This article was downloaded by: [New York University] On: 03 June 2015, At: 01:42 Publisher: Routledge Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
Journal of Sports Sciences Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/rjsp20
Discriminating talent-identified junior Australian football players using a video decision-making task a
a
b
a
Carl T. Woods , Annette J. Raynor , Lyndell Bruce & Zane McDonald a
School of Exercise and Health Sciences, Edith Cowan University, Joondalup, Western Australia, Australia b
School of Medical Sciences, Royal Melbourne Institute of Technology University, Melbourne, Australia Published online: 28 May 2015.
Click for updates To cite this article: Carl T. Woods, Annette J. Raynor, Lyndell Bruce & Zane McDonald (2015): Discriminating talentidentified junior Australian football players using a video decision-making task, Journal of Sports Sciences, DOI: 10.1080/02640414.2015.1053512 To link to this article: http://dx.doi.org/10.1080/02640414.2015.1053512
PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions
Journal of Sports Sciences, 2015 http://dx.doi.org/10.1080/02640414.2015.1053512
Discriminating talent-identified junior Australian football players using a video decision-making task
CARL T. WOODS1, ANNETTE J. RAYNOR1, LYNDELL BRUCE2 & ZANE MCDONALD1 1
School of Exercise and Health Sciences, Edith Cowan University, Joondalup, Western Australia, Australia and 2School of Medical Sciences, Royal Melbourne Institute of Technology University, Melbourne, Australia
Downloaded by [New York University] at 01:42 03 June 2015
(Accepted 15 May 2015)
Abstract This study examined if a video decision-making task could discriminate talent-identified junior Australian football players from their non-talent-identified counterparts. Participants were recruited from the 2013 under 18 (U18) West Australian Football League competition and classified into two groups: talent-identified (State U18 Academy representatives; n = 25; 17.8 ± 0.5 years) and non-talent-identified (non-State U18 Academy selection; n = 25; 17.3 ± 0.6 years). Participants completed a video decision-making task consisting of 26 clips sourced from the Australian Football League game-day footage, recording responses on a sheet provided. A score of “1” was given for correct and “0” for incorrect responses, with the participants total score used as the criterion value. One-way analysis of variance tested the main effect of “status” on the task criterion, whilst a bootstrapped receiver operating characteristic (ROC) curve assessed the discriminant ability of the task. An area under the curve (AUC) of 1 (100%) represented perfect discrimination. Between-group differences were evident (P < 0.05) and the ROC curve was maximised with a score of 15.5/26 (60%) (AUC = 89.0%), correctly classifying 92% and 76% of the talent-identified and non-talent-identified participants, respectively. Future research should investigate the mechanisms leading to the superior decision-making observed in the talent-identified group. Keywords: talent identification, expert performance, team sports, discriminant analysis
Introduction Identifying and quantifying a successful performance within multidisciplinary team sports is complex, as it typically stems from an advantageous blend of both skill (motor and perceptual) and physiology (Farrow & Raab, 2008). Nevertheless, Launder (2001, 2013) defined the main elements that contribute to effective play in such sports, referred to as the model of a skilful player. Specifically, this model characterises the intricate elements needed to succeed within such sports as physical, technical and/or tactical. Consequently, the contribution that each element has to an overall successful performance within such sports can be independently analysed (Launder, 2001). For example, within Australian football players, specific physical and technical skills have been shown to discriminate talent-identified juniors and are thus important components to assess when identifying potential success (Keogh, 1999; Robertson, Woods, & Gastin, 2014; Woods, Raynor, Bruce, McDonald, & Collier, 2014; Woods, Raynor, Bruce, & McDonald, 2014). However, to date, it is unknown if talent-identified
junior Australian football players possess a superior contextual decision-making skill despite its suggested importance for success in the game (Berry, Abernethy, & Côté, 2008; Farrow & Raab, 2008; Gréhaigne, Godbout, & Bouthier, 2001). This is of interest with regard to talent identification, as tactically gifted players may compensate for physical and/ or technical inefficiencies with a superior decisionmaking skill (Abbott & Collins, 2004; Farrow & Raab, 2008). Additionally, it is suggested that when junior players are playing at a higher level, their “survival” in the team may rely on their tactical skill rather than their physical and/or technical skills (Farrow & Raab, 2008). Addressing this pertinent gap may lead to a more robust means for identifying talent in junior Australian football players when coupled with more traditional physical and technical measurement tools. The appealing features of sports-specific decisionmaking (such as the spatial and temporal constraints imposed on players) can be the very features that provide the greatest experimental difficulties. These difficulties are commonly avoided through the use of
Correspondence: Carl Woods, School of Exercise and Health Sciences, Edith Cowan University, Joondalup, WA, Australia. E-mail: [email protected] © 2015 Taylor & Francis
