Documente Academic
Documente Profesional
Documente Cultură
CITATION
Woods, W. C., Hayes, D. J., Meyer, F., Kardan, O., & Berman, M. G. (2016, July 21). Dynamic
Effects on Elite and Amateur Performance. Sport, Exercise, and Performance Psychology.
Advance online publication. http://dx.doi.org/10.1037/spy0000070
Sport, Exercise, and Performance Psychology © 2016 American Psychological Association
2016, Vol. 5, No. 3, 000 2157-3905/16/$12.00 http://dx.doi.org/10.1037/spy0000070
have examined intrapersonal variation in psychological processes, there has been little
This document is copyrighted by the American Psychological Association or one of its allied publishers.
exploration into variation in the behaviors that comprise athletic performance itself.
Understanding the extent to which performance behaviors change as a function of
opponent and time is critical to understanding how fluctuations in psychological
processes may influence performance. Three studies examined the influence of dy-
namic, within-person variation on athletic performance. In Study 1, we detected dynamic
effects on tennis players’ serving speed. In Study 2, we showed similar dynamic effects in
user inputs by elite e-sport players. In Study 3, we replicated the dynamic effects on
performance in a sample of amateur e-sport players. Findings provided preliminary evi-
dence that dynamic variability in performance represents a significant proportion of
variance.
Measurement of athletic performance is an tions (Shavelson & Webb, 1991). There is,
area of special interest within sports psychol- however, a growing body of recent research that
ogy. Many investigations have explored the ex- has utilized dynamic, within-person study de-
tent to which trait-like, between-person differ- signs to explore how change over time and
ences predict divergent performance outcomes situation is related to performance outcome in
(Gaudreau, Nicholls, & Levy, 2010). These sport. To date, however, these within-person
studies typically examine how individuals with investigations of dynamic effects have focused
one characteristic differ from individuals with on the extent to which variations in psycholog-
another characteristic. For example, an investi- ical processes predict performance outcome.
gator may be interested in performance differ- Understanding the extent to which athletic per-
ences between men and women, or between formance itself is malleable across time and
athletes with high anxiety and low anxiety. opponents may provide insight into the how
Trait-level research typically obtains only one intrapersonal shifts in psychological processes
measurement for each individual and is unable affect performance outcome in sport.
to account for variations in individual affect, Trait-level research in sports psychology has
thought, and action over time and across situa-
been successful in identifying intrapersonal, in-
terpersonal, and situational characteristics that
predict superior athletic performance. For ex-
ample, investigations have described how tes-
William C. Woods, Daniel J. Hayes, Francisco Meyer, tosterone (Aguilar, Jiménez, & Alvero-Crus,
Omid Kardan, and Marc G. Berman, Department of Psy- 2013) and cortisol (Filaire, Alix, Ferrand, &
chology, The University of Chicago.
The authors thank Brian Lakey for methodological assis- Verger, 2009) levels change before and after
tance. intense competition. Beyond physiology, ath-
Correspondence concerning this article should be ad- letes with strong feelings of precompetition
dressed to William C. Woods or Marc G. Berman, Depart-
ment of Psychology, University of Chicago, 416 Green
vigor are likely to have superior performance
Hall, Chicago, IL 60637. E-mail: willcwoods@gmail.com compared with those with less vigor (Beedie,
or bermanm@uchicago.edu Terry, & Lane, 2000). Situational effects on
1
2 WOODS, HAYES, MEYER, KARDAN, AND BERMAN
performance have also been documented such one half a standard deviation unit better than
as the lower anxiety levels of team-sport ath- those who were not (Jokela & Hanin, 1999).
letes relative to those in individual sports (Mar- More recently, Hagtvet and Hanin (2007) ex-
tens, Burton, Vealey, Bump, & Smith, 1990). amined the consistency of athlete emotion pro-
Although trait-level research has documented files across strong and poor performances and
intrapersonal, interpersonal, and situational ef- reported a consistent profile for successful per-
fects on performance, there are conflicting find- formances but no pattern within unsuccessful
ings that highlight the limitations of between- performances.
person research. For example, anxiety may be Similarly, a study of elite youth soccer play-
beneficial to some athletes, but deleterious to ers reported low interplayer agreement on the
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
others (Hanin, 1997). Similarly, Anshel and supportiveness of coaches (Rees, Freeman,
This document is copyrighted by the American Psychological Association or one of its allied publishers.
Wells (2000) showed that the overall use of Bell, & Bunney, 2012). Across three studies, the
coping is not as effective a predictor of perfor- authors reported that idiosyncratic personal
mance outcomes as the specific type of coping taste explained more of the variance in judg-
strategy deployed. These inconsistencies sug- ments of coaches’ supportiveness than did a
gest that the influence of typically characterized trait-like tendency of athletes to see coaches as
traits may be more situationally dependent than more or less supportive. Furthermore, these id-
previously theorized. If some athletes thrive at iosyncratic perceptions of supportiveness ex-
higher levels of anxiety, while others do not, tend beyond an athlete’s tendency to view all
comparisons of the effect of relative levels of coaches as supportive. A recent replication by
anxiety will not be meaningful. In order for Coussens, Rees, and Freeman (2015) reported
trait-like differences to be robust, relative levels similarly strong dynamic variability in support-
of a characteristic need to have similar effects iveness judgments across two studies. Athletes
across individuals (i.e., main effect). showed little agreement as to which coaches
When a characteristic’s effect is not consis- were supportive or agreeable. Most of the vari-
tent across individuals, it may be beneficial to ation was accounted for by idiosyncratic rela-
examine how different levels of that character- tionships (i.e., relational influences) between
istic afect the same individuals over time, for athletes and coaches. Furthermore, the rela-
example, how shifts in an athlete’s anxiety are tional aspects of support were correlated with
linked to changes in outcome performance. This hypothesized antecedents of perceived support.
