• No results found

Chapter 3: An Inventory of Existing Representations of Spectrum Conditions . 43

3.5 Conclusion: Film Inventory


neurotypical characters is worthwhile as the ASD character is a commodity, acting as the key to survival or finding the truth.

The genre of a film has relevance to both the box office sales and the influence on recall of the film (Redfern, 2012). If the scenes with ASD portrayals found in horror, thriller and action films (n=9) are more memorable due to their emotive nature, they could also be associated with negative feelings or attitudes towards people on the spectrum. The demonic association of autism characteristics in films and references to ‘being in their own world’ that are often heard in news media or documentary descriptions could assimilate into images of autism that actually reduce the public’s openness or positive attitudes towards affected persons. With the high entertainment value and large box office sales of films in these genres, there exists a realistic probability that using a disorder to drive the plot could perpetuate negative stereotypes.



Matching the industry’s motivations with authentic portrayals could increase the production of ASD representations that have the potential to dispel the public’s misconceptions about ASD and thereby avoid reducing the complexity of autism affectedness to a few emotive snapshots (Haller, 2000). Since media both influences and reflects societal beliefs, it is necessary to understand how productions contribute to ‘education’ about autism for the lay viewer. With this in mind, I will next explore both the visual and aural associations to the term autism as they are presented in the 15 films that identified a character as



Chapter 4 Overview - Characters: the Traits they Portray and the Things They Say about Autism Spectrum Conditions

This chapter is comprised of two parts. Chapter 4A describes the analysis of character traits and 4B explores the character dialogue relative to the autism spectrum. This chapter examines how cultural artefacts like film communicate new information about autism. The information presented through film then

‘scaffolds’ (Vygotsky, 1986) the adoption of homogeneous (one-dimensional) and confining diagnostic characteristics for the inexperienced viewer.

Subsequently, using a constructivist lens the quality and quantity of autism-related characteristics portrayed by film characters are explored, and the relevance of character dialogue to the construction of new knowledge for inexperienced viewers is examined.



Chapter 4A: Accuracy of Filmic Character Representations of Autism Conditions

4.A.1 Focussed Literature Review: Accuracy

Stereotyping has consequences for individuals in both personal and practical terms and occurs across a plethora of processes (Duvdevany, Rimmerman, &

Portowicz, 1995). For example, there is debate regarding the stereotyped representation of the ‘normal distribution’ of disability in film, particularly the representation of ASD (Farnall & Smith, 1999; Hartman, 2006; Murray, 2008;

Osteen, 2008; Draaisma, 2006; Baker, 2008). Although there is a small body of literature that discusses how disability and disorder, including ASD, are portrayed in film, the discussion is usually focused on the plot utility of the disorder and the social stigma associated with the representation (Murray, 2008;

Baker, 2008). This debate rests in subjective observation of stereotyped archetypes, and criticisms of the function of autism as a formulaic plot driver (Baker, 2008; Draaisma, 2009; Osteen, 2008; Murray, 2006). In an effort to contribute empirical evidence to the debate, I investigate the authenticity of ASD film portrayals using a scientifically based assessment tool (the Childhood Autism Rating Scale, Second Edition, (CARS2)). This tool is utilised by practitioners to determine the presence and severity of spectrum characteristics and is used in this stage of Study One to conduct an analysis of filmic representations of ASD.

The use of a clinical tool in a social realm, such as film, may seem counterintuitive as these arenas rarely overlap when working with persons affected by autism.

For example, there are limitations arising from the lack of control over influential variables in the practitioner’s social realm. However, although autism may be diagnosed clinically the people on the spectrum exist in the social sphere and therefore there may be educational merit derived from applying a ‘clinical’ tool to a ‘social’ representation. The potential benefit is found in the departure from the controlled, scientific, and manipulable observation of autism in clinical settings wherein film arguably mirrors the highly variable and uncontrolled


‘social presence’ of autism more likely encountered by the inexperienced individual.

Where the degree of the presentation of ASD characteristics is ordinarily determined through the gathering of information from professionals, parents, teachers, and the individual using a variety of tools (e.g., the ADOS, the CAPS, the CARS2), the brief and limited observation afforded by the interaction between the viewer (rater) and the character may be more reflective of ‘real life’.

