Quantifying Social Skill Deficits and Strengths Profiles in Autistic Youth

Abstract

While social difficulties in autism are well-established, questions remain regarding whether these represent challenges in acquiring or performing such skills, reduced social strengths, or a unique distribution across these domains (i.e., social profile). This study empirically derived social profiles of 211 autistic and non-autistic youth (M_age= 13.50; Autistic N= 150; Male N = 151; 85.3% White). Assessments occurred between 2016 and 2020. Results showed that autistic youth exhibit significantly more social acquisition and performance deficits and fewer strengths than non-autistic youth (ds = −.44 − .65). Performance deficits were most – and acquisition deficits least – prominent within autistic profiles, potentially implicating longstanding theoretical models of social difficulties in autism, and supporting new, idiographic approaches for conceptualizing, assessing, and treating social challenges.

Challenges in social interaction are considered the coherent, defining feature of autism (Hobson, 2014), and often represent a source of subsequent difficulties in this population. Indeed, difficulties engaging in prosocial behaviors impact the development and maintenance of interpersonal relationships and often lead to poor long-term outcomes such as loneliness and mental health difficulties across neurotypical (Allen et al., 2015) and autistic people (Schiltz et al., 2021). Because of this, a deeper understanding of what exactly comprises social functioning, how these skills develop, and accurate methods of assessment is crucial for this population.

It has long been assumed that autistic individuals struggle primarily with enactment of consistent and successful social behavior, often called social acquisition. This presumption has guided both theoretical and clinical understanding of autism (i.e., claims that autistic individuals never or very rarely engage in prosocial behavior; Russel et al., 2012), as well as intervention approaches, many of which rely on the explicit teaching of social behavior to aid youth in acquiring behaviors they are presumed not to have the ability to otherwise produce (Matson et al., 2007). Notably, the approach of utilizing intervention approaches with the aim of increasing acquisition has led to a predominance of interventions that didactically teach rules and norms directly (i.e., so that youth are aware of how to execute behaviors that are seemingly novel to them), even when such approaches are acknowledged as insufficient for achieving durable changes in social interaction (Bishop-Fitzpatrick et al., 2017; Lerner & White, 2015). Indeed, recent work highlights that while many autistic youth do acquire knowledge (i.e., they can state what a correct behavior should be) of some social skills through didactic social skills interventions (Bottema-Beutel et al., 2018; Gates et al., 2017; Guivarch et al., 2017), many continue to struggle with consistently engaging in those behaviors, exhibiting so-called social performance deficits (Lerner & Mikami, 2012; Mendelson et al., 2016; White et al., 2007).

The distinction between social acquisition and social performance deficits undergirds many leading theories of social functioning in typical and atypical populations (Beauchamp & Anderson, 2010; Crick & Dodge, 1994; Mendelson et al., 2016). Despite the foundational role of distinguishing between these two types of social deficits (Bishop-Fitzpatrick et al., 2017; Gresham, 1981; Marro et al., 2019; Mendelson et al., 2016), direct investigation into their prevalence and correlates in youth populations has been hampered by the lack of methods to detect them – that is, they have relied upon a theoretical - but not empirical - distinction. In fact, few studies have sought to directly examine the presumption that social acquisition (and not social performance) deficits are most characteristic of autism at all. However, a recently-developed methodology allows for these deficits to be examined more directly (Gresham et al., 2010), and has been used in clinical and non-clinical populations (Aduen et al., 2018; Gresham et al., 2010). Thus, it provides promising opportunities to quantitatively explore the distribution of these types of social deficits in autism, thereby opening the door to interrogation of the foundations of longstanding theories of social functioning in autism (Bishop-Fitzpatrick et al., 2017; Koenig et al., 2009; Mesibov, 1984) and providing opportunities for identifying and targeting the operative deficits at play for a given individual. The current study aims to utilize this new methodology to better characterize social functioning differences between autistic and non-autistic youth.

Social Acquisition vs. Social Performance

Difficulties with social interaction in youth have been theorized to result from impairments in the acquisition of social behaviors (e.g., never or very rarely engaging in a behavior, arising from a lack of procedural knowledge of how to utilize a given social skill, or not being able to distinguish which social skill is appropriate to a given situation; Gresham, 1981; Gresham et al., 2010), failure to consistently perform a given social skill at the appropriate time (e.g., knowing how and when to perform a social behavior, but not reliably taking advantage of the opportunity to do so as a result of interfering cognitive or perceptual processes; Guivarch et al., 2017; Keifer et al., 2020), or both. The functional meaning of this distinction (never vs. sometimes engaging in a given behavior) arises from a disciplinary emphasis on observable behaviors: whether an individual can engage in a given behavior can be inferred from whether or not they have ever demonstrated the procedural skill to do so (Aduen et al., 2018).

Deficits in social acquisition are thought to arise from several potential sources. Most frequently, social acquisition deficits are attributed to lack of social knowledge (i.e., not knowing what behaviors one should deploy in a given situation, or how to engage in a given social skill at all; Gresham et al., 2010). Social performance deficits arise when an individual does not reliably enact a (functionally important) social behavior, despite having sufficient knowledge of it to do so at least some of the time (Gresham, 1981). Many hypotheses exist regarding the source of social performance deficits. For instance, a failure to enact a known social behavior can arise because an individual has not successfully understood another’s emotional state, has missed a given social cue, is not processing information quickly enough, or for one of a myriad of other reasons, sometimes called performance-related factors (Gresham, 2016; Mendelson et al., 2016). However, the unifying feature of these factors is that they impede enactment of prosocial behavior that an individual otherwise knows (and has demonstrated that they know) how to do.

Though successful utilization of these two processes (i.e., knowing what to do, then utilizing skills that enable enactment of such known behavior) is essential for successful social functioning (Gresham et al., 2010), social acquisition and social performance skills are conceptually distinct, and the literature consistently indicates that they should be considered as such when assessing social problems and developing intervention plans to improve social functioning (Frey et al., 2011; Gresham, 2016). Indeed, the preponderance of general (Crick & Dodge, 1994) and clinical population-specific (Beauchamp & Anderson, 2010; Mendelson et al., 2016) models of social competence rely on a distinction between social acquisition and social performance as a core means of specifying between-person variation in functional social outcomes. Nonetheless, to date, few tools exist to distinguish these constructs during assessment and intervention, particularly for clinical populations such as those with autism.

