Choking under pressure: The comparison of kick order deterioration between soccer penalty shootouts of international tournaments and national tournaments

(1)

1

Choking under pressure: The comparison of kick order

deterioration between soccer penalty shootouts of

international tournaments and national tournaments

Name: Roy Christiaan Honingh

Student ID: 11847581

BSc: Business Administration

Faculty: Economics and Business

Supervisor: Dhr. drs. R. (Rob) van Hemert

(2)

Statement of Originality

This document is written by Roy Christiaan Honingh who declares to take full responsibility for the content of this document.

I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

Abstract

Choking under pressure means performing worse than expected in situations with a high degree of perceived importance. Evidence was found for choking under pressure in soccer shootouts in the three biggest international tournaments (World Cup, European Championship and Copa America). Fewer goals were scored on kicks that were more immediately decisive to the outcome (kick numbers 4 to 9) compared to the initial kicks. However, this finding was based on 409 observations in which kick number 6 to 9 only counted for 28 observations. This study examines whether this kick order deterioration is generalizable to shootouts in national tournaments. More specifically, it is hypothesized that choking under pressure also occurs for the most decisive kick numbers. A total of 2179 penalty shots were included in this research. The results showed no direct negative relationship between kick number and

outcome. However, evidence was found for performance decline of kick number 6 to 17 in the (semi)finals compared to earlier tournament rounds. So, kick order deterioration is not generalizable to the national tournaments. Instead, this research suggests that the pressure in national tournaments is lower than international tournament by proving that choking only occurs in the last knockout stages of the tournament.

(4)

Introduction

The Laws of the Game (The International Football Association Board [IFAB], 2019) in soccer state that when a winner has to be declared and two teams are tied after extra time in a

tournament, the penalty shootout (or kicks from the penalty mark) is used to decide the winner. The ball is placed eleven meters from the goal and the player gets the opportunity to score against the opponents’ goalkeeper.

The shootout, which consists of an alternating series of penalty kicks, is won by the team with the highest goal tally after n kicks per team (n = 5). In the event of a tie after five penalty kicks each, the shootout progresses to a ‘sudden death’ by increasing the n in iterative fashion (n = n + 1) until one team obtains a higher goal tally after an equal number of kicks taken per team. The team to strike first is determined after the end of the extra time in the match by the toss of a coin (McGarry & Franks, 2000).

The goal is 2.44 meters high and 7.32 meters wide and the goal keeper is not allowed to stand in front on the goal line. The ball reaches the goal in approximately 0.2 seconds which makes it a rather unfair play in which the player is expected to score (IFAB, 2019).

While ‘expected to score’ sounds rather easily, research confirmed that this is not always the case. Jordet, Hartman, Visscher & Lemmink (2007) analyzed penalty shootouts in the three biggest international tournaments (World Cup, European Championship and Copa America) and confirmed kick order deterioration during the shootouts. The fourth until the nineth shot were scored significantly less often (conversion of 73.6%) with regards to the first three shots (conversion of 82.5%) of each team. Furthermore, the most decisive shots were performed significantly worse than the first three shots. The researchers explained that importance of the outcome, with the fourth shot onwards being potential decisive shots, adds extra stress, which reduces the performance of the shooters.

This phenomenon is known as ‘choking under pressure’. Choking under pressure can be defined as performing worse than expected in situations with a high degree of perceived importance (Baumeister, 1984; Beilock & Grey, 2007). Stressors related to outcome (e.g. the prospect of missing the penalty) constitute major environmental sources of stress. According to Jordet & Elferink-Gemser (2012), even longer waiting time in the mid-circle, with

cumulative exposure to high levels of stress, increases pre-shot anxiety levels and this negatively affects shot performance.

(6)

6

Beilock and Carr (2001) elaborated on the choking under pressure definition by arguing that there is an important mechanism via which too much arousal leads to worse performance. They explained this mechanism as the switch from automatic behavior to consciously controlled behavior. This switch can be caused by anxiety and self-consciousness about performing, which is a result of increased performance pressure. The authors believe that this explicit monitoring theory is especially important for sensorimotor skills and therefore could apply during penalty shootouts in soccer tournaments. Tasks like taking a penalty kick are highly practiced, so it might be better to take the shot without too much thought (automatic) than thinking about the task (controlled).

Furthermore, Baumeister (1984) found support for the explicit monitoring theory with experimenting two different kinds of pressure. First, pressure coming from self-presentational concerns caused by competition and observers. Second, pressure coming from monetary incentives. Both types seem to apply to penalty kicks. In penalty shootouts, a reason for self-presentational concern is that the outcome of the penalty kick could be of great importance to the outcome of the shootout (and thus the game). Self-presentational concern could also arise with the large amounts of spectators in the stadiums as well as the (live) broadcasting on (inter)national television. Finally, to elaborate on pressure coming from monetary incentives, large financial rewards can be gained by winning games and conceding to further rounds in soccer tournaments.

Interestingly, Jordet et al. (2007) found that the average fifth shot in all international tournaments had a higher (nonsignificant) chance of scoring than the second, third and fourth shot. An explanation was that the fifth shot was decisive for whether you win the shootout or not. Therefore, soccer trainers would use a tactic in which the fifth player would also be one of their best skilled players. McGarry & Franks (2000) elaborated on this by arguing that the best five ranked penalty shooters from the on-field players should be assigned to the first five penalty kicks in the reverse order of ability. That is, the fifth best penalty shooter should take the first penalty kick, the fourth best penalty shooter should take the second penalty kick, and so on. In the event of a sudden death, the next highest ranked on-field player should be assigned to the next penalty kick until the shootout ends. So, for this tactic to be successful, players should be ranked a priori on their penalty-taking ability.

To complement, Jordet & Hartman (2008) found that when the (final) fifth shot could be the winning shot, performance was significantly better. To explain this final shot, the type

(7)

7

of shot was classified into positive and negative valence shots. Shots where a goal instantly leads to a victory were classified as positive valence shots and shots where a miss instantly leads to a loss as negative valence shots. Then, avoidance behavior, defined as looking away from the goalkeeper or preparing the shot quickly (thus speeding up the wait), occurred more with negative valence shots than shots with positive valence. Thus, players with negative valence shots performed worse than players with positive valence shots.

However, both the researches by Jordet et al. (2007) and Jordet & Hartman (2008) were based on data from multiple international tournaments. The purpose of this study is to investigate whether the conclusions based on the international tournaments are generalizable to national tournaments. Does the kick order deterioration relationship, explained by Jordet et al. (2007) also occur in national tournaments where the perceived importance of the outcome is less than for international tournaments?

