Autism Spectrum Disorder (ASD) is a neurodevelopmental condition involving significant deficits in social communication and the presence of restricted, repetitive or stereotyped patterns of behavior, interests, or activities (American Psychiatric Association [APA] 2013). The prevalence of ASD has grown over the past two decades and is currently estimated at 1 in 59 children (Baio et al. 2018). As the rate of ASD diagnosis grows, it is becoming increasingly important to accurately identify co-morbid symptoms of depression, anxiety, and hyperactivity in the context of ASD in order to understand symptom expression and develop useful interventions.

Self-report is commonly used for the assessment of depression and anxiety (Hagopian and Jennett 2014; Woolley and Curtis 2007); however, the validity of self-reports of these internalizing issues within the autism spectrum have been questioned (Baron-Cohen 2002; Mazefsky et al. 2011). [At least two theoretical models concerning ASD, Alexithymia (Bird and Cook 2013) and Theory of Mind (Spek et al. 2010), are consistent with these potential issues with self-report of internalizing symptoms.]

Self-report may also be involved in the assessment of ADHD—especially as individuals enter adolescence and their autonomy from caregivers and introspective abilities increase (Danckaerts et al. 1999); however, there are currently no case-control studies that explore how youth with ASD report symptoms of ADHD or other externalizing issues.

High-Functioning Autism Spectrum Disorder: Depression, Anxiety, and ADHD

Under the fifth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-5; APA 2013), ASD encompasses previous diagnoses of autistic disorder, Asperger’s disorder (AD), and pervasive developmental disorder-not otherwise specified (PDD-NOS) from the fourth edition-text revision of the DSM (DSM-IV-TR; APA 2000). The term high-functioning autism spectrum disorder (HFASD) distinguishes the group of individuals who both meet criteria for ASD and have intact language and cognitive abilities [i.e., no lower than the borderline range (Volker 2012)].

Students with HFASD are at increased risk for co-morbid psychiatric conditions such as depression, anxiety, and ADHD (Mazefsky et al. 2011). In general, students with HFASD are more likely to experience clinically relevant levels of depression and anxiety than the general community of same-age peers (Kim et al. 2011). Hypothesized factors that may contribute to the development of these internalizing symptoms include negative social experiences and repeated social failures; relative to other students on the autism spectrum, those who are higher functioning tend to have greater levels of social and self-awareness, and greater interest in social relationships, which can make them especially vulnerable to experiencing symptoms of depression and anxiety (Lopata et al. 2010). Additionally, symptoms of ADHD (e.g., hyperactivity and inattention) are also common in individuals with autism spectrum disorder (APA 2013; Jang et al. 2013; Mazzone et al. 2012). Consistent with these increased risks, Volker et al. (2010) found that mean scores of depression, anxiety, hyperactivity, and attention problems on the Behavior Assessment System for Children-Second Edition Parent Rating Scales (BASC-2 PRS; Reynolds and Kamphaus 2004) were all significantly higher for students (ages 6–16 years) with HFASD compared to typically developing children matched by age, gender, and ethnicity.

Depression in ASD/HFASD

Researchers generally recognize that depression is more common among individuals with ASD compared to the general population (Matson and Williams 2014). The prevalence rate of depression in children and adolescents with ASD has ranged from 1.4 to 30%, though particular clinical study samples have reported higher rates (Anderson et al. 2015). However, researchers suggest that depression occurs, or is detected, less frequently in individuals with ASD who have lower intellectual abilities (Anderson et al. 2015). Similar to what is suspected for anxiety, levels of depression may be higher in youth with HFASD because they may be more aware of their social difficulties than youth with lower functioning ASD, which can contribute to negative thinking and experiencing depressive symptoms (De-la-Iglesia and Olivar 2015).

Depression in youth with HFASD can generate negative outcomes, as it may lead to an increased risk of suicidal ideation (Matson and Williams 2014). For example, Mayes et al. (2013) found that youth in an ASD sample showed suicidal ideation at a rate 28 times higher than youth in a typically developing comparison sample. Thus, the importance of identifying appropriate methods of assessment for this issue is clear.

