Based on the theoretical framework of Risk Perception-Efficacy Appraisal-Behavioral Response, the study aims to explore the group heterogeneity in high-temperature health risk perception and behavioral responses among college students in Chongqing, so as to provide a scientific basis for implementing differentiated health interventions.

A multi-stage cluster sampling method was used to select 856 students from five universities in Chongqing.Data were collected using a validated questionnaire.One-way analysis of variance and independent samples t-test were used to analyze individual differences, Pearson correlation analysis was employed to examine relationships between variables, multiple linear regression was used to identify influencing factors, a structural equation model was constructed to validate the theoretical pathways, and the Bootstrap method was applied to test mediating effects.

In the risk perception dimension, the severity score of high-temperature health hazards among college students (3.28±0.89) was higher than that of susceptibility score (2.94±0.93).Efficacy appraisal showed that the response efficacy score was the highest (3.91±0.81).In behavioral responses, adaptive behaviors were most prominent (5.43±2.75), while emergency preparedness behaviors were the lowest (2.71±1.33).The structural equation model validated the pathway of "threat appraisal → efficacy appraisal → behavioral response" (χ²/df=2.49, RMSEA=0.05).Self-efficacy played a fully mediating role between threat appraisal and behavioral response, with a mediation effect value of 0.10(95%CI=0.02-0.19).K-means cluster analysis categorized the subjects into three groups, with the core barriers to behavior being economic constraints (22.3%), lack of motivation (34.8%), and insufficient cognition (34.1%), respectively.

Decision-making regarding high-temperature health behaviors among college students follows psychological pathway of "cognition-appraisal-action" with self-efficacy serving as a key mediating variable driving behavioral change.Targeted interventions should be implemented for groups with different characteristics.