Downloaded by [New York University] at 01:42 03 June 2015
2
C. T. Woods et al.
video-based experimental procedures which allow a researcher to control the environmental conditions, as well as ensuring that players view the same scenarios under similar environmental conditions (Berry et al., 2008; Bruce, Farrow, Raynor, & Mann, 2012). Nonetheless, it is crucial that the experimental design underpins the sports-based decision-making performance, as the chosen scenarios must provide a situation enabling the identification of possible outcomes (e.g. option generation) before the decision is selected (Mascarenhas, Collins, & Mortimer, 2005; Raab & Johnson, 2007). Additionally, to explore the expertnear-expert differences when attempting to identify talent, Farrow and Raab (2008) suggest that the scenarios should reflect game structures that are commonly encountered by players in match play (i.e. kicking/passing the ball in from defence or into attack). It has been reported that contextual decision-making is a process whereby expert decision-makers use superior visual search and option-generation strategies when compared to their lesser skilled counterparts (Hepler & Feltz, 2012; Raab & Johnson, 2007; Ward, Ericsson, & Williams, 2013). However, prior to investigating these processes in greater detail and in accordance with the expert performance approach (Ericsson & Smith, 1991; Williams & Ericsson, 2005), we must first establish if a difference in decision-making ability exists between specific athletic cohorts. As such, this study focuses on the product rather than the process leading to the decision made and is consequently underpinned by stage one of the expert performance approach (Ericsson & Smith, 1991; Williams & Ericsson, 2005) where the product or outcome (in this case decision-making skill) is captured and compared between two athletic cohorts. The aim of the study is to examine if contextual decision-making skill is discriminative of talent-identified junior Australian football players. The outcome of this study will therefore contribute to the development of a more robust talent identification programme whereby coupled with additional assessments of physical and technical ability; a more comprehensive means for the identification of talent may be developed in junior Australian football players. Methodology From an initial sample of 85 under 18 (U18) West Australian Football League (WAFL) players within an age of 17.8 ± 0.8 years, two groups, namely talent-identified (n = 25; 17.8 ± 0.5 years) and non-talent-identified (n = 25; 17.3 ± 0.6 years), were selected. The talent-identified sample consisted of 25 participants who had been selected onto the 2013 WAFL State U18 Academy squad
(elite talent development programme), whilst the non-talent-identified sample consisted of 25 participants randomly chosen using the random number generation package in Excel (Microsoft, Inc.) from the remaining cohort of 60 WAFL U18 players not selected onto the Academy squad. Chi-square analysis indicated that there was no relative age effect for either group for quartile distribution (talentidentified, χ23 = 2.66, P > 0.05; non-talent-identified, χ23 = 2.89, P > 0.05) or half year distribution (talent-identified, χ23 = 0.33, P > 0.05; non-talentidentified, χ23 = 1.87, P > 0.05). At the time of recruitment all participants were injury free and partaking in regular training sessions. The relevant Human Research Ethics Committee provided ethical approval with all participants and parents/guardians (if under 18 years of age) providing informed consent prior to testing. Each video clip was sourced from Australian Football League (AFL) game-day footage which was filmed from a camera set-up behind the goals, with this footage being obtained with permission from one AFL club. These clips were from an offensive perspective (i.e. the team was in possession of the ball at the decision-making moment) and were inclusive of ball movements into attack (inside 50 m plays), passages of play from defence (including kick-ins) and passages of play through the midfield. A total of 18 AFL games from the 2012 season were viewed by the chief investigator with clips being initially included in the task if the player in possession of the ball had a minimum of three and maximum of five teammates available to kick or handball the ball to. This minimum and maximum was used in an attempt to control the number of possible decisions participants potentially had to make to help allow a more concise reflection of decision-making skill. This inclusion criterion initially identified 52 possible clips with each clip edited using Adobe Premiere Elements version 9 (Adobe Systems Incorporated, Australia) to provide a lead time of approximately 15 s prior to the critical decision-making moment. These clips were then independently reviewed by three highly experienced state-level coaches; each with a minimum of 10 years’ coaching at senior state and/or AFL level. Moreover, these coaches had not coached either the talent-identified or non-talent-identified participants reported within this study to help limit a potential coaching bias influencing a participant’s decision-making choice. Each coach was independently asked to identify who the player in possession of the ball should pass to. Only clips where all three coaches agreed upon the passing option were included in the final version of the task; with this secondary inclusion criterion resulting in 26 applicable clips.
3
2.15.1 (R Developmental Core Team, Vienna, Austria). Psychometric properties of the video decision-making task The test–retest reliability of the video decision-making task was assessed by having 10 random participants from the talent-identified sample complete the entire task on two separate occasions separated by five days. No feedback was provided to these participants between or post trials in an attempt to control any external factors which may have influenced their decisions. The same test protocols as described above were followed, with two-way mixed intra-class correlation coefficients (ICC) indicating an ICC range of 0.88–0.99; thus, reflecting a high level of test–retest reliability. Results
100
Between-group differences were evident for the main effect of status on the decision-making task (F(1, 48) = 43.009, P = 0.000; d = 1.36) with the mean score for the talent-identified group being 18 ± 1.8 /26 (69%) compared to only 12 ± 2.5 /26 (46%) for the non-talent-identified group. According to the AUC presented in Figure 1, the ROC curve was maximised with a score of 15.5/26. Consequently, 23 of the talent-identified participants (92%) had a score greater than 15.5, whilst only 6 of the non-talent-identified participants (24%) had a score greater than 15.5. Thus, a cut-
15.5 (76.0%, 92.0%)
60
AUC: 89.0%
40
Sensitivity (%)
80
•
20
Participants were tested in groups of no more than five players at a time with the participants seated 4 m from a 140 cm flat screen television and separated from one another by approximately 50 cm. Two invigilators were present at the time of testing to ensure participants’ choices were their own. Prior to starting the main trial, the participants viewed five practice clips to familiarise themselves with the task requirements. Each participant was provided with an A4-sized booklet that consisted of a coloured screenshot of each decision-making trial. These screenshots were separated by a blank page to ensure the participants could not view the next decisionmaking trial prior to witnessing it on screen. Upon presentation of the decision-making trial, the clip was frozen for 3 s and the participants were cued on screen to “turn the page and circle your choice on the screenshot provided”. Consequently, a coloured A4 image of the decision-making scenario was provided on the recording sheet, in which the participants were required to circle their preferred passing option (ranging from three to five possible passing options). Following the 3 s, the screen was occluded and the participants were cued to “turn your page in readiness for the next clip”. A dichotomous scoring system was employed, with “1” being given to correct choices and “0” to incorrect choices. The criterion value for analysis was the individual participant’s score out of 26 trials. The main effect of “status” (two levels: talent-identified, non-talent-identified) was assessed using a one-way analysis of variance for the task criterion value (total score) using SPSS software (Version 19. SPSS Inc., USA), with the Type-I error rate set at P < 0.05. The effect size of “status” was calculated using Cohen’s d statistic with an effect size of d = 0.20 considered small, d = 0.50 moderate and d > 0.80 large (Cohen, 1998). To examine the discriminant ability of the task, a sensitivity analysis was undertaken. Consequently, a receiver operating characteristic (ROC) curve was produced to plot the truepositive rate against the false-positive rate; with an area under the curve (AUC) being calculated. Importantly, an AUC of 1 (100%) is representative of perfect discriminant potential for a binary outcome. The point on the curve in which the area under it is maximised and is thus reflective of the optimal discrimination potential was considered the value at which a “cut-off” might be acceptable for identifying talent. Here, the ROC curve was used to produce such a cut-off indicator by using the total score obtained on the video decision-making task as a binary classifier. This statistical technique for identifying talent has been previously reported elsewhere (Woods, Raynor, Bruce, McDonald, & Collier, 2014). All modelling and visualisation was undertaken using the statistical computing software R version
Score (AU)
0
Downloaded by [New York University] at 01:42 03 June 2015
Discriminating talent in junior AF
100
80
60 40 Specificity (%)
20
0
Figure 1. ROC curve indicating the point at which the greatest area under the curve has occurred through the use of the decisionmaking score to discriminate talent in U18 Australian football players.
