1.1 Aim of Study
The aim of this study is to compare the technical attributes of the same player, playing at both an elite level for their respective club and International representations. The study will include the analysis of two different players playing in the Barclays Premier League and the EURO 2008 Championships. This measure will be based on a subjectively drawn continuum that will analyse a player’s technical performance throughout the game.
It will be investigated whether or not there is a varying level of performance for players playing at club and International level, and if so, where these differences occur.
1.2 Hypothesis
The hypothesis states that based on the unfamiliarity of surroundings at the EURO 2008 Championships, a lower level of technically ability will be observed in international performance.
1.3 Assumptions
It is assumed that all players selected are elite level players, all players are playing to win and that Club and International level competition is of a similarly high standard.
1.4 Limitations
Limitations surrounding the nature of this research can include the word limit and submission date imposed by the school of sport. Injury to key players, conditions and the surrounding environment may have an impact on the performance of the individuals being analysed. As the footage is recorded from satellite television, no control on camera angle or the actual footage being shown is granted to the analyst therefore, it is possible for certain instances to be missed from the analysis.
1.5 De-Limitations
A de-limitation to the study undertaken is that broad generalisations cannot be made, due to the limited amount of data gathered. The design is also specific to an attacking player, therefore can only be applied in certain instances. The design and structure of the technique rating system is not endorsed by a recognised Sports Scientist other than the researcher undertaking the study.
1.6 Definition of Terms
Behaviour: A player’s chosen action when involved in a game for example, a pass.
Technique: The action completed by an individual, against a normative scale.
2. Review of Literature
Sports performance analysis is primarily concerned with investigating relevant aspects of player and/or team performance in competitive sport. These aspects include technical aspects, tactics, patterns of play and work-rate (O’Donoghue, 2005).
2.1 Notational Analysis in Football
Some of the earliest hand notations in football can be contributed to the extensive work done by Reep and Benjamin, who collected data from 3,213 matches between 1953 and 1968. The results were taken from numerous English League and World Cup matches and recorded using systematic observation methods. They were concerned with the frequency of passing sequences, shots at goal and goals scored. They reported that 80 per cent of goals resulted from a sequence of three passes or fewer and that 50 per cent of all goals came from possession gained in the final attacking quarter. The results found by the analysis of Reep and Benjamin consequently led to the pioneering of the British long-ball game, giving strength to the argument for coaches to implement the tactic of ‘direct play’.
Taking these finds further, Bate (1988) explored how chance featured in football, and its relation to the tactics and strategies employed by teams. It was found that 94 per cent of goals scored at all levels of international football, resulted from a sequence of 4 or fewer passes. This added backing to the finds of Reep and Benjamin (1968) with Bate concluding that in order to become successful, teams should:
- Play the ball forward as often as possible.
- Reduce square and backwards passing to a minimum.
- Increase the number of long passes forward and forward runs with the ball.
- Play the ball into space as often as possible.
Again these finds credit the tactic of ‘direct play’ which appeared prominent in the British game for many years, thus discarding the more continental style of possession football and short passing.
More recently, James et al (2002) cited a need to enhance the applied benefit of notational analysis idiographic assessment of teams; strategies are required in order to establish meaning normative profiles (Hughes et al., 2001). Therefore an investigation was conducted with the aim to assess the strategies of a team in both domestic and European competition over the course of a competitive season. This was achieved with the analysis of a professional British soccer team using a computerised behavioural measurement package with focus toward the frequency and duration of possession in designated areas of the pitch. The results demonstrated a difference in the teams’ tactical characteristics with more play in pre-defensive areas during European competition at the expense of pre-offensive areas compared to domestic matches. Clear strategic differences were observed with attacking play found to occur more frequently down the right-hand side of the pitch in domestic in comparison to Europe. This evidently demonstrates that different strategies are implied at both individual and team levels as a function of the nature of the competition. It can be stated from these results that Individual roles therefore appear to be dictated by playing position, team tactics and game circumstances (James et al., 2002).
2.2 Factors Determining Success in Football
Ali (1988), designed a hand notation system that recorded 13 basic factors that occur throughout a football match including: dribbling, short pass, long pass, goal, shot on target, shot off target, offside, ball intercepted by goalkeeper, header on target, header off target, intercepted short pass, intercepted long pass and position of restarts. The aim of the system was to identify any specific patterns of attack and how successful each pattern was in determining the outcome of the match. Therefore the system solely focused on the sequences that occurred in the attacking half of the field, with the patterns recorded onto a prepared pitch diagram to show the exact location of each sequence. The data was entered into a computer using X and Y coordinates from the pitch diagram in order to compare the relation to pattern and constituent. To determine the influence of each attacking sequence, the final action of each pattern was then analysed from all the sequences collected. From this system Ali found that attacks using width were more likely to result in success as opposed to attacks instigated through the centre of the field. Plays that had large numbers of passes in a sequence increased the likelihood of a score where as the most common outcome from a long pass was offside.
Contradicting these results were the findings of Hughes et al., (1988). They proclaimed that successful teams approached the final 6th of the pitch by playing the ball largely in the more central areas. This was revealed from a system whereby a methodology was constructed using 24 performance variables to differentiate between successful and unsuccessful teams from the 1986 World Cup. The main characteristics of play when in possession of the ball were measured with successful teams being denoted as those of which progressed through to the semi-final stages of the competition where as unsuccessful teams failed to qualify from within the 1st group stages. Along with attacking more centrally, Hughes et al,. (1988) found that successful teams have more touches when in possession of the ball and had more shots on goal from within the penalty area. Unsuccessful teams were found to frequently play the ball to the wide areas of the pitch in their own defensive zones whilst continually opting to dribble with the ball in comparison to the more successful sides.