Anxiety in ASD/HFASD

There is considerable evidence to suggest that youth with ASD experience significantly higher levels of anxiety compared to normative samples (Bellini 2004; van Steensel and Heeman 2017). More surprisingly, van Steensel and Heeman (2017) found that youth with ASD also have higher levels of anxiety compared to clinically referred youth. According to a meta-analysis by van Steensel et al. (2011), encompassing studies between 1979 and 2011, prevalence rates for children and adolescents with ASD who have at least one comorbid anxiety disorder (e.g., Obsessive–Compulsive Disorder, Social Anxiety Disorder, Specific Phobia, Generalized Anxiety Disorder, Separation Anxiety Disorder, Panic Disorder, and/or Agoraphobia) ranged from 7.5 to 84.1% (M = 39.6%). As stated above, researchers such as Lopata et al. (2010) and Ung et al. (2013) suggested that this is because individuals with HFASD are more likely to have greater social awareness, insight, and interest in social relationships. However, they still have considerable difficulties with social interactions and relationships, which can lead to the development of anxiety. The presence of co-morbid anxiety disorders in youth with ASD is associated with increased risk of functional impairment, which can lead to lower quality of life (Ung et al. 2013), may exacerbate interpersonal challenges, and contribute to behavior problems (White et al. 2012).

ADHD within ASD/HFASD

The DSM-5 was the first edition of the DSM that allowed for co-morbid diagnoses of ADHD and ASD. Prior to this edition, DSM-IV-TR’s trumping rule [i.e., “…a diagnosis of Attention-Deficit/Hyperactivity Disorder is not made if Autistic Disorder is present” (APA 2000, p. 74)] resulted in a lack of consistent co-morbidity data concerning the prevalence of ADHD within cases of ASD. Mazzone et al. (2012) stated that, although the trumping rule existed, research practice and theoretical models suggested that the comorbidity between ADHD and ASD was real, relevant, and of frequent occurrence. More recently, Jang et al. (2013) reported that ADHD is one of the most common comorbid disorders in individuals with ASD—with rates ranging widely from 14 to 78% across studies. Pitzianti et al. (2016) suggested that the comorbidity between HFASD and ADHD is related to impairment in executive functioning. Existing research also suggests that individuals with comorbid ASD and ADHD are more likely to have more severe psychiatric difficulties (Jang et al. 2013).

Source Differences in Reports of Internalizing Symptoms within ASD/HFASD

Researchers in the field assess depression and anxiety in youth with ASD using a variety of different methods (e.g., questionnaires, rating scales, interviews, etc.) and rely on different sources (e.g., parent, clinician, teacher, self-report, etc.) for assessment information. As such, findings concerning agreement between source reports are mixed. Some studies (Farrugia and Hudson 2006; Ozsivadjian et al. 2014) have found evidence to support that youth with HFASD report similar levels of internalizing issues when compared to other reporters (i.e., parents, teachers, and clinicians). However, the majority of studies that have explored the level of agreement between youth with HFASD and other raters on scales of depression and anxiety have found that youth with HFASD tend to report lower levels of depression and anxiety compared to other raters. On average, youth with HFASD also tend to report similar levels of depression and anxiety as their typically developing peers. As previously mentioned, current research consensus is that many youth with HFASD experience depression and anxiety at higher rates than the normative, typically developing population (Bellini 2004; Kim et al. 2011; Matson and Williams 2014; van Steensel et al. 2011); however, this is not typically reflected well when youth with ASD self-report their symptoms of depression and anxiety (Barnhill et al. 2001; Bitsika and Sharpley 2015; Kaat and Lecavalier 2015; Lopata et al. 2010; White et al. 2012). Additionally, child development research suggests that the accuracy of self-reports depend on a child’s cognitive development, language skills, understanding of emotions, self-awareness, and perception of others (Farrugia and Hudson 2006), which are often areas of difficulty for children with ASD.

