© ACAPS, 2023

1 Introduction

Performance analysis has been used in team sports for a long time. The rationale for using it is to provide unbiased information to achieve a greater understanding of sport performance (O’Donoghue, 2010), to support coaches’ informed decision making during the evolving game (Mouchet & Duffy, 2018), and to inform coaching strategies for future matches.

Here we are focused on team sports performance where the inference of causality of the performance is difficult to establish due to the complexity of interactive behaviors that characterizes it; where many factors can be relevant to the results (Hughes, 2004). Therefore, performance analysis plays an important role in team sports coaches’ learning processes (Camiré, Trudel, & Forneris, 2014) by the association of their coaching experience with objective data. In this study, we are focused on team sports performance where the inference of causality of the performance is difficult to establish due to the complexity of interactive behaviors that characterizes it; where many factors can influence the results (Hughes, 2004).

Coaching has been established as an important field for sport (Gilbert & Trudel, 2004; Jones & Corsby, 2015; Cope, Partington, & Harvey, 2017). Numerous studies have focused on the coach’s in-game decision-making processes and have shown that it can be improved by the use of performance analysis (e.g., Cushion, Ford, & Williams, 2012; Lorains, Ball, & MacMahon, 2013). Information on key performance indicators available to the coach might, therefore, allow more informed decision making and interventions on his/her’s part. This requires the coach to have important competencies for managing the on-going evolution of the game to deal with the complexity, uncertainty and temporal pressure, demanding the ability to “read and react to the field”, to observe and respond to events appropriately (Lara-Bercial et al., 2017) and to make reactive decisions (Mouchet, Morgan, & Thomas, 2018, p. 975).

In many team sports, coaches can provide feedback during breaks or timeouts (Mason, Farrow, & Hattie, 2020), during half-time team talks (Mouchet & Maso, 2018), or throughout the course of the game (Mouchet, Harvey, & Light, 2014). In volleyball, “each team may request a maximum of two time-outs” (FIVB, 2016, p. 36), which requires coaches to provide short and meaningful feedback for their players between rallies.

Focusing on elite volleyball, here we aim to provide a scientific analysis about a key aspect of the volleyball game, the side out attack, to confirm or contradict some coaching experience-based beliefs.

1.1 Volleyball side out

Due to the rule-imposed exchange of ball possession between the teams, the game of volleyball can be characterized as a sequential structure composed by several complexities (Palao, Santos, & Ureña, 2002), defined by the net-crossing of the ball. The side out is the response to the opponent’s serve including serve-reception, setting and attack actions (Costa et al., 2011; Costa, Afonso, Brant, & Mesquita, 2012). In a volleyball rally, most points are won by the receiving team, with around 50% of the plays ending at this point (Degrenne, 2019). In a recent study, Drikos, Barzouka, Nikolaidou, & Sotiropoulos (2021) found that side out attack effectiveness was one of two variables that discriminated winning a volleyball set or tournament, corroborating previous results that establish it as the game action most correlated with team success (Palao, Santos, & Ureña, 2004; Yiannis & Panagiotis, 2005; Zetou, Moustakidis, Tsigilis, & Komninakidou, 2007).

However, we cannot envision the attack action out of context. The setter is the key element of the team’s offensive organization (Graça & Mesquita, 2002), using diverse strategies to place a given attacker in the best position possible to score (see Fig. 1 for an illustration of setter choices), taking space and time into account in his decision making; space as in where on court the attack occurs, and time, as in the set’s tempo, which relates to the ball flight-time, from shorter – first, to longer – third attacking tempos (Selinger & Ackermann-Blount, 1986).

One of the strategies used by the setter is the combination of first and second attacking tempos in different zones (court areas). This strategy aims to put the attacker in numeric equality (1 vs. 1) or superiority (1 vs. 0) against the opponents (Queiroga, Matias, Greco, Graça, & Mesquita, 2005), creating uncertainty in the opponent’s block line (i.e., front-row players).