Harris and Reilly (1988) concerned themselves with the positions taken up by the attackers of a team, in relation to the positioning of the defence, and the overall success of each attacking sequence. This was done by producing an index to demonstrate the ratio of attackers to defenders in each particular instance, whilst simultaneously assessing the space between a defender, and an attacker in possession of the ball. This was then taken further and analysed in correspondence with attacking success, with a successful attack resulting in a goal, intermediate attacks created a non scoring shot at goal and unsuccessful attacks had no shot as an outcome. From the finds Harris and Reilly suggested that successful attacks resulted from a positive creation of space, where an attacking player passes a defender on the field of play; whilst an unsuccessful attack tended to result due to failure of using the space effectively, likely to be due to good defensive organisation.
A study conducted by Luhtanen et al., (2001) featured selected offensive and defensive variables of outfield players and goalkeepers in the tournament of EURO 2000, attempting to examine any correlation between performance results and the teams final tournament ranking. Over 2000 actions per game were collected using a computerised notation system with all the individual’s actions calculated to give means for each of the teams per selected variable. The results found France to be the highest ranked nation in passing, receiving the ball, runs with the ball and tackles, which evidently coincides with them winning the EURO 2000 competition, thus being ranked as the number 1 team. Due to their consistent high ranking throughout several performance variables, Luhtanen et al., (2001) concluded that France was worthy winners of EURO 2000 and that the best team won. However, in the overall ranking of variables, Italy came 13th yet only just lost in the final during extra-time, whilst the Netherlands came first in ball possession, second in the amount of passes and shot counts at the same time as being placed highly in the corresponding successful executions, but lost to Italy in the semi-final. The low interception, duels and tackling count received by the Netherlands can therefore be attributed as a direct result from the nature of the passing and possession game embarked upon by the team, since when a team is in control of the ball for large periods of play, there is no need for interceptions or tackles to be made.
2.3 Evaluation of Individuals Performance
The analysis of individual performance is highly important within football as a team is effectively constructed from a collective of individuals, all of whom will have different characteristics and varying roles to partake. This was identified by Reilly and Holmes (1983) who conducted 2 studies looking into the level of skill distribution in football. It was achieved from the analysis of 6 non professional matches, looking at skill performances as being either successful or unsuccessful. To add to this study, a cross section of skills tests was carried out on a group of 40 adolescent males from a variety of outfield positions. The findings from the two tests indicated that:
- Success rate of each skill is dependant on pitch location
- In defensive areas is where the highest skill success rate occurs
- Midfield players demonstrated more superior test scores to defenders
Reilly concluded that the greater the distance to the opponents’ goal, the more time available on the ball, giving explanation to the high skill success rate in defensive areas as less pressure is applied.
According to a technical analysis of outfield players conducted by Dufour (1993), on the ball playing time can be divided into:
50.6% Intercepting, 22.4% Passing, 18.7% Controlling Ball, 4.5% Tackling, 2.4% Shooting and 1.4% on other activities. This provides some indication into the main aspects that comprise an outfield player’s role; however the lack of specification regarding playing position makes it slightly inaccurate for use as an individual template as for instance, strikers will generally spend more time shooting then defenders, etc.
James et al., (2002) highlighted the importance of individual analysis stating that it produces a much finer grained overall team evaluation. The study of 21 matches played by the same team over a variety of competitions was aimed at identifying the different roles individuals may take across differing circumstances. More defensive play was observed within individual players along with passes that involve less risk of conceding ball possession. This is characteristic of the nature of play within the top European club competitions, where retaining ball possession appears more frequent then in league matches.
2.4 Positional Demands in Football
A high percentage of the studies conducted around the positional demands in football have a tendency to focus on the physiological demands placed upon individuals (e.g. Reilly and Thomas, 1976; O’Donoghue and Parker, 2001). Some initial research conducted into the technical demands of football playing positions by Dunn et al., (2003) reported that the technical demands for each playing position varied accordingly and were dependant upon the zonal area in which the player functioned. Williams et al., (2003) made links between the physical and technical demands placed upon each of the outfield playing positions. Dead ball skills, clearances and headers were found to be most important to defenders, with ball skills and passing skills being most important to midfielders. Although these studies give some insight into the specific demands of each playing position, the data presented only accounts for the frequency of each technical skill measured, with no reference to skill outcome (i.e. success rates).
When creating performance profiles for football, it is imperative that positional differences are taken into consideration. Taylor et al (2004) conducted a study of 22 league and cup matches for a professional football team with the aim of constructing valid positional performance profiles. By using performance indicators, validated by coaches and experts, they were able to produce behavioural and performance profiles for the positions of fullback, centre back, midfield and forward. Their analysis found significant differences in the frequency of behaviours performed between each of the playing positions’, however similarities were apparent in the finds with regards to the outcome of some behaviours. This suggests that while there are obvious differences in the technical demands of each playing position, mere inter-positional profiles are not as functional as intra-positional profiles, when accounting for individual strengths and weaknesses to be analysed together with recognition to specific roles.
Probert (2007) looked in detail into the technical demands required between specific positions of play, with regards to the successful execution of independent behaviours. Positional classes were used to group performers into goalkeepers, defenders, midfielders and strikers, with comparisons being made between successful and unsuccessful teams. Significant differences in the frequency distributions of defenders, midfielders and strikers were evident in the results’, however no significant differences were reported between the frequency distribution, and the technical rating between successful and unsuccessful teams. These finds highlight the importance of player selection with regards to positioning, and the necessity for players to train the right attributes which feature predominantly in their chosen position.
Although football players and coaches’ alike understand the technical components required to play in each position, there still appears to be a distinct lack of literature available to reinforce their knowledge, which is not just based upon opinion. Wiemeyer (2003) undertook an interview based investigation consisting of 14 coaches from varying participation levels in an attempt to establish technical positional demands within football. From the results gathered, a profile of positional technical requirements can be formed:
Table 1. Technical requirements of positions (Wiemeyer, 2003)
By combining these technical requirements along with a behavioural or action rating system a specific performance profile for a striker can be constructed, providing an overall player rating for any given performance.