Source Differences in Reports of ADHD/Externalizing Symptoms

To our knowledge, no previous research findings have addressed how children and adolescents with ASD self-report their symptoms of ADHD (e.g., hyperactivity, inattention, etc.). Given this situation, the review here is focused on research findings regarding the self-report of youth with ADHD. As with the assessment of depression and anxiety, the clinical assessment of externalizing symptoms, such as hyperactivity, commonly involves multiple informants and methods. Overall, prior research (Cantwell et al. 1997; Danckaerts et al. 1999; Hartung et al. 2005; Rietz et al. 2016) provides evidence that individuals with ADHD report fewer symptoms and impairments that are associated with ADHD when compared to parent report. However, there is evidence that individuals with ADHD report higher levels of impairment and more symptoms than typically developing controls (Fischer et al. 1993; Klimkeit et al. 2006; Lufi and Parish-Plass 1995). These research findings indicate that individuals with ADHD report some symptoms of their ADHD, but may be underreporting the intensity or frequency of those symptoms in comparison to parent raters. Although there are currently no research studies that have examined how individuals with HFASD may self-report on symptoms of ADHD, the possibility of a similar trend is reasonable. Externalizing behaviors are generally more directly observable than internalizing issues, could be self-reported with less reliance on internal perceptions, and those with HFASD may encounter more direct feedback from the environment concerning externalizing behaviors like hyperactivity or aggression.

Current Study

The purpose of the current study was to replicate and expand on the Lopata et al. (2010) case-control study. Like Lopata et al., the present study examined, in terms of symptom levels and source differences, BASC-2 self-report and parent report ratings of depression and anxiety in cases with HFASD and matched typically developing (TD) controls. In addition, the present study examined possible generalization of the findings to additional self-report measures of depression (i.e., CDI) and anxiety (i.e., MASC), and began to extend the generality of the findings to externalizing issues using both self-ratings and parent ratings of hyperactivity. Five primary hypotheses were tested. Each represents an anticipated pattern of results within a separate 2 × 2 within-subjects ANOVA.

Hypotheses 1 and 2

For both depression and anxiety, BASC-2-PRS ratings will be significantly higher than BASC-2-SRP ratings in the HFASD condition, and significantly higher than the BASC-2-PRS ratings in the TD condition—after controlling for age, sex, and parent education. Also for both constructs, self-report (as assessed by the BASC-2-SRP) will be similar and non-substantively different (i.e., no greater than a small effect size and means within the average range) across the HFASD and TD conditions—after controlling for age, sex, and parent education.

Hypotheses 3 and 4

Separately for depression (as measured by the CDI total score) and anxiety (as assessed by the MASC total score), self-report will be similar and non-substantively different (i.e., no greater than a small effect size and means within the average range) across the HFASD and TD conditions—after controlling for age, sex, and parent education.

Hypothesis 5

BASC-2-PRS ratings of hyperactivity will be significantly higher than BASC-2-SRP ratings of hyperactivity in the HFASD condition, and significantly higher than the BASC-2-PRS ratings of hyperactivity in the TD condition—after controlling for age, sex, and parent education.

Method

Participants

A total of 88 youths ages 8–16 years old participated in the current study. Included in the sample were two groups. Group one was composed of 44 children with HFASD (age M = 10.33 years, SD = 1.86) and group two was composed of 44 TD children (age M = 10.30 years, SD = 1.84). The TD children were matched to the HFASD cases based on age, sex, and parent education. The majority of participants in each group were male (81.8%) and Caucasian (HFASD = 86.4%, TD = 93.2%). The average years of parent education was 15.72 (SD = 1.90) in the HFASD condition, and 15.68 (SD = 1.82) in the TD condition, ranging from 12 to 19+ years in each group.

