Social skills characteristics of students with autism spectrum disorder

Abstract

Objectives: This study examines differences in social skills among children with autism spectrum disorder (ASD). In order to investigate these differences, social skills were associated with variables like gender, age, intellectual disability, language development, and school type.

Methods: For the purposes of the study a total of 63 students with ASD attending primary and secondary special education units were selected in Northern Greece. A structured questionnaire was filled in by their teachers.

Results: The results showed major differences between children with ASD and intellectual disability and those without. Likewise, verbal children obtained higher scores than non-verbal. These higher scores indicate better social skills. Age, gender, and school type differentiated the scores of the groups only in a few factors of the questionnaire.

Conclusion: Intellectual disability and language are variables that clearly influence the socialization of children with ASD.

Introduction

One of the core symptoms of autism spectrum disorder (ASD) is the deficit in social functioning (American Psychiatric Association 2013). This deficit includes difficulties in initiating or joining social activities, difficulties in understanding others’ viewpoints, engaging in inappropriate behaviors, lack of eye contact, distance from people, non-functional use of language, and a lack of communicative gestures and others. (Kanner 1973; Coggins et al. 1983; Landa et al. 2007; Ghaziuddin 2008; O’Connor and Kirk 2008; White and Roberson-Nay 2009; Bölte et al. 2011; Callahan et al. 2011; Gillis et al. 2011; Orsmond et al. 2013; Barbaro and Dissanayake 2013).

Research suggests that, despite the fact children with ASD have some common characteristics, the population with ASD presents tremendous heterogeneity in its adaptive behavior (Burack and Volkmar 1992). A review of the literature suggests that several variables exert a differential influence on this population’s social behavior, including gender, age, level of intellectual functioning, and language skills.

One variable potentially affecting the social skills of a child with ASD is its gender. For example, Dworzynski et al. (2012) used the Childhood Autism Spectrum Test in order to compare males and females of 10–12 years old. With intelligence effects partialled out, the females displayed fewer social and communicative deficits than the males. Similar results were reported by Lai et al. (2012) and Head et al. (2014). More specifically, in the latter study, the female adolescents scored higher than the male adolescents in a ‘friendship questionnaire’ (Head et al. 2014). Moreover, the females with ASD demonstrated similar scores to those of typically developing males. Higher friendship questionnaire scores indicate higher levels of sociability, emotionality, and friendship, which denote fewer deficits in adaptive behavior. Furthermore, in the study by Hiller et al. (2014), using the broad social criteria presented in the DSM-IV-TR, the female adolescents seemed to demonstrate more social behaviors than the boys (e.g. in maintaining a reciprocal conversation, being able to initiate, etc.). In contrast, Baron-Cohen and Wheelwright (2004), Carter et al. (2007) and Harrop et al. (2014) did not find any differences between males and females with ASD in aspects of socializing. Males and females with comparable childhood autistic symptoms showed similar cognitive deficits in social domains (Kopp and Gillberg 1992). Also, Lai et al. (2012) showed that regardless of sex, adults with ASD presented impaired facial emotion recognition abilities, which are necessary for socializing successfully with other people. On the other hand, Frazier et al. (2013) and Dean et al. (2014) suggested that there is a specific female ASD phenotype which emerges with more severe symptoms, including greater social communication impairment than males and lower social acceptance and prominence.

