Analysis of the variables that predict serve efficacy in young volleyball players

1 1 ANALYSIS OF THE VARIABLES THAT PREDICT SERVE 2 EFFICACY IN YOUNG VOLLEYBALL PLAYERS. 3 4 Carmen Fernandez-Echeverria1, Alexander Gil2, Alberto Moreno1, Fernando Claver1 5 and M. Perla Moreno1 6 7 1Faculty of Sport Sciences. University of Extremadura 8 2Faculty of Physical Activity and Sport Sciences. Saint Antonio Catholic University of 9 Murcia. 10 11 12 13 14 15 16 17 Abstract. 18 The main objective of this research study was to analyse, in volleyball, the 19 prediction level of different serve variables on its efficacy. The sample 20 was comprised of 2038 serve actions, corresponding to 36 female teams 21 participating in the Spanish Championship female in 2010 in the Under- 22 14 and Under-16 categories. The independent variables were: serve zone, 23 serve type, striking technique, in-game role of the server, reception zone, 24 receiver player and serve direction. The dependent variable was serve 25 efficacy. The multinomial logistic regression applied showed that all the 26 variables considered in the study, except for serve direction and striking 27 technique, predicted the efficacy of serve. The probability of increasing 28 serve efficacy was achieved through: executing standing serves, serving to 29 the space between players and serve reception by players other than the 30 libero. These results can guide coaches in volleyball player training 31 processes, with respect to the serve action. 32 33 Key words: Notational analysis, volleyball, performance, multinomial 34 logistic regression. 35 36 37 38 39 40 41 42 43 44 45 46 47 48 2 ANALYSIS OF THE VARIABLES THAT PREDICT SERVE 1 EFFICACY IN YOUNG VOLLEYBALL PLAYERS. 2 3 4 Abstract. 5 The main objective of this research study was to analyse, in volleyball, the 6 prediction level of different serve variables on its efficacy. The sample 7 was comprised of 2038 serve actions, corresponding to 36 female teams 8 participating in the Spanish Championship female in 2010 in the Under- 9 14 and Under-16 categories. The independent variables were: serve zone, 10 serve type, striking technique, in-game role of the server, reception zone, 11 receiver player and serve direction. The dependent variable was serve 12 efficacy. The multinomial logistic regression applied showed that all the 13 variables considered in the study, except for serve direction and striking 14 technique, predicted the efficacy of serve. The probability of increasing 15 serve efficacy was achieved through: executing standing serves, serving to 16 the space between players and serve reception by players other than the 17 libero. These results can guide coaches in volleyball player training 18 processes, with respect to the serve action. 19 20 Key words: Notational analysis, volleyball, performance, multinomial 21 logistic regression. 22 23 24 1. Introduction. 25 26 Volleyball is a sport where attack prevails over defence (Moreno et al., 2007). Different 27 modifications have been made to the rules of the game, in an attempt to offset this 28 imbalance, but always favouring defence: increase of the serve zone (1998) and the 29 incorporation of a new player, the libero (1998). The inclusion of the libero has led to 30 greater effectiveness both in reception (Joao et al., 2006; Callejón and Hernández, 31 2009) and in defence (Mesquita et al., 2007), although imbalance between attack and 32 defence still exists. 33 34 Due to this imbalance between attack and defence, one of the game actions that teams 35 consider to be very important is the serve (Moreno, et al., 2007), as an improvement in 36 its effectiveness may avoid or limit the construction of the opponent’s attack, and 37 therefore, favour defence. Thus, the serve has gone from being an action used to start 38 the game, to being an attack weapon that is able to score a direct point [ace] (Quiroga et 39 al., 2010). 40 41 The serve, just like attacking and blocking, has a purpose-determined nature and may be 42 decisive in a team’s performance (Drikos et al., 2009; Patsiaouras et al., 2009). The 43 benefit of the serve is not just that a point is immediately scored, but rather, the serve 44 has an influence on the later development of the game. Thus, it has been observed that 45 the execution of a good serve (power serve or with a tactical intention), affects the 46 reception performance (Quiroga et al., 2010; Quiroga et al., 2012) and the attacking 47 options of the opposite team, reducing first tempo attacks and increasing second tempo 48 3 attacks (Katsikadelli, 1996; Papadimitriou et al., 2004). This influence of the serve on 1 the attack options causes an increase in blocking performance, significantly increasing 2 triple block (Palao et al., 2004), which would facilitate the defence. 