Social acquisition & performance in Autism.

Social acquisition deficits are often the default assumption in the literature on social functioning in autism (Bishop-Fitzpatrick et al., 2017). Such acquisition difficulties have frequently been assumed to result from a dearth of social knowledge; indeed, there is a long history of identifying autistic youth as either not engaging in given prosocial behaviors or not knowing how to do so (Mesibov, 1984). However, the possibility that autistic children already know skills that are not being consistently being performed has seldom been considered (Bishop-Fitzpatrick et al., 2017; Keifer et al., 2020; Koenig et al., 2009).

Recent work suggests that focusing solely on social acquisition deficits is either incorrect (Guivarch et al., 2017; Lerner & Mikami, 2012) or insufficient (Keifer et al., 2020; McMahon et al., 2013; Mendelson et al., 2016) in characterizing social challenges among autistic individuals. Moreover, conceptualizations, and subsequent interventions, focusing solely on social acquisition deficits in autism generally fail to consider the processes related to social performance necessary for skill generalization (Bottema-Beutel et al., 2018; Choque Olsson et al., 2017; Ke et al., 2017; Lerner & White, 2015), and as a result, individuals continue to struggle socially outside of the specific interventions. Additionally, many measures purporting to comprehensively assess social skills interventions focus specifically on social acquisition, while not fully considering whether the identified behaviors are being applied reliably across settings (Gates et al., 2017; Wolstencroft et al., 2018).

More recently, efforts have been made to better understand social performance deficits in autism (Corbett et al., 2016; Lerner & Mikami, 2012; Marro et al., 2019) and evidence continues to point towards social performance deficits as particularly impairing (Keifer et al., 2020; Mendelson et al., 2016). Indeed, recent work has highlighted social performance of skills (e.g., planning, processing, and enacting behaviors) as necessary, if not more important, for successful social interaction (Corbett et al., 2016; Lerner & Mikami, 2012).

The primary conceptualization of social performance deficits in autism also relates to social knowledge – specifically that it may be present but underutilized. Though autistic individuals may develop appropriate social knowledge (i.e., know what to do), differences in social performance may inhibit the reliable enactment of known social skills (Gates et al., 2017; Jonsson et al., 2016; Ke et al., 2017), rendering gains in knowledge as being less relevant during in-vivo social situations (Callenmark et al., 2014; Mendelson et al., 2016).

Importantly, the social performance deficit model aligns with other contemporary theories of autism. For instance, a child who knows how to initiate a conversation but does not always exhibit typical signs of motivation to do so (Chevallier et al., 2012; Jaswal & Akhtar, 2019) may present with social performance challenges; so, too, might a teen who struggles to reliably predict responses to given actions in their environment (Cannon et al., 2021), or one who misreads facial emotional cues (Lozier et al., 2014) engage in an effective social response less consistently than their peers. Thus, ample current empirical and theoretical work on social competence in autism points to social performance deficits as being fundamental; however, little work has sought to empirically identify such challenges at all, let along distinguish them within an autism profile. However, a recently-developed scoring technique of a common assessment tool offers the novel opportunity to evaluate the distribution of social acquisition and performance deficits (as well as concurrent skills) in youth — creating a Social Skills Deficits and Strengths Profile (SSDSP).

Social Skills Strengths and Deficits Profile

Building off theories of acquisition and performance deficits (Gresham, 1981), Gresham et al. (2010) developed an alternative scoring method for the Social Skills Improvement System (SSIS; Gresham & Elliott, 2008) to investigate base rates of these phenomena in the general population. The Social Skills Improvement System (Gresham & Elliott, 2008) is commonly used to examine social skills difficulties for autistic and non-autistic children and enables targeted assessment of several important areas of social functioning.

Based on a combination of frequency of skill performance and subjective importance ratings of skill behavior, social skills can be classified into acquisition deficits, performance deficits, and social skill strengths (Gresham et al., 2010). In this framework, acquisition deficits are those important social behaviors that are not performed because a child has failed to learn the skill and therefore cannot perform it. Performance deficits are skills in which the child knows what the skill is but does not readily nor consistently performing it (i.e., needs assistance, only performs under certain conditions, etc.). Strengths are those that are reliably and consistently performed (Gresham et al., 2010).

Using this method, base rates have been demonstrated in a pre-school (Frey et al., 2011), a school-aged general population (Gresham et al., 2010), and an ADHD sample (Aduen et al., 2018). In school-aged and ADHD populations (Aduen et al., 2018; Gresham et al., 2010), rates of acquisition deficits were low in both, while performance deficits were significantly higher in ADHD than non-ADHD youth. Importantly, these studies provided essential information about social skills presentations, and specifically allow more detailed understanding of social skills difficulties in the ADHD population, which could be applied to future assessment and social skills intervention planning (Aduen et al., 2018).

Despite long conjecture between the role of acquisition and performance deficits in social difficulties for autistic youth (Bishop-Fitzpatrick et al., 2017; Mendelson et al., 2016), to date no study has directly investigated the distribution of social deficits in autistic individuals using this framework. Such an investigation would be an essential step towards answering basic questions long hypothesized, but rarely tested in the field, such as: do autistic youth exhibit more social acquisition deficits than their peers? Do they exhibit such deficits more frequently than social performance deficits, or even than strengths? These fundamental questions guide conceptualizations of the nature of the social challenges experienced by autistic people, and guide treatment decisions, so there is urgent need to address them. Therefore, this study sought to characterize the relative contribution of acquisition and performance deficits in autistic youth as instantiated in the SSDSP.

