EDP3004

Contents lists available at ScienceDirect

Research in Autism Spectrum Disorders

journal homepage: www.elsevier.com/locate/rasd

Social cognition and Reading comprehension in children and
adolescents with autism spectrum disorders or typical development

Nancy S. McIntyrea,⁎, Tasha M. Oswalda,b,e, Emily J. Solaria,b, Matthew C. Zajica,
Lindsay E. Lerroa,f, Claire Hughesc, Rory T. Devined, Peter C. Mundya,b

a School of Education, One Shields Avenue, University of California, Davis, 95616, USA
b UC Davis Medical Investigations of Neurodevelopmental Disorders (MIND) Institute, 2825 50th Street, Sacramento, California, 95817, USA
c Department of , University of Cambridge, Downing Street, Cambridge, CB3EB, UK
d School of , University of Birmingham, Edgbaston, Birmingham, B15 2TT, UK
e Tasha M. Oswald is now at Portia Bell Hume Behavioral Health and Training Center, 5776 Stoneridge Mall Road, Suite 340, Pleasanton, California,
94588, USA
f Lindsay E. Lerro is now at The Swain Center, 795 Farmers Lane #23, Santa Rosa, California, 95405, USA

A R T I C L E I N F O

Number of completed reviews is 2

Keywords:
ASD
Social cognition
Theory of mind
Oral language
Reading comprehension

A B S T R A C T

Background: Many individuals with autism spectrum disorders (ASD) exhibit social cognitive
impairments in the development of theory of mind (ToM), or the ability to attribute mental states
to oneself and others. ToM has been shown to relate to reading comprehension for children and
adolescents with typical development (TD) and with ASD. This study examined the relation
between reading comprehension, word recognition, oral language, and ToM for higher-func-
tioning children and adolescents with ASD (HFASD) as compared to those with TD.
Method: 70 children with HFASD and 40 children with TD, aged 9–17 years, participated in the
study. In order to describe the HFASD as compared to the TD sample, a series of ANOVAs and
ANCOVAs were conducted. Multiple regression analyses were conducted with reading compre-
hension as the outcome variable. Separate regression models (TD & HFASD) were run with IQ,
word recognition, oral language, and two ToM measures (Happé’s Strange Stories and the Silent
Films Task) as predictors.
Results: The TD group performed better than the HFASD group on all standardized and experi-
mental measures. Regression analyses revealed that after controlling for IQ, word recognition,
and oral language, both ToM measures predicted unique variance in reading comprehension in
the HFASD, but not the TD, sample. Furthermore, the TD and HFASD groups displayed different
patterns of significant predictors of reading comprehension.
Conclusions: This study suggests that in addition to oral language and higher-order linguistic
comprehension, social cognition is an important factor to consider when designing reading in-
terventions for students with ASD.

1. Introduction

Social and communication difficulties are hallmark characteristics of individuals with autism spectrum disorder (ASD;
Christensen, Baio, & Van Naarden Braun, 2016) and have been linked to underlying social cognitive impairments in the development
of theory of mind (ToM). ToM is the ability to attribute mental states (e.g., beliefs, desires, intentions) to oneself, and infer others’

https://doi.org/10.1016/j.rasd.2018.06.004
Received 5 August 2017; Received in revised form 2 June 2018; Accepted 11 June 2018

⁎ Corresponding author at: University of California, Davis, One Shields Avenue, Davis, CA 95616, USA. Tel.: +1 530 848-6702; fax: +1 530 752-6135.
E-mail address: [email protected] (N.S. McIntyre).

Research in Autism Spectrum Disorders 54 (2018) 9–20

Available online 29 June 2018
1750-9467/ Published by Elsevier Ltd.

