2.1. Study Design
In the initial phase of the investigation, the authors of the study collaborated with futsal experts to gather opinions regarding futsal-specific agility. The authors presented the basic idea, technical details, and theoretical boundaries of test execution (i.e., available equipment) and asked the experts to express their opinion about the most appropriate scenario which will be convenient for the purpose of testing of the futsal specific CODS and RAG for all field players (e.g., defenders, wingers, and pivots). The applied knowledge obtained allowed the development of relatively simple testing procedures that evaluated different agility components of the sport (please see Procedures section below for details).
This study comprised of a repeated measurement design in order to define the reliability of the newly developed tests. Additionally, throughout the cross-sectional design, this study compared basic anthropometric variables, sprinting-, CODS- and RAG-performances from two groups of futsal players of varying performance-levels in order to identify the construct validity of the applied tests. First, a randomly selected subgroup of 11 players performed the agility protocols on two separate testing occasions (e.g., test-retest) that were separated by 6 days to define inter-testing reliability. The intra-testing reliability was obtained from the results of all the participants involved in the study (n = 32). Finally, the groups based on performance (i.e., team-level and top-level; see further for more details about dividing into performance-levels) were compared on dependent variables (body mass, body height, sprinting-, CODS- and RAG-performance).
Thirty-two male professional futsal players (age = 26.2 ± 5.2 years; body height = 182.1 ± 6.0 cm, body mass = 77.4 ± 8.0 kg) from three futsal teams, competing at the highest national level in Croatia, voluntarily participated in the study. Although more players were originally tested (e.g., 35), in this study, the participants were selected based on the following criteria: Minimum 7 years of active involvement in futsal; older than 18 years of age; free from injury or illness; and have regularly performed standard training for at least three weeks prior. The goalkeepers were not included in this investigation. For the purpose of this study, the total sample was divided into two groups based on the performance level: Top-level players (12 players) and team-level players (20 players). The top-level players were those who met at least one of the three following criteria: (1) They were members of senior-level national futsal team over the last two years; (2) were members of the junior-level national futsal in the last competitive season (<18 years); (3) participated in the Union of European Football Association (UEFA) Futsal Champions League over the last two years, which is the highest competition level for futsal teams in Europe. The team-level players were those who were not grouped as the top-level players (please see previous criteria) and who were members of the teams participating at the highest national competitive level for the observed season.
During the timeframe of testing, the players participated in 6–7 training sessions and one official game each week. One specific session per week was exclusively dedicated to weight training, pre-habilitation, and exercises designed to improve aerobic-anaerobic endurance. The remaining sessions involved technical-tactical preparation (60% of training focus), combined with small sided drills and skill-based games (20%) and training games (20%).
The ethics board of the first author’s institution provided approval of the research experiment (Ethical Board Approval No: 2181-205-02-05-14-001). All the participants were informed of the purpose, benefits and risks of the investigation. The participants voluntarily took part in the testing after they provided written consent.
The variables in this study included body mass (BM), body height (BH), 10-m sprint (S10M), as well as the newly developed futsal-specific CODS- and RAG-tests. All the assessment methods followed standardized procedures and were performed with calibrated equipment by experienced evaluators.
For the S10M test, the participants were placed 1m behind the trigger line with their body leaned forward. The first timing gate (Powertimers 300, Newtest Oy, Oulu Finland, Core serial number: 08310013) remained on the trigger line (0 m), and the second timing gate was placed at the finish line (10 m). Each timing gate had reflecting sensors at 1 m height. The participants were told not to include backward movements at the start and to sprint at maximal speed the whole distance with avoiding a dive finish. Three sprint trials with a rest period of 2 min between each were performed. The inter-trial reliability based on the results of three attempts was high (Intraclass correlation coefficient = 0.90), and the best achievement of each participant was used for the analyses.
The agility-components were tested by newly developed futsal-specific CODS and RAG tests. The performance during CODS and RAG followed two procedures: (1) The participants had to touch the ball at the precise moment a change-of-direction occurred (CODS_T and RAG_T, respectively); and (2) the participants dribbled a ball during the execution of each test (CODS_D and RAG_D, respectively). All tests had a Y shaped pattern with the distances specified in the Figure 1
a for tests which involved dribbling; Figure 1
b for tests which involved ball touching). The timing for the RAG tests began when the participants crossed the initial infrared signal. At that moment, a hardware module lit one 30 cm high cones (A or B). As no prior indication was provided for the RAG tests, the participants had to quickly notice the specific lit and react accordingly. Thus, the RAG performance was non-planned. For the CODS tests, the participants had advanced knowledge on which cone will light up, and therefore were able to pre-plan the movement template.
For the RAG_D and CODS_D, the participants were instructed to dribble a ball (Figure 1
a) to a marked circle on the ground in front of the designated cone. The participants left the ball within the circle and then changed their direction to run back to the starting line as quick as possible (Figure 1
a). For the CODS_T and RAG_T, the participants had to run to the ball, which was placed in front of the cone, touch it with the sole of the foot and then run back through the infrared signal to stop the timer (Figure 1
b). The RAG and CODS tests were performed over five trials with either the known scenario (for CODS_D and CODS_T), or unknown scenario/template (for RAG-D and RAG_T). In some previous investigations, the authors used live people and video as stimuli for the change of direction in RAG testing to assure more realistic testing procedure [28
]. However, this study decided to use light-stimuli because it allowed standardized conditions for all players. Specifically, this was one of the first studies which included ball-dribbling in test execution, and therefore, this study tried to ensure a controlled testing environment.
