Recent research suggests that non-autistic people in South Korea are less willing to engage with autistic people than their counterparts in the United States (US; Kim et al., 2021) or the United Kingdom (UK; Mac Cárthaigh & Lopez, 2020). Planning to keep one’s distance from autistic people is an indicator of public stigma, which includes labeling people as different, stereotyping or making assumptions about differences, distinguishing between us and them, and discriminating against them (Goffman, 1963; Link & Phelan, 2001). Public stigma reduces the social status and material opportunities available to people who are stigmatized (Johnson & Joshi, 2016). Many stigmatized people internalize stigma, viewing themselves as less than others. Given that internalized stigma contributes to poor mental health among autistic individuals (Botha & Frost, 2020), there has been increasing interest in understanding factors that contribute to stigma towards autistic people such as limited autism knowledge and/or contact with autistic people (e.g., Gillespie-Lynch et al., 2019; Kim et al., 2021).

An emerging body of cross-cultural research suggests that public autism stigma, measured by desired social distance with autistic individuals, is heightened in South Korea (Kim et al., 2021), Japan (Someki et al., 2018), Lebanon (Obeid et al., 2015), and Malaysia (de Vries et al., 2020) compared to the US or UK. This cross-cultural pattern aligns with speculation that stigma more generally may be higher in more collectivistic (i.e., prioritizing group harmony and cohesion) than individualistic (i.e., prioritizing independence and individual success) cultures (e.g., Papadopoulos et al., 2013). However, associations between individuals’ cultural values and their desired social distance from autistic people do not support this interpretation. In Lebanon, South Korea, and the US, heightened individual-level horizontal collectivism (or valuing cooperation between ingroups) has been associated with decreased stigma, and heightened vertical individualism (or valuing individual attempts at dominance) has been associated with increased stigma (Gillespie-Lynch et al., 2019; Kim et al., 2021).

In addition to acceptance of inequality (i.e., vertical orientation), many other factors also contribute to stigma. Some factors appear to contribute to stigma similarly across cultures (e.g., heightened vertical orientation, less accurate autism knowledge, and reduced pleasantness and frequency of previous contact with autism; Kim et al., 2021), while other factors may only be associated with autism stigma in some cultural contexts. For example, cultural tightness (or the strength of a society’s norms and rejection of deviant behaviors) was only associated with stigma towards the label “autism” in South Korea and not in the US (Kim et al., 2021). In contrast, in-group favoritism and out-group derogation, which are commonly associated with prejudice toward other minority identities, were associated with stigma towards the label “autism” in the US but not in South Korea. To better understand when and why public stigma towards autistic people is heightened in South Korea vs. the US, the current study examined if characteristics of non-autistic people (e.g., autism knowledge) and/or autistic people (e.g., support needs) contribute to stigma towards adults exhibiting unlabeled characteristics associated with autism in South Korea and the US.

Why Focus on Autism Stigma in South Korea

South Korea is a relatively ethnically, racially, and culturally homogeneous country (Kim-Rupnow, 2005). Such homogeneity may contribute to heightened cultural tightness (Gelfand et al., 2011). Indeed, cultural tightness is higher among South Koreans than Americans (Kim et al., 2021), which may cause individuals who exhibit behaviors that diverge from social norms (e.g., autistic individuals) to become targets of discrimination in South Korea (Kim-Rupnow, 2005). Kim et al. (2021) found that cultural tightness was positively associated with stigma towards the label “autism” in South Korea. However, South Koreans reported greater stigma towards the label “autism” than Americans, even after accounting for rater characteristics such as cultural tightness, vertical orientation, autism knowledge, and pleasantness and quantity of contact with autistic people.

It is possible that limited exposure to the neurodiversity movement, which has been shifting perspectives about autism in many countries since coming into being in the late 1990s (Kapp, 2020), may be magnifying cross-cultural variations in autism stigma between South Korea and the US. The neurodiversity movement frames autism and other neurological differences as valuable aspects of human diversity (Kapp et al., 2013; Singer, 2016). Neurodiversity advocates condemn efforts to “cure” autism and other neurological conditions and instead advocate for support to help autistic individuals thrive. The neurodiversity movement may be increasing autism acceptance by reframing stigmatizing misconceptions, e.g., that autism is an illness, with the recognition that autism is often a valued aspect of people’s identities (Grinker, 2020). However, the neurodiversity movement’s reframing of autism as “a way of being” (Sinclair, 2012) is unlikely to be equally appealing in all cultural contexts.

Indeed, prior research suggests that the neurodiversity movement’s framing of autism as a pervasive and enduring aspect of personhood may clash with at least some responses to autism in South Korea. For example, South Korean mothers may often reject an autism diagnosis for their child due to multiple intersecting considerations including viewing the “autism” label as an arbitrary classification that does not account for their child’s unique strengths and capacity for growth and as both permanent and stigmatizing (e.g., limits the educational opportunities of autistic children and the marriageability of relatives; Grinker & Cho, 2013). This desire to view autism as transient and context-dependent may be a key reason that autism has often been misdiagnosed in South Korea as reactive attachment disorder, which is caused by negative parenting practices (Kang-Yi et al., 2013). Viewing autism in terms of temporary limitations that are not intrinsic to a person is not consistent with the neurodiversity movement’s recognition of autism as a consistent aspect of a person’s neurology. Additionally, widespread prioritization of academic and professional competitiveness, concerns about how autistic individuals may influence other students’ academic productivity, and high pressure to conform in South Korea (Grinker & Cho, 2013; Kim et al., 2021) could lead Koreans to view the neurodiversity movement as not well suited to Korean society. Therefore, we hypothesized that agreement with the neurodiversity movement would be lower in South Korea than in the US and that lower neurodiversity movement endorsement would contribute to heightened autism stigma in South Korea relative to the US.