Hypotheses

The current study sought to utilize the SSDSP model to characterize the relative contribution of acquisition deficits, performance deficits, and social skill strengths to social functioning in autistic youth. Although the use of the SSDSP has been validated in pre-school (Frey et al., 2011), school-age (Gresham et al., 2010), and ADHD youth (Aduen et al., 2018), this is the first study to attempt to systematically investigate and differentiate social acquisition vs. performance deficits in autistic youth using this alternative scoring method.

Therefore, we first aimed to examine the psychometric properties of the SSDSP for examining social functioning in autistic and non-autistic youth. We hypothesized that the SSDSP would (1a) show adequate internal consistency in autistic and non-autistic youth and (1b) show similar convergent and predicted divergent validity in autistic and non-autistic youth. Next, when using the SSDSP to assess between-group differences, we hypothesized that (2a) autistic youth would exhibit significantly more acquisition deficits than non-autistic youth, (2b) autistic youth would show significantly more performance deficits than non-autistic youth, and (2c) autistic youth would show fewer social skill strengths than non-autistic youth. Third, though acquisition deficits have long been implicated in autism-related social problems (Bishop-Fitzpatrick et al., 2017; Koenig et al., 2009; Mesibov, 1984), a growing literature suggests performance deficits are more strongly implicated in social functioning difficulties in autism (Keifer et al., 2020; Lerner & Mikami, 2012; Mendelson et al., 2016). Thus, when using the SSDSP to assess within- autistic group profiles, we hypothesized that (3a) autistic youth would exhibit more performance deficits than acquisition deficits, (3b) more performance deficits than skill strengths, and (3c) more acquisition deficits than skill strengths.

Method

Procedures

Informed consent was obtained from all study participants. Data were collected between January 2016 and March 2020. Data were obtained from two larger studies (one assessment and one treatment study; treatment study data included baseline assessment only) at a major university in the New York metro area. All parents participated in a detailed series of evaluations, including the CASI-5, the BASC-2 and the SSIS-RS. Each child participated in the Autism Diagnostic Observation Schedule-Second Edition (ADOS-2; Lord et al., 2012), which is the gold-standard diagnostic system for autism spectrum disorders, and was administered by a certified, clinically trained examiner. Those scoring above the clinical cutoff were assigned to the autistic group, those below were assigned to the non-autistic group. Participant IQ data was also available based on Kaufman Brief Intelligence Test-Second Edition assessment scores. Demographic data (e.g., gender, age, socioeconomic status, race and ethnicity, prior diagnostic information) were completed by parents and was used to assess divergent validity, in which we expected age, sex assigned at birth, parent education and income, and race (white vs. non-white), to be uncorrelated with SSDSP domains. Non-English-speaking participants and participants with IQ < 70 were excluded.

Participants

Data were aggregated across several research studies conducted in the Atlantic region of the United States. The sample consisted of data from 211 children (Autistic N= 150) aged 7.73 to 17.88 (M= 13.50, SD= 2.36; Male N= 151; See Table 1). Sample ethnicity included youth who identified as White (85.3%), Black or African American (2.8%), Asian (5.2%), Hispanic or Latino (8.1%), other (4.7%), or declined to answer (1.9%). All participants and their parents were English-speaking (See Table 1 for detailed information per group).

Table 1.

Mean Percentages of Social Skills Deficits and Strengths Profiles

Variable	Autistic			Non-Autistic			Cohen’s d (95% CI)	p b
	M		SD	M		SD
N (Male/Female)	150 (115/35)			61 (36/25)				.01
Age	13.47		2.53	13.55		1.87	0.03	.83
Ethnicity:
Caucasian		84.7%			86.9%			.68
Hispanic		5.3%			14.8%		.16c	.02
Black/African American		2.7%			3.3%			.81
Asian		6.0%			3.3%			.42
Other		5.3%			1.6%			.23
Decline to answer		2.0%			1.6%			.86
ADOS CSa	7.77		1.89	1.90		.86	−3.53	<.001
SSIS SS	80.14		15.82	91.57		19.78	0.67	< .001
FSIQ	100.79		16.26	106.85		14.39	0.39	.01
SSDSP
Acquisition	9.23		11.73	4.45		8.39	−.44 (−.74 – −.14)	<.01
Performance	34.39		15.83	24.80		20.67	−.55 (−.85 – −.25)	<.001
Strength	15.59		15.83	28.51		25.36	.68 (.37 – .98)	<.01

Note. CS = Comparison Score. FSIQ = Full Scale Intelligence Quotient. M = Mean. SD = Standard Deviation. SSDSP = Social Skills Deficits and Strength Profile. SSIS = Social Skills Improvement Scale Standard Scores. Participants were able to select multiple ethnic groups based on their child’s identity.

^aTotal N = 210; Autistic N = 150; Non-Autistic N=60.

^bCategorical variables were compared via χ² analyses and continuous variables were compared via independent samples t-tests.

^cEffect size represents Cramer’s V.

Measures

Autism Diagnostic Observation Schedule- Second Edition (ADOS-2; Lord et al., 2012)

The ADOS-2 is a standardized researcher-administered assessment consisting of standardized activities and questions measuring communication, social interaction, and play skills, which has proven to be a reliable and valid measure for assessing autism symptoms (Lord et al., 2012). Modules 3 and 4 of the ADOS-2 is used to assist in the diagnosis of autism spectrum disorders in verbally fluent children and adults and were administered by research-reliable examiners for study inclusion. The modules involve structured and unstructured social situations that allow for observation of spontaneous social behaviors and response to social cues as well as restrictive and repetitive behaviors (RRBs). The ADOS-2 describes the participant’s reaction to social “presses” during the assessment session and can help to evaluate the presence of symptoms associated with a diagnosis of autism. The ADOS-2 Comparison Score (CSS), a standardized, continuous conversion of ADOS-2 raw scores indicating autism-related symptoms, was also used as a measure of convergent validity. Higher ADOS-2 CSS scores indicate higher levels of autism symptomatology.

