Abstract: Objective To test the reliability and validity of the Professional Grief Reaction Scale among nursing interns based on Item Response Theory (IRT) and Classical Test Theory (CTT). Methods A convenience sample of 450 nursing students with internship experience was surveyed by using the Professional Grief Reaction Scale. Both IRT and CTT were used to evaluate the scale’s reliability and validity. Results The final scale consisted of two dimensions: Emotional Response (14 items) and Physical and Cognitive Response (16 items). IRT indicated that the information-weighted mean square fit statistic (Infit MNSQ) and the unweighted mean square fit statistic (Outfit MNSQ) for all items ranged from 0.840 to 1.400. The average difference between item difficulty and individual ability was 0.750 logits. The threshold differences between the options "somewhat agree" and "fairly agree" were small. Differential item functioning values for interns of different genders and ages were all below 0.640 logits. The reliability of individuals and items ranged from 0.840 to 0.980, and the separation indices ranged from 2.290 to 6.860. According to CTT, the cumulative variance contribution rate of the two factors was 52.472%; the model fit met requirements, and the convergent validity was reliable. The Cronbach’s alpha coefficient for the scale was 0.931, and for the two dimensions, that was 0.865 and 0.955, respectively. Conclusion The revised Professional Grief Reaction Scale demonstrates good psychometric properties and can be used to assess professional grief among nursing interns.

Key words: professional grief, nursing intern, reliability and validity, Item Response Theory, Classical Test Theory