INTRODUCTION

With over 1.2 billion fans and more than 80 million players, tennis is one of the most popular sports in the world (International Tennis Federation, 2021). Tennis is also a highly complex sport that requires not only general endurance but also speed, agility, upper body power, and coordination (Filipcic & Filipcic, 2005; Kramer et al., 2017; Robertson et al., 2018). Apart from professional tennis players needing these perfor-mance prerequisites to defend their top positions in the world (Reid & Schneiker, 2008), it is also necessary to understand which performance characteristics are already relevant for tennis players at an early age. This may not only contribute to a successful future tennis performance (Till & Baker, 2020); it may also orientate a best mover to this particular sport according to the particular strengths of his or her individual performance profile. Importantly, the better the individual talent characteristics match the (future tennis) demands, the higher the chances that the beginners will achieve success and satisfaction in this complex sport. This assumption is underlined by Suppiah et al. (2015), who state that a wrong choice can never be compensated for by training. Engaging in an unsuitable sport might not only be detrimental to fun but also lead to drop outs ahead of time. Conversely, if children like a recommended sport, the talented athletes could transform their physical, physiological, and psychological gifts through a long term process of diligent learning, deliberate practice, and an extended amount of high quality training into optimal achievements (Davids & Baker, 2007; Pion, 2015).

With the help of physical fitness and motor competence tests, young athletes are analyzed and their future potential is assessed (Kramer, Huijgen, Lyons, et al., 2016; Ulbricht et al., 2015). These so called sports orientation and talent identification (TID) campaigns have become increasingly common in recent years (Johnston et al., 2018). This is also due to national sports organizations investing more and more effort into the systematic identification of talented young players. In a professionalized competitive environment, a relaxed approach no longer appears acceptable (De Bosscher et al., 2008), and talent identification could become the key to national elite sport performance. Therefore, TID is designed to identify promising young athletes at an early stage (Hohmann & Seidel, 2003; Pion, 2015). Thus, the testing and scouting of athletes begin as soon as they enter a sport. On average, the general sport entry age of professional athletes is about 9 years (8.5 ± 2.5 years: Güllich & Emrich, 2014; 9.1 ± 3.7 years: Vaeyens et al., 2009). In this context, tennis and other racket sports are known to have an early starting age (5‒8 years) (Faber et al., 2016). Studies by Li et al. (2020) have shown that 75% of all Top 300 tennis players began playing tennis between 3 and 7 years of age, and only 4% of the Top 300 tennis players started after the age of 10. This might be related to the fact that, firstly, the former group of athletes had more time to freely gain competitive experience, and secondly ‒ particularly before puberty ‒ there are sensitive learning phases that promote motor learning and thus offer the opportunity for the acquisition of technical skills (Knudsen, 2004). In addition, a tennis education starting at a young age gives the coaches a longer observation period; this reduces talent selection errors during early adolescence and enhances practitioners’ talent identification decisions. Accord-ingly, it is not surprising that the evaluation of young athletes begins when they are under nine years (U9) of age (Potočnik et al., 2020; Tomkinson et al., 2017). However, because in many cases these young athletes have only little technical experience so far, these TIDs often consist of several generic test items rather than specific, more technically demanding skill tests (Hohmann et al., 2018; Niessner et al., 2020). This approach is supported by studies by Faber et al. (2020), who found a small but significant correlation between more sport specific, coordinative technical skills (e.g., speed while dribbling or aiming at a target) and previous training hours in racket players aged between 8 to 10 years old. In contrast, generic tests such as sprinting or standing long jump had no significant relationship with training volume.

While there are already some studies on the success of TID in other sports, such as soccer (see Sarmento et al., 2018), in tennis it is still questionable whether TID campaigns in the U9 age group (Hohmann et al., 2018; Pion, 2015) can provide any information whatsoever about later tennis performance. Although there are many studies on TID in junior tennis (U12‒U18) and in professional tennis (Baiget et al., 2016; Van Den Berg et al., 2006), there are either no studies or hardly any on younger age groups (e.g., U9). This might be related to the fact that for this young age group, which lacks a developed tournament and ranking system, no proper differentiation of better and weaker performance groups can be establish-ed (Siener & Hohmann, 2019). The assessment of different tennis performance groups usually arises with the entrance into the junior ranking system of the U12 and is then often determined by the ranking position (Ulbricht et al., 2016). For this reason, it has been common practice to use as a template for TID in youth professional adult tennis players’ profiles, collected from cross sectional studies (Hohmann & Seidel, 2003). This assumes that the same invariant skills are crucial for tennis success in both age groups. However, this has not yet been adequately demonstrated by long term studies (Baker et al., 2020). Therefore, for a prognostically valid evaluation of U9 test performances, either a prospective study design must be employed or these test performances must be considered retrospectively (see also Mostaert et al., 2020; Till et al., 2015). Either way, both approaches are highly time consuming and only possible in a longitudinal study.