Focusing on Stigma Towards Autistic Behaviors

We focused on stigma towards unlabeled characteristics associated with autism rather than the label “autism” as recommended by stakeholders in South Korea (Grinker & Cho, 2013, p. 58):

Mothers suggested that our research on autism would be more feasible if we did not use the word autism at all, or used it only “in secret,” by which they meant we should talk about the symptoms but not the classification itself. This sentiment was shared by teachers as well.

However, explorations of stigma toward autistic behaviors may be complicated by the diversity of characteristics associated with autism. Although Autism Spectrum Disorder is considered one broad diagnosis by the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5; American Psychiatric Association, 2013), autistic people are so diverse that autism is increasingly conceptualized as “autisms” or “an autism constellation” (Fletcher-Watson & Happé, 2019). Given this diversity, all autistic people may not experience the same level of stigma. However, existing stigma literature has barely scratched the surface of the autism constellation. Most studies examining “autism stigma” have either asked participants to evaluate a vignette about one autistic individual who exhibits unlabeled (e.g., Nevill & White, 2011) or labeled autistic characteristics (Gardiner & Iarocci, 2014; O’Connor et al., 2020) or to indicate desired social distance from a generic diagnostic label (e.g., “autistic person”; Gillespie-Lynch et al., 2015; Lu et al., 2021).

Such universalizing approaches leave us largely in the dark about why different autistic people might experience different levels of stigma (and thus provide limited information about how we can help protect the most vulnerable autistic people from discrimination). According to the DSM-5 (APA, 2013), autism is defined by social communication and interaction difficulties and restricted interests and repetitive behaviors (RIRBs). Non-autistic people may have different conceptualizations (or levels of stigmatization) about different types of autistic behaviors. For instance, RIRBs (e.g., flapping hands) may be more stigmatized than social difficulties (e.g., difficulty making friends) because they are more visibly recognizable and therefore seem like a more explicit violation of social norms.

Further, approximately 25–30% of autistic people do not communicate with others using spoken speech (i.e., non-speaking; Anderson et al., 2007; Lord et al., 2004; Norrelgen et al., 2014). Yet, non-speaking autistic people have been insufficiently represented in autism research and advocacy more generally (Bailin, 2019; Peña, 2019), and stigma towards non-speaking autistic people has been almost entirely overlooked in the autism stigma literature. Therefore, a primary aim of the current study was to examine predictors of stigma towards non-speaking autistic people, as well as those experiencing primarily social difficulties or RIRBs. Although being non-speaking would have been one way to demonstrate the social communication difficulties needed for an autism diagnosis in the DSM-IV, difficulties with structural language have become a specifier rather than an aspect of the core diagnostic criteria for ASD in the DSM-5 (APA, 2013). Therefore, we regarded being non-speaking as representing a different domain than social difficulties, rather than as a subset of social difficulties.

The level of support an autistic individual needs may also influence the stigma they experience. Some people may seek to distance themselves from autistic people to whom they think they would need to make greater efforts to adapt. Indeed, non-autistic undergraduates in the US expressed heightened social distance towards an individual exhibiting characteristics of Asperger’s Syndrome (e.g., perseverating on one topic and looking down when talking to others) relative to an individual showing milder social difficulties (e.g., shy and quiet) or no social difficulties (Butler & Gillis, 2011). Whether the character was labeled as having “Asperger’s disorder” did not impact stigma. Similarly, college students in the US and Lebanon expressed more desire to keep their distance from an autistic peer exhibiting “disruptive” behaviors (e.g., often calling out in class) compared to an autistic peer who exhibited “withdrawn” behaviors (e.g., frequently staring at an iPad rather than engaging; Gillespie-Lynch et al., 2021b). In this case, labeling the students as autistic led to lower stigma than leaving the vignettes unlabeled. Perceived dangerousness, but not perceived responsibility for one’s condition, was associated with greater stigma. In a study in which non-autistic college students rated their first impressions of 20 autistic and 20 non-autistic adults (Sasson & Morrison, 2019), first impressions of both autistic and non-autistic adults were more negative when the people being rated had higher levels of autistic traits.

In direct contrast to the research with adult raters just discussed, middle school students in the US expressed more willingness to interact with a child actor exhibiting “severe” signs of autism (e.g., flapping hands, rocking body, smelling a ball, and immediate echolalia) compared to a child exhibiting “milder” autistic behaviors (e.g., plays basketball, talks excessively about one topic, and uses literal language; Griffin, 2018). The middle school participants were not informed that the child was autistic. The author speculated that the children were less interested in engaging with the child with “milder” autistic characteristics because he did not clearly have a disability and therefore the children expected him to be able to “control his behavior.” Consistent with this interpretation, autistic middle school children in the UK with better social-communication abilities reported higher levels of victimization than autistic youth who experienced more social-communicative difficulties (Rowley et al., 2012). Similarly, in a very small sample of ten autistic adolescents and young adults in the US, lower levels of autistic traits were associated with self-reports of having experienced more autism stigma (Shtayermman, 2009). Various possible explanations have been proposed to explain these associations, including greater social awareness and social experience among autistic people with more advanced social-communication skills.