Laios & Moustakidis (2011) stated that when in side out, a team wins the point in the first attack 54.5% of the time. Given the impact of success in this game context on winning, it is important to know how expert setters deal with unsuccessful side out attacks in their subsequent decision making. Expert decision makers are known as having “a larger knowledge base, and superior anticipation in order to make faster and more accurate decisions” (Baker, Coté, & Abernethy, 2003; Lorains et al., 2013, p. 111). Their decision making behaviour can inform coaches’ future empirically based cues to their team when faced with an unsuccessful side out attack.

Fig. 1 Setter’s choices in side out.

1.2 Setter’s decision-making

Coaches may have different beliefs about playing strategies after a lost attack. Some coaches might persuade the setter to use the same attacker, to avoid confidence issues, while others may want to change the attack strategy, asking the setter to use another attacker. Theoretically, these views could be linked with “hot hand” and “cold hand” paradigms (Gilovich, Vallone, & Tversky, 1985). The “hot hand” paradigm supposes that a player who has succeeded a few times before, is more likely to attain success in the next action. In contrast, the “cold hand” paradigm supposed that a player who has failed a previous action is more likely to fail again. This paradigm has been used to prove the existence of this phenomenon. However, as explained by Raab, Gula, & Gigerenzer (2012), the review of Bar-Eli, Avugos, & Raab (2006) did not find consensus on this matter.

In the present study, our goal is to verify how the setter’s decision-making in the second side out is associated with the offensive organization in the first lost side out, and how this might relate to the second side out success.

There is a lot of information in the literature about the way coaches provide feedback to players (e.g., Bortoli, Bertollo, Messina, Chiariotti, & Robazza, 2010), with a predominant technical orientation on coach interventions (Pereira, Mesquita, & Graça, 2009, 2010), but little is known about their tactical feedback content. Our goal is to provide empirical support to volleyball coach’s decision-making processes in the key game context of side out, based on expert-level setters’ decision making. This will permit coaches to provide more accurate feedback to the setters in real-time after a lost side out. We focus not only on the sequential aspect of the setter’s decision making, but also on the second side out success, i.e., the potential to support, or contradict, the “cold hand” paradigm and coaches’ beliefs.

2 Material and methods

2.1 Sample

We analyzed 44 matches from the 2014 and the 2018 Mens’ World Championships, and the 2016 Olympic Games. The analysis was performed from the perspective of the receiving team (i.e., when this team was in side out). To satisfy the study’s goal, we selected only the side out sequences lost by the receiving team (first side out) and the subsequent rally (second side out), resulting in a sample of 499 game sequences, and 998 rallies.

2.2 Instruments and procedures

An observational design was chosen for this study. Matches were fully recorded from behind the baseline, from a higher ground perspective in the stands without changing the camera position, for posterior analysis and event coding.

Data was collected using the software SportsCode by Hudl (V11, Lincoln, Nebraska, US). In the software a predefined specific code window that allowed the analysis of the different oppositions in a rally was developed (see Fig. 2).

Every match was analyzed with this code window. All the indicators used have been created for a doctoral dissertation (Degrenne, 2019). For this study, we used three main categories: (a) the attack zone; (b) the attack tempo; (c) the side out attack outcome

All matches were analyzed by only one observer. At the end, a database was created and exported to Excel software where each line was a rally. For this study, this database was reshaped to have a new specific database where each line of records corresponded to two sequential rallies played with a given team in side out. The first side out was always lost by the receiving team, i.e., they remained in side out in the subsequent play.

Fig. 2 Code window used for the study.

2.3 Variables

The variables included in this study are described in Table 1.

Considering that they were not related to the study’s purpose, free balls, setter attacks, and service errors were not included in the study’s sample.

First, second and third attack tempo were categorized according to Selinger & Ackermann-Blount’s (1986) proposal. The first tempo is when the attacker jumps before the set and is ready to hit the ball when the setter is touching the ball, the 2nd tempo is when the attacker jumps simultaneously with the set, and the 3rd tempo is when the attacker starts his approach when the ball is close to peak trajectory height.

Fig. 3 Attacking zone in volleyball.