For the purpose of this study, the performance on the dominant and non-dominant sides were analyzed. To determine the dominant and non-dominant side, the mean value for all B-cone performances (i.e., executions on the right side), and all A-cone performances (i.e., executions on the left side) for each participant and each executed test (e.g., CODS_D, CODS_T, RAG_D, and RAG_T) was calculated first. The performance side with the lower mean value was determined as the dominant side for each executed test (for each player separately). As a result, eight agility-performances were recorded as follows: RAG with dribbling performed on dominant- (RAG_DD), and non-dominant sides (RAG_DND); RAG with ball touching at the moment of change of direction performed on dominant- (RAG_TD), and non-dominant sides (RAG_TND); and the corresponding performances on each CODS test (CODS_DD, CODS_DND, CODS_TD, and CODS_TND, respectively).
The measuring of the CODS -and RAG-performances was performed by a hardware device based on an ATMEL micro-controller (model AT89C51RE2; ATMEL Corp, San Jose, CA, United States). A photoelectric infrared sensor (E18-D80NK) served as an external time triggering input, and light emitting diodes were used as outputs. The photoelectric infrared sensor (see Figure 1
—IR) has a response time of less than 2 ms and a digital output signal. The sensor’s distance for detection ranged from 3 to 80 cm with the capability of detecting transparent objects. The sensor was connected with a microcontroller IO port (Figure 1
—IR). The device was connected to a PC operated on Windows 7 operating system.
The testing was performed in an indoor facility on plastic turf where all participants regularly trained and competed. All players were tested at the same location at the approximately the same time of the day in order to avoid diurnal variations (between 16:00 and 19:00). The standardized warm up was performed before testing that included (in order): 5 min of light jogging; 5 min of combined lunges, jumps, one-leg hops, and changes of direction; and 5 min of dynamic stretching exercises. The testing was arranged in groups of 3–4 participants which allowed for the appropriate rest intervals between the tests and trials. A day before the data from CODS and RAG were collected, all players were familiarized with the procedures. The original testing field scenarios were applied, but players were instructed to perform each given test at a submaximal effort to identify the most appropriate movement pattern. Prior to the familiarization trials, each test was briefly and verbally presented to the participants by one of the authors of the study. All participants performed 2–3 familiarization trials of each test. The same day of familiarization, the participants were tested for sprinting capacity and anthropometrics variables. The CODS and RAG tests were performed at maximal effort for the data collection over the two days that followed familiarization. The CODS and RAG tests were performed in a random order, in which half of the sample performed CODS first and RAG second, and vice versa for the other half. All testing trials were performed after either protocol was initiated. The rest periods between each trial and each test was 2–3 min and 4–5 min, respectively. The trial for familiarization, sprint performance, and anthropometric data among the 11 players who performed the test-retest reliability procedures occurred 1 week prior to the other subjects. The CODS and RAG data collection for this cohort took place over the following two days in randomized order, according to the previously explained protocol. Then, this group was re-tested on CODS and RAG during the subsequent week, together with the other subjects.
2.4. Statistical Analysis
All the data were log-transformed to reduce the non-uniformity of error, and normality was tested using the Kolmogorov–Smirnov test procedure. The homoscedasticity for all variables was tested by Levene’s test. The descriptive statistics included the means and standard deviations presented as the true results for each variable (non-log-transformed).
The intraclass correlation coefficient (ICC) was used for analyzing the relative reliability, and the coefficient of variation (CV) was used for analyzing the absolute reliability [29
]. The ICC was calculated from the variance estimates derived from a repeated measures analysis of variance, where the test and retest results were observed as repeated measures. The calculation of inter-testing ICC was done on the basis of the results of those participants who participated in two testing sessions (i.e., the 11 participants in the test-retest cohort).
Further, to define the intra-testing reliability (i.e., reliability across the trials in one testing session), the means and SDs for all trials and all study participants were used. A repeated measures analysis of variance over the testing trials with the corresponding Tukey post hoc test, were used to assess systematic errors (e.g., learning, fatigue) among the trials. Further, the intra-testing ICCs and CVs were calculated as previously suggested [30
]. The ICC ≥ 0.75 and CV ≤ 10% were considered to indicate good (appropriate) reliability of studied CODS and RAG tests [32
The Pearson’s product moment correlation coefficients were calculated to define the associations between RAG and CODS performances.
The construct validity of the tests was assessed by comparing performance groups in each test. The differences between performance-groups were calculated using 2-sided t
-test for independent samples. Additionally, the effect sizes (ES; Cohen’s d) for the differences in sprinting- and agility-performances between the performance-group were calculated, and interpreted using the following qualitative descriptors: <0.2 as trivial, 0.21–0.49 as small, 0.50–0.79 as medium, >0.79 as strong [35