Contrasting associations across studies suggest a key reason why autism stigma might differ across cultural contexts. People may be less likely to recognize autism as a valid reason for not following norms in some cultural and developmental contexts. Phelan and colleagues (2008) proposed three main functions shared by both stigma and prejudice (which they viewed as manifestations of the same overarching construct): (1) domination (e.g., keeping people who have less power down so that the powerful can continue to profit from their oppression), (2) norm enforcement (e.g., using punishment to force people to conform), and (3) disease avoidance (e.g., avoiding people because they are viewed as ill). They asserted that the norm enforcement function of stigma could only be applied when people are viewed as capable of adapting to norms. This distinction provides a possible explanation for why autism stigma was positively associated with cultural tightness in South Korea; perhaps Koreans are less likely to view autism as a valid reason for not adjusting to norms than Americans.

The Current Study

To the best of our knowledge, only one prior study compared stigma towards unlabeled and labeled characteristics associated with autism across cultural contexts (Lebanon and the US; Gillespie-Lynch et al., 2021b). This omission is problematic because culture shapes how we interpret others’ behaviors and may thus cause specific characteristics of autism to be more or less accepted in different cultural contexts (Golson et al., 2022). For example, refraining from making expressive facial expressions is often considered an appropriate social behavior in South Korea due to the influence of Confucianism in the past (Lee & Lee, 2012). This suggests that showing a few social characteristics of autism may not be associated with heightened stigma in South Korea, as long as they are not marked by the “autism” label, which is more stigmatized in South Korea than in the US (Kim et al., 2021). Alternatively, even small deviations from social norms might evoke a desire to keep one’s distance in South Korea relative to the US, given the greater emphasis in Korea on abiding by social norms.

Therefore, using vignettes depicting characters who primarily exhibited a specific domain of autistic behavior (social difficulties, being non-speaking, or RIRBs), we conducted a cross-cultural comparison of stigma toward unlabeled autistic behaviors in South Korea and the US. We examined the following research questions (RQs):

  1. RQ 1a.

    Does non-autistic adults’ stigma towards characters in vignettes depicting unlabeled behaviors associated with autism vary as a function of the character’s support needs?

    We hypothesized that non-autistic people would report heightened stigma toward characters with greater support needs because high support needs may place greater demands on others.

  2. RQ 1b.

    Do associations between stigma and support needs vary across countries?

    We did not have a hypothesis regarding potential variations across countries in associations between support needs and stigma due to the lack of relevant prior research.

  3. RQ 2.

    Which characteristics of non-autistic adults (i.e., raters) are associated with their stigma towards unlabeled behaviors associated with autism?

    As in past studies (e.g., Gillespie-Lynch et al., 2019; Kim et al., 2021; Obeid et al., 2021), we hypothesized that being Korean, having inaccurate autism knowledge, reporting less positive and frequent contact with autistic individuals, expressing greater acceptance of inequality (and not collectivism), and less accurate identification of autism would predict stigma toward unlabeled behaviors indicative of autism. We also hypothesized that cultural tightness would predict stigma among South Korean participants only, similar to Kim et al.’s (2021) findings. Further, we hypothesized that reduced neurodiversity endorsement, but not neurodiversity awareness, would be associated with more stigma. We did not have specific hypotheses regarding demographic variables (age and education level) due to mixed findings in previous studies (e.g., Aubé et al., 2021; Campbell et al., 2004; Kim, 2021; Surmen et al., 2015).

For all RQs, we examined if patterns remained consistent or diverged across different domains of autistic behaviors. We did not have specific hypotheses about how findings might differ depending on the domains of behaviors focused on because previous research did not examine this issue. However, a recent vignette study revealed initial evidence that job applicants exhibiting RIRBs (i.e., sensory sensitivities and rigidity) might be rated more negatively than those exhibiting social difficulties (McMahon et al., 2021).

Method

Participants

A total of 842 adult (> 18 years old) participants (490 Americans and 352 Koreans) were recruited from Amazon’s MTurk and DataSpring (i.e., a crowdsourcing marketplace similar to MTurk, recruiting Asian research panels), respectively, in September 2021 and completed our online survey. We included two attention check items (“please mark strongly disagree for this item”) and excluded data from 101 American and 75 Korean participants who failed at least one attention check. We also asked participants to identify their autism diagnostic status twice, eliminating data from 33 American and 13 Korean participants who identified as autistic on one item but not on the other.Footnote 1 Additionally, because we aimed to assess autism stigma among non-autistic individuals only, we excluded data from the 116 American and five Korean participants who self-identified as autistic (i.e., identifying as autistic in both questions regardless of having a formal autism diagnosis) from analyses.Footnote 2 Two-hundred and forty American and 259 Korean participants were included in the analysis. Table 1 presents detailed demographic information on participants included in the final analysis.