Behavior Assessment System for Children – Second Edition (BASC; Reynolds & Kamphaus, 2004)

Parents completed the BASC-2, which is a widely used measure of youth psychopathology and behavior, which yields T-scores for each scale. The Social Skills subscale of the BASC-2 was used to index social skills symptoms. Further, this measure has been used to assess the construct validity of the SSDSP in ADHD and non-ADHD youth and has demonstrated expected correspondence with SSDSP domains (Aduen et al., 2018). Higher scores indicate better social functioning. The Social Skills subscale was used to represent parent-reported symptoms and as a measure of convergent validity.

Child and Adolescent Symptom Inventory – Fifth Edition (CASI-5; Gadow & Sprafkin, 2012)

The CASI-5 is a parent-report behavior rating scale of children’s symptoms of behavioral disorders, which are directly derived from the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5; American Psychiatric Association, 2013). This measure has demonstrated strong psychometrics in autistic populations (Gadow & Sprafkin, 2012; Kim et al., 2019). The CASI-5 includes an Autism severity score and an ADHD Combined severity score, where higher scores indicate more informant-endorsed symptoms of a DSM-5 diagnosis for these disorders. Established internal consistency of the CASI-5 subscales are generally satisfactory (Kim et al., 2019). The Autism and ADHD Severity scores were used to represent parent-reported symptoms and as a measure of convergent validity.

Kaufman Brief Intelligence Test-Second Edition (KBIT-2; Kaufman, 2004)

The KBIT-2 is a brief measure of intelligence used with individuals ages 4–90. The KBIT-2 was used as a screener for study eligibility and includes measures of verbal, non-verbal, and overall composite of intelligence quotient (IQ). Standard scores are calculated using age-specific norms and is commonly used to measure intelligence in autistic populations (Marro et al., 2019). The KBIT-2 Composite Scores were used to represent the participants’ full-scale intelligence (FSIQ) and as a measure of convergent validity.

Social Skills Improvement System (SSIS; Gresham & Elliott, 2008)

Parents completed the SSIS, a measure which allows assessment of social skills, problem behaviors, and academic competence. The method of social skills evaluation in the SSIS, and thus the resulting SSDSP, relies on the assumption that that we can gather information on knowledge and mastery of a social skill based on how often an individual observably performs that skill, as well as based on the raters’ subjective importance of the behavior (Gresham et al., 2010). 46-items on the SSIS measure social skills specifically, which were used to create the SSDSPs in the present study. Parents rated each of these 46 items twice, first on the frequency of skill performance and second on the importance of that same skill. Frequency was assessed on a 4-point Likert scale (never, seldom, often, almost always) and importance was assessed on a 3-point Likert scale (not important, important, critical). Higher scores indicate higher social skills frequency and higher importance. The internal consistencies of the frequency and importance ratings in this sample were excellent (Cronbach’s α = .96 and .94).

SSIS SSDSP Alternative Scoring Procedure.

Using the Gresham et al. (2010) alternative scoring method, SSDSPs were created to assess the distribution of social skill deficits and strengths for each individual using the following procedure. Based on the frequency and importance rated by parents on the 46-items assessing social functioning, items were categorized into one of three mutually exclusive domains (acquisition, performance, strength). Items receiving a frequency rating of “never” and an importance rating of “important or critical” were identified as acquisition deficits; items receiving a frequency rating of “seldom” and an importance rating of “important or critical” were identified as performance deficits; items receiving a frequency rating of “almost always” and an importance rating of “important or critical” were identified as social skill strengths. Those items not meeting the above criteria were labeled as neither a strength nor a deficit and were not assessed in the current study.

After all items were categorized, each participant received a score based on the percentage of items meeting each SSDSP domain; these scores were then averaged into group percentages for within and between-groups analyses. Higher percentages in the acquisition and performance deficit domains indicated more difficulty in these categories, while higher percentages in the strengths domain indicated more competent social functioning.

Previous work has established precedent for this alternative scoring method and the resulting SSDSP has shown reliability and validity in pre-school and school-aged children, as well as an ADHD clinical sample (Aduen et al., 2018; Frey et al., 2011; Gresham et al., 2010). However, this has not yet been established in a clinical sample consisting of autistic youth, thus we investigated the psychometric properties of the SSDSP.

Data Analytic Plan

As this is the first time the SSDSP is being used to explore social differences in autistic populations, these analyses can be considered a relatively exploratory effort, despite operating within a formal hypothesis-testing (i.e., confirmatory) analytic framework. First, we investigated the internal consistency of the SSDSP (hypothesis 1a) using Cronbach’s Alpha (α). An α >.70 is considered to have good reliability (Kline, 1999). Second, to further test hypothesis 1a, we examined convergent validity by considering the correlation of the SSDSP profiles (p <.05) with theoretically-linked variables (FSIQ, ADOS-2 CSS, CASI-5 Autism Scores, CASI-5 ADHD Scores, and BASC Social Skills), and divergent validity by demonstrating null relations with theoretically-unrelated variables (child age, child biological sex, education, marital status, income, and race). Next, to test hypothesis 2, that autistic youth would have significantly higher average percentages of acquisition and performance deficits and significantly fewer average percentages of social skills strengths, a repeated-measures analysis of variance (RM-ANOVA) was conducted to assess if mean differences exist between diagnostic group (autistic vs. non-autistic) and SSDSP scores. Following the significant F-tests, pairwise comparisons were used to assess where the differences lie. In order to gauge the magnitude of effects, effect size statistics (Cohen’s d) were calculated: 0.2 is considered to be a small effect, 0.5 is considered a medium effect, and above 0.8 is considered a large effect (Cohen, 1992). Further post-hoc analyses of covariance (ANCOVAs) were conducted to assess whether obtained effects maintained after controlling for plausible confounds. Finally, to examine the average autism-specific SSDSP profile, paired-samples t-tests examined within-autistic and non-autistic group distributions (hypothesis 3).