T

http://www.sciencedirect.com/science/journal/17509467

https://www.elsevier.com/locate/rasd

https://doi.org/10.1016/j.rasd.2018.06.004

https://doi.org/10.1016/j.rasd.2018.06.004

mailto:[email protected]

https://doi.org/10.1016/j.rasd.2018.06.004

http://crossmark.crossref.org/dialog/?doi=10.1016/j.rasd.2018.06.004&domain=pdf

mental states in order to understand and predict their behavior (Baron-Cohen, Leslie, & Frith, 1985; White, Hill, Happé, & Frith,
2009). Impairments in ToM are evident across developmental stages and functioning levels in individuals with ASD, and these deficits
remain evident throughout adolescence and into adulthood, even among individuals without comorbid intellectual disabilities
(Baron-Cohen, Jolliffe, Mortimore, & Robertson, 1997; Kaland, Callesen, Møller-Nielsen, Mortensen, & Smith, 2008; White et al.,
2009).

ToM enables one to understand that oral and written language are about the expression of thoughts, emotions, and desires, as well
as the interpretation of intended meaning rather than literal meaning (Tomasello, 2010). Relatedly, performance on ToM tasks is
associated with oral language skills (Happé, 1995) and communicative competence (Hale & Tager-Flusberg, 2005). ToM and lan-
guage have been posited to have a bidirectional relation in development (Miller, 2006; Slade & Ruffman, 2005). For instance,
evidence from typically developing (TD) children suggests that ToM is a contributing factor to semantic development (Baldwin &
Moses, 2001; Birch & Bloom, 2002). On the other hand, research also suggests that exposure to, and engagement in, conversations
about mental states influence children’s ToM development (Dunn, Brown, Slomkowski, Tesla, & Youngblade, 1991; Ruffman, Slade, &
Crowe, 2002). Furthermore, a large body of evidence has demonstrated that general language skills (e.g., syntax and semantics)
contribute to performance on ToM tasks, including false belief tasks (Astington & Jenkins, 1999; Cutting & Dunn, 1999; Happé, 1995;
Slade & Ruffman, 2005), and that mastery of mental state words such as “think” and “know” are related to children’s ToM devel-
opment (Bartsch & Wellman, 1995; Shattz, Wellman, & Silber, 1983).

Challenges with ToM have the potential to impact performance in a variety of contexts, including educational settings. While data
indicate that many children with ASD have difficulty understanding the mental states or beliefs of others, there is a dearth of
information about how this could specifically impact academic achievement for these children. Reading achievement, which is also
related to language development, underpins learning in many academic milieus and serves as a conduit to overall academic
achievement. This study investigates the relation between ToM and reading achievement. Understanding the influence of language
and ToM on reading in school-aged children with ASD may provide insights into syndrome specific academic challenges.

2. Reading framework and reading achievement in ASD

The Simple View of Reading framework (Gough & Tunmer, 1986) posits that successful reading comprehension is the product of
proficient word reading skills and oral language comprehension and that poor reading comprehension may reflect weaknesses in
either, or both, domains. Prior research has demonstrated the utility of this framework for understanding reading skill in TD readers
or those with reading disorders (e.g., Catts, Adlof, & Weismer, 2006; Joshi & Aaron, 2000; Tunmer & Chapman, 2012), as well as
those with ASD (Jones et al., 2009; McIntyre et al., 2017; Norbury & Nation, 2011; Ricketts, 2011). Studies with children with ASD
have demonstrated that proficient reading comprehension is a particular challenge that impacts between 33 and 65% of ASD samples
(Jones et al., 2009; Lucas & Norbury, 2014; McIntyre et al., 2017; Nation, Clarke, Wright, & Williams, 2006; Ricketts, Jones, Happé, &
Charman, 2013). Many children with ASD exhibit impairments in oral language development (Eigsti, de Marchena, Schuh, & Kelley,
2011; Tager-Flusberg, 2006), and while considerable variability in both domains of the Simple View has been demonstrated, a strong
relation between structural language components (i.e., phonology, semantics, syntax, morphology) and reading comprehension has
been established (Lindgren, Folstein, Tomblin, & Tager-Flusberg, 2009; McIntyre et al., 2017; Norbury & Nation, 2011; Ricketts et al.,
2013). In several studies, children with ASD and structural language impairments performed significantly more poorly on measures of
word recognition, word decoding, and overall reading comprehension (Lindgren et al., 2009; Lucas & Norbury, 2014; Norbury &
Nation, 2011). Data from studies by Ricketts et al. (2013) and Williams, Minshew, and Goldstein, (2015) indicated that for children
with ASD and adequate sentence level language skills, reading comprehension impairments may be related to syndrome specific
higher order inferential and cognitive processing challenges. Individuals with ASD have demonstrated difficulty with verbal rea-
soning, inference generation, and answering questions about inferences (Lucas & Norbury, 2015; Norbury & Nation, 2011; Saldaña &
Frith, 2007; Tirado & Saldaña, 2016); this is particularly evident when needing to use ToM to make inferences about social in-
formation regarding emotional states, mental states, or intentionality (Bodner, Engelhardt, Minshew, & Williams, 2015; Happé, 1994;
Kaland et al., 2008; Le Sourn-Bissauoi, Caillies, Gierski, & Motte, 2009).

