發展思覺失調症患者之工作社交技巧電腦化多階段角色扮演測驗

Development of a computerized multistage role-play testing of employment social skills in patients with schizophrenia

黃怡靜(Yi-Jing Huang)

指導教授 ： 謝清麟

國立臺灣大學/醫學院/職能治療研究所/博士(2018年)

https://doi.org/10.6342/NTU201800786

摘要

背景與目的：工作社交技巧 (employment social skills) 為影響思覺失調症患者求職與工作適應的主因之一。完整、精確且快速地評估患者之工作社交技巧，可協助臨床人員擬訂職業復健計畫與呈現復健成效。角色扮演測驗為當今最為推薦之工作社交技巧評估方式，然而目前缺乏內容完整、對於施測者負擔小、具備良好心理計量特性的工作社交技巧角色扮演測驗。此限制嚴重影響專業人員快速且精準地掌握患者問題，使得社交技巧訓練與職業重建的成效難以進一步提升。為克服現有測驗之限制，本研究之目的有二：(1) 發展一套適用於思覺失調症患者之完整、快速暨精準的「工作社交技巧電腦化多階段角色扮演測驗」(COmputerized Multistage role-play Testing of Employment Social Skills, COMTESS)。(2) 初步驗證 COMTESS 之心理計量特性（含信度、收斂效度與已知族群效度）與施測效率。
方法：研究者以二階段研究分別發展與驗證COMTESS。第一階段中，研究者依據以下六步驟發展COMTESS：(1) 文獻回顧：彙整過去研究中常見的工作社交技巧相關行為與觀察指標。(2) 編製初版題庫：依文獻彙整結果編寫社交情境劇本與計分量尺。(3) 專家審核題庫：邀請專家評分初版題庫中各題情境之重要性並提供修改建議。(4) 建置施測介面：以投影片放映檔設計施測介面與施測平台。(5) 題庫驗證：施測題庫於慢性思覺失調症患者，並以單向度羅序分析驗證各向度題目之模型適配度。(6) 選題並建構COMTESS：以多向度羅序分析估計題庫難度與患者能力參數，並挑選分派題與高/低難度測驗組以構成COMTESS。
第二階段中，研究者使用患者的能力估計標準誤 (standard error) 計算COMTESS之個別與平均信度。收斂效度以COMTESS能力值與個人與社會功能量表 (Personal and Social Performance Scale, PSP) 總分之相關性驗證。已知族群效度則以蒙特利爾認知評估 (Montreal Cognitive Assessment) 為效標，使用t檢定與Cohen’s d比較有/無明顯認知障礙組間COMTESS能力之差異顯著性與大小。COMTESS之施測效率以患者作答題庫所需之時間估算。
結果：第一階段中，研究者發展94題初版題庫，其中「個人社交技巧」向度與「職務相關社交技巧」向度各有47題，每題的計分量尺為0~4分。初版題庫經28位精神復健與職業重建領域之專家審核，共刪除14題較不重要或不適切的題目，保留80題題目，其中2題作為練習題，其餘78題作為正式測驗題。共144位思覺失調症患者參與題庫驗證。題庫中所有題目皆符合羅序模型之假設。然而，原計分量尺之階難度順序錯置，故研究者將每題之計分量尺修改為0~2分。研究者自二向度題庫中，分別挑選1題分派題與高難度及低難度測驗組各4題構成COMTESS（共18題）。每位患者僅須作答2題分派題與二向度各4題高/低難度測驗組，共10題。
第二階段初步驗證COMTESS之結果顯示COMTESS二向度皆具有可接受的個別信度（二向度皆有97.8%的患者信度 ≥ 0.70）與良好的平均信度 （二向度皆為0.80）。COMTESS二向度能力與PSP總分之相關係數皆為0.41，顯示COMTESS之收斂效度良好。在有/無明顯認知障礙之患者間，COMTESS二向度能力皆有顯著差異 (p < 0.001) 且差異量大 (d = 0.91~0.92)，顯示COMTESS具有良好的已知族群效度。COMTESS之施測時間預估僅需9.9分鐘，顯示COMTESS之施測效率高。
結論：COMTESS為第一個針對思覺失調症患者之工作社交技巧所發展的電腦化角色扮演測驗。COMTESS完整包含「個人社交技巧」向度與「職務相關社交技巧」向度。COMTESS利用電腦化多階段測驗之優勢，每位患者僅須作答10題，約12.8%的題庫題數（78題），施測時間僅需近10分鐘，且COMTESS具備良好信度、收斂效度與已知族群效度，顯示COMTESS可完整、快速且精準地評量思覺失調症患者之工作社交技巧，且可有效區辨不同認知程度患者間的工作社交技巧差異。此外，透過電腦化施測，可大幅減少施測者訓練時間與施測負擔，並增加臨床施測的可行性。