4
C. T. Woods et al.
off score of 15.5/26 successfully discriminated 92% of the talent-identified participants (true positives) and 76% of the non-talent-identified participants (true negatives).
Downloaded by [New York University] at 01:42 03 June 2015
Discussion To the author’s knowledge, this is the first study to examine if contextual decision-making skill could be used to discriminate talent in U18 Australian football players. Results indicated that talent-identified U18 participants were more accurate than their nontalent-identified peers when making game-based decisions. Consequently, the vast majority of the talent-identified participants appear to have a greater decision-making skill than the non-talent-identified group; highlighting the importance of objectively assessing decision-making skill when identifying talent in the game. Unlike many studies that have previously been reported in the literature (Berry et al., 2008; Lorains, Ball, & MacMahon, 2013; Raab & Johnson, 2007; Ward et al., 2013), this study design had an intended focus on the product of decision-making rather than the processes underpinning it. To date, identifying talent in junior Australian football players has traditionally been more subjective (i.e. watching players in match play) which can be unreliable and requires a high level of expertise (Burgess, Naughton, & Norton, 2012). Thus, the methodology and results presented within the current study provide a highly practical objective means for identifying players with a high level of decision-making skill. Although the reported sample size does not statistically allow for such an analysis to take place within the current study, it would be interesting to investigate potential differences between positional groups. Nevertheless, anecdotally we noted that the two best decision-makers within the talent-identified cohort were midfield (e.g. nomadic) players, whilst three of the six non-talent-identified participants who were misclassified were also midfield players. Further, the two talent-identified participants who were misclassified were not midfield players. This particular finding is intriguing given that the clips were inclusive of ball movements into attack, passages of play from defence and passages of play through the midfield, thus not being localised to a specific field position. Moreover, there did not appear to be a trend in the decision-making skill between the key position players and the passages of play (e.g. the decision-making accuracy of forwards did not appear to be more accurate than other positions when viewing ball movements into attack). Nevertheless, given the apparent trend noted in the midfield type players, certain field positions
may require or develop a greater holistic decisionmaking skill than others. Consequently, analysis on a positional specific level may hold important considerations for team selection strategies. For example, it could determine the position that is most suited to the players acquired set of skills. Further, it would be worthwhile for future positional analyses to incorporate a task that consists of both offensive clips (i.e. the team is in possession of the ball at the decisionmaking moment) and defensive clips (i.e. the team without possession of the ball at the decision-making moment) to identify if certain field positions lend themselves to a certain type of decision-maker. The superior game-based decision-making skill noted by the talent-identified participants may stem from a variety of reasons and in keeping with the second stage of the expert performance approach it would be of interest to further examine the now pertinent question of why this difference between the groups exists. Interestingly, of the 25 talent-identified participants, 16 were a part of the WAFL State Academy squad at an U16 level and had thus been a part of the State Academy for at least 2 years, whereas none of the non-talent-identified sample represented the WAFL State Academy at an U16 level. Each of these 16 talent-identified players was successfully classified according to the sensitivity analysis undertaken. Consequently, the prolonged superior training environment that the majority of the talent-identified group were exposed to may have led to greater developmental opportunity. Specifically, the training drills employed by the coaches within the State Academy may have been more environmentally open to allow for a greater physiological and cognitive loading (Farrow, Pyne, & Gabbett, 2008), thus enhancing their decision-making skill in a range of contexts. Additionally, such exposure may have also led to superior option-generation strategy and eye movement, allowing the talent-identified participants to filter irrelevant visual cues and isolate the correct choice more consistently (Raab & Johnson, 2007; Ward et al., 2013). It would be interesting for future research to investigate the second stage of the expert performance approach and examine the underlying mechanisms that may contribute to the superior decision-making performance noted by the talent-identified players within this study. A better understanding of this decisionmaking process may allow developmental coaches to design and deliver more effective training activities to maximise the development of a player’s decisionmaking potential. Interestingly, no relative age effect was reported, providing some slight discrepancy to the findings of Coutts, Kempton, and Vaeyens (2014). However, this discrepancy may stem from slight
Downloaded by [New York University] at 01:42 03 June 2015
Discriminating talent in junior AF methodological differences, namely the comparisons made between samples. Specifically, Coutts et al. (2014) compared the birthdates of an athletic population to that of the general population, which one could assume would be normally distributed, whilst the current study compared a performance measure within a relatively homogenous sample of athletes. Additionally, the relative age effect has been shown to have a negligible influence on tactical skill qualities (Schorer, Baker, Busch, Wilhelm, & Pabst, 2009) and thus we can assume that the relative age did not contribute to the superior game-based decision-making as seen in the talent-identified group and thus may not be a factor that acutely influences the development of decision-making in the game. Given the multidisciplinary requirements of Australian football, it was not totally unexpected to note that there were six non-talent-identified participants who possessed the same decision-making skill as the talent-identified group. Specifically, as described by Launder (2001, 2013), a successful performance within multidisciplinary team sports requires a range of physical, technical and tactical elements, and consequently a superior performance within one of these elements (i.e. tactical) does not always reflect overall skilfulness or potential for success. This alludes to a likely compensation phenomenon, in that a player’s performance in a specific skill is either advantageously or disadvantageously accounted for by their performance in another (Bartmus, Neumann, & de Marées, 1987; Tranckle & Cushion, 2006). Moreover, it is possible that the superior decision-making skill shown by the six nontalent-identified participants was outweighed by lesser physical and/or technical skills that therefore contributed to their non-selection (Woods, Raynor, Bruce, McDonald, & Collier, 2014; Woods, Raynor, Bruce, & McDonald, 2014). Conclusion Our study has indicated that talent-identified U18 Australian football players are more accurate at making correct game-based decisions when compared to their non-talent-identified peers on a video decisionmaking task. Specifically, we were able to correctly discriminate 92% and 76% of the talent-identified and non-talent-identified participants, respectively, through the use of a video decision-making task. This study provides coaches and talent scouts alike with a robust objective means of measuring decisionmaking skill in the game; an element which may traditionally have been subjectively assessed when attempting to identify talent. Nevertheless, it would be interesting for future research to examine decision-making from a positional-specific perspective in an attempt to ascertain if certain players may be
5
better suited to specific field positions or if playing specific positions does in fact enhance a player’s decision-making potential. Additionally, given the range of elements required to succeed within the game, such analysis should be undertaken in conjunction with physical and technical measurement tools in an attempt to optimise the potential identification of talent. Acknowledgements The authors would like to thank the West Australian Football Commission for their support and assistance with data collection. No financial support was required or provided for this study.