2.5 Use of Match Analysis by Coaches
Match analysis is used to give immediate feedback to a player and also provides the coach with greater detail on a performance. Hughes (1996) identifies the four main purposes of match analysis as:
- Analysis of Movement
- Technical/Mechanical Evaluation
- Tactical Evaluation
- Statistical Compilation
The increased demand for elite athletes to continually perform at the highest level means appropriate and accurate feedback must be provided following a performance. Coaches use feedback to help develop and improve the skills of their performers. Franks and Goodman (1986) proposed that coaching relies heavily upon analysis in order to effect improvement in a performance.
Due to the pressure of football management, notational analysis is being increasingly utilized in an attempt to bring success (Partridge and Franks, 1997). Many coaches believe information derived from advances in technological research to be invaluable (Liebermann et al., 2002). A coaches’ interpretation of events that occur throughout a match may not always be accurate, due to the bias they feel towards their own team. Therefore team strategies should be based on something more substantial than opinion (Bate, 1988).
With the advances in computer-based match analysis systems, quantitative performance data can be integrated with video images (O’Donoghue et al., 1995). The systems can easily be tailored for soccer analysis using performance indicators of relevant interest to the coach. Along with the statistical data provided to the performer, the captured footage can be screened several times over and slowed down. Any mistakes can then be rectified by the analyst (Carling et al., 2005). According to Franks and Goodman (1986), an objective quantification of important events that occur during a game is critical for a complete post match analysis. The analysis obtained should be used by the coach to facilitate motor skill acquisition and improve performance.
2.6 Relation of Literature to Study
Despite the vast amounts of literature available on soccer within the field of notational analysis, there will always be openings for further research to be conducted on account of the continual advancements within sports science. Bate (1996) stated that “if an accurate analysis of the technical attributes of each player position was able to be established, then the results could significantly influence team selection and coaching sessions” (cited in Probert, 2007).
The need for further research in this particular field is noted by Grehaigne et al., (2001) suggesting for a need to move beyond merely describing behaviours, and progress towards the actual ability to predict performance. Predictive models can provide some indication of what can be expected in future performances (Potter and Hughes, 2001), therefore being more beneficial to both the coach and performer.
Franks and McGarry (1996) views that the ability to provide information on an individual’s technical performance using scientifically approved predictive models, as opposed to opinion based comments, can significantly modify playing behaviour and promote successful performance.
2.7 Aims of Study
This research aims to deliver a behavioural rating system for a striker or offensive midfielder that allows for clear comparisons in performance between League and International level matches. It will aim to investigate whether or not there is a varying level of performance for players playing at club and International level, and if so, where these differences occur.
3. Methodology
3.1 Introduction
This study was conducted as a post-event hand notation system, looking at individual player analysis within Premier League and Euro 2008 football matches. The system was established using a performance profile designed to evaluate a players’ contribution at Club level, in comparison to International level football.
3.2 Equipment
3.21 Hardware
All Premier League matches analysed were from the 07/08 League season and were recorded from a combination of Sky Sports and Setanta Sports. The matches were recorded with a Samsung DVD SR270M Digital DVD Recorder and burnt onto blank DVD-R discs. For the analysis to be carried out the discs were inserted into an ACER Extensa 5510 Laptop for viewing. The data was recorded onto the specifically designed, A4 paper, data collection sheet, using a pen.
3.22 Software
All games were viewed via windows media player, allowing for easy pause and rewind functions. The International fixtures were taken from:
Whereby for a small fee, full reruns of any selected fixture from the EURO 2008 finals can be purchased, and linked into windows media player for viewing.
3.221 Pilot Study
A pilot study was conducted to test the data collection procedure and to familiarise with the process. All fixtures were analysed by the same analyst, allowing for higher accuracy in the analysis of behaviours and events. The first 15 minutes of an International game between Germany and England was viewed with some immediate problems on the rating of behaviours becoming apparent, for example, it is relatively impossible to rate the successfulness of a block. It is either a successful block, or an unsuccessful block. This prompted several changes into the scoring system employed to correct any problems that may be encountered.
3.222 Final System
The final system design was specific to a striker / attacking player, such as the ones selected for analysis. It is important to understand this, as a defender analysed by the same system would be likely to produce distinctly different results. This specificity is however required, in order to gain a more sound understanding of the technical differences players may adopt in different teams, positions and situations.
3.223 Operational Definitions
The operational definitions can be used for analysis of all outfield players, as the behaviours integrated into the system cover most of the actions conducted by players during games. In order to note down the behaviours performed more efficiently, a key was used to denote a players actions:
P = Pass – Transferring the ball from one player to another.
S = Shot – An attempt to score.
T = Tackle – Challenging an opponent for the ball.
D = Dispossessed – Losing possession to an opponents’ tackle.
I = Interception – Cutting out an opponents attempt to pass.
B = Block – Blocking a shot at goal with any part of the body other than head.
DR = Dribble – Keeping possession of the ball (while moving) against an opponents attempt to retrieve it.
AC = Aerial Challenge – Attempt to play the ball in the air.
CL = Clearance – Playing the ball up-field or into touch to prevent goal scoring opportunities for opponents.
CR = Cross – Horizontal or vertical delivery into opposition area.
RB = Rule Breach – Foul conceded.
The scoring system bases itself on the outcome of the player’s actions, not on the perceived technique in which the skill was performed. In order for this to be a fair test, behaviours were categorised and scored differently;
Pass, Cross, Dribble and Arial Challenge are scored as:
+3 = Successful completion of behaviour resulting in a direct goal assist.
+2 = Successful completion of behaviour resulting in a direct goal attempt.
+1 = Successful completion of behaviour, but no direct goal attempt.
-1 = Possession conceded.
-2 = Opposition have a goal attempt as a direct result of possession conceded.
-3 = Opposition score a goal as a direct result of possession conceded.
Shots at goal are scored as:
+3 = Shot results in a goal.
+2 = Shot forces the keeper into making a save.
+1 = Shot is blocked by an outfield opponent.
-1 = Shot is off-target, but taken from outside the 18 yard box.
-2 = Shot is off-target, from inside the 18 yard box, but outside the 6 yard box.
-3 = Shot is off-target from inside the 6 yard box.
Tackles were scored as:
+1 = Ball possession won as a result from the challenge.