In reference to the age effects on the socialization of children with ASD, the research results also seem to be contradictory. A number of studies indicate improvement in social adjustment as individuals with ASD grow older (Rutter et al. 1970; DeMeyer et al. 1973; Bartak and Rutter 1973; Ando and Yoshimura 1979). DeMeyer et al. (1973) found that the social and conversational skills of mildly impaired children with ASD improved with age. Similar results regarding age were obtained by Bartak and Rutter (1973) in reference to social responsiveness, including eye contact, play, and facial expression. Freeman et al. (1999) found that Social Skills scores on Vineland Adaptive Behavior Scales changed as an age function. In a comparison with three different age groups using the Vineland Socialization Scale, Fenton et al. (2003) found that the group including the oldest children presented significant improvement. On the other hand, Garfin et al. (1988) did not find any significant differences between children and adolescents with ASD in most items in the Childhood Autism Rating Scale, including relationships with people. Similar results were found in the study by Klin et al. (2007), where there was no association between social skills and age, suggesting stable levels throughout chronological age. In a longitudinal study involving children with ASD, Vineland Socialization scores were found to decrease over time, meaning that socialization deteriorates with age (Szatmari et al. 2003). Meanwhile, according to Stone and Caro-Martinez (1990), there is a difference between interactions of children with ASD with peers and their interactions with adults, in the sense that after the age of five, children with ASD tend to improve their interactions with adults but their relationships with peers remain unsuccessful.

There are a variety of studies exploring the association between social skills and cognitive function in children with ASD, with or without intellectual disability. Wing (1981a) and Kenworthy et al. (2010) found a high correlation between the severity of intellectual disability and social interaction. The main conclusions were that, the more severe the brain damage, the higher was the risk that the parts of the brain responsible for social interaction and social skills might be affected. This is explained by the fact that the inadequate cognitive functioning imposes constraints on the adaptive functioning. Obviously, an individual is not expected to function at a social level that is significantly more advanced than their intellectual level. So, greater cognitive difficulties are associated with greater social responsiveness difficulties and poorer social skills (McMahon and Henderson 2014; Ratcliffe et al. 2015). Nevertheless, these have not been consistently confirmed by other studies. For instance, Carter et al. (1996) claims that even high-functioning individuals with ASD show significant deficits in adaptive behaviors. Freeman et al. (1999) concluded that social skills are independent from initial IQ scores. There is evidence that social deficits in high-functioning children with ASD are more extensive than would be predicted on the basis of IQ alone. For example, Liss et al. (2001) found that children with high-functioning autism are more impaired in the area of socialization than IQ matched controls. Numerous studies (Rutter 1978; Wing and Gould 1979; Cohen et al. 1986; Fein et al. 1986; Frith 2003; Lord and Ward 1993; Happé 1994; Sigman and Kasari 1995) converge to the conclusion that social deficits of children with ASD are clearly more serious than would be expected on the basis of cognitive functioning.

Language seems to be another important predictor of outcomes in the domain of socialization of children with ASD (Szatmari et al. 2003). Klin et al. (2007) reported that communication was strongly related to Verbal IQ (VIQ). Bölte and Poustka (2002) indicated similar findings in a sample of high-functioning children with autism. According to Black et al. (2009), stronger VIQ is associated with fewer social deficits. Kobayashi et al. (1992) and Venter et al. (1992) support the view that expressive language level is probably the strongest predictor of outcome in the socialization of children with ASD, at least in individuals beyond the preschool level, as it affects almost every aspect of social interaction (Happé 1995; Mahoney et al. 1998). Perry and Factor (1989), in their study of the psychometric properties of the Vineland Adaptive Behavior Scales and the AAMD Adaptive Behavior Scale for a sample with ASD found that there is a high degree of association between language/communication and socialization areas. Likewise, Strid et al. (2012), in a study on social communication in speaking and non-speaking children with ASD, found that the non-speaking children had reduced performance on all social communicative measures. Nevertheless, in the study by Carter et al. (1998), using the Vineland Adaptive Behavior Scales, non-verbal and verbal individuals with ASD did not differ in socialization. In the same study, it was proved that sometimes children with ASD and high VIQ with large vocabularies and fluent speech seem to be more competent in interactions than they actually are.

Concerning the school placement policy for ASD in Greece and its relationship to social deficits, not enough previous research or results exist to suggest any kind of link between the two. Informally, children with ASD and major difficulties (such as the existence of intellectual disability and the non-existence of language) are placed in special schools or in inclusive classes. The rest of the children with ASD attend typical classes with a supporting teacher. Accordingly, under this policy we would expect that children in typical classes would exhibit less difficulties in socialization. This may be partially true, but what are the possibilities that findings of this kind might be due to the greater chances they have of interacting with other peers?