However, it is recognized that a clinician would likely use a battery of tests to inform diagnosis rather than a singular instrument. Nevertheless, the utilisation of an observational tool, like the CARS2, applied to a filmic portrayal of ASD does not allow for manipulation of stimuli or the control of variables that tend to reflect actual processes. Furthermore, to explore the functionality of an assessment tool, the authenticity of the filmic representation and the influence of previous knowledge, it is necessary to represent both the arena the diagnosis is achieved in (clinical) and the realm in which it exists (social). In this part (A) of this chapter I report the findings that reflect both an education specialist’s and a clinical psychologist’s ratings of filmic portrayals of ASD using the CARS2 instrument. This process provides insight, from both the social and the clinical perceptions, into the ‘individual’ display of ASD characteristics within the filmic context.

4.A.2 Method: Accuracy

Most of the ASD assessment tools are specifically designed to detect and quantify the presence, frequency and severity of specific behaviours to inform a diagnosis according to the characteristics outlined in the DSM-IV; this is usually achieved via a clinical interview. However, for this stage of the study, an assessment tool that relied on observable characteristics was required; thus, the CARS2 instrument was selected. The CARS2 is an observation-based tool where manipulation of stimuli is not essential, thereby allowing for the observation of autism-related characteristics in film.



The CARS2 is composed of 15 behavioural areas or categories, 14 of which relate to diagnostic characteristics found in the DSM-IV (American Psychiatric Association, 1994) and one for ‘overall impression’. There are two versions of the CARS2, with one appropriate for use amongst the general autism population (Standard version) and one appropriate for use amongst the high functioning autism population (High Functioning version) (Schopler, Van Bourgondien, Wellman, & Love, 2010). In addition to the two versions there is a form for parents and caregivers to complete which would provide information across contexts however it would be impractical and ineffectual to adopt the perspective of a character’s caregiver and therefore this was not used for this study. For each of the 15 behavioural categories there is a potential score from one to four, where one represents age-appropriate or no evidence of difficulty, and four represents severely abnormal (Schopler, Van Bourgondien, Wellman,

& Love, 2010). For each behavioural area a mean score is provided that represents the average score expected for the population with a diagnosis of ASD (Schopler et al, 2010).

Most currently available assessment tools (e.g., ADOS) require specific training courses for their implementation or some level of interaction with the individual, family, and professionals involved. Since the goal of this analysis was not to inform diagnosis but to provide objective evidence about the filmic representation of ASD, an intensive clinical tool such as ADOS was deemed neither necessary nor feasible to use.

The CARS2 was selected for this analysis because it can be used by “well versed professionals independently of interviews” (Schopler et al, 2010: page 5), and it adjusts for the age of the assessed individual. The CARS2 is recommended for assessing persons from childhood through adulthood (Department of Human Services, Victoria State Government). In addition, it has been determined to have a 95% inter-rater reliability (Schopler et al, 2010).


In accordance with the instructions for using CARS2, when the DVD descriptor indicated the film portrayed autism, the Standard version of CARS2 was used;

for films that portrayed high functioning autism or Asperger Syndrome, the High Functioning version of CARS2 was used. The categorical CARS2 scores, the diagnosis the film contends it represents, and the normal distribution of these characteristics in the ASD affected population were then compared to determine whether the film authentically represented ASD.

The inclusion criteria for the films selected for rating using the CARS2 were: the film had been released in cinema; the synopsis on the DVD explicitly stated autism, Asperger Syndrome or ASD; and the film had been released prior to December 31, 2010. This resulted in the identification of 15 films (see Table 4.A.1).

The CARS2 instrument was applied to the filmic character portrayals in the 15 selected films. Two raters independently assessed the filmic character portrayals: an education specialist with a background in theory and strategies for supporting students with ASD, and a clinical psychologist with a practice focus on ASD assessment. Both assessors had experience using the CARS2 assessment tool. The raters viewed each film independently and in random order one time. In the instance that more than one character was on the spectrum the main character was rated (for example, Mozart and the Whale-Donald was rated10) Assessment scores were entered into SPSS (SPSS v.17.0 2008) and analysed for inter-rater reliability using the Kappa statistic. The 15 categories on each of the versions of the CARS2 were analysed for outliers.

