Moral judgments in online discourse are not biased by gender

Abstract

Abstract The interaction between social norms and gender roles prescribes gender-specific behaviors that influence moral judgments. While previous work has demonstrated the existence of gender-bias in judgments, these studies are mainly based on controlled experiments that may not reflect real-world decision-making processes. Here, we study how moral judgments are biased by the self-disclosed gender of the protagonist of a story. Using data from , a Reddit community with 17 million members who share first-hand experiences seeking community judgment on their behavior, we employ machine learning techniques to match stories describing similar situations that differ only by the protagonist’s gender. We find no direct causal effect of the protagonist’s self-disclosed gender on the received moral judgments, except for stories about “friendship and relationships”, where male protagonists receive more negative judgments. Our findings complement existing correlational studies and suggest that gender roles may exert greater influence in specific social contexts. These results have implications for understanding sociological constructs and highlight potential biases in data used to train large language models.