The present study starts at the earliest possible point in time at which a distinction between diffe-rent performance groups in tennis can be made. From U12, tennis players are able to qualify for the national junior tennis rankings of the German Tennis Federation and thus stand out from other athletes. These junior tennis rankings cover the age range up to U18 (Deutscher Tennis Bund [DTB], 2020). Within the group of junior tennis players, two performance groups ‒ junior ranked players (RPs) and non ranked players (NPs) ‒ can be distinguished. Based on these two performance groups, it is possible to retrospectively analyze the performance shown in the physical fitness and motor competence tests of the U9. Thus, the aim of this long-term retrospective study was to compare the initial U9 physical fitness and motor competence test performances of today’s junior ranked and non‒ranked tennis players from the national tennis ranking list of the German Tennis Federation (U12‒U18) and, based on this, to address whether TID in U9 seems at all possible. Overall, this study makes the first attempt to close the gap in research on TID at U9 and to investigate the validity of talent prediction by physical fitness testing at this young age. If it transpires that at this young age, there are already significant differences in performance characteristics between later successful and less successful tennis players, these characteristics can be used to make a talent forecast for players when they are only 8 years old.

MATERIALS AND METHODOLOGY

General study design

In this retrospective (long-term) study, we examined whether there is a relationship between the playing success achieved by junior tennis players (U12‒U18) and this cohort’s general childhood performance (U9). For this purpose, 174 junior tennis players were divided into two groups based on their current playing success: (A) 16 players who have achieved a position in the official national tennis ranking lists of the German Tennis Federation (RPs); and (B) 158 players who have not achieved a position on the rankings (NPs).

All these players (A+B) had already been tested in the U9 category in two anthropometric tests, six physical fitness tests, and three motor competence tests (Fig. 1). This now allows us to analyze whether the two performance groups had already showed differences in their performance characteristics at the age of 8 and, if so, in which generic tests the differences were particularly evident. The results can then indicate whether certain characteristics/predictors detectable as early as U9 may point to later success at junior age and thus make TID worthwhile at this earlier stage.

Figure 1. Study Overview

Participants

A total of 174 junior tennis players (U12‒U18; M = 156.3 months, Min = 132 months, Max = 206 months) were included in this study. The sample consists of 62 girls () and 112 boys (). Among these junior tennis players, 16 RPs (n = 11, n = 5) achieved a place in the official national junior tennis ranking list of the German Tennis Federation (DTB, 2020). To be included in this junior ranking, a player must have at least 10 wins in junior ranking matches or have already earned a position in the adult rankings. All other junior tennis players (NPs; n = 158, n = 101, n = 57) were registered in clubs and actively participated in regional and local club competitions and tournaments but did not achieve the necessary number of wins in official ranking matches to earn a place on the official national junior ranking list of the German Tennis Federation.

As part of the study, all participants and their parents were fully informed about the content of the testing and the resulting studies, and their consent was obtained before the study began. The study and research design were in line with the Declaration of Helsinki and was approved by the ethics committee of the Municipality of Fulda (Germany), as well as the State Office of School Education of Fulda (Germany).

Measurements

Physical Fitness and Motor Competence Tests of U9 Testing

All junior tennis players were already tested at U9 with a test battery of two anthropometric (body weight and height) tests; six physical fitness tests (sprint, flexibility, arm and upper body strength, leg power, and endurance performance); and three mixed motor competence tests (coordination, balance, and ball throw performance). Each of the standardized tests was performed according to existing test protocols, which include a detailed description of the test items, the exact test set up, the demonstration of the test item, the execution of the test phases, and the measurements (Bös et al., 2009; Siener et al., 2021):

20-m Sprint

The sprint performance was recorded using a 20-m sprint (~21.9 yards). In each of the two possible attempts, the test persons started 0.3 m before the starting line. The time was stopped by means of light gates (Brower Timing Systems; Draper, USA). The test met an objectivity value of .86 and a reliability value of .96 (Bös et al., 2009).

