somatic age diagnostics that incorporate global body measures such as weight, standing height, and sitting height (Mirwald et al., 2002) enable to accurately assess one’s maturity status. Instead, further indicators of maturity status such as skeletal age should be referred to that consider further facets in addition to, for example, height and weight, and therefore, may be less susceptible to measurement errors in populations beyond normal level of body size. For this purpose, skeletal age assessment by ultrasound diagnostics seems promising. Meanwhile, mobile ultrasound devices exist that estimate an individuals’ skeletal age based on three various spots and/or positions of the hand (Rachmiel et al., 2017; Utczas et al., 2017) and have been shown to be reasonably accurate when compared to reference diagnostics in comparable settings in soccer (Leyhr et al., 2020; Rüeger et al., 2022).

In addition, potential confounders that might have an impact on the study results need to be discussed. First, the study setting and competition level of players (i.e., selection tournament of U15 national team trialists) naturally resulted in a rather small sample size and represents a common issue within talent research (e.g., Hecksteden et al., 2021) restricting the statistical power of the analyses and the generalizability of the study’s results.

Second, factors that are related to the environment of the investigated players might have influenced the study results. For instance, the support of the family seems to be an important resource for talent development as it provides a player with social support and, therefore, influences a player’s performance and his or her chance of being selected (e.g., Lenze et al., 2023). Further, there is some indication that it might play a role whether a player was born in a rural or urban area. This is an issue currently being discussed in talent research as a potential birthplace effect (e.g., Maayan et al., 2022).

Third, person-related confounders are also worthy of consideration. In the current study, players’ ethnicity was not assessed. Ethnic differences in growth and maturation serve as a potential source of error for the estimation of players’ maturity status. However, the variance in growth and maturation appears to be much greater within ethnic groups than between them. Grgic et al. (2020) were able to demonstrate differences in skeletal age development among European and African children which might imply the need to consider ethnicity also in the estimation of maturity status. In contrast, Timme et al. (2017) stated that there is no demonstrable effect of ethnicity on skeletal maturity. Therefore, any potential influence of ethnicity and the associated necessity of adjustments, for example, to the somatic age estimation formulas, remains unclear but represents a research gap that should be addressed in future studies. Then, the years of playing basketball (and other sports) as well as the training load in every day practice may comprise confounders that might have affected the study outcomes (e.g., the investigated motor performances). With regard to previous experiences in early stages of development, Güllich and Barth (2023) suggest in a recent systematic review and meta-analysis that in individual and team sports higher-performing youth athletes were engaged in talent promotion programs at younger ages than their lower-performing counterparts. However, higher-performing senior athletes seem to be involved in such programs at older ages compared to lower-performing senior athletes. Specifically for basketball, Arede, Esteves, et al. (2019) demonstrated differences in physical parameters (e.g., jumping, sprinting) between U13 basketball players who were less and more specialized players in the ages of six to ten years. Further, analyzing real game performance data in addition to the considered motor performance indicators may allow for more in-depth results with regard to the investigated research questions. In this context, observational instruments can be used to evaluate players’ technical and tactical in-game performances in the matches played at the tournament (Rösch et al., 2022). Moreover, game-related statistics (e.g., points scored, assists, or turnovers) may be analyzed, which are conveniently recorded in elite youth basketball and regularly utilized by coaches to inform selection decisions in practice (Butterworth et al., 2013; Rösch et al., 2021). For instance, Arede, Fernandes, et al. (2021) demonstrated that performance data (i.e., minutes played, and points scored at the FIBA U16 European Championships as well as points scored at the Portuguese National Championships) were associated with players’ selection status for the Portuguese U18 youth national team. Going beyond, the authors found group-based differences in the data depending on players’ maturity timing.

The consideration of game-related statistics may provide further insights into the investigated selection processes when utilized as criterion variables. The choice of the criterion variables for operationalizing success has a decisive role in the evaluation of the present study’s findings. Within the present study, we utilized the coaches’ selection decisions for further talent development purposes of the German U15 national team (i.e., training camp) to differentiate between successful and non-successful players. On the one hand, this corresponds to the common procedure of making binary decisions (i.e., yes or no) as part of a selection. On the other hand, it would also be important to consider, for example, the actual game performance of the players as a criterion in addition to this binary decision criterion (Bergkamp et al., 2019; Berri et al., 2011; Rösch et al., 2021). For example, game observations or common performance indices of a basketball game (i.e., game-related statistics) would lend themselves to evaluation in the context of further studies.

In addition to such a concurrent validation that was followed within this study, a prognostic validation seems warranted, investigating the performances of players who may have the potential to be successful athletes in adulthood (Johnston et al., 2021). Specifically, the extent to which maturity-related characteristics also influence future success or the predictive power of motor tests for future success should be investigated. To pursue such an approach, a longitudinal study design is required. Doing so makes it possible to determine whether the examined players will perform at an elite level in (e.g., competing in domestic or international leagues on club level, reaching German senior national team). But, and not less important, a longitudinal approach that includes a regular monitoring of maturation-related information and performance over the course of the talent development process of a player would enable to investigate whether there are certain selection stages where maturation’s impact on talent selection is higher than in other stages. This means, for instance, to analyze in which selection stage (e.g., at the club or regional level) or at which specific age late-maturing players are more likely to be de-selected already prior to the investigated national selection tournament (where maturation had a certain, but indeed smaller impact in selection).