Despite the numerous studies that exist which investigate the relationship between different variables and social skills of children with ASD, the empirical findings are seen to be inconsistent and often contradictory. This outcome can be related to considerable variations in the methods of analysis, sample size, variable selection, data quality, and diversified characteristics of control groups. The purpose of the present study is to theoretically study, empirically investigate, and critically evaluate the key points in the social skills profile of children with ASD, in order to help shed light on the seemingly complex relationship between the social skills of these children and variables like gender, age, language ability, and level of cognitive functioning. Other relationships that are also tested are those between the social skills of children with ASD and the type of school they attend. The present paper intends to enhance the earlier research and expand on the previous empirical findings, contributing further useful insight into the heterogeneity of the social skills of children with ASD and then into the understanding of their difficulties in order to create appropriate interventions. To the authors’ knowledge, this is one of the first few empirical studies, both within Greece and outside it, to focus on the relationships between critical variables and the social skills of children with ASD as most of the previous studies that are cited focused only on some of these variables. In this study, an attempt to examine all these factors was made in order to indentify the ones with the major predictive value in the social skills of children with ASD.

Moreover, the study attempts to provide a range of quantified outcomes on these critical issues, especially in the context of Greece.

The aims of this paper are as follows:

• (i)To assess whether there are any relationships between variables like the gender, age, existence or not of intellectual disability, different language level and school type of children with ASD and the different scores they obtain in specific social skills.
• (ii)To identify the variables which differentiate the social abilities of the children with ASD in the most factors of the questionnaire and the specific areas they confront the most social difficulties, in order to create the appropriate intervention programs for the specific domains of socialization in the future.

Methods

Research questions

The research questions were as follows:

Are there differences between the answers of the sample for the created categories (variables) concerning gender, age group, place of abode, intellectual disability, language, and school type?

Which of these variables influence the most factors? Which specific factors are influenced by each variable?

Participants

Identity of the children with ASD

The demographic characteristics of the core students included gender and age, place of abode (i.e. residence in an urban or rural location), school type, and the existence of a diagnosis for intellectual disability. Also, student’s language ability was included, which was tested on a 3-point scale: non-verbal, use of extremely confined language (only single words), and verbal.

The study cohort comprised 63 children with ASD between the ages of 6 and 17 years [19 females (30.2%) and 44 males (69.8%)]. The sample size is representative for the Northern Greece’s schools and totally valid. All participants were required to have received a diagnosis of ASD from a certified public diagnostic center for children with special needs and learning difficulties. The diagnostic characterization of the sample is based on a multidisciplinary and differential evaluation and assessment. In the multidisciplinary evaluation different tools were used by the multidisciplinary team to determine the diagnosis of ASD for these children.

Twenty-four (38.1%) of the participants were between 6 and 11 years old and 39 of them (61.9%) 12–17 years old (Table 1).

Table 1. Demographic characteristics of children with ASD.

Characteristic	n	%
Gender
Male	44	69.8
Female	19	30.2
Age
6–11 years old	24	38.1
12–17 years old	39	61.9
Area
Urban	52	82.5
Rural	11	17.5
School type
Special school	36	57.1
Typical class with supporting teacher	25	39.7
Inclusion class	2	3.2
Diagnosis for intellectual disability from certified public diagnostic centers
Yes	37	58.7
No	26	41.3
Language level
Verbal	41	65.1
Non-verbal	20	31.7
One Word	2	3.2

Most of the children in the sample – 52 (82.5%) of them, to be specific – lived in urban areas, while 11 of them (17.5%) in rural areas. As regards school type, 36 children (57.1%) attended a special school, 25 (39.7%) attended typical classes with a supporting teacher and, 2 (3.2%) attended an inclusion class (Table 1).

