Attitudes towards autistic characters in mainstream series and connections with viewers’ own characteristics: an exploratory study Open Access

Autism is a neurodevelopmental disorder characterised by repetitive behaviours and difficulties in communication and social interactions (World Health Organization, 2019). According to a recent systematic review by Zeidan and colleagues, approximately one out of every 100 children is diagnosed with autism spectrum disorder (ASD) around the world. The review also found that the prevalence of diagnoses has increased since 2012. This increase is attributed to better detection and identification, as well as increased awareness about the condition (Zeidan et al., 2022). Although much attention is given to autism in children, there has also been research on the prevalence, assessment and diagnosis of autism in adults. Another recent systematic review (Huang et al., 2020) reported that the prevalence of adult autism is approximately 11 out of every 1,000 people. Similarly, this review found that the prevalence has been increasing over the past few years.

In recent years, not only the prevalence of autism but also the representation of autistic characters in mainstream media increased. Mainstream media has a significant effect on how people perceive and judge mental health and illness (Dietrich et al., 2006). Audiences can be influenced through emotional engagement, repeated exposure and the creation of parasocial relationships and interactions (i.e. nonreciprocal connections and attachment that can happen with both real-world people, such as celebrities on social media, or fictional characters). In the case of autism, the increased portrayal in recent years might have led to a more positive awareness and understanding of autism and the specific characteristics (Fontes and Pino-Juste, 2022; Nordahl-Hansen and Øien, 2021). Yet, creating awareness through media and fictional characters also bears disadvantages. While portrayal of autism seems to become more positive in books and on social media, representation is often still stigmatising in media such as newspapers and in movies and TV shows (Mittmann et al., 2024). For example, the average autistic character on TV is mostly white and male, and there is an unproportionally high prevalence of savantism, meaning that the autistic character has extraordinary skills (Nordahl-Hansen et al., 2018; Dean and Nordahl-Hansen, 2022). This might dilute the realistic situation of autism, as in reality, less than 30% of people with autism have savant skills (Howlin et al., 2009) and there exists a diverse distribution of all genders and races (Pham et al., 2022). While parasocial relationships and interactions can lead to lower stigma, they can also have a negative impact through negative self-comparison (Bond, 2021; Hoffner and Bond, 2022), which is more likely if the portrayal is more stigmatising or unrealistic.

Mostly, the conclusions and recommendations by studies investigating the phenomenon of media portrayal are that media should attempt to aim for a higher level of realism and less stigmatisation in their depiction of autism. Yet, these studies seldomly consider viewers’ attitudes towards those series and what series and attributes about autistic characters viewers prefer. Furthermore, research is lacking about how these preferences relate to personal characteristics, for example, the viewers’ own autistic tendencies.

Young adults are a main target group for series watching. In Austria, 65% of people under the age of 30 years claim to watch series several times a week (Spectra, 2020), and young adults are a main consumer age group of streaming services such as Netflix (Stoll, 2022). Nowadays, the main source for series is indeed streaming sites, which have overcome traditional TV watching (Sadana and Sharma, 2021; Gupta and Singharia, 2021). These streaming sites are also called over-the-top (OTT) platforms. For example, Netflix – currently the most in-demand OTT service – had a total of 223 million subscribers worldwide in 2022, with a total net income of $5.044bn (Molenaar, 2023). Young adults are therefore a target group that is potentially highly influenced by the content of mainstream series. At the same time, the selection of a particular series out of the vast variety the platforms offer may be influenced by personal (autistic) characteristics: For example, gender stereotypes or viewers’ sex could also play a role in series consumption behaviour. Studies found that specific genres are more popular with male (e.g. action, horror) or female (e.g. romantic, drama) viewer groups and that men in general prefer media content depicting male protagonists (Greenwood, 2010; Daalmans et al., 2017).

Therefore, the aim of this study was to examine connections of media consumption behaviour and personal characteristics. Because of the exploratory nature of the study, we did not form hypotheses. Our research questions were as follows:

The current study is a secondary analysis of a study examining self-diagnosis of autism and media consumption [blinded for review]. We used a cross-sectional online questionnaire. After sociodemographic questions such as age, gender, and nationality, the subsequent section explored media consumption. Participants were presented with a selection of 20 different series, comprising 15 series featuring a character with autism or autism-like symptoms and five distraction items. Participants could indicate which series they have watched (from never to almost all episodes). For each watched series, participants were asked to rate how much they liked the series and the character. Lastly, an emotion recognition task (ERT) was administered as an objective indicator for autistic tendencies within the participant. Participating in this online study took about 20–30 minutes.