Sideward Jumping

The number of two-legged jumps that a participant could perform between two 50-cm x 50-cm (1 cm ≈ 0.4 inches) squares within 15 s was measured. Only jumps where none of the boundary lines was touched counted. In total, two attempts were made, with a break of at least 2 min. between each attempt. The mean value of both tests was used for further calculations. The objectivity of this test is .99 and the reliability is .89 (Bös et al., 2009).

Balancing Backwards

The participants balanced backwards on a 6-cm, 4.5-cm, and 3-cm wide wooden beam. Two attempts were made on each beam, and the number of steps (feet fully raised) before leaving the beam was counted. A maximum of eight steps/points could be achieved per beam, so that the total maximum number of points for this test task is limited to 48. The test achieved an objectivity of .99 and a reliability of .73 (Bös et al., 2009).

Standing Forward Bend

In this flexibility test, the participants have to try to reach as low as possible with their hands. The ground height is evaluated as 0 cm, and everything that goes beyond that (below ground level) is entered as a positive value. The achieved value had to be held for at least 3 sec. in each of the two attempts. The test achieved an objectivity of .99 and a reliability of .94 (Bös et al., 2009).

Push-Ups

In this test, the participants had to perform as many push-ups as possible over 40 sec. An execution was only evaluated if the technique was correct and the test person subsequently lay back down in the starting position. Only one attempt was performed. Bös et al. (2009) rated the objectivity with .98 and the reliability with .69.

Sit-Ups

The sit-up test also evaluated the number of correctly performed sit-ups in 40 sec. Only one attempt was performed. The objectivity of this test was .92 and the reliability was .74 (Klein et al., 2012).

Standing Long Jump

In the standing long jump test, the jumping distance was measured in centimeters. Each test person had two attempts, of which the better attempt was recorded. Between both attempts, a complete break was ensured. The test achieved an objectivity of .99 and a reliability of .89 (Bös et al., 2009).

Ball Throw

In the ball throw test, the throw distance was measured in centimeters orthogonal to the line of release. The test persons threw with an 80-gm ball from a standing position three times in a row. As in the previous tests, the best value was evaluated. The test could be evaluated in our own studies (n = 1800) with a reliability of .77.

6 min Endurance Run

In the endurance test, the participants tried to run as many laps as possible around a 9-m x18-m volleyball field in 6 min. The achieved distance was noted in meters. The test was carried out by a total of 15 people at the same time. The objectivity of this test is .87 and the reliability is .92 (Bös et al., 2009).

All tests were conducted by qualified personnel during regular school hours (8‒12 am), and a uniform warm-up was held before starting. The 6-min endurance run was always the last test in the test series.

STATISTICAL ANALYSES

For all analyses, the software SPSS (version 26; SPSS Inc, Chicago, IL, USA) was used.

A univariate ANOVA found that age affects the data and that the performance of U9 players increased significantly with age. In order to avoid this age bias, bivariate regressions to age in months were used to z standardize the test value residuals separately for both genders (Siener et al., 2021). Since this procedure poses the risk that in homogeneous groups certain test values are distorted by the group composition, the results of about 4000 children predominantly not from sporting clubs (see Hohmann et al., 2018; Tomkinson et al., 2017) were additionally used for z standardization. Thus, all data are available as age and gender independent z values. To ensure a better comparison of the sprint data, the z values were additionally multiplied by “−1”, thus turning the better sprint results into positive z values.

T-Tests and a MANCOVA were used to check whet-her the test results differed significantly between the gender and tennis success groups. A covariate weight was chosen. Effect sizes for partial eta squared (ƞ2) smaller than 0.01 are interpreted as trivial, effect sizes between 0.01 and 0.059 are small, between 0.06 and 0.139 are moderate, and values higher than 0.14 are large.

To gain a better insight into the influence of the individual test items on later tennis success, bivariate correlations were also calculated. The Spearman correlations were classified according to the following pattern: trivial (0 ‒ 0.1), small (0.1 ‒ 0.3), moderate (0.3 ‒ 0.5), and large (0.5 ‒ 0.7).

RESULTS

The initial U9 test results of the junior tennis players show that in almost all test tasks, RPs perform significantly better than NPs (Table 1). Particularly noteworthy here are the standing long jump, the sideward jumping, the balancing, the endurance run, and the ball throw, which all have values of p < 0.001 (t-Tests). In sideward jumping, the maximum value of the RPs was five jumps (12.5%) higher than the maximum value of the NPs. In the ball throw and the standing long jump, the later better athletes had a clear advantage in their youth. In both tests, approximately 84% of the NPs did not reach the average test result of RPs. In the sideward jumping, the balancing, and the endurance run, 84% of the RPs were also above the average result of the weaker performance group. For the forward bend test task (p = 0.158) and body weight (p = 0.910), no significant differences in the two performance groups could be found. In addition, the later ranked players are on average 4 months older than their weaker tennis colleagues (p < 0.01).