-1 = Ball possession not won as a result from the challenge.
If a player is Dispossessed of the ball or commits a Rule Breach then a score of -1 is given as these behaviours cannot realistically be regarded as successful. If a player makes an Interception, Block or Clearance then a score of +1 is awarded as these behaviours cannot realistically be regarded as unsuccessful.
The pitch was also divided up into distinct areas to allow a more detailed analysis of where certain behaviours occur during a game. Player position also becomes more apparent, giving insight into the positions taken up and where they may be more successful.
Figure 1. Pitch division example
3.224 Reliability Study
The results of the intra observer reliability can be seen within the results section, in tables 3 and 4. This test of intra reliability was achieved by using the Chi-Squared test of independence and the percentage error statistic, within the Microsoft Excel package.
For trials to be reliable, the calculated P value from a Chi squared test are required to be of a value of 0.95 or higher, to indicate that there is agreement between the trials and therefore also reliability at the p>0.95 level of significance (Bland and Altman, 1986).
The percentage error statistic is used to calculate the amount of error for each variable involved in the observation system (Hughes et al., 2002). The calculated values are required to be of 5% or lower in order to fall below the accepted 0.05% level.
3.3 Data Population
Only 3 Premier League players competed in more than approximately 250 minutes of football during the EURO 2008 tournament. This is due to a combination of player injury, players being rested, team selection, and team elimination from the tournament. With one of the players in question being a defender, this left only 2 players of which the analysis could be conducted. All the Premier League Fixtures are from the 2007/08 season and all International matches are from the European Championships 2008 (EURO 2008). All matches in which the 2 players being analysed competed in were placed into a hat, and 3 League and 3 International matches were selected at random:
Player A Player B
Portugal V Turkey Spain V Sweden
Portugal V Czech Republic Spain V Italy
Portugal V Germany Spain V Germany
Manchester United V Fulham Newcastle United V Liverpool
Arsenal V Manchester United Liverpool V Aston Villa
Liverpool V Manchester United Liverpool V Manchester United
3.4 Procedure
For Premier League match analysis, the required DVD was inserted into the Laptop, and the game brought to the screen via Windows Media Player. The Player Being analysed was recorded onto the data collection sheet, with their proposed position noted from the team selection screen. The first 10 minutes of the game were then played with no analysis occurring to confirm that the player was in fact in the correct position. The final 15 minutes were then played to confirm that the player completed over ¾ of the required 90 minutes of the game.
For International match analysis, the same procedure was utilised however, instead of inserting DVD’s into the Laptop, the internet browser was used to bring up the website () from where the desired game can be selected for analysis and played through Windows Media Player in the exact same way as the Premier League fixtures.
Once the player was identified as being in the correct position, analysis of the match could begin from the start. Every time the player in question performed an action, the game was paused and the behaviour recorded along with the following data onto the data collection sheet shown:
Table 2. Example of Data Collection Sheet
3.5 Data Processing
Data were tabulated and then entered into the Microsoft Excel computer package for analysis. The Chi squared test of independence was used to determine whether differences between the results were statistically significant. The Chi-Squared test of independence is a statistical test which provides the significance value of the difference between the observed and expected results. It provides the probability of a Type 1 Error (Thomas and Nelson, 2001). For an identified difference to be significant, the reported P value must be 0.05 or lower (p<0.05) to be accepted at the 95% level of significance (Vincent, 1999).
4.0 Results
4.1 Reliability
The Chi-Squared test of independence and the percentage error statistic were used to calculate the intra observer reliability.
Table 3. Intra-observer action observation reliability between T1 and T2
The Chi square P value of 0.99 indicates strong reliability does occur between T1 and T2 for intra-observer action observation at the 95% level of significance.
The percentage error between the 2 data sets is at 4.7%, which is below the accepted 5% level demonstrating strong reliability.
Table 4. Intra-observer technique rating reliability between T1 and T2
A P value of 0.99 again shows strong reliability between tests T1 and T2 at the 95% level of significance. A low percentage error statistic of 1.6% is again below 5% and at an acceptable level.
4.2 Processed Data
4.21 Player Action Distribution
Figures 1 to 4 respectively show the action distribution for both Player A and Player B across their international and league performances. Figures 5 and 6 show the comparison between league and international performances, for players A and B, across the mean action distribution.
Figure 2. Player A mean action distribution for international performances
Figure 3. Player A mean action distribution for league performances
Figure 4. Player B mean action distribution for international performances
Figure 5. Player B mean action distribution for league performances
Figure 6. Player A mean action distribution comparison
Figure 7. Player B mean action distribution comparison
Figure 6 shows the mean action distribution comparison for Player A. The only significant difference observed is the large increase in shots on goal during international matches. Figure 7 shows the mean action distribution for Player B. Several distinct differences can be observed with more aerial challenges, passes and dispossessions occurring during international matches and more crosses occurring in league games.
- Comparison of the Mean Quality Rating of Techniques for Selected Variables
Figure 8. Player A comparison of the mean quality rating of techniques
Figure 9. Player B comparison of the mean quality rating of techniques
Figures 8 and 9 show the mean quality rating for selected techniques for both Players’ A and B. Figure 8 clearly shows a higher technique rating by Player A for shooting during league matches in comparison to that of international matches. Figure 9 also shows a higher technique rating for Player B in shooting during international matches in comparison to league matches.
4.23 Distribution of Technique Rating Across Selected Performance Variables
Figure 10. Distribution of player A’s Aerial Challenge technique ratings
Figure 11. Distribution of player A’s Crossing technique ratings
Figure 12. Distribution of player A’s Dribbling technique ratings
Figure 13. Distribution of player A’s Passing technique ratings
Figure 14. Distribution of player A’s Shooting technique ratings
Player A shows a trend of +1 rating’s as the highest for all behaviours except shooting where a rating of +2 is the highest. A distinct lack in shooting is apparent in the league games, however a very good success rate for this behaviour is observed.