4.A.3 Results: Accuracy

The overall mean ratings of the educational specialist (ES) and the clinical psychologist (CP) were virtually identical by both category (ES M=3.01, CP M=2.87) and Total raw score (ES M=42.27; CP M=42.41; t=-.153, p> .10). Similar

10 Only one character was rated per film to maintain consistency and limit the possibility of comparative bias (judging one character as more or less able based on the portrayal of the secondary character).



results have been reported for the mean of the ratings by ‘professionals in different disciplines’ in the CARS2 manual (Schopler et al., 2010: page 79). Inter-rater reliability was examined for ratings made by the two independent professionals for the 15 film characters. Consistent with Schopler et al. (2010) an unweighted coefficient kappa was used to determine agreement between raters. Inter-rater agreement of 87.4% was obtained for Total Scores on the CARS2-ST (r=.874, p=.000). The inter-rater agreement increased to 93.5%

(r=.935, p=.002) when foreign language films were excluded from analysis. This agreement is consistent with that reported by Schopler et al (2010: page79).

The weighted kappa estimates were calculated for the pair of raters. This statistic evaluates categorical, absolute agreement between raters across the categories for each item but also ratings values of that are very similar (.5 on either side) (Schopler et al., 2010). The results indicate a median level of agreement of .686 (Table 4.A.1). This indicates a moderate agreement for median level of category scores on the CARS2 Standard version, with 100% agreement for severity grouping (minimal-to-no symptoms of ASD; mild-to-moderate symptoms of ASD;

and severe symptoms of ASD). The Kappa for the category scores of CARS2 HF version for the educational specialist and the clinical psychologist was substantial showing a median level of .651, indicating agreement beyond what would be expected by chance (Viera & Garrett, 2005), again with 100%

agreement for severity grouping. In both cases the Kappa was slightly lower than those reported by Schopler et al (2010) which was .71 for the CARS and .73 for the CARS2-HF.


Table 4.A.1: Median ratings for each rater by CARS2 category

Category Rater 1 Rater 2

Relating 3.5 3.14

Imitation 2.42 2.21

Emotion 3.35 3.21

Body 2.64 2.92

Object 3.14 2.92

Adapt 2.78 2.64

Visual 2.5 3.21

Listen 2.9 2.85

Taste 2.5 2.42

Fear 3 3.2

Verbal 3.57 3.21

Non-verbal 3.5 2.85

Activity 2.21 2

Intellect 3.64 2.92

Overall 3.4 3.21

The films selected, the DVD box descriptor, and the severity ratings according to the CARS2 assessment can be found in Table 4.A.2. The results indicate that 13 of the 15 films scored in the ‘Severe Symptoms of ASD’ group on the CARS2. In terms of the DVD descriptors, 10 used the word ‘autistic’, two used the word

‘autism’, two used the term Asperger Syndrome and one used Asperger Syndrome and autism.



Table 4.A.2: Comparison of contended portrayal and CARS 2 severity score

Film Title DVD Box Descriptor Severity Groupings According to CARS2 Scores

Ben X Mildly autistic Severe

The Black Balloon Autistic brother Charlie Severe

Bless the Child Thought to be autistic Minimal-to-no

Chocolate Young autistic girl Severe

Guarding Eddy Autistic sports jabbering kid

Mild-to-moderate Killer Diller An autistic savant Severe

Mary and Max Asperger’s Syndrome Severe Mercury Rising Nine-year-old autistic boy Severe Molly Autistic, retarded sister Severe Mozart and the

Whale A math savant Asperger’s

Syndrome, a form of autism


My Name is Khan Has Asperger’s Syndrome Severe

Ocean Heaven Childhood autism Severe

Rainman Autistic savant Severe

Silent Fall An autistic child Severe

Snowcake Suffers from autism Severe

The CARS2 reports ‘interpretive categories associated with CARS2-ST or CARS2- HF’, describing the T-score range in terms of “levels of autism-related symptoms compared to those with an autism diagnosis” (Schopler et al, 2010, p 45). The T-score range as identified by Schopler and colleagues (2010, p 45) for “average level of autism-related symptoms compared to those with an autism diagnosis”

is 45-54. In other words, an individual who has a T-score of 48 would be classified as exhibiting autism-related symptoms similar to those expected in the population with an autism diagnosis (e.g., the mid-range of the distribution curve). Figure 4.A.1 shows the T-score range for each film character in relation to the normal distribution curve for the population with an autism diagnosis indicated in the CARS2 manual.


