Beyond the Mood Swings

Tracking Anxiety and Social Cognition Across the Menstrual Cycle

By Nikki Rademaker

Overview

For my Master thesis in Creative Intelligence & Technology at Leiden University, daily anxiety and affective Theory of Mind (ToM) performance were studied across menstrual phases. I tracked state anxiety for two full cycles in participants with a regular menstrual cycle by sending out daily surveys containing the State-Trait Anxiety Inventory - S (STAI-S) ( Spielberger et al., 1972). Furthermore, affective ToM performance was tested twice per cycle by short story tests based on the Adult Faux Pas Recognition Test (FPRT) ( Stone et al., 1998).

Tools

• Qualtrics (surveys and tests)
• Excel (daily tracking and ToM scoring)
• Python (data preprocessing and visualization)
• JASP (statistics)

The Project

The menstrual cycle is a hormonal process that prepares the female body for a possible pregnancy. It repeats monthly if pregnancy does not occur and consists of four phases: menstrual, follicular, ovulation, and luteal. Research on the menstrual cycle has long been neglected, resulting in major gaps in understanding its patterns and effects. These gaps persist due to the topic's complexity, its underexplored nature, and the lack of a standardized, reliable research method (Sharp et al., 2023; Critchley et al., 2020). Therefore, this study examined how hormonal changes across the menstrual cycle affect individuals' general anxiety levels and affective ToM abilities. The goal was to understand whether specific cycle phases influence mental health and social-cognitive processing. Anxiety was examined across the menstrual, follicular (with and without menstruation), and luteal phases, while ToM was measured in the mid-follicular and mid-to-late luteal phases.

The following bullet points describe the entire process of doing this research:

• Finding participants: Recruited via flyering, social media, and personal networks. Participants had to speak English, have regular cycles, and not use hormonal birth control. See the project website.
• Setup: Surveys and tests ran in Qualtrics with secure, ID-based data storage. Participants gave consent and completed a short survey on age, diagnoses, and cycle details.
• Daily tracking: I updated an Excel file daily to monitor cycle days, ToM test windows, and missed entries. Participants were removed after completing two cycles.
• Daily surveys: Used STAI-S for anxiety plus questions on menstruation, sleep, stress, and medication. Sent daily at 20:00, open for 48 hours with reminders. Scores were calculated automatically.
• ToM tests: Four affective ToM tests (based on Adult FPRT), two per cycle on days 7-10 and 21-24.
• Data preprocessing and analysis: Merged survey and cycle data to align daily entries with menstrual phases. Missing data were imputed using the LOCF method. Phase-level averages were computed for STAI-S, and ToM responses were manually scored. Anxiety level differences were tested with a repeated-measures ANOVA across phases. ToM scores were compared between mid-follicular and mid-to-late luteal phases using paired-sample t-tests.

Analyzing the data led to the figures shown below. As can be seen in the first bar chart, anxiety varied significantly across phases. Anxiety was reported to be highest during the menstrual phase (low estrogen levels), and lowest during the rest of the follicular phase following menstruation (high estrogen levels). A repeated-measures ANOVA confirmed a significant effect of phase on anxiety. Post hoc comparisons showed that anxiety was significantly higher during menstruation compared to the follicular phase. Luteal levels were moderate without significant differences from other phases.

The ToM bar chart shows no significant differences between phases. The total ToM score reflected overall social understanding, the Empathy ratio measured emotional sensitivity toward others, and the Understanding Inappropriateness ratio measured awareness of socially inappropriate remarks. All remained stable across phases.

Challenges and Reflection

Managing participants was one of the biggest personal challenges for me. Many dropped out during the study, which I knew to expect but still found stressful and discouraging. It required constant follow-up, reminders, and adjustments to keep the study running smoothly. Balancing daily data collection with my own schedule was also demanding, especially when tracking multiple participants manually. It took a significant amount of time each day, but I learned to be patient, plan ahead, and document every step carefully.

A limitation of the study is the small and changing sample size, which reduced statistical power. Self-reported cycle data and the absence of hormonal measurements may also have affected precision. Despite these hurdles, the project taught me how to manage real-world data, handle uncertainty, and stay resilient throughout the challenges of human-subject research.

Abstract

Research on how hormonal changes across the menstrual cycle influence anxiety and social cognition remains limited. This study examined whether naturally cycling individuals experience varying levels of state anxiety and affective Theory of Mind (ToM) performance across menstrual phases. For two full menstrual cycles, participants completed daily surveys measuring state anxiety using the State-Trait Anxiety Inventory. They also completed a total of four affective ToM tests based on the Adult Faux Pas Recognition Test, conducted during the mid-follicular and mid-to-late luteal phases. A repeated measures ANOVA showed a significant difference in anxiety between menstrual phases. The mean anxiety score was highest during menstruation and significantly greater than during the follicular phase. The lowest anxiety was found in the follicular phase excluding menstruation. No significant difference was observed between the luteal phase and the other phases. The paired samples t-test showed no significant difference in affective ToM performance between the mid-follicular and mid-to-late luteal phases. This suggests that the menstrual cycle does not influence affective ToM performance. In conclusion, the findings indicate that the menstrual cycle affects anxiety levels but not affective ToM. However, the small sample sizes (n = 20 for anxiety, n = 17 for ToM) and phase predictions based on self-reported cycle data limit the generalizability of these results. The findings contribute to closing the data gap on how the menstrual cycle affects everyday anxiety fluctuations and take a step toward exploring potential phase effects on social cognition.