Figure 15. Distribution of player B’s Aerial Challenge technique ratings
Figure 16. Distribution of player B’s Crossing technique ratings
Figure 17. Distribution of player B’s Dribbling technique ratings
Figure 18. Distribution of player B’s Passing technique ratings
Figure 19. Distribution of player B’s Shooting technique ratings
Player B shows a trend of +1 rating’s as the highest for all behaviours except shooting where +2 and -2 are equally the highest. A significant difference can be observed in +1 rated crosses for Player B when comparing league to international matches.
4.24 Mean Behaviour Rating and Frequency for First and Last 15 Minutes of Matches
Figure 20. Average behaviour rating for first and last 15 minutes for Player A
Figure 21. Average behaviour frequency for first and last 15 minutes for Player A
Figure 20 shows a higher behaviour rating occurs in the last 15 minutes of international matches with no difference observed between the first and last 15 minutes during club matches. Figure 21 shows a higher behaviour frequency for both club and international matches during the first 15 minutes.
Figure 22. Average behaviour rating for first and last 15 minutes for Player B
Figure 23. Average behaviour frequency for first and last 15 minutes for Player B
Figure 22 displays a perceivably low average behaviour rating during club matches in the last 15 minutes. This is accompanied by a low behaviour frequency count displayed in Figure 23.
4.25 Mean Player Rating Across Performance
Figure 24. Mean player rating across performances
Overall player rating across performances shows the overall mean rating for league and international matches. This is done by adding all the player ratings from the analysed games and comparing the averages. This gives an overall and clear view that both performers score slightly higher in international performances compared to their respective league performances.
5.0 Discussion
5.1 Reliability
Hughes and Franks (2004) state that the reliability of a system should be demonstrated clearly, in a way in which is compatible with the intended analysis of the data. According to Atkinson and Nevill (1998) any research conducted within the field of sport should include the use of statistical methods for analysing the reliability of the data obtained. This experiment incorporated the use of the Chi-squared test of independence along with the percentage error statistic to determine that intra observer reliability was achieved. Chi-squared was chosen as it provides a significance value of the difference between the expected and observed result (Thomas and Nelson, 2001) between two or more sets of nominal data. By combining this with the percentage error statistic, it allowed the amount of error for each variable between trials to also be established (Hughes et al., 2002).
5.11 Intra-Observer Reliability
Tables 3 and 4 indicate that the intra-observer reliability between T1 and T2 can be assumed according to the Chi-squared statistic (p≥0.95) as well as the percentage error statistic (p>0.05). The only slight discrepancy occurred in identifying if the player had slightly overrun with the ball when dribbling but managed to keep possession with their team, instead of intentionally passing the ball to a team mate. This only made minor significance as to the results outcome with reliability being assumed by both statistical tests and did not feature prominently throughout the experiment.
5.12 Evaluation of Reliability
The ability to collect data with minimum measurement error is vital during any research (Atkinson and Nevill, 1998). Having data which possess reliable characteristics implies that individual measurements can be taken and used for future research with confidence (Hopkins, 2000). Reliability tests on both behaviour observation and technique rating proved positive allowing reliability to be assumed. Intra-observer reliability was found to be present by both the Chi-squared and percentage error statistic tests.
5.2 Player Action Distribution
Figures 2 to 5 show the exact behaviour distribution for Players A and B across both league and international performances. This is displayed as a percentage of the player’s total on the ball actions and is useful as it allows for the relative distribution of techniques across performances (Probert, 2007). Figures 2 and 3 clearly demonstrate that Passing is the most prominent behaviour utilised by Player A with 40% or greater of their on the ball time spent completing this behaviour, in both international and league performances. The only notable significant differences (p<0.05) appear in Shooting (8% difference) and Aerial Challenges (5% difference) between the action distribution for both sets of data. Although there appears to be a significant difference in the action distribution for shooting (p<0.05), it only accounts for an average of 6% of the total on the ball activity across all performances, which is a relatively small fraction of the Players on the ball time. The findings from figures 2 and 3 is backed up by the analysis conducted by Probert (2007) who found that 5% of a striker’s on the ball time is spent attempting to shoot for goal, something which is vital for both the individual and the team (Smith, 1973; Van Lingen, 1997; Wiemeyer; 2003). This would therefore encourage that a significant amount of training time be dedicated to shooting drills and practices for strikers and attacking players alike. Figures 4 and 5 display several percentage changes across the performances for Player B, with 7 out of the 8 action distributions observed differing by a figure of 5% or greater. This indicates a change in the actions undertaken by Player B over the performances in international and league matches, hinting at a possible contrast in the style of play.
The comparison between the mean action distributions for Players A and B can be seen in figures 6 and 7. This type of graph is useful as it provides an indication as to whether or not any correlation between the action distributions over international and league performances occur. This is distinctly clear in Figure 6 with the only anomaly appearing to occur in the action distribution of Shooting. The high frequency count for passing is again highlighted in figure 6, showing it to be the most utilised attribute by Player A. Although Passing accounts for such a high percentage of on the ball actions, it does not feature as one of the most important components of a striker’s overall priorities, ranking only 7th according to Smith (1973).
According to figure 7, the most utilised attribute for Player B varies across performance. Aerial Challenges feature as the most frequently used action during league performances (27%) followed by Passing (23%) and Dribbling (17%). Taking this in comparison with international performances where Dribbling is the most utilised attribute (23%) followed by Aerial Challenges (21%) and Passing (17%). This indicates clear changes in performance, likely to be caused by a general change in the style of play for the team, alongside adjustments in the role undertaken by the performer. A higher than anticipated percentage of time spent Shooting was observed with an average of 8.5% of overall on the ball time spent shooting on goal. Changes in the percentage of action distribution occurs frequently across almost all variables barring Tackling and Shooting with a 6% difference in the distribution of Aerial Challenges, Dribbling, Passing and Crossing along with a 5% difference in the distribution of Rule Breaches and Dispossessions observed for Player B.
