Study 1 Design:
The experiment involves two groups of subjects. The first group is the “workers” who perform in a math addition task for four rounds. In the last round, they choose among three different payment schemes: self-tournament, other tournament and piece rate. The workers are paid based on a randomly picked round.
In a later experiment, we recruit another set of subjects – “firms.” Firms’ first task is to perform the math task for five minutes for one round – so that they would be familiarized with the nature of the task that the workers finished. We pay them 20 cents per correctly answered problem. Their second task is to choose a person who performed under the three different payment scheme rounds and in a final choice round. The Firms decide who to hire based on their choice in the final round and the worker’s fourth-round score determines the Firm’s payoff. The lab study involves three treatments: male treatment, female treatment, and neutral treatment.
In a version of this Study 1, firms decide who to hire based on their choice in the final round and the worker’s first-round score determines the Firm’s payoff, keeping all other aspects the same.
In gender (male and female) treatments, the Firm first learns the gender of their potential workers. That is, we tell them that among three only male/only female workers, and they hire one. And that each of those males/females has chosen different compensation schemes in their final rounds: one of them competed against herself, one of them competed against another individual, and one of them chose not to compete. We then ask them who they would like to hire (whose round 1 or round 4 performance determines their own payoff).
In neutral treatment, firms are matched to three individuals whose gender is not known. The rest of the procedures are identical to the gender treatment. In the end, we ask the question of where we learn the believed gender of the worker.
At the end of the experiment, subjects are asked about their belief about their own and the hired worker's performance in the task.
Study 2 Design:
In Study 2, online subjects (MTurk workers) are told that a friend of theirs is seeking for advice about what to include his/her cover letter for a job s/he is interested in applying. The advice seeker can have three competitive traits (s/he is sometimes self-competitive, is sometimes other-competitive, and sometimes refrains from competitions). The advice giver reads all these traits and what they mean in detail. We implement four treatments: the work can involve independent decision-making or can require teamwork, and the gender of the advice seeker varies. We incentivize the advice by paying subjects extra if their suggestions match with the majority of the participants.
In an extension to Study 2 (Study 2-B), we conduct the same experiment with some modifications. First, we change the word “challenge” with the word “compare” to soften the language of the competition paragraphs. In order to endorse higher ambition in the non-competitor paragraph and to make the three types of (non)-competitors more equal in that regard, we add a phrase where the candidate communicates that they are a good performer also in the non-competition paragraph. Finally, we run Study 2-B on Prolific.
Study 3 Design:
In Study 3, the participants rate different cover letters based on their employability. Keeping everything else constant, we vary the gender of the applicant and their competitive tendencies (self-competitive, non-competitive, other-competitive). Additionally, we vary the nature of the work condition: the work can involve independent decision-making or require teamwork. Each subject rate two cover letters based on their likelihood of being invited for an interview, likelihood of being hired, and the likelihood of getting a promotion to an upper-level position. The way we incentivize those ratings involves a comparison of the subjects' ratings to an expert's ratings. The closer their ratings are to the expert's, the higher the subjects earn.