Experimental Design Details
Part 1. “Lost at Sea” Task
a. Individual work
Participants first work individually on a task called“Lost at Sea”. The scenario outlines a survival situation in which a group of individuals becomes isolated after an accident. These survivors have ten items salvaged from the accident and are tasked with ranking these items based on their importance for the group’s survival. This task has been used in previous work on leadership (e.g. Thomas-Hunt and Phillips (2004); Born, Ranehill and Sandberg (2022)). To enhance the precision of confidence measures, we have adapted this task from ranking 10 items to a pairwise comparison of 5 pairs of items. Participants are presented with 5 pairs of items and must select, for each pair, which item they believe is more critical for survival. After each of the 5 questions, participants are asked to report their confidence in their answer (“local confidence”), using a scale ranging from 0 (random choice) to 100 (absolute certainty). Additionally, for the first question, participants are required to write a few lines justifying their answer to ensure their engagement and familiarity with the task. Participants' answers are then compared to those provided by a panel of survival experts. Participants receive €2 for each answer that matches the expert's answer and €0 otherwise. They do not learn about their score until the very end of the study. Before starting the task, participants must answer a set of comprehension questions correctly.
b. Confidence elicitation
Participants are asked to estimate their performance in Part 1.a. They provide two estimations: the number of correct responses (“global confidence”, on a scale from 0 to 5) and their belief about the probability of their performance being in the top 50% of a reference group (other participants who have previously completed the same task) (“relative confidence”, on a scale ranging from 0% to 100%).
c. Confidence manipulation
Participants receive a noisy feedback signal about their relative performance in Part 1.a. They are (truthfully) told that the computer draws a ball from one of two virtual urns that contain 20 balls of two different colors. If their performance was actually in the top 50%, then the ball would come from an urn with 15 black balls and 5 red balls. If their performance was not in the top 50%, then the ball would come from an urn with 15 red balls and 5 black balls. Thus, although a black ball is good news about performance, the signal is noisy and some subjects randomly receive a ball of the wrong color. This follows the method used by Schwardmann and Van der Weele (2019).
d. Post-signal confidence elicitation
We elicit participants’ belief about the probability of their performance being in the top 50% of a reference group using the exact same procedure as in Part 1.b.
All confidence elicitation questions are incentivized. Participants are told at the start of Part 1.b. that one of these questions will be randomly selected and added to their payment for Part 1.
Part 2. “Desert survival” Task
a. Appointment of a group leader
Participants are informed that they will perform a task that is similar to the “Lost at sea” scenario, but involves a different survival situation. They are also told that they are grouped with 3 other participants in the session, with whom they will work for this task. Before starting Part 2.b, a group leader will be appointed. The role of the leader is to answer each question first, providing a message along with their answer. They thus have the opportunity to persuade the group to adopt their viewpoint. Group members receive the leader’s answer and message, and provide their own answers following (or not) the leader’s input. The group's payment depends on the average number of correct responses provided by group members (including the leader)
Answers are evaluated in the same manner as in Part 1.a, and group members (including the leader) receive the same payoff for the task.
Once participants are informed of these procedures, they indicate how much they want to become the leader, on a scale between 1 and 10 (“Willingness to Lead”). The group members who indicate the highest motivation are selected to become group leaders. In case of a tie, the computer randomly selects the group leader.
After stating their willingness to lead, all group members are informed about whether they have been selected to be the group leader or not.
b. Group work
Participants complete the “Desert Survival” task, similar to Part 1.a but involving a new survival scenario. They are presented with 5 pairs of items and must select the more important item in each pair. After each of the 5 questions, they are asked to report their confidence in their answers, using a scale from 0 (random choice) to 100 (absolute certainty).
Participants must answer a set of comprehension questions correctly before stating their WTL and beginning the task.
c. Confidence elicitation:
Although payoff for the task is determined at the group level, each participant answers all questions individually, which allows to elicit again individual beliefs about performance. Participants are asked to estimate their performance in Part 2.b., providing two estimations: the number of correct responses (“global confidence”) and their their belief about the probability of their performance being in the top 50% of a reference group (other participants who have previously completed the same task) (“relative confidence”, on a scale ranging from 0% to 100%).
They are also asked an additional question regarding their beliefs about: the number of other group members who have submitted the same answers as theirs (for leaders), the percentage of participants in the session who have submitted the same answers as theirs (for followers). Each guess is incentivized, and one of these questions is randomly selected to determine a bonus added to the participant's payment for Part 2.
d. Evaluation of Leader
Every group members' answers are displayed on the screen. Leaders' answers are highlighted in red, and participants' own answers are in bold. Group members are asked to evaluate the leader's performance on a scale from 1 to 10, and leaders are asked to evaluate their own performance as a leader on the same scale.
Group members rate the extent to which they found the leader's justification message convincing on a scale from 1 (not at all convincing) to 10 (very convincing).
Part 3 - IAT Test, Questionnaire, and Payment
a. IAT
Participants take an implicit association test (IAT), eliciting the strength of their implicit associations between leadership and being male.
b. Questionnaire
Participants answer questions about their nationality, gender identity, willingness to take risks, confidence about their ability to persuade others, beliefs about gender differences in task performance, attitudes on gender and leadership, personality traits using a reduced version of the Big Five Inventory (BFI-10), and previous leadership experience. They are also asked to provide a brief motivation for why they wanted to become the leader or not. We also ask questions about the confidence manipulation in Part 1.c.
c. Hypothetical Scenario - Responsibility
In order to gain insights on reasons that motivate or deter people to lead in different leadership environments, we present participants with a hypothetical leadership scenario mimicking the responsibility treatment that we had in a previous data collection. We ask them about their WTL to lead in this scenario where the role of the leader is not to influence others, but to take decisions on behalf of others. We also ask them to pick the 3 reasons that most people who cite when asked about what would motivate or deter them to lead in the hypothetical scenario presented. This question is incentivized, and correct answers are based on our previous data collection in which we implemented the hypothetical scenario presented.
d. SVO task
We elicit participants’ social preferences using the six primary SVO Slider items (Murphy et. al, 2011).
e. Feedback and Payment
Participants receive feedback on their performance in Part 1 and 2. They also have the opportunity to view the experts' solutions to better understand their scores. Finally, one the two first parts of the experiment is randomly selected to determine the participant's final payment, according to the rules described earlier.