3 4 Due to the importance of the serve in the development of the game, many different 5 research studies have been developed to determine the typical characteristics of the 6 serve and how this affects the opposite team, both at a high level and in formative 7 stages. To this end, the typical variables of the serve action have been analysed, such as 8 the serve zone (Quiroga et al., 2010; Travlos, 2010), the serve type (Marcelino et al., 9 2011; Costa et al., 2012), the serve direction (Gil et al., 2011b; Marcelino et al., 2011), 10 the in-game role of the server (Afonso et al., 2012) and even some more kinematic type 11 variables such as the serve speed (Coleman et al., 1993; Quiroga et al., 2012) or 12 situational variables, such as the set number (Campos et al., 2014) and moment in the 13 set when the serve is executed (Marcelino et al., 2012). 14 15 Another of the variables most frequently analysed in connection with the serve, are the 16 variables related to reception such as the receiver player (Quiroga et al., 2012), the 17 reception zone (Lidor et al., 2007; Afonso et al., 2010) and the receiving technique 18 (Afonso et al., 2012). 19 20 The majority of the research studies conducted on the serve are descriptive studies, 21 which analyse the frequency of occurrence of each serve type, or whose aim is to 22 discover the existing association between different serve variables. These studies have 23 been carried out both at high level (Papadimitriou et al., 2004; Moreno et al., 2007; 24 Quiroga et al., 2010), and in formative stages (Gil et al., 2011a; Costa et al., 2012; 25 Dávila-Romero et al., 2012). 26 27 Descriptive statistics help us characterise the game actions, but recently and so as to 28 discover which factors determine performance in the different game actions, the 29 multinomial logistic regression analysis has started to be used, thus permitting an 30 estimation of the predictive range of the variables. The application of this statistical test 31 in the analysis of sporting performance gives rise to greater comprehension of the 32 fundamental dynamics of the sport (Nevill et al., 2002). However, this test has largely 33 been applied, until now, in high-level research studies (Marcelino et al., 2011; Afonso 34 et al., 2012; Costa et al., 2012). 35 36 Due to the reduced number of studies conducted on formative stages and to the 37 implications that these studies may have, we believe it advisable to conduct studies that 38 provide relevant information regarding the serve action for the training process of 39 players in formative stages, taking technical and tactical factors into account. 40 41 Thus, the main objective of our research study was to analyse the prediction level of 42 different serve variables (serve zone, serve type, striking technique, in-game role of the 43 server, reception zone, receiver player and serve direction), on the serve efficacy, in 44 formative stage volleyball. 45 46 The hypotheses of this study are: 47 4 - The variables: serve zone, serve type, striking technique, in-game role of the server, 1 reception zone, receiver player and serve direction, will act as predictors for the serve 2 efficacy. 3 4 - Standing serves will increase the serve efficacy compared with jumping serves. 5 6 - Serves made to the space between players will increase the serve efficacy, compared 7 with serves made to the mid field zone. 8 9 - Serves made to a player other than the libero will increase the serve efficacy, 10 compared with serves made to the libero. 11 12 2. Methods 13 14 2.1 Sample 15 16 The study sample was comprised of 2038 serve actions, corresponding to the 17 observation of 36 teams participating in the Spanish Championship female in the 18 Under-14 and Under-16 categories. The number of teams and serve actions observed 19 per category were distributed as follows: Under-14 female (18 teams/1029 serve 20 actions) and Under-16 female (18 teams/1009 serve actions). To guarantee data 21 homogeneity, the serves analysed correspond to the observation of one match played by 22 each team to be studied. The results of the observed matches where three sets to two and 23 three sets to one, these results assure similar level of play in both matches. Furthermore, 24 due to the championship been played in neutral field for both teams, it was not 25 necessary to take into account whether the teams were playing at home or away. 26 27 The protocol was fully approved by the Research Ethics Committee of the University of 28 Extremadura (Spain). All players and their parents or guardians were fully informed 29 about the study, and they signed a consent form. 30 31 2.2 