Participants were recruited online through the research group’s network and via distribution by the Österreichische Gallup-Institut GmbH panel. The study received approval from [blinded for review] (EK-Nr. 1004/2023), and all participants granted their informed consent by ticking a box indicating their agreement at the beginning of the questionnaire. Because of the study’s exploratory nature, no power analysis was conducted prior to recruitment. Yet, given that more than 300 participants were recruited, the study is likely to have sufficient power.

Series depicting autism: We examined 15 different series depicting an autistic character or a character exhibiting autistic characteristics without explicitly being labelled autistic (when used collectively, we will call them autistic characters for the rest of this paper). Table 1 shows characteristics of the included series and characters. Note that one of the series was a reality show (Love on the Spectrum). We added five distraction items depicting a character with a different mental condition. These were Stranger Things, Mr. Robot, 13 Reasons Why, BoJack Horseman and Sex Education. All series were selected by the research team based on online research of popular series depicting a character with autism, disclosed as such (canonically) or with autism-like symptoms but without detailed information about a diagnosis (non-canonically). If autism-like symptoms are present even without diagnosis, it was determined by the research team, based on identifying a series via online research and viewing sequences of the show to confirm the information detected in the web search. The distraction items were selected via online research. The series had to contain a character with a mental health issue other than autism. Again, mental health symptoms are present even without diagnosis, as was determined by the research team, based on identifying a series via online research and viewing sequences of the show to confirm the information detected in the web search.

Emotion recognition task: We used the ERT as an objective means of assessing autism (Kessels et al., 2014). We chose this task instead of a self-administered questionnaire because challenges in recognising emotions are a distinctive trait of autism that can be objectively measured (Yeung, 2022). The ambition to measure autistic tendencies objectively was based on former findings indicating that consuming series depicting autistic characters can increase self-diagnosis of autism (Mittmann et al., 2023). Law Smith et al. (2010) showed in their study, including a clinical sample of “high-functioning Autsim” participants, that emotion recognition overall is impaired in the clinical sample compared to typically developed participants. Even if not all emotions at all levels of intensity were equally affected. Harms et al. (2010) conclude in their review that individuals with ASD experience obvious difficulties in labelling and matching emotions, and even if they perform similarly to controls in some cases, the decoding process of facial expressions is different in all studies reviewed. Similarly, Monteiro et al. (2017) concluded that in the reviewed EEG/ERP studies, subjects with ASD showed reduced amplitudes and delayed latencies in all conditions in comparison to typically developed individuals. The findings in ERTs might be as heterogenic as the methods used to determine them; it is still more likely to find deficits in facial emotion recognition in ASD than not (Harms et al., 2010). In the ERT, participants are presented with 48 distinct facial expressions and are required to determine whether each face conveys surprise, happiness, sadness, fear, disgust or anger across four levels of intensity/difficulty. To streamline the measure, we reduced the original 96-item set (consisting of two male and two female faces) by excluding two sets of faces (one male and one female) across all difficulty levels; therefore, the possible range was 0 (all emotions were classified wrong) – 48 (all emotions were classified right) points. ERT scores can range from 0 to 48. Higher scores mean better facial expression recognition.

Media “Liking”: We asked participants how much they “liked” the character as well as the series with a single question for each series on a five-point Likert scale. We used the German questions “Wie gut hat Ihnen der Character/die Serie gefallen?” (How did you like the character/series?), which is a general statement and does not focus especially on “likeability” or “sympathy” but also includes finding a character or series interesting or fascinating.

For statistical analysis, SPSS 28.0.1.0 was used (IBM Corp, 2021). We conducted a Mann–Whitney U-test to compare the overall viewer numbers. We conducted a Chi²-test to compare the viewer numbers of series with non-savant characters. We conducted t-tests to compare the mean “character liking” scores. As grouping variables, we used sex (male, female) and the ERT scores (ERT high and low scorer groups).

Our sample consisted of a total of 348 participants between 18 and 30 years (mean age: 24.49 years, SD = 3.71). A total of 117 participants identified as male, 213 as female, 17 as non-binary and 1 person did not disclose sex/gender. The 18 non-binary/non-disclosed cases were excluded from the analyses about sex differences. The majority of participants was Austrian (n = 291).

The mean ERT score in the overall sample was 30.13 (SD = 5.48). Female participants had significantly higher mean ERT scores (mean: 31.16, SD = 5.11) than male participants (mean: 28.25, SD = 5.80), t(328) = −4.725, p < 0.001. By separating the sample in ERT high and low scorers (ERT ≤25, ≥35), we generated 2 subsamples to perform an extreme group comparison by excluding the middle section of mean ± SD approximately 30 ± 5. The ERT high scorer sample consisted of 78 participants with a mean ERT score of 36.94 (SD = 2.26), and the ERT low scorer sample consisted of 68 participants with a mean ERT score of 22.06 (SD = 3.22). Both subsamples were not normally distributed. Therefore, all conclusions should be interpreted with caution.

The 15 included series that depict characters with autism or autism-like characteristics are displayed in Table 1. Seven of the series have male autistic characters. Characters are canonically autistic in seven series (meaning that they are officially declared autistic in the series). In six series, the main character is especially gifted in a specific field or shows signs of savantism; from these six autistic characters, four are male and two female. The Big Bang Theory was the series with the most viewers (91.38%), followed by Sherlock and Bones. The least popular series were As We See It and Ella Schön. The most liked series are The Queen’s Gambit, Sherlock and The Good Doctor. The most liked characters were Beth Harmon (The Queen’s Gambit) and Sherlock Holmes (Sherlock). The lowest character liking scores were found for Ella Schön (Ella Schön) and Matilda Moss (Everything’s Gonna Be Okay).

We compared the character liking scores of male and female viewers over the series characters (Figure 1). We conducted t-tests to compare the mean “liking of the character scores”. We only found a significant difference for the character Sam Gardner from Atypical, t(89) = −3.31, p < 0.001, who was more liked by female participants. None of the other character liking scores showed significant sex differences.

We compared the percentage of viewers between male and female participants over the series with and without savantism or a main character who is especially gifted in a specific field (Figure 2). We first conducted a Mann–Whitney U-test to compare the overall viewer numbers, without a focus on the sex of the viewers. The result indicates that series depicting characters with savantism have significantly more viewers than series with non-savant characters overall (z = −3.193, p < 0.001). We then conducted a Chi²-test to compare the viewer numbers of series with savant and non-savant characters in male and female participants. The result indicates that there is no significant difference in the number of male and female viewers of series with non-savant characters (χ² = (8, N = 348) = 12.661, p = 0.284) and no significant difference in series depicting characters with savantism (χ² = (5, N = 348) = 3.728, p = 0.589).

We compared the percentage of viewers between participants with ERT scores of 25 and lower as well as 35 and higher over the shows with and without (Figure 3). We conducted a Mann–Whitney U-test to compare the overall viewer numbers. The result indicates that series depicting characters with savantism have significantly more viewers in the ERT low and ERT high scorer groups than series with non-savant characters (z = −3.185, p < 0.001). We conducted a Chi²-test to compare the viewer numbers of series with non-savant characters in the ERT low and ERT high scorer groups. The result indicates that significantly more ERT low scorers are watching series with non-savant characters (χ² (8, N = 348) = 25.552, p < 0.001) than ERT high scorers. Series depicting characters with savantism show no difference in the viewer numbers between ERT high scorers and ERT low scorers (χ² (5, N = 348) = 0.358, p = 0.996).

We compared the “character liking scores” between participants with ERT scores of 25 and lower as well as 35 and higher over the series with and without savantism or a main character who is especially gifted in a specific field (Figure 4). We conducted t-tests to compare the mean “character liking” scores. We only found a significant difference for the character Beth Harmon from The Queen’s Gambit, t(58) = −2.891, p = 0.005, who was more liked by ERT high scorers. All other characters showed no significant difference between ERT high scorers and ERT low scorers.

This study explored media consumption behaviour of a general audience of young adults regarding series depicting a canonically or non-canonically autistic character. We examined which attributes are attractive for viewers and connections to their own autistic characteristics, sex and ERT scores. We found that series depicting a character with savantism have more audience than series with non-savant characters overall. We could not find sex differences in the preference for series depicting a character with or without savantism. Accordingly, we found no significant sex differences for the character liking scores except for the character Sam Gardner from Atypical, who was more liked by female participants. Sam Gardner, as a male character who is strongly involved with his overprotective mother and his female therapist, has problems finding a girlfriend and is bullied in school, and is likely a character male viewers might not want to identify with or relate to. We could not replicate the tendency of men to prefer male over female characters (Greenwood, 2010; Daalmans et al., 2017) as we could neither find great differences in liking scores nor in viewer numbers corresponding to characters’ sex. Our findings are more consistent with the outcomes of a German study, which found that differences between male and female viewers in genre preference tend to be overestimated (Wühr et al., 2017).

There is no difference between ERT-high-scorer and ERT-low-scorer considering series with savant characters, which means series featuring non-savant characters tend to have a smaller audience compared to series with savant characters, indicating that they cater to a more specific and niche market. However, it is interesting to note that a significantly higher number of individuals with low scores in the ERT, and therefore a higher tendency for autism, prefer series that portray autism in a more realistic manner. This suggests that these series appeal to individuals even without an official autism diagnosis, possibly because they can relate to certain events or behaviours depicted within the series or exhibited by the autistic characters. Interestingly, we found no significant difference in the “character liking scores” between ERT high scorers and ERT low scorers except for the character Beth Harmon from The Queen’s Gambit. The authors can only speculate that viewers with lower ERT scores might find it harder to relate to Beth Harmon as she is a woman struggling with addiction to tranquilisers and alcohol, and she is referred to and refers to herself as “crazy”, and therefore her (autistic) character is cast in a negative light. The continued reinforcement of narrow and stereotypical portrayals of autism may heighten the pressure on autistic individuals to hide their traits to avoid being stigmatised (Orm et al., 2023). This masking or camouflaging goes hand in hand with mental health problems for the affected persons (Cook et al., 2021).

All included series can be classified as OTT series. This means that instead of watching the series from a TV station, the series can be consumed via web-based video platforms. This new form of series watching has mostly overtaken traditional sources such as cable TV (Sadana and Sharma, 2021; Gupta and Singharia, 2021) and has therefore several implications. In Austria, the most common OTT platforms in 2022 were Netflix, Amazon Prime and Disney+ (Statista Research Department, 2023a), and in the age group of older adolescents and young adults, videos were mostly consumed via YouTube, Netflix and Amazon Prime (Statista Research Department, 2023b). This might have influenced consumption of series that are not available on these platforms (e.g. Ella Schön, which is available in the ZDFmediathek, or Everything’s gonna be okay, which streams on Hulu).

While this study gives valuable insights into how a general population of young adults consumes OTT series with autistic characters, the results are limited in several ways. First, because of the exploratory character of the study, our results can only be seen as initial findings paving the way for more detailed research questions. For example, there could be more detailed research around other characteristics about a series such as how length or number of seasons or network impact effects. Furthermore, there could be more specific research on what makes characters attractive to which viewer groups. In our sample, we found significant sex differences and significant differences in ERT scores for two specific characters, Sam Gardner and Beth Harmon. It would be interesting to know what makes those characters especially interesting to specific groups and what characteristics appeal to all viewers. Second, the study was conducted with mostly Austrian participants and with all participants being from German-speaking countries. This means that especially media consumption behaviour cannot be generalised to other countries, as OTT platforms offer a different range of series in different countries. Third, the interactivity of such platforms is a valued factor for consumers (Chang and Chang, 2020). For content selection, OTTs often suggest new series to watch based on series the user has already watched. This means that some people might watch more series depicting autistic characters based on their previous consumption behaviour. While this might have influenced the consumption behaviour of participants of our study, it applies to most OTT platforms and is hence applicable for all the participants equally.

Our finding that series depicting savantism have more viewers than series depicting non-savant characters does not clearly support the call for a more realistic portrayal of autism in mainstream media. Notably, this is also true for participants with low ERT scores who may have autistic tendencies themselves. Hence, the main implication is that there might be a disconnect between what is considered “good” from professionals and experts by experience and what is more enjoyable to watch for a general population regardless of their autistic tendencies.

Our study shows that non-stigmatising series are more often watched by individuals with autistic tendencies. While it is important for autistic individuals to see characters on TV that they can connect with and identify with, it would also be desirable to achieve a de-stigmatising effect in a general population – but this can only happen if the general population is inclined to watch these series. Additionally, the hugely commercialised media industry naturally relies on their viewership to watch and like their series. Hence, they need to create characters that are enjoyable to watch for a broad range of audiences, which might lead to a discrepancy between representative portrayals of autism and demand for viewership. Arguably, non-realistic representation holds true for many genres and series that are popular.

Therefore, one question remains unanswered: How can mainstream media bridge the creation of non-stigmatising and realistic characters that are still as appealing to viewers as more savant ones, so that series depicting a diverse range of autistic characters are also especially interesting for a general audience who do not self-identify as autistic? Future research should examine best practices and form recommendations for the portrayal of autism in mainstream series, considering not only research around non-stigmatising portrayal but also including knowledge about preferences of viewers.