Abstract
Autistic and non-autistic observers form less favorable impressions of autistic adults relative to non-autistic adults, yet only non-autistic adults report lower interest in socially interacting with them. This discrepancy may relate to differences in how autistic and non-autistic people value and interpret social traits, with autistic people viewing characteristics such as “awkwardness” in neutral or positive terms rather than as a deterrent to interaction. Here, 24 autistic adults and 24 non-autistic adults rated themselves on 20 traits, their preferences for these traits in others, and their perception of them in an “average” person. Autistic participants rated themselves as more awkward, less socially keen, and less socially “normal” than non-autistic participants. Self-ratings and preferences in others correlated on three attributes—awkward, socially keen, and socially “normal”—with the autistic group demonstrating a large effect (r = .70) for partners similar to their self-perception of “normal”. Collectively, these findings suggest that common social descriptors like awkward have different appeals for autistic people and should not be universally assumed to be undesirable characteristics. Future studies should seek to replicate these findings with larger and more diverse samples and more established measures.
Autism is diagnosed in part by “persistent deficits in communication and social interaction” (APA, 2013), yet external factors, including stigma and treatment by non-autistic people, are increasingly being recognized as significant contributors to autistic social disability (Botha & Frost, 2020; Turnock et al., 2022), Stigma towards autism contributes to social exclusion, reduces personal and professional opportunities, and increases stress and mental health challenges for autistic people (Botha et al., 2022: Han et al., 2022). One way that autism-related stigma manifests among non-autistic people is in negative impressions of the social, expressive, and communicative characteristics of autism (Mitchell et al., 2021; Turnock et al., 2022). These impressions, particularly judgments of “awkwardness,” strongly differentiate non-autistic evaluations of autistic and non-autistic people (Grossman, 2015) and are associated with reduced social interest in interacting with autistic adults (Sasson et al., 2017). In turn, this process creates barriers to social inclusion and reduces opportunities for autistic people (Mitchell et al., 2021).
Autistic adults also evaluate autistic people as more awkward than non-autistic controls (DeBrabander et al., 2019). Unlike non-autistic adults, however, these impressions are not associated with reduced interest in socially interacting with autistic people (DeBrabander et al., 2019). This finding suggests that autistic adults may not hold biases against “awkward” social presentations the way that non-autistic people do. Indeed, autistic adolescents and adults do not express the non-autistic preference for interacting with other non-autistic people (Chen et al., 2021; Morrison et al., 2020), are less likely to object to autistic social differences (Sng et al., 2020), and often show higher empathy (Komeda et al., 2019), greater rapport (Crompton et al., 2020a), and better communication (Crompton et al., 2020b) with other autistic people compared to non-autistic ones. Such findings suggest that similarity in dispositional characteristics, rather than individual social “skill,” underlies affiliation, connection, and understanding for autistic and non-autistic people (Milton, 2012).
The term “awkward” is a loosely defined descriptor signifying broad social, expressive, and/or motor characteristics that deviate from privileged norms (Tashiro, 2017), or uncomfortable moments of social disconnection common to interactions between members of stigmatized and non-stigmatized groups (Hebl et al., 2000). Autistic people can differ socially in several observable ways that may increase perceptions of awkwardness, particularly among their non-autistic peers. They often use gaze differently and engage in less eye contact (Trevisan et al., 2017), demonstrate non-normative expressivity in their faces (Faso et al., 2015) and voices (Hubbard et al., 2017), exhibit different preferences in communication (Crompton et al., 2021) and conversational dynamics (Rifai et al., 2022), and move in atypical ways (Lum et al., 2021). For autistic people who are often marginalized and struggle in non-autistic social spaces, these “awkward” social presentations and communication styles may not be viewed negatively but rather may be perceived as indicators of potential social compatibility (Granieri et al., 2020). This raises the possibility that certain traits such as “awkwardness” may confer different values and appeal to autistic and non-autistic people, with autistic but not non-autistic adults viewing it as a sign of homophily and potential social connection (Black et al., 2022).