style of research typically utilizes within-person Athletes who viewed a coach as unusually
designs to detect dynamic processes in affect, agreeable and competent also tended to view
thought, and action. that same coach as unusually supportive. These
Recently, within-person research designs studies provide evidence that psychological
have been deployed to understand how changes processes in sport are subject to dynamic pro-
in an athlete’s psychological processes can af- cess variance.
fect performance. One of the most cited models In addition, studies of dynamic shifts in psy-
is the individual zones of optimal functioning chological processes have been linked to per-
(IZOF; Hanin, 1997). IZOF combines dynamic formance outcome. Gaudreau, Nicholls, and
and trait-level research to understand how pos- Levy (2010) explored the extent to which dy-
itive and negative emotions influence perfor- namic variation in coping predicted perfor-
mance. In applied settings, sport psychologists mance outcomes in golfers. When controlling
develop an athlete’s unique profile of optimal for trait-level differences in athletic ability, dy-
emotion using positivity–negativity and opti- namic variation in the coping utilized by the
mality– dysfunctionality dimensions. For exam- athletes strongly predicted subjective judgments
ple, moderate anxiety may be negative and op- of performance and moderately predicted objec-
timal for one athlete, but negative and tive performance outcomes. In a similar study,
dysfunctional for another. Thus, the closer an Doron and Gaudreau (2014) investigated the
athlete is to being “in the zone” of optimal extent to which prior performance affected psy-
emotion, the better his or her performance. This chological processes, and whether these in turn
effect has been extensively documented with a predicted later outcomes, in elite fencers in a
meta-analysis of over 6,000 athletes, finding simulated competition. Previous performance
that those who were “in the zone” performed at outcomes were predictive of shifts in the psy-
DYNAMIC EFFECTS ON PERFORMANCE 3
chological processes of the athletes, but the the investigation. In other words, dynamic pro-
psychological processes were not predictive of cess designs are interested in how an individual
later outcomes. These studies reflect recent suc- is different from herself in a given characteristic
cesses in applying within-person research de- at one observation relative to another observa-
signs to detect dynamic effects on sports per- tion.
formance. The current investigation used the framework
These dynamic effects investigations are, of the Social Relations Model (SRM; Kenny,
however, somewhat limited. Although shifts in 1994; Kenny, Kashy, & Cook, 2006) to differ-
cognitive processes such as affect and coping entiate stable, between-person and dynamic,
were tracked within each athlete, similar varia- within-person influences on athletic perfor-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
tion in athletic process behaviors was not mea- mance. Trait-like effects can be conceptualized
This document is copyrighted by the American Psychological Association or one of its allied publishers.
sured. It would be useful, for example, to know as the difference in mean scores on a perfor-
the extent to which a professional tennis play- mance variable between athlete A and athlete B,
er’s serving speed varies over time and oppo- for example, the tendency of athlete A to always
nent. Understanding shifts in process perfor- perform better than athlete B. These influences
mance behaviors would allow researchers to will henceforth be referred to as athlete effects.1
differentiate shifts in outcome due to dynamic Dynamic variance can come from two sourc-
psychological processes from shifts due to vari- es: opponents and time. The social sources are
ation in performance. To date, however, dy- measured as Opponent effects and Athlete ⫻
namic variation in process-related performance Opponent interactions. Opponent effects reflect
behavior has not been explored systematically. the extent to which a given opponent tends to
Therefore, the aim of the current investigation elicit the same change in performance by all
was to identify the extent to which process-
athletes.2 Athlete ⫻ Opponent interactions re-
related athletic behaviors vary across opponents
flect idiosyncratic differences in performance in
and over time. Study 1 assessed the variation in
particular athlete– opponent pairs (i.e., rela-
a single process-related behavior in elite tennis
tional influences). For example, whereas oppo-
players at an international event. Study 2 ex-
nent E may elicit a strong change in process
plored variation in all possible process-related
behaviors in world-class e-sport players in a behaviors in athlete A, the same opponent may
global tournament. Finally, in order to test not elicit any shift in athlete B. When athletes
whether our findings generalize beyond elite do not share the same opponents, Opponent
performers, Study 3 features a sample of ama- effects and Athlete ⫻ Opponent interactions are
teur e-sport players. confounded into a single effect (i.e., competi-
tion effect). Thus, competition effects reflect the
General Methodological Considerations extent to which some opponents elicit greater
(or worse) performance than others.
Measurement of dynamic, within-person In addition to the influence of opponents on
variation of behavior is different from investi- variation in process performance, we were also
gations focused on trait-like, between-person interested in changes in process behavior over
differences, because dynamic effects research time. In tennis, a set is composed of up to 12
focuses on what trait-level research conceptual- games, and is won when one player wins 6
izes as measurement error (Shavelson & Webb, games with a 2-game advantage. (Note: If play-
1991). Trait-level research divides participants ers are tied at 6 games all, the set is decided by
into groups based upon different levels of a a tiebreaker.) In electronic sport, the terms “set”
common characteristic, such as anxiety level. In and “game” both refer to a single game session.
this design, differences within groups are essen- In both professional tennis and e-sport, a match
tially error; ideally, group members would share consists of several sets, and is usually won on a
identical levels of anxiety in order to more “best of” basis (e.g., best of three sets). Addi-
precisely examine the role of high or low anx-
iety in predicting outcome. In dynamic process 1
Athlete effects are identical to perceiver effects in the
designs, each participant is treated as her own SRM.
group of observations, and shifts in a given 2
Opponent effects are identical to target effects in the
characteristic across situations are the focus of SRM.