Table 1 Participant characteristics
Full size table

Procedure

After completing an online consent form, participants read three unlabeled vignettes representing three domains of autistic characteristics: social difficulties, being non-speaking, and RIRB. Participants were randomly assigned to read either low- or high-support vignettes within each domain. They indicated their desired social distance from the character in each vignette and selected diagnostic labels for each character. Subsequently, all participants completed surveys assessing autism knowledge, pleasantness and quantity of previous contact with autistic individuals, cultural orientation, cultural tightness, awareness and endorsement of the neurodiversity movement, and demographics (i.e., age, gender, education level, and ethnicity). A professional translator translated the instruments that had not previously been translated (i.e., the vignettes and the items on and descriptions of the neurodiversity movement) from English to Korean. SYK, who is fluent in both Korean and English, back-translated the translation and compared the original survey with the back-translated survey to revise and finalize the Korean version of the survey. SYK, who is Korean, chose gender-neutral names for the Korean vignettes, and the professional translator confirmed that Korean names are gender-neutral. Supplementary Material A presents the full list of items and vignettes used in this study. The Institutional Review Board of Duksung Women's University approved all data collection procedures.

Measures

Vignettes and Social Distance Scale (SDS)

We developed six vignettes depicting a character with a gender-neutral name exhibiting autistic characteristics. Each vignette focused on one of three domains: social difficulties, RIRB, and non-speaking. Each domain was further divided into low vs. high support needs. Characters with low support needs were depicted as living an independent life with a job, while characters with high support needs were depicted as unemployed and being supported by close family members in terms of daily chores and going out. A doctoral-level autism researcher, who has lived in both South Korea and the US for more than 10 years, confirmed that the vignettes appropriately reflected low and high support needs in both Korean and American cultures. See Supplementary Table S1 for the number of participants who read each vignette type.

The development of vignettes was guided by the description of social interactions and RIRBs in the DSM-5 and the authors’ expertise and experiences with autism. No vignette labeled the character’s diagnosis, and each vignette matched the other vignettes in terms of the numbers of autistic characteristics described, strengths, clues suggesting support needs, and other demographic information. We aimed to have each vignette depict someone who could be autistic (i.e., someone who has both social difficulties and RIRBs). We highlighted the primary focus of a given vignette by referring to it multiple times while suggesting that the non-prioritized domain was present much more subtly. For example, the vignettes depicting RIRBs included multiple RIRBs with a single reference to potential social difficulties (i.e., attending a peer group to have an opportunity to socialize). An autistic male adult reviewed all vignettes and indicated that the vignettes accurately described behaviors and characteristics that could be exhibited by autistic people. Supplementary Table S2 lists each vignette’s components.

After reading each vignette, participants responded to a Social Distance Scale (Bogardus, 1933). We adapted a social distance scale used in prior work (Gillespie-Lynch et al., 2015), which included seven items assessing one’s willingness to interact with an autistic individual at varying degrees of intimacy. We adapted this measure so that participants indicated their willingness to interact with the character described in the vignette (i.e., changed the wording from a generic “autistic person” to “the character described in the vignette”). Additionally, we revised and added items to capture participants’ willingness to interact with the character in a wide range of contexts (i.e., professional, studying, friendship, and dating contexts).Footnote 3 All 12 items utilized a 5-point Likert response scale, with higher scores representing higher levels of stigma. The alpha coefficients of the SDS for the social, non-speaking, and RIRB domains were 0.88, 0.88, and 0.90, respectively (0.87, 0.87, and 0.90 for the US subsample and 0.83, 0.87, and 0.88 for the South Korean subsample).

Identification of Likely Diagnosis

After responding to the Social Distance Scale for each of the three vignettes, participants were asked to choose the single most likely diagnosis of each character from several options: attention-deficit/hyperactivity disorder (ADHD), autism or Asperger syndrome, intellectual disability (ID), learning disability, cerebral palsy, other, and no diagnosis.Footnote 4

Participatory Autism Knowledge-Measure (PAK-M)

Autism knowledge was measured using PAK-M, an instrument adapted from the Autism Awareness Survey (Stone, 1987) through collaboration with autistic students (Gillespie-Lynch et al., 2021a). Participants responded to 29 statements (e.g., “Autistic people show affection”) scored on a 5-point Likert scale, and higher mean scores represented more accurate autism knowledge. The PAK-M was reported to have acceptable internal consistencies in the US, Lebanon, and South Korea (Gillespie-Lynch et al., 2021a; Kim et al., 2021). The alpha coefficient of the PAK-M in the current study was 0.82 (0.62 for the US and 0.87 for South Korea).

To better understand the low internal consistency of the PAK-M in the US sample, we checked the internal consistency of the non-reverse-scored (i.e., regular) and reverse-scored items, separately, because simultaneous use of regular and reverse-scored items may introduce uncontrolled variance due to participants’ response styles, which impact consistency and factor structure (Suárez-Alvarez et al., 2018; Vigil-Colet et al., 2020). The alpha coefficient of regular items was 0.88 (0.88 for the US and 0.73 for South Korea), and that of reverse-scored items was 0.88 (0.93 for the US and 0.69 for South Korea). We repeated all analyses with the version that consisted of only regular items and again with the version that consisted only of reverse-scored items. We note in footnotes if using the regular or reverse-scored versions altered findings from main analyses conducted with the full measure.

Previous Contact

We measured the quantity of participants’ previous contact with autistic individuals with an item, “How often do you spend time with an autistic individual?” to which participants responded on a 5-point Likert scale (1, “don’t spend time”; 5, “very often”). We also assessed the types of previous contact by asking participants to indicate any relationships they have had with an autistic person (e.g., your romantic partner is autistic, your co-worker is autistic, or your acquaintance is autistic). We coded participants into either knowing or having no experience with autistic individuals. Participants were also categorized into those who are a nuclear family member (i.e., parents, children, or siblings of an autistic person) of an autistic person and those who are not.