Further, the present study only included male youth basketball players. However, gender-specific differences in biological maturation require similar investigations as in the present study with female samples. Here, it is by no means clear as to whether findings in male athletes can be transferred to females. For instance, Baxter-Jones et al. (2020) highlight that maturation-related physiological changes in females, such as increased relative fat mass, widening of hips, and breast development (see also Barbour-Tuck et al., 2018; Sherar et al., 2011; Siervogel et al., 2003), may not be conducive to performance. So, it might be assumed that especially the maturation process within females might also come along with disadvantages regrading performance and as such it may be hypothesized that later maturing females may be equally selected as earlier maturing females. Indeed, there is a limited number of studies investigating the influence of biological maturation on selection and/or performance in female basketball samples. While one recent study indicates its significant influence on performance indicators such as jumping, endurance, and sprinting (Gryko et al., 2022), further similar studies are needed in the future to create more scientific evidence on the influence of biological maturation on talent selection processes in female athletes.

In addition, the playing position displays an important facet to consider. Especially in basketball the position a player is assigned to (at least partly) depends on maturation-related anthropometric features. When observing players during the selection process, coaches evaluate possible positions where players might optimally play. While players with a higher maturity status might be taller and then are frequently used as centers, shorter players (associated with lower maturity status) are more likely appointed as guards (Arede, Ferreira, et al., 2019; Te Wierike et al., 2015). Coaches at the selection tournament have mentioned that basically more guards are presented there. As a smaller player who is then mainly used as a guard, it is therefore reasonable to assume that the selection chances for these players are then lower. Consequently, future studies on the influence of maturation on selection and motor performance in youth basketball should therefore differentiate between the specific requirements of playing positions (Ivanović et al., 2022).

Conclusion and practical implications

The current study found a strong pre-selection bias regarding relative age and biological maturity timing in the investigated sample of elite youth basketball players. Furthermore, again for the selection of the youth national team within the tournament, selected players showed higher values in maturity-related anthropometric data (i.e., height, weight) as well as with regard to biological maturation itself. Consequently, coaches working with youth players in preceding selection stages (i.e., regional and club level), but also on the national level should raise their awareness to players’ relative age and biological maturation when looking at players’ potential. Here, focusing on, for example, players’ biological maturation as an additional information for the evaluation of their performance might help coaches to prevent late-maturing players from being de-selected due to a currently delayed physical development. Strategies such as bio-banding (i.e., maturity matching of players in competition) and Futures programs (i.e., national teams for late maturing players) should be considered for integration within the existing programs. Such strategies have been shown to be successful in helping to retain and transition talented late maturing players and presenting both early and late maturing players with new challenges and learning opportunities (Malina et al., 2019). Bio-banding in basketball has also been shown to promote a style of play that places greater emphasis on technical and tactical ability and exploration of space (e.g., Arede, Cumming, et al., 2021).

The present study provides evidence for an added value of using maturation-corrected scores when analyzing motor performance diagnostics only to some extent. However, the use of such scores seems to provide a reasonable additive information for coaches when evaluating player performance. Indeed, a regular assessment of players’ biological maturation enables stakeholders involved in talent selection processes to create normative values for motor performance indicators according to biological maturation and would enable them to compare players’ test results with specific reference values (e.g., early, on-time, late) in the specific population (Ramos et al., 2021). In doing so—while anthropometric data needed for the estimation of biological maturity is usually acquired in motor diagnostics anyway (Gál-Pottyondy et al., 2021a)—with only little effort possible maturity-related bias in the performance data can be avoided by simple corrections. As a result, coaches or scouts get an additional view on the players which might put certain players (i.e., late-maturing players) into coaches’ focus which would have been deselected otherwise.

Author contributions

Conceptualization, D.L., D.R., O.H.; data curation, D.R., D.L.; formal analysis D.R., D.L.; investigation, D.L., D.R.; methodology, D.L., D.R., O.H.; project administration, D.L., D.R.; resources, D.R.; supervision, D.L., O.H.; validation, D.L., D.R.; visualization, D.R., D.L.; writing—original draft, D.L., D.R.; writing—review and editing, D.L., D.R, S.P.C., O.H. All authors contributed to the article and approved the submitted version.

Acknowledgments

We thank the German Basketball Federation (Deutscher Basketball Bund, DBB) for the provision of data and the valuable support. We also thank our colleagues for productive discussions and critical reading.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

A de-identified version of the raw data supporting the findings of this study will be made available by the authors upon reasonable request.

Additional information

Funding

The author(s) reported there is no funding associated with the work featured in this article.