So far, several causes for choking under pressure have been discussed. To summarize, the importance of the outcome of the shot, the accumulative waiting time in the mid-circle, the explicit monitoring theory, self-presentational concerns, monetary incentives, the best skilled player tactic and valence could have a significant (negative) impact on shot

performance. However, this study will not control for all possible causes discussed above as the unit of analyses are the findings of the paper of Jordet et al. (2007).

The conclusions of Jordet et al. (2007) were based on a really small sample size. For example, shot six until nine only had 28 observations. On top of that, the finding of the kick order deterioration was on a multivariate analysis with only a limited number of controls. The quality of the player and negative and positive valence were not taken into account. So, as a result of the ‘choking under pressure’ definition, in which teams perform worse than expected in situations with a high degree of perceived importance (Baumeister, 1984; Beilock & Grey, 2007), the first hypothesis will be a similar check for the findings of Jordet al. (2007) in national tournaments. Furthermore, the findings will thus be controlled for the quality of the players taken the relating kicks, the differences in experience of the players and also for valence to draw more accurate conclusions about the decisive kicks in this relationship.

Hypothesis 1. Kick number negatively relates to shot performance in national tournaments.

(8)

8

Secondly, Baumeister (1984) supplied evidence with his monetary incentive experiment, such that external financial incentives increase pressure which resulted in worse performance. Wells & Skowronski (2012) found similar results in the prestigious PGA golf tournaments in which professionals on the PGA tours scored significantly worse in the (final) fourth round compared to the third round. ‘Par’ is a term in the game to denote the number of strokes that a golfer should require to finish a hole ( i.e. 3, 4 or 5). If a golfer requires fewer strokes to complete a hole, he or she is under par. Whereas, if the golfer requires more strokes to complete a hole, he or she is over par. In the (final) fourth round, the scores were higher, meaning that the golfers took more shots to complete the holes. Moreover, the closer a player was to a tournament lead, the larger his choking score. Keeping in mind that in this final round the prices were to be distributed among the top-ranked players of the tournament, increasing the degree of perceived importance of performance in the final round.

To further find out whether higher potential rewards resulting in inferior shot

performance in the biggest international soccer tournaments is generalizable to national cups, a second situational factor will be considered, namely the round of the tournament. Given the fact that the (financial) external rewards increase after a soccer team proceeds to the next round in the tournament, the round of tournament is an accurate indicator to test whether shot performance differs among the different stages of the tournament.

Jordet et al. (2007) found evidence for significantly inferior shot performance in the most prestigious tournament (the World Cup) compared to the smaller intra-continental tournaments (European Championship and Copa America). Although only concluding that significantly less goals are scored in the biggest tournament, this finding suggests that when the more is at stake, the higher the perceived pressure, resulting in worse performance. This also corresponds with the definition of choking under pressure by Baumeister (1984) and Beilock & Grey (2007). This potential relationship will be tested in the second hypothesis.

Hypothesis 2. Round of tournament negatively relates to shot performance in national tournaments

(9)

9

There is a reason to believe that the kick order deterioration is highest in the final stages of the tournament. Due to the lack of arguments found in the Jordet et al. (2007) article, round of tournament could be an important moderator to actually examine whether higher incentives, as a result of preceding to further rounds, lead to worse kick order performance. Combined with the findings of Baumeister (1984) and Wells & Skowronski (2012) who emphasize that increased pressure on decisive moments is higher in final stages of a tournament, which results in more choking, it is predicted that kick order deterioration in shootouts is stronger in later tournaments rounds. This prediction is tested in the third hypothesis.

Hypothesis 3. The negative relationship between kick number and shot performance is moderated by round of tournament, such that this relationship is stronger for later rounds in the tournament.

(10)

10

Methods

Data

In this paragraph the methods used to conduct this research and the way of collecting the data will be discussed. Because the data is considered to build upon current knowledge, it is considered a deductive research. The data consisted of 203 shootouts taken in 17 different national soccer cups in Europe. A total of 2179 penalty shots, taken between seasons 2008-2009 to 2019-2020, were included in this research. The average age of the players

participating in the shootouts was 27.4 years old (SD = 4.4). The different positions of the players in the shootout were divided in defenders (n = 671) midfielders (n = 778) and forwards (n = 730).

Out of all attempts, 271 shots were considered as negative valence, where missing the shot means losing the game. Whereas, 162 shots were considered as positive valence, where scoring the shot means winning the game. All the other shots (n = 1746) were considered as neutral shots, because they were not the decisive shots that determine either termination or progression.

Furthermore, the least penalties taken in a shootout were 6 shots in total. The highest amount of shots were 34 in a single shootout. Unfortunately, as a result of a lack of several data (unknown age and/or position), some shots and even full shootouts were taken out of the dataset, which resulted in possible uneven amounts of shots per kick number. The total shots were further subdivided in different kick number groups. Every kick number group has its own amount of shots. As a result of cleaning the data, kick number three had the most observations (n = 402). However, as the total amount per kick number decreases, kick numbers 6 until 17 were combined (n = 388).

The total shots were unequally divided between the different rounds of tournament. The most shots (n = 992) were taken in the round of 16. Furthermore, the quarter finals counted for 476 shots, the semifinals for a total 424 shots and the finals for 287 shots. Data collection took place online using www.transfermarkt.com and were collected by several fellow students and a supervisor.

(11)

11 Variables

The dependent variable was shot performance. Only scored penalties were considered a success (goal = 1), whether a shot was missed or saved, was considered a fail (miss = 0). Two independent variables were considered. The first one was kick number, which subdivides the total amount of kicks in 6 groups. The second independent and also moderator variable is round of tournament, which has four measures: The round of 16, the quarter finals, the semifinals and the final.

Furthermore, three control variables were used to rule out alternative explanations. To control for quality and experience, two control variables were used: Position and age. Age is measured by taking the second year of the particular season minus the year of birth of the player. Moreover, age was divided in 3 dummy variables by taking the mean age and one standard deviation on either side of the mean. This resulted in ‘young’ (17 to 23 years old), ‘medium’ (24 to 31 years old) and ‘old’ (32 to 48 years old). Position corresponds to the specific task of the players and had dummy variables for defender, midfielder and forward. The last control variable was valence. Valence had three dummy variables: Neutral shots, positive valence shots and negative valence shots. Hereby positive valence corresponds to the shot where scoring means winning the shootout, whereas negative valence means missing is losing.

