decision-making accuracy in reactive agility

DECISION-MAKING ACCURACY IN REACTIVE AGILITY: QUANTIFYING THE COST OF POOR DECISIONS GREG J. HENRY,1 BRIAN DAWSON,1 BRENDAN S. LAY,1

AND

WARREN B. YOUNG2

1

School of Sport Science, Exercise and Health, University of Western Australia, Crawley, Australia; and 2School of Human Movement and Sport Sciences, University of Ballarat, Victoria, Australia ABSTRACT

Henry, GJ, Dawson, B, Lay, BS, and Young, WB. Decisionmaking accuracy in reactive agility: Quantifying the cost of poor decisions. J Strength Cond Res 27(11): 3190–3196, 2013— Decision-making accuracy and the time cost of incorrect responses was compared between higher- (n = 14) and lowerstandard (n = 14) Australian footballers during reactive agility tasks incorporating feint and nonfeint scenarios. Accuracy was assessed as whether the subject turned in the correct direction to each stimulus. With skill groups pooled, decision accuracy at the first (or only) stimulus (decision time 1) was 94 6 7%, and it decreased to 83 6 20% for the second stimulus (decision time 2; p = 0.01; d = 0.69). However, with skill groups separated, decision accuracy was similar between groups at decision time 1 (higher 95 6 6% vs. lower 92 6 7%; p = 0.6; d = 0.42), somewhat better in the higher-standard group at decision time 2 (88 6 22% vs. 78 6 17%; p = 0.08; d = 0.50). But the decrease in accuracy from decision time 1 to 2 was significant in the lower-standard group only (92 6 7% to 78 6 17%; p = 0.02; d = 1.04). However, with skill groups pooled but agility times examined exclusively in trials involving correct or incorrect decisions, incorrect decisions at decision time 1 during feint trials resulted in a shorter agility time (1.73 6 0.24 seconds vs. 2.03 6 0.39 seconds; p = 0.008; d = 0.92), whereas agility time was significantly longer in feint (incorrect at decision time 2 only; 2.65 6 0.41 seconds vs. 1.97 6 0.36 seconds; p , 0.001; d = 1.76) and nonfeint trials (1.64 6 0.13 seconds vs. 1.51 6 0.10 seconds; p = 0.001; d = 1.13). Therefore, although decision-making errors typically worsen reactive agility performance, successful anticipation of a feint can produce performance improvements. Furthermore, higher-standard footballers are less susceptible to such feints, perhaps because of superior anticipation. Training to improve decision-making accuracy, particularly involving feint move-

Address correspondence to Greg Henry, [email protected]. 27(11)/3190–3196 Journal of Strength and Conditioning Research Ó 2013 National Strength and Conditioning Association

3190

the

ments, may therefore principally benefit lesser-skilled players and should be practiced regularly.

KEY WORDS feint, change of direction, anticipation, Australian football

INTRODUCTION

E

xpert performers are often more able than lesserskilled players to recognize opponent postural and motion information earlier in the execution of a skill (3,4,11,24). In addition, higher-skilled athletes commonly demonstrate greater decision-making accuracy and superior ability to anticipate game events and opponent actions (1,7,12,24). Consequently, in tackle-based sports, superior decision-making speed and accuracy might have important practical benefits by allowing players to more quickly and accurately change direction in response to an opponent’s movements (2,4), thus helping to either apply or avoid tackles. Although higher-skilled players from various sports have demonstrated faster decision-making speed during reactive agility tests (2,4,17,20,21,23) and several studies (2,4,5) have reported the frequency of inaccurate responses, few of these studies have specifically investigated response accuracy, how it may vary between different athletes, and the resulting impact of decision-making errors on agility performance. This is despite the evidence demonstrating that expert performers respond to the movements of their opponents more accurately in other domains (6,7,12). Consequently, it is currently unclear whether, in addition to faster decision-making speed, higher-performance players exhibit superior decision-making accuracy during high-intensity multimovement agility tasks. Previous reactive agility research has also typically incorporated single-turn tasks, but when pursued, players commonly use a feint to deceive opponents into turning in the wrong direction, producing a time advantage for the offensive player (13,15,16). Such feints often involve 2 closely spaced physical or ball movements by the attacking player, requiring a similar response from a defensive player (15,16). Little research has, however, considered how the use of a feint impacts agility performance, and none has examined the impact on decision accuracy in agility tasks. Only 1 study has reported that response accuracy was maintained in higher-skilled Rugby players during feint trials but was decreased significantly in