5.3 Comparison of the Quality Rating of Techniques for Selected Variables
The mean quality rating of techniques was established to help eliminate the bias that can occur from simply showing frequency distribution graphs. The coding of performance indicators according to their relative success or unsuccessful implementation has been a prominent feature of several studies (Rico and Bangsbo, 1997; Pearce and Hughes, 2001; Probert, 2007). No exact quantitative analysis has been conducted whereby comparisons into the successful executions of technique on the same individual player, across international and club competition. This ability to conduct an exact objective quantification of critical events is vital for a complete analysis (Franks and Goodman, 1986).
Figure 8 shows a comparison of the mean quality rating of techniques employed by Player A for selected performance variables, during international and league matches. A mean score of -0.1 can be observed for the Aerial Challenge variable in international matches, which is significantly lower (p<0.05) than the 0.5 average attained for the same performance variable in league matches. Add this to the findings from the mean distribution (figure 6) in which Player A displays a 2.3 increase from 7.0 for international to 9.3 in league matches, for the same performance variable. This increase can be attributed to a tactical implementation into team strategy whereby long passes are purposely aimed to Player A during league games, resulting in more Aerial Challenges. This theory is further backed up by the results found in the analysis of Reep and Benjamin (1968) who found a trend in English League matches whereby a ‘long ball’ tactic was uncovered and accredited. From figure 8, it can be seen that the highest mean quality rating occurs for the performance variable of shooting, in both international and league matches being 1.2 and 2.5 respectively. This gives backing to the importance of the performance variable of shooting in strikers, as this is the highest technical rating observed for Player A, suggesting strong links between the practicing and development of a strikers ability to shoot.
In contrast to the quality rating of techniques observed for Player A, from figure 9 it can be seen that Player B has the lowest mean quality rating in the performance variable of Shooting, during league matches (0.1). This contradicts the findings from the international games where a quality rating of 1.6 was found. With no difference in player position during testing and no significant differences observed in the mean action distribution for Shooting, it could possibly be suggested that this low rating is due to the fact that Player B may have been suffering from or recently recovering from an injury, during the time period in which the league performances were analysed. This would undoubtedly affect a player’s performance and confidence in front of goal.
5.4 Distribution of Technique Rating Across Selected Performance Variables
Figure 10 shows the distribution of Player A’s Aerial Challenge technique ratings. The graph shows that the highest frequency observed for Aerial Challenges in league matches, is a technique rating of +1. This makes the likelihood of Player A being successful when completing an Aerial Challenge in league competition. In contrast, the highest frequency observed for Aerial Challenges in international matches is a technique rating of -1, suggesting that more Aerial Challenges are completed unsuccessfully at an international level. Figure 10 is backed up by the finds of Yamanaka et al (1993) who found that British teams performed significantly more headers in both their own and opponents half of the pitch compared to more continental teams. This goes some way to justifying the theory that a tactic of direct play is utilised in the league performances observed for Player A.
It would seem appropriate that a high frequency of Aerial Challenges in league games is followed by a high frequency in completed Crosses, as a percentage of Aerial Challenges will be resultant from balls delivered in and around the 18 yard box. This is confirmed with a frequency count n=7 observed at a rating of +1 in league games compared to n=4 for the quality rating in internationals. A frequency count n=4 is also observed at a rating of +2, indicating that several goal scoring opportunities were created as a result of Player A’s Crossing.
Wiemeyer (2003) identified that an effective striker is one who possesses good one to one play against an opponent. The ability to run with the ball both unopposed and under pressure can create problems for the oppositions defence and increase the probability of creating an attacking option (Probert, 2007). This is confirmed in figure 12 where the highest frequency is observed at the technical rating of +1 for both international and league performances. This demonstrates that Player A has a great ability at successfully dribbling with the ball to good effect for the team.
The majority of passes observed by Player A are at the quality rating of +1. This indicates that Player A’s passing is predominantly successful, however only a minority of passes create goal scoring opportunities for the team (n=4). As previously stated, passing is not necessarily a priority for a striker, and being strategically placed higher up the field can make passing more difficult with more pressure applied as a player advances towards an opponent’s goal. Reilly and Holmes (1983) state that the greater the distance to the opponents’ goal, the more time available on the ball, giving explanation to the high skill success rate in defensive areas as less pressure is applied.
From figure 14, a distinct lack of Shooting is apparent in league matches as apposed to the results obtained for the same performance variable in international games. Even so, a high success rate can be seen with all Shots taken in league games resulting in a technical rating of either +2 or +3. It can therefore be said that Player A appears to be a more clinical finisher in league competition in comparison to international games. A high percentage of Shots attempted during international matches resulted in a quality rating of either +1 or +2 (n=7 and n=8 respectively). This could possibly reflect a higher standard of goalkeeping and defending during international competition with a total of n=15 Shots either being blocked by a defender or saved by the goalkeeper during international matches, in comparison with n=2 for the analysed league games.
Returning to the viewpoint that English League Football is dominated by a more direct play approach, Player B also completes more Aerial Challenges at club level as opposed to international competition (figure 15). N=27 for +1 and n= 14 for -1 rated Aerial Challenges during league competition compared with n=18 for +1 and n= 6 for -1 when on international duty. This adds further evidence to the notion of direct play, however if this tactic is purposely employed, a frequency count of n=14 for -1 rated Aerial Challenges suggests that there could be room for improvement within this performance variable through training and coaching.
Figure 16 and 17 shows the distribution of Player B’s Crossing and Dribbling quality ratings respectively. A high frequency of +1 rated crosses from international matches is displayed (n=7) in comparison to league games (n=1). This contradicts the theory suggesting that a high frequency in Crosses is proportional to a high frequency of Aerial Challenges, as the overall amount of crosses completed by Player B in league competition is equivalent to n=3. This could again be due to the performer undertaking different roles within the international team, such as being asked to provide width and crosses for oncoming midfielders. As previously highlighted, successful Dribbling is an important attribute to a striker (Wiemeyer, 2003) and Player B scores high in the frequency for the +1 quality rating for the variable of Dribbling, suggesting a high-quality dribbling attribute is possessed by the performer.