Such a finding would indicate that autistic people, like non-autistic people, have preferences for similarity that drive social affiliation (McPherson et al., 2001), and indicate that certain social descriptors like “awkwardness” may not be universally viewed negatively by all people. Here, building upon previous studies of self-perception of personality in autism (Robinson et al., 2017; Schriber et al., 2014), we test whether autistic and non-autistic adults differ in 1) how they view themselves on personality and character traits; 2) their preference for these traits in others; and 3) their perception of these traits in an “average” person. We also explore whether self-assessments on traits such as “awkward” relate to preferences for the same traits in other people.
Methods
Participants
Forty-eight adults (24 autistic and 24 non-autistic) participated in this study. This sample size was determined based on the similar sample sizes of related studies examining self and other trait assessment in autism (e.g., Johnson et al., 2009; Schriber et al., 2014; Usher et al., 2018) that were sufficient for detecting medium to large effects. Thirteen autistic adults were recruited from a university registry of adults with diagnoses of autism or autism spectrum confirmed by ADOS-2, Module 4, revised algorithm (Hus & Lord, 2014). The other eleven autistic participants were recruited from the university and the local community (a large metropolitan area in the Southwest) and self-reported that they had previously received a clinical diagnosis on the autism spectrum (i.e., Autism Spectrum Disorder or Asperger Syndrome) that could not be ADOS-confirmed due to Covid-related precautions. Non-autistic adults were undergraduate students participating for course credit who were selected from a larger screened sample to be demographically comparable to the autistic sample on age and gender (for sample demographics, see Table 1). The current study was approved by the university's Institutional Review Board.
Sample demographics
Measures and procedures
Self and other autism personality scale (SOAPS)
A novel questionnaire consisting of 20 personality traits as well as characteristics relevant to autism was compiled for this study based upon validated items from Five Factor Models of Personality (Mullins-Sweatt et al., 2006) and prior work assessing impressions of autistic people (Sasson et al., 2017): Awkward, Attractive, Trustworthy, Dominant, Likeable, Intelligent, Orderly, Extroverted, Direct, Socially Keen, Socially Normal, Comfortable, Expressive, Confident, Patient, Empathetic, Warm, Manipulative, Mature, and Funny. Items are rated on a Likert scale from 1 (extremely low) to 5 (extremely high), with 3 indicating neither high nor low. Synonyms for high and low levels of each item are provided for context and clarity.
Three versions of the SOAPS were administered that were identical, except participants either rated themselves (i.e., “Rate yourself…”) on each of the 20 traits, or they rated how much they prefer the traits in other people (i.e., “How much do you like these traits in other people?”), or they rated the traits in an “average” person. All three versions of the SOAPS demonstrated high internal consistency, with Cronbach's alphas of .85, .82, and .83 for the self, preferred others, and “average” person versions, respectively.
Data were collected online using REDcap software. Participants provided informed consent before completing three versions of the SOAPS in a randomized order. Items within each version were also presented in a randomized order.
Analysis plan
Although the autistic and non-autistic groups each completed three versions of the SOAPS, primary analyses were restricted to between groups comparisons to test study aims concerning group differences in self-ratings, preferences in others ratings, and perceptions of the “average person”. A MANOVA was first conducted to determine whether the autistic and non-autistic groups differed in their overall self-ratings on the SOAPS. The alpha level for these MANOVAs was set at .05. A significant result would indicate that autistic and non-autistic participants viewed their personality characteristics differently, and follow-up independent t-tests would be used to specify the items on which they differ. The same process was then repeated with the “preferences in others” version of the SOAPS, and then again with the “average-person” version. To correct for multiple comparisons, the alpha level for follow-up tests was set to .01. Given the relatively low power of the current study, results with p-values ≤ .05 that did not survive correction are also noted but should be considered preliminary and not interpreted with the same level of confidence. Directional Bayes factors are included to aid with interpretation. Bayes factors can provide an estimation of the evidence in favor of the alternative hypothesis even when power is relatively low, allowing for more nuanced interpretation of the data (Rouder et al., 2009). Values over 3 and under 10 indicate moderate support of the alternative hypothesis, between 10 and 100 indicate strong support, and over 100 indicate extreme support (van Doorn et al., 2021). Cohen's d effect sizes are also provided for all group comparisons.