Data analysis and predictions

Kick number descriptive statistics were provided per kick number group, conversion rate and number of shots. Also, the descriptive statistics for the round of tournament were provided per round of tournament, the conversion rate and the number of shots. Same has been done for all control variables. For moderation, descriptive statistics were provided of kick number

conversion rates per round of tournament. To examine the association between the two independent variables kick number and round of tournament on shot performance, Pearson correlation coefficients were calculated. A correlation between 0.10 and 0.30 was considered weak, 0.30-0.50 moderate and above 0.50 strong. Furthermore, two multivariate associations were considered using a binary logistic regression analysis (Enter method). One with the two independent variables as predictors and shot performance as outcome variable. And the second multivariate associations examined the difference in the effect of the moderator on the relationship between kick number and shot performance. In order to control for age, position and valence, these variables were added to model as covariates.

(12)

12

ODDs Ratio’s (OR) were used as the relative measure of effect. The statistical analysis was performed by the use of SPSS (version 25), where the level of significance was set at 0.05.

The first hypothesis will test the main effect of kick number on outcome. The second hypothesis will the test the main effect of round of tournament on outcome. The third

hypothesis will test the interaction, namely the moderating effect of the round of tournament on the relationship between kick number and shot performance.

To sum up, it is expected that there will be enough evidence for the first hypothesis, in which the more decisive kick number will choke more under pressure. Also, it is expected that the increased external pressures in further tournaments rounds will result in lower shot performance. For the third hypothesis, it is expected that there will be a significant kick order deterioration difference between the round of 16 and the final, most decisive, stage. This difference will be the strongest in the combination of the final rounds and the most decisive kick number (kick number 6 to 17) with regards to this kick number in the round of 16.

(13)

13

Results

Table I. Correlation table

. Mean Outcome Kick 1 Kick 2 Kick 3 Kick 4 Kick 5 Kick 6 to 17

Round of 16

Quarter Semi Final Defender Midfielder Forward NV PV No Valence 17-23 years old 24-31 years old 32- 48 years old Outcome .73 — Kick 1 .18 .007 — Kick 2 .18 -.011 -.226** — Kick 3 .18 -.003 -.226** -.226** — Kick 4 .17 .004 -.218** -.218** -.218** — Kick 5 .12 .026 -.174** -.174** -.175** -.168** — Kick 6 to 17 .16 -0.21 -.203** -.203** -.204** -.196** -.157** — Round of 16 .46 .037 .006 .006 .007 .010 .00 -.030 — Quarter .22 -.030 .007 .007 .006 .010 .012 -.042* -.483** — Semi .19 -.021 -.006 -.006 -.007 -.014 -.012 .046* -.449** -.260** — Final .13 .006 -.010 -.010 -.010 -.010 .00 .043* -.356** -.206** -.191** — Defender .31 -.011 -.106** -.024 -.053* .038 -.002 .159** .023 -.035 .009 -.001 — Midfielder .36 -.009 .069** .042 .021 -.035 -.067** -.044* .007 -.016 .006 .001 -.497** — Forward .34 .020 .034 -.018 .031 -.002 .070** -.111** -.030 .051* -.015 .00 -.473** -.529** — NV .12 .008 -.179** -.179** -.158** .004 .287** .292** .019 -.028 .022 -.019 .068** -.054* -.011 — PV .07 .00 -.135** -.135** -.121** .041 .318** .091** .001 .007 -.007 -.002 .035 -.007 -.027 -.107** — No valence .80 -.048* -.058** -.009 .046* .013 .011 -.001 .009 .005 .004 -.024 -.008 -.005 .012 .029 -.022 — 17-23 years old .20 -.005 -.007 -.007 -.041 .034 .013 .013 -.034 .094** -.087** .038 .005 .026 -.031 .022 .026 -.026 — 24-31 years old .63 -.010 .014 -.013 .019 -.030 -.018 .027 -.087* -.006 .110** .006 -.001 -.012 .013 -.005 -.015 .006 -.648** — 32-48 years old .17 .018 -.009 .025 .018 .003 .009 -.048* .147** -.091** -.049* -.048* -.004 -.012 .016 -.017 -.009 .020 -.227** .594** —

(14)

14

Table I shows the correlation table of all variables included in this research. There is a weak correlation between the dependent variable and the independent variables. Although

insignificant, kick numbers two, three and 6 to 17 are negatively correlated to outcome. Same assumption holds for the different rounds of tournaments, where the quarter final and

semifinal are negatively corelated to outcome, whereas the round of 16 and the final have slightly positive correlations.

Regarding the control variables, although insignificant, it seems like the forwards are more positively correlated to outcome than midfielder and defenders do. For the different age groups, only the most experienced players show a slightly positive correlation (insignificant). Interestingly, the only significant weak correlation to outcome is of the ‘no valence’ group (r = -.048).

Control variable defender shows significant correlation with kick number 1(r = -.106) and significant positive correlation with kick 6 to 17 (r =.159), suggesting that slightly less defenders take the first shot and most defenders take the shots after the fifth. On the opposite, forwards have a significant positive correlation with kick number 5 (r = .070) and a

significant negative correlation with kick number 6 to 17(r = -.111). This may suggest that the best shooters, are chosen to take the most decisive shot (kick number 5). The control variable negative valence shows significant correlation with defender (r = .068), what is more or less in line with defenders taking shots in the ‘sudden death’ stage in the shootout; kick number 6 to 17. Even though very weak, the older players show a significant negative correlation (r = -.048) to kick number 6 to 17. Overall, this underlines the importance of a multivariate analysis which is depicted in table II on the next page. The results will be discussed per hypothesis.

(15)

15 Table II. Multivariate associations with regard to shot performance (1/2)

Analyses was performed using a logistic regression model. 2179 kicks were included in the analysis. None of the control variables were significantly related to shot performance.

