本研究為以影片示範三級教學模式之初級介入，係以過去文獻為基石，奉現今研究品質指標為圭臬，結合輔助科技之趨勢，發展具社會效度之初級影片示範教學，並以洗手技能為例教導臺灣與美國自閉症幼兒生活自理技能。洗手係生活自理技能之一，亦為社會重要技能，然而，國內外相關嚴謹之研究報告付諸闕如，教學大多仰賴高資源及高支持量的密集成人提示或高科技精密儀器，社會效度多不明確，難以符合實務教學之需求。因此，本研究針對自閉症幼兒，使用最具高實證基礎的影片示範教學為唯一的介入元素，並採用三層級中所需資源和支持量最少之初級介入；研究設計採多重基準線跨研究對象設計，並參考學術界已有共識之單一受試研究品質指標、社會效度指標及影片示範教學過程特徵與影片製作要素。本研究對象分處臺灣和美國，共有五位三至六歲自閉症男性幼兒符合條件並全程參與。研究者或蒐集資料的教師記錄研究對象看完影片後洗手總步驟數與每一步驟完成情形。結果顯示，本研究具有良好的觀察者間一致性、過程效度、內在效度與外在效度，五位幼兒中有四位在介入後有顯著且穩定之行為改變，而且維持成效良好及有部分類化效果，顯示初級影片示範教學能有效教導自閉症幼兒學會洗手技能。本研究發現，此初級介入適合有功能性語言且會簡單句子，無影響學習的自我刺激功能之行為或固著物的自閉症學習者。本研究亦討論初級影片示範教學介入適合的對象、未來實施三層級影片示範教學模式應用與成本效益、社會效度等如何落實於實務之議題及未來研究方向，以供相關研究與實務專業人員參考。

Purpose: This three-tier video modeling intervention model aimed to explore the effects of a primary tier generic video modeling to teach hand washing skill to children with autism in Taiwan and in the United States. Self-help skill, such as hand washing, is a socially important skill. However, autism literatures on teaching such skill are lack of rigor and social validity. Besides, most existing literatures relied on labor-intensive or high technology equipment which involved high resources and support. This study utilized video modeling, the most high evidence level of the autism evidence-based practices, and the least amount of resource and support involved primary tier of the Three Tier Video Modeling Intervention Model. Method: This study abided the academic consensus of single subject design quality indicators, social validity quality indicators, video modeling procedural characteristics and video components to develop a generic video for children with autism to learn hand washing. A multiple baseline across participants design was used to teach five 3 to 6 year-old male children with autism who met the inclusion criteria to independently wash their hands. The intervention protocol was rather easy to follow. The participant children were asked to view a 3-minute-video demonstrating the 9 steps of hand washing from a peer model and provided the participants the opportunity to practice the skill in 40 seconds. The total numbers and each of the independent performed steps were recorded. Results: The results showed that 4 of the 5 participants significantly and stably changed their skill behavior when and only when intervention was implemented. Maintenance and partial generalization effect were also observed. The validities (internal validity, external validity, and social validity) and reliabilities (inter-observer agreement and fidelity) of this study were high. Discussions and implications: The following themes and the implications for future research were discussed: a) the observed phenomenon of delay effects; b) a dialogue with Rosenberg’s primary tier video modeling study; c) none-responders with self-stimulus function behavior and the absence of functional expression in sentences; d) the application of the three-tier video modeling intervention model and the cost-effectiveness analysis; and e) the social validity and its relationship to bridging the research and practice.