Erratic shooting has previously been reported in studies where the coding of performance indicators according to their relative success or unsuccessful implementation has been a prominent feature. It may be universally accepted by coaches that strikers could have 2 erratic shots for every goal scored that could win the game, as opposed to creating a high frequency of mediocre shots which have no general influence (Probert, 2007). This is apparent in figure 19 where a high frequency count is observed for technical ratings ranging from -2 through to +3.
More successful shooting is apparent during international fixtures with two goals observed at international competition and zero goals observed in league games.
5.5 Mean Behaviour Rating and Frequency for First and Last 15 Minutes of Matches
Throughout the regulatory 90 minutes of a soccer game, the team performances along with an individuals performance fluctuates as good and bad spells occur during matches. By solely focusing on the data for the mean behaviour rating and the mean frequency count of behaviours occurring in the first and last 15 minutes of performances, some indication towards the areas of the game where the performers in question are most successful (or unsuccessful) may be gained. This sort of information may be taken into account by coaches when calculating substitutions and tactical changes.
From figure 20, a higher behaviour rating can be seen to occur for Player A in the last 15 minutes of international matches as opposed to the first 15 minutes of play. No differences in the mean behaviour rating were observed at club level; however, an inferior mean behavioural rating is displayed for both the first and last 15 minutes of club matches in comparison to that of international games. Figure 21 indicates that Player A has a higher behaviour frequency for both club and international matches during the first 15 minutes in comparison to the last. In the last 15 minutes of international games, a high behaviour success rating is apparent, yet the behaviour frequency is at its lowest during this time. This could point towards the opposing teams being disorganised in the final minutes of the game, possibly fatigued, leaving time and space for Player A to exploit. It can be interpreted by a coach that Player A does not appear to under perform over the latter stages of a game.
Figure 22 displays a perceivably low mean behaviour rating in club matches during the last 15 minutes (n=0.16) in comparison to the high mean behaviour rating observed in the last 15 minutes of international competition (n=0.52). This is accompanied by a low behaviour frequency count during the last 15 minutes of club matches (n=6.33), displayed in Figure 23. This can be seen to give strength to the argument that Player B was injured or recovering from injury during the league analysis, and therefore was lacking in physical fitness.
5.6 Mean Player Rating Across Performance
Overall player rating across performance shows the overall mean rating for league and international matches. This is done by adding all the player ratings from the analysed games and comparing the averages. This gives an overall and clear view that both Player A and Player B scored slightly higher in international performances (n=42.7 and n=19 respectively) compared to their individual league performances (n=37 and n=13 respectively). Although no reasoning can justify the finds of this experiment as to the reasoning behind the results displayed, further analysis could help understand the nature of these results.
6.0 Conclusion
6.1 Findings of Study
The study consisted of a hand notation system with an accepted level of intra-observer reliability under the Chi-squared statistic (p>0.95) for both behaviour frequency observation and technique success rate.
Significant differences (p<0.05) were found in the action distribution for both Player’s A and B, across the respective international and league performances.
Comparison of the mean quality rating of techniques showed significant differences in the selected performance variables of Aerial Challenges and Shooting for both Player A and Player B.
Mean behaviour rating and frequency for the first and last 15 minutes of games demonstrated significant differences (p<0.05) in international performances for Player A, and international and club performances in Player B.
Significant differences (p<0.05) were observed across the mean player rating for performance for league and international competition with both Players A and B performing at a higher mean rating in international level. This contradicts the hypothesis that the higher expectation levels and unfamiliarity of surroundings at the EURO 2008 Championships, a lower level of technically ability will be observed.
6.2 Recommendations for Players and Coaches
From the results reported, it can be stated that significant differences do occur between individual performances for club and international representations. The results could have a practical impact in two ways:
- By encouraging coaches to implement the variables in which individual performers excel more frequently into team strategies.
- By highlighting the varying levels of performance for individual players at the beginning and end of matches, and the affects this has on their input to the team.
6.3 Future Recommendations
This test could be assisted by analysing more games, consequently producing more data which may influence any significant values (Hughes et al., 2002).
A further study may wish to alter the technical rating system to allow for one individual system by which all performance variables can be scored by, instead of several individual scoring systems.
A Kruskal-Wallis and Mann Witney post hoc test (Vincent, 1999) could be conducted to calculate the significance of differences between variables. This type of statistical test may create further significance between the average technical ratings.
References
Ali, A.H. (1988). A statistical analysis of tactical movement patterns in soccer. In Science and Football (edited by T, Reilly., A. Lees., K. Davids. and W.J. Murphy.), pp. 302-308. London: E&FN Spon.
Atkinson, G. and Nevill, A.M. (1998). Statistical methods for assessing measurement error (reliability) in variables relevant to sports medicine. Sports Medicine, 26 (4), 217-237.
Bangsbo, J. (1997). The physiology of intermittent activity in football. In Science and Football III (edited by T. Reilly, J. Bangsbo & M. Hughes), pp. 43-53. London: E&FN Spon.
Bate, D. (1996). Soccer skills practice. In Science and Soccer (edited by T. Reilly.), pp. 227-241. London: E&FN Spon.
Bate, R. (1988). Football chance: Tactics and Strategy. In Science and Football (edited by T, Reilly., A. Lees., K. Davids. and W.J. Murphy.), pp. 293-301. London: E&FN Spon.
Bland, J.M. and Altman, D.G. (1986). Statistical methods for assessing agreement between two methods of clinical agreement. The Lancet, Feb: 8, 307-310.
Dunn, A., Ford, P. and Williams, M. (2003). A technical profile of different playing positions. Insight, 6(4), 41-45.
Franks, I.M. (1997) Use of feedback by coaches and players. In Science and Football III (edited by T. Reilly, J. Bangsbo & M. Hughes), pp. 267-278. London: E&FN Spon.
Franks, I.M. and Goodman, D. (1986). A systematic approach to analysing sports performance. Journal of Sports Sciences, 4, 49-59.
Franks, I.M. and McGarry T. (1996). The science of match analysis. In Science and Soccer (edited by T. Reilly.), pp. 363-375. London: E&FN Spon.