Pleasantness of previous contact was assessed with an item “In the past, were your overall experiences with individuals with ASD pleasant?” to which participants responded on a 7-point Likert scale (1, “strongly disagree”; 7, “strongly agree”).

Cultural Tightness

Four items from Gelfand et al.’s (2011) Tightness-Looseness Scale, which measures the degree to which social norms within nations are defined and enforced, were used to measure cultural tightness (e.g., “There are many social norms that people are supposed to abide by in this country”). Participants responded on a 6-point Likert scale, with a higher score representing higher levels of cultural tightness. The alpha coefficient was 0.88 (0.82 for the U.S. and 0.91 for South Korea). Participants completed two additional items, but these items were dropped from analyses because they reduced the internal consistency of the instrument to 0.67 (0.59 for the U.S. and 0.73 for South Korea). Any changes in main analysis results due to excluding these items are reported in footnotes (See Supplementary Materials A for the full list of included and eliminated items).

Cultural Orientation Scale

Triandis and Gelfand (1998)’s Culture Orientation Scale was used to measure participants’ cultural orientation. The Culture Orientation Scale consists of four subscales: vertical collectivism (e.g., ‘It is important to me that I respect the decisions made by my groups’), vertical individualism (e.g., ‘Winning is everything), horizontal collectivism (e.g., ‘I feel good when I cooperate with others’), and horizontal individualism (‘e.g., I often do ‘my own thing’). Each subscale consists of four items scored on a 7-point scale. Each individual’s total individualism score was subtracted from their collectivism score to yield their relative degree of collectivism vs. individualism; higher scores represented higher degree of collectivism. Each individual’s total vertical orientation scores were subtracted from their horizontal orientation scores to yield their relative degree of vertical vs. horizontal orientation; higher scores represented a higher degree of vertical orientation. The alpha coefficient of the overall scale was 0.91 (0.84 for the U.S. and 0.83 for Korea); the alpha coefficients of all subscales were higher than 0.80.

Neurodiversity Movement Awareness and Endorsement

We created two items measuring participants’ neurodiversity awareness and their endorsement of the neurodiversity movement. Participants responded either yes or no to a question, “Do you know what the neurodiversity movement is?” Subsequently, participants read descriptions of the neurodiversity movement and responded to a question “how much do you agree with the neurodiversity movement?” on a 4-point scale (1, strongly disagree, 4, strongly agree).

Demographic Information

Participants responded to five demographic questions asking about their gender, age, ethnicity, education level, and diagnostic status.

Analytic Approach

To account for the large number of analyses conducted, we applied an alpha level of 0.005 to all analyses and considered p-values between 0.05 and 0.005 as suggestive based on the recommendations of Wasserstein et al. (2019). First, we computed independent samples t-tests by country to examine cultural differences in all variables. We then conducted a one-way ANOVA examining if the stigma varies across domains of autistic behaviors highlighted (i.e., social vs. non-speaking vs. RIRB).Footnote 5

To address the first RQs, we conducted regression analyses by domain predicting stigma towards autistic characteristics with country, support needs condition, and their interaction as predictors. To address the second RQ, we first examined zero-order correlations in the combined sample between stigma in each domain and rater characteristics to identify variables that were initially associated with stigma. Then, we included variables that were suggestively correlated (p < 0.05) with stigma in their respective domain in the regression model. We entered country first, followed by at least suggestively correlated variables frequently examined in previous autism attitudes literature (autism knowledge, pleasantness, and quantity of contactFootnote 6), personality traits (cultural orientation and tightness), correct autism identification, demographic variables, and neurodiversity-related variables. We eliminated 5, 4, and 4 outliers with studentized residuals > 3 (UCLA, n.d.) from regressions conducted in social, non-speaking, and RIRB domains, respectively.Footnote 7 All three final models met the assumptions of multicollinearity (examined using variance inflation factor and condition number), homoscedasticity (tested using White's test for heteroskedasticity and Breusch-Pagan test for heteroskedasticity), and normality of residuals (assessed using kernel density estimation and standardized normal probability plot) based on the Statistical Methods and Data Analytics (UCLA, n.d.). Finally, we conducted exploratory zero-order correlations between autism stigma toward each vignette and key variables among both the combined sample and by country to further explore if the associations between rater characteristics and stigma vary across contexts.

Results

Initial Comparisons Across Countries and Domains

Exploratory t-tests showed that Koreans reported heightened stigma towards unlabeled autistic behaviors across vignettes, less accurate autism knowledge, and less frequent and pleasant contact with autistic individuals compared to Americans (ps < 0.0001). Koreans were also less likely to report that they were aware of the neurodiversity movement (χ2 = 80.8, p < 0.0001) and reported less neurodiversity endorsement (Mann–Whitney U = 9.0, p < 0.0001) than Americans. When examining only those who were aware of the neurodiversity movement, Koreans still reported less neurodiversity endorsement than Americans (Mann–Whitney U = 2.8, p = 0.005). Koreans also reported heightened cultural tightness relative to Americans (p = 0.001).

