据自闭症儿童看护人报告，缺乏角色扮演活动有助于提高这种技能。Sancho, Sidener T.， Reeve & Sidener D.，(2010)发现自闭症儿童能够成功地模仿在不同场景中展示不同角色角色的模型。这是通过视频建模获得的，通过该方法，一个模特展示不同技能的动作被记录下来，并展示给孩子们，然后观察他们试图准确地复制脚本。然而，本研究的不足之处在于，它没有对不同情况下的play角色进行概括。相比之下，Dupere, MacDonald和Ahearn(2013)进行的一项研究通过量化参与者在一组玩具中与未经训练的角色玩耍的情况，解决了这一局限性。这是通过“合适的循环”的实现来完成的，它涉及到模型演示角色，这些角色适用于多个字符，但只适用于一个字符。参与者模仿这些循环语言或动作的次数被用来推断自闭症儿童展示角色扮演刺激泛化的能力。然而，这两项研究都没有承认样本量小的局限性(一项有3名参与者，另一项只有2名参与者)，这使得很难将结果推广到更广泛的人群。不管怎样，视频建模的使用为这两项研究都提供了优势，因为它允许视频片段被多次播放，这可以节省时间、模型和资源。更重要的是，这种方法被证明更适合那些在分散注意力的环境中难以集中注意力的自闭症儿童，而且视频中包含了第一人称视角——描绘成人的手在玩玩具——以及背景声音中的语音脚本。(插入引文说孩子会分心)。两种方法都是一致的，即组成一个基线，让孩子们玩没有强化或视频提示的玩具，作为一个标准来比较视频显示后所描绘的仿制品。桑丘等人(2010)报告了10次中平均6.3次的正确模仿，而基线测试期间没有脚本语言。同样，在训练后的阶段，对未经训练的字符进行模仿的情况也有所增加
A lack of role-playing activities reported by caretakers of autistic children facilitates the need for an increase of this skill. Sancho, Sidener T., Reeve & Sidener D., (2010) found that autistic children were able to successfully imitate models displaying roles of various characters in distinct settings. This was obtained through video modelling, whereby actions of a model demonstrating different skills are recorded and shown to children, who are then observed attempting to replicate the scripts accurately. However, this study was disadvantaged in that it did not accommodate for generalisation of the play roles in different situations. In contrast, a study conducted by Dupere, MacDonald and Ahearn (2013) addressed this limitation by aiming to quantify the occurrence of participants playing with untrained characters amongst a set of toys. This was done through the implementation of ‘suitable loops’, which involved the model demonstrating roles that are applicable to multiple characters, only on one character. The number of times participants imitate speech or actions of these loops was used to infer the ability of autistic children to demonstrate stimulus generalisation for role play.However, both studies fail to acknowledge the limitation of a small sample size (one with 3 participants, and one with only 2), which makes it difficult to generalise the results to the wider population. Regardless, the use of video modelling provides a strength for both studies, as it allows the footage to be played multiple times, and this can save time, efforts of models and resources. More importantly, this method was proven to be more suitable for autistic children who struggle to pay attention to a particular procedure in distracting environments, and the videos contained first person perspective – depicting adult hands playing with toys – and vocal scripts in the background sound. (INSERT CITATION THAT SAYS KIDS GET DISTRACTED).Both methods were consistent, composing a baseline with children asked to play with the toy sets with no reinforcements or video prompts, to use as a standard to compare the imitations portrayed after video is shown. An average of 6.3 correct imitations out of 10 was reported by Sancho et al. (2010), compared with no scripted verbalisations during the baseline session. Similarly, an overall increase in the usage of imitation for untrained characters was reported in the post-training sessions.