The distribution for filmic representations of autism-related symptoms is skewed to the left, with most films being in the 60-70 range; this suggests that these films are representing ‘very high levels’ or ‘extreme levels’ of autism- related symptoms.

Figure 4.A.2: The difference in median score between raters and expected median by CARS2 category 0

1 2 3 4 5 6 7 8

70+ 60-70 55-59 45-54 40-44 25-39 24

Number of films in each range

CARS 2 T-score range

Film T-score

Figure 2: Indicates Normal Distribution Curve of symptoms for those with Autism Diagnosis in the population Figure 4.A.1: The normal distribution curve for behaviour categories in the actual population with autism versus the distribution of behaviour categories for the filmic population

0 0.5 1 1.5 2 2.5 3 3.5 4

Raters CARS2



An analysis of the median scores by category indicates that the film representations of autism-related symptoms were higher for each CARS2 category than the reported median (Figure 4.A.2), with the exception of the

‘imitation’, ‘visual’, and ‘activity’ categories. The ‘imitation’ challenges and atypical ‘activity levels’ that would be expected for the population within the

‘severe group’ were portrayed below the median score as indicated by the CARS2, whereas the ‘visual’ category was similar to the median score for the population with an autism diagnosis.

4.A.4 Discussion: Accuracy 4.A.4.1 Film representations

The application of a clinical tool to a social and non-manipulable presentation of autism has both merits and challenges. The character portrayals were found to be more severe in their display of autism symptomology than would be expected in the average autism affected population as assessed using the CARS2 (Schopler et al, 2010; Gillberg, 1991). This obscured representation of the autism spectrum as shown in the distribution curve for filmic portrayals is skewed to the left, indicating that the portrayals are extreme versions of autism affectedness. Furthermore, there was a higher than average presentation of symptoms in nearly every CARS2 category than would be anticipated for the average population affected by ASD, as reported by Schopler and colleagues (2010).

These findings reveal the dominance of ‘extreme representations’ in film, thereby providing evidence that the film industry does portray outliers or at least aspects of exceptionalities beyond those that would be expected in the normal distribution. For instance, portrayals of outliers included those with unfathomable abilities (omnipotent or superhuman powers) in films such as Bless the Child and Chocolate, or those shown to have no intent or capacity in films like Molly (before surgery ‘cured’ her of autism) or Mercury Rising. There are obvious misrepresentations of the autism and Asperger Syndrome diagnoses


in film. For example, certain characteristics such as ‘poor quality’ eye contact (one of the symptoms of ASD) are shown as total avoidance (e.g., Mercury Rising or Silent Fall), or savant skills are portrayed as integral to the ASD diagnosis (e.g., Rainman, Mozart and the Whale, and Killer Diller).

Further complicating the picture of autism painted by the film industry is the fact that some of the character portrayals, in spite of their dramatic representation of profound symptomology, would not warrant a clinical diagnosis (Conn & Bhurgra, 2012). This phenomenon is supported by Schopler and colleagues (2010), “it is possible for an individual to obtain a CARS2-ST total raw score of 30 or higher…and not qualify for a DSM-IV diagnosis of PDD (Pervasive Developmental Disorder)” (p 44). An individual that scores 30 or higher but does not warrant a PDD diagnosis may have other conditions that cause similar behaviours or may score severely in certain categories but not at all in others. Irrespective of this possibility, the film industry has identified the character as having a diagnosis of ASD in the DVD synopsis, which suggests for the lay viewer that the representation of autism in the film is irrefutably present and should be accepted.