3. Relation between ToM and reading achievement

Emerging research highlights the impact of individual differences in ToM on TD children’s experiences in school and the pathways
that might link ToM to academic success. One pathway Hughes and Devine (2015) reviewed posited links between ToM and me-
tacognitive skills such as understanding that texts have an intended meaning and are representational, like mental states, and,
therefore, are open to different, sometimes incorrect, interpretations by different people. Lecce, Zocchi, Pagnin, Palladino, and
Taumoepeau (2010) reported that, when controlling for verbal abilities, individual differences in TD children’s ToM predicted later
meta knowledge about reading including understanding the attitudes and differences of people as readers, the various goals a reader
might have for a text, knowledge about different types of texts and their characteristics, and possible strategies that can be applied to
different types of texts. Kim (2015) provided evidence that ToM indirectly predicted reading comprehension through its significant
relation with listening comprehension in TD kindergarteners, supporting the hypothesis that ToM tasks capture inference making and
complex social reasoning skills that are important for understanding story characters’ and authors’ beliefs and intentions. Reading
literary fiction has been shown to promote ToM, empathy, and social development in TD children (Doise, Mugny, James, Emler, &
Mackie, 2013; Mar, Tackett, & Moore, 2010) and adults (Kidd & Castano, 2013; Mar, Oatley, & Peterson, 2009), suggesting that
exposure to fiction may change how individuals think about themselves and others. Taken together, these data indicate that ToM and

N.S. McIntyre et al. Research in Autism Spectrum Disorders 54 (2018) 9–20

10

reading comprehension are interrelated and, therefore, deficits in one might suggest deficits in the other.
Although within the ASD literature ToM deficits have been heavily studied for more than two decades, research investigating the

relation between individual differences in ToM and reading has been limited. However, examining whether individual differences in
ToM underlie interactions with, and comprehension of, texts for individuals with ASD is important since understanding the specific
factors impacting reading comprehension is a crucial step in ASD academic achievement research. Emerging research suggests that
reading comprehension difficulties are part of the social communication phenotype of individuals with ASD without intellectual
disabilities, or higher-functioning school-aged children with ASD (HFASD), demonstrating significant negative associations between
reading development and ASD symptom intensity or social functioning (Estes, Rivera, Bryan, Cali, & Dawson, 2011; Jones et al.,
2009; McIntyre et al., 2017; Ricketts et al., 2013). Narratives in particular may be challenging for children with ASD due to their
delays in development of ToM (Randi, Newman, & Grigorenko, 2010; Ricketts, 2011). Ricketts et al. (2013) examined social cog-
nition, operationalized by two advanced ToM tasks, as a predictor of reading comprehension. Their sample of 100 adolescents with
ASD ranged in age from 14 to 16 years and had nonverbal IQs ranging from 53 to 126 (M = 90.37, SD = 18.61). Utilizing two ToM
measures, a verbal-based ToM task (Strange Stories; Happé, 1994) and a non-verbal ToM task assessing the ability to interpret
intentions of silent animations of two interacting cartoon triangles (Frith-Happé animations; Abell, Happé, & Frith, 2000), Ricketts
et al. (2013) found that ToM was a unique predictor of reading comprehension even after controlling for word reading and oral
language for adolescents with ASD. This finding provided evidence that the additional factor, ToM, contributes to variance in reading
comprehension in ASD that is not accounted for by the Simple View of Reading.