Variables Shot Performance

Odds ratio (p-value) Kick number Kick 1 Kick 2 Kick 3 Kick 4 Kick 5 Kick 6 to 17 1.143 (0.470) 1.048 (0.794) 1.085 (0.653) 1.125 (0.497) 1.305 (0.172) 1.000 (Ref.) Round of Tournament Round of 16 Quarter Semi Final 1.000 (Ref.) 0.797 (0.069) 0.831 (0.154) 0.945 (0.710) Age Young Medium Old 1.000 (Ref.) 1.040 (0.755) 0.954 (0.770) Position Defender Midfielder Forward 1.000 (Ref.) 1.010 (0.933) 1.094 (0.468) Valence Neutral NV PV 1.000 (Ref.) 1.029 (0.873) 0.948 (0.801)

(16)

16

The first hypothesis is about the negative relationship between kick number and shot performance. Table III presents the descriptive statistics of the different kick number groups. In the twelve years considered in this research, a total of 2179 shots have been performed in the 17 national cup. A great amount of the shots were converted into a goal (72.9%, n = 1588). Out of all kick number subgroups shot 5 was conversed the most (76.1%). The

different conversion rates between all kick numbers in the national tournaments seemed very constant with the lowest conversion rate being 70.7% (N = 338). Interestingly, when

excluding the first kick number, kick number 2 to kick number 5 show actually no kick order deterioration. Instead, a kick order improvement can be observed (better visualized in figure II). However, according to the first multivariate analysis, there is no significant kick order deterioration nor kick order improvement between the different kick number groups. Although table III shows a decrease in shot performance after kick number 5 (-5.4%), this difference is not significant enough (OR = 1.305, P = 0.172). So, there is not enough evidence to support the first hypothesis.

Table III. Conversion rate per kick number

Figure II. %-scoring per kick number

Kick number N N-scoring %-scoring

1 401 295 73.6% 2 401 288 71.8% 3 402 292 72.6% 4 378 277 73.3% 5 259 197 76.1% 6 to 17 338 239 70.7% Total 2179 1588 72.9%

(17)

17

Considering the control variables, Table IV, V and VI show the descriptive statistics. The results show that there is very little difference between the subgroups of each control variable. For position, there is a slightly higher conversion of forward players compared to midfielders and defender. Furthermore, neither do the different age groups show interesting variation in conversion rates. More interesting is the fact that the negative valence group have a higher conversion rate (n = 200, 73.8%) that the positive valence shots (n = 118, 72.8%). However, the first multivariate analysis shows no significant effect of the different ages of the player, nor the different positions of the player on outcome. Also, the differences in valence are of no significance with regards to the outcome.

Table IV. Conversion rate per position

Table V. Conversion rate per age group

Table VI. Conversion rate per valence group

Position N N-scoring %-Scoring

Defender 671 484 72.1%

Midfielder 778 563 72.4%

Forward 730 541 74.1%

Total 2179 1588 72.9%

Age N N-scoring %-Scoring

17-23 (young) 432 314 72.7% 24- 31 (medium) 1372 1005 73.3% 32-48 (old) 375 269 71.7% Total 2179 1588 72.9%

Valence N N-scoring %-Scoring

Neutral shots 1746 1270 72.7%

Negative valence 271 200 73.8%

Positive valence 162 118 72.8%

(18)

18

The second hypothesis is about the negative relationship between round of tournament and shot performance. The conversion of the different rounds of tournament are in table VII (better visualized in figure III). From the different rounds of tournament, the conversion rate is highest for the round of 16 (74.7%, n = 741 ), as expected. There is a decrease in

performance after the round of 16 (-4.3%). But for further rounds, an increase in performance is perceived. With regards to the round of 16, performance in all further tournament round was worse, with the quarter finals scoring the least (70.4 %, n = 476).

Furthermore, the only big difference in outcome is that of the last 16 compared the quarterfinals. However, as the multivariate analysis indicates, this difference is not significant enough (OR = 0.797, P = 0.069). Combined with the fact that the performance even

(insignificantly) increased after the quarter finals, there is no support for hypothesis two.

Table VII. Conversion rate per round

.

Figure III. %-scoring per round of tournament Round of Tournament N N-scoring %-Scoring Last 16 992 741 74.7% Quarter 476 335 70.4% Semi 424 301 71% Final 287 211 73.5% Total 2179 1588 72.9%

(19)

19

The third hypothesis is about the positive, moderating effect of round of tournament on the negative relationship between kick number and shot performance. Table VIII shows the different conversion rates of the different kick numbers per round of tournament. Considering the fifth shot in the final stages was interesting, as the conversion rate was the highest (85.29%). However, this high conversion rate was only based on 34 observations. Furthermore, the semifinals and the finals show a big difference in conversion between the most decisive shots (kick number 6 to 17) and kick number 5. Also, for the round of 16, this difference is almost equal to kick number 5. To examine whether there are significant differences, a second multivariate analysis had been done (table IX). The table shows the differences of the round of 16 compared to the semifinal and final. Interestingly, there is a significant difference in outcome when comparing the round of 16 with the semifinals and finals. Kick number 5 in the semifinals and finals scored significantly better than kick number 6 to 17 (OR = 2.319, P = .021) while this difference in the round of 16 is more or less the same (OR = 1.002, P = 0.996).

Furthermore, the control variable position also shows a significant difference for the semifinals and finals, in which the forward players score significantly more than defenders (OR= 1.708, P= 0.017). Whereas this difference is not significant for the round of 16 (OR= 0.822, P= 0.298). This might suggest that the forward players perform better in later

tournament rounds.

All in all, according to the analysis, there is actually evidence for the third hypothesis. With regards to kick number 5 and kick number 6 to 17, the kick order deterioration is thus stronger in later tournaments.

Table VIII. Conversion of kick number per round of tournament Round of tournament %-scoring Kick 1 %-scoring Kick 2 %-scoring Kick 3 %-scoring Kick 4 %-scoring Kick 5 %-scoring Kick 6 to 17 Round of 16 76.22% 76.76% 74.73% 69.89% 75.42% 75.35% Quarter 72.22% 73.33% 67.78% 73.26% 66.67% 66.67% Semi 72.37% 59.21% 71.05% 79.71% 82.98% 66.25% Final 68.00% 70.00% 76.00% 76.60% 85.29% 69.64% Total 72.20% 69.83% 72.39% 74.87% 77.59% 69.48%

(20)

20

Table IX. Multivariate associations with regard to shot performance (2/2)

Variables Variables Odds-Ratio (P-value)

Round of 16 Kick number

Kick 1 Kick 2 Kick 3 Kick 4 Kick 5 Kick 6 to 17 1.167 (0.598) 1.199 (0.532) 1.078 (0.793) 0.806 (0.415) 1.002 (0.996) 1.000 (Ref.) Age group Young Medium Old 1.000 (Ref.) 0.921 (0.654) 1.089 (0.729) Position Defender Midfielder Forward 1.000 (Ref.) 0.924 (0.669) 0.822 (0.298) Valence Neutral NV PV 1.000 (Ref.) 1.141 (0.615) 1.210 (0.556) No round of 16 (Semifinal and final)

Kick number Kick 1 Kick 2 Kick 3 Kick 4 Kick 5 Kick 6 to 17 1.076 (0.808) 0.761 (0.350) 1.115 (0.715) 1.587 (0.125) 2.319 (.021*) 1.000 (Ref.) Age Young Medium Old 1.000 (Ref.) 1.086 (0.720) 0.669 (0.151) Position Defender Midfielder Forward 1.000 (Ref.) 1.152 (0.490) 1.708 (0.017*) Valence Neutral NV PV 1.000 (Ref.) 0.970 (0.921) 0.874 (0.718)

(21)

21

Discussion

Summary

The aim of this research was to examine whether kick order deterioration in shootouts from international tournaments is generalizable to shootouts in national tournaments.