A one-way repeated ANOVA revealed that there was a significant effect of domain on stigma, F = 10.64; p < 0.001. Tukey’s post-hoc tests revealed that participants reported significantly more stigma toward characters exhibiting primarily RIRBs (M = 2.89) than characters exhibiting primarily social (M = 2.73; p < 0.0001) or non-speaking characteristics (M = 2.78; p = 0.003). There was evidence suggestive of heightened stigma towards non-speaking relative to social characteristics (p = 0.02). A one-way repeated ANOVA also revealed that there was a suggestive effect of domain on whether participants identified the characters as having a disability other than autism (p = 0.007). Follow-up analyses revealed that the percentages of participants identifying the characters as having a disability other than autism in the non-speaking and social domains were higher than that in the RIRB domain (p = 0.004 and 0.002, respectively). The percentages of participants identifying the characters as having a disability other than autism in the non-speaking and social domains did not differ from each other (p = 0.83). There was no effect of domain on the percentage of participants identifying the characters having autism or no disability (both ps > 0.05).

Exploring Country, Support Needs, and their Interaction as Predictors of Stigma

Separate regression analyses predicting stigma toward each of the domains of unlabeled behaviors from country, support needs condition, and their interaction demonstrated that Korean nationality predicted heightened autism stigma in all three domains (ps < 0.0001). Heightened support needs predicted greater autism stigma only in the RIRB domain (p = 0.004).Footnote 8 The interaction between country and support needs did not predict stigma towards social (p = 0.28) and RIRB characteristics (p = 0.19). Evidence suggestive of an interaction between country and support needs was evident for non-speaking characteristics (p = 0.009). Koreans reported greater stigma toward the non-speaking character with high support needs (M = 3.28) than toward the character with low support needs (M = 2.84; t = -5.19; p < 0.001). There was no difference between Americans’ stigma toward the non-speaking character with high support needs (M = 2.56) and low support needs (M = 2.46; t = -1.04; p < 0.30). See Table 2 for the results of regression analyses predicting stigma with support needs, country, and their interaction.

Table 2 Regression analysis predicting stigma toward unlabeled autistic characteristics with support needs, country, and their interaction
Full size table

Do Individual Differences Account for Differences in Stigma Across Countries?

Exploratory zero-order correlations between stigma and other characteristics in each domain can be seen in Supplementary Tables S3-S5. In the regression predicting stigma in the social domain from individual differences (Table 3), less accurate autism knowledge and lower neurodiversity movement endorsement (both ps < 0.001) predicted greater stigma. Korean nationality (p = 0.02) and less frequent contact (p = 0.01) were suggestively associated with greater stigma towards social characteristics.

Table 3 Summary of regression analysis by domain predicting stigma toward unlabeled autistic characteristics
Full size table

In the regression predicting stigma in the non-speaking domain, less accurate autism knowledge (p = 0.001) and less frequent contact (p < 0.001) predicted greater stigma. Country was no longer associated with stigma in this domain (p = 0.50). Evidence suggestive of associations between stigma towards non-speaking characteristics and the interaction between country and support needs was observed (p = 0.04).

Less accurate knowledge (p < 0.001), lower neurodiversity movement endorsement (p < 0.001), and less frequent contact (p < 0.001) predicted greater stigma in the RIRB domain.Footnote 9, Footnote 10 Support needs (p = 0.003) remained associated with stigma in this model, while country (p = 0.01) and greater cultural tightness (p = 0.02) suggestively predicted heightened stigma towards RIRBs.

Exploratory Analyses: Do Individual Differences Associated with Stigma Vary Across Contexts?

See Supplementary Table S6 ~ S23 for the correlation matrices of the combined sample, American participants, and Korean participants for each vignette. Zero-order correlations split by vignette and country revealed that less accurate autism knowledge and less neurodiversity endorsement were associated with greater stigma toward a character exhibiting social difficulties and high support needs among Americans (all ps < 0.005). Evidence suggestive of associations between stigma and collectivism (p = 0.01) and vertical orientation (p = 0.01) were observed among Americans. No factors were associated with stigma towards this character among Koreans. Less autism knowledge and neurodiversity awareness and endorsement and greater vertical orientation and education were correlated with heightened stigma toward a character exhibiting social difficulties and low support needs among Americans (all ps < 0.005). Evidence suggestive of associations between autism stigma, less accurate identification of autism in the vignette character, not having an autistic nuclear family member, and less collectivism was also observed among Americans (all ps < 0.05). Only cultural tightness was suggestively associated with stigma towards this character among Koreans (p = 0.03).

Less autism knowledge, frequent contact, and neurodiversity endorsement (all ps < 0.005), and greater vertical orientation (suggestive; p = 0.03) and collectivism (suggestive; p = 0.02) were correlated with heightened stigma towards a non-speaking character with high support needs among Americans. Not having a nuclear member was suggestively associated with greater stigma towards this character among Koreans (p = 0.03). Among Americans, less knowledge (p = 0.002) and less neurodiversity endorsement (p = 0.001) was correlated with heightened stigma toward a non-speaking character with low support needs. Among Koreans, greater cultural tightness (p = 0.004) and not having a nuclear family member (p = 0.04) were significantly and suggestively, respectively, associated with heightened stigma towards this character.

Less frequent contact with autistic people, less neurodiversity endorsement (all ps < 0.005), and less collectivism (suggestive; p = 0.04) were associated with greater stigma towards a character exhibiting RIRB and high support needs among Americans. Evidence suggestive of negative associations between stigma, neurodiversity endorsement, autism knowledge, and pleasantness of contact were observed among Koreans (all ps < 0.05). Among Americans, less knowledge (p < 0.001), less neurodiversity endorsement (p = 0.0001), and greater vertical orientation (suggestive; p = 0.01) were correlated with heightened stigma toward a character exhibiting RIRB and low support needs. Greater cultural tightness, less collectivism, and less neurodiversity endorsement were suggestively associated with stigma towards this character among Koreans (all ps < 0.05).