Grehaigne, J.F., Mahut, B. and Fernandez, A. (2001). Qualitative observation tools to analyse soccer. International Journal of Performance Analysis in Sport - E, Vol. 1: No 1, pp.52-61.
Hopkins, W.G. (2000). Measures of reliability in sports medicine and science. Sports Medicine, 30 (1), 1-15.
Hughes, M. (1993). Notation Analysis in Football. In Science and Football II (edited by T. Reilly., J. Clarys. and A. Stibble), pp.151-159. London: E&FN Spon.
Hughes, M. (1996). Notational analysis. In Science and Soccer (edited by T. Reilly.), pp. 343-361. London: E&FN Spon.
Hughes, M. (1998). The application of notational analysis to racket sports. In Science and Racket Sports II (edited by A. Lees, I. Maynard, M. Hughes and T. Reilly), pp. 211-220. E and FN Spon: London.
Hughes, M. and Franks, I, M. (2004). Notational Analysis of Sport (2nd Edition). Routledge: London.
Hughes, M., Robertson, K. and Nicholson, A. (1988). Comparison of patterns of play of successful and unsuccessful teams in the 1986 World Cup for soccer. In Science and Football (edited by T, Reilly., A. Lees., K. Davids. and W.J. Murphy.), pp. 363-367. London: E&FN Spon.
Hughes, M., Cooper, S.M. and Nevill, A. (2002). Analysis procedures for non parametric data from performance analysis. International Journal of Performance Analysis in Sport - E, Vol. 2: No 1, pp.6-20.
James, N., Mellalieu, S.D. and Hollely, C. (2002). Analysis of strategies in soccer as a function of European and domestic competition. International Journal of Performance Analysis in Sport - E, Vol. 2: No 1, pp.85-103.
Leibermann, D.G., Katz, L., Hughes, M.D., Bartlett, R.M., McClements, J. and Franks, I.M. (2002). Advances in the application of information technology to sport performance. Journal of Sports Sciences, 20, 755-769.
Luhtanen, P., Belinskij, A., Hayrinen, M. and Vanttinen, T. (2001). A comparative tournament analysis between the Euro 1996 and 2000 in soccer. International Journal of Performance Analysis in Sport - E, Vol. 1: No 1, pp.74-82.
O’Donoghue, P.G., Boyd, M., Lawlor, J. and Bleakley, E.W. (2001). Time –motion analysis of elite, semi-professional and amateur soccer competition. Journal of Human Movement Studies, 41, 1-12.
Olsen, E. and Larsen, O. (1997). Use of match analysis by coaches. In Science and Football III (edited by T. Reilly, J. Bangsbo & M. Hughes), pp. 209-220. London: E&FN Spon.
Partridge, D. and Franks, I.M. (1997). The use of computer video interactive analysis in the sport of soccer: changing individual performance by providing quantitative and qualitative feedback. In Notational Analysis of Sport I & II (edited by M.Hughes.), pp. 99-111. CPA: UWIC: Cardiff.
Partridge, D., Mosher, R.E. and Franks, I.M. (1993). A computer assisted analysis of technical performance – a comparison of the 1990 world cup and intercollegiate soccer. In Science and Football II (edited by T. Reilly., J. Clarys. and A. Stibble), pp.221-231. London: E&FN Spon.
Pearce, M. and Hughes, M. (2001). Substitutions in Euro 2000. In pass.com (edited by M.Hughes. and I.M. Franks.), pp. 303-315. CPA: UWIC: Cardiff.
Pollard, R., Reep, C. and Hartley, S. (1988). The quantitative comparison of playing styles in soccer. In Science and Football (edited by T, Reilly., A. Lees., K. Davids. and W.J. Murphy.), pp. 309-315. London: E&FN Spon.
Probert, G. (2007). A technical analysis of elite male soccer players by position and success. Unpublished Bsc dissertation. Cardiff U.K. UWIC, 68 pages.
Reep, C. and Benjamin, B. (1968). Skill and Chance in Association Football. Journal of the Royal Statistical Society, 131, 581-585.
Reilly, T. and Holmes, M. (1983). A Preliminary analysis of selected soccer skills. Physical Education Review, 6(1), 64-71.
Reilly, T. and Thomas, V. (1976). A motion analysis of work-rate in different positional roles in professional football match play. Journal of Human Movement Studies, 2, 87-97.
Rico, J. and Bangsbo, J. (1997). Coding system to evaluate actions with the ball during a soccer match. In Notational Analysis of Sport I & II (edited by M.Hughes.), pp. 85-90. CPA: UWIC: Cardiff.
Smith, M. (1973). Success in Football. Butler and Tanner: London.
Taylor, J.B., Mellalieu, S.D. James, N. (2004). Behavioural comparisons of positional demands in professional soccer. Department of Sports Science, University of Wales Swansea, Swansea, United Kingdom. Correspondence concerning this article should be addressed to Joseph B. Taylor, Department of Sports Science, Vivian Tower, University of Wales Swansea, Singleton Park, Swansea, SA2 8PP, United Kingdom. Tel: 00-44-1792-29-5075, Fax: 00-44-1792-51-3171, Email:[email protected].
Thomas, J.R. and Nelson, J.K. (2001). Research Methods in Physical Activity (4th Edition. Human Kinetics:USA.
UEFA Official website (2009)
(accessed last on 23rd January 2009)
Van Lingen, B. (1997). Coaching Soccer. Reedswain, Spring City: USA.
Vincent. W.J. (1999). Statistics in Kinesiology (2nd Edition). Human Kinetics: USA
Wiemeyer, J. (2003). Who should play in which position in soccer? Empirical evidence and unconventional modelling. International Journal of Performance Analysis in Sport - E, Vol. 3: No 1, pp.1-18.
Yamanaka, K., Liang, D,Y.and Hughes, M. (1997). An analysis of the playing patterns of the Japan national team in the 1994 World Cup qualifying match for Asia. In Science and Football III (edited by T. Reilly, J. Bangsbo & M. Hughes), pp. 221-228. London: E&FN Spon.