The results did not find support for the first hypothesis, which states that kick number negatively relates to shot performance in national tournaments. This indicates that there is no evidence for actual deterioration in shot performance between the first and the last shots in national tournaments. Even though the last shots were the decisive shots, the results cannot find enough support that choking under pressure occurs in these stages of the shootout.

To further examine whether the round of tournament negatively relates to the shot performance in national tournaments, the second hypothesis was created. However, also this second hypothesis did not find enough evidence for support. This indicates that the shootouts do not differ in the different stages in the tournament. Even though the final stages could result in more choking under pressure, this was not the case.

After assuming that the first and the second hypotheses could potentially show a decline in performance, the third hypothesis stated that the negative relationship between kick number and shot performance is moderated by round of tournament, such that this

relationship was stronger for later rounds in the tournament. Interestingly, the findings show support in favor of this hypothesis. Initially, there was insignificant difference in kick number outcomes for the round of 16. However, when comparing this with the semifinals and the finals combined, there was a significant decrease in performance after kick number 5. And since kick number 6 to 17 are considered to be the ‘sudden death’ stage, this finding is in line with the pressure experiments of Baumeister (1984). Pressure coming from self-presentational concerns arise as a result of the increasing amount of observers in further tournament rounds. Moreover, pressure coming from monetary incentives plays a role, as the further a team gets in a tournament, the more external rewards are to be earned. So choking under pressure does also occur in national tournaments.

Furthermore, to rule out alternative explanations, the control variables age, position and valence, showed no significant direct effect on shot performance in national tournaments. However, when comparing the round of 16 with the semifinal and final, there is a significant difference in performance in favor of forward players, who scored significantly better than

(22)

22

defenders. So, there is an assumption that the fifth shot were mostly taken by forward players, who have the highest scoring ability.

To continue, this is also in line with the correlations table discussed earlier. Defenders show a stronger correlation to kick 6 to 17, whereas forward players show a significant negative correlation with kick 6 to 17. Moreover, the forward players show a slightly higher correlation to outcome than the other position groups. On top of that, there is significant positive correlation between the forward players and kick number 5. This suggests that the forward players are chosen to take the (potentially) last shot in the shootout, with the idea of securing the victory.

The findings for the first hypothesis were not in line with most of the previous research done on this topic. This research was initially designed to trivialize and to build on the results of Jordet et al. (2007). The difference in outcome in this study and the study of Jordet et al. (2007) may be explained by the fact that Jordet et al. (2007) only studied penalty shootouts from 1976 until 2004. Therefore, the international tournaments (Word Cup,

European Championship and Copa America) from 2004 onwards were not taken into account. Although they were in an international context, this research is based on the recent 12 years of national shootouts.

One thing in line with Jordet et al. (2007) is that kick number five actually performed better than all other kick numbers. However, they also found direct evidence for kick order deterioration in international tournaments, whereas this research did not found this in national tournaments. They also found that shot performance was significantly worse in the most prestigious tournament (World Cup) compared to the smaller intra-continental tournaments (European Championship and Copa America). The second hypothesis was based on this finding, but did not find a direct performance decline for the higher rounds of a national tournament.

Furthermore, the valence theory from Jordet & Hartman (2008), which stated that shots labeled as positive valence shots scored significantly better that shots labeled as negative valence negatively, did not find any evidence in this research. Moreover, the shots labeled as negative valence performed actually better than the shots labeled as positive valence. Jordet & Hartman (2008) based their findings on international tournaments and can therefore not be generalized to national tournaments. So, this research did not find evidence for the valence theory in a national context. Jordet & Hartman (2008) based their conclusions

(23)

23

on the most prestigious tournaments, where the players might perform under constant high pressure circumstances in every tournament round. Because the situational factors differ in the national tournaments, the valence theory cannot be rejected.

Alternative explanations

Even though this research consisted of a lot more observations that the Jordet et al. (2007) paper, it could not predict direct causality of the different variables on shot performance in the national tournament. However, it seems like the round of tournament has a moderating effect on the relationship between the variables and the outcome. This could mean that there is choking under pressure in international tournaments in such a way that with referring to national tournaments, the pressure is perceived as higher. In this way the choking occurs less in national tournaments and explains the findings of this research, such that choking only occurs in the most intense rounds of tournament (semifinals and finals) and not for the quarter finals and round of 16s. Whereas, on the international level, the level of pressure is higher and more constant.

Kick number 5 being scored the most may refer to soccer trainers choosing their best player to take the (potentially) decisive fifth kick. This assumption was not only made by Jordet et al. (2007) but applied by McGarry& Franks (2000) too. However, the little

differences in shot performance between the kick numbers may be the most logical result of teams that already practiced before the game, keeping in mind that a cup match could potentially result in a shootout.

Another alternative explanation of the insignificant results may be that the quality of the goalkeeper is lower in some national tournaments compared to others, which could eventually explain the insignificant shot performance differences. Moreover, this could also explain why the conversion on the World Cup is the lowest. World’s best goalkeepers are performing at that stage.

Furthermore, this research suggests that control variables like position and age are important to take into the multivariate analyses, as Jordet et al. (2007) did not. In this way, the results of this research may be more accurate and deviate from the results of Jordet et al. (2007)

(24)

24 Limitations

Although the data is collected in a time span of 12 years and the data counts enough

observations to draw conclusion, the research is still a study in one context. It only included 17 different European soccer tournaments. Furthermore, in the utopian situation where all variables are perfectly measured, there is still a chance the calculated outcome of a penalty kick will not occur. This is the consequence of the error term, which contains many forms of good and bad luck such as still scoring badly shot penalty kicks or missing a penalty even though the goalkeeper chose a different side to dive (Chiappori, Levitt & Groseclose, 2002). This is the fundamental argument for saying that the outcome of a penalty kick is the same as the outcome of a lottery ticket. This is not true, although luck can have an inevitable role.