4. Rationale for current study

The current study has two aims: (1) to replicate the finding of Ricketts et al. (2013) that ToM predicts unique variance in reading
comprehension beyond that explained by word reading and oral language in a sample of children with HFASD, and (2) to extend their
work by examining this relation in an HFASD sample comprised of a broader age range and to determine if differential relations exist
when comparing an HFASD with a TD control group. Our hypothesis was that after controlling for full-scale IQ (FIQ), word reading,
and oral language, ToM indices would explain significant unique variance in both the HFASD and TD groups.

5. Methods

5.1. Participants

This longitudinal research was conducted in compliance with the Institutional Review Board and written parental consent and
child assent was obtained prior to data collection. The sample consisted of 70 (58 male) children with HFASD and 35 (22 male) TD
children, aged 9 to 17 years. Enrolled participants were recruited from a university research subject tracking system, the local
community through school districts, and word of mouth. Exclusionary criteria included parent report of a syndrome other than ASD
(e.g., Fragile X), a neurological disorder (e.g., epilepsy), psychotic symptoms (e.g., hallucinations), significant sensory or motor
impairment, or any major medical disorder that could be associated with extended absences from school. Individuals were included
in the HFASD sample if they had a community diagnosis of ASD, confirmed by trained researchers using the Autism Diagnostic
Observation Schedule, Second Edition (ADOS-2; Lord et al., 2012), and a FIQ estimate between 75–130 on the Wechsler Abbreviated
Scales of Intelligence-II (WASI-2, Wechsler, 2011). Participants in the TD group did not have a community diagnosis of, or meet
criteria by parent questionnaire for, ASD and had an FIQ estimate between 75–130.

Table 1
Descriptive statistics and significant group differences in full sample.

N
Measures

HFASD
70
M (SD)

TD
35
M (SD)

ANOVA
F

p η2p

Age 12.50 (2.13) 12.84 (2.30) 0.55 .46 .01

IQ
VIQ 95.64 (15.19) 107.29 (12.31) 15.47 < .001 .13 PIQ 103.36 (14.82) 113.54 (11.74) 12.56 < .01 .11 FIQ 99.16 (13.65) 111.66 (10.34) 22.76 < .001 .18 ASD Diagnostic Measures ADOS-2 Social Affect 8.43 (3.43) – – – – ADOS-2 RRB 2.53 (1.21) – – – – ADOS-2 Total Score 10.87 (3.71) – – – – SCQ Lifetime Total 21.13 (7.02) 2.37 (2.28) 236.04 < .001 .70 SRS 82.00(10.49) 45.00 (9.03) 314.84 < .001 .76 Note. HFASD = high-functioning autism spectrum disorders; TD = typically developing; M = mean; SD = standard deviation; VIQ = verbal IQ; PIQ = performance IQ; FIQ = full-scale IQ; ADOS-2 =Autism Diagnostic Observation Scale, Second Edition; RRB = Restricted and Repetitive Behaviors; SCQ = Social Communication Questionnaire, Lifetime Edition, total raw score; SRS = Social Responsiveness Scale, T-scores. N.S. McIntyre et al. Research in Autism Spectrum Disorders 54 (2018) 9–20 11 All participants included in the HFASD sample met criteria on the ADOS-2 and scored significantly higher in ASD symptomatology than those in the TD group on the parent report questionnaires as shown in Table 1. Demographic data presented in Table 2 indicate that the ratio of boys to girls in the HFASD sample, approximately 4.8:1, is similar to national prevalence rates (Christensen et al., 2016). Generally, participants’ mothers in both groups completed at least some college, with 30% in the HFASD and 23% in the TD sample completing graduate school. The groups were matched on age (see Table 1 for descriptive statistics). However, full scale IQ (FIQ), verbal IQ (VIQ), and nonverbal performance IQ (PIQ) were significantly lower in the ASD group than TD group, with moderate to large associated effect sizes. We took two complementary approaches to tackling this methodological limitation: (a) "controlling" for FIQ in statistical analyses and (b) creating closely matched subsamples prior to statistical analysis. 5.2. Procedures and measures Participants were recruited to take part in a longitudinal study of social and academic development. Data reported are from assessment sessions that were conducted by members of trained research group in a university-based child assessment laboratory during 2.5-hour sessions. Diagnostic, demographic, and IQ data were collected at the first time point. Reading, oral language, and ToM measures were collected 15 months later at the second time point. All assessments were standardized except the experimental ToM measures; age-normed standard scores are reported on standardized measures. 5.2.1. Diagnostic measures The ADOS-2 (Lord et al., 2012) is a diagnostic assessment for ASD that has a strong predictive validity against best estimate clinical diagnoses (Charman & Gotham, 2013). Trained research personnel administered Module 3 or 4 to confirm community ASD diagnoses through the evaluation of two core domains: Social Affect and Restricted and Repetitive Behaviors. The cutoff score of seven places a child on the autism spectrum. Parent report questionnaires were administered to provide converging evidence of ASD symptomatology in the HFASD sample and to rule out ASD symptomatology in the TD sample. The SCQ Lifetime version (Rutter, Bailey, & Lord, 2003) is a 40-item parent report rating developmental symptoms of ASD in children four years and older. Scores greater than 15 suggest autism. The SRS (Constantino & Gruber, 2005) is a 65-item parent questionnaire rating dimensions of communication, social behavior, and the repetitive/stereotypic behaviors characteristic of ASD. T-scores between 60–75 indicate deficiencies in reciprocal social behavior in the mild-moderate range, while those above 76 fall in the severe range. 5.2.2. Cognition The WASI-2 (Wechsler, 2011) provided an estimate of nonverbal and verbal cognitive skills. The verbal composite (VIQ) was comprised of two subtests, Vocabulary and Similarities, which measured expressive vocabulary and abstract semantic reasoning respectively. The Performance composite (PIQ) was comprised of two nonverbal subtests, Block Design and Matrix Reasoning, which measured visual spatial processing and problem solving. The four subtests combine to provide an age-normed standard score (M = 100, SD = 15) measurement of full scale IQ (FIQ). Table 2 Demographics across subgroups. HFASD TD N 70 35 % % Gender Male 83 63 Female 17 37 Ethnicity African American 0 0 Asian 4 0 Caucasian 67 74 Caucasian Plus One Other Ethnicity 13 11 Hispanic/Latino/a 9 6 Native American 0 0 Native Hawaiian/Pacific Islander 0 3 Other 4 0 Decline to State 3 6 Mother's Highest Level of Education Some High School 