4 WOODS, HAYES, MEYER, KARDAN, AND BERMAN
tionally, because sets are crossed with athletes, assigned to dynamic influences, reducing the
we are able to measure the main effect of set on proportion of variance due to between-person
all athletes and the Athlete ⫻ Set interactions, athlete effects.
which are the extent to which particular sets
influence the performance of particular athletes.
Study 1
For example, an interaction would occur if the
third set in a match always brought out stronger Study 1 examined the influence of trait-like
performance in athlete A, but did not affect the and dynamic variability on athletic performance
performance of athlete B. in a sample of elite tennis players.
The effects of time can also be present within
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
style of match for women and best-of-five style lete ⫻ Set effect, the highest order interaction,
for men. We could have chosen sets at random served as the error term (Kenny, 1994; Kenny et
within a match when an athlete had played in al., 2006). An effect is significant when its
more than two sets. However, this would have lower bound 95% confidence interval does not
resulted in sets not being fully crossed with include 0. Effects are significantly different
athletes, which would have confounded the from each other when their 95% confidence
main effect of set with the Athlete ⫻ Set inter- intervals do not overlap.
action. To maximize our chances of detecting
all possible influences on performance, we Results and Discussion
elected to analyze the first two sets for each
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
match for all subjects. The goal of Study 1 was to estimate the
This document is copyrighted by the American Psychological Association or one of its allied publishers.
Statistical analyses. Dynamic variability extent to which professional tennis players’ per-
was calculated using the VARCOMP procedure formance varies over time and opponent, re-
in SPSS (Version 22.0). Because our opponents
flecting dynamic variability, and the relative
were not fully crossed with athletes, we exam-
magnitude of dynamic and trait-like influences.
ined the influence of competition effects on
process behaviors (e.g., Lakey, Orehek, Hain, & Variance component estimation (see Table 1)
VanVleet, 2010; Lakey & Tanner, 2013). Al- found a significant athlete effect, which ex-
though fully crossed designs allow for more plained the majority of the variance in serving
thorough variance partitioning, few athletic speed. In other words, some tennis players tend
tournaments are structured such that all athletes to always have faster serves than others. In
play each other, and we determined that the addition, we also detected significant dynamic
high ecological validity of these real-world influences on serving speed. The significant
tournaments compensated for the limitations of competition effects indicated that the serving
the study design. The data were structured as a speed of tennis players was influenced to some
one-with-many design (Kenny et al., 2006), extent by their opponents. The Athlete ⫻ Set
with Opponents nested within Athletes ⫻ Set. effect was not significant, indicating that serv-
Each athlete formed a level of the random Ath- ing speed did not change significantly over the
letes factor; each opponent formed a level of the course of sets. The athlete effect was signifi-
random Opponents3 factor; each set formed a cantly larger than the competition effect (p ⬍
level of the random Set factor. Sets represented .05).
the replication factor. This design produced five The large athlete effects reflected the extent
effects: (a) athlete, (b) opponent4 nested within to which some athletes consistently serve at
athlete, (c) set, (d) Athlete ⫻ Set, and (e) [Op- higher speeds than others. Thus, most of the
ponent nested within athlete] ⫻ Set. Athlete variance in serving speed was due to differences
effects reflect stable, trait-like differences in between players, rather than the differences in
performance. Opponent nested within athlete an athlete’s performance from one opponent to
effects (i.e., competition effects) represent vari-
the next. For example, Athlete A in our sample
ation in performance across opponents. The
might consistently serve at 150 KPH, and her
Athlete ⫻ Set effect represents the interaction
of sets with each athlete’s performance. For the serve tends to vary about 5 KPH from one
purposes of this study, opponent nested within
athlete, and Athlete ⫻ Set, serve as indicators of
dynamic variation. The set effect does not rep- 3
Within our design, matches are completely confounded
resent between-person or within-person influ- within opponent. Although ideally all effects could be mea-
sured in a generalizability design, this confounding reflects
ences on performance, and was nonsignificant, a real-world reality within sport: One’s opponent does not
and is thus excluded from discussion for brev- typically change during the course of a single match.
ity. Separate VARCOMPs were performed for 4
Because opponents are not fully crossed in a one-with-
males and females to check for gender effects. many design, effects due to opponent and effects due to
The results for both groups were identical, and Athlete ⫻ Opponent interaction are confounded in a single
effect (i.e., competition effect). Different study designs,
thus we proceeded with a single, mixed sample. such as round robin style designs, allow the researcher to
This design yields only one observation per cell. distinguish opponent effects from Athlete ⫻ Opponent in-
Consequently, the opponent nested within Ath- teractions.