Variables like the quality of the pitch and the weather circumstances will eventually also contribute to the results of a penalty kick and had not been taken into account in this research.

The quality of the goalkeeper has also not been taken into account. Baumann, Friehe & Wedow (2011) argued that the quality of the goalie as well as the quality of the player is of importance for the outcome of the penalty kick. Although the data consisted of 2179

observations, this could have had a significant effect on the outcome in this research because the longest shootout was 34 shots in the Greek Cup. One could assume that the goalkeepers who are active in the Greece league are qualitatively worse than goalkeepers playing in the Spanish league. Moreover, this research is based on the overall performance of the 17 different cups and did not consider the different performances of the different countries.

To add, this research controlled the experience and quality of the player to his age and position, assuming that older players, as well as players in position ‘forward’ would perform better than other groups, which was only the case when comparing the different tournament rounds. Another way of defining the quality is defining the actual market worth of the player. And moreover, compare the market worth of the whole team and consider the ‘better’ team. On the one hand, one could assume that the qualitative better team scores higher in the overall performance during the shootout. On the other hand, qualitative worse teams could be more satisfied when keeping the score tied and be more motivated to go into the shootout. This situation could thus also influence the shot performance.

Another limitation in this research is the absence of considering the home and away team. A supportive home crowd for example, can negatively influence the performance of a

(25)

25

sport team. This because supportive audience create an unintended form of pressure which can consequently disrupt the automatic process of performing a specific task (Wallace, Baumeister & Vohs, 2005; Dohmen, 2008). So a supportive home crowd could have a strengthening impact on the kick order deterioration for this research. On the contrary, professional athletes tend to have more success at their home stadium. Some plausible explanations are referee biases, travel time and motivational factors (Courneya & Carron; 1992, Nevill & Holder, 1999). These arguments suggest that there are more factors that could influence shot performance in shootouts.

In addition to the home crowd example, the stadium capacity could reinforce the relationship between home crowd and shot performance. Baumeister (1984) already argued that self-presentational concerns caused by observers increases pressure. Most spectators are obviously supporting the home team. Building on the arguments of Dohmen (2008), this suggests that the more spectators are in the stadium, the higher the potential choking for decisive kick number or in decisive rounds of tournament. Moreover, some shootouts were taken on neither a home, nor an away ground, but on neutral ground. This research only focuses on the outcome and did not control for home, away or neutral shootouts.

Jordet, Hartman & Vuijk (2012) found evidence that for teams who lost the previous shootout in a tournament, scored consequently worse and teams who won a previous shootout scored significantly better. This research did not control for this assumption.

Final limitation of this research is the level of fatigue. Penalty shootouts always take place after 120 minutes (90 minutes of regular time plus 30 minutes of extra time). Rahnama et al. (2003) demonstrated that exercise simulating the work rate reduces the capability of muscles to produce force. Furthermore, Mohr et al. (2004) showed that players perform less high-intensity running and sprinting in the second than in the first half of extra time. Probably because of a lower muscle temperature. The physiological changes may influence the

neuromuscular coordination and the required shooting skills for a successful penalty kick (Masuda et al., 2005). Cian et al. (2000) also provided evidence that prolonged exercise of two hours leads to a decline in several cognitive functions. So, the level of fatigue might have explained the different shot performances of the players.

(26)

26 Future research

First of all, the multivariate logistic regression models only explained 0.4% to 0.5% of the variance in the outcome of a penalty kick. So, to minimize the error term and to explain more percent of the variance in the outcome, future research could include the limitations discussed earlier and examine whether the findings differ from the findings in this research.

Second, future research could investigate whether the moderation effect of round of tournament is even stronger in the international context. In the national context only the semifinals and finals showed significant kick order deterioration.

Third, this research suggests that forward players score more in the decisive stages of the tournament. Future research could investigate what would be the optimal line up for a shootout in national tournaments and also for international tournaments. It could be that these findings deviate.

Lastly, to make the findings in this research even more generalizable, it is recommended to include shootout data from other continents of the world.

Practical implications

There is no direct kick order deterioration in the 17 national tournaments considered in this study. But this research showed that the choking is significantly higher in later rounds of the tournament. Furthermore, there is evidence that the forward players perform better under these circumstances than midfielders or defenders do. So, the results in this study are still useful for practitioners. This study thus suggests the arguments issued by McGarry & Franks (2000). That is that players should be ranked a priori on their penalty taking ability. So, the fifth best player should take the first penalty, the fourth best player should take the second penalty, and so on.

Furthermore, it is important for soccer trainers to not consider a shootouts as a lottery. It would be very useful to train shootouts. For example, soccer trainers still need to be aware of how to cope with stressors that may negatively influence shot performance. Thus, this study highly recommends to practice the shootouts in the actual circumstances as if it is an actual match so that players learn how to cope with stressors.

(27)

27 Conclusion

Previous research on kick order deterioration in international tournaments suggests that the more decisive a penalty kick is, the higher the chance of choking under pressure, resulting in lower actual shot performance. If choking under pressure occurs in the international context, then it could also occur in the national context. The purpose of this study was thus to

investigate whether the conclusions based on the international tournaments are generalizable to national tournaments.

The research question was: Does the kick order deterioration relationship, explained by Jordet et al. (2007), also occur in national tournaments where the perceived importance of the outcome is less than in international tournaments? The conclusions from Jordet et al. (2007) were based on a small sample (n = 409), whereas the conclusions of this study are founded on a much larger dataset (n = 2179). There is evidence to conclude that the kick order deterioration is not generalizable to the national tournaments. Instead, this research suggests that the pressure in national tournaments is lower than international tournament by proving that there is no choking under pressure in the initial knockout stages (round of 16 and quarter finals). For later, more decisive rounds choking actually does occur, as pressures related to the outcome increase.

Only the shots in most decisive rounds of tournament show that there is a significant difference in performance between the decisive and indecisive shots. So, choking under pressure also occurs in national tournaments.

Coming back to the start of this research: Still, the goal is 2.44 meters high and 7.32 meters wide, the goal keeper is not allowed to stand in front of the goal line and the ball reaches the goal in approximately 0.2 seconds. However, in the situation of taking a penalty in national (semi)finals from the sixth kick onwards, scoring is definitely not obvious. This study shows that kick order deterioration occurs significantly in such situation.