Results

Internal Consistency

After the transformation of the SSIS items into the three SSDSP domains (i.e., acquisition, performance, strengths), which demonstrated good internal consistency for acquisition deficits (α =.90), performance deficits (α =.88), and social skill strengths (α =.94). For the autistic group alone, SSDSP domains demonstrated good internal consistency for acquisition deficits (α =.89), performance deficits (α =.83), and social skill strengths (α =.92) domains. For the non-autistic group, SSDSP domains demonstrated good internal consistency for acquisition deficits (α =.90), performance deficits (α =.93) and social skill strengths (α =.96) domains.

Convergent Validity

Acquisition Deficits.

For the entire sample, acquisition deficits correlated with all theoretically linked variables as expected, with the exception of FSIQ, which was uncorrelated with acquisition deficits (see Table 2). For the autistic group alone, all theoretically linked variables correlated as expected with acquisition deficits, with the exception of FSIQ and ADOS CSS scores, which were both uncorrelated with acquisition deficits in the autistic group. For the non-autistic group, all theoretically linked variables correlated as expected, with the exception of ADOS CSS scores, which were uncorrelated with acquisition deficits in the non-autistic group.

Table 2.

Correlation Matrix Showing Relationships Between SSDSP Subscales (Acquisition Deficits, Performance Deficits, and Strengths) and Theoretically Linked Variables (Convergent Validity)

	Acquisition Deficits	Performance Deficits	Strength	FSIQ	ADOS-2a CSS	CASI-5a Autism	CASI-5b ADHD	BASC-2c Social Skills
Acquisition Total	-	.32**	−.36**	−.09	.21**	.46**	.36**	−.52**
Autistic		.24 **	−.34 **	.00	.06	.42 **	.32 **	−.53 **
Non-Autistic		.45 **	−.36 **	−.29 **	.16	.58 *	.40 **	−.42 **
Performance Total		-	−.73**	−.19**	.31**	.49**	.43**	−.63**
Autistic			−.68 **	−.07	.16	.13	.29 **	−.55 **
Non-Autistic			−.76 **	−.34 **	.26 *	.80 **	.61 **	−.67 **
Strength Total			-	.19**	−.36**	−.43**	−.36**	.73**
Autistic				.04	−.18 **	−.10	−.19 *	.67 **
Non-Autistic				.34 **	−.38 **	−.51	−.52 **	.77 **

Note. FSIQ = Full Scale Intelligence Quotient. ADOS-2 = Autism Diagnostic Observation Scale, Second Edition. CASI-5= Child & Adolescent Symptom Inventory, Fifth Edition. BASC-2 = Behavior Assessment System for Children, Third Edition.

^*p < .05

^**p ≤ .01

^***p ≤ .001.

^aTotal N = 210; Autistic N = 150; Non-Autistic N=60.

^bTotal N = 59; Autistic N = 45; Non-Autistic N = 14.

^cTotal N = 187; Autistic N= 134; Non-Autistic N = 53.

Performance Deficits.

For the entire sample, performance deficits were correlated with all theoretically linked variables. For the autistic group alone, all theoretically linked variables correlated as expected with performance deficits, except for FSIQ, ADOS CSS, and CASI-5 Autism scores, which were all uncorrelated with performance deficits in the autistic group. For the non-autistic group alone, all theoretically linked variables correlated with performance deficits.

Social Skills Strengths.

For the entire sample, strengths were correlated with all theoretically linked variables. For the autistic group alone, all theoretically linked variables were correlated with strengths except for FSIQ and CASI-5 Autism scores, which were uncorrelated with social skill strengths in the autistic group. For the non- autistic group, all theoretically linked variables correlated as expected, with the exception of CASI-5 Autism scores, which were all uncorrelated with social skill strengths in the non-autistic group.

Divergent Validity

Acquisition deficits.

For the entire sample, all theoretically unlinked variables were not correlated with acquisition deficits (see Table 3). For the autistic group alone, all theoretically unlinked variables were not correlated with acquisition deficits, except for parent income, which exhibited a positive relation with acquisition deficits. For the non-autistic group, all theoretically unlinked variables were not correlated with acquisition deficits, except for child age, which was positively correlated with acquisition deficits in the non-autistic group.

Table 3.

Correlation Matrix Showing Relationships Between SSDSP Subscales (Acquisition Deficits, Performance Deficits, and Strengths) and Theoretically Unlinked Variables (Divergent Validity)

		Child Age	Child Sexa	Parent Educationa	Parent Marital Statusa	Parent Incomea	Race – Whitea	Race- Non-Whitea
Acquisition	Total	−.06	.01	−.03	−.03	.13	.05	−.10
	Autistic	−.12	−.09	−.06	−.04	.16 *	.11	−.13
	Non-Autistic	.26 *	.19	.09	.10	−.05	−.15	.02
Performance	Total	−.08	.01	.04	.02	.06	.05	−.02
	Autistic	−.13	−.07	.12	.01	−.03	.07	−.04
	Non-Autistic	.04	.03	−.12	.16	.24	.04	.03
Strength	Total	.09	−.08	−.01	−.03	−.07	−.06	.05
	Autistic	.15	.02	−.14	−.01	−.03	−.10	.06
	Non-Autistic	−.04	−.10	.19	−.20	−.13	−.04	.01

Note.

^avalues in these columns represent point biserial correlations.

^*p < .05

^**p ≤ .01

^***p ≤ .001

Performance Deficits.

For the entire sample, all theoretically unlinked variables were not correlated with performance deficits. For the autistic group alone, all theoretically unlinked variables were not correlated with performance deficits. For the non-autistic group, all theoretically unlinked variables were not correlated with performance deficits.

Social Skill Strengths.

For the entire sample, all theoretically unlinked variables were not correlated with social skill strengths. For the autistic group alone, all theoretically unlinked variables were not correlated with social skill strengths in the autistic group. For the non-autistic group, all theoretically unlinked variables were not correlated with social skills strengths.

Between-Group Differences

There was a main effect of group on SSDSP domain scores (F (2, 208) = 11.22, p <.001), such that diagnostic group had an effect on average percentage of SSDSP scores. Further probing investigated between-group differences within each SSDSP domain.

Acquisition Deficits.