Disclosure statement No potential conflict of interest was reported by the authors. References Abbott, A., & Collins, D. (2004). Eliminating the dichotomy between theory and practice in talent identification and development: Considering the role of psychology. Journal of Sports Sciences, 22, 395–408. doi:10.1080/02640410410001675324 Bartmus, U., Neumann, E., & de Marées, H. (1987). The talent problem in sports. International Journal of Sports Medicine, 8, 415–416. Berry, J., Abernethy, B., & Côté, J. (2008). The contribution of structured activity and deliberate play to the development of expert perceptual and decision-making skill. Journal of Sport and Exercise Psychology, 30, 685–705. Bruce, L., Farrow, D., Raynor, A. J., & Mann, D. (2012). But I can’t pass that far! The influence of motor skill on decision making. Psychology of Sport and Exercise, 13, 152–161. doi:10.1016/j.psychsport.2011.10.005 Burgess, D., Naughton, G., & Norton, K. (2012). Quantifying the gap between under 18 and senior AFL football: 2003–2009. International Journal of Sports Physiology and Performance, 7, 53–58. Cohen, J. (1998). Behavioural sciences. Hillsdale, NJ: Lawrence Erlbaum Associates. Coutts, A. J., Kempton, T., & Vaeyens, R. (2014). Relative age effects in Australian Football League national draftees. Journal of Sports Sciences, 32, 623–628. doi:10.1080/02640414. 2013.847277 Ericsson, K. A., & Smith, J. (1991). Prospects and limits of the empirical study of expertise: An introduction. In A. Ericsson & J. Smith (Eds.), Toward a general theory of expertise: Prospects and limits (pp. 1–38). New York, NY: Cambridge University Press. Farrow, D., Pyne, D., & Gabbett, T. J. (2008). Skill and physiological demands of open and closed training drills in Australian football. International Journal of Sports Science and Coaching, 3, 489–499. doi:10.1260/174795408787186512 Farrow, D., & Raab, M. (2008). Recipe to become an expert in decision making. In D. Farrow, J. Baker, & C. MacMahon (Eds.), Developing sport expertise. Researchers and coaches put theory into practice (pp. 137–154). London: Routledge. Gréhaigne, J.-F., Godbout, P., & Bouthier, D. (2001). The teaching and learning of decision making in team sports. Quest, 53, 59–76. doi:10.1080/00336297.2001.10491730
Downloaded by [New York University] at 01:42 03 June 2015
6
C. T. Woods et al.
Hepler, T. J., & Feltz, D. L. (2012). Take the first heuristic, selfefficacy, and decision-making in sport. Journal of Experimental Psychology: Applied, 18, 154–161. Keogh, J. (1999). The use of physical fitness scores and anthropometric data to predict selection in an elite under 18 Australian rules football team. Journal of Science and Medicine in Sport, 2, 125–133. doi:10.1016/S1440-2440(99) 80192-3 Launder, A. (2001). Player practice: The games approach to teaching and coaching. Champaign, IL: Human Kinetics. Launder, A. (2013). Play practice: Engaging and developing skilled players from beginning to elite. Champaign, IL: Human Kinetics. Lorains, M., Ball, K., & MacMahon, C. (2013). Expertise differences in a video decision-making task: Speed influences on performance. Psychology of Sport and Exercise, 14, 293–297. doi:10.1016/j.psychsport.2012.11.004 Mascarenhas, D. R., Collins, D., & Mortimer, P. (2005). Elite refereeing performance: Developing a model for sport science support. Sport Psychologist, 19, 364–381. Raab, M., & Johnson, J. G. (2007). Expertise-based differences in search and option-generation strategies. Journal of Experimental Psychology: Applied, 13, 158–170. Robertson, S., Woods, C. T., & Gastin, P. (2014). Predicting higher selection in elite junior Australian rules football: The influence of physical performance and anthropometric
attributes. Journal of Science and Medicine in Sport, in-press. doi:10.1016/j.jsams.2014.07.019 Schorer, J., Baker, J., Busch, D., Wilhelm, A., & Pabst, J. (2009). Relative age, talent identification and youth skill development: Do relatively younger athletes have superior technical skills? Talent Development and Excellence, 1, 45–56. Tranckle, P., & Cushion, C. J. (2006). Rethinking giftedness and talent in sport. Quest, 58, 265–282. doi:10.1080/ 00336297.2006.10491883 Ward, P., Ericsson, K. A., & Williams, A. M. (2013). Complex perceptual-cognitive expertise in a simulated task environment. Journal of Cognitive Engineering and Decision Making, 7, 231–254. doi:10.1177/1555343412461254 Williams, A. M., & Ericsson, K. A. (2005). Perceptual-cognitive expertise in sport: Some considerations when applying the expert performance approach. Human Movement Science, 24, 283–307. doi:10.1016/j.humov.2005.06.002 Woods, C. T., Raynor, A. J., Bruce, L., & McDonald, Z. (2014). The use of skill tests to predict status in junior Australian football. Journal of Sports Sciences, 33(11), 1132–1140. doi:10.1080/02640414.2014.986501 Woods, C. T., Raynor, A. J., Bruce, L., McDonald, Z., & Collier, N. (2014). Predicting playing status in junior Australian football using physical and anthropometric parameters. Journal of Science and Medicine in Sport, 18(2), 225–229. doi:10.1016/j. jsams.2014.02.006
Journal of Sports Sciences Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/rjsp20
Discriminating talent-identified junior Australian football players using a video decision-making task a
a
b
a
Carl T. Woods , Annette J. Raynor , Lyndell Bruce & Zane McDonald a
School of Exercise and Health Sciences, Edith Cowan University, Joondalup, Western Australia, Australia b
School of Medical Sciences, Royal Melbourne Institute of Technology University, Melbourne, Australia Published online: 28 May 2015.