1 0 Completed High School 3 0 Some College 24 26 Completed College 33 43 Some Graduate School 7 3 Completed Graduate School 30 23 Decline to State/Unavailable 2 5 N.S. McIntyre et al. Research in Autism Spectrum Disorders 54 (2018) 9–20 12 5.2.3. Reading The Gray Oral Reading Tests – Fifth Edition (GORT-5; Wiederholt & Bryant, 2012) assessed reading comprehension and yielded an age-normed standard score. The assessment is comprised of 16 progressively more difficult narrative or expository reading passages read aloud by the child, followed by 5 open-ended questions that required an oral response. Some questions rely on recall of details while others require higher order processing such as synthesis of the main idea, explaining causal relations, or making inferences. Word reading was assessed with the Test of Word Reading Efficiency, Second Edition (TOWRE-2, Torgesen, Wagner, & Rashotte, 2012), measuring accuracy and fluency of sight word recognition (Sight Word Efficiency: SWE) and phonemic decoding (Phonemic Decoding Efficiency: PDE). Participants read as many real words (SWE) or decodable nonwords (PDE) as they were able to in 45 s. The TOWRE-2 also yielded an Index score which is a standardized composite score comprised of the SWE and PDE subtests. 5.2.4. Oral language Receptive vocabulary was measured by the Wechsler Individual Achievement Test, 3rd edition (WIAT-III; Breaux, 2009). The assessment requires one to point to the picture that depicts the meaning of progressively more semantically sophisticated words, and yields an age-normed standard score. Linguistic comprehension skills such as those necessary to listen to and utilize narrative structure to organize and retell gist and verbatim details were measured with the Story Recall subtest of the Wide Range Assessment of Memory and Learning, Second Edition (WRAML-2, Sheslow & Adams, 2003). This assessment yielded an age-normed standard score. The Language Composite used in subsequent analyses was created by averaging participants’ receptive vocabulary and story recall standard scores. 5.2.5. Theory of mind ToM was measured by two advanced ToM measures that were appropriate for children and adolescents; these tasks were more complex than the false belief tasks used with young children as they required participants to apply their mental state understanding to contextualized social scenarios. The Strange Stories (Happé, 1994; White et al., 2009) task was comprised of five short stories; text was presented on a computer screen while a recorded voice read the story aloud. The stories used in this study depicted social scenarios involving deceptions, double bluffs, or misunderstandings. Participants were required to verbally respond to one question per story about a character’s beliefs, desires or behavior based on their imputed mental state. The second measure, the Silent Films Task, was created by Devine and Hughes (2013) to study advanced ToM in middle childhood and adolescence in TD individuals and to be an analogous task to Strange Stories. It was comprised of five short silent film clips presented on a computer screen; the clips were drawn from a silent comedy film by Harold Lloyd called Safety Last (Roach, Newmeyer, & Taylor, 1923). Immediately after each clip, one or two questions about the characters’ beliefs and intentions appeared on the screen and were also read aloud by the examiner. Each task was coded by one of two trained experimenters using the coding scheme developed for Strange Stories (Happé, 1994). For both measures, responses were scored using a 0-1-2 system where 0 represented an incorrect or “don’t know” response, 1 represented partial credit for reference to lower order mental states, and 2 represented a complete, correct answer that was based on an explicit inference about characters’ beliefs or intentions. Participants could receive a maximum score of 10 on Strange Stories and 12 on Silent Films. Raw scores were averaged together for each participant to form a ToM Composite score. For each ToM task, a trained experimenter coded the full sample and a second trained experimenter coded 15% of the sample. Inter-rater reliability was good for both Strange Stories (Cohen’s kappa = 0.97) and Silent Films (Cohen’s kappa = 0.85). 