Autistic youth obtained significantly more — twice as many — social acquisition deficits than non-autistic youth, yielding a small effect size difference between them (Table 1; Figure 1). This effect was maintained after controlling for age, IQ, and sex (p <.01).

Figure 1. Mean Percentages of Social Skills Deficits and Strengths.

Note. Bars represent 95% Confidence Intervals. Cross-domain between-group analyses revealed that autistic youth obtained larger average percentages of performance deficits than the non- autistic youth obtained average social skills strengths (t(209) = 2.03, p <.05).

Performance deficits.

Autistic youth received significantly higher ratings of social performance deficits than non-autistic youth, yielding a medium effect size difference between them. This effect was maintained after controlling for age, IQ, and sex (p < .001).

Social skills strengths.

Autistic youth obtained significantly fewer ratings of social skill strengths than non-autistic participants, yielding a small effect size difference between them. This effect was maintained after controlling for age, IQ, and sex (p < .001).

The magnitude of the differences between autistic and non-autistic youth across each of the SSDSP domains did not differ (i.e., the difference between the size of the group differences was not significant; all p > .05).

Within-Autistic SSDSP Differences

After analyzing between-group differences, the within-autistic SSDSP was analyzed. Performance deficits were the most prevalent aspect of autistic social functioning according to the SSDSP. Autistic youth exhibited performance deficits that were three times as prevalent as acquisition deficits (t(149) = −17.80, p <.001) and twice as prevalent as social skill strengths (t(149) = 7.92; p <.001). Additionally, acquisition deficits were the least prevalent feature of the autistic SSDSP, with more prevalent performance deficits and social skill strengths (t(149) = −3.44, p <.001).

Discussion

This was the first study to empirically derive SSDSPs to characterize the relative distribution of social acquisition deficits, social performance deficits, and social skill strengths in social functioning in autistic youth. The SSDSP yielded strong psychometric properties (internal consistency, convergent and divergent validity) in autistic and non-autistic youth, supporting its utility for measuring these constructs. Autistic youth exhibited more social skills deficits and fewer strengths than non-autistic youth; while social acquisition deficits were considerably more common than in non-autistic youth, they still represented a minority of the SSDSP. Within the autistic group, performance deficits were by far the most prevalent aspect of the SSDSP, pointing towards these deficits as the most salient feature for characterizing social functioning in autistic youth.

Psychometric Properties of the SSDSP

We first sought to examine its psychometric properties for examining social functioning in autistic and non-autistic youth, specifically looking at internal consistency and convergent and divergent validity. This study supports the internal consistency of the SSDSP across populations. Additionally, this study provides good evidence for its relation to convergent, theoretically linked variables as well as divergent, theoretically unlinked variables. Overall, results support the validity of the SSDSP in the present sample.

Convergent Validity

SSDSP domains correlated as expected with most theoretically linked variables. BASC-2 social skills correlated as expected with SSDSP domains. These results are consistent with similar analyses conducted by Aduen and colleagues (2018) when investigating the SSDSP in ADHD and non-ADHD youth.

FSIQ correlated negatively with performance deficits and positively with social skills strengths across groups. Though FSIQ correlated with all SSDSP domains in the non- autistic group alone, it did not correlate with any SSDSP domain in the autistic group alone. Positive relationships between social skills and FSIQ are often reported (Eaves & Ho, 2008). These findings add nuance to this literature, suggesting that such associations may be driven by positive associations between IQ and strengths in non- autistic youth (i.e., intelligence may be confounded with social functioning within the bounds of normative variation in social functioning), yet these same associations may not be true for autistic youth (i.e., within a population evincing broader range of social functioning).

Autism symptoms, regardless of informant, correlated with all three SSDSP domains across the total sample. In the autistic group, however, acquisition deficits correlated only with parent-report (CASI-5) autism symptoms, and social skill strengths correlated only with clinician-report (ADOS-2 CSS), while performance deficits correlated with neither. This suggests that parents of youth with more autism symptoms may tend to overrepresent their children’s social problems as a lack of acquisition (i.e., they never and or do not know how to enact a given behavior) rather than difficulties enacting known behaviors in different settings (White et al., 2007). Conversely, it may be that the standardized setting of the ADOS-2, which minimizes confounding (e.g., sensory) input, may provide a distilled opportunity for autistic youth to reveal their strengths most clearly. This highlights the utility of the SSDSP to obtain a more detailed picture of which aspects of autism symptoms relate to which elements of social functioning.

Divergent Validity

With regard to divergent validity, no relationships were found between SSDSP domains and most theoretically unlinked demographic factors (i.e., age, sex assigned at birth, education, or marital status). However, parent income did correlate with acquisition deficits in the autistic group, suggesting that higher income families are more likely to say that their children with autism are not engaging in important social skills because they don’t know what to do – perhaps reflecting a more instrumental view of social difficulties.

Overall, the SSDSP appears to be a reliable tool in examining social functioning in autistic and non- autistic youth and can lead to interesting insights on how the makeup of the SSDSP may vary across groups in relation to other psychometric assessments and to other factors.

Autism vs. Non-Autism SSDSP

This study next sought to directly examine the distribution of the SSDSP between autistic and non- autistic groups. These findings demonstrate that parents of autistic youth report that their children exhibit more acquisition deficits, more performance deficits, and fewer social skill strengths than non- autistic youth – consistent with the pathognomonic social challenges of autism (Hobson, 2014). However, these findings add nuance to this framework. Acquisition deficits were more prevalent in the autistic group than non- autistic group, supporting work showing that autistic youth may have more difficulty acquiring and retaining social skills than non-autistic youth (Bishop-Fitzpatrick et al., 2017; Mesibov, 1984). However, results also revealed a much higher rate of performance deficits in the autistic group compared to the non-autistic group, indicating that autistic youth may possess knowledge about many functional social skills that are not being observably implemented in social situations, over and above rates that are common for non- autistic youth. This lends support to theories that suggest that social skills deficits in autistic youth can more frequently be attributed to performance-related factors, like reduced social information processing speed (Mendelson et al., 2016).