TM

Journal of Strength and Conditioning Research

Copyright © National Strength and Conditioning Association Unauthorized reproduction of this article is prohibited.

the

TM

Journal of Strength and Conditioning Research lower-skilled players, indicating that lesser-skilled players are less able to discriminate deceptive movements from the legitimate ones (11). However, although these subjects were encouraged to visualize themselves as the defender and to make a physical movement in response to the video, the measured response was a written one, and therefore, perception was not coupled to a sport-specific action and the associated time pressure (11). Such a coupling is important when examining potential expert advantage in sport because differences are more apparent during such ecologically valid tasks (7,12,18). The use of feints are common in sport; therefore, the first aim of this study was to test the hypothesis that such tactics will negatively impact decision-making accuracy in all players because of the increased cognitive and motor challenge. However, it is also anticipated that a larger decrease in decision accuracy will be observed in lesser-skilled players because decision-making accuracy and susceptibility to deceptive movements has previously been observed to be superior in higher-skilled players (5,11). Also, while intuitively appealing to expect that decisionmaking errors will result in a significant time cost, even if movement direction is corrected, this has not been experimentally confirmed. Therefore, there is currently an incomplete understanding of what the full costs of errors (or error corrections) are in an agility context and how this might impact overall agility performance. For example, whether time deficits brought about by decision-making errors during response selection and programming can be mitigated during the motor response, because of superior physical attributes such as speed and power, remains unclear (4,5,8,23). Consequently, the time cost of response errors was also examined to investigate the hypothesis that decision-making errors will lead to significant decreases in agility performance and that these deficits will not be overcome during the motor segment of the reactive agility task. More understanding regarding the effectiveness of feints during agility tasks and whether the accuracy of the response can discriminate between players of different standards should also be gained from this study. In addition, for the first time, identifying factors such as decision-making accuracy that discriminate between more and less successful performers will be quantified to provide some empirical justification for reactive agility and other game simulation training. Better insight into whether decision-making and physical training practices have the potential to improve decision accuracy, enhance anticipation of deceptive movement, or improve the response to a feint may also be provided. This might assist strength and conditioning coaches to provide appropriate training advice to athletes about the physical and cognitive aspects that influence agility performance.

METHODS Experimental Approach to the Problem

This study is the first to examine the effect that a feint during agility tasks has on decision-making accuracy and subsequent

| www.nsca.com

reactive agility performance. To achieve this, a video-based reactive agility test incorporating feint and nonfeint trials was completed by 2 groups (higher/lower standard) of Australian footballers. Initially, to test the hypothesis that the increased cognitive challenge brought about by a feint would decrease decision accuracy, the 2 skill groups were pooled and decision accuracy was compared between the different decision points in feint and nonfeint trials. The 2 skill groups were then separated to examine the relative impact of feints on decision accuracy to test the hypotheses that higher-standard players would not only have superior decision-making accuracy but also better maintain their levels than the lower-standard players after the introduction of deceptive movements. Finally, by quantifying and comparing the dependent variables of agility, decision, and movement times on trials where subjects initially turned in the wrong direction (against trials with correct responses), the impact of poor decision-making was quantified, to measure the relative impact on each of the subcomponents of agility. Subjects

An a priori power analysis (GPower V3.0.1, GPower, Dusseldorf, Germany) revealed that a sample size of 26 resulted in statistical power of 0.80 at an alpha level of 0.50 and an effect size of r = 0.5. Therefore, to allow for dropout, 28 male subjects were recruited and divided according to their level of participation in Australian football. The higher-standard footballers (n = 14, mean 6 SD: age, height, and mass of 23 6 3 years, 182 6 7 cm, and 81 6 9 kg, respectively) were semiprofessional Australian footballers, whereas the lowerstandard players were midgrade amateur players (n = 14, mean 6 SD: age, height, and mass of 21 6 2 years, 178 6 4 cm, and 77 6 8 kg, respectively). The Human Research Ethics Committee of The University of Western Australia (UWA) approved the study design (RA/4/3/1114), and subjects were informed of the risks and subsequently gave informed consent. Procedures