Measures and procedures

The empirical evaluation is based on a structured questionnaire which was divided into two parts. The first part included information relating to the students’ gender and age, their place of abode and the type of school they attended. It also asked whether the student had a diagnosis for intellectual disability. Children’s IQ was assessed with WISC-III at the certified centers. Finally, at the end of this part of the questionnaire, teachers were asked to indicate their students’ language ability using a 3-point scale: non-verbal, use only of single words, verbal. The original 5-point scale for language ability (Bellini and Hopf 2007) was not used owing to the results of the pilot study, which led to the conclusion that the 3-point scale was far easier and more convenient for teachers to use in order to assess children’s language ability. Teachers indicated their students’ language ability, based on their observation as ‘nonverbal’ when a student has no language at all, ‘extremely confined language’ when a student usually uses one-word language to communicate functionally and ‘verbal’ when students use more than one word to communicate. The students’ language ability was determined only on the teachers’ report.

The second part was based on the Autism Social Skills Profile (ASSP) (Bellini and Hopf 2007). This instrument includes 49 items that are rated on a 4-point Likert scale ranging from never = 0 to very often = 3, with high scores corresponding to positive social behaviors. The internal consistency of the measure is reportedly high (a = 0.926). Also, according to Bellini’s study (2007), after an examination of the factor loadings, three factors were identified which were labeled as Social Reciprocity, Social Participation/Avoidance, and Detrimental Social Behaviors (Bellini and Hopf 2007).

We recruited our participants from special education units. In order to test teachers’ comprehension and objectivity in answering this questionnaire, an initial pilot study was conducted with 20 teachers in Northern Greece. The preliminary teacher responses and feedback, as well as the structure of the questionnaire, were compared with relevant empirical input gained from relevant international studies (e.g. Ando and Yoshimura 1979; Black et al. 2009; Head et al. 2014). The questionnaire was translated into Greek and then back-translated into English and the appropriate adaptations were made in order to ensure the instrument was as accurate as the original one. The final version of the questionnaire was evaluated by the research team.

The questionnaires were answered by primary and secondary special education teachers who were teaching students with ASD in special schools, ordinary classes with supportive teaching, and inclusion classes in Northern Greece during the academic year 2013–2014. After approaching the headteachers and informing them about the aim of the study, the first of the researchers briefed the teachers and asked them whether they would like to participate in the present study. Out of the 150 teachers who were briefed, 66 consented to take part in the study. Of the 66 questionnaires collected, 3 were excluded because they contained more than three missing items, according to Bellini’s and Hopf’s (2007) Preliminary Analysis of Psychometric Properties. So, the 63 remaining questionnaires concern the assessments of one teacher to one student.

The participants were encouraged to contact the research team with any queries regarding the completion of the questionnaire.

ASSP was preferred as it is a very useful tool that can be briefly answered by parents or teachers. Moreover, through the pilot study that was conducted, it proved that ASSP is fully comprehensible with great accuracy by the teachers. The majority of other tools are used in a variety of populations and include items for many domains of skills, while ASSP includes specific items about social skills of children with ASD that are investigated in this study and provide the opportunity to discover children’s with ASD-specific abilities and disabilities in this study. Also, ASSP is a tool with great psychometric properties and a reportedly high internal consistency (Bellini and Hopf 2007).

Statistical analysis

Statistical analysis of the data was performed using the standard statistical package for the social sciences (SPSS, version 22). Also, descriptive statistics, reliability Cronbach’s test and factor analysis. Additionally, One-Way Analysis of Variance (ANOVA) was applied in order to conduct multiple comparisons, to analyze the differences among group means and to examine the influence of the variables on the factors of ASSP. Bonferroni test was also used as a post hoc test complementing One-Way ANOVA for exploring which of the groups has the greatest value for each factor.

Results

Factor analysis

For the forty-nine (49) questions of the questionnaire, factor analysis was utilized by implementing the principal components analysis and the varimax rotations. The number of the factors was determined using the criterion of eigenvalues, which should be greater than one. Also, the maximum iterations number of convergence should be equal to 40. The factors were created on the basis of the rotated component matrix and the results of the factor analysis showed that there were 10 factors. The cumulative variance of rotation sums of squared loadings was 77.76%. The factors are presented below (Table 2).

Table 2. Description of factors.

Factor	Description	Variance (%)
Social reciprocity	Skills that are necessary for maintaining a successful reciprocal social interaction and behaviors which indicate that the child takes into consideration the other person’s thoughts and feelings during interaction	20.16
Social participation	Skills that are crucial for social engagement in social interactions	15.30
Detrimental social behavior	Items related to inappropriate verbal behavior during social interactions	9.52
Asking questions	Asking questions about other people and topics, and asking for help	8.29
Recognizing social cues	Concerns the child’s ability to read the clues during interactions with others in order to react appropriately	4.69
Solitary activities	How often the child engages in solitary activities	4.51
Social anxiety/fear	The negative experiences and feelings that children experience in social interactions	4.28
Social receptivity	Items relating to a child’s permission for it to be assisted by others and to the question of whether it is manipulated by others	3.94
Social initiations	The appropriate timing that the child exhibits in social initiations	3.84
Group activities	The frequency of a child’s preference to participate in group activities in the presence of other children	3.22

The labeling of the 10 factors proved to be an arduous task as some of them were broad and seemingly diverse. The factors were: (1) Social Reciprocity, which includes the skills that are necessary for maintaining a successful reciprocal social interaction and behaviors which indicate that the child takes into consideration the other person’s thoughts and feelings during interaction; (2) Social Participation, which is related to skills that are crucial for social engagement in social interactions; (3) Detrimental Social Behavior, which includes items related to inappropriate verbal behavior during social interactions; (4) Asking Questions, that is, asking questions about other people and topics, and asking for help; (5) Recognizing Social Cues, which concerns the child’s ability to read the clues during interactions with others in order to react appropriately; (6) Solitary Activities, which are based on a single question about how often the child engages in solitary activities; (7) Social Anxiety/Fear, which concerns the negative experiences and feelings that children experience in social interactions; (8) Social Receptivity, which includes two items relating to a child’s permission for it to be assisted by others and to the question of whether it is manipulated by others; (9) Social Initiations, which refers to the appropriate timing that the child exhibits in social initiations; and finally (10) Group Activities, which are based on one item only and are related to the frequency of a child’s preference to participate in group activities in the presence of other children (Table 3).

Table 3. Descriptive statistics for the factors.

Factor	Mean	Std. Deviation
Social reciprocity	0.87	0.71
Social participation	1.16	0.72
Detrimental social behavior	1.36	0.86
Asking questions	0.90	0.84
Recognizing social cues	1.57	0.60
Solitary activities	1.30	0.99
Social anxiety/fear	2.01	0.74
Social receptivity	1.93	0.42
Social initiations	1.40	1.05
Group activities	1.41	0.94

Notes: Questionnaire’s items are rated on a 4-point Likert scale (ranging from never = 0 to very often = 3), high means correspond to positive social behavior in specific social domains.

Reliability of the instrument

Internal consistency for the whole scale and each of the sub-scales was examined using Cronbach’s alpha and the results are as follows: Whole scale: a = .952; Social Reciprocity: a = .956; Social Participation: a = .935; Detrimental Social Behaviors: a = .864; Asking Questions: a = .754; Recognizing Social Cues: a = .754; Social Anxiety: a = .706; and Social Receptivity: a = .755. Cronbach’s alpha wasn’t available for three factors (Solitary Activities, Social Initiations, and Group Activities) due to the fact that these factors consisted of one item only. Hence, the value of Cronbach’s alpha for each category was above 0.70, which is the minimum value in order for research findings to be accepted as consistent. In our case, it may be observed that the questionnaire was consistent both in whole and in part.

Social profile of children with ASD

Gender

Higher scores in specific factors indicate better social skills than other groups in these social domains. In order to find out which gender had the highest score concerning the answers based on the factors, a One-Way ANOVA test was conducted. There were no significant differences in the mean between the girls and the boys. The results of the test showed no population mean differences in the factors concerning gender. However, only the Social Anxiety/Fear factor seemed to have statistically significant population mean differences. In the post hoc analysis with Bonferroni, the girls (2.34) seemed to have better social skills (F = 6.03, p < .05), instead of the boys (1.86).

Age group

The differences between older and younger children with ASD in terms of their social skills and deficits were also examined. A One-Way ANOVA test was conducted to discover which age group (children from 6 to 11 years old with ASD compared with teenagers from 12 to 17 years old) had better skills concerning the answers based on the factors. The difference in responses for the Social Initiations (F = 13.228, p < .05) and Solitary Activities (F = 8.277, p < .05) factors was statistically significant. The results of the Bonferroni test showed that the older age group (12–17 years old) had betters social skills instead of the younger age group (6–11 years old).

Intellectual disability

The next variable tested with One-Way ANOVA was intellectual disability. Thirty-seven students with intellectual disability and 26 students without it were taken into consideration. Therefore, it was determined whether the scores for the factors were the same between these two groups. The answers for the factors Social Reciprocity (F = 11.082, p < .05), Social Participation (F = 7.758, p < .05), Detrimental Social Behavior (F = 5.240, p < .05), Asking Questions (F = 4.232, p < .05), Recognizing Social Cues (F = 6.019, p < .05), and Social Receptivity (F = 5.971, p < .05) were found not to be the same between the intellectual disability group and the group with no disability. Therefore, it was possible to explore which group had the greatest value for each factor. The Bonferroni methodology was used. It is clear that there was a statistically significant population mean difference between the students who had intellectual disability and the students without intellectual disability. The scores were higher for the students without intellectual disability instead of the students with intellectual disability. So, students without intellectual disability have better social skills than students with intellectual disability. Table 4 shows the mean of the two groups for each of the six factors for which there were statistically significant differences.

Table 4. ANOVA between children with I.D. and children with typical I.Q.

Factor	Intellectual disability	No intellectual disability	F	p
Social reciprocity	0.64	1.20	11.08	0.001
Social participation	0.96	1.45	7.76	0.007
Detrimental social behavior	1.15	1.65	5.24	0.023
Asking questions	0.72	1.15	4.23	0.044
Recognizing social cues	1.42	1.79	6.02	0.017
Social receptivity	1.82	2.07	5.97	0.017

Language level

One-Way ANOVA and post hoc criterion showed differences in the means of the three groups (verbal, non-verbal, one-word language) in the factors: Social Reciprocity (F = 17.527, p < .05), Social Participation (F = 10.876, p < .05), Detrimental Social Behavior (F = 38.748, p < .05), Asking Questions (F = 16.302, p < .05), and Group Activities (F = 6.947, p < .05). Thus, the language type with the greatest value for each factor was examined. To be specific, in the case of Social Reciprocity, the verbal type had a greater score than the other categories (verbal = 1.18, one-word language = 0.47, non-verbal = 0.27). In the case of Social Participation, the verbal type seemed to have better skills (verbal = 1.43, one-word language = 0.90, non-verbal = 0.64). As for Detrimental Social Behavior, the one-word language category had a higher score (verbal = 1.74, one-word language = 2.67, non-verbal = 0.45). In the case of the Asking Questions factor, the verbal type had a higher score than the other categories (verbal = 1.26, one-word language = 0.38, non-verbal = 0.21). Finally, in the case of Group Activities, the verbal type seemed to have better skills (verbal = 1.68, one-word language = 0.00, non-verbal = 1.00). Table 5 shows the mean of the three groups for each of the five factors for which there were statistically significant differences.

Table 5. ANOVA for verbal group, one-word language group, and non-verbal group.

Factor	Verbal group	One-word language group	Non-verbal group	F	p
Social reciprocity	1.18	0.47	0.27	17.53	0.000
Social participation	1.43	0.90	0.64	10.88	0.000
Detrimental social behavior	1.74	2.67	0.45	38.75	0.000
Asking questions	1.26	0.38	0.21	16.30	0.000
Group activities	1.68	0.00	1.00	6.95	0.002

School type

In looking for differences in the means of the three groups in respect of the school they attended (typical class with supporting teacher, special school, inclusion class), the One-Way ANOVA test showed differences in population mean only for the Detrimental Social Behavior (F = 3.167, p < .05) and Social Receptivity (F = 5.573, p < .05) factors. This means that a post hoc analysis was required only for these factors. To be specific, students at a general type of school had a higher score (1.68) than students who attended a special type of school (1.15) and children who were students in an integrated class (1.00). These findings were for Detrimental Social Behavior. As for Social Receptivity, students at a general type of school had better social skills (2.12) than students who attended a special type of school (1.82) and students in an integrated class (1.50).

Discussion

Past studies have postulated that social skills deficits belong to the core symptoms of ASD. Although the population of students with ASD is characterized by heterogeneity and different levels of social adaptation, there is no clear picture of the importance and influence that the specific variables mentioned above have in social skills of students with ASD. So a core objective of this study was to examine the influence of these specific variables on the socialization of students with ASD and then, on the basis of these results, to be able to predict the abilities and the difficulties of social skills of our future students, in order to create the most appropriate programs and interventions for them.

Therefore, according to the teachers’ ratings obtained in the context of the present study, and knowing that boys with ASD usually have more severe symptoms than girls (Wing 1981b), we attempted to determine whether gender in our sample was a variable that differentiated the level of social abilities. The results did not indicate any specific differences in socializing abilities between boys and girls. The only difference was found in the Social Anxiety/Fear factor, where the girls scored higher than the boys. Due to the fact that lower scores in this factor indicate more anxiety and fear (as we reversed the original scores for the items of this factor in order to have the same direction with the rest of the items), the boys seemed to have had more negative experiences from previous social interactions than the girls. These negative experience leads to social anxiety and fear for future interactions. This is probably why girls with ASD tend to interact more with peers and behave more suitable than boys (Wing 1981b; Kopp and Gillberg 1992).

Age is also a variable that is usually related to changes in children’s behavior, due to development and maturity. Apart from maturity and familiarity with social conditions and interactions, a further implication of our results could be that older children with ASD will have taken part in more intervention programs and will have had more opportunities to interact with others. In the present study, adolescent students scored higher in the factors of Social Initiations and Solitary Activities than the younger children. So it seems that younger children exhibit poor timing in their social initiations and engage less in solitary activities and hobbies than older children. In general, we did not find many differences between the two age groups in respect of the manner in which they socialize. So we may conclude that the social deficits in ASD are quite persistent (Garfin and McCallon 1988; Szatmari et al. 2003; Klin et al. 2007).

In contrast to age, intellectual disability seems to clearly differentiate the social skills of children with ASD. Important differences were found between students with ASD and intellectual disability and students presenting only ASD in 6 of the 10 factors. Specifically, differences were found in the following factors: Social Reciprocity, Social Participation, Detrimental Social Behavior, Asking Questions, and Recognizing Social Cues. Children diagnosed with both ASD and intellectual disability have worse skills in these factors than children with only ASD. So, as far as Social Reciprocity is concerned, it seems that children with intellectual disability exhibit greater difficulty engaging in successful reciprocal social interactions and they do not take into consideration the thoughts and feelings of others during interactions. Moreover, they exhibit difficulties in the domain of Social Participation as they do not engage in social interactions as much as children with ASD and typical IQ. Also, concerning the Detrimental Social Behavior factor, the results show that children with ASD and intellectual disability behave in a socially inappropriate manner more frequently than other children. This kind of behavior contributes to negative peer experiences and leads directly to adverse social interactions (Bellini and Hopf 2007). Furthermore, children with intellectual disability seem to ask fewer questions about people and topics and to ask for help less frequently than other children. Their scores in the Recognizing Social Cues factor suggest that they do not have the ability to recognize the facial expression of others or to understand their humor and they even misinterpret other people’s intentions, leading to inappropriate behaviors. Finally, concerning the Social Reciprocity factor, it seems that they are manipulated by others more easily compared with children with typical IQ.

These differences between the two groups (the one with intellectual disability and the other with typical IQ) were quite expected (Mayes and Calhoun 2007), as children’s cognitive deficits impose serious constraints on social skills. Besides, we have to bear in mind that many of these skills include, and are based on, cognitive processes such as perception, theory of mind, and memory (McMahon and Henderson 2014; Ratcliffe et al. 2015). Also, it should be noted that the diagnosis for intellectual disability was performed by certified centers and that the authors did not have a clear idea of the accurate functionality of each child or its specific Intelligence Quotient. So, readers should treat these results with great caution.

Language also seems to be an important variable that is related to social skills of children with ASD. Differences were found in five factors: Social Reciprocity, Social Participation, Detrimental Social Behavior, Asking Questions, and Group Activities. Non-verbal children had by far the most deficits in four of these factors (Social Reciprocity, Social Participation, Detrimental Social Behavior, and Asking Questions), in comparison with the other two groups (verbal and one-word language). This may be taken to mean that the non-existence of language affects the skills that are important for maintenance and participation in interactions or discussions with others and is related to more inappropriate behaviors. The children in the verbal group had the greatest scores in four of these factors (Social Reciprocity, Social Participation, Asking Questions, and Group Activities), while the children with one-word language had the middle scores in three of the factors (Social Reciprocity, Social Participation, and Asking Questions), the greatest one in Detrimental Social Behavior and the lowest in Group Activities.

Similar results from other studies (Wing and Gould 1982; Denckla 1986; Perry and Factor 1989) suggest that the development or not of language in children with ASD affects every aspect of their socialization (Happé 1995; Mahoney et al. 1998), and that this is probably a reason why children with no language lack social skills. However, a number of items described skills that premised the existence of some type of language (e.g. the making of inappropriate comments, maintaining ‘give and take’ in conversations, etc.). Taking this into account, their low scores were to be expected.

Another variable tested was the role of the type of school that the students with ASD attended, and its relation with their social skills. The most important differences, among the three groups (students in a typical class with a supporting teacher, students in a special school, and those in an inclusion class), were found in the Detrimental Social Behavior and Social Receptivity factors. Specifically, children attending typical classes achieved the highest scores, children in special schools reached the middle scores, and the students in integration classes achieved the lowest in these factors. So it seems that children in typical classes behave more appropriately in the social context and do not let others manipulate them, while children in integration do.

Although there are no previous studies of the role of school type in the social skills of students with ASD, in the context of the Greek educational system, children with more serious ASD symptoms are usually placed in special schools or integration classes, while high-functioning children with ASD are placed in typical classes with supporting teachers. Although we did not find major differences in the social skills of the three groups, it seems that our results confirm the policy regarding students’ school placement. However, it is impossible to determine whether these results are ultimately due to this policy or due to the fact that children in typical classes have more opportunities to interact with other typical children. Also, these results could have been attributed to the supporting teachers and the facilitation they provide in these interactions in the classroom.

In conclusion, it seems that Intellectual Disability is the most important variable which strongly influences the children’s with ASD socialization. Also, language is the second one in the row. Concerning, the rest of the variables (type of school, age, and gender) influence only a few factors and so it seems that they do not significantly differentiate the social skills of children with ASD.

Limitations and suggestions for future research

Although the results of the present study provide valuable information regarding a number of different variables and their relations with the social skills of students with ASD, a number of limitations should be noted. The first limitation stems from the relatively small number of participants. This limitation decreases the generalization of the results. We are already running a new study on a nationwide scale in Greece in order to achieve more definitive results.

The second limitation lies in the fact that the sample was not homogeneous. It consisted of children with ASD who varied in the severity of their problems. For example, we did not have an accurate level of IQ and language. Therefore, our results on the effect of intellectual disability on social skills should be treated with caution. Future research should confirm children’s diagnoses and the functionality of their speech.

The third limitation lies in the fact that the assessment of social skills was based on a single teacher’s observations for each child, thus making it rather subjective. In the future, similar studies should take into account more than one assessment for each child.

Finally, the questionnaire of the present study included several questions related to verbal communication. Therefore, it would be preferable to use a tool for children with ASD language deficits in future studies, in order to evaluate their social skills.