4.A.4.2 Educational implications

Given the extreme portrayals of autism in films, the educational value of filmic representations of ASD is questionable. There is, however, a distinct possibility that utilising a tool that informs diagnosis by critically analysing the presence of ASD symptoms and their impact on the filmic character, could be beneficial.

This possibility was explored through the application of the CARS2 observation assessment tool by two professionals from different disciplines, a clinical psychologist and an education specialist. Both assessors found that the filmic representations of ASD among the population of film characters were more severe than would be expected for the average population with a diagnosis of ASD. Significantly, however, the representation of profound disability was less evident when looking at the raters’ scores for 15th CARS2 category, ‘overall impression’, which considered the character portrayal holistically rather than by specific characteristics. In the ‘overall impression’ category there was very high



agreement between the education specialist and the clinical psychologist, represented in an impressive r= .951, p=.000 for the Overall Impression scores, which are a better picture of the overall individual than the categorical measures of one kind of behaviour. T h e results can be interpreted to show that with relative accuracy well- versed practitioners can assess the impact of autism on an individual’s daily living activities, even when those individuals are fictional. In other words, experienced practitioners can agree on the severity of the autism portrayed in film because they have an understanding from their breadth of experience with real world ASD affectedness of the benefits of the portrayed strengths and the hindrances of the portrayed challenges. Thus, while maintaining a cautious consideration of the extreme presence of ASD characteristics portrayed in the films, the findings suggest the possibility of using filmic representations of ASD for educative purposes.

A growing body of literature is exploring the use of film for educative purposes.

Entertainment-Education Theory (EET) “provides the opportunity to study mass media as agents of both entertainment and persuasion” (Singhal & Rogers, 2002:

page 119). EET is also used to understand the influence of entertainment media on the knowledge and attitudes of viewers (Moyer-Gusé, 2008) via the assessment of EET-based interventions. Although this section of the thesis did not focus on EET, it is noteworthy that researchers are continuing to use it to explore the educational value of film. Research regarding the use of filmic representations to improve professional practice has also increased; for example, Wiertelak (2002) incorporated full-length feature films into neuroscience classrooms to promote content mastery and critical thinking skills.

In a similar way, through the use of an empirical assessment tool, this study has shown that filmic characters can be used to promote identification of characteristics related to ASD.

A further benefit to practitioners in using film representations of ASD is the static nature of the medium. Being able to refer back to the exact behaviours or interactions depicted in film offers opportunities for discussion in educational situations. Repeatedly referring to and replaying behaviours, communications,


and interactions can promote accuracy in the identification of antecedent events (events that come just before the behaviour) and general learning amongst the viewers. While the exaggerated and often inaccurate representation of ASD characteristics and behaviours in the 15 films examined using the CARS2 would likely make it easier for new practitioners to collect data on behavioural categories, it would probably not facilitate the development of a coherent understanding of the diversity in presentation of characteristics displayed by the Autistic Community.

4.A.4.3 Limitations: Accuracy

The benefits of applying assessment tools to filmic portrayals of a disorder outweigh the challenges, however these challenges should still be acknowledged.

The challenges are illustrated through the slightly lower inter- rater agreement found in this analysis than that reported for the CARS2 when used in a clinical setting. A number of factors reduced the likelihood of high inter-rater reliability. Firstly, the sample size was small and, as noted previously, some of the films were in languages other than English which required the use of a number of skills when performing the assessment; this served to distract from the task of observing behaviours. For example, reading subtitles and trying to understand cultural nuances make it challenging to determine whether a particular behaviour is deviating from the expectations by age. When analysis was conducted without the foreign language films, levels of agreement were higher than that reported in the CARS2 Manual. Secondly, some of the films, for example Molly and Bless the Child, began by showing a character with autism and then moved away from the diagnosis and diagnostic symptomology. This made it difficult to rate the characters because of the dramatic shift in idiosyncrasies, which further exacerbates the unrealistic representation of ASD.

Finally, when applying an assessment tool to films in independent settings (e.g., each rater viewing and scoring in their private residence) there are extenuating factors that may influence rating outcomes. These extenuating factors include, but are not limited to: other people in the area; pauses in viewing such as bathroom breaks; affection for, or aversion to, actors; and perceived