Instrumentation

Behavior Assessment System for Children, Second Edition—Parent Rating Scale (BASC-2-PRS) and Self-Report of Personality (BASC-2-SRP)

Participants were assessed using the BASC-2 (Reynolds and Kamphaus 2004), which is a multi-method assessment system used to evaluate behavior and self-perceptions, along various adaptive and clinical behavior dimensions. The BASC-2 Parent Rating Scales (PRS) and Self-Report of Personality (SRP) were used in the study and are described here. Parents either completed the PRS-Child (PRS-C) for children ages 6–11 years old or the PRS-Adolescent (PRS-A) for adolescents ages 12–21 years old. Youth participants either completed the SPR-Child (SRP-C) for children ages 8–12 years old or the SRP-Adolescent (SRP-A) for adolescents ages 12–21 years old. For the PRS and SRP clinical scales, a T score in the 60–69 range is considered “at-risk” and a score that is ≥ 70 is in the “clinically significant” range. For the adaptive scales, a T score in the 31–40 range is considered “at-risk” and a score that is ≤ 30 is in the “clinically significant” range (Reynolds and Kamphaus 2004).

Though a variety of clinical and adaptive scales are available on the BASC-2-PRS and the BASC-2-SRP, only the Anxiety, Depression, Hyperactivity, Atypicality, and Attention Problems Scales from the PRS and SRP were used in the current study. [Atypicality was included because of its strong associated elevations in both ASD (see Hass et al. 2012; Volker et al. 2010) and ADHD; (Harrison et al. 2011).] In the BASC-2 normative sample, PRS coefficient alphas for the ages covered in the present study ranged from .81 to .85 for Anxiety, .86 to .88 for Depression, .82 to .86 for Hyperactivity, .79 to .85 for Atypicality, and .85 to .88 for Attention Problems; SRP coefficient alphas for the ages covered in the present study were .86 for Anxiety and .76 for Hyperactivity, and ranged from .84 to .88 for Depression, .83 to .84 for Atypicality, and .76 to .78 for Attention Problems. Concurrent validity evidence for these scales is presented in the manual (Reynolds and Kamphaus 2004).

The Multidimensional Anxiety Scale for Children (MASC)

Anxiety was assessed using a self-report measure, the MASC (March et al. 1997), which was designed to screen for anxiety symptoms in children. This instrument is completed by the child and consists of four scales—Physical Symptoms, Harm Avoidance, Social Anxiety, and Separation/Panic. In addition to these scales, the MASC provides a Total Anxiety Score; an Anxiety Disorder Index, which identifies children in need of clinical assessment; and an Inconsistency Index, which is used to identify random or careless responses by the child. Only the Total Anxiety Score was used in the present study. Norm referenced T scores, with a normative mean of 50 and standard deviation of 10, are calculated from the summed item raw scores. According to the MASC manual, the interpretation of T scores is as follows: 56–60 = slightly above average; 61–65 = above average; 66–70 = much above average; above 70 = very much above average (March and MHS Staff 1997). Internal consistency ranged from .87 to .89 for the MASC Total Anxiety Score (March and MHS Staff 1997). Preliminary evidence of discriminant, convergent, and divergent validity is reported in the manual (March and MHS Staff 1997) and by March et al. (1997).

Children’s Depression Inventory (CDI)

Depression was assessed using the CDI (Kovacs 2003), which is a self-report measure for children. Only the CDI Total Score was used in the present study. Total raw scores, based on the sum of all item scores, can range from 0 to 54 with higher scores indicative of greater depression severity. According to the CDI manual, the interpretation of norm referenced T scores is as follows: 56–60 = slightly above average; 61–65 = above average; 66–70 = much above average; above 70 = very much above average (Kovacs 2003). In the normative sample, internal consistency for the CDI Total Score was .86 (Kovacs 2003). The manual provides evidence generally supporting the concurrent validity of the CDI with similar measures of depression or related constructs (e.g., anxiety and self-esteem; Kovacs 2003).