Click for updates To cite this article: Carl T. Woods, Annette J. Raynor, Lyndell Bruce & Zane McDonald (2015): Discriminating talentidentified junior Australian football players using a video decision-making task, Journal of Sports Sciences, DOI: 10.1080/02640414.2015.1053512 To link to this article: http://dx.doi.org/10.1080/02640414.2015.1053512
PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions
Journal of Sports Sciences, 2015 http://dx.doi.org/10.1080/02640414.2015.1053512
Discriminating talent-identified junior Australian football players using a video decision-making task
CARL T. WOODS1, ANNETTE J. RAYNOR1, LYNDELL BRUCE2 & ZANE MCDONALD1 1
School of Exercise and Health Sciences, Edith Cowan University, Joondalup, Western Australia, Australia and 2School of Medical Sciences, Royal Melbourne Institute of Technology University, Melbourne, Australia
Downloaded by [New York University] at 01:42 03 June 2015
(Accepted 15 May 2015)
Abstract This study examined if a video decision-making task could discriminate talent-identified junior Australian football players from their non-talent-identified counterparts. Participants were recruited from the 2013 under 18 (U18) West Australian Football League competition and classified into two groups: talent-identified (State U18 Academy representatives; n = 25; 17.8 ± 0.5 years) and non-talent-identified (non-State U18 Academy selection; n = 25; 17.3 ± 0.6 years). Participants completed a video decision-making task consisting of 26 clips sourced from the Australian Football League game-day footage, recording responses on a sheet provided. A score of “1” was given for correct and “0” for incorrect responses, with the participants total score used as the criterion value. One-way analysis of variance tested the main effect of “status” on the task criterion, whilst a bootstrapped receiver operating characteristic (ROC) curve assessed the discriminant ability of the task. An area under the curve (AUC) of 1 (100%) represented perfect discrimination. Between-group differences were evident (P < 0.05) and the ROC curve was maximised with a score of 15.5/26 (60%) (AUC = 89.0%), correctly classifying 92% and 76% of the talent-identified and non-talent-identified participants, respectively. Future research should investigate the mechanisms leading to the superior decision-making observed in the talent-identified group. Keywords: talent identification, expert performance, team sports, discriminant analysis
Introduction Identifying and quantifying a successful performance within multidisciplinary team sports is complex, as it typically stems from an advantageous blend of both skill (motor and perceptual) and physiology (Farrow & Raab, 2008). Nevertheless, Launder (2001, 2013) defined the main elements that contribute to effective play in such sports, referred to as the model of a skilful player. Specifically, this model characterises the intricate elements needed to succeed within such sports as physical, technical and/or tactical. Consequently, the contribution that each element has to an overall successful performance within such sports can be independently analysed (Launder, 2001). For example, within Australian football players, specific physical and technical skills have been shown to discriminate talent-identified juniors and are thus important components to assess when identifying potential success (Keogh, 1999; Robertson, Woods, & Gastin, 2014; Woods, Raynor, Bruce, McDonald, & Collier, 2014; Woods, Raynor, Bruce, & McDonald, 2014). However, to date, it is unknown if talent-identified
junior Australian football players possess a superior contextual decision-making skill despite its suggested importance for success in the game (Berry, Abernethy, & Côté, 2008; Farrow & Raab, 2008; Gréhaigne, Godbout, & Bouthier, 2001). This is of interest with regard to talent identification, as tactically gifted players may compensate for physical and/ or technical inefficiencies with a superior decisionmaking skill (Abbott & Collins, 2004; Farrow & Raab, 2008). Additionally, it is suggested that when junior players are playing at a higher level, their “survival” in the team may rely on their tactical skill rather than their physical and/or technical skills (Farrow & Raab, 2008). Addressing this pertinent gap may lead to a more robust means for identifying talent in junior Australian football players when coupled with more traditional physical and technical measurement tools. The appealing features of sports-specific decisionmaking (such as the spatial and temporal constraints imposed on players) can be the very features that provide the greatest experimental difficulties. These difficulties are commonly avoided through the use of
Correspondence: Carl Woods, School of Exercise and Health Sciences, Edith Cowan University, Joondalup, WA, Australia. E-mail: [email protected] © 2015 Taylor & Francis
Downloaded by [New York University] at 01:42 03 June 2015
2
C. T. Woods et al.
video-based experimental procedures which allow a researcher to control the environmental conditions, as well as ensuring that players view the same scenarios under similar environmental conditions (Berry et al., 2008; Bruce, Farrow, Raynor, & Mann, 2012). Nonetheless, it is crucial that the experimental design underpins the sports-based decision-making performance, as the chosen scenarios must provide a situation enabling the identification of possible outcomes (e.g. option generation) before the decision is selected (Mascarenhas, Collins, & Mortimer, 2005; Raab & Johnson, 2007). Additionally, to explore the expertnear-expert differences when attempting to identify talent, Farrow and Raab (2008) suggest that the scenarios should reflect game structures that are commonly encountered by players in match play (i.e. kicking/passing the ball in from defence or into attack). It has been reported that contextual decision-making is a process whereby expert decision-makers use superior visual search and option-generation strategies when compared to their lesser skilled counterparts (Hepler & Feltz, 2012; Raab & Johnson, 2007; Ward, Ericsson, & Williams, 2013). However, prior to investigating these processes in greater detail and in accordance with the expert performance approach (Ericsson & Smith, 1991; Williams & Ericsson, 2005), we must first establish if a difference in decision-making ability exists between specific athletic cohorts. As such, this study focuses on the product rather than the process leading to the decision made and is consequently underpinned by stage one of the expert performance approach (Ericsson & Smith, 1991; Williams & Ericsson, 2005) where the product or outcome (in this case decision-making skill) is captured and compared between two athletic cohorts. The aim of the study is to examine if contextual decision-making skill is discriminative of talent-identified junior Australian football players. The outcome of this study will therefore contribute to the development of a more robust talent identification programme whereby coupled with additional assessments of physical and technical ability; a more comprehensive means for the identification of talent may be developed in junior Australian football players. Methodology From an initial sample of 85 under 18 (U18) West Australian Football League (WAFL) players within an age of 17.8 ± 0.8 years, two groups, namely talent-identified (n = 25; 17.8 ± 0.5 years) and non-talent-identified (n = 25; 17.3 ± 0.6 years), were selected. The talent-identified sample consisted of 25 participants who had been selected onto the 2013 WAFL State U18 Academy squad
(elite talent development programme), whilst the non-talent-identified sample consisted of 25 participants randomly chosen using the random number generation package in Excel (Microsoft, Inc.) from the remaining cohort of 60 WAFL U18 players not selected onto the Academy squad. Chi-square analysis indicated that there was no relative age effect for either group for quartile distribution (talentidentified, χ23 = 2.66, P > 0.05; non-talent-identified, χ23 = 2.89, P > 0.05) or half year distribution (talent-identified, χ23 = 0.33, P > 0.05; non-talentidentified, χ23 = 1.87, P > 0.05). At the time of recruitment all participants were injury free and partaking in regular training sessions. The relevant Human Research Ethics Committee provided ethical approval with all participants and parents/guardians (if under 18 years of age) providing informed consent prior to testing. Each video clip was sourced from Australian Football League (AFL) game-day footage which was filmed from a camera set-up behind the goals, with this footage being obtained with permission from one AFL club. These clips were from an offensive perspective (i.e. the team was in possession of the ball at the decision-making moment) and were inclusive of ball movements into attack (inside 50 m plays), passages of play from defence (including kick-ins) and passages of play through the midfield. A total of 18 AFL games from the 2012 season were viewed by the chief investigator with clips being initially included in the task if the player in possession of the ball had a minimum of three and maximum of five teammates available to kick or handball the ball to. This minimum and maximum was used in an attempt to control the number of possible decisions participants potentially had to make to help allow a more concise reflection of decision-making skill. This inclusion criterion initially identified 52 possible clips with each clip edited using Adobe Premiere Elements version 9 (Adobe Systems Incorporated, Australia) to provide a lead time of approximately 15 s prior to the critical decision-making moment. These clips were then independently reviewed by three highly experienced state-level coaches; each with a minimum of 10 years’ coaching at senior state and/or AFL level. Moreover, these coaches had not coached either the talent-identified or non-talent-identified participants reported within this study to help limit a potential coaching bias influencing a participant’s decision-making choice. Each coach was independently asked to identify who the player in possession of the ball should pass to. Only clips where all three coaches agreed upon the passing option were included in the final version of the task; with this secondary inclusion criterion resulting in 26 applicable clips.
3
2.15.1 (R Developmental Core Team, Vienna, Austria). Psychometric properties of the video decision-making task The test–retest reliability of the video decision-making task was assessed by having 10 random participants from the talent-identified sample complete the entire task on two separate occasions separated by five days. No feedback was provided to these participants between or post trials in an attempt to control any external factors which may have influenced their decisions. The same test protocols as described above were followed, with two-way mixed intra-class correlation coefficients (ICC) indicating an ICC range of 0.88–0.99; thus, reflecting a high level of test–retest reliability. Results
100
Between-group differences were evident for the main effect of status on the decision-making task (F(1, 48) = 43.009, P = 0.000; d = 1.36) with the mean score for the talent-identified group being 18 ± 1.8 /26 (69%) compared to only 12 ± 2.5 /26 (46%) for the non-talent-identified group. According to the AUC presented in Figure 1, the ROC curve was maximised with a score of 15.5/26. Consequently, 23 of the talent-identified participants (92%) had a score greater than 15.5, whilst only 6 of the non-talent-identified participants (24%) had a score greater than 15.5. Thus, a cut-
15.5 (76.0%, 92.0%)
60
AUC: 89.0%
40
Sensitivity (%)
80
•
20
Participants were tested in groups of no more than five players at a time with the participants seated 4 m from a 140 cm flat screen television and separated from one another by approximately 50 cm. Two invigilators were present at the time of testing to ensure participants’ choices were their own. Prior to starting the main trial, the participants viewed five practice clips to familiarise themselves with the task requirements. Each participant was provided with an A4-sized booklet that consisted of a coloured screenshot of each decision-making trial. These screenshots were separated by a blank page to ensure the participants could not view the next decisionmaking trial prior to witnessing it on screen. Upon presentation of the decision-making trial, the clip was frozen for 3 s and the participants were cued on screen to “turn the page and circle your choice on the screenshot provided”. Consequently, a coloured A4 image of the decision-making scenario was provided on the recording sheet, in which the participants were required to circle their preferred passing option (ranging from three to five possible passing options). Following the 3 s, the screen was occluded and the participants were cued to “turn your page in readiness for the next clip”. A dichotomous scoring system was employed, with “1” being given to correct choices and “0” to incorrect choices. The criterion value for analysis was the individual participant’s score out of 26 trials. The main effect of “status” (two levels: talent-identified, non-talent-identified) was assessed using a one-way analysis of variance for the task criterion value (total score) using SPSS software (Version 19. SPSS Inc., USA), with the Type-I error rate set at P < 0.05. The effect size of “status” was calculated using Cohen’s d statistic with an effect size of d = 0.20 considered small, d = 0.50 moderate and d > 0.80 large (Cohen, 1998). To examine the discriminant ability of the task, a sensitivity analysis was undertaken. Consequently, a receiver operating characteristic (ROC) curve was produced to plot the truepositive rate against the false-positive rate; with an area under the curve (AUC) being calculated. Importantly, an AUC of 1 (100%) is representative of perfect discriminant potential for a binary outcome. The point on the curve in which the area under it is maximised and is thus reflective of the optimal discrimination potential was considered the value at which a “cut-off” might be acceptable for identifying talent. Here, the ROC curve was used to produce such a cut-off indicator by using the total score obtained on the video decision-making task as a binary classifier. This statistical technique for identifying talent has been previously reported elsewhere (Woods, Raynor, Bruce, McDonald, & Collier, 2014). All modelling and visualisation was undertaken using the statistical computing software R version
Score (AU)
0
Downloaded by [New York University] at 01:42 03 June 2015
Discriminating talent in junior AF
100
80
60 40 Specificity (%)
20
0
Figure 1. ROC curve indicating the point at which the greatest area under the curve has occurred through the use of the decisionmaking score to discriminate talent in U18 Australian football players.