(28)

28

Reference list

Baumann, F., Friehe, T. & Wedow, M. (2011). General ability and specialization: Evidence from penalty kicks in soccer. Journal of Sports Economics, 12(1), 81-105.

Baumeister, R. F. (1984). Choking Under Pressure: Self-Consciousness and Paradoxical Effects of Incentives on Skillful Performance. Journal of Personality and Social Psychology, 46(3), 610- 620.

Beilock, S. L. & Carr, T. H. (2001). On the Fragility of Skilled Performance: What Governs Choking Under Pressure? Journal of Experimental Psychology: General, 130(4), 701-725.

Beilock, S. L., & Gray, R. (2007). Why do athletes choke under pressure? In G. Tenenbaum, & R. C. Eklund (Eds.), Handbook of sport psychology, 425–444.

Chiappori, P. A., Levitt, S. & Groseclose, T. (2002). Testing mixed-strategy equilibria, when players are heterogeneous: The case of penalty kicks in soccer. American Economic Review, 1138-1151.

Cian, C. J., Koulmann, N., Barraud, P. A., Raphael, C., Jimenez, C. & Melin, B. (2000). Influence of variations in body hydration on cognitive functioning: Effect of hyperhydration, heat, stress and exercise-induced dehydration. Journal of Psychophysiology, 14, 29-36.

Courneya, K. S. & Carron, A. V. (1992). The home advantage in sports competitions: A Literature review. Journal of Sport and Exercise Psychology, 14(1), 13-27.

Dohmen, T. J. (2008). Do professionals choke under pressure? Journal of Economic Behavior & Organization, 65(3), 636-653.

Jordet, G., & Elferink-Gemser, M. (2012). Stress, Coping, and Emotions on the World Stage: The Experience of Participating in a Major Soccer Tournament Penalty Shootout.

Journal of Applied Sport Psychology, 24(1), 73-91.

Jordet, G. & Hartman, E. (2008). Avoidance motivation and choking under pressure in soccer penalty shootouts. Journal of Sport & Exercise Psychology, 30, 450-457. Jordet, G., Hartman, E., Visscher, C. & Lemmink, K. A. P. M. (2007). Kicks from the penalty

mark in soccer: The roles of stress, skill, and fatigue for kick outcomes. Journal of Sport Sciences, 25(2), 121-129.

Jordet, G., Hartman, E. & Vuijk, P. J. (2012). Team history and choking under pressure in major soccer penalty shootouts. British Journal of Psychology, 103, 268-283.

(29)

29

Masuda, K., Kikuhara, K., Demura, S., Katsuta, S. & Yamanaka, K. (2005). Relationship between muscle strength in various isokinetic movements and kick performance among soccer players. Journal of Sports Medicine and Physical Fitness, 45, 44-52. McGarry, T. & Franks, I. M. (2000). On winning the penalty shootout in soccer. Journal of

Sports Sciences, 18(6), 401-409.

Mohr, M., Krustrup, P. & Bangsbo, J. (2004). Muscle temperature and sprint performance during soccer matches – beneficial effects of re-warm up at half time. Scandinavian journal of Medicine and Science in Sports, 15, 136-143.

Nevill, A. M. & Holder, R. L. (1999). Home advantage in sport. Sports Medicine, 28(4), 221-236.

Rahnama, N., Reilly, T., Lees, A. & Graham-Smith, P. (2003). Muscle fatigue induced by exercise simulating the work rate of competitive soccer. Journal of Sports Sciences, 21, 933-942.

The International Football Association Board [IFAB]. (2019). Laws of the game 2018/2019. Zurich: FIFA.

Wallace, H. M., Baumeister, R. F. & Vohs, K. D. (2005). Audience support and choking under pressure: A home advantage? Journal of sports sciences, 23(4), 429-438. Wells, B. M. & Skowronski, J. J. (2012). Evidence of Choking Under Pressure on the PGA

(30)

30

Appendix

*step 1= all variables come directly out of Excel. Actions in Excel: Kick number 6 to 17 were combined, Round of tournament is divided in the four categories, Positions were being

combined, Valence was being combined, age was divided in three groups and the -99 have been taken out of the dataset.

*step 2 = recoding age in Young, Medium and Old. I previously called this Age_group, but I cannot found the syntax. So, I did it again with Age_group_2. Please be aware that I used Age_Group throughout my analysis.

DATASET ACTIVATE DataSet1.

RECODE age (32 thru Highest=3) (Lowest thru 23=1) (24 thru 31=2) INTO Age_group_2. EXECUTE.

*Step 3= Frequencies for all variables DATASET ACTIVATE DataSet1.

FREQUENCIES VARIABLES=outcome kicknumber Tournament_Round age_group Position NV PV NV_and_PV

/STATISTICS=STDDEV MEAN SUM /ORDER=ANALYSIS.

*Step 4= Descriptives for all variables

DESCRIPTIVES VARIABLES=outcome kicknumber Tournament_Round age_group Position NV PV NV_and_PV

/STATISTICS=MEAN SUM STDDEV MIN MAX.

*step 5 = Chart builder - graph of kick order on outcome

GGRAPH

/GRAPHDATASET NAME="graphdataset" VARIABLES=kicknumber MEAN(outcome)[name="MEAN_outcome"]

MISSING=LISTWISE REPORTMISSING=NO /GRAPHSPEC SOURCE=INLINE.

(31)

31

BEGIN GPL

SOURCE: s=userSource(id("graphdataset"))

DATA: kicknumber=col(source(s), name("kicknumber"), unit.category()) DATA: MEAN_outcome=col(source(s), name("MEAN_outcome")) GUIDE: axis(dim(1), label("kicknumber"))

GUIDE: axis(dim(2), label("Mean outcome"))

GUIDE: text.title(label("Simple Bar Mean of outcome by kicknumber")) SCALE: cat(dim(1), include("1", "2", "3", "4", "5", "6"))

SCALE: linear(dim(2), include(0))

ELEMENT: interval(position(kicknumber*MEAN_outcome), shape.interior(shape.square)) END GPL.

*Step 6 = Chart builder- graph of tournament round on outcome

GGRAPH

/GRAPHDATASET NAME="graphdataset" VARIABLES=Tournament_Round MEAN(outcome)[name="MEAN_outcome"]

MISSING=LISTWISE REPORTMISSING=NO /GRAPHSPEC SOURCE=INLINE.

