Background: Good change-of-direction sprint performance is considered important for basketball players. Regarding the components of a change-of-direction task, a correlation between the linear sprint and concentric power output can be assumed. Previous studies have shown heterogeneous correlation coefficients between different change-of-direction (COD) tests and linear and jump performance.
Methods: Therefore, 47 trained youth basketball players (13.2 ± 3.6 years; 34 males and 13 females) were recruited. Linear sprint (10 and 20 m), squat jump, and COD sprint tests (the zig-zag test and triangle test) were conducted. One-tailed Pearson correlation analysis was used to assess the relationship of sprint and jump performance with COD tests, and two-tailed analysis was used to assess the relationship between the different COD tests.
Results: This study showed that both linear sprint variables and concentric power output explained the variance in the different COD tests in a homogeneous manner (r2 = 0.14 to 0.42). The correlation coefficients did not statistically differ between the male and female players, and the different COD tests correlated moderately with one another (r2 = 0.42 - 0.46). For both analyses, the intercorrelation results did not differ significantly between sexes and analyzed correlation coefficients.
Conclusion: It is recommended that the squat jump, 10- to 20 m linear sprint, zig-zag and triangle tests are included in batteries of tests conducted in youth basketball players. COD tests with the same structure may have similar physiological requirements; however, these tests have task-specific characteristics. Therefore, coaches and sport scientists must review and select different tests with logical validity based on sport-specific requirements.
Athletic performance, Sports, Exercise test
In team sports, players must perform several high-intensity activities during a game [1,2]. These actions are important for scoring and even determining the outcome of a game . Many authors have suggested that power, agility and good sprint performance are important characteristics for basketball players [4,5]. Basketball players have to perform different movements every two seconds . Often, athletes run straight ahead and cut - they brake, change directions, and accelerate again - with a directional change of less than 90 degrees . Therefore, COD tests are of great interest for training purposes. The zig-zag test (ZZT) is commonly used in team sports, such as basketball . However, different authors have recommended different COD tests [3,9,10]. For example, COD tests such as the triangle test (TriT) are commonly used in team sport assessments . Different COD tests last for different durations, cover different distances, and require different degrees of directional changes and are therefore influenced by different variables [11,12]. Standardizing all COD tests does not seem possible due to the many available variations. The length and approximate total time of the TriT (10 m [m], 2.9 to 3.5 s)  and ZZT (approximately 20 m, 5 to 6 s)  demonstrate the different requirements of these tests .
CODs consist of an acceleration phase, which is similar to the acceleration phase in a sprint. This phase is typically followed by a deceleration/braking phase caused by eccentric muscle work  and a directional change (turn) that involves the manipulation of the base of support relative to the center of mass to apply an external force and reacceleration in a new direction [14,15]. Logically, a correlation between CODs and linear sprints as well as concentric power output can be assumed. However, a variety of correlation coefficients between different CODs and 10 m - 20 m linear sprints (LS) (r2 = 0.22 - 0.56) and jump performance (r2 = 0.06 - 0.22) have been reported in the literature [9,13,16-18]. This variety in correlation coefficients is also true for the correlation of the ZZT with the squat jump (SJ) (r2 = 0.01 - 0.26 - 0.79) [19,20], 10 m LS (r2 = 0.05 - 0.17) and 20 m LS (r2 = 0.03 - 0.21) [8,19]. Little and Williams  showed that 20 m LS performance has a higher impact on ZZT performance (r2 = 0.21) than 10 m LS performance (r2 = 0.12). Kadlubowski, et al.  showed a moderate correlation between TriT left turn (TritL) and SJ performance (r2 = -0.19) and a strong correlation between TriT right turn (TriTR) and SJ performance (r2 = -0.28). The authors found a strong correlation between TriTL and 10 m LS performance (r2 = 0.39) and TriTR and 10 m LS performance (r2 = 0.26).
To the authors' knowledge, no studies have 1) analyzed the influence of LS and jump performance on COD performance without decision-making processes (as indicated by the TriT and ZZT) in youth basketball players or 2) compared the correlations between the different COD tests. It was hypothesized that COD test performance is influenced moderately by both SJ and LS performance. The 20 m LS performance explains more of the ZZT variance than 10 m LS performance and vice versa for the TriT (depending on the total distance of the respective tests).
To answer the research question, the tests were carried out on two testing days. On testing day 1, the SJ and LS were conducted with a 15-min break between the tests. On testing day two, the participants were randomly assigned to one of two groups, which performed the TriT and ZZT in a different order and included a 15-min break between the tests. All participants were used to perform all the tests because they usually perform them semiannually.