Next, pearson's r correlations were used to explore whether participants preferred other people who were similar to how they viewed themselves. To minimize the number of analyzes conducted, correlations were only conducted on the traits in which autistic and non-autistic participants differed at p ≤ .05. Finally, to determine whether autistic and non-autistic participants differed in how they view themselves relative to their perception of the “average” person, paired sample t-tests were conducted for each group on the self and “average” person for the traits in which the groups differed on both self and “average” person ratings.
Positionality statement
This study was conceived, administered, and interpreted by a collection of autistic and non-autistic researchers. The co-first authors of this paper identify as autistic and/or autistic/ADHD, and another author identifies as neurodivergent.
Results
The autistic and non-autistic groups did not significantly differ on age (t (46) = .22, p = .93), self-identified gender (X2 (2, N = 48) = .65, p = .72), self-identified race (X2 (3, N = 48) = 4.83. p = .19), and maternal education (both groups 4.3 on a 6-point scale, t (46) = 0.0, p = 1.0).
A full comparison of autistic and non-autistic ratings of personality traits in the self, in preferred others, and in the perceived average person can be viewed in Table 2. The overall MANOVA for self-ratings was significant (F (1, 46) = 4.13, p < .001), and follow-up independent sample t-tests indicated that the autistic group rated themselves as significantly more awkward (p < .01; BF = 25.35), less socially keen (p < .001; BF > 100), less socially normal (p < .001, BF > 100), and less funny (p < .01, BF = 12.2) than non-autistic participants. Similarly, the overall MANOVA for ratings of trait preferences in others was significant (F (1, 46) = 2.13, p = .03). Follow-up independent sample t-tests indicated that the autistic group preferred other people who are less socially normal (p = .02; BF = 6.10), less extroverted (p = .03; BF = 4.56), more awkward (p = .04; BF = 3.56), more intelligent (p = .04; BF = 3.17), and less socially keen (p = .05; BF = 2.76) than non-autistic participants, though these comparisons did not survive correction for multiple comparisons. Finally, the overall MANOVA for ratings of traits in an “average” person was not significant (F (1, 46) = 0.78, p = .719), with the groups only significantly differing in the autistic participants perceiving an “average” person to be more confident (p = .01; BF = 8.27) and more socially keen (p < .01; BF = 22.17) than non-autistic participants, with more extroverted (p = .02, BF = 4.87) and less awkward (p = .05; BF = 2.97) also emerging but not surviving correction.
Comparison of autistic and non-autistic ratings of personality traits in the self, in preferred others, and in the perceived average person.
Displayed are mean ratings on the Self, Preferred Others, and Perceived Average Person versions of the “Self and Other Autism Personality Scale”. The range of each item was 1 (extremely low) to 5 (extremely high).
Note: * ≤ .05; ** ≤ .01; ** ≤ .001
NA: Non-Autistic.
Pearson's r correlations between self and preferred other ratings for the three traits on which autistic and non-autistic participants differed showed significant positive associations: awkward (r = .39, p < .01), socially keen (r = .32, p = .03), and socially normal (r = .66, p < .001). When self-other correlations were conducted independently on each group, effect sizes were generally larger in the autistic compared to the non-autistic group, but due to reduced power relative to the full sample only the self-other correlation for socially normal reached significance (autistic group: r = .70, p < .001; non-autistic group: r = .53, p < .01).
Finally, whereas the autism group rated themselves significantly higher than their perceived average on awkwardness (t (23) = 4.87, p < .001) and lower on being socially keen (t (23) = -.383, p < .001), the non-autistic group demonstrated the opposite pattern on socially keen: they rated themselves as significantly more socially keen than their perceived average (t (23) = 5.13, p < .001). Self and perceived average ratings on awkward did not significantly differ for the non-autistic group (t (23) = -.15, p = .88).
