The superstar effect on perceived performance in professional football: An online experiment

Full text

Contents lists available at ScienceDirect
Journal of Economic Psychology
journal homepage: www.elsevier.com/locate/joep
Brief report
The superstar effect on perceived performance in professional
football: An online experiment
Yu Pan a,∗, Marco Henriques Pereira a, Carlos Gomez-Gonzalez a,b, Helmut M. Dietl a
aUniversity of Zurich, Department of Business Administration, Plattenstrasse 14, 8032 Zurich, Switzerland
bUniversity of Lausanne, Faculty of Political and Social Sciences, Sport Sciences Institute, Unicentre, 1015 Lausanne, Switzerland
A R T I C L E I N F O
JEL classification:
C91
D91
L83
PsycINFO Classification:
3040
Keywords:
Superstar effect
Perceived performance
Evaluation bias
Experiment
A B S T R A C T
We conduct a novel experiment to investigate whether football superstars consistently receive
more favorable evaluations than non-superstars. Engaging 500 participants from Prolific, we
randomly assign them to evaluate the same football videos with either visible or obscured
players. In the control group, where players are visible, superstars receive lower performance
ratings than non-superstars, challenging common perceptions. This trend is more intensified
in the treatment group, where obscured identities result in even lower ratings for superstars,
relative to non-superstars, suggesting a diminished superstar premium. These findings provide
causal experimental evidence contributing to the literature on evaluation bias and the superstar
effect.
1. Introduction
The winner-take-all phenomenon (Cook & Frank,2010;Rosen,1981), evident in industries like entertainment (Salganik et al.,
2006), academia (Bol et al.,2018), and business (Saez & Zucman,2020), shows how a few superstars capture the majority of
rewards. This phenomenon is particularly pronounced in professional sports, where a few individuals consistently rank among the
world’s highest-paid athletes and accumulate the most individual prices (Bourg & Gouguet,2023). In general, even when their work
is shared or similar, the superstar receives more credit than a relative unknown (Rossiter,1993). This is why minor differences in
talent or popularity can magnify significant earnings disparities, favoring superstars (Adler,1985;Rosen,1981). Such preferential
treatment is primarily driven by consumer preferences for superstars (Brandes et al.,2008;Humphreys & Johnson,2019), further
magnified by extensive media coverage that enhances their visibility and strengthens their market position (Adler,2006).
Concepts from sociology and psychology further describe the formation and influence of superstars. The Matthew effect, as
theorized by Merton (1968), describes how well-regarded individuals accrue greater recognition and credit, highlighting external
influences on cumulative advantage. The halo effect emphasizes the internal impact of reputation on individual perception and
judgment (Thorndike,1920). In this paper, we investigate how superstar status may bias audience perceptions by using sports
highlights as experimental material. The sports setting is ideal to comparably examine the evaluations of high-skilled individuals
who face the same tasks on a standardized playing field (Arrondel et al.,2019;Dilmaghani,2022;Morgulev et al.,2019). Our paper
examines whether superstars receive more favorable evaluations than non-superstars in a setting with comparable performance.
We conduct an experiment with 500 participants recruited from Prolific. The participants are tasked with evaluating 10 football
highlight videos, which include five videos of superstars and five videos of non-superstars. We randomly assign the participants
∗Corresponding author.
E-mail address: yu.pan@business.uzh.ch (Y. Pan).
https://doi.org/10.1016/j.joep.2024.102776
Received 8 March 2024; Received in revised form 24 September 2024; Accepted 3 November 2024
Journal of Economic Psychology 106 (2025) 102776
Available online 13 November 2024
0167-4870/© 2024 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license
( http://creativecommons.org/licenses/by/4.0/ ).

Y. Pan et al.
to one of two groups: the control group, where participants watch all 10 videos with visible players, and the treatment group,
where participants watch the same 10 videos but with obscured players. In the treatment group, we utilize Ultralytics YOLOv8,
an advanced deep-learning and computer-vision tool, to anonymize players while maintaining the integrity of their movement.
We measure performance through the ‘‘blind tasting’’ method, where evaluators rate performance without recognizing the players’
identities, similar to wine tastings where the wines’ identities are not disclosed (Goldstein et al.,2008). This experimental design
allows us to measure the causal effect of superstar status on performance evaluations.