關鍵字

思覺失調症 ； 工作社交技巧 ； 電腦化多階段測驗 ； 角色扮演 ； 心理計量

並列摘要

Background and purposes: Employment social skills can be defined as the capability to interact appropriately with others or deal with social situations that are required at work. For patients with schizophrenia, a lack of employment social skills has been recognized as one of the major barriers in seeking employment, adapting to a new job, and learning job-specific skills. Thus, accurate, precise, and efficient assessments of employment social skills is the basis for successful vocational rehabilitation. However, there are no content-comprehensive, low-burden, and psychometrically sound measures of employment social skills. Such a limitation has restricted clinicians in efficiently capturing the problems of patients, thereby compromising the treatment effects of social skills training and vocational rehabilitation. To ameliorate the drawbacks of the existing measures, this dissertation had 2 purposes: (1) to develop a content-comprehensive, efficient, and precise measure for assessing employment social skills in patients with schizophrenia; i.e., the COmputerized Multistage role-play Testing of Employment Social Skills (COMTESS); (2) to preliminarily examine the reliability, convergent validity, known-group validity, and efficiency of the COMTESS.
Methods: This dissertation contained 2 stages. Stage 1: the development of the COMTESS was divided into 6 steps: (1) literature review; (2) development of the draft item bank; (3) expert review of the draft item bank; (4) construction of the test interface and platform; (5) Rasch analysis of the item bank; (6) selection of routing items and hard/easy modules to establish the COMTESS.
Stage 2: we used the standard errors for patients’ ability estimates to calculate the individual reliabilities and average reliability of the COMTESS. We investigated the correlations between the scores of the COMTESS and the Personal and Social Performance Scale (PSP) to examine the convergent validity of the COMTESS. To examine the known-group validity, we divided the patients into groups with and without significant cognitive impairments by the Montreal Cognitive Assessment. We further used independent t test and Cohen’s d to test the significance and amount of differences in the COMTESS scores between the two groups. The efficiency of the COMTESS was estimated by the average administration time per item multiplied by the number of items.
Results: In Stage 1, we developed a total of 94 items equally divided into the domains of “personal social skills” and “task-related social skills”. Each item was rated based on patients’ both non-verbal and verbal expressions by a 5-point scale (0~4 points). The draft item bank was reviewed by 28 experts. Fourteen items that were less important or inappropriate were deleted, and 80 items remained in the item bank. Thereafter, the item bank were examined in a total of 144 patients with schizophrenia using Rasch analysis. The items fitted the model well. However, the rating scale was modified as a 3-ponit scale (0~2 points) due to the reversed step difficulties. Finally, we selected 1 routing item and 4 items in each of the hard and easy modules, from each domain of item bank, to establish the COMTESS (a total of 18 items).
In Stage 2, the average reliabilities of the two domains of the COMTESS both were 0.80, and 97.8% patients had reliabilities of ≥ 0.70 in both domains. These results indicate that the COMTESS had sufficient reliability. The correlation between the scores of the COMTESS and the PSP was 0.41, which indicates good convergent validity. Significant (p < 0.001) and large (d = 0.91~0.92) differences in the COMTESS scores were found between the groups with and without significant cognitive impairments, which indicate good known-group validity. The administration time of the COMTESS was about 9.9 minutes, which indicate that the COMTESS is an efficient test.
Conclusion: The COMTESS may be the first computerized multistage role-play testing of employment social skills for patients with schizophrenia. The COMTESS comprehensively contains two domains of “personal social skills” and “task-related social skills”. Our results show that the COMTESS has sufficient reliability, convergent validity, and known-group validity. Moreover, the administration time of the COMTESS is only around 10 minutes. These findings imply that the COMTESS is able to comprehensively assess employment social skills in a timely and precise fashion and to discriminate the levels of employment social skills among patients. Besides, computerized testing can largely reduce the training time and burden of the administrators and increase the feasibility.

並列關鍵字

schizophrenia ； employment social skills ； computerized multistage testing ； role play ； psychometrics

參考文獻

勞動部統計處（2018）。105年身心障礙者勞動狀況摘要分析。台北市：勞動部統計處。取自 https://www.mol.gov.tw/statistics/2462/19476/19060/。

衛生福利部統計處（2018）。身心障礙人數按類別及年齡別分。台北市：衛生福利部統計處。取自 https://dep.mohw.gov.tw/DOS/cp-2976-13825-113.html。

Aaronson, N., Alonso, J., Burnam, A., Lohr, K. N., Patrick, D. L., Perrin, E., & Stein, R. E. (2002). Assessing health status and quality-of-life instruments: Attributes and review criteria. Quality of Life Research, 11, 193-205.

American Psychiatric Association. (2013). Desk reference to the diagnostic criteria from DSM-5™. Arlington, VA: American Psychiatric Publishing, Inc.

Andrich, D. (1978). A rating formulation for ordered response categories. Psychometrika, 43, 561-573.