All tests were conducted indoors on a carpeted sprung wooden floor and took place during in the first 2 months of the playing season. All testing occurred in the early evening on a nontraining day, and subjects were requested to refrain from strenuous physical activity on these days. Furthermore, subjects did not consume food or caffeinated drinks in the 2 hours before a test session but were able to drink water freely. Agility trials used a video-based reactive agility test designed around an initial 458 change of direction, described in detail elsewhere (8) (Figure 1). Data were collected using a video camera (DCR-VX2100E; Sony Corporation, Tokyo, Japan) positioned 2 m behind the start line and infrared timing gates (Fitness Technology, Adelaide, Australia) interfaced with a computer loaded with customized “Agility” software (School of Sport Science, Exercise and Health, University of Western Australia, Crawley, Australia). The Agility computer was networked with another containing “Vidplay” software (School of Sport Science, Exercise and Health, University of VOLUME 27 | NUMBER 11 | NOVEMBER 2013 |

3191

Copyright © National Strength and Conditioning Association Unauthorized reproduction of this article is prohibited.

Decision-Making Accuracy in Reactive Agility

Figure 1. Schematic illustration of the video reactive agility test.

Western Australia) that controlled the stimulus video. The Vidplay computer was in turn linked to a roof-mounted projector displaying a life-sized image of a player onto a 4 3 2.5m screen, positioned 16 m from the start line. Previously, this protocol was found to have good test-retest (coefficient of variation of 1.4% and intraclass correlation coefficient of 0.81) and intrarater reliability (coefficient of variation of 5.2% and intraclass correlation coefficient of 0.99) (8). After a standardized warm-up, 5 straight 4-m sprint trials were completed, and the fastest time was used to calculate abort time and to manipulate stimulus/delay times in subsequent agility trials. After an explanation and demonstration of the reactive agility test procedure, subjects completed 5 submaximal practice trials, including 1 turn each to the left and right, another with no turn, and 2 feint trials (1 to each side). After another 5-minute recovery period, the 14 experimental trials commenced, randomly ordered before the assessment, consisting of 3 turns each to the left, right, feint left/turn right, feint right/turn left, and 2 with no turn. The no turn option was included to increase spatial and temporal uncertainty to be more reflective of game demands. The data from the no-turn trials was later discarded, resulting in 12 trials remaining for analysis (6 nonfeint and 6 feint). For agility trials, triggering of the start gate initiated the video stimulus, and at a time equal to each subject’s 4-m

3192

the

sprint time, the projected stimulus person changed direction (or feinted, then turned), to which the subject responded by turning and sprinting through the appropriate exit gate in a simulated attempt to tackle. Additionally, an abort feature ensured a near-maximal sprint during the approach run. If the 3 m per abort gate was not reached before the abort time elapsed, the trial was aborted, occluding the stimulus, and the trial was repeated. The abort time was equal to the 4-m sprint time assessed previously and, along with manipulation of the stimulus video length, assisted in limiting the stimulus presentation point to the 3- to 4-m distance zone for each subject. Subjects were instructed to sprint forward, respond to the stimulus as quickly and accurately as possible, and then sprint to the exit gate. The recorded video was analyzed using siliconCOACH PRO (V6, Dunedin, New Zealand) software, allowing times to be recorded (60.02 seconds) and data yielded from both the video record and timing gates, which provided the following key measures, all reported in seconds: 1. 3-m time—Time from start to the 3-m abort gate. 2. Decision Time 1 (DT1)—The time from the first stimulus presentation (the only turn in nonfeint trials or the first turn in the feint trials) until the response initiation by the test subject, both of which were defined as “the first definitive lateral movement of the foot which initiates the change of direction.” 3. Decision Time 2 (DT2)—The time from the second stimulus (the second/real turn in the feint trials) until the initiation of the second response by the subject. Unlike DT1, this was defined as the ground contact of the foot that initiates the turn, a definition used previously (4). The different definitions for DT1 and DT2 were necessary as the orientation of the subject was such that the preferred reference point of initial lateral movement of the foot was not readily identifiable in the video for the second response. 4. Total time—Total elapsed time from start to finish gates. 5. Agility Time—Total time 2 3-m time. 6. Movement Time—Time from response initiation (the second response in feint trials) until triggering an exit gate.