BEGIN GPL

DATA: Tournament_Round=col(source(s), name("Tournament_Round"), unit.category()) DATA: MEAN_outcome=col(source(s), name("MEAN_outcome"))

GUIDE: axis(dim(1), label("Tournament_Round")) GUIDE: axis(dim(2), label("Mean outcome"))

GUIDE: text.title(label("Simple Bar Mean of outcome by Tournament_Round")) SCALE: cat(dim(1), include("1", "2", "3", "4"))

SCALE: linear(dim(2), include(0))

ELEMENT: interval(position(Tournament_Round*MEAN_outcome), shape.interior(shape.square))

(32)

32

*Step 7 = Chart builder - 3D graph of tournament round and kick number on outcome (not a clear graph, so eventually not used)

GGRAPH

/GRAPHDATASET NAME="graphdataset" VARIABLES=kicknumber MEAN(outcome)[name="MEAN_outcome"]

Tournament_Round MISSING=LISTWISE REPORTMISSING=NO /GRAPHSPEC SOURCE=INLINE.

BEGIN GPL

DATA: kicknumber=col(source(s), name("kicknumber"), unit.category()) DATA: MEAN_outcome=col(source(s), name("MEAN_outcome"))

DATA: Tournament_Round=col(source(s), name("Tournament_Round"), unit.category()) COORD: rect(dim(1,2,3))

GUIDE: axis(dim(1), label("Tournament_Round")) GUIDE: axis(dim(2), label("kicknumber"))

GUIDE: axis(dim(3), label("Mean outcome"))

GUIDE: text.title(label("Simple 3-D Bar Mean of outcome by kicknumber by Tournament_Round"))

SCALE: cat(dim(2), include("1", "2", "3", "4", "5", "6")) SCALE: linear(dim(3), include(0))

SCALE: cat(dim(1), include("1", "2", "3", "4"))

ELEMENT: interval(position(Tournament_Round*kicknumber*MEAN_outcome), shape.interior(shape.square))

END GPL.

*Step 8 = comparing the means of all variables

MEANS TABLES=outcome BY kicknumber Tournament_Round age_group Position NV PV NV_and_PV

(33)

33

*Step 9= Make correlation table for all seperate variables

CORRELATIONS

/VARIABLES=outcome Kick_1 Kick_2 Kick_3 Kick_4 Kick_5 Kick_6_to_17 last_16 quarter semi final

defender midfielder forward NV PV No_Valence Age_17_to_23 Age_24_to_31 Age_32_to_48

/PRINT=TWOTAIL NOSIG /STATISTICS DESCRIPTIVES /MISSING=PAIRWISE.

*Step 10= binary logistic regression, only categorical variable is age group (young, medium old), others are all 1 or 0 (no significance)

LOGISTIC REGRESSION VARIABLES outcome

/METHOD=ENTER Kick_1 Kick_2 Kick_3 Kick_4 Kick_5 Kick_6_to_17 last_16 quarter semi final

age_group defender midfielder forward NV PV No_Valence /CONTRAST (age_group)=Indicator

/CRITERIA=PIN(.05) POUT(.10) ITERATE(20) CUT(.5).

*Step 11= binary logistic regression, all categorical (no significance) (1st multivariate

analysis)

/METHOD=ENTER kicknumber Tournament_Round age_group Position NV_and_PV /CONTRAST (kicknumber)=Indicator /CONTRAST (Tournament_Round)=Indicator(1) /CONTRAST (age_group)=Indicator(1) /CONTRAST (Position)=Indicator(1) /CONTRAST (NV_and_PV)=Indicator(1) /CLASSPLOT /PRINT=GOODFIT ITER(1)

(34)

34

*Step 12: Binary logistic regression of all variables apart to see difference in ratios (with reference groups)

LOGISTIC REGRESSION VARIABLES outcome /METHOD=ENTER kicknumber

/CONTRAST (kicknumber)=Indicator /CLASSPLOT

/PRINT=ITER(1) CI(95)

/CRITERIA=PIN(0.05) POUT(0.10) ITERATE(20) CUT(0.5).

LOGISTIC REGRESSION VARIABLES outcome /METHOD=ENTER Tournament_Round

/CONTRAST (Tournament_Round)=Indicator(1) /CLASSPLOT

LOGISTIC REGRESSION VARIABLES outcome /METHOD=ENTER age_group

/CONTRAST (age_group)=Indicator(1) /CLASSPLOT

LOGISTIC REGRESSION VARIABLES outcome /METHOD=ENTER Position

/CONTRAST (Position)=Indicator(1) /CLASSPLOT

LOGISTIC REGRESSION VARIABLES outcome /METHOD=ENTER NV_and_PV

(35)

35

/CLASSPLOT

*Step 13= split files for the different rounds

SORT CASES BY last_16.

SPLIT FILE LAYERED BY last_16.

*Step 14= Compare means of different kick numbers

MEANS TABLES=outcome BY Kick_1 Kick_2 Kick_3 Kick_4 Kick_5 Kick_6_to_17 /CELLS=MEAN COUNT STDDEV SUM.

*Step 15= I continued doing this for all rounds of tournaments to examine potential increasing kick order deterioration

SORT CASES BY quarter.

SPLIT FILE LAYERED BY quarter.

SORT CASES BY semi.

SPLIT FILE LAYERED BY semi.

SORT CASES BY final.

SPLIT FILE LAYERED BY final.

(36)

36

*Step 16= split files again for round of 16

SORT CASES BY last_16.

SPLIT FILE LAYERED BY last_16.

*Step 17= Select cases: Quarter = not 1 (for comparing the round of 16 with the semi and final)

USE ALL.

COMPUTE filter_$=(quarter = not 1).

VARIABLE LABELS filter_$ 'quarter = not 1 (FILTER)'. VALUE LABELS filter_$ 0 'Not Selected' 1 'Selected'. FORMATS filter_$ (f1.0).

FILTER BY filter_$. EXECUTE.

*Step 18= binary logistic regression (2nd multivariate analysis)

/METHOD=ENTER kicknumber age_group Position NV_and_PV /CONTRAST (kicknumber)=Indicator

/CONTRAST (age_group)=Indicator(1) /CONTRAST (Position)=Indicator(1) /CONTRAST (NV_and_PV)=Indicator(1) /CLASSPLOT

/PRINT=GOODFIT ITER(1) CI(95)

Choking under pressure: The comparison of kick order deterioration between soccer penalty shootouts of international tournaments and national tournaments