Variables 32 33 The following independent variables were considered in our study: (a) serve zone, 34 defined as the zone from where the serve is carried out, covering a 9-metre wide space 35 located behind the baseline of the court and as an extension to the sidelines of the court, 36 differentiating three zones of origin (zone 1, zone 6, zone 5); (b) striking technique, 37 defined as the type of serve technique used by the player, considering the flight 38 trajectory of the ball after striking (power or float); (c) serve type, defined as the type of 39 serve used by the player, considering the location of the player at the time of contact 40 with the ball (standing or jump); (d) reception zone (figure 1), defined as the zone where 41 the serve is received (lane 1, lane 6, lane 5 and space between players); (e) in-game role 42 of the server, defined as the in-game role of the player serving (setter, receiver-attacker, 43 middle attacker, opposite); (f) receiver player defined as the in-game role of the player 44 who the serve is aimed at for reception (forward-attacker, libero and other players); (g) 45 serve direction, defined as the direction determined by the serve depending of the serve 46 zone and reception zone (parallel, mid cross-court and long cross-court). 47 48 5 Lane 5 Lane 6 Lane 1 1 Figure 1. Zone where the serve is received 2 3 The dependent variable considered in our study was; serve efficacy, defined as the 4 performance or effect obtained with the serve. In order to assess the effectiveness, the 5 FIVB statistics system was used, adapted from Coleman (1975), grouping together 6 some values that were differentiated in this system. Three levels were established: (I) 7 permits attacking, “Serve that permits the construction of an attack, with all the attack 8 options or with limited options"; (II) does not permit attacking, "Serve that makes it 9 impossible to construct the attack, which is the equivalent to serving a free ball"; (3) 10 Point, “Serve that directly scores a point". 11 12 The aforementioned variables were measured via the systematic observation of the 13 serve. 14 15 2.3 Procedures 16 17 Initially, and to guarantee the validity of the instrument, a group of four researchers – 18 experts in analysing volleyball performance – developed a first version of the 19 instrument, in agreement with the theoretical framework and based on existing 20 bibliography. Later, this instrument was used by a group of 10 volleyball coaches, thus 21 enabling the creation of a final version of the instrument that would guarantee its 22 validity for the study (Okoli and Pawlowski, 2004). 23 24 The data were later collected on video. The matches were recorded using a SONY 25 HDR-XR155 digital camera (M2TS format). This camera was located at one of the ends 26 of the court, guaranteeing a height of 5 m above floor level and a distance of 7 metres 27 behind the baseline, to obtain an optimal line of sight. 28 29 For the reliability of the observation, after collecting the video footage and previously to 30 the coding process, one observer with experience in this function were trained to 31 observe and encode game actions, carried out a training process, using, in the different 32 training sessions, samples with different characteristics, and exceeding 10% of the total 33 sample (203 serves), indicated by Tabachnick and Fidell (2007). The intra-observer 34 Cohen’s Kappa values reached, in the observation of all the variables, were higher than 35 .81 (in table 1), in the sixth training session, which was the minimum value considered 36 to be almost perfect agreement (Landis and Koch, 1977). To guarantee the time 37 Between players Between players 6 reliability of the measurement, the same coding was developed on two occasions, with a 1 time difference of 10 days, obtaining Cohen’s Kappa values of over .81. 2 3 Table 1. Reliability of intra-observer 4 Intra-observer Intra-observer (with a space interval of 10 days) Serve zone 0.92 1.00 Striking technique 0.88 0.92 Serve type 0.92 0.92 reception zone 0.92 0.92 In-game role of the server 1.00 1.00 Receiver player 1.00 1.00 Serve direction 0.92 1.00 5 6 2.4. Statistical Analysis 7 8 Previously, the absence of multicollinearity was verified through the level of tolerance 9 and the inflation factor (VIF). The tolerance values of the model are greater than 50%, 10 so there are no collinearity problems, as the values are a long way from 0. The inflation 11 factor values of the model are less than 5, which implies an absence of collinearity, as 12 the acceptable value of VIF is 5 or less than 5 (Kleinbaum et al., 1988). 