TM

Journal of Strength and Conditioning Research

Copyright © National Strength and Conditioning Association Unauthorized reproduction of this article is prohibited.

the

TM

Journal of Strength and Conditioning Research

| www.nsca.com

TABLE 1. Mean 6 SD decision accuracy (% correct) at DT1 and DT2 for HS and LS footballers (effect size shown in parentheses).* Difference 6 90% confidence interval (effect size)

DT1 accuracy DT2 accuracy

Group

Accuracy

HS vs. LS

HS DT1 vs. DT2

LS DT1 vs. DT2

HS LS HS LS

95 92 88 78

6 6 6 6

364 (0.42) 10 6 13 (0.5)

7 6 10 (0.44)

14 6 9† (1.04)

6 7 22 17

*DT1 = response to the first stimulus in the feint trials or the only stimulus in the nonfeint trials; DT2 = response to the second stimulus in the feint trials; HS = higher standard; LS = lower standard. †Correct responses at DT2 significantly less than at DT1 in lower standard, p , 0.05.

7. Decision Accuracy—Percentage of total trials where the initial reaction was in the correct direction. Statistical Analyses

Decision accuracy in the trial and skill groups was compared using independent t-tests. Agility, decision, and movement times were compared using independent t-tests between groups wholly inclusive of either incorrect or correct decisions to determine the time/performance cost of wrong turns. Where appropriate, a p value of ,0.05 was used to determine significance. To further interpret the differences between mean values, Cohen’s effect sizes (d) were calculated and interpreted based on the criteria of Hopkins (9), where 0.0 = trivial, 0.2 = small, 0.6 = moderate, 1.2 = large, 2.0 = very large, and 4.0 = nearly perfect. Additionally, the smallest worthwhile change was calculated and used in conjunction with differences in mean values and 90% confidence intervals (CI) to determine the practical importance of the differences. The smallest worthwhile change for agility time was determined to be 0.015 seconds based on a similar length reactive agility

test using a comparable population (21) and, based on another reactive agility test (4), was 0.016 and 0.017 seconds, respectively, for decision and movement times. A spreadsheet was used to compare the changes in the mean values and CIs against the smallest worthwhile change and the likelihood that the change is large enough to have an important practical impact, positively or negatively, and was expressed using the descriptors ,1%, almost certainly not; 1–5%, very unlikely; 6–25%, unlikely; 26–75%, possibly; 76– 95%, likely; 96–99% very likely; .99%, almost certainly (10).

RESULTS Decision Accuracy

With the skill groups pooled, decision accuracy at DT1 (94 6 7%) was higher (d = 0.69; p = 0.01) than DT2 (83 6 20%). Furthermore, with the skill groups separated, decision accuracy was similar between the playing groups at DT1 (p = 0.6; d = 0.42) but somewhat better in the higher-standard group at DT2 (p = 0.08; d = 0.50) (Table 1). Finally, whereas decision accuracy at DT2 was significantly lower than at DT1 for the

TABLE 2. Mean 6 SD agility, decision, and movement times for nonfeint trials wholly inclusive of correct or incorrect decisions at the DT1.*

Nonfeint agility time Nonfeint DT1 Nonfeint movement time

DT1 decision

n

IC C IC C IC C

7 147 8 160 7 160

Time (sec) 1.64 1.51 0.41 0.31 1.19 1.12

6 6 6 6 6 6

Difference in mean 6 90% confidence interval

d

Practical inference†

0.13 6 0.07

1.13

0.10 6 0.06

0.81

0.07 6 0.07

0.50

Almost certainly detrimental (100/0/0) Very likely detrimental (99/1/0) Likely detrimental (88/9/3)

0.13z 0.11 0.15z 0.09 0.16 0.11

*DT1 = response to the first stimulus in the feint trials or the only stimulus in the nonfeint trials; C = correct; IC = incorrect. †With reference to the smallest worthwhile change of 0.015, 0.016, and 0.017 seconds for agility, decision, and movement times,

respectively. Practical inference stated with reference to C vs. IC. zIC trials significantly slower than C, p , 0.05.