6 WOODS, HAYES, MEYER, KARDAN, AND BERMAN
Table 1
Variance Components, Confidence Intervals, and Proportion of Variance
Explained in Study 1
Variance 95% confidence Proportion of
Serving speed component interval variance
Between-person
Athlete 234.63 153.88–315.38 .86ⴱ
Within-person
Competition 22.23 12.44–32.02 .08ⴱ
Athlete ⫻ Set 1.67 ⫺2.09–5.42 .01
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
ⴱ
p ⬍ .05.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
mance in general that may not be as apparent if another strong test of the importance of dy-
tennis were compared with a similar sport. namic variability in performance behaviors.
Thus, although the status of electronic sport in APM is useful because it encapsulates all
comparison with traditional sport is a source of possible performance behaviors, yet it is limited
ongoing controversy (Jonasson & Thiborg, because it also includes unnecessary repetitions
2010; Witkowski, 2012), we believe that com- in keystrokes or mouse clicks as well as mis-
paring the statistical properties of performance takes. Although it may be possible to conduct
between tennis and e-sport professionals will an evaluative screening on APM to count or
provide interesting insights for both domains. exclude poor inputs, we decided against this.
Although some players may inflate their APM,
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
ation in performance as athletes move between mouse inputs over the course of a game than
This document is copyrighted by the American Psychological Association or one of its allied publishers.
opponents. The Athlete ⫻ Set effects are the others. There were also two sources of dynamic
interaction between particular players and par- influences: competition and athlete x set effects.
ticular sets. Finally, the Athlete ⫻ Segment Put another way, the APM of professional e-
effect represents the interaction of the time sport players was influenced in part by the com-
course of matches with each athlete’s perfor- petitor, and also by the set within a match. For
mance. The Set and Segment effects do not example, the first set may pull for unusually
reflect between- or within-person influences on strong APM in some players, but not others.
performance and were nonsignificant, and are The athlete effect explained the majority of the
thus excluded from discussion for the sake of variance in APM, and was significantly larger
brevity. For the purposes of this study, compe- than competition effects and Athlete ⫻ Set ef-
tition effects, Athlete ⫻ Segment, and Ath- fects (p ⬍ .05). The Athlete ⫻ Set interaction
lete ⫻ Set operate as indicators of dynamic was significantly larger than the competition
variation. This design yields only one observa- effect (p ⬍ .05).
tion per cell, and thus the highest order interac- The results of Study 2 represent a partial
tion was the error term (Kenny, 1994; Kenny et replication of Study 1. Trait-like differences
al., 2006). Because we had 100 estimations of accounted for the vast majority of the variance
each effect, we used those estimates to construct in both tennis serving speed and electronic sport
95% confidence intervals around the median actions per minute. In addition, both studies
variance component estimate using the percen- found small but significant dynamic effects on
tile method. An effect is significant when its process performance variables. Nevertheless,
lower bound 95% confidence interval does not the sources of these effects differed. In Study 1,
include 0. Effects are significantly different all of the dynamic variation was due to compe-
when their 95% confidence intervals do not tition effects. For both tennis and StarCraft 2
overlap. players, performance was influenced by their
Table 2
Median Estimates of Variance Components, Confidence Intervals, and
Proportion of Variance Explained in Study 2
Variance component 95% confidence Proportion of
Actions per Minute (median) interval variance
Between-person
Athlete 4409.33 4218.24–4434.86 .77ⴱ
Within-person
Competition 396.58 323.35–428.44 .06ⴱ
Athlete ⫻ Set 544.02 470.78–575.87 .09ⴱ
Athlete ⫻ Segment 33.43 .00–502.42 .01
Note. Negative variance component estimates are constrained to zero by SPSS at the end of
the VARCOMP operation. Because we constructed 95% confidence intervals using these
estimates, the lower bound will always be constrained to zero.
ⴱ
p ⬍ .05.
DYNAMIC EFFECTS ON PERFORMANCE 9
a deliberate choice by players to enact particular dicted that the more experienced players would
This document is copyrighted by the American Psychological Association or one of its allied publishers.
strategies based upon knowledge of their oppo- show stronger athlete effects than the novices.
nents. For example, in trying to exploit his Finding differences in the stable and dynamic
opponent’s perceived weaknesses, player A effects of novices and experts in Study 3 would
may deploy more aggressive strategies against suggest that our earlier results reflected elite
opponent B, while using more conservative skill. On the other hand, if the results of Study
strategies against opponent C. If more aggres- 3 converged with those of Studies 1 and 2, we
sive strategies require frequent, precisely con- would have evidence that the larger athlete ef-
trolled attacks, they may require consistently fects and significant but smaller competition
higher APM to execute than more conservative effects are a part of performance in general. As
strategies. we were particularly interested in how expertise
E-Sport performance was also influenced by impacts the relative magnitudes of athlete and
an interaction between the players and particu- competition effects, we compared results across
lar sets, meaning that the first set may always skill levels.
elicit more APM in some players, but not oth- This study has implications for research on
ers. It is possible that this Athlete ⫻ Set inter- the effects of video game playing on cognitive
action is a statistical representation of varying processes, which has been criticized for violat-
strategy from set to set. For instance, in a typical ing the rules of experimental design by targeting
match Player A may characteristically utilize a experienced video game players (Boot, Blakely,
rapid attack strategy requiring high APM for a & Simons, 2011) and relying on self-reports of
short amount of time in the first set, while using expertise rather than objective metrics (Latham,
a slower strategy requiring a more relaxed APM Patston, & Tippett, 2013). Comparing the find-
over a longer period in the second set, while ings of Study 2’s professional sample with
Player B might engage in a high APM strategy Study 3’s amateur sample may shed light on
in second sets, but not first. This strategy vari- how to define video game experience more
ation by each athlete represents an Athlete ⫻ clearly.
