2020-08-122024-05-15https://scholars.lib.ntu.edu.tw/handle/123456789/664888摘要：背景：情緒知覺 (emotion perception, EP) 為人們藉由外顯刺激（如臉部表情或聲調），以推測他人情緒狀態之能力。通常而言，EP包含7種情緒狀態（快樂、悲傷、生氣、厭惡、害怕、驚訝及平靜），並以視覺（如臉部表情辨識）及聽覺（如聲調情緒辨識）資訊為主要之接收/判讀途徑。思覺失調症患者常有中至重度之EP缺損，影響其精神症狀、社會功能及生活品質。然而，現有之EP測驗有6項缺失（未完整包含7種情緒、未完整評估視覺及聽覺EP能力、未詳盡提供個別向度分數、未校正受試者性別影響、缺乏以華人為刺激物之測驗，以及心理計量特性大多未知），嚴重影響施測效能及結果解讀。此外，由於EP向度數繁多（7向度，視/聽覺途徑），且信度及效率不易兼具，故完整且精準之測驗，很可能題數過多，限制其可行性。電腦適性測驗 (computerized adaptive testing, CAT) 可挑選並僅施測與受試者能力相符之題目，以提升施測效率，但未犧牲信度。CAT可藉由多向度羅序模型 (multidimensional Rasch model) 為基礎，可同時評估多種能力、運用向度間相關以提升精準度，且可校正性別偏差之優勢，極具潛力可改善前述困境。 目的：本研究之目的為發展適用於思覺失調症患者之視覺及聽覺EP之CAT (Computerized Adaptive Testing of Facial and Voice Emotion Perception Test, CAT-FIVE)，並驗證其心理計量特性，包含：再測信度、隨機測量誤差、練習效應、建構效度、生態效度、已知族群效度及反應性。 方法：本研究預計以3年二階段完成：（一）以3步驟發展CAT-FIVE：(1) 發展CAT-FIVE題庫：先自台灣地區華人情緒與相關心理生理資料庫之「專業表演者臉部表情常模資料（視覺）」及「基本情緒聲調資料庫（聽覺）」，分別挑選7種情緒之照片及聲調資料做為候選題，並測試於至少200位患者及100位健康成人。於刪除模型適配度不佳與具性別差異功能函數 (differential item functioning, DIF) 之題目後，剩餘題目組成CAT-FIVE題庫；(2) 挑選CAT-FIVE終止條件：以模擬分析，挑選兼具高信度及高效率之終止條件，以最佳化CAT-FIVE之施測效能；(3) 建構CAT-FIVE之施測介面：依前述結果發展CAT-FIVE電腦施測平台。（二）驗證CAT-FIVE之心理計量特性：招募至少160位患者，接受CAT-FIVE及社會功能效標評估共2次（間隔4週），以症狀穩定者之資料，檢驗CAT-FIVE之再測信度、隨機測量誤差、練習效應及生態效度；並以症狀改善者之資料，驗證其反應性。 預期結果：我們預期CAT-FIVE可提供完整（包含7種情緒與視/聽覺途徑）、詳盡（提供個別途徑/情緒向度之分數）、精準（平均信度 &#8805; 0.90）、有效（具個別單向度、能呈現病患與健康成人應有之差異，且評估結果能反應個案真實生活功能）、敏銳（能偵測個案之EP功能改變）、快速（施測時間短，< 10分鐘），且不受練習效應及性別偏差影響之EP評估。因此，CAT-FIVE將有助於提升臨床及研究人員評估患者EP功能之效能，以利規劃個別化療程以及追蹤治療成效。 本研究創新之處有三：(1) CAT-FIVE將是國際間第一個同時評估視覺及聽覺EP之CAT。(2) CAT-FIVE可克服傳統測驗之缺失，提供快速、完整、詳盡、準確、有效且敏銳之EP評估。(3) CAT-FIVE之發展可做為發展/改良其它評估工具之參考，以改善現有測驗之不足。<br> Abstract: Background: Emotion perception (EP) is an ability to recognize others’ emotional status through observable information (e.g., facial expression or voice tone). The EP conceptually contains 7 basic emotions (happiness, sadness, anger, disgust, fear, surprise, and calm) that can be recognized through information from visual (facial emotion recognition) and voice (voice emotion recognition) channels. Patients with schizophrenia tend to have moderate to severe deficits of EP, which affect their symptom severity, social function, and quality of life. However, the commonly used EP measures have at least 6 weaknesses (measuring only parts of basic emotions, assessing EP based on only visual or voice channel, lack of individual score for each emotion, possible biases due to examinees’ gender, lack of stimuli from Asian people, and unknown psychometric properties), limiting its score interpretations. In addition, given the EP contains numerous domains (7 emotions and two channels) and the difficulty of achieving both high reliability and efficiency, an ideal EP assessment is very likely to be lengthy with poor utility. Computerized adaptive testing (CAT) can administer only the items appropriate to patients’ ability, which can improve efficiency without scarifying reliability. Moreover, CAT is operated based on multidimensional Rasch model, which can further improve reliability and efficiency by considering inter-domain correlations and control the biases due to examinees’ gender. Accordingly, CAT appears promising to solve the aforementioned challenge. Purposes: We aim to develop a computerized adaptive testing assessing facial and voice emotion perception (CAT-FIVE) and examine its psychometric properties, including test-retest reliability, random measurement error, practice effect, construct validity, ecological validity, known-groups validity, and responsiveness. Methods: This 3-year study contains two phases. First, we will develop the CAT-FIVE with 3 steps. (1) Forming the item bank. The candidate items will be selected from opened databases and will be validated on at least 200 patients with schizophrenia and 100 healthy adults. After removing misfit items to multidimensional Rasch model and considering the biases due to examinees’ gender, the remaining items will be used to form the item bank. (2) Determining the best set of stopping rules. To optimize the reliability and efficiency, 16 candidate sets of stopping rules will be proposed and compared to select the best one with both high reliability and efficiency for the CAT-FIVE. (3) Constructing the administration system of the CAT-FIVE. Second, we will examine the psychometric properties of the CAT-FIVE in at least 160 patients. Expected results: We expect that the CAT-FIVE will provide comprehensive (7 emotions with visual and voice channels), informative (individual score for each emotion), valid (fitting well to multidimensional Rasch model, demonstrating different scores for patients and healthy adults, being able to infer patients’ real-life performances), responsive (sensitive to examinees’ changes of EP levels), unbiased (no biases due to examinees’ gender), and efficient (< 10 minutes) EP assessments. Accordingly, the CAT-FIVE can effectively improve the efficacy of EP assessments in both clinical and research settings. Innovation: This study has 3 innovations. First, the CAT-FIVE will be the first CAT assesses both visual- and voice-based EP. Second, the CAT-FIVE can overcome the weakness of the traditional EP tests and provide comprehensive, reliable, valid, unbiased, responsive, and efficient EP assessments. Third, the development of the CAT-FIVE can be an example of using advanced measuring techniques to develop or refine assessments.思覺失調症臉部情緒辨識情緒聲調辨識心理計量特性電腦適性測驗schizophreniafacial emotion recognitionvoice emotion recognitionpsychometric propertiescomputerized adaptive testing