This paper makes two main contributions. First, it provides novel experimental evidence that addresses endogeneity issues
prevalent in previous studies on evaluation bias in sports. Previous research identifies biases favoring high-status individuals, such as
in baseball refereeing (Kim & King,2014), Olympic Games umpiring (Waguespack & Salomon,2016), and the voting processes for the
FIFA Ballon d’Or and NBA All-Star (Anderson et al.,2020;Biegert et al.,2023). However, these studies may face some endogeneity
issues that prevent strong causal conclusions. Kausel et al. (2019) addresses endogeneity by using penalty shoot-out outcomes as
external shocks to measure outcome bias in subjective performance ratings. By randomly assigning participants to control and
treatment groups and using advanced computer vision techniques, we isolate the impact of performance from biases related to player
status. This approach provides clearer causal inferences on evaluation bias favoring superstars among non-professional audiences.
Second, this paper shows how the halo effect supplements our understanding of status-bias accumulation and the formation of
distinctions between superstars and non-superstars. Research in managerial settings shows that subjective performance assessments
are often swayed by factors unrelated to actual performance, such as a person’s past achievements or popularity (Bellé et al.,2017;
Maske et al.,2021). Our paper expands this concept to sports, revealing that such factors can influence audiences’ evaluations
of a player’s on-pitch performance. The halo effect provides a causal mechanism for how superstar status influences performance
evaluations. It suggests that the overall impression of a person (in our case, the superstar status) creates a cognitive bias that affects
the evaluation of their specific attributes or actions. By manipulating the visibility of players’ identities, we can isolate and test
whether superstar status leads to biased performance evaluations.
We observe two primary findings. First, superstars, on average, receive lower ratings than non-superstars in both the control and
treatment groups. This pattern suggests an underestimation of non-superstars’ performances, particularly in highlight selections, and
implies that for non-superstars to achieve recognition comparable to superstars, non-superstars may need to outperform superstars.
The result aligns with previous research showing how media coverage enhances superstar winner-take-all dynamics (Koenig,2023).
Second, when we obscure players’ identities in the treatment group, the ratings for superstars decrease compared to non-superstars.
This change implies the superstar premium is diminished when players’ identities are concealed. These findings are robust after
controlling for demographic information, club preference, and scored-type categories.
2. Methodology
2.1. Criteria for video selection
We select video highlights from the annual compilation of the Premier League (www.premierleague.com), featuring the best
goals from the 2021–22 season. This selection allows us to exploit differences between superstars and non-superstars with videos of
comparable performance quality.
Previous empirical studies using sports data have investigated the superstar effect (Deutscher et al.,2023;Franck & Nüesch,
2012). Acknowledging both exceptional talent (Rosen,1981) and disproportionate popularity (Adler,1985) as key factors in defining
a ‘‘superstar’’, we utilize performance ratings from WhoScored as a proxy for talent (Özdemir et al.,2022) and Instagram follower
counts as a proxy for popularity (Budzinski & Gaenssle,2018).
We employ a composite index, combining talent and popularity, to identify superstar players, defining the top five players
by composite score as superstars. Non-superstars are selected based on quartiles of the composite-index ranking, ensuring a
representation of player performance and popularity levels. Our selection includes superstars such as Cristiano Ronaldo and
Mohamed Salah, as well as non-superstars like John McGinn and Ollie Watkins (as illustrated in Fig. 1).
We categorize highlight videos into three specific types: long-shot, solo, and long-pass goals. This facilitates an equitable
comparison between superstars and non-superstars, based on the nature of the goals scored. Table 1lists the 10 selected players
and the matches from which the highlight videos are compiled.
2.2. Survey and participants
We use EFS Survey (www.unipark.com) to design a questionnaire with embedded videos. Our study involves 500 UK participants
from Prolific, equally divided between 250 women and 250 men, with an average age of 41.92 years. Because of a higher proportion
of women on Prolific (Douglas et al.,2023), we use the gender-balancing feature of the platform, aligning with the male-dominated
demographic common in football viewership. Participants from Prolific are directed to the EFS Survey. The study compensates £0.75
per participant, with an anticipated completion time of five minutes. Based on technical difficulties observed during our pre-test,
we restrict participation to tablet and desktop users.