Set interaction.
Study 2 replicated Study 1’s findings that Method
elite competitive performance is primarily a
function of trait-like differences in the speed of Materials. Study 3 examined performance
the performers. However, in both samples, per- in amateur players of StarCraft 2, the e-sport
formance was also dynamically influenced. De- described in Study 2.
spite these compelling findings, Studies 1 and 2 Participants. Our sample included 289
featured professionals, representing only the top amateur StarCraft 2 players. StarCraft 2 features
of the range of ability in their respective do- an online multiplayer matching system that
mains. Although we found common effects be- matches players on the basis of their perfor-
tween both samples, it is unknown whether mance history. Players are divided into seven
large trait-like effects and smaller yet significant leagues on the basis of their skill relative to
dynamic influences are features of elite perfor- other players. For the current study, between 34
mance or of performance in general. Thus, in and 51 players from each league were sampled.
Study 3, we analyzed the performance of ama- Match data were gathered from a publically
teur e-sport players in an attempt to detect dy- accessible StarCraft 2 replay website where
namic influences. players post their match replays and can com-
10 WOODS, HAYES, MEYER, KARDAN, AND BERMAN
pare with other players: ggtracker.net. First, 100 Results and Discussion
replays from each league were chosen at ran-
dom from the database. Matches that were In Studies 1 and 2 we detected large and
shorter than 5 minutes were excluded. If at least stable athlete effects on performance, and
30 players appeared in 2 of the matches chosen smaller, but significant, dynamic influences—
at random, then the drawing was complete. If not knowing if these effects were a property of
fewer than 30 players appeared in 2 matches, performance in general, or an attribute of expert
then another 50 matches were chosen at ran- performance specifically. The goal of Study 3
dom, with care taken not to redraw the same was to determine if these patterns of influences
matches. This process was repeated to select on performance would be apparent across a
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
matches from each league. Because players are range of skill levels, and they were. Across the
This document is copyrighted by the American Psychological Association or one of its allied publishers.
only identified by their in-game handle, demo- entire range of amateur skill levels, there were
graphic information for the selected players is both large athlete effect and smaller, but signif-
unknown. Due to limitations of the database, we icant, dynamic effects, consistent with the first
were only able to get one match per athlete– two studies (see Table 3). Within the dynamic
opponent pair. influences, as in Study 2, a significant portion of
Measures. Our process performance vari- performance was attributable to competition ef-
able, actions per minute, was identical to that fects. Thus, regardless of skill level, some op-
used in Study 2. We used the same segment ponents elicited more APM than others. Repli-
aggregation process as well. cating Studies 1 and 2, the athlete effect was
significantly larger than the competition effect
Statistical analyses. The design of Study 3
(p ⬍ .05). As in Study 2, there were no signif-
differred slightly from that of Study 2 because
icant Athlete ⫻ Segment interactions.
of differences in the data available. Because
In addition to identifying significant influ-
each player had only one match with each op-
ences on performance, we also looked within
ponent, we are unable to detect set or Athlete ⫻
proportions of athlete and competition effect
Set effects. Thus, the design has five effects: (a) estimates to see if there were differences be-
athlete, (b) opponent nested within athlete, (c) tween leagues. Interestingly, there were no dif-
segment, (d) Segment ⫻ Athlete, and (e) [Op- ferences between leagues in either effect, even
ponent nested within athlete] ⫻ Segment. We between the highest and lowest skill levels. On
ran the VARCOMP procedure (SPSS 22.0) on the other hand, examination of the variance
each league’s sample separately. As before, we totals demonstrated that there were differences
omit discussion of the segment effect because it in the players assigned to each league. On av-
is not significant and is not the focus of this erage, there tended to be much more variety in
article. As in Study 2, we had 100 estimates of the performances of players in the platinum
each effect because of the segment aggregation league compared with that of the bronze league,
procedure, and these 100 estimates were used to for example. Yet although the players in each
construct 95% confidence intervals. An effect is league were different, and their performances
significant if its 95% confidence interval does were different, the magnitude of athlete and
not include 0. Effects are significantly different competition effects was consistent across all
from each other if their 95% confidence inter- samples. This finding, in conjunction with the
vals do not overlap. similar findings of Studies 1 and 2, provided
In addition to estimating effects for each sam- evidence that dynamic effects are a component
ple, we were also interested in potential differ- of performance in general, rather than an aspect
ences in the proportions of variance attributable of elite performance.
to effects between samples. Consequently, we The similarity of findings in Studies 2 and 3
constructed 95% confidence intervals around suggested that, although much smaller than ath-
the median proportion of variance estimate us- lete effects, competition effects may be influen-
ing the same method that we used for the vari- tial across all levels of expertise within e-sport
ance component estimations. Groups differ in performance. We hypothesized above that com-
the proportion of variance attributable to partic- petition influences may in part reflect choices in
ular effects if the 95% confidence intervals of the strategies deployed by players. Neverthe-
their proportions do not overlap. less, although it is very likely that experienced
DYNAMIC EFFECTS ON PERFORMANCE 11
Table 3
Median Estimates of Variance Components, Confidence Intervals, and Proportion of Variance Explained
in Study 3
Variance 95% confidence interval Proportion of 95% confidence interval
component for variance variance for proportion of
Actions per minute (median) components (median) variance
Between-person
Athlete effects
Bronze (novices) 1082.72 985.51–1102.30 .74ⴱ .57–.77
Silver 1748.90 1694.99–1783.09 .66ⴱ .51–.71
.70ⴱ
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
e-sport players deploy varied strategies from to occasion from APM due to opponent. APM
match to match, it seems less likely that novices influences due to occasion are likely to have
do the same. It is possible that the source of more impact at lower levels of skill, when play-
dynamic effects shifts over time from inconsis- ers are still mastering keyboard commands.
