This thesis applies multinomial logistic regression to analyze data from the Customer Satisfaction (CSAT) Survey within the Human Resource function at PT X. The CSAT Survey measures employee satisfaction with specific services received by PT X employees. Currently, the business process primarily considers quantitative feedback, often overlooking qualitative feedback. When qualitative feedback is considered, a phenomenon known as score-sentiment mismatch can occur, where the numerical score does not accurately reflect the actual satisfaction level. This study aims to explore whether demographic factors significantly influence this phenomenon and to examine commenting behavior, specifically the tendency to provide comments in the CSAT survey. The demographic data analyzed includes gender, marital status, age group, region, and team size. The analysis revealed that gender, generation group, and region significantly contribute to inconsistent score-sentiment predictions in the CSAT survey. Males, Baby Boomers, G