Discussion

Our findings generalize upon prior research by: (1) demonstrating that heightened stigma towards the label autism in South Korea versus the US (Kim et al., 2021) is also apparent for unlabeled autistic characteristics representing social, non-speaking, and RIRB domains, and (2) determining that, as in prior work focused on predictors of stigma towards the label autism (e.g., Kim et al., 2021), greater autism knowledge and more contact with autistic people predicted greater stigma towards at least some unlabeled autistic behaviors. Our findings extend upon prior research by: (1) examining stigma towards non-speaking autistic people, who are underrepresented in autism research more generally and autism stigma research in particular, (2) examining the degree to which autism stigma varies in relation to autistic people’s support needs, and (3) exploring cross-cultural variations in awareness of and agreement with the neurodiversity movement in relation to autism stigma.

Are Support Needs of Autistic People Associated with Stigma?

We observed partial support for our hypothesis that participants would report heightened stigma toward characters with higher support needs. While this pattern was apparent in baseline correlations, once country and the interaction between country and support needs were accounted for in regression analyses, support needs were only associated with stigma in the RIRB domain and not in the social and non-speaking domains. Evidence suggestive of an interaction between support needs and country was observed in the non-speaking domain. While South Koreans stigmatized a non-speaking character with high support needs more than a non-speaking character with lower support needs, Americans did not differentiate based on the support needs of non-speaking characters. Together, these findings suggest that what “support needs' means may vary as a function of the type of characteristic being examined and the cultural context in which it is being examined. Future mixed-methods research should examine intersections between how people interpret different types of “support needs” and stigma in different cultural contexts.

Stigma towards characters exhibiting primarily RIRBs was higher than stigma towards both non-speaking people and those exhibiting primarily social difficulties. Our findings build on prior work demonstrating that job applicants exhibiting RIRBs might be rated more negatively than those exhibiting social difficulties (McMahon et al., 2021) and that non-autistic adults may be more likely to avoid people whose behaviors may be perceived as more disruptive (Gillespie-Lynch et al., 2021b). The characterization of RIRBs in our RIRB vignettes included focused interests, sensory differences, difficulty adjusting to change, and stimming (e.g., jumping up and down and flapping one’s hands). Given recent evidence that focused interests may not impact stigma towards autistic people (Stockwell et al., 2021), future studies should explore if specific types of repetitive behaviors are associated with heightened autism stigma.

Which Rater Characteristics Are Associated with Stigma?

Initial regressions including only country, support needs, and their interaction showed that Koreans reported higher autism stigma than Americans across all three domains. Except for accurate identification of autism, the other hypothesized individual-level variables (i.e., more inaccurate autism knowledge, less positive and frequent contact with autistic individuals, and greater acceptance of inequality and cultural tightness) were correlated with stigma in at least one domain. Once these individual differences of raters were included in models, impacts of nationality on stigma were attenuated for social and RIRB characteristics and no longer apparent in the non-speaking domains.

In regressions including rater characteristics, autism knowledge was associated with stigma across all three behavioral domains as in previous studies (Gillespie-Lynch et al., 2019; Kim et al., 2021; Obeid et al., 2015). Having accurate knowledge may dispel negative stereotypes about autism, resulting in reduced stigma (Gillespie-Lynch et al., 2015). Frequency of contact with autistic individuals was significantly associated with stigma in the non-speaking and RIRB domains and suggestively associated with stigma in the social domain. Frequent high-quality contact with marginalized people is one of the most effective strategies for reducing stigma (Corrigan & Penn, 1999).

Given that support for the neurodiversity movement was associated with reduced stigma towards social and RIRB characteristics, our findings suggest that the neurodiversity movement is indeed associated with reduced stigma towards autistic people. However, baseline correlations by vignette and country revealed that neurodiversity movement support was associated with reduced stigma in all domains among Americans, and was only associated with reduced stigma towards a character exhibiting RIRBs among Koreans. Neurodiversity movement support was also low among Koreans, even after accounting for reduced awareness of the neurodiversity movement. These findings suggest that at least some principles of the neurodiversity movement are not well-aligned with current social norms in South Korea. Therefore, creative work is needed to adapt existing autism trainings, which often emphasize the principles of the neurodiversity movement (e.g., Gillespie-Lynch et al., 2021a; Jones et al., 2021), for South Korean contexts.

As was the case in baseline correlations examining stigma towards the label “autism” (Kim et al., 2021), exploratory zero-order correlations revealed that cultural tightness was correlated with stigma towards unlabeled autistic characteristics only among Koreans. Further, among Koreans, cultural tightness was associated with stigma towards characters exhibiting low support needs across all three behavioral domains, but not towards their counterparts with high support needs. These findings align with Phelan and colleagues’ (2008) reminder that the norm enforcement function of stigma can only be applied when people are viewed as capable of adapting to norms. Our findings suggest that South Koreans are more likely to expect autistic people with lower support needs to adapt to social norms than people with higher support needs. Given that attribution theory proposes that the perceived responsibility for negative behaviors influences emotional and sympathetic responses, resulting in stigmatization (Weiner, 1993) but perceived responsibility was not associated with stigma in a combined sample of college students in Lebanon and the US (Gillespie-Lynch et al., 2021b) and was associated in China (Lu et al., 2021), findings suggest that attribution theory may be more relevant for understanding autism stigma in more culturally tight contexts, like South Korea and China.