tency due to inexperience (i.e., learning curve) Thus, this confounded influence may have had
to conscious modification of APM as a player an impact on our estimates of competition ef-
gains experience. Unfortunately, because our fects. However, because the estimates of com-
data were cross-sectional, we were unable to petition effects did not differ between leagues,
address this hypothesis. Additionally, having this seems unlikely.
multiple observations with each opponent Our findings diverged from previous findings
would have allowed us to distinguish APM due suggesting that random variability is less pro-
12 WOODS, HAYES, MEYER, KARDAN, AND BERMAN
pants of various expertise levels made a series mance process behaviors themselves vary
This document is copyrighted by the American Psychological Association or one of its allied publishers.
of handball throws. It is possible that athlete and across time and opponents. In particular, it is
competition effects behave differently depend- critical to understand how performance-process
ing on the performance variable of interest. behaviors vary within individuals and the rela-
The results of Study 3 complement ongoing tive magnitude of this dynamic variation. The
research with video game players to understand current studies offer preliminary evidence of
cognitive processes such as attention and visual significant variation in performance process be-
processing. Often, this research relies on com- haviors across opponents and over time. These
paring experienced players with novices (e.g., results were consistent across two very different
Glass, Maddox, & Love, 2013) or video game sporting domains, suggesting generalizability.
players with nonplayers (e.g., Irons, Remington, Across three studies, trait-like athlete effects
& McLean, 2011). This literature has become were the predominant influence on perfor-
controversial, however, because of lack of mance. Replicating this finding in two distinct
agreement on what constitutes expertise in re- domains, as well as across levels of expertise,
search on video game players (Latham, Patston, suggested that these stable effects may be an
& Tippett, 2013). For example, Basak, Boot, important feature of performance in general,
Voss, and Kramer (2008) trained older adults on beyond tennis and e-sport. It is possible that a
a video game for 23.5 hours before comparing a large portion of trait-like influences stems from
host of cognitive processes to those of controls, physical performance in general. The repetitive
whereas Glass and colleagues (2013) compared movements and planned behaviors involved in
players given 40 hours of training with nonplay- performance lend themselves to consistency.
ers. Similarly, there are concerns that the results Furthermore, future research should explore
of research on experienced video gamers may factors that enhance or inhibit the mostly con-
have been impacted by demand characteristics, sistent nature of performance and whether per-
particularly if participants knew why they were forming close to an athlete’s overall average is
recruited for the studies. Given that expert per- beneficial to performance outcome. Put another
formers are ideal for understanding links be- way, is being more consistent, and varying less,
tween video game playing and cognitive skill a valuable skill in sport performance? Findings
Towne, Ericsson, and Sumner (2014), it is vital of these studies may have implications for train-
that researchers have empirically supported ing programs. Athletes may benefit from learn-
methods of distinguishing experts from nonex- ing to consciously alter their consistency.
perts. The present finding that the performance Although the majority of the variance in per-
behaviors of experienced and novice players formance for tennis and e-sport was attributable
have very similar magnitudes of athlete and to stable differences between performers, we
competition effects highlights the difficulty of reliably detected a significant influence of op-
defining what expertise with a video game is. ponent, whereby some opponents pulled for
Looking simply at variation in performance be- faster serves and APM than others. These com-
haviors will likely be inadequate. petition effects were statistically distinguishable
from trait-like differences, which are stable over
General Discussion opponent and time. It is possible that the signif-
icant competition influences on process-related
For the last several decades, the sport psy- behavior represent conscious, strategic manipu-
chology literature has explored the extent to lation. It is not difficult to imagine a tennis
DYNAMIC EFFECTS ON PERFORMANCE 13
player purposefully varying her serving speed dynamic effects in performance itself, in addi-
to target a perceived weakness in her opponent, tion to dynamic effects on psychological pro-
or a StarCraft 2 player utilizing strategies with cesses. For example, Gaudreau, Nicholls, and
varying APM against different opponents. Un- Levy (2010) reported that dynamic shifts in
fortunately, because the Athlete ⫻ Opponent golfers’ coping strategies predict performance
interaction was confounded with the opponent outcome. Similarly, Doron and Gaudreau
effect into competition effects, we were unable (2014) linked prior performance outcomes to
to determine whether, for example, some oppo- later perceived control and affect. Even so,
nents pull for more APM in performance across these studies focus on variability in psycholog-
all players, or whether shifts in performance are ical processes rather than performance itself.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
due to idiosyncratic relationships between par- The present study documented dynamic effects
This document is copyrighted by the American Psychological Association or one of its allied publishers.