5.3. Data analysis In order to describe the HFASD as compared to the TD sample, a series of Analyses of Variance (ANOVAs) and Analyses of Covariance (ANCOVAs) controlling for FIQ, were conducted in SPSS version 25 to calculate sample means and standard deviations and to examine diagnostic group differences in (a) reading comprehension, (b) word reading and decoding, (c) oral language, and (d) ToM. Alpha levels below 0.05 were considered statistically significant for all analyses. Effect size for these analyses was calculated as partial eta squared (η2p) to measure the strength of association between variables. For this statistic, values between .01 and .05 are considered a small effect size, values between .06 and .14 are considered medium effect size, and above .14 are considered a large effect size. Next, correlations among the reading, FIQ, language and ToM variables were examined for strength of associations between reading comprehension and predictor variables in both diagnostic groups. Hierarchical multiple regression analyses were conducted separately for each diagnostic group with reading comprehension as the outcome variable. A series of regressions were run in the HFASD full sample (n = 70) and TD sample (n = 35) controlling for FIQ at step 1. Next, the Word Reading Index was entered as step 2, followed by the Language Composite as step 3 in each model. Finally, since all three ToM variables were significantly correlated with reading comprehension in the HFASD sample, Strange Stories, Silent Films, or ToM Composite was entered as step 4 in Model 1, 2 and 3 respectively. None of the ToM variables were significantly correlated with reading comprehension in the TD sample, however, to allow between-group comparisons, the ToM Composite score was entered as step 4 in the TD sample. Due to the difficulty of recruiting a sample of typically developing children matched on FIQ to a large sample of higher func- tioning school-aged children with ASD, covariance analyses controlled for FIQ. However, Miller and Chapman (2001) raised concerns about the limitations of covariance analyses, thus we addressed this in supplemental analyses in which we utilized a more rigorously controlled FIQ-matched subsample of individuals with HFASD (n = 37) to explore the consistency of the main findings. All analyses as described above were run for the matched samples. N.S. McIntyre et al. Research in Autism Spectrum Disorders 54 (2018) 9–20 13 6. Results 6.1. Full sample Descriptive statistics and diagnostic group differences on the standardized and experimental measures are presented in Table 3. HFASD means on the standardized word recognition, decoding, and oral language measures were in the average range as compared to the tests’ norming samples, but the group mean on the reading comprehension measure was approximately one standard deviation below average. The TD sample means were within the average to above average range on all standardized reading and language variables. Raw scores were computed for the ToM tasks and measures of normality of the distributions in each sample were examined. The Strange Stories task was designed for ASD populations, and was normally distributed in the HFASD sample, with skewness of -0.59 (SE = 0.29) and kurtosis of -0.54 (SE = 0.57). Fifty-four percent of the TD sample in this study scored 90% or higher on this measure, although skewness of -0.91 (SE = 0.40) and kurtosis of -0.27 (SE = 0.78) was within normal limits. Overall, this task was not an area of deficit for this TD sample. The Silent Films task was normally distributed in both samples: TD skewness of 0.05 (SE = 0.40) and kurtosis of -0.72 (SE = 0.57), and in ASD skewness of -0.49 (SE = 0.29) and kurtosis of -0.94 (SE = 0.78). The ToM composite variable was normally distributed in both samples [ASD skewness of -0.70 (SE = 0.29) and kurtosis of -0.54 (SE = 0.57; TD skewness of -0.41 (SE = 0.40) and kurtosis of -0.50 (SE = 0.78)] and is a more robust measure of ToM. The TD group performed significantly better than the HFASD group on all standardized and experimental measures when FIQ was not controlled. With FIQ controlled, significant differences remained in Reading Comprehension, Sight Word Recognition, Word Reading Index, Receptive Vocabulary, Story Recall, Language Composite, and the ToM composite score. Correlations between observed variables are presented in Table 4. In the HFASD group, reading comprehension was moderately to Table 3 Descriptive statistics and diagnostic group differences on standardized and experimental variables for full HFASD (n = 70) and TD (n = 35) samples. Variable HFASD M (SD) TD M (SD) ANOVA F η2p ANCOVA F η2p Read Comp 7.44 (2.86) 10.51 (2.29) 30.47*** .23 8.69** .08 PDE 94.01 (15.67) 104.63 (13.57) 11.67** .10 1.47 .01 SWE 89.50 (14.51) 102.83 (14.10) 20.05*** .16 4.97* .05 Word Read Index 91.24 (14.39) 103.97 (13.67) 18.87*** .16 3.95* .04 Vocab 105.38 (16.86) 120.38 (11.17) 21.72*** .19 4.59* .05 Story Recall 91.98 (18.11) 110.00 (9.84) …