4
C. T. Woods et al.
off score of 15.5/26 successfully discriminated 92% of the talent-identified participants (true positives) and 76% of the non-talent-identified participants (true negatives).
Downloaded by [New York University] at 01:42 03 June 2015
Discussion To the author’s knowledge, this is the first study to examine if contextual decision-making skill could be used to discriminate talent in U18 Australian football players. Results indicated that talent-identified U18 participants were more accurate than their nontalent-identified peers when making game-based decisions. Consequently, the vast majority of the talent-identified participants appear to have a greater decision-making skill than the non-talent-identified group; highlighting the importance of objectively assessing decision-making skill when identifying talent in the game. Unlike many studies that have previously been reported in the literature (Berry et al., 2008; Lorains, Ball, & MacMahon, 2013; Raab & Johnson, 2007; Ward et al., 2013), this study design had an intended focus on the product of decision-making rather than the processes underpinning it. To date, identifying talent in junior Australian football players has traditionally been more subjective (i.e. watching players in match play) which can be unreliable and requires a high level of expertise (Burgess, Naughton, & Norton, 2012). Thus, the methodology and results presented within the current study provide a highly practical objective means for identifying players with a high level of decision-making skill. Although the reported sample size does not statistically allow for such an analysis to take place within the current study, it would be interesting to investigate potential differences between positional groups. Nevertheless, anecdotally we noted that the two best decision-makers within the talent-identified cohort were midfield (e.g. nomadic) players, whilst three of the six non-talent-identified participants who were misclassified were also midfield players. Further, the two talent-identified participants who were misclassified were not midfield players. This particular finding is intriguing given that the clips were inclusive of ball movements into attack, passages of play from defence and passages of play through the midfield, thus not being localised to a specific field position. Moreover, there did not appear to be a trend in the decision-making skill between the key position players and the passages of play (e.g. the decision-making accuracy of forwards did not appear to be more accurate than other positions when viewing ball movements into attack). Nevertheless, given the apparent trend noted in the midfield type players, certain field positions
may require or develop a greater holistic decisionmaking skill than others. Consequently, analysis on a positional specific level may hold important considerations for team selection strategies. For example, it could determine the position that is most suited to the players acquired set of skills. Further, it would be worthwhile for future positional analyses to incorporate a task that consists of both offensive clips (i.e. the team is in possession of the ball at the decisionmaking moment) and defensive clips (i.e. the team without possession of the ball at the decision-making moment) to identify if certain field positions lend themselves to a certain type of decision-maker. The superior game-based decision-making skill noted by the talent-identified participants may stem from a variety of reasons and in keeping with the second stage of the expert performance approach it would be of interest to further examine the now pertinent question of why this difference between the groups exists. Interestingly, of the 25 talent-identified participants, 16 were a part of the WAFL State Academy squad at an U16 level and had thus been a part of the State Academy for at least 2 years, whereas none of the non-talent-identified sample represented the WAFL State Academy at an U16 level. Each of these 16 talent-identified players was successfully classified according to the sensitivity analysis undertaken. Consequently, the prolonged superior training environment that the majority of the talent-identified group were exposed to may have led to greater developmental opportunity. Specifically, the training drills employed by the coaches within the State Academy may have been more environmentally open to allow for a greater physiological and cognitive loading (Farrow, Pyne, & Gabbett, 2008), thus enhancing their decision-making skill in a range of contexts. Additionally, such exposure may have also led to superior option-generation strategy and eye movement, allowing the talent-identified participants to filter irrelevant visual cues and isolate the correct choice more consistently (Raab & Johnson, 2007; Ward et al., 2013). It would be interesting for future research to investigate the second stage of the expert performance approach and examine the underlying mechanisms that may contribute to the superior decision-making performance noted by the talent-identified players within this study. A better understanding of this decisionmaking process may allow developmental coaches to design and deliver more effective training activities to maximise the development of a player’s decisionmaking potential. Interestingly, no relative age effect was reported, providing some slight discrepancy to the findings of Coutts, Kempton, and Vaeyens (2014). However, this discrepancy may stem from slight
Downloaded by [New York University] at 01:42 03 June 2015
Discriminating talent in junior AF methodological differences, namely the comparisons made between samples. Specifically, Coutts et al. (2014) compared the birthdates of an athletic population to that of the general population, which one could assume would be normally distributed, whilst the current study compared a performance measure within a relatively homogenous sample of athletes. Additionally, the relative age effect has been shown to have a negligible influence on tactical skill qualities (Schorer, Baker, Busch, Wilhelm, & Pabst, 2009) and thus we can assume that the relative age did not contribute to the superior game-based decision-making as seen in the talent-identified group and thus may not be a factor that acutely influences the development of decision-making in the game. Given the multidisciplinary requirements of Australian football, it was not totally unexpected to note that there were six non-talent-identified participants who possessed the same decision-making skill as the talent-identified group. Specifically, as described by Launder (2001, 2013), a successful performance within multidisciplinary team sports requires a range of physical, technical and tactical elements, and consequently a superior performance within one of these elements (i.e. tactical) does not always reflect overall skilfulness or potential for success. This alludes to a likely compensation phenomenon, in that a player’s performance in a specific skill is either advantageously or disadvantageously accounted for by their performance in another (Bartmus, Neumann, & de Marées, 1987; Tranckle & Cushion, 2006). Moreover, it is possible that the superior decision-making skill shown by the six nontalent-identified participants was outweighed by lesser physical and/or technical skills that therefore contributed to their non-selection (Woods, Raynor, Bruce, McDonald, & Collier, 2014; Woods, Raynor, Bruce, & McDonald, 2014). Conclusion Our study has indicated that talent-identified U18 Australian football players are more accurate at making correct game-based decisions when compared to their non-talent-identified peers on a video decisionmaking task. Specifically, we were able to correctly discriminate 92% and 76% of the talent-identified and non-talent-identified participants, respectively, through the use of a video decision-making task. This study provides coaches and talent scouts alike with a robust objective means of measuring decisionmaking skill in the game; an element which may traditionally have been subjectively assessed when attempting to identify talent. Nevertheless, it would be interesting for future research to examine decision-making from a positional-specific perspective in an attempt to ascertain if certain players may be
5
better suited to specific field positions or if playing specific positions does in fact enhance a player’s decision-making potential. Additionally, given the range of elements required to succeed within the game, such analysis should be undertaken in conjunction with physical and technical measurement tools in an attempt to optimise the potential identification of talent. Acknowledgements The authors would like to thank the West Australian Football Commission for their support and assistance with data collection. No financial support was required or provided for this study.