ticular athletes and particular opponents. in performance process in two different do-
For professional e-sport players, but not pro- mains, tennis and e-sport, and across a range of
fessional tennis players, our results showed a skill levels, from novice to professional, in one
significant Athlete ⫻ Set interaction, meaning domain. Future investigations should utilize dy-
that particular sets pulled for different levels of namic effects in both psychological variables
APM for particular athletes. In contrast, both and performance itself in order to understand
samples had significant competition influences. the role that within-person variation can have on
It is possible that the differences in dynamic performance outcomes. For example, a study of
variance components between tennis and e- tennis players could investigate whether dy-
sport professionals may be due to differences in namic shifts in affect and anxiety could be
tennis and StarCraft 2 performance variables. In linked to dynamic shifts in serving speed and
tennis, the serve is one of a constellation of measure the impact of both dynamic variation
behaviors that make up a match. In StarCraft 2 of psychological processes and performance on
and e-sport in general, keyboard and mouse the outcome of matches.
actions represent all of the possible process- The results of the current study suggested that
related behaviors. We may have been able to competitive performance-process in e-sport is
detect an Athlete ⫻ Set interaction if we were subject to significant dynamic variation. Given
able to examine all possible behaviors within earlier findings, it is important to consider a
tennis, or alternatively, these effects may disap- potential effect of solo gameplay. In a study on
pear if we were to examine only a particular the impact of social interaction on performance,
input or subset of inputs by e-sport players. In barely acquainted dyad members were asked to
addition, another possible reason for the differ- play a team-based video game together, and rate
ence is that APM in StarCraft 2 is measured the affect and feelings of supportedness elicited
over time, whereas serving speed represents dis- by their partners (Woods, Lakey, & Sain, 2015;
crete events. We encourage future researchers Study 3). Performance in the video game was
to continue to develop performance metrics, mostly attributable to stable differences in skill
particularly for e-sport. among participants, but another large portion of
Given the relative magnitude of the variance the variance was attributable to interactions be-
explained by athlete and competition effects, we tween dyad members. When a dyad member
can conceptualize performance, at least for ten- elicited unusually favorable affect and high
nis and e-sport players, to be determined pri- feelings of supportedness, that dyad member
marily by stable differences in ability between also elicited unusually strong performance in
performers, but also by the small but significant the participant. In the cooperative video game,
influence of the competitor. Furthermore, be- dynamic influences accounted for 35% of the
cause there were similar magnitudes of effects variance in performance, whereas in our sam-
at all levels of experience as examined in Study ples, dynamic influences never accounted for
3, it is possible that this structure of influences more than 15%. Although these findings may
generalizes to performance in general. simply result from the use of different video
To date, a handful of studies have investi- games and different performance metrics, it is
gated dynamic effects in sport. The current in- important to consider that cooperative perfor-
vestigation contributes to this literature by un- mance could elicit more dynamic variability
derscoring the importance of considering than the solo performances included in the cur-
14 WOODS, HAYES, MEYER, KARDAN, AND BERMAN
rent study. Future research should investigate There are several limitations to the current
the extent to which the magnitude of dynamic investigation. First, our research design did not
influences is affected by solo, dyad, and team allow for opponents to be fully crossed with
performances, such as doubles tennis. athletes, and thus the main effect of opponent
Furthermore, since dynamic variation effects was confounded with the athlete opponent in-
have been reported outside of sport perfor- teraction. Future investigations could utilize de-
mance, contrasting findings on dynamic varia- signs like a round robin that allow for separating
tion on performance across different tasks may these two effects. Second, because tennis and
yield implications for future investigations. e-sport professionals engage in quite different
Gross and colleagues (2015) demonstrated that process-related behaviors, we had to roughly
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
student performance on quizzes could be fore- equate serving speed in tennis with APM in
This document is copyrighted by the American Psychological Association or one of its allied publishers.
casted in part by idiosyncratic preferences for e-sport as process behaviors, noting that
particular professors. When a professor elicited whereas serving speed is only one of many
unusually high evaluations from a student, that tennis behaviors, APM represents the entire
student performed unusually well on that pro- spectrum of e-sport behaviors. Future research
fessor’s quiz, beyond his or her generally ten- into e-sport performance should attempt to cre-
dency to perform well on all quizzes, and be- ate subsets of input behaviors that may more
yond the professor’s tendency to elicit higher closely be equated with occasional behaviors
scores by all students. Research into dynamic like tennis serves. Alternatively, researchers
variability in ordinary social interaction has might develop a performance metric that inte-
demonstrated that most of the variance in per- grates the many aspects of tennis play, rather
ceived support, affect, and negative affect is than a single metric such as serves. A third
attributable to the interaction partner, whereas limitation is that, because Study 3 relied on only
positive affect is approximately equally socially one observation per opponent, we were not able
influenced and trait-like (Lakey & Tanner, to replicate the Athlete ⫻ Set effect found in
2013). At the dynamic level, all of these con- Study 2. Nevertheless, even with these differ-
structs are highly correlated, meaning that in- ences, we replicated findings showing both
terlocutors who elicit favorable affect also elicit strong athlete effects and significant competi-
stronger feelings of being supported. When dy- tion effects in tennis professionals, e-sport pro-
namic variance was controlled, correlations be- fessionals, and e-sport amateurs. Finally, our
tween trait-level tendencies to view others as research was limited in that we did not link
supportive and to experience favorable affect dynamic effects on performance process to per-
were either nonsignificant or inconsistent across formance outcome, which will be an important
studies. Understanding differences and similar- next step if such effects do affect outcome.
