Beyond Sentiment: Evaluating Gendered Language Using Zero-Shot Text Classification
Abstract: Situational Judgment Tests (SJTs) present hypothetical job-related situations to assess judgment and decision-making skills. Using zero-shot text classification, we replicated previously established findings on sentiment effects and further explored the influence of gendered language on participant responses in SJTs. Our study demonstrates that negative sentiment in action statements lowers effectiveness ratings and increases response variability. Contrary to gender schema theory, we found no evidence that gender-congruent phrasing led to higher effectiveness ratings. These findings underscore the potential of zero-shot text classification for refining SJT item development and mitigating unintended biases.
Paper Type: Short
Research Area: NLP Applications
Research Area Keywords: Ethics, Bias, Fairness, NLP Applications, Sentiment Analysis, Human-Centered NLP
Contribution Types: Reproduction study, Approaches to low-resource settings, Data analysis
Languages Studied: English
Submission Number: 1747
Loading