Discussion
The current study examined whether autistic and non-autistic adults’ self-perception of their character traits relate to their preference for the same traits in other people. Results showed that autistic adults rated themselves as more awkward, less socially “normal,” and less socially keen than non-autistic adults. Additionally, there was preliminary evidence that autistic participants preferred other people who are more awkward, less socially “normal”, and less socially keen than did non-autistic participants, though these effects did not survive correction. However, self-ratings and preferred-other ratings on all three of these social traits were significantly correlated, with autistic participants showing a particularly strong correspondence between how “normal” they viewed themselves and how “normal” they preferred their social partners to be.
These findings suggest that normative assumptions about the desirability of certain social traits may not extend in the same way to social preferences for autistic people. They also help explain why autistic people do not share the non-autistic tendency to associate awkwardness with reduced social interest (DeBrabander et al., 2019) and often establish better rapport and communicate more effectively with other autistic people (Crompton et al., 2020a, 2020b). Autistic people, like all people, may prefer and connect better with those who overlap with them in experiences, sensibilities, and modes of understanding (Milton, 2012). One implication of this finding is that uniformly rank-ordering social traits according to normative standards may fail to recognize how compatibility on these traits facilitates connection and affiliation, particularly between neurologically divergent people. Clinicians, educators, and other professionals encouraging, instilling, or training non-autistic social value systems should consider whether routes to social well-being may differ for autistic people.
Interestingly, group differences in self and preferred other ratings were generally constrained to traits related to social presentations and social understanding (e.g., awkward, socially normal, socially keen). Self and preferred other ratings for most other traits (e.g., mature, trustworthy, patient) did not differ between groups, indicating that autistic adults primarily differentiate themselves from non-autistic adults on social descriptors common to autism (Grossman, 2015; Humphrey & Lewis, 2008), with greater similarity between groups on less autism-relevant traits. Although these findings suggest that autistic adults may accurately assess themselves as socially different from the norm, further analyses indicated that autistic and non-autistic people even differ in their perception of what the “norm” is. Autistic participants perceived “average” people to be less awkward, more extraverted, more socially keen, and more confident than did non-autistic participants, which may reflect internalized beliefs in autistic participants about their own social characteristics and signify that they anchor their perception of the “average” to how they view themselves. Autistic people may also demonstrate semantic autobiographical differences in which they are more likely to perceive themselves in ways others have described them in the past (Robinson et al., 2017), and these descriptions may emphasize differences from typicality.
Findings from this study should be interpreted in the context of some limitations. The non-autistic sample was drawn from university psychology students, the autistic sample consisted of individuals with lower support needs, and both groups were majority White and young adults, all of which limit generalizability. Further, although the autistic and non-autistic participants were similar demographically, the overall sample size was modest and may have been underpowered to detect more subtle effects. For example, some effects did not survive correction for multiple comparisons, particularly group differences in autistic and non-autistic “preferences in others”. Although Cohen's d and Bayes factors for these items correspond to moderate effects, these findings should be considered preliminary and warrant replication before interpreting with confidence. Additionally, the modest sample size precluded the ability to examine moderating effects of gender, age, and other demographic characteristics. Finally, a novel survey (SOAPS) was used to measure trait assessment. Although items on this survey were taken from established measures, demonstrated high internal consistency, and produced several large effects, the SOAPS itself requires further validation. Future studies may also want to examine whether autistic and non-autistic participants interpret the SOAPS differently, as has been demonstrated on other self-report measures (Gernsbacher et al., 2017), and whether results would differ if participants reported traits that they like in a person they know rather than in the more hypothetical context used here.
In conclusion, findings from this study suggest that autistic and non-autistic adults differ in some of their trait perceptions and preferences, particularly in those related to sociability. Specifically, traits associated with social “normality” not only best differentiated how autistic and non-autistic participants viewed themselves, but also correlated with their preferences for the same social characteristics in other people. These findings challenge assumptions of “universal desirability” of social traits and align with a conceptualization of autistic sociability that emphasizes relational compatibility rather than individual skills or traits as determinative of social outcomes.
Footnotes
Declaration of conflicting interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