2.4 Reliability

After notation, data was exported to an Excel data sheet. In this sheet, each line corresponded to a game sequence played, and the columns corresponded to the variables notated. We later exported the data to SPSS Statistics 26 package for statistical analysis.

One observer, the first author, performed the analysis of the full sample. He is a national level credited French coach, with 12 years’ experience in coaching. Also, he has a master’s degree in Sport Coaching and a PhD in Sports Sciences. These skills and experiences qualified him as an expert observer in volleyball. A second expert observer was available to perform reliability checks. This observer had identical skills to those described for the first observer and he is an assistant coach in the male third level in France.

For the observation reliability procedures, 554 rallies were analyzed (55% of the sample) and Cohen’s Kappa was applied. Overall, we found good reliability with a coefficient of 0.85 for the Offensive Organization.

2.5 Data analysis

The in situ and real competitive context of data collection led us to the use of nominal-type variables collected by expert volleyball observers. Inferentially, we used Chi-square statistics and assessed effect sizes with Cramer’s V. Cramer’s V was considered weak when values varied between 0.1 and 0.3, moderate when values varied between 0.3 and 0.5, moderate, and strong for values above 0.5 (Cohen, 1988). In the four Chi-square analyses, the assumptions for test use were satisfied, i.e., there were no expected cell Ns of zero, and the maximum of cells with an expected N below five was 20%.

3 Results

We targeted decision-making in a second side out setting, after a first lost side out rally. Initially, we looked at how Zone relationship and Tempo relationship were associated, to get an insight into the behavior in terms of the spatial and time relations. Afterwards, we analyzed how Zone relationship and Tempo relationship related to 2SO. Finally, to extend the analysis of the setter decision making to the cold/hot hand paradigm, we analyzed the association of Replay relationship with 2SO.

3.1 Association between Zone relationship and Tempo relationship

The association between Zone relationship and Tempo relationship was statistically significant, though with a small effect size (χ² = 44.429; p < 0.001; ES = 0.17). The crosstabulation of this association is depicted in Table 2.

For Zone relationship, FF was the most frequent occurrence (N = 178, 35.7%), followed by FB (N = 144, 28.9%), and BF (N = 122, 24.4%), with similar frequencies. By far BB was the least frequent category (N = 55, 11%). In Tempo relationship, Same1 and Same3 were almost vestigial (N = 20, 4%, and N = 40, 8%, respectively). For the other categories, Same2 was the most recurrent one (N = 169, 33.9%), followed closely by Faster (N = 153, 30.7%), and Slower (N = 117, 23.4%).

In relation to setting In front after already having done that in the lost side out (FF), a higher frequency than expected was found for reusing a first tempo attack (Same1), and a lower frequency than expected was found for reusing a second tempo attack (Same2), which contributed to the significance of the association. Changing, from the first side out to the next, the direction of the set relates to the use of a slower attack tempo in the second side out, but with a lower frequency than expected in BF, and a higher frequency than expected in FB. Finally, reusing a backset also contributes to the significance of the association, with a lower frequency than expected when the second side out set was slower, and with a higher frequency than expected when a second attack tempo was used in both side outs.

3.2 Association between Zone relationship and 2SO

The association between Zone relationship and 2SO is depicted in Table 3. This association was not statistically significant (χ² = 10.814; p = 0.094). To be noted, Win was the most frequent 2SO outcome, accounting for more than half of the occurrences (N = 271, 54.3%). In 106 cases (39.1%), Win was achieved through the reuse of a In front attack after the lost side out. Also, FB and BF, i.e., changing the direction of the set accounted for 53.3% of the total sample.

3.3 Association between Tempo relationship and 2SO

The association between Tempo relationship and 2SO was statistically significant, though with a small effect size (χ² = 27.355; p = 0.001; ES = 0.17). The crosstabulation of this association is depicted in Table 4.

As mentioned previously, Win was the most recurrent 2SO (73.4%), and it was achieved, either by using a Faster tempo (N = 102, 37.6%), or by Same2 (N = 97, 35.8%).

Faster tempos had higher Win frequencies than expected and lower Continuity frequencies than expected. On the contrary, Slower tempos had lower Win frequencies and higher Continuity frequencies than expected. There were also higher frequencies than expected of Continuity outcomes on the reuse of a First tempo attack (Same1) in the second side out.

3.4 Association between Replay relationship and 2SO

The association between Replay relationship and 2SO was statistically significant, though with a small effect size (χ² = 30.606; p = 0.006; ES = 0.18). The crosstabulation of this association is depicted in Table 5.

The attacker used in the first side out was only reused in the second side out (Replay) in one fourth of the sample (N = 130, 26%). When opting to Replay an attacker, the tendency was clearly to reuse Z4, which was the option used in more than half of the Replay cases (N = 68, 52.3% of the Replay cases). As for No replay, setter decision-making was for the most part oriented into changing setting direction (N = 262, 71%), given FB (N = 142, 38.5%) and BF (N = 120, 32.5%) were the most recurrent categories. Within No replay, when keeping setting direction from the first to the second side out, FF was most prevalent (N = 94, 25.5%), with BB having a vestigial presence within the No replay sample (N = 13, 3.5%).

Considering the association of Replay relationship with 2SO, replaying Z1 had higher frequencies than expected in the Win outcome. In contrast, replaying Z2 had lower frequencies than expected in the Win outcome, but higher frequencies than expected in the Continuity outcome. When the setter did not replay an attacker, although not as recurrent as FB and BF, in 65 out of 94 occurrences (69.1%) FF resulted in a Win outcome, which was higher than expected. Inversely, the lower frequency than expected of FB resulting in a Win outcome also contributed to the significance of the association. Cumulatively, their tendency was inversed with respect to continuity outcomes, with FF having a lower frequency and FB having a higher frequency of the Continuity outcome than expected.

4 Discussion

Our goal was to verify how the setter’s decision-making in the second side out was associated with the offensive organization in the first lost side out, and how this might relate to the second side out success. We wanted to provide empirical support to volleyball coaches’ decision-making process in side out, based on expert-level setter decision making. We focused not only on the sequential aspect of the setter decision making, but also on the second side out success, i.e., supporting, or contradicting, the “cold hand” paradigm and possibly coaches’ beliefs.

We found that, for the most part, setters maintain the use of the second tempo during the second side out attack (33.9%). This decision had no effects on the outcome. The second most used option was to accelerate the offensive organization (30.7%) which in turn was related with a high value of Win in SO2. These two game sequences represent 73.4% of the Win situations in SO2. These results allow us to argue that fast or accelerating attack tempo sequences is one of the key aspects to enhance offensive efficacy, in accordance with Afonso, Mesquita, Marcelino, & Da Silva (2010).

Zone relationship was not associated with SO2. Strategically, the setter’s goal is to introduce uncertainty in the opponent’s blockers. For every Zone relationship category, the highest frequency was for the Win category of SO2, which is probably an expression of the high level of this game context, and the variability is zone use of the setter’s setting pattern. The higher values of FF within the Win category of SO2 might attest to that, with a larger number of setting possibilities in front of the setter, than in the back (see Fig. 1). Additional evidence in that regard is the FB and BF use, i.e., changing from back-set to the front, and vice-versa, which seems to be a recurrent strategy for the setters, accounting for 53.3% of the total sample, and 49% of the Win category, within SO2.

On the other hand, playing faster than in the first side out is significantly associated with more Win in SO2, reinforcing previous research’s findings in relation of the use of faster attack tempos with success in volleyball side out (Millán-Sánchez, Parra-Royón, Benítez, & Ureña Espa, 2020).

To identify trends in the setter’s decision-making, we considered if the same player was used, or not, in the second side out, and, if so, which player was replayed, and if not, did the setter use the inversion (i.e., changing set direction) or not.

First, our results showed that the setter used the same player in both side out attacks in only a quarter of the sample. This information can be linked with the “cold hand” belief because the setter gave most of the balls to another player when the previous attack failed. This fact could be explained by a decreasing self-confidence of the attacker after a failure. Another more tactical explanation could be the necessity for the setter to put their attackers in an advantageous situation to make the point. For that, he/she must maintain some uncertainty in the opponent’s mind by using different offensive combinations. Nevertheless, when the setter chose to replay the same attacker, our results show that in more than one half of the sample he replayed the zone 4 attacker. This option can be explained by: him being the most adaptable and efficient attacker on the field (Millán-Sánchez, Morante Rábago, & Ureña Espa, 2017); the zone 4 location on court, being the farthest attacking zone in front of the setter, forcing larger displacements on the opponent middle blocker; and/or the association of slower attack tempos with this attack zone (second and third; see Costa et al., 2017) related with the possible constraints on the setting option of the quality of the previous ball contact. However, the most effective decision was to replay zone 1, which could be related to the larger distance between the hitter and the opponent’s block. On the contrary, replaying zone 2 was the least effective decision. This information adds nuances and context to the data provided by Millán-Sánchez et al. (2017), where the opposite hitter was more efficient in zone 2 than in zone 1; possibly not, when he is replayed. These differences could also be due to the different sample, which in Millán-Sánchez et al. (2017) case was analyzed without differentiating between side out attack and counterattack. Nonetheless, the setter decision making seems to be more tactically driven than driven by a potential belief in the cold or hot hand paradigms.

When the setter is not replaying the same attacker, the inversion is by far the most recurrent option with 71% of the No replay sample. Within the No replay sample, the inversion Front-Back (38.5%) is the least efficient decision but it could be conceived by the setter as the safest. These results reinforce those of Millán-Sánchez et al. (2017) where the prevalence of the use of the opposite hitter in attack is questioned in terms of its efficacy, as opposed to outside hitters. Also supporting the potential role of the outside hitter in SO2, the most effective decision in No replay settings was to keep playing in front of the setter. This can be linked with the use of the outside hitter in zone 4 or of the first tempo in zone 3, which can lead to a numeric equality (1 vs. 1) or superiority (1 vs. 0) against the opposing blocking line.

These findings can be extremely useful for future volleyball coaching. The first element is that expert setters are only replaying a player in one quarter of the situations and when they do, they tend to use the outside hitter, i.e., zone 4. This information can allow better blocking organization by reducing the attention given to the previous attacker if he was not in zone 4. The second element is that the reuse of the opposite hitter from zone 1 was related with more Win in SO2. Finally, the third element is that when the setter is not replaying an attacker, he should continue playing in front of him because it is associated with a Win SO2.

When considering these findings, sample size should be taken into consideration. We have selected a sample from three competitions, but after the inclusion process, the sample was reduced from 7581 rallies to 998. Future studies should consider a larger sample.

5 Conclusions

In the present study we showed that setter’s decision making after a lost side out was associated with their previous choices, both in terms of space and time. The Zone relationship was not associated with SO2, but the Tempo relationship was. Accelerating from the first to the second side out lead to more SO2 Win than expected.

Our results can be useful for a coach to provide feedback in real-time to his setter after a lost side out. We discovered that setters were prone to not replay an attacker, but when doing so they tended to reuse zone 4 (i.e., the outside hitter). In this situation the most recurrent setting behavior was the inversion, i.e., changing set direction, even though the most effective decision was to keep playing in front of the setter. These elements can be used between rallies by the coach to give instructions to his setter based on the first side out. For example, if the first side out has been lost after playing in Z1, the setter should replay with this same player.

Match performance analysis can be a sound source of information for grounded coaching decision making. We have shown that expert setter behavior, captured by performance analysis, can inform coaches in the future, whether for athlete development, practice, or match intervention.

This approach is an original contribution to knowledge in the field of coaching in volleyball.

Conflicts of interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Acknowledgement

Thanks to Professor Kevin Morgan, from the School of Sport and Health Sciences at Cardiff Metropolitan University for his proofreading.

References

All Tables

All Figures

	Fig. 1 Setter’s choices in side out.
In the text

	Fig. 2 Code window used for the study.
In the text

	Fig. 3 Attacking zone in volleyball.
In the text