Notably, both groups demonstrated higher rates of performance deficits as compared to acquisition deficits. Similar patterns have emerged in past research, in which performance deficits outpaced acquisition deficits in clinical and non-clinical samples (Aduen et al., 2018; Gresham et al., 2010). This suggests that the pattern of elevated performance deficits in autism may simply represent an exaggeration of a normative SSDSP pattern, rather than one that is unique to autism.

Social skill strengths were also less prevalent in autistic youth. This finding offers a more detailed understanding of the differences between autistic and non- autistic youth social functioning overall. Past research has typically focused on gross findings that autistic youth have more deficits than non- autistic youth (Kapp, 2019), rather than examining differences in consistently-performed, functionally important social behaviors. This highlights the capability for the SSDSP to respond to the call of the autistic community to have research on the social lives of people with autistic that explicitly examine strengths (Bottema-Beutel et al., 2021).

The SSDSP in Autism

Within the autistic sample, consistent with our hypotheses, we found that autistic youth exhibit much higher performance deficits than either acquisition deficits or social skill strengths. Specifically, performance deficits were twice as prevalent as social skills strengths, and three times as prevalent as acquisition deficits; this indicates a large gap between those social skills that autistic youth are aware of (i.e., can perform) and those that are actually, usually performed.

These findings suggest that acquisition deficits are the least prevalent feature of autistic social deficits, while performance deficits, which require differentiated treatment approaches to ameliorate, are primarily driving social skill difficulties. Again, these findings may not conform with longstanding models of social competence, which suggest that social acquisition deficits are more influential in social functioning difficulties (Bishop-Fitzpatrick et al., 2017), but support emerging work that specifically identify performance deficits as impairing as well (Keifer et al., 2020; Lerner & Mikami, 2012; Mendelson et al., 2016; Wolstencroft et al., 2018).

While acquisition deficits were almost twice as prevalent in the autistic group compared to the non- autistic group, they still remained a small minority of the SSDSP within the autistic group and were much less prevalent than performance deficits. Notably, they were also less prevalent than social strengths – indicating that it may in fact be more salient and impactful to map and reinforce the strengths of autistic individuals than to consider their acquisition deficits.

It is notable, then, that acquisition deficits are frequently discussed when considering the profiles of autistic youth (Bishop-Fitzpatrick et al., 2017). This may be because they are relatively more frequent compared to autistic individuals, and so therefore more notable. However, in this study they still remained absolutely infrequent within the autistic population. These results highlight the need to carefully consider the SSDSP of autistic youth, and specifically, the differences between acquisition and performance deficits. Utilization of a tool such as the SSDSP offers the ability to make this distinction and avoids the potential oversimplification that autistic youth simply struggle socially.

Theoretical & Clinical Implications

The SSDSP provides concrete support for core models of social competence that point to the critical nature of both acquisition and performance deficits in social functioning overall (Beauchamp & Anderson, 2010; Crick & Dodge, 1994; Mendelson et al., 2016) and the SSDSP provides direct, quantitative investigation into social skills deficits in autistic youth.

These findings also add to the growing literature supporting performance deficits as the most important feature of autism-specific social skills difficulties and the need to consider these when making clinical decisions. These results are consistent with emerging evidence that suggest that performance deficits are more impairing for social functioning difficulties for autistic youth (Lerner & Mikami, 2012; Mendelson et al., 2016; White et al., 2007). Specifically, this highlights that measures and interventions focusing on primarily on assessing and targeting social skill acquisition deficits are incorrect and insufficient in both characterizing social challenges as well adequately providing the means for social skill improvement.

As a clinically useful tool, the SSDSP shows promise of providing precise social skills assessment that is easy to use. Having an alternative scoring system, applicable to a commonly used measure of social skills, with the ability to produce detailed information on social skills difficulties has the potential to allow clinicians and researchers to tailor intervention to best target areas of need. Frey and colleagues (2011) began to investigate the use of the SSIS-RS to target and guide social skills intervention for pre-school children by identifying the skills that require differentiated intervention strategies based on being identified as an acquisition deficit or a performance deficit. However, a framework such as this has yet to be implemented using the SSDSP in school-aged autistic youth. This would be an important and clinically useful step for research to explore.

Beyond the alternative scoring method discussed in this paper, the development of more accurate measures and assessment tools to evaluate social performance difficulties should be explored in future research. Specifically, methods of assessment could leverage observational coding techniques to examine the relation between professed knowledge and observed behavior in a variety of ecologically valid settings. Likewise, the use of machine learning/computer vision technologies could be used to objectively capture a wider range of behaviors and setting in vivo to better model these differences. In addition to more precise measures of social performance, the construct overlap between social knowledge (knowing what to do socially) and social acquisition (ever engaging in a given social behavior) also bears further examination: while they are similar (and one may depend on the other), they are not interchangeable. Future research would benefit, then, from also leveraging this tool along with direct examinations of social knowledge (i.e., asking youth what they know) and acquisition (i.e., using social press procedures to examine if a child ever exhibits a given behavior) to better disentangle them in the study of social profiles in autism.

Results of this study have the potential of guiding autism-specific clinical research as well. Many interventions that have been developed for autistic youth specifically target acquisition deficits and fail to provide naturally occurring situations in which the performance of a behavior can be modeled, practiced, and reinforced (Gates et al., 2017; Ke et al., 2017; Wolstencroft et al., 2018). With the SSDSP indicating that most autistic -group social deficits are attributable to difficulties performing already known behaviors, the current amalgamation of strategies employed in social skills interventions may not efficiently target areas of need. For example, if an individual has mostly performance-based deficits, requiring opportunities to perform social skills and be reinforced for doing so, intervention curricula that predominantly target acquisition deficits (i.e., teaching social rules) may be ineffective, or even counterproductive (Bottema-Beutel et al., 2018). This underscores the importance of considering the impact of performance deficits and the methods used to develop them. It is a necessity for researchers attempting to craft interventions that best improve social skill deficits to carefully consider the impact of performance deficits and the methods that best improve these skills. The use of the SSDSP for evaluating intervention efficacy, and for guiding treatment, would offer an opportunity to provide more successful treatment options, and to our knowledge, no study has attempted to evaluate treatment efficacy pre- and post-intervention using this method. Future research should aim to investigate the use of this tool as an evaluative measure.