ities between ordinary social interaction and Sport psychology research has been domi-
purposeful competition may shed light into how nated by examinations of how trait-like differ-
purposeful performance works, and how train- ences among athletes can explain differences in
ing might be enhanced. performance outcomes. Yet, a growing body of
Significant dynamic influences across com- research has begun to investigate how changes
petitive domains and skill levels raise implica- within athletes’ psychological processes can af-
tions for future investigations of training pro- fect athletic performance outcomes. What has
grams. Because across all studies, some not been explored is the extent to which perfor-
opponents elicited stronger performances than mance-process behaviors in sport also change
others, athletes may benefit from being assigned over opponents and time. In three studies, we
training partners who elicit varying levels of reported preliminary evidence of important
change in a player’s typical performance, for variability in the performance of elite profes-
example, one who elicits much higher than av- sional athletes as well as amateur e-sport play-
erage serving speeds. Additionally, because ers. Study 1 found large, stable athlete effects
some opponents do pull for greater perfor- and significant, dynamic variation in serving
mances than others, it seems that athletes would speed by elite tennis players such that some
benefit from variety in practice partners more opponents elicited faster serves than others.
generally, in order to learn to mitigate undue Study 2 reported similarly large athlete effects
negative influences. and significant, dynamic effects in the fre-
DYNAMIC EFFECTS ON PERFORMANCE 15
quency of keyboard and mouse inputs by pro- Filaire, E., Alix, D., Ferrand, C., & Verger, M.
fessional e-sport players, with dynamic variabil- (2009). Psychophysiological stress in tennis play-
ity once again reflecting changes in opponent. ers during the first single match of a tournament.
Study 3 replicated earlier findings with profes- Psychoneuroendocrinology, 34, 150 –157. http://
dx.doi.org/10.1016/j.psyneuen.2008.08.022
sional performers in a sample of amateur e-sport
Gaudreau, P., Nicholls, A., & Levy, A. R. (2010).
players over a range of skill levels. Future re- The ups and downs of coping and sport achieve-
search may benefit from integrating measure- ment: An episodic process analysis of within-
ment of dynamic variation in process-related person associations. Journal of Sport & Exercise
behaviors in the quest to develop more effective Psychology, 32, 298 –311.
interventions to optimize performance out- Glass, B. D., Maddox, W. T., & Love, B. C. (2013).
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
theory when social influences are isolated from expertise levels. Journal of Motor Behavior, 39,
trait influences. Personality and Social Psychology 409 – 421. http://dx.doi.org/10.3200/JMBR.39.5
Bulletin, 36, 132–142. http://dx.doi.org/10.1177/ .409-422
0146167209349375 Shavelson, R. J., & Webb, N. M. (1991). Generaliz-
Lakey, B., & Tanner, S. (2013). Social influences in ability theory: A primer. Newbury Park, CA: Sage.
negative thinking and affect. Cognitive Therapy Towne, T. J., Ericsson, K. A., & Sumner, A. M.
and Research, 37, 160 –172. http://dx.doi.org/10 (2014). Uncovering mechanisms in video game
.1007/s10608-012-9444-9 research: Suggestions from the expert-perfor-
Latham, A. J., Patston, L. L., & Tippett, L. J. (2013). mance approach. Frontiers in Psychology, 5, 161.
Just how expert are “expert” video-game players? http://dx.doi.org/10.3389/fpsyg.2014.00161
Assessing the experience and expertise of video- Wang, H., Xia, B., & Chen, Z. (2015). Cultural
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
game players across “action” video-game genres. difference on team performance between Chinese
This document is copyrighted by the American Psychological Association or one of its allied publishers.
Frontiers in Psychology, 4, 941. http://dx.doi.org/ and Americans in multiplayer online battle arena
10.3389/fpsyg.2013.00941 games. In P. L. P. Rau (Ed.), Cross-cultural design
Martens, R., Burton, D., Vealey, R. S., Bump, L. A., applications in mobile interaction, education,
& Smith, D. E. (1990). Development and valida- health, transport and cultural heritage (pp. 374 –
tion of the Competitive State Anxiety Inventory-2. 383). New York, NY: Springer International Pub-
In R. Martens, R. S. Vealey, & D. Burton (Eds.), lishing. http://dx.doi.org/10.1007/978-3-319-
Competitive anxiety in sport (pp. 117–190). Cham- 20934-0_35
paign, IL: Human Kinetics. Witkowski, E. (2012). On the digital playing field:
Müller, H., & Sternad, D. (2004). Decomposition of How we “do sport” with networked computer
variability in the execution of goal-oriented tasks: games. Games and Culture, 7, 349 –374. http://dx
Three components of skill improvement. Journal .doi.org/10.1177/1555412012454222
of Experimental Psychology: Human Perception Woods, W. C., Lakey, B., & Sain, T. (2015). The role
and Performance, 30, 212–233. http://dx.doi.org/ of ordinary conversation and shared activity in the
10.1037/0096-1523.30.1.212 main effect between perceived support and affect.
Rees, T., Freeman, P., Bell, S., & Bunney, R. (2012). European Journal of Social Psychology. Advance
Three generalizability studies of the components online publication. http://dx.doi.org/10.1002/ejsp
of perceived coach support. Journal of Sport & .2149
Exercise Psychology, 34, 238 –251.
Schorer, J., Baker, J., Fath, F., & Jaitner, T. (2007). Received December 7, 2015
Identification of interindividual and intraindividual Revision received April 16, 2016
movement patterns in handball players of varying Accepted May 4, 2016 䡲