VOLUME 27 | NUMBER 11 | NOVEMBER 2013 |

3193

Copyright © National Strength and Conditioning Association Unauthorized reproduction of this article is prohibited.

Decision-Making Accuracy in Reactive Agility

TABLE 3. Mean 6 SD agility, decision, and movement times for feint trials wholly inclusive of correct or incorrect decisions at either DT1 or DT2.* n Decision at DT1 only Feint agility time Feint DT1 Feint movement time Decision at DT2 only Feint agility time Feint DT2 Feint movement time

Time (sec)

IC C IC C IC C

13 127 13 127 13 127

1.73 2.03 0.23 0.27 0.98 1.26

6 6 6 6 6 6

0.24z 0.39 0.12 0.14 0.13z 0.31

IC C IC C IC C

6 134 21 135 6 134

2.65 1.97 0.44 0.24 1.66 1.21

6 6 6 6 6 6

0.41§ 0.36 0.13§ 0.12 0.46§ 0.29

Difference in mean (%) 6 90% confidence interval

d

0.30 (15) 6 0.18

0.92

0.04 (15) 6 0.07

0.3

0.27 (22) 6 0.12

1.14

0.67 (34) 6 0.31

1.76

0.19 (82) 6 0.92

1.56

0.45 (37) 6 0.21

1.18

Practical inference† Almost certainly beneficial (100/0/0) Possibly beneficial (72/19/9) Almost certainly beneficial (100/0/0) Almost certainly detrimental (100/0/0) Almost certainly detrimental (100/0/0) Almost certainly detrimental (100/0/0)

*DT1 = response to the first stimulus in the feint trials or the only stimulus in the nonfeint trials; DT2 = response to the second stimulus in the feint trials; C = correct; IC = incorrect. †With reference to the smallest worthwhile change of 0.015, 0.016, and 0.017 seconds for agility, decision, and movement times, respectively. Practical inference stated with reference to C vs. IC. zIC trials significantly faster, p , 0.05. §IC trials significantly slower, p , 0.001.

lower-standard players (p = 0.02; d = 1.04), it was similar at both time points for the higher-standard group (p = 0.2; d = 0.44). Decision-Making Cost

In the nonfeint trials, where initial responses were in the wrong direction, DT1 (p = 0.004; d = 0.81) and agility times (p = 0.001; d = 1.13) were longer than when players turned in the correct direction, whereas movement time was similar (p = 0.12; d = 0.50) (Table 2). Similarly, in feint trials where incorrect decisions were made at the second decision point (DT2), decision (p , 0.001; d = 1.56), agility (p , 0.001; d = 1.76), and movement times (p , 0.001; d = 1.18) were all longer than if the initial movement direction was correct (Table 3). In contrast, on feint trials where the first of the 2 reactions (FeintDT1) was incorrect, the agility (p = 0.008; d = 0.92) and movement times (p = 0.002; d = 1.14) were shorter (Table 3).

DISCUSSION Initially, this study examined the effects of a feint on decisionmaking accuracy during a reactive agility task incorporating feint and nonfeint trials. Results showed that decision accuracy at the first decision time (Table 1, DT1), which was before the feint in those trials, was higher than observed in some studies using video-based stimuli and real opponents (2,5) but lower than in others (4). This is consistent with the research showing that the nature of the stimulus and the scenario involved may exert some influence over decision accuracy because decision-making ability is strongly linked to previous experience and familiarity with specific situations

3194

the

(12,19). Accordingly, coaches and researchers should consider the nature of the scenario used as the stimulus during both agility training and assessments to ensure that it is related to the specific needs of their sport. However, a key aim of this study was to examine the impact of a feint on decision accuracy, with results revealing that, with the skill groups pooled, accuracy decreased significantly at the second decision point (DT2) of a multimovement task. This finding agrees with our first hypothesis, which asserted that the inclusion of deceptive movements would decrease decision accuracy. This is a novel finding in an agility setting and is likely due to the uncertainty produced by the feint, thereby producing an increased cognitive and motor challenge. Alternatively, the speed, difficulty, and effort involved in fast complex multimovement motor tasks might reduce the energy available for the cognitive component, negatively impacting decision clarity, and in turn decision accuracy (7,8). Together, these findings collectively offer strong support for coaches to encourage the use of this tactic by offensive players, as it should lead to a time/distance advantage for the offensive player. Furthermore, because perceptual and anticipatory skills are trainable (1,6), including those using deceptive movements (18), these findings also support the need to incorporate sports-specific feints into reactive agility training to potentially enhance response speed and reduce error rate by defensive players by exposing them to scenarios they are likely to encounter during match play. Another aim of this study was to investigate the relative impact of deceptive movements on higher- and lower-standard

