Gender Bias in Professional Tennis

How to cite

Pan, Yu; Gomez-Gonzalez, Carlos; Dietl, Helmut. 2024. "Gender Bias in Professional Tennis." AEA RCT Registry. AEARCTR-0014001. https://doi.org/10.1257/rct.14001-1.2

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Yu Pan* , Carlos Gomez-Gonzalez , Helmut Dietl

* Primary investigator

Registry ID
AEARCTR-0014001
DOI
https://doi.org/10.1257/rct.14001-1.2
First registered
Last updated
Status
In development
Trial dates
2024-07-26 – 2024-08-08
Countries
Private
Registry record
http://www.socialscienceregistry.org/trials/14001

Abstract

We employ a novel experimental design to investigate gender biases in performance evaluations, specifically examining whether women receive less recognition for their successes and face harsher penalties for their mistakes compared to men. Utilizing advanced artificial intelligence-generated content (AIGC), we manipulate the perceived gender of players, allowing participants to evaluate identical performances attributed to different genders. This methodology isolates and measures the causal effects of gender bias. Our study involves Prolific participants assessing tennis highlights and unforced errors, ensuring evaluations are based solely on performance while varying perceived gender. Grounded in Regulatory Focus Theory, we explore how promotion focus (associated with ambition and risk-taking, typically linked to men) and prevention focus (associated with caution and responsibility, typically linked to women) influence these evaluations. Study 1 examines if men are more generously rewarded for successes, represented by highlights, while Study 2 investigates if women are more severely penalized for failures, represented by unforced errors. We provide novel experimental evidence on the mechanisms of gender bias in performance assessments.

Intervention

Our intervention consists of presenting participants with anonymized and gender-reversed video clips of professional tennis matches. The videos are manipulated using advanced artificial intelligence-generated content (AIGC) to create gender-neutral and gender-specific animations. Participants are randomly assigned to one of three treatment groups: (1) watching videos with players represented as unisex robots, (2) watching videos with players animated with distinctly male characteristics, and (3) watching videos with players animated with distinctly female characteristics. This setup allows us to isolate and measure the impact of perceived gender on performance evaluations.

Experimental design

This study utilizes a randomized controlled trial design to investigate gender bias in professional tennis performance evaluations. Participants are recruited through the Prolific platform and are randomly assigned to one of four groups: a control group watching standard tennis videos, or one of three treatment groups watching videos where players are depicted as unisex robots, or with distinctly male or female characteristics. The study examines two core questions: whether people reward men and women differently for high performance (Study A), and whether they penalize men and women differently for mistakes (Study B). This design ensures that the same video content is evaluated under different perceived gender conditions, allowing us to isolate the causal effects of gender bias.

Design details

Randomization method
Randomization will be conducted by survey platform automatically.
Randomization unit
The unit of randomization is the individual participant. Each participant is independently assigned to one of the treatment groups or the control group.
Clusters
Not applicable, as the randomization unit is the individual participant, not clusters.
Observations
We plan to recruit approximately 4,000 participants in total.
Arms
Study 1 (High Performance Evaluation) Control Group: 400 participants Manipulation Check Group: 400 participants Treatment Group 1A (Gender-Neutral Robots): 400 participants Treatment Group 1B (All-Men Animation): 400 participants Treatment Group 1C (All-Women Animation): 400 participants Study 2 (Mistake Evaluation) Control Group: 400 participants Manipulation Check Group: 400 participants Treatment Group 2A (Gender-Neutral Robots): 400 participants Treatment Group 2B (All-Men Animation): 400 participants Treatment Group 2C (All-Women Animation): 400 participants
Primary outcomes
The key outcome variables of interest in this experiment are: 1. Highlight Rating: The rating given to video highlights by participants, measured on a scale from 1 to 7. This assesses perceptions of high performance. 2. Mistake Rating: The severity rating given to video highlights of mistakes by participants, measured on a scale from 1 to 7. This assesses perceptions of errors.
Sponsors
Name: GRC Short Grant, University of Zurich

Cited by

No citation data available for this registration yet.

Keywords