Previous research compares individuals at different levels of their environment—gender, nationality, or socioeconomic status—assuming these categories shape emotional experiences. Behind these findings lies an implicit theoretical claim: at some level of social organisation, shared norms and affordances become strong enough that emotional experiences begin to cohere. This study proposes a bottom-up approach to empirically test how individuals naturally group in response to a period of prolonged global stress. Using clustering, a robust machine learning tool, we grouped individuals based on their patterns of emotional experiences. The data, spanning 38 countries and measuring 20 emotions during the first wave of the COVID-19 pandemic, allowed us to identify meaningful clusters. Sociocultural factors (cultural values, country-specific factors, SES and gender) were used to predict cluster membership, to determine where in the sociocultural landscape emotions are most influenced.

We identified five distinct groups of individuals based on their emotional experience patterns, with SES emerging as the strongest predictor. We also identified regional similarities which could not be explained with our chosen predictors. These findings offer empirical support for a more nuanced view of emotional experience, one that acknowledges both shared cultural patterns and the significant influence of individual lived realities. Methodologically, this study demonstrates a novel, data-driven approach to understanding emotional experience beyond predefined demographic categories, offering a replicable pathway for future research on emotional variation across social contexts.