TM

Journal of Strength and Conditioning Research

Copyright © National Strength and Conditioning Association Unauthorized reproduction of this article is prohibited.

the

TM

Journal of Strength and Conditioning Research Australian footballers. Previously, higher-standard athletes were reported to exhibit greater decision-making accuracy than lower-standard athletes when viewing or performing tasks including (11) and not including deceptive actions (1,7,12,24). However, this was not observed in the current study where decision-making accuracy during at each decision point, both feint and nonfeint trials, although generally higher in the higher-standard players, was not significantly different between the playing groups, with only small associated effect sizes (Table 1). Although, there was a significant decrease in accuracy from decision time 1 (before the feint) to decision time 2 (after the feint) in the lower-standard players, yet there was little change in the decision accuracy in the higher-standard players with the inclusion of the feint. The observation that decision-making accuracy in lesserskilled athletes was more adversely impacted after the introduction of a feint is consistent with previous findings (11) using video occlusion coupled with a written response. Moreover, these findings also support our second hypothesis, which contended that any skill group differences would increase with the inclusion of a feint because of greater decision-making skill in higher-standard players (1,7,12,24). Accordingly, deceptive agility movements appear to have only minor impact on decision accuracy in higher-skilled footballers, whereas lowerskilled players appear less able to effectively recognize and respond to feints, leading to a significant reduction in decision-making accuracy. Accordingly, the small differences between the groups at the first decision time indicates the players in this study seem equally skilled in recognizing cues and predicting actions when the task was a simple single-turn activity. This may possibly be because these single stimuli were not sufficiently difficult to challenge more than the universal perceptual and cognitive skills common to many team-sport athletes (11). In contrast, coupling deceptive movements and multiple turns might increase the perceptual, cognitive, and motor challenge to the point where lesser-skilled players are less able to effectively distinguish and interpret the available cues and maintain decision accuracy, leading to larger decreases in decision accuracy. Furthermore, the ability to maintain decision accuracy in the higher-standard players is likely a result of superior ability to identify and utilize opponent postural and motion information earlier and more accurately in the execution of a skill (11,24). Additionally, visual search strategies such as gaze fixation frequency and duration and scanning patterns may also play a role, with the lower-standard players not attending to the appropriate cues when the second stimulus was presented (22,24). However, other factors such as superior pattern recall, strategic decision-making ability, or enhanced situational expectations (24) are less likely to have been responsible because the current task was a one-on-one scenario and so the opportunity to draw on these experiences were limited. Finally, the impact of poor decisions and subsequent turning errors were also examined to determine if such errors would increase overall response times. Although

