Continental basketball competitions are organized by the FIBA with the same rules. Specific information of teams in each continent is found in Table 1. Archival data of the 471 games in the Championships of female basketball of America, Africa, Asia, and Europe in 2013, 2015, and 2017 were obtained from the open-access official FIBA records (available at https://archive.fiba.com). Players who played less than ten minutes in a single game were excluded from the samples (Zhang et al., 2017), which lead to a sample of 9,208 game observations.

In order to test the validity of data sets, a subsample of 20 games (final score differences equal to or less than 10 points) was randomly selected and observed by two experienced analysts (basketball coaches with more than 5 years of experience in basketball performance analysis). The results were compared with the gathered data on the website and perfect intra-class correlation coefficients (ICC = 1.0) were obtained for free-throws, two-point and three-point (both made and missed), offensive and defensive rebounds, turnovers, steals, blocked shots, and personal fouls. For assists and steals, the results were lower but still very acceptable (ICC = 0.83). There was a formal approval of all procedures from the Local Institution of Research Review Board.

The game-related statistics were transformed to per-minute statistics (original statistics/min × 40 min) according to the amount of time players were on the court (Zhang et al., 2017). Continental players were divided into three groups according to playing positions in Table 2. Based on previous research (Sampaio et al., 2006), a total of 14 variables were selected to quantify the technical performance (Sampaio et al., 2006; Zhang et al., 2017). Definitions of these variables can be found in Table 3.

Generalized mixed linear modeling was then realized with Proc Glimmix in the University Edition of Statistical Analysis System (version SAS Studio 3.6). The variables playing position, game outcome, game type, and team and opponent quality were included in the modeling as the fixed effects. Random effects for player name and team identity were added to account for repeated measurement on the players and teams. Separate Poisson regressions were run for each of the continental championships in the modeling, taking the value of each of the fourteen technical variables as dependent variables.

Playing position, game outcome, and game type were all included as nominal predictor variables in the modeling. Playing positions have three levels (Center, Forward, and Guard), game outcome with two levels (win and loss), and game type with two levels (balanced and unbalanced: point difference above and not above 10 points). The effect of team and opponent quality was estimated by the difference in the log of the team’s ranking in the Championships as a predictor (Yi et al., 2018).

Uncertainty in the true effects of the predictors was evaluated using non-clinical magnitude-based inferences as implemented in the spreadsheet accompanying the package of materials for generalized mixed modeling with SAS Studio (Hopkins, 2016). Estimated magnitudes and their confidence limits were expressed in standardized units and were assessed qualitatively with the following scale: <0.2 trivial, 0.2–0.6 small, 0.6–1.2 moderate, 1.2–2.0 large, >2.0 very large (Hopkins et al., 2009). Standardization was achieved by dividing the estimated effect by the between-player standard deviation, which was derived from the mixed model by adding the variance for the true differences between players to the team-to-team variance within players before taking the square root. Effects were deemed clear if the 90% confidence interval did not include substantial positive and negative values simultaneously. Clear effects were reported with a qualitative likelihood that the true effect was either substantial or trivial (whichever probability was greater) using the following scale: <0.5% most unlikely, 0.5–5% very unlikely, 5–25% unlikely, 25–75% possibly, 75–95% likely, 95–99.5% very likely, and >99.5% most likely (Hopkins et al., 2009).

Centers performed more 2ptA and 2ptM from America, Asia, and Europe than forward. All the regions’ forward had a higher number of 3ptA while this difference in Europe and Asia was more significant (ES: 0.89–0.97, moderately). Except for African centers, those from other three regions recorded a moderately higher number of TR (ES: 0.69–0.86).

Asian and European centers achieved a moderately-to-large higher number of 2ptA, 2ptM, OR, and DR than that of guards (ES = 0.90–1.30). In all continental championships, centers showed a small-to-moderately fewer number of 3ptA, 3ptM, and AST (ES: 0.44–1.09).

Forward from Africa and Asia showed higher values of 2ptA (ES = 0.53, small), 2ptM (ES = 0.49, small), OR (ES = 0.55, small), and DR (ES = 0.53, small) than guards. Guards from Africa and America had a greater number of 3ptM (ES = 0.32, small) compared with those in other two regions. In addition, American guards committed more turnovers than forward (ES = 0.37, small).

European centers performed more 2ptM and 2ptA than forward, which was supported by Ibáñez et al. (2018) and Madarame (2018b). In order to enhance the offensive efficiency, European teams seemed to be more patient and tried to seek the best opportunity to shoot under a slow tempo. In addition, centers often execute more two-point shots because they stay near the basket and have the natural advantage of stature as well as a higher field-goal shooting percentage (Sampaio et al., 2006; George et al., 2009; Courel-Ibáñez et al., 2017). Therefore, coaches should be aware of the importance of 2-point shooting for European centers and set specific defensive strategies such as double team or front deny.

The study found that only African forward showed a trivial difference with centers in the number of 2ptM and 2ptA, which may reveal that they were played in a traditional style which emphasized on the 2-point field goals or middle-range shots. However, this type of traditional basketball did not bring any advantage to African teams because it has been corroborated by Madarame (2018b) that the number of 2ptM in African championships was the lowest among the four regions. In fact, those versatile players who can be efficiently qualified in multiple positions are more favored by coaches and managers because their malleability allows the recruitment processes and they line up more flexibly, especially in this era of “small ball” (Sampaio et al., 2006; Teramoto and Cross, 2010, 2018; Zhang et al., 2019a; Zhai et al., 2020). In this regard, personnel scouts should be cautious to add an African forward on the wish list.