Additionally, it is important to note that parent report measures in autism intervention research are also susceptible to detection bias (e.g., biases in reported skills based on parent-awareness of diagnostic or treatment group; Sandbank et al., 2020), which may limit interpretability of the results. Future research utilizing parent-reported SSDSPs (or similar tools) to investigate intervention effectiveness should aim to utilize designs that mitigate such bias (e.g., informants blinded to treatment condition via active control designs or online self-paced interventions; Kang et al., 2019; Libsack et al., 2022), or examine the SSDSP as reported by more distal reporters (e.g., teachers).

Relatedly, although these results suggest a need for increased attention to performance deficits, our between-group results still point to significant differences in acquisition deficits, with autistic youth having twice as many acquisition deficits than non- autistic youth. Together, these underscore that although performance deficits require increased attention, acquisition deficits in autistic youth should not be ignored either. Recent work investigating the impact of knowledge-based interventions (i.e., targeting acquisition deficits) and performance-based interventions (i.e., targeting performance deficits) suggest that, though knowledge-based interventions are rarely sufficient to address performance deficits, performance-based interventions can improve performance skills while also improving acquisition deficits without didactic teaching methods (Marro et al., 2019). Additionally, recent research evaluating social skills interventions also underscore an important perspective: autistic youth report preferring performance-based activities (Bottema-Beutel et al., 2016). Future research should continue to explore intervention approaches that match the needs and preferences of autistic youth.

The SSDSP also shows promise for use in developing individualized, targeted treatment plans. The alternative scoring method offers the opportunity to evaluate precisely which functionally important skills are acquisition deficits, which are performance deficits, and which are strengths. By evaluating the SSDSP skills, clinicians could assess which skills appear to need a more performance-based approach and which may need more knowledge-based training.

Limitations

Several limitations should be considered. First, this study only included parent-report of social skills difficulties. Although parent-report provides important information about social skills difficulties, both teacher- and self-report could provide important added information about differences in social presentation across settings (i.e., generalizability) and with different observers, as well as important understanding of the self-perception of social skills from the perspective of autistic youth. Ratings from specific informants have the potential to carry several biases, including informant attributions (e.g., different understandings of causes of difficulties), perspectives (e.g., information based on informant’s unique relationship with the participant), and motivation (e.g., rating a child’s behavior based on what the informant wants the outcome to be, such as receipt of treatment; De Los Reyes & Kazdin, 2005). That said, use of this methodology across similar contexts (and pattern of associations found here) suggests that these results are unlikely to be uniformly attributable to parent report bias. Further, parents may also be influenced by central tendency bias - they may be more likely to rate their child’s skills towards the middle of the scale as opposed to the lower or upper extremes (i.e., never or always; Saal et al., 1980). That said, results indicate that parents in this sample did utilize the full range of the scale, as evidenced by the high rate of both acquisition deficits and strengths displayed (which require the use of upper [“Never”] and lower [“Almost Always”] extremes to be categorized as such). Regardless, the use of multiple informants to evaluate including social skills in the assessment of youth is both ideal to help control for above-mentioned biases and necessary to better understand the potential clinical application of the SSDSP. Future research should aim to explore autistic SSDSPs with varying informants.

Further, our sample had limited racial/ethnic diversity. In order to better serve underrepresented populations, future research must assess a more representative sample. The systemic exclusion of minoritized groups in autism research impacts diagnosis (Mandell et al., 2007), service access and delivery (Straiton & Sridhar, 2021), and perpetuates health disparities (Bishop-Fitzpatrick & Kind, 2017). Further, future research should aim to better understand social functioning difficulties and the use of this tool in identifying the distribution of deficits and strengths in youth identifying as female, transgender, and gender non-binary.

As have been mentioned in past studies using this method (Aduen et al., 2018; Gresham et al., 2010), there are some limitations related to the SSDSP alternative scoring method that should be addressed. As mentioned, the scoring method relies on the informant’s ratings of frequency of behavior and the subjective importance of that skills. This assumes that one can judge acquisition of a skill based on how often it is observably performed. However, it is possible that a child may know how to do a skill she has never performed. Additionally, relying on a single rater’s perspective inexorably introduces concerns of rater bias (De Los Reyes et al., 2015). A parent of a child with a limited social repertoire may place higher importance on skills that are performed more frequently, or on skills their child does not value; they may also place lower importance on skills that are more difficult for the child. Though there is typically agreement that social skills are important, there are key informant discrepancies on which skills are the most important, and on how important they are (Rankin et al., 2016). However, this method has shown validity in measuring social functioning in several populations, including in the current study. Comparison of this tool to measures that directly measure social acquisition and performance may offer more insight into its incremental utility.

Lastly, in research examining autistic social differences, one must consider whether it is in fact social skills difficulties that result in challenges, or social stigma and resulting pressures to conform to non-autistic social norms. As an example, preliminary evidence suggests a positive relation between individuals conforming to non-autistic social norms (i.e., passing as non-autistic) and internalizing problems such as anxiety and depression (Libsack et al., 2021). Future research should further investigate methods of assessment, and resulting social supports, that focus on improving outcomes to better tailor to the needs of autistic individuals.

Conclusion

The current study was the first to characterize the parent-reported SSDSP within and among autistic and non- autistic youth. This work provides support for the growing literature highlighting the prominence of social performance deficits in understanding social profiles of autistic youth. Specifically, results support that social performance deficits contribute overwhelmingly to social skills problems in autistic youth; indeed, they were three times as prevalent as acquisition deficits, which had long been assumed to be the primary driver of social difficulties. This study indicated that the SSDSP may be a valid and promising tool for obtaining detailed information on social functioning in autism, and for guiding targeted and individualized interventions. Overall, this study highlights the importance of taking a more nuanced, empirically-grounded perspective to examining social strengths and challenges in autism.