Practical Implications

This study has practical implications for designing anti-stigma interventions. Training that aims to increase autism knowledge and endorsement of the neurodiversity movement and provide frequent high-quality contact with autistic people could mitigate autism stigma. However, the lack of consistent associations between neurodiversity awareness and autism stigma in the current study indicates that providing only basic facts about the neurodiversity movement during autism trainings is unlikely to mitigate autism stigma; rather, helping people understand the strengths and benefits of the neurodiversity movement may begin to promote support for the neurodiversity movement, and, consequently, a reduction in autism stigma.

Since high cultural tightness was only suggestively associated with stigma among Koreans, helping Koreans understand how the social appropriateness of different types of behaviors varies across contexts by introducing direct and mediated contact with autistic people who are successful in idiosyncratic ways may help reduce autism stigma in South Korea. Indeed, anecdotal accounts suggest that autism stigma was reduced via a popular show, Marathon, which depicted an autistic person who was happy and successful as a runner (Hwang & Charnley, 2010). Helping Koreans understand the functions served by different autistic behaviors could also help broaden social norms in South Korea. For example, training could highlight work by Kapp et al., (2019) exploring the importance of stimming as a coping mechanism. By emphasizing the functions of “deviant” behaviors, autism training designed for more culturally tight contexts can help people understand that behaviors that look different on the surface may have similar underlying functions to behaviors they themselves do. Finally, interventions to reduce autism stigma should strive to comprehensively address the full spectrum of autism, by ensuring fair representation of non-speaking autistic people and people with other co-occurring conditions.

Limitations

These findings should be considered in light of the following limitations. First, similar to the limitation of previous cross-cultural studies (Kim et al., 2021; Someki et al., 2018), by dichotomizing participants as “Korean” or “American,” we treated each country as a proxy for a single culture. This does not account for the diversity within countries, particularly the US, which is an ethically, culturally, and racially diverse country. Second, data collection was conducted during the COVID-19 pandemic, and participants’ experiences with COVID-19 may have impacted the findings of the study particularly because ecological catastrophes may activate cultural tightness (Jackson et al., 2019). Third, most participants who self-identified as autistic and were subsequently excluded from the analysis were American. It is possible that some Korean participants were in fact autistic but did not reveal their diagnosis or were unaware of their diagnosis due to the low autism awareness and heightened autism stigma in South Korea.

Fourth, one instrument, the adapted Social Distance Scale, was used to measure autism stigma, but the stigma is influenced by various contextual factors, such as the types of relationships expected to be maintained with autistic individuals. For instance, some non-autistic individuals may be more willing to casually spend time with an autistic individual than to start a business together. Fifth, although vignettes allow researchers to manipulate specific components of variables to examine how participants’ responses vary (Gould, 1996), there are drawbacks to using vignettes. Characters described in the vignettes do not comprehensively capture autistic behaviors and characteristics, and participants’ responses to those depicted in the vignettes might not be applicable to other autistic individuals with different characteristics. Sixth, particularly for those who thought that the characters have a disability other than autism, their attitudes toward other disabilities (e.g., ID) may have influenced their ratings of the desired social distance, but we did not examine this due to the small number of participants who identified characters as having a disability other than autism when split by diagnosis type.

Seventh, because we provided the definition of the neurodiversity movement only after participants answered the neurodiversity awareness question, some participants who indicated that they were aware of the neurodiversity movement may not have actually had an accurate understanding of the concept. Finally, the alpha of the PAK-M was relatively low among American participants, raising concerns about the consistency of findings related to autism knowledge. Future research should examine if and how reverse-coded items impact internal consistencies and factor structures of autism knowledge measures across cultures (perhaps via cultural differences in response styles; Suárez-Alvarez et al., 2018; Vigil-Colet et al., 2020).

Conclusions

The current study demonstrates that stigma towards unlabeled autistic behaviors is heightened in South Korea relative to the US, as has been evident in past work examining stigma towards the label “autism.” However, findings also indicate that the mechanisms underlying autism stigma are complex and context-dependent. Extending upon prior work suggesting that individual differences may contribute more to autism stigma than country differences (e.g., Gillespie-Lynch et al., 2019), when characteristics of autistic individuals (i.e., how much support they need and the types of autistic behaviors they demonstrate) and individual differences among raters (autism knowledge, contact quantity, and neurodiversity endorsement) were accounted for, raters’ nationality did not predict stigma towards unlabeled autistic characteristics in the non-speaking domain in the current study. Exploratory correlations revealed that the patterns of associations between other variables and stigma towards autistic characteristics often differed as a function of participants’ nationality as well as the types of autistic characteristics highlighted in the vignettes. Findings suggest that stigma towards the same condition, autism, may vary as a function of cultural context and the degree to which an autistic individual is perceived as needing help from others. These findings highlight the need for further studies examining how contexts influence associations between rater characteristics and autism stigma (e.g., using mediation/moderation analyses).

Together, these findings highlight the need to develop culturally adapted anti-stigma trainings that include representations of non-speaking autistic people and other people who are often overlooked in autism research, e.g., those with ID (Russell et al., 2019). We suggest that the efforts to increase autism knowledge and help individuals appreciate and endorse the neurodiversity movement (e.g., by helping people understand both autistic strengths and the underlying reasons for behaviors that may look different) may help reduce autism stigma in both South Korea and the US.