13 14 A multinomial logistic regression analysis was performed to obtain the estimated 15 occurrence probabilities of the dependent variable based on the values of the 16 independent variables. The level of significance was established as p< .05. 17 18 The association of each independent variable with the dependent variable (crude odds 19 ratio) was verified to determine which variables presented statistical significance and 20 thus enter them into an adjusted model (adjusted odds ratio). This step provides 21 information about which independent variables contribute to predicting the dependent 22 variable. According to Field (2009), the interpretation of the odds ratio (OR) values 23 indicate that if they are greater than 1 as the predictor increases, the odds of the outcome 24 occurring also increase. On the other hand, a value less than 1 indicates that as the 25 predictor increases, the odds of the outcome occurring decrease. 26 27 3. Results. 28 29 It was initially verified (in table 2), through an inferential analysis using Chi-cuadrado 30 (x2), that all the variables, except for the striking technique, had a significant association 31 (p<.05) with efficacy. 32 33 Table 2. Association of independent variables with the dependent variable. 34 X2 P. V de Cramer Serve zone 17.316 .002 .047 Striking technique 5.812 .055 .039 Serve type 11.158 .004 .054 reception zone 151.223 .000 .140 7 In-game role of the server 16.328 .012 .046 Receiver player 90.589 .000 .109 Serve direction 57.250 .000 .086 1 2 In table 3, we present the results over the complete multinomial logistics regression 3 model. 4 5 Table 3. Adjusted model for serve effectiveness. 6 Serve variables Permits attackinga % Does not permit attacking % OR Crude OR Adjusted P Point % OR Crude OR Adjusted P Serve zone Zone 1 58.7% 22.8% 0.879 (0.697-1.109)c 0.909 (0.706-1.171)c .460 18.5% 0.802 (0.626-1.027)c 0.811 (0.621-1.058)c .123 Zone 5 48.9% 29.2% 1.352 (0.987-1.1853) 1.531 (1.083-2.165) .016 21.9% 1.140 (0.809-1.605) 1.218 (0.840-1.766) .299 Zone 6 b 54.5% 24.1% 21.4% Serve type Standing 54.1% 25.6% 1.566 (1.167-2.103) 1.623 (1.193-2.280) .002 20.2% 1.064 (0.800-1.415) 0.992 (0.734-1.341) .960 Jump b 50.5% 18.3% 21.2% Reception zone Lane 1 58.1% 23.5% 0.910 (0.674-1.229) 0.858 (0.602-1.225) .400 18.3% 1.116 (0.798-1.562) 1.134 (0.771-1.668) .524 Space between players 34.5% 25.3% 1.644 (1.103-2.450) 1.138 (0.668-1.939) .634 40.2% 4.125 (2.840-5.990) 3.247 (1.936-5.443) .000 Lane 5 56.4% 22.0% 0.878 (0.680-1.134) 0.871 (0.641-1.182) .374 21.5% 1.352 (1.027-1.779) 1.439 (1.039-1.994) .028 Lane 6 b 57.9% 25.8% 16.3% In-game role of the server Setter 52.5% 27.0% 1.383 (1.032-1.854) 1,457 (1.075-1.975) .015 20.5% 1.211 (0.883-1.662) 1.240 (0.892-1.725) .201 Attacker receiver 52.9% 26.6% 1.352 (1.033-1.769) 1.436 (1.091-1.890) .010 20.5% 1.201 (0.900-1.602) 1.251 (0.930-1.682) .139 Opposite 59.1% 17.8% 0.808 (0.550-1.188) 0.869 (0.584-1.292) .488 23.1% 1.213 (0.841-1.749) 1.335 (0.909-1.960) .141 Middle attacker b 59.0% 22.0% 19.0% Receiver player Forward attacker 57.5% 21.8% 1.631 (1.122-2.373) 1.583 (1.050-2.388) .028 20.7% 2.271 (1.486-3.472) 1.798 (1.133-2.852) .013 Others players 50.0% 27.5% 2.359 (1.669-3.335) 2.267 (1.588-3.237) .000 22.5% 2.836 (1.899-4.235) 2.627 (2.153-4.178) .000 Libero b 71.9% 16.7% 11.4% 8 Serve direction Long Cross court 44.2% 26.2% 1.414 (1.045-1.913) 1.571 (1.021-2.415) .040 29.6% 2.025 (1.493-2.745) 1.333 (0.863-2.058) .194 Mid Cross court 58.4% 23.8% 0.973 (0.769-1.231) 1.076 (0.839-1.380) .563 17.8% 0.923 (0.713-1.196) 1.000 (0.764-1.310) .997 Parallel b 57.2% 23.9% 18.9% ”a” Category of reference for the dependent variable. “b” Category of reference of the independent variables. 1 “c” Numbers in brackets refer to the 95% confidence interval. 2 3 All the variables presented (serve zone, serve type, in-game role of the server, reception 4 zone, receiver player), except for the serve direction, show significant association, and 5 therefore predict the serve efficacy in the adjusted model (Table 3). 6 7 Executing the serve from zone 1 instead of from zone 6, increases the frequency 8 (OR=1.53) of the serve not permitting the construction of a subsequent attack on the 9 opposite team (does not permit attacking/permits attacking ratio). 10 11 Executing the standing serve instead of the jump serve, increases the frequency 12 (OR=1.63) of the serve not permitting the construction of the subsequent attack on the 13 opposite team (does not permit attacking/permits attacking ratio). 14 15 Aiming the serve at the seam or to lane 5 instead of to lane 6, increases the frequency 16 (OR=3.24 and OR=1.43, respectively) of obtaining an ace with the serve (point/permits 17 attacking ratio). 18 19 If the serve is executed by the setter and the attacker receiver, instead of being executed 20 by the middle attacker, this increases the frequency (OR=1.45 and OR=1.43, 21 respectively) of the serve not permitting the construction of the subsequent attack on the 22 opposite team (does not permit attacking/permits attacking ratio). 23 24 If the serve is received by the forward attacker or by another receiver, instead of by the 25 libero, this increases the frequency (OR=1.58 and 2.35, respectively) of the serve not 26 permitting the construction of an attack (does not permit attacking/permits attacking 27 ratio). Furthermore, the serve received by the forward attacker or by another receiver, 28 instead of by the libero, increases the frequency (OR=1.79 and OR=2.62, respectively) 29 of obtaining an ace with the serve (point/permits attacking ratio). 30 31 4. Discussion 32 33 The main objective of this research study was to analyse the prediction level of different 34 serve variables in volleyball, on its efficacy, in formative stages. 35 36 The first hypothesis indicated that the variables: serve zone, serve type, striking 37 technique, in-game role of the server, reception zone, receiver player and serve 38 direction will act as predictors of the serve efficacy. The results of our study do not 39 confirm this hypothesis, as the variables, striking technique and serve direction do not 40 predict its efficacy. 41 42 9 Furthermore, the variables considered in our study have been used in previous research 1 studies, showing a significant association with the serve efficacy, both at formative 2 stages (García-Tormo et al., 2006; Gil et al., 2011b), and in high level volleyball 3 (Afonso et al., 2012; Quiroga et al., 2012). At a high level volleyball, predictive studies 4 indicate that the serve type, receiver player and reception zone variables act as 5 predictors of the serve efficacy (Afonso et al., 2012), whilst in our study in formative 6 stages, there is a greater number of predictive variables of the serve efficacy. These 7 results clearly show that, in formative stages, there is a greater number of variables that 8 affect the serve efficacy than at high level volleyball. It is thus advisable to work with 9 different variables (serve zone, serve type, striking technique, in-game role of the 10 server, reception zone, receiver player) to improve the performance of this technical- 11 tactical action. Whilst, at a high level volleyball, the variables that affect the serve 12 efficacy are more particularize (Quiroga et al., 2012), favouring much more specific 13 training. 14 15 The second hypothesis indicated that standing serves will increase the serve efficacy 16 with respect to jump serves. The results of our study confirm this hypothesis, as it is the 17 standing serve, instead of the jump serve that increases the likelihood of the opposite 18 team not being able to construct the attack. 19 20 In line with our results, García-Tormo et al. (2006) and Gil et al. (2011a) found a 21 significant association between the variables, serve type and efficacy, indicating that the 22 least missed serve is the floater standing type. It is also the serve type that fosters play 23 continuity and is commonly used in formative stages. It is also more frequent in female 24 than in male categories. 25 26 However, research studies performed at high level volleyball differ from our results, as 27 they showed that the jump serve is the most effective serve type (Asterios et al., 2009; 28 Palao et al., 2009; Costa et al., 2012). 29 30 Along this line, in high level female volleyball, the most commonly used type of serve 31 is the jump serve (Costa et al., 2012), highlighting that the greater percentage of aces is 32 served with the power jump serve and with the tennis floater serve, the first of these 33 being the fastest serve and the one with the largest number of errors in execution with 34 respect to other serve modalities (Quiroga et al., 2012). 35 36 This shows differences between the serve type used in elite women’s volleyball, the 37 jump serve, and the serve used in formative stages in female category, that is, the 38 standing serve (Vescovi and Dunning, 1999). These differences may be due to the fact 39 that, in formative stages, the technical mastery of the jump serve is limited due to the 40 anthropometric characteristics (Stamm, 1999) and physical-physiological characteristics 41 (Côté et al., 2003) of female players. These aspects may, in formative stages, slow 42 down the jump serve and make it less aggressive, not obtaining the same efficacy as at a 43 high level volleyball. 44 45 The third hypothesis indicated that serves made to the space between players, will 46 increase the serve efficacy, compared with those made to the mid field area. The results 47