Human behaviour critically shapes infectious disease transmission, yet many local-scale models neglect behavioural dynamics. We address this gap by simulating how Shared Social Identity (SSI) and the observed adherence of others affect pathogen exposure in crowd settings. We utilize the pedestrian simulation software Vadere, which integrates an aerosol transmission model. In this model, infectious agents emit aerosol clouds that susceptible agents can inhale. Mask-wearing mitigates this process by reducing both the pathogens exhaled by infectious agents and the amount inhaled by susceptible ones. We simulate a ticket-checkpoint scenario with 100 susceptible agents. First, a "role-model group", (20 agents) passes, either fully adhering or fully not adhering to mask-wearing. Next, an "imitating group" (80 agents), including one infectious agent, enters. These agents identify to a certain degree with the role-model group, measured through a parameter called shared social identity (SSI). Based on SSI and observed adherence of the other group, we sample each agent's mask-wearing probability from survey data collected as part of the UK Government’s Events Research Programme. Results show that when the role-model group wears masks, adherence in the imitating group remains high regardless of SSI, resulting in a very few highly exposed agents (around 1). However, when the role-model group is non-compliant, SSI is decisive. Agents with low SSI tend to ignore the non-compliant role-model group and continue to adhere, keeping the number of highly exposed agents limited (around 4). In contrast, agents with high SSI align with the non-adherence of the role models, significantly increasing the number of highly exposed agents (around 12). These findings indicate that shared social identity amplifies the effect of observing others' non-adherence, highlighting the importance of considering social dynamics when designing interventions to reduce exposure at crowded events.