To be self-evident, defensive rebound is the key part from defense to offense and it is also the necessary prerequisite to fast break or assists. In fact, defensive rebounds have become one of the vital indicators for winning in all levels of competitions and teams with a higher competitive level have a better ability to transform defensive rebounds to points (Gómez et al., 2006, 2008; Leicht et al., 2017; Paulauskas et al., 2018). Because of the advantage of playing position and activity area, centers always collect more defensive rebounds than forward (Sampaio et al., 2006; Zhang et al., 2019b). Our studies indicated that this difference was more significant in America, which may imply a whole team strategy that centers secured rebounds with fast break created by forward and guards. Therefore, opponents’ coaches could take advantage of these characteristics to mention forward to strive for offensive rebounds to get more scoring opportunities.

In addition, centers committed more fouls than forward. Generally, personal fouls were influenced by not only the player’s ability associated with individuals’ perception and judgment to the opponents’ action but also the playing positions (Sindik and Jukić, 2011). It seemed reasonable that centers committed more fouls than forward because they usually stay near the basket and are frequently challenged by opponents through various techniques and tactics. However, this trend could not be applied in American championships because no clear difference showed between these two positions. These findings are likely a consequence of established defensive strategy, which means American teams emphasized on the balance of defense in each position while others focused more on inside defense.

Guards from Europe and Asia secured less offensive rebounds than those of their centers. To some extent, offensive rebounds represented not only athletes’ motivation for competition but also a certain game strategy, which could generate more attacking opportunities (Ibáñez et al., 2018). In the defensive point of view, offensive rebounds represented the weakness of defensive players (Ibáñez et al., 2018). Thus, players should be encouraged to get more offensive rebounds when they play against European and Asian guards.

As a crucial indicator to game outcomes, assists were considered widely to be a measure of a player’s technique and a team’s overall tactical awareness (Gómez et al., 2008; Ibáñez et al., 2008, 2018; Paulauskas et al., 2018). In addition, assists were often performed by guards because they spent the most live playing time in possession of the ball (e.g., passing and ball handling) (Ferioli et al., 2020). The present study pointed out that the differences of assists between centers and guards in America and Europe were clearer than those in Asia and Africa, possibly because in more intense championships (e.g., America and Europe) (Madarame, 2018b), guards play the role of undertaking more responsibility for organizing and dominating the ball on offense (Sampaio et al., 2006). However, it is worth noting that there may be two different types of assists because American guards usually created opportunities by isolation, while European guards often relied on team tactics (Ibáñez et al., 2018). Thus, defensive deployment should be differentiated from one to the other when facing these two types of assists.

Teams or players with an excellent efficiency of 3-point goals could be a huge threat to the defender, forcing them into a dilemma between close-out to a 3-point line or drop for a dribble action (Sampaio et al., 2006; Teramoto and Cross, 2010, 2018; Zhang et al., 2019a). Generally, guards showed a better performance of 3-point shooting than forward (Sampaio et al., 2006; Sindik and Jukić, 2011). Our results indicated that compared with guards from Asia and Europe, those from America and Africa recorded a higher number of 3ptA and 3ptM than their forward. It may be attributed to the faster game rhythm where guards play the main role of controlling the game pace and dominating long-distance shooting (Ibáñez et al., 2018; Madarame, 2018b). Therefore, when playing against American or African teams, defensive players should place more emphasis on guards’ 3-point shots.

It has been proven that turnovers have an important effect on the game results, especially in women’s competitions (Leicht et al., 2017). As expected, guards had more turnovers than forward in all regions, which was supported by the study of Sampaio et al. (2006) and Vázquez-Guerrero et al. (2020) who stated that guards were more vulnerable to commit turnovers because they usually stay near the perimeter with higher pressure and performed the highest number of high-intensity accelerations and decelerations, especially in the game with a faster tempo (Ibáñez et al., 2018; Madarame, 2018a,b). Notably, American guards, compared with those in other three regions, recorded more turnovers than forward. This result is likely a consequence of the existing characteristics of game pace of American competitions and may also be attributed to the tactical strategies or technical habits that American guards spent more time in control of a live ball by holding or dribbling it (Madarame, 2018b; Ferioli et al., 2020). Consequently, American coaches should consider increasing the proportion of forward’ possession of the ball to relieve defensive pressure on guards. The main limitation of this study is that the positions of the players are determined through the team roster on the official website, and we hypothesize that players’ positions are fixed through the whole tournament. However, coaches may adjust the line-up to adapt to the demands of the game during playing time, which means that a guard could be arranged to play as a forward. Therefore, future studies are encouraged to identify players’ positions by more precise approaches such as video observation or other analysis software. Moreover, all the game-related statistics were obtained from the box score which only display regular indicators. Further studies may wish to excavate some data in relation to technical or tactical behavior such as shooting types, defensive habits, types of screens, and timing of cuts. Furthermore, all the comparisons were made specifically within each continental competition, which means further research can develop a comparison in a simultaneous tournament (e.g., Basketball World Cup and Olympic Games).