Participants are randomly assigned to one of two groups: a control group and a treatment group. Fig. 2shows screenshots of two
identical frames to illustrate the experimental conditions. All participants watch 10 videos in a random order: In the control group
the players are visible (Fig. 2(a)), and in the treatment group the players are obscured (Fig. 2(b)); there is otherwise no difference
between the two groups. To minimize club and player recognition, we apply a black-and-white filter to the spectator stands in both
Journal of Economic Psychology 106 (2025) 102776
2

Y. Pan et al.
Fig. 1. Selection of superstar and non-superstar.
Table 1
List of selected players and matches (2021–22 season).
No. Name Club Opponent
Superstar
1 Mohamed Salah Liverpool Manchester City
2 Cristiano Ronaldo Manchester United Tottenham Hotspur
3 Kevin De Bruyne Manchester City Leeds United
4 Son Heung-min Tottenham Hotspur Manchester City
5 Harry Kane Tottenham Hotspur Leeds United
Non-superstar
6 John McGinn Aston Villa Watford
7 Ollie Watkins Aston Villa Brighton & Hove Albion
8 Naby Keïta Liverpool Crystal Palace
9 Harvey Barnes Leicester City Leeds United
10 Enock Mwepu Brighton & Hove Albion Liverpool
groups. For the treatment group, we employ the computer-vision model YOLOv8 (Jocher et al.,2023) and apply Gaussian blur to
all players for the duration of the videos. We ensure that any identifiable features of the players are effectively obscured, while
maintaining the clarity of the players’ movement in a competitive game environment.
After watching each video, participants rate the performance of the player who scored the goal on a scale ranging from 1 (poor)
to 7 (outstanding), and then they answer a series of demographic and football-related questions (Online Appendix A includes the
questionnaire). This experimental design allows us to identify the causal effect of the halo effect associated with superstar status. By
randomly assigning participants to groups with visible or obscured players, we can isolate how the knowledge of a player’s superstar
status causally influences performance evaluations.
Our comprehensive dataset encompasses ratings from these 500 participants across each of the 10 video highlights, resulting
in a total of 5,000 observations. Table 2provides detailed demographics of the participants, including age, education level, player
recognition, and club preferences. Participants, on average, spent less than 5 min completing the survey. Furthermore, 51% reported
having a favorite club, while 42% indicated having a least favorite club. Based on this information, we create dummy variables to
control when a (least) favorite club scores or concedes a goal. Online Appendix B provides further details about ‘‘preference controls’’.
2.3. Empirical strategy
To assess the effect of superstar status on how participants rate players’ performances, we use the following regression model:
Rating𝑖𝑗 =𝛼0+𝛽1Superstar𝑖𝑗 +𝛽2ObscuredPlayers𝑖𝑗
+𝛽3(Superstar𝑖𝑗 ×ObscuredPlayers𝑖𝑗 ) +𝑋𝑗+𝛼𝑖+𝜖𝑖𝑗
In this model, Rating𝑖𝑗 represents the rating given to the video highlight 𝑖by the participant 𝑗. The variable Superstar𝑖𝑗 is a
binary indicator of whether the video features a superstar player, while ObscuredPlayers𝑖𝑗 indicates whether the participant (in
Journal of Economic Psychology 106 (2025) 102776
3

Y. Pan et al.
Fig. 2. Example of videos.
Table 2
Descriptive statistics of participants.
Overall Treatment group Control group
Mean SD Min Max Obs Mean SD Obs Mean SD Obs
Age 41.92 14.25 18 81 500 42.09 14.52 245 41.75 14.01 255
Years of education 15.29 2.16 6 18 497 15.23 2.25 244 15.35 2.07 253
Recognition (yes) 0.21 0.41 0 1 500 0.13 0.33 245 0.29 0.45 255
Favorite club 0.51 0.50 0 1 500 0.56 0.50 245 0.46 0.50 255
Least favorite club 0.42 0.49 0 1 500 0.49 0.50 245 0.35 0.48 255
Time (in seconds) 281.29 147.57 142 1612 500 281.13 155.51 245 281.44 139.81 255
Note: Summary statistics by group: Overall, treatment group, and control group.
the treatment group) watches the video with obscured players. Superstar𝑖𝑗 ×ObscuredPlayers𝑖𝑗 is an interaction term. 𝑋𝑗includes
control variables, participant demographics, and club preferences. 𝛼𝑖represents the scored-type fixed effects of the video highlight 𝑖.