| www.nsca.com

decreases in performance were observed in some examples, the opposite was also seen. For example, as hypothesized, incorrect decisions made during nonfeint trials (nonfeint DT1) and at the second decision time (DT2) in the feint trials did result in significant worsening of agility time (Tables 2 and 3). Moreover, this increase was caused through a lengthening in both decision and movement time, as movement direction was corrected. Accordingly, the longer decision time indicates some delay and confusion in information processing or response selection, whereas the longer movement time implies a delay in response execution (14,15). Furthermore, to illustrate the practical significance of the decrease in performance resulting from incorrect decisions, we determined that the mean velocity during the nonfeint trials was approximately 5.3 m$s21 (agility time in correct decision trials measured over 8 m), which meant that the worsening of agility time from incorrect decisions equated to a distance cost for the defensive player of approximately 0.7 m. Similarly, assuming that the feint response adds a further 2 m in distance covered (1 m for the step in the wrong direction and another for the first corrected step), the mean velocity of 5.1 m$s21 during feint trials (agility time in correct trials measured over 10 m) meant incorrect decisions resulted in a defensive player losing approximately 3 m on their opponent. Subsequently, this highlights the significant advantage that can be gained by an offensive player successfully deceiving an opponent into turning the wrong direction. Such an advantage may allow them to pass a ball more easily or score. Another important observation was that on 15 of 21 occasions when there were errors at DT2 during feint trials, the subjects did not complete the trial but gave up once they realized they were deceived. The practical significance of this, if repeated on a sporting field is clear, it would have resulted in an offensive player running freely and potentially into an attacking position. Furthermore, as illustrated above on those error trials that were corrected, the difficulty in arresting the momentum built up in the wrong direction meant completed movement times were much slower than for correct trials, resulting in significant losses of distance on their opponents. Consequently, deceptive movements can have several significant negative effects on agility performance. First, decision accuracy decreases; second, recognizing those mistakes might lead to players giving up a chase; and finally, even where movement direction is corrected, there is a greatly increased time and distance deficit to overcome. Therefore, the use of feints in agility and other game actions by offensive players should be strongly encouraged, and defensively, athletes should also regularly practice against these, to reduce error rates and improve response speed. In contrast to nonfeint DT1 and DT2 in the feint trials, incorrect decisions at the first decision point in feint trials (feint DT1, Table 3) inadvertently lead to significant improvements in agility time. In those instances, responding in the “wrong” direction to the first stimulus ultimately proved to be correct upon presentation of the second VOLUME 27 | NUMBER 11 | NOVEMBER 2013 |

3195

Copyright © National Strength and Conditioning Association Unauthorized reproduction of this article is prohibited.

Decision-Making Accuracy in Reactive Agility stimulus, negating the need for a second turn and leading to faster movement times. Indeed, the faster movements produced in this instance equate to a distance advantage of 1.2 m, but this time to the defensive player, which would greatly improve their chances of making a successful tackle. Conceivably, this “error” at the first turn might be because of successful anticipation of the feint, but because the subjects’ intent is unknown, it is uncertain if they anticipated or guessed incorrectly but “got lucky” and found themselves moving in the correct direction once the second turn was initiated. Nonetheless, these results provide an insight into the potential practical benefit of effective anticipation, which if successful, may shift any advantage of evasive movements, and feints to the pursuing player. In conclusion, the introduction of a feint into a reactive agility task reduces decision accuracy that, in most cases, leads to substantial reductions in overall performance for the pursuing player. However, successful anticipation of a feint may improve performance and shift the advantage to the pursuing player. Finally, the use of a feint also results in a significant decrease in decision-making quality in lowerstandard Australian football players but does not affect higherstandard players. Accordingly, these results provide support for the potential for all players, but especially lesser-skilled players, to improve performance by incorporating sport-specific agility training drills that include feints from both an offensive and defensive point of view. Such training could improve the ability to anticipate, recognize, and respond to simple and more complex agility movements and other deceptive actions.

PRACTICAL APPLICATIONS Where incorrect responses to single and multiturn agility movements results in significant time and distance being lost by defensive players when pursuing an offensive player, traininginduced improvements in decision-making speed and accuracy can enhance the ability to avoid or make tackles that might influence a game. Therefore, in addition to physical preparation, coaches should devote some training time to improving game and situational awareness, and decision-making quality and speed. This should incorporate a variety of reactive agility and other game simulation tasks, which include numerous feint scenarios to expose players to this tactic and reduce its influence on decision accuracy and agility performance (6,18). To achieve this, coaches can use video-based perceptual training and also on-field reactive agility training using game-specific stimuli such as real opponents using context-specific movements such as one-on-one pursuit and/or evasion drills. Additionally, coaches should modulate the cognitive challenge over time by moving from relatively simple one-on-one activities to more complex multiplayer scenarios or small-sided games depending on the needs of the athlete and the sport.