Disclosure statement No potential conflict of interest was reported by the authors. References Abbott, A., & Collins, D. (2004). Eliminating the dichotomy between theory and practice in talent identification and development: Considering the role of psychology. Journal of Sports Sciences, 22, 395–408. doi:10.1080/02640410410001675324 Bartmus, U., Neumann, E., & de Marées, H. (1987). The talent problem in sports. International Journal of Sports Medicine, 8, 415–416. Berry, J., Abernethy, B., & Côté, J. (2008). The contribution of structured activity and deliberate play to the development of expert perceptual and decision-making skill. Journal of Sport and Exercise Psychology, 30, 685–705. Bruce, L., Farrow, D., Raynor, A. J., & Mann, D. (2012). But I can’t pass that far! The influence of motor skill on decision making. Psychology of Sport and Exercise, 13, 152–161. doi:10.1016/j.psychsport.2011.10.005 Burgess, D., Naughton, G., & Norton, K. (2012). Quantifying the gap between under 18 and senior AFL football: 2003–2009. International Journal of Sports Physiology and Performance, 7, 53–58. Cohen, J. (1998). Behavioural sciences. Hillsdale, NJ: Lawrence Erlbaum Associates. Coutts, A. J., Kempton, T., & Vaeyens, R. (2014). Relative age effects in Australian Football League national draftees. Journal of Sports Sciences, 32, 623–628. doi:10.1080/02640414. 2013.847277 Ericsson, K. A., & Smith, J. (1991). Prospects and limits of the empirical study of expertise: An introduction. In A. Ericsson & J. Smith (Eds.), Toward a general theory of expertise: Prospects and limits (pp. 1–38). New York, NY: Cambridge University Press. Farrow, D., Pyne, D., & Gabbett, T. J. (2008). Skill and physiological demands of open and closed training drills in Australian football. International Journal of Sports Science and Coaching, 3, 489–499. doi:10.1260/174795408787186512 Farrow, D., & Raab, M. (2008). Recipe to become an expert in decision making. In D. Farrow, J. Baker, & C. MacMahon (Eds.), Developing sport expertise. Researchers and coaches put theory into practice (pp. 137–154). London: Routledge. Gréhaigne, J.-F., Godbout, P., & Bouthier, D. (2001). The teaching and learning of decision making in team sports. Quest, 53, 59–76. doi:10.1080/00336297.2001.10491730
Downloaded by [New York University] at 01:42 03 June 2015
6
C. T. Woods et al.
Hepler, T. J., & Feltz, D. L. (2012). Take the first heuristic, selfefficacy, and decision-making in sport. Journal of Experimental Psychology: Applied, 18, 154–161. Keogh, J. (1999). The use of physical fitness scores and anthropometric data to predict selection in an elite under 18 Australian rules football team. Journal of Science and Medicine in Sport, 2, 125–133. doi:10.1016/S1440-2440(99) 80192-3 Launder, A. (2001). Player practice: The games approach to teaching and coaching. Champaign, IL: Human Kinetics. Launder, A. (2013). Play practice: Engaging and developing skilled players from beginning to elite. Champaign, IL: Human Kinetics. Lorains, M., Ball, K., & MacMahon, C. (2013). Expertise differences in a video decision-making task: Speed influences on performance. Psychology of Sport and Exercise, 14, 293–297. doi:10.1016/j.psychsport.2012.11.004 Mascarenhas, D. R., Collins, D., & Mortimer, P. (2005). Elite refereeing performance: Developing a model for sport science support. Sport Psychologist, 19, 364–381. Raab, M., & Johnson, J. G. (2007). Expertise-based differences in search and option-generation strategies. Journal of Experimental Psychology: Applied, 13, 158–170. Robertson, S., Woods, C. T., & Gastin, P. (2014). Predicting higher selection in elite junior Australian rules football: The influence of physical performance and anthropometric
attributes. Journal of Science and Medicine in Sport, in-press. doi:10.1016/j.jsams.2014.07.019 Schorer, J., Baker, J., Busch, D., Wilhelm, A., & Pabst, J. (2009). Relative age, talent identification and youth skill development: Do relatively younger athletes have superior technical skills? Talent Development and Excellence, 1, 45–56. Tranckle, P., & Cushion, C. J. (2006). Rethinking giftedness and talent in sport. Quest, 58, 265–282. doi:10.1080/ 00336297.2006.10491883 Ward, P., Ericsson, K. A., & Williams, A. M. (2013). Complex perceptual-cognitive expertise in a simulated task environment. Journal of Cognitive Engineering and Decision Making, 7, 231–254. doi:10.1177/1555343412461254 Williams, A. M., & Ericsson, K. A. (2005). Perceptual-cognitive expertise in sport: Some considerations when applying the expert performance approach. Human Movement Science, 24, 283–307. doi:10.1016/j.humov.2005.06.002 Woods, C. T., Raynor, A. J., Bruce, L., & McDonald, Z. (2014). The use of skill tests to predict status in junior Australian football. Journal of Sports Sciences, 33(11), 1132–1140. doi:10.1080/02640414.2014.986501 Woods, C. T., Raynor, A. J., Bruce, L., McDonald, Z., & Collier, N. (2014). Predicting playing status in junior Australian football using physical and anthropometric parameters. Journal of Science and Medicine in Sport, 18(2), 225–229. doi:10.1016/j. jsams.2014.02.006