Standard errors are clustered at the participant level. The coefficients of interest are 𝛽1, assessing the performance-rating difference
between superstars and non-superstars, and 𝛽3, evaluating the altered superstar effect when players’ identities are obscured.
Our regression framework quantifies the superstar effect as a signal in performance evaluation, highlighting how the visibility of
a superstar’s identity influences audience evaluations. We control for specific characteristics of each video and adjust for potential
biases among participants to enhance the robustness and validity of our findings.
3. Results
Fig. 3shows how the ratings of superstars decrease when their identities are obscured. Under the same treatment, non-superstars’
ratings decrease less and even increase in the cases of John McGinn and Ollie Watkins. Fig. 4presents aggregated results, categorized
under superstar and non-superstar, within the control group and treatment group. The results suggest a preference towards non-
superstars in both groups. The change from visible to obscured player identity results in an increased rating disparity between
superstars and non-superstars, growing from −0.111 (5.605–5.716) to −0.234 (5.488–5.722).
Table 3summarizes the estimation results. The superstar coefficient is negative and statistically significant across all outcomes.
This suggests that superstar players generally receive lower ratings than non-superstars, regardless of the visibility of players. The
coefficients for obscured players are not statistically significant, suggesting that obscuring player identities does not significantly
change the ratings. This result aligns with our experimental design goal of effectively masking superstar features without
compromising the clarity of their movements in a competitive game environment. The coefficient of the interaction term superstar
×obscured players is negatively significant at the 1% level across all outcomes, suggesting that obscuring players’ identities in the
treatment group leads to a greater decrease in superstar ratings compared to non-superstars.
The inclusion of scored-type fixed effects in column 4 (Table 3) enhances the robustness of our findings. This method controls
for variations in goal-scoring scenarios, isolating the influence of superstar status and visibility from other variables. The model
accounts for demographic characteristics and individual preferences, enhancing the accuracy of our analysis in identifying biases in
football performance ratings.
We report additional results and robustness checks for respondent biases in Online Appendix B. Participants show a preference
for videos of their favorite clubs scoring and of solo goals (Table B1). Our results are robust to various respondent bias checks (Table
B2) and a randomization inference exercise (Figure B1).
Journal of Economic Psychology 106 (2025) 102776
4

Y. Pan et al.
Fig. 3. Comparison of ratings by players.
Fig. 4. Comparison of ratings by superstar and non-superstar.
Our findings indicate that, on average, superstars are rated lower than non-superstars. When players’ identity is visible, in the
control group, the disparity between the ratings of superstars and non-superstars is smaller than when the players’ identities are
obscured in the treatment group. This means that when the identity of players is not visible, superstars’ ratings decrease, while
non-superstars’ ratings do not.
4. Discussion and concluding remarks
This study provides novel insights into evaluation biases in professional football. We examine whether superstars consistently
receive more favorable evaluations than non-superstars. Unexpectedly, our findings reveal that superstars receive lower performance
ratings than non-superstars, regardless of whether their identities are visible or obscured in the videos. This suggests that non-
superstars may need to outperform superstars to obtain similar recognition in highlight selections. When selecting highlights, media
may favor a player’s popularity over performance, as popularity is directly related to entertainment income (Brandes et al.,2008;
Humphreys & Johnson,2019). To address potential selection bias in this process, future research could employ qualitative methods
to analyze discrepancies between media evaluations and general audience perceptions (Budzinski & Gaenssle,2018), or use causal
machine learning techniques to objectively measure scoring difficulty across players (Brox & Lechner,2024).
Furthermore, obscuring player identities in the treatment group results in a marked decrease in ratings for superstars as
compared to non-superstars. This change points to a diminished superstar premium, revealing a bias that overrates their performance
Journal of Economic Psychology 106 (2025) 102776
5

Y. Pan et al.
Table 3
Main results.
