本研究的目的是同時採用驗證性因素分析以及羅序模式以驗證職能治療綜合評量表於台灣精神障
礙者之心理計量特性。本研究之個案來自於北部某醫學中心的精神科病房，共321 位。包含142 位男性(44.2%)，平均年齡為30.2 歲(標準差＝9.43)。本研究採用的職能治療綜合評量表為全國精神科職能治療師經常使用之功能評估。研究資料以WINSTEPS 3.56、AMOS7.0 及SPSS 17.0 軟體進行統計分析。結果顯示社交行為次量表符合單一建構效度，但是其精確度較差。項目的難易度排序，精神障礙者在3 個次量表(一般行為次量表、社會行為次量表、工作行為次量表)中最難的題目分別為「活動量」、「團體中角色」及「細動作協調」。社會行為次量表於不同性別、教育及診斷族群間並無任何系統性誤差，一般行為及工作行為次量表於教育程度或性別上各有二題具有項目難度差異性。本研究結果顯示，未來需繼續修訂一般行為以及工作行為次量表，以期符合單一建構之要求。應增加題目以改善社會行為次量表之精確性。

The aim of this study was to examine the construct validity, accuracy, and differential item functioning (DIF) of the Comprehensive Occupational Therapy Evaluation (COTES) with Rasch measurement model and confirmatory factor analysis (CFA). A total of 321 persons with mental illness were recruited from an inpatient unit of a teaching hospital in northern Taiwan. There were 142(44.2%) males and 179(55.8%) females. The mean
age of the subjects was 30.2 years (SD=9.43). The WINSTEPS 3.56, AMOS 7.0 and SPSS 17.0 were used for ata analysis. The results showed that only the social behavior subscale fitted with the model requirement for
both CFA and Rasch model. But the accuracy of the subscale was not good. The most difficult endorsed items for three subscales were the activity level (general behavior subscale), group roles (social behavior subscale) and fine motor coordination (task behavior subscale). There was no DIF item in the social behavior subscale; two DIF items were found for the general behavior and task behavior subscales. Therefore, there are needs to revise the general behavior and task behavior subscales to fit the requirement of unidimensionality. More items are needed in the social behavior subscale to increase its accuracy.analysis