Whowill lead? Evidence on leadership exposure and the gender gap in aspirations

How to cite

Cairo, Sofie. 2025. "Whowill lead? Evidence on leadership exposure and the gender gap in aspirations." AEA RCT Registry. AEARCTR-0017194. https://doi.org/10.1257/rct.17194-1.0

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

Sofie Cairo*

* Primary investigator

Registry ID
AEARCTR-0017194
DOI
https://doi.org/10.1257/rct.17194-1.0
First registered
Last updated
Status
In development
Trial dates
2025-11-06 – 2026-08-01
Countries
Private
JEL codes
J16, J24, J71, D03
Registry record
http://www.socialscienceregistry.org/trials/17194

Abstract

There is limited prior evidence on how to close the gender gap in leadership aspirations, which emerges already in adolescence (Alan et al. 2020), persists among university graduates and entry-level employees, and deepens further over careers (Azmat et al., 2025; Haegele, 2024).

This motivates our research question and intervention. The question we try to answer is: Can individual leadership aspirations be exogenously stimulated by randomly assigning individuals to a leadership role on a team that has to solve an abstract collaborative task requiring both communication and decision-making. The channel through which this random leadership experience may stimulate willingness to lead is through making own leadership preferences and ability salient to the randomly selected individual, as well as emphasizing typical elements of a leadership role, namely a specific title, certain responsibilities, some private information, and economic incentives.

While leadership tasks in the workplace are typically not randomly assigned, but are rather a strong signal and predictor of future promotions (Bircan et al, 2024), we investigate if these leadership experiences in and of themselves stimulate individuals' willingness to take on leadership roles. We hypothesize that such leadership experiences may be particularly relevant to groups of employees who are underrepresented among managers, and as such have fewer role models, leading to uncertainty and lower confidence in own fit and skills to lead.

Specifically, we conduct an RCT on university students, a large share of which business students, expecting to work in private sector companies after graduation. The RCT is framed as a training session on teamwork and collaboration, centered around playing two rounds of a newly developed multiplayer online game in groups of four at the experimental lab of University of Copenhagen.

Intervention

We conduct an RCT on university students. The RCT is framed as a training session on teamwork and collaboration, centered around playing two rounds of a newly developed multiplayer online game in groups of four at a gaming center in Copenhagen.

In round one of gaming, we randomly assign group leaders who are granted a title, certain responsibilities, access to private information, and individual performance-based economic incentives. After round one, preferences for playing the leader in round two are elicited. Individuals are asked if they are willing to be the leader of the team (1), Would accept to be promoted to leader (2), are indifferent between being a leader or a team member (3), or prefer to be a team member (4). Preference elicitation is incentivized and binding. Participants are informed that teams will be reshuffled and that within each new team, the participant expressing the stronger leadership preference (equivalent to a lower number) will be promoted with ties being settled randomly, and that individuals selecting the fourth option will never be asked to lead. That is, based on their expressed preference, participants are put first, second, or third inline to be the leader of their new team. In the predefined round two groups, the group member with stronger leadership preferences is then promoted to be the leader in accordance with instructions, and round two, which is identifcal to round one, is played.

The experimental data are combined with pre-elicited measures of skills and preferences for leadership and gaming. Comparing willingness to lead among randomly selected leaders and randomly selected team members allows us to derive the causal effect of treatment on leadership preferences.

Experimental design

We conduct an RCT on university students. The RCT is framed as a training session on teamwork and collaboration, centered around playing two rounds of a newly developed multiplayer online game in groups of four at a gaming center in Copenhagen.

In round one of gaming, we randomly assign group leaders who are granted a title, certain responsibilities, access to private information, and individual performance-based economic incentives. After round one, preferences for playing the leader in round two are elicited. Individuals are asked if they are willing to be the leader of the team (1), Would accept to be promoted to leader (2), are indifferent between being a leader or a team member (3), or prefer to be a team member (4). Preference elicitation is incentivized and binding. Participants are informed that teams will be reshuffled and that within each new team, the participant expressing the stronger leadership preference (equivalent to a lower number) will be promoted with ties being settled randomly, and that individuals selecting the fourth option will never be asked to lead. That is, based on their expressed preference, participants are put first, second, or third inline to be the leader of their new team. In the predefined round two groups, the group member with stronger leadership preferences is then promoted to be the leader in accordance with instructions, and round two, which is identifcal to round one, is played.

The experimental data are combined with pre-elicited measures of skills and preferences for leadership and gaming. Comparing willingness to lead among randomly selected leaders and randomly selected team members allows us to derive the causal effect of treatment on leadership preferences.

Design details

Randomization method
For round one of gaming, the randomization is based on allocation of random numbers (Excel) to participants, and then choosing the 5 men with highest numbers as team leaders, and the 5 women with highest numbers as team leaders, and then distributing the rest randomly into teams starting from the participant with the highest number. For round one of gaming, again random numbers are drawn and individuals are distributed into teams starting from the highest number.
Randomization unit
For the first round of gaming, we randomly select 5 male group leaders and 5 female group leaders, and then distribute the rest of the students as team members to their teams. Next, we randomly form round two teams, but we can only assign roles after eliciting leadership preferences in the midsurvey. The team member who has stronger leadership preferences is promoted to leader.
Clusters
The treatment is not clustered.
Observations
I am enrolling 520 students in total, that it 130 teams of four participants each. I randomly pick 65 male leaders and 65 female leaders.
Arms
The treatment arm has 130 observations, whereas the control arm has 390 observations. When separating by gender, the treatment arm for women (men) has 65 observations, while the control arm has at least 65 women (men).
Primary outcomes
The key outcomes that we study are share of treated (randomly exposed to a leadership experience in round one of gaming) relative to controls (team members in round one of gaming) who express a willingness to lead in round two of computer gaming. Also, we measure the quality of leaders who are promoted after a leadership experience relative to leaders who are promoted after a team member experience.
Secondary outcomes
Secondary outcomes relate to the mechanisms behind leadership quality and leadership preferences. Generally we analyze communication and collaboration metrics and transcriped team communication data. These include: Share of talk time, number of interruptions, share of time spent actively listening and more.

Cited by

No citation data available for this registration yet.

Keywords