Dependent variable: rating
(1) (2) (3) (4)
Superstar −0.111∗∗∗ −0.106∗∗∗ −0.113∗∗∗ −0.113∗∗∗
(0.029) (0.029) (0.029) (0.029)
Obscured players 0.009 0.017 0.014 0.015
(0.059) (0.059) (0.059) (0.059)
Superstar ×Obscured players −0.126∗∗ −0.135∗∗ −0.142∗∗∗ −0.141∗∗∗
(0.042) (0.042) (0.042) (0.042)
Mean of dependent variable 5.632 5.638 5.638 5.638
Demographics controls √ √ √
Preference controls √ √
Scored-type FE √
Observations 4978 4860 4860 4860
R20.010 0.017 0.019 0.070
Standard errors in parentheses.
∗𝑝 <0.05,∗∗ 𝑝 <0.01,∗∗∗ 𝑝 <0.001.
irrespective of actual performance levels. Although the effect size of approximately 0.1 is modest, nuanced differences can drive
considerable disparities in exposure and wages between superstars and non-superstars. The results support previous findings
from blind experiments, where consumers show a similar bias for pricey and labeled wines (Goldstein et al.,2008). Our results
indicate that superstars do not necessarily perform and contribute more than non-superstars (Jedelhauser et al.,2023); rather,
they are perceived to be better. However, our experimental design only addresses the quality of goals in highlights. This approach
inevitably overlooks other performance-related factors leading to unequal outcomes between superstars and non-superstars, such as
contribution size and consistency.
Our experiment provides robust evidence for how superstar status influences performance evaluations through the halo effect.
By manipulating player identity visibility, we show that evaluators’ prior knowledge of a player’s superstar status causally affects
performance ratings, independent of actual performance. This finding contributes to the literature on the causal mechanism of
the halo effect in performance evaluations (Bellé et al.,2017;Dennis,2007). Our results suggest that evaluators can be biased by
an individual’s status, granting an evaluation premium to superstars or those with higher status (Kim & King,2014;Liao,2021;
Nufer,2019). This causal link between status and evaluation bias enhances our understanding of subjective evaluation processes,
highlighting how established perceptions of status can shape perceptions of performance.
We acknowledge two main limitations in our study. First, the complex nature of team sports makes measuring individual
performance challenging. We overcome this issue by focusing on highlights that include the best goals of a given year, making the
performance of goalscorers comparable. The limited number of highlight videos and score types may constrain the generalizability
of our results. Future research may use individual sports, where performance can be clearly attributed to the individual, to further
explore these dynamics. Additionally, we do not control for other factors such as physical appearance that may contribute to
superstars receiving favorable evaluations (Parshakov et al.,2024).
Second, the manipulation of features in video format presents more challenges than in static images. While we successfully
employ Ultralytics YOLOv8 for our ‘‘blind tasting’’ methodology, an advancement over Adobe Premiere Pro (Gomez-Gonzalez et al.,
2023), future research could benefit from employing more advanced AI techniques for player transformation. An exemplary model
for such transformation is the gender portrayal technique used by Marcel and Orange in their video campaigns (Ewe,2023).
Technological advances could enable a deeper exploration of biases, mirroring the approaches used in studies like the manipulation
of beauty (Póvoa et al.,2020), and helping develop fair performance measures to overcome evaluation biases (Sarlis & Tjortjis,
2020).
Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared
to influence the work reported in this paper.
Appendix A. Supplementary data
Supplementary material related to this article can be found online at https://doi.org/10.1016/j.joep.2024.102776.
References
Adler, M. (1985). Stardom and talent. American Economic Review,75(1), 208–212, https://www.jstor.org/stable/1812714.
Adler, M. (2006). Stardom and talent. In V. A. Ginsburg, & D. Throsby (Eds.), vol. 1,Handbook of the economics of art and culture (pp. 895–906). Elsevier,
http://dx.doi.org/10.1016/S1574-0676(06)01025- 8.
Journal of Economic Psychology 106 (2025) 102776
6

Y. Pan et al.
Anderson, C. J., Arrondel, L., Blais, A., Daoust, J. F., Laslier, J. F., & Van der Straeten, K. (2020). Messi, Ronaldo, and the politics of celebrity elections: Voting
for the best soccer player in the world. Perspectives on Politics,18(1), 91–110. http://dx.doi.org/10.1017/S1537592719002391.
Arrondel, L., Duhautois, R., & Laslier, J. F. (2019). Decision under psychological pressure: The shooter’s anxiety at the penalty kick. Journal of Economic Psychology,
70, 22–35. http://dx.doi.org/10.1016/j.joep.2018.10.008.
Bellé, N., Cantarelli, P., & Belardinelli, P. (2017). Cognitive biases in performance appraisal: Experimental evidence on anchoring and halo effects with public
sector managers and employees. Review of Public Personnel Administration,37(3), 275–294. http://dx.doi.org/10.1177/0734371X17704891.
Biegert, T., Kühhirt, M., & Van Lancker, W. (2023). They can’t all be stars: The Matthew Effect, cumulative status bias, and status persistence in NBA All-Star
elections. American Sociological Review,88(2), 189–219. http://dx.doi.org/10.1177/00031224231159139.
Bol, T., de Vaan, M., & van de Rijt, A. (2018). The Matthew effect in science funding. Proceedings of the National Academy of Sciences,115(19), 4887–4890.
http://dx.doi.org/10.1073/pnas.1719557115.
Bourg, J.-F., & Gouguet, J.-J. (2023). Superstars: Why does the winner take all? In Socio-economics of sport – a critical analysis. Université de Limoges,
http://dx.doi.org/10.25965/ebooks.484.
Brandes, L., Franck, E., & Nüesch, S. (2008). Local heroes and superstars: An empirical analysis of star attraction in German soccer. Journal of Sports Economics,
9(3), 266–286. http://dx.doi.org/10.1177/1527002507302026.
Brox, E., & Lechner, M. (2024). Teamwork and spillover effects in performance evaluations. arXiv preprint arXiv:2403.15200 https://arxiv.org/abs/2403.15200.
Budzinski, O., & Gaenssle, S. (2018). The economics of social media (super-)stars: An empirical investigation of stardom and success on YouTube. Journal of
Media Economics,31(3–4), 75–95. http://dx.doi.org/10.1080/08997764.2020.1849228.
Cook, P. J., & Frank, R. H. (2010). The winner-take-all society: Why the few at the top get so much more than the rest of us. Penguin Random House.
Dennis, I. (2007). Halo effects in grading student projects. Journal of Applied Psychology,92(4), 1169–1176. http://dx.doi.org/10.1037/0021-9010.92.4.1169.
Deutscher, C., Neuberg, L., & Thiem, S. (2023). Who’s afraid of the GOATs?–Shadow Effects of tennis superstars. Journal of Economic Psychology,99, Article
102663. http://dx.doi.org/10.1016/j.joep.2023.102663.
Dilmaghani, M. (2022). Chess girls don’t cry: Gender composition of games and effort in competitions among the super-elite. Journal of Economic Psychology,89,
Article 102482. http://dx.doi.org/10.1016/j.joep.2022.102482.
Douglas, B. D., Ewell, P. J., & Brauer, M. (2023). Data quality in online human-subjects research: Comparisons between MTurk, Prolific, CloudResearch, Qualtrics,
and SONA. PLoS One,18(3), 1–17. http://dx.doi.org/10.1371/journal.pone.0279720.
Ewe, K. (2023). A viral French ad shows how women’s soccer can be just as exciting as men’s. Time, Retrieved from https://time.com/6295047/viral-france-
advertisement-soccer- gender-orange.
Franck, E., & Nüesch, S. (2012). Talent and/or popularity: What does it take to be a superstar? Economic Inquiry,50(1), 202–216. http://dx.doi.org/10.1111/j.1465-
7295.2010.00360.x.
Goldstein, R., Almenberg, J., Dreber, A., Emerson, J. W., Herschkowitsch, A., & Katz, J. (2008). Do more expensive wines taste better? Evidence from a large
sample of blind tastings. Journal of Wine Economics,3(1), 1–9. http://dx.doi.org/10.1017/S1931436100000523.
Gomez-Gonzalez, C., Dietl, H., Berri, D., & Nesseler, C. (2023). Gender information and perceived quality: An experiment with professional soccer performance.
Sport Management Review,27(1), 45–66. http://dx.doi.org/10.1080/14413523.2023.2233341.
Humphreys, B., & Johnson, C. (2019). The effect of superstars on game attendance: Evidence from the NBA. Journal of Sports Economics,21,http://dx.doi.org/
10.1177/1527002519885441.
Jedelhauser, F., Flepp, R., & Franck, E. (2023). Overshadowed by popularity: The value of second-tier stars in European football. Journal of Sports Economics,
24(8), 1026–1054. http://dx.doi.org/10.1177/15270025231187880.
Jocher, G., Chaurasia, A., & Qiu, J. (2023). Ultralytics YOLOv8. https://github.com/ultralytics/ultralytics, Version 8.0.0.
Kausel, E. E., Ventura, S., & Rodríguez, A. (2019). Outcome bias in subjective ratings of performance: Evidence from the (football) field. Journal of Economic
Psychology,75, Article 102132. http://dx.doi.org/10.1016/j.joep.2018.12.006.
Kim, J. W., & King, B. G. (2014). Seeing stars: Matthew effects and status bias in Major League Baseball umpiring. Management Science,60(11), 2619–2644.
http://dx.doi.org/10.1287/mnsc.2014.1967.
Koenig, F. (2023). Technical change and superstar effects: Evidence from the rollout of television. American Economic Review: Insights,5(2), 207–223.
http://dx.doi.org/10.1257/aeri.20210539.
Liao, C. H. (2021). The Matthew effect and the halo effect in research funding. Journal of Informetrics,15(1), Article 101108. http://dx.doi.org/10.1016/j.joi.
2020.101108.
Maske, M. K., Sohn, M., & Hirsch, B. (2021). How managerial accountability mitigates a halo effect in managers’ ex-post bonus adjustments. Management
Accounting Research,51, Article 100738. http://dx.doi.org/10.1016/j.mar.2021.100738.
Merton, R. K. (1968). The Matthew effect in science: The reward and communication systems of science are considered. Science,159(3810), 56–63. http:
//dx.doi.org/10.1126/science.159.3810.56.
Morgulev, E., Azar, O. H., & Bar-Eli, M. (2019). Does a ‘‘comeback’’ create momentum in overtime? Analysis of NBA tied games. Journal of Economic Psychology,
75, Article 102126. http://dx.doi.org/10.1016/j.joep.2018.11.005.
Nufer, G. (2019). "Say hello to Halo": the halo effect in sports. Innovative Marketing,15, 116–129. http://dx.doi.org/10.21511/im.15(3).2019.09.
Özdemir, A., Dietl, H., Rossi, G., & Simmons, R. (2022). Are workers rewarded for inconsistent performance? Industrial Relations: A Journal of Economy and
Society,61(2), 137–151. http://dx.doi.org/10.1111/irel.12292.
Parshakov, P., Gasparetto, T., Votintseva, N., & Shakina, E. (2024). Beyond the pitch: Exploring the role of beauty in soccer player salaries. Journal of Economic
Psychology,101, Article 102709. http://dx.doi.org/10.1016/j.joep.2024.102709.
Póvoa, A. C. S., Pech, W., Viacava, J. J. C., & Schwartz, M. T. (2020). Is the beauty premium accessible to all? An experimental analysis. Journal of Economic
Psychology,78, Article 102252. http://dx.doi.org/10.1016/j.joep.2020.102252.
Rosen, S. (1981). The economics of superstars. American Economic Review,71(5), 845–858, https://www.jstor.org/stable/1803469.
Rossiter, M. W. (1993). The Matthew Matilda effect in science. Social Studies of Science,23(2), 325–341, https://www.jstor.org/stable/285482.
Saez, E., & Zucman, G. (2020). The rise of income and wealth inequality in America: Evidence from distributional macroeconomic accounts. Journal of Economic
Perspectives,34(4), 3–26. http://dx.doi.org/10.1257/jep.34.4.3.
Salganik, M. J., Dodds, P. S., & Watts, D. J. (2006). Experimental study of inequality and unpredictability in an artificial cultural market. Science,311(5762),
854–856. http://dx.doi.org/10.1126/science.1121066.
Sarlis, V., & Tjortjis, C. (2020). Sports analytics–Evaluation of basketball players and team performance. Information Systems,93, Article 101562. http:
//dx.doi.org/10.1016/j.is.2020.101562.
Thorndike, E. L. (1920). A constant error in psychological ratings. Journal of Applied Psychology,4(1), 25–29. http://dx.doi.org/10.1037/h0071663.
Waguespack, D. M., & Salomon, R. (2016). Quality, subjectivity, and sustained superior performance at the Olympic Games. Management Science,62(1), 286–300.
http://dx.doi.org/10.1287/mnsc.2014.2144.
Journal of Economic Psychology 106 (2025) 102776
7