Thirty-four male (13.1 ± 0.4 years old; 175.3 ± 8.8 cm; 60.8 ± 11.4 kg) and thirteen female (13.6 ± 0.5 years old; 170.2 ± 6.6 cm; 57 ± 4.1 kg) youth basketball players were recruited from different under 13-year-old (U13) and under 14-year-old (U14) elite youth training centers. The youth basketball teams played in the highest junior divisions in Germany and were part of a state selection team. The basketball players performed 3 to 4 basketball sessions per week (approximately 1.5 training sessions/day) and competed on the weekend. All participants had played basketball since their early childhood and therefore were very highly trained relative to other individuals of the same age. The participants did not participate in fatiguing training sessions for a minimum of 3 days before testing. None of the players reported any injuries at the time of testing.
All participants and their parents were informed about the experimental risks involved with the research. All participants and their parents provided written informed consent to participate in the present study. Furthermore, approval for this study was obtained from the institutional review board (German University of Health & Sport, No. 01/2019.92002800). The study was performed with human participants in accordance with the Helsinki Declaration.
The warm-up for the testing day consisted of nonspecific running at low-to-medium intensity for approximately 5 minutes (min). Then, coordination exercises, such as running with lifted knees, heeling, and side stepping, were performed for approximately 5 min. Afterwards, the athletes completed a 5-min dynamic stretching program (standing scales, hand walks, lunge steps with twisting and lateral lunges with rotation). Subsequently, 3 acceleration runs covering approximately 30 m were performed with short walking breaks. Overall, the total warm-up time on each test day was 20 min.
The subjects performed three attempts per COD test, with a 2-min break between each test. A description of the test setups can be found in the following previous studies: ZZT  and TriT . The total distance of the ZZT was 20 m, with 3 changes in direction, and the total distance of the TriT was 10 m, with 2 changes in direction. In contrast to the ZZT, which involves turning in both directions, the TriT was carried out for each of the two directions separately since the test only allows one turning direction. If the pylons or hurdle bars were knocked down or touched during COD testing, a follow-up attempt was completed. The tests were separated by a 15-min break. LS performance was measured for distances of 10 and 20 m. Every athlete had three attempts. Between each completed sprint, the athletes received a 2-min break . The time was measured for all tests with a double-timing gate system (wk7 time watch, Ditzingen, Germany). The starting point was marked with a small cap 0.75 m away from the starting gate to avoid early triggering, e.g., by a hand movement or a bent body position. The subjects chose independently when the measurement began according to the activation of the barriers. Thus, the reaction time was excluded from the measurement.
Jumping performance was measured using a contact mat (Refitronic, Schmitten, Germany) that operates as a switch. This system send information to the computer regarding whether the mat was loaded. From this information, the flight time and the jump height were determined for all jumps. The jump height was calculated from the flight time (gt2/8; g = the gravitational acceleration [9.81 m·s-2] and t = flight time). The squat jump was initiated at a knee angle of 90° without countermovement, and the hands were positioned on the hip. The participants performed 6 trials, and the best result was used for analysis. Between every jump, the athletes received a 1-min break.
The data were analyzed using SPSS 26.0. (IBM, Ehningen, DE, Germany). The significance level for all statistical tests was set at < 0.05. The descriptive statistics for all measures are presented as the mean ± standard deviation (SD). For the soccer players (n = 47), the Kolmogorov-Smirnoff test for normality confirmed that group's data were normally distributed. Reliability analyses were performed using the Intraclass Correlation Coefficient (ICC) and 95% confidence interval insert (95% CI). Furthermore, a bivariate one-tailed Pearson correlation analysis was used to assess the relationship between linear sprint and concentric power performance and performance for the different COD tests. To determine significant differences in the correlation coefficients between subgroups (male vs. female), the data were z'-transformed according to the Fisher method. The difference between the two transformed values after standardization was assessed for significance (). A bivariate two-tailed Pearson correlation analysis was used to assess the relationship between the different COD tests. Additionally, the inter correlation was assessed by correlation analysis with different COD tests. The correlation coefficients are displayed as the coefficients of determination (r2). Benjamini and Hochberg's method was used to control the study-wise false discovery rate to be 0.05 . The best time for each test was used for the statistical analysis. ICCs > 0.7 were classified as acceptable .
The Kolmogorov-Smirnoff test results showed that all parameters were normally distributed. COD performance for all the different tests, the ICC and the 95% CIs for the performance tests are displayed in (Table 1). The ICCs of the tests were greater than 0.76, indicating very strong to nearly perfect correlations and therefore good reliability according to a previous study .
Table 1: Reliability of the performance variables. View Table 1
The one-tailed Pearson correlation coefficients between LS and SJ performance and COD performance are shown in (Table 2). The correlation coefficients between the male and female subgroups did not differ significantly (p < 0.05). Therefore, all correlation coefficients presented correspond to the entire group. The highest correlation coefficients were observed between 10 m and 20 m LS performance (r2 = 0.83) and between ZZT and TriTL performance (r2 = 0.46). Moreover, the least significant coefficients were observed between TriTR and 10 m LS performance (r2 = 0.14) and between TriTR and SJ performance (r2 = -0.17). After adjusting for the study-wise false discovery rate for the values, all coefficients remained significant.
Table 2: One-tailed determination coefficients for linear sprint and concentric power performance and performance for the change-of-direction tests. View Table 2
The two-tailed Pearson correlation coefficients between the COD tests show the following results: there is a strong correlation between ZZT and TriTL performance (r2 = 0.46), as well as between ZZT and TriTR performance (r2 = 0.42). The analysis also showed a strong correlation between TriTR and TriTL performance (r2 = 0.44). The intercorrelation coefficients were not significantly different between sexes. For the pooled group the intercorrelation was calculated with z = 0.35 (p = 0.36).
The data of this study showed that 1) both linear sprint variables and concentric power output explain the variance in performance for different COD tests in a homogeneous manner (r2 = 0.14 - 0.42), 2) the correlation coefficients do not differ statistically between male and female players and 3) TriTR, TriTL and ZZT correlate moderately with one another (r2 = 0.42 - 0.46); for both analyses, the intercorrelation results showed no significant differences.
Interestingly, the correlation coefficients did not differ significantly for the influence of 10 m and 20 m LS and SJ performance on performance for different COD tests. Notably, there were no significant differences in the coefficients between the male and female players. SJ and 20 m LS performance correlated slightly more strongly with COD performance than 10 m LS performance (r2 = 0.26 - 0.41 vs. r2 = 0.17 - 0.42 vs. r2 = 0.14 - 0.29, respectively) for all COD tests, but the 95% CI did not indicate significant differences between coefficients. In general, this finding is not in line with those of other studies that showed a higher correlation of 10 m LS performance than SJ performance with COD performance (r2 = 0.19 - 0.56) [9,16-18,25]. Additionally, Little and Williams  showed that the 20 m LS performance has a larger impact on ZZT performance (r2 = 0.21) than 10 m LS performance (r2 = 0.12) does, which could not be confirmed in this study. In addition, it should be noted that SJ and sprint performance are correlated in a moderate to strong manner in different studies [26,27], so it is possible that inter correlations influence the calculated coefficients. The absence of significant differences in the coefficients between sprint performance and concentric power and for the COD tests may indicate that the two CODs are not heterogeneous enough to reveal differences. In general, the authors suggest that the strongest correlations between two variables are assumed to indicate the highest possible agreement in terms of neuronal, metabolic and/or morphological demands . Nevertheless, the results of this study indicate a moderate-to-strong influence of speed-strength variables on performance in the 10- to 20 m distance COD tests with 2- to 3 directional changes, independent of the direction of rotation.
The analysis of the correlations between the 3 COD tests in terms of performance showed strong coefficients (r2 = 0.42 - 0.46), with no significant differences between coefficients. This seem to be in line with the findings of Kadlubowski , who also showed a strong correlation between the TriTR and TriTL (r2 = 0.26). However, slightly more variance was found in the current study, likely due to study population being more heterogeneous and including younger athletes (13.1 ± 0.6 vs. 18.5 ± 4.5 years old). All tests in this study had a minimal difference in the distance, number and degree of COD. The similar neuronal, metabolic and/or morphological demands are supported by high coefficients, with non significant differences between coefficients. Nevertheless, 42 - 46% of variance was shown in this study, indicating there are still task-specific characteristics, even though there were small differences in the test structure (i.e., degree/direction of rotation, running speed) . Therefore, the results are far from the level, to assume a generalizability of the COD tests in team sports .
A limitation of this study is that a different number of male (n = 34) and female (n = 13) subjects were included in this study, which may provide an explanation for the heterogeneous correlation coefficients. However, the data set is valuable because only a few studies have investigated and compared young female and male basketball players. It should also be noted, that levels of reliability corresponding to ICCs of 0.72 to 0.81 can be considered acceptable . An ICC of 1.0 is expected when measuring the same skill. Therefore, an explained variance of 52% - 66% indicates a certain variability of the COD test and means that the results must be viewed with a certain degree of caution.
The results of this investigation, in general, show that COD performance is influenced by linear sprint performance and concentric power output. However, it is recommended that the SJ, LS, and COD tests are included in the batteries of tests performed by youth basketball players. COD tests with the same structure may have similar physiological requirements. Therefore, the tests assessed in this study do not seem to measure a 'general' ability to change direction but rather task-specific performance. Therefore, coaches and sport scientists must review and select different tests with logical validity based on sport-specific requirements. In general, plyometric and strength training is recommended to increase COD performance and should be supplemented by training for the corresponding target exercise.