Wechsler Intelligence Scale for Children-4th Edition (WISC-IV)

Cognitive ability was assessed using a four-subtest short form of the WISC-IV (Wechsler 2003). This short form consisted of Vocabulary, Similarities, Block Design, and Matrix Reasoning subtests. The short form composite IQ had an internal consistency of 0.95 and correlated 0.92 with the full test Full Scale IQ. The formula for the short form IQ, and reliability and validity coefficients, were calculated via methods provided by Tellegen and Briggs (1967) using standardization information in the WISC-IV technical manual as input.

Procedure

The children with HFASD were all tested during the pretesting period as they waited to begin participation in one of several intensive social intervention programs for children with HFASD. Consistent with Lopata et al. (2008) and Volker et al. (2010), a three-step multi-stage screening procedure was used to confirm the diagnostic status of the children in the HFASD group and to ensure that functional levels were sufficient to benefit from interventional programming of a cognitive-behavioral nature. This process included (a) parent submission of diagnostic reports from outside clinicians, (b) independent review of the diagnostic reports by two licensed psychologists with agreement that the child met ASD criteria (using a structured checklist covering DSM-IV-TR diagnostic criteria consistent with the autism spectrum), (c) cognitive assessment and observation, and (d) agreement after final independent review of all data by two licensed psychologists. If both independently determined that the child’s results were consistent with a HFASD, then the child was included in the study. These children completed the BASC-2-SRP, the MASC, and the CDI, along with several other instruments, during the pretesting phase of the treatment study. All self-report rating measures were read to the children by the examiner to assure accuracy and guard against reading difficulties. Their parents completed the BASC-2-PRS and a demographic questionnaire as part of the pretesting packet for the treatment study.

Children in the TD group were recruited from the community through public announcements, posted flyers, and word of mouth. These children completed the WISC-IV, the BASC-SRP, the CDI, and the MASC as part of a battery of tests given to those who volunteered to be typical case controls. (As above, self-report rating measures were read to each child by the examiner.) Their parents completed the BASC-2-PRS, an educational and developmental history form, and a diagnostic screening interview to rule out possible psychiatric conditions or developmental disorders. For their participation, children in the TD group were compensated with a gift card good for small purchases at a local merchant.

Design and Data Analysis Plan

This study was a case-control design comparing two groups of participants (i.e., those who have HFASD vs. TD participants who do not have a HFASD). TD participants were matched to the HFASD participants by age, sex, and years of parent education. For data analytic purposes, it is a 2 × 2 within-subjects design (parent vs. child × HFASD vs. TD). The dependency inherent in parent–child dyads (both rating the child) and the dependency created through case matching warranted within-subjects analyses along both dimensions.

A priori power analyses were conducted, using G* Power (Faul et al. 2007), to determine the number of participants needed to achieve at least 80% power in the current study for depression and anxiety analyses (estimates for hyperactivity were not available) based on Lopata et al. (2010) effect size estimates. The results across four F test interaction terms indicated that the largest sample size needed to achieve at least 80% would be 40 matched parent/child pairs (n = 80).

Depression, anxiety, and hyperactivity ratings from 44 matched pairs were analyzed using 2 × 2 within-subjects ANOVAs for each dependent variable (i.e., five ANOVAs). In one analysis, depression was measured by the BASC-2-PRS and the BASC-2-SRP; in another, it was measured by the BASC-2-PRS and the CDI. Anxiety was measured by the BASC-2-PRS and the BASC-2-SRP in one analysis and by the BASC-2-PRS and the MASC in another. Finally, hyperactivity was measured by the BASC-2-PRS and the BASC-2-SRP. These five ANOVAs were subjected to a Bonferroni correction and tested at a comparison-wise alpha of p < .01, keeping the experiment-wise error at p < .05. (BASC-2 atypicality and attention problem scores were also examined in secondary analyses without additional alpha correction.)

Results

Data Analyses

Demographic differences between the HFASD and TD control groups were investigