The Impact of the Menstrual Cycle on Bargaining Behavior

The study consists of two phases. The first phase involves tracking participants’ menstrual cycles over a three-month period. Participants are recruited three months before the online bargaining session, during which they report the start date of their last menstruation. Using this information, along with their average cycle length over the tracking period, we estimate the start of their next menstrual cycle. A three-month tracking period is necessary, as medical literature suggests this is the minimum duration required for accurate cycle measurements.

In the second phase, participants take part in an online bargaining experiment. They are randomly assigned to anonymous pairs and engage in a bargaining game over ten rounds. Each participant is given a unique anonymous ID (i.e., their Prolific ID) to ensure confidentiality throughout the study.

We recruit respondents from Prolific who reside in the EU and US and meet the following eligibility criteria: women aged 18 to 42 who are fluent in English, do not use contraceptives, and have regular menstrual cycles. To ensure eligibility, an initial pre-screening survey will be conducted on Prolific to collect general information on participants’ menstrual cycles, contraceptive use, and hormonal intake. Only individuals who meet the inclusion criteria will be invited to participate in the online experiment and the menstrual cycle tracking phase.

2025-03-24

2025-07-25

Measures of the bargaining process and outcomes:

For the bargaining process, we analyze informed players' initial offers, uninformed players' initial demands, and both players' concessions during bargaining. Initial offers and demands capture an individual's bargaining attitudes before they are affected by the negotiation partner's behavior. To investigate the interactive nature of bargaining, we look at concessions during the simultaneous bargaining stage.

Bargaining outcomes are measures of negotiation success. We analyze negotiators' final payoffs, deal rates, and payoffs conditional on reaching an agreement.

Initial offers and demands are the proposals made by informed and uninformed players in the initial bargaining phase, which are made in private and before they start interacting with the other player.

Concessions are changes in offers and demands, respectively, that show a negotiator's willingness to give up part of their own stake to reach an agreement and make the bargaining partner better off.

Conditional payoffs are the payoffs conditional on reaching an agreement. Deal rates refer to the frequency of agreements, and Final payoffs are the payoffs including disagreement.

We collect data on (i) demographics (e.g., age, education, and work status) and (ii) risk and social preferences. These variables will be used as controls to assess the robustness of our analysis.

As a robustness check, we will also adjust the window frame for ovulation. This approach will help assess the sensitivity of our findings and further reinforce the robustness of our study's conclusions. Furthermore, to ensure the reliability and validity of our findings, we will incorporate a self-reported measure of bargaining to complement the results obtained from the behavioral measure.

Finally, as an additional robustness check, we will perform the entire analysis separately for informed and uninformed players, using bootstrapped cluster-robust standard errors with 1,000 repetitions.

The study consists of two phases. The first phase involves tracking participants’ menstrual cycles over a three-month period. Participants are recruited three months before the online bargaining session, during which they report the start date of their last menstruation. Using this information, along with their average cycle length over the tracking period, we estimate the start of their next menstrual cycle. A three-month tracking period is necessary, as medical literature suggests this is the minimum duration required for accurate cycle measurements.

In the second phase, participants take part in an online bargaining experiment. They are randomly assigned to anonymous pairs and engage in a bargaining game over ten rounds. Each participant is given a unique anonymous ID to ensure confidentiality throughout the study.

I. Menstrual cycle tracking phase: Participants will monitor their menstrual cycles for three months. Each month, they will report the start date of their period, the duration of their cycle in days, and any irregularities they experience, such as irregular bleeding, missed periods, or the initiation of contraceptive use. This information enables us to accurately determine each participant’s cycle length and menstrual phase at the time of the experimental online survey (as in \cite{Buser2012}). To maintain anonymity, participants will submit their cycle data through an online Qualtrics survey using their anonymous Prolific ID. This ID allows us to link data from the tracking phase to the online survey while ensuring confidentiality.

Participants will receive tracking reminders via Prolific’s anonymous messaging tool, which allows experimenters to communicate securely with registered participants. This ensures consistency in data collection throughout the study. A few weeks before the three-month tracking period ends, participants will receive suggested time slots for scheduling their online experiment session. This step is crucial to maintaining a balanced sample across the three main menstrual phases (Ovulation, Premenstrual, and other phases).


II. Online Experiment: Following the tracking phase, participants will take part in an experimental online survey that includes a bargaining game designed to simulate real-world economic scenarios, such as salary negotiations. In this game, two female participants negotiate the division of a fixed sum of money (the "pie size"). While the total amount is known to both players, only one participant—the informed player—knows the actual value of the pie, whereas the other—the uninformed player—does not. Participants are randomly assigned to one of these roles and retain the same role throughout the experiment. The negotiation process focuses on determining the uninformed player's share of the pie.

The game consists of ten rounds, with participants being randomly paired in each round using a random-stranger matching protocol. A bargaining round concludes when the uninformed player's request matches the informed player's offer within a time limit, resulting in a deal. If an agreement is reached, the uninformed player receives the agreed share, while the informed player's earnings equal the remaining portion of the pie. If no agreement is reached within the fixed time frame, both players receive nothing.

At the end of the experimental online survey, we also assess participants’ risk and social preferences, which are controlled for in the analysis.

Additional measures and questionnaire: At the end of the experiment, we elicit demographics (age, nationality, education, and income). Additionally, we will obtain information about their current menstrual cycle to verify the data obtained from their prior menstrual cycle tracking. Lastly, we also ask hypothetical questions about their bargaining and risk attitudes.


Computer will replace disconnected players in the Bargaining game:
To avoid losing the data a whole matching group due to disruptions caused by connectivity issues, a computer-controlled player will be introduced if a participant loses connection after a bargaining round has begun. In these cases, a computer player takes over the disconnected participant’s role for the remainder of the rounds. The replacement is not pre-assigned and is triggered only in real-time. Participants are informed on-screen if they are interacting with a computer player rather than a human player and about the strategy the computer player uses. The computer player behaves according to fixed, randomized rules: in the initial bargaining stage, if assigned the role of an informed player, it proposes a random value between £0.00 and the actual pie size in the respective round; if uninformed, it selects a random value between £0.00 and £24.00 (the maximum pie size). In the simultaneous bargaining stage, the computer moves the slider to a random position every 8 seconds until second 24 with no change afterwards. If a round involving a computer player is selected for payment, participants receive the payoff corresponding to the outcome of that round.
In the main analysis, we will include all matching groups, even those where a computer player replaced a disconnected participant. However, we will exclude only the specific rounds in which a computer was actively involved in the interaction.
As a robustness check, we will re-run the analysis excluding entire matching groups that had any computer replacement, unless doing so results in too few observations for meaningful regression analysis.

Participants are randomly assigned to the roles of informed and uninformed players in the bargaining game and maintain the same role throughout the 10 bargaining periods. In addition, they are grouped into matching groups of eight players (depending on session attendance) and are re-randomized into new pairs at the start of each bargaining period. In each period, they are matched to a new other participant with the opposite role. Participants may be matched with the same other participant in multiple rounds, but they will never be matched with the same participant in consecutive rounds and they are informed about it.

Participants will be matched into groups of 8 (4 informed, 4 uninformed) for the bargaining task. This group size balances statistical power, session feasibility, and real-time matching constraints.

No

120 groups of 8 participants (N = 960), we expect to obtain 160 informed and 160 uninformed participants per phase.

960 (480 informed players and 480 uninformed players)

120 groups of 8 participants (N = 960), we expect to obtain 160 informed and 160 uninformed participants per phase.

Menstrual Phase Assignment: Participants are pre-assigned to matching groups based on three menstrual phase categories—Ovulation, Pre-menstrual, and a combined Menstrual/Post-menstrual category—estimated using self-reported cycle data collected across prior survey waves. The matching logic prioritizes two objectives: (1) maximizing the inclusion of participants in the ovulation phase, given its short temporal window, and (2) maintaining approximate balance across phase categories and roles (informed/uninformed) within each 8-person group.

Assignment to sessions is based on the participants who are connected at the time of the session, and dynamic group formation is handled automatically by oTree. The platform allows for flexible, real-time allocation of participants into matching groups while preserving the intended balance in phase and role distributions whenever possible.

Although over 1,200 participants have been tracked, we conservatively plan for a final sample size of N = 960 to account for attrition and potential no-shows. This oversampling strategy increases the likelihood of forming complete, well-balanced groups during live sessions. All participants who attend a scheduled session will be matched following the same stratified randomization protocol, with priority given to maintaining the desired balance across menstrual phases and player roles.

Participants will be matched into groups of 8 (4 informed, 4 uninformed) for the bargaining task. This group size balances statistical power, session feasibility, and real-time matching constraints. Based on power simulations, this structure allows us to reach our target distribution across menstrual phases with a manageable number of sessions. For example, with 120 groups of 8 participants (N = 960), we expect to obtain 160 informed and 160 uninformed participants. This yields 80 individual observations per treatment-phase cell, enabling us to detect a continuous effect size of d = 0.3 with 90.4% power and a 14 percentage point difference in deal rates with 84.4% power (α = 0.05), accounting for within-group dependence (i.e., assuming each observation reflects a group-level average for a given menstrual phase, rather than an individual-level measure).