1. Abernethy, B, Wood, J, and Parks, S. Can the anticipatory skills of experts be learned by novices. Res Q Exerc Sport 70: 313–318, 1999. the

3. Farrow, D and Abernethy, B. Can anticipatory skills be learned through implicit video-based perceptual training? J Sports Sci 20: 471–485, 2002. 4. Farrow, D, Young, W, and Bruce, L. The development of a new test for reactive agility for netball: A new methodology. J Sci Med Sport 8: 52–60, 2005. 5. Gabbett, T, Kelly, J, and Sheppard, J. Speed, change of direction speed and reactive agility of rugby league players. J Strength Cond Res 22: 174–181, 2008. 6. Gabbett, T, Rubinoff, M, Thorburn, L, and Farrow, D. Testing and training anticipation skills in softball fielders. Int J Sports Sci Coach 2: 15–24, 2007. 7. Grehaigne, J-F, Godbout, P, and Bouthier, D. The teaching and learning of decision making in sports. Quest 53: 59–76, 2001. 8. Henry, G, Dawson, B, Lay, B, and Young, W. Validity of a reactive agility test for Australian football. Int J Sports Physiol Perform 6: 534–545, 2012. 9. Hopkins, WG. A scale of magnitudes for effect statistics. In: A New View of Statistics, 2006. Available at: http://sportsci.org/resource/stats/ effectmag.html. Accessed May 20, 2009. 10. Hopkins, WG. A spreadsheet for Deriving a confidence interval, Mechanistic Inference and Clinical Inference from a P Value. Sportscience 11: 16–20, 2007. Available at: sportsci.org/2007/wghinf. htm. Accessed June 25, 2009. 11. Jackson, R, Warren, S, and Abernethy, B. Anticipation skill and susceptibility to deceptive movement. Acta Psychol (Amst) 123: 355–371, 2006. 12. Mann, D, Williams, AM, Ward, P, and Janelle, CM. Perceptualcognitive expertise in sport: A meta-analysis. J Sport Exerc Psychol 29: 457–478, 2007. 13. McClymont, D. Beating the hesitation or “stutter” step. In: Science and Football V: The Proceedings of the Fifth World Congress on Science and Football. Reilly T, Cabri J, and Arau´jo D, eds. London, United Kingdom: Routledge, 2005. pp. 399–407. 14. Schmidt, R and Gordon, GB. Errors in motor responding, “rapid” corrections, and false anticipations. J Mot Behav 9: 101–111, 1977. 15. Schmidt, R and Lee, T. Motor Control and Learning: A Behavioural Emphasis. Champaign, IL: Human Kinetics, 2005. 16. Schmidt, R and Wrisberg, C. Motor Learning and Performance. Champaign: Human Kinetics, 2008. 17. Serpell, BG, Ford, M, and Young, WB. The development of a new test of agility for rugby league. J Strength Cond Res 24: 3270–3277, 2010. 18. Serpell, BG, Young, WB, and Ford, M. Are the perceptual and decision-making components of agility trainable? A preliminary investigation. J Strength Cond Res 25: 1240–1248, 2011. 19. Sheppard, J. An evaluation of a new test of reactive agility. In: School of Human Movement and Sports Sciences. Ballarat, CA: University of Ballarat, 2004, p. 134. 20. Sheppard, JM, Barker, M, and Gabbett, T. Training agility in elite rugby players: A case study. J Aust Strength Cond 16: 15–19, 2008. 21. Sheppard, JM, Young, W, Doyle, T, Sheppard, TA, and Newton, RU. An evaluation of a new test of reactive agility and its relationship to sprint speed and change of direction speed. J Sci Med Sport 9: 342–349, 2006. 22. Singer, RN, Williams, MA, Frehlich, SG, Janelle, CM, Radlo, SJ, Barba, DA, and Bouchard, LJ. New frontiers in visual search: An exploratory study in live tennis situations. Res Q Exerc Sport 69: 290–296, 1998. 23. Veale, JP, Pearce, AJ, and Carlson, JS. Reliability and validity of a reactive agility test for Australian football. Int J Sports Physiol Perform 5: 239–248, 2010.

REFERENCES

3196

2. Duvnjak-Zaknich, DM, Dawson, B, Wallman, KE, and Henry, GJ. Effect of caffeine on reactive agility time when fresh and fatigued. Med Sci Sports Exerc 43: 1523–1530, 2011.

24. Williams, AM and Ford, P. Expertise and expert performance in sport. Int Rev Exerc Sport Psychol 1: 4–18, 2008.

TM

Journal of Strength and Conditioning Research

Copyright © National Strength and Conditioning Association Unauthorized reproduction of this article is prohibited.