In examining the raw scores of the two perfor-mance groups separately by gender, it is notable that no significant group difference can be found for boys (n = 112) in the test scores for body weight (p = 0.149), height (p = 0.8), sprint (p = 0.106), push-ups (p = 0.065), and forward bends (p = 0.111). All other test results showed significant group differences (p < 0.01). Girls (n = 62) showed comparable results. Only the test results for body height (p = 0.044) and sprint (p = 0.013) were also significant in contrast to the results for boys.

Also, after eliminating the age effect, RPs achieved better z-values than the NPs in all test items (Fig. 2). Except for the 20-m sprint, however, the ascending order of the single tests was almost identical in both groups. Also, in regard to the order of the RPs’ tests, the ball throw (Mz = 2.03) was in first place, followed by the standing long jump (Mz = 1.53), the endurance run (Mz = 1.42), and sideward jumping (Mz = 1.41). For the ball throw, female tennis players (Mz = 2.95, n = 5) performed significantly better (p < 0.05) than their male counterparts (Mz = 1.61, n = 11). The same holds true for body height (Mz = 0.92, Mz = 0.07) and the standing long jump (Mz =1.88, z = 1.37). The anthropometric measures of the RPs were Mz = 0.34 (body height) and Mz = −0.30 (body mass). RPs at U9 were, therefore, on average slightly taller and lighter than the NPs. However, while in a t-test comparison, significant differences between the two performance groups can be found in almost all generic test items (with the exception of p Forward bends = 0.086), there are no significant differences in the anthropometric test values of body mass (p = 0.078) and height (p = 0.234).

Table 1. ‒ Descriptive statistics for the former U9 test results of the junior tennis players.

Figure 2. Initial U9 test performances (z-values) of junior ranked tennis players, non- ranked tennis players, and non-athletes (* p ≤ 0.05; ** p ≤ 0.01).

The MANCOVA (Table 2) shows that body mass only had a significant (p < 0.05) influence on the test values of balancing, sit ups, and the endurance run. Gender only had a significant influence on the ball throw in the analyses (p Gender = .012; p Performance*Gender = .056). All other hypotheses concerning a difference between the genders had to be rejected. As already suspected in Fig. 2, a partial Eta squared of ƞ2 = .157 shows the strongest effect size of group differences in the ball throw. The standing long jump, the endurance run, and sideways jumping show moderate effect sizes, while all other effects are small. It is notable that the standard deviation (SD) for the RPs fluctuates strongly. For the NPs, on the other hand, the SD of almost 1, which is usual for z values, is achieved.

Looking at the box plots of RPs and NPs for the different test tasks (Fig. 3), it is notable that the RPs have the biggest advantage, especially in the ball throw. Nevertheless, not all of the 16 RPs can show very high values. There is also one athlete with a z-value of z = −1.2 and three athletes with z values below z = 1.0 (range of 6.06). All other RPs show above-ave-rage throwing performances. Such large fluctuations in performance cannot be seen in test items the endurance run (range of 2.3) and balancing (range of 2.16). Here the results are comparatively close to each other, and none of the RP values is below a z-value of z = 0. Also, in body mass, no major fluctuations can be detected in the RPs (range of 1.15). The results show that in the ranked player group 75% (12/16) of the tennis players at the age of 8 yrs achieved a very good test result (z ≥ 1.0) in the standing long jump, the endurance run, and the ball throw. In addition to the three tests mentioned, more than 50% of the ranked junior tennis players were initially (when U9) able to achieve a very good test score in the sideward jumping (68.8%) and sit ups (56.3%).

Table 2. ‒ Results of the MANCOVA for the different former U9 test disciplines

Apart from the forward bends, no significant results can be seen for the correlations of tennis ranking success with body height or body mass (Table 3). All other test values achieved significant correlations, mostly in the moderate range.

The ball throw reaches the highest correlation value with r = .360, followed by the standing long jump (r = .287), and the endurance run (r = .296). However, these two values cannot be considered a moderate result. Also, all other test values have small correlation effects.

Figure 3. Boxplots of the former U9 test results (z-values) of junior non-ranked tennis players and ranked tennis players.

Table 3. ‒ Bivariate correlations of the different U9 test results and the chance to achieve a ranking position in the official youth national ranking lists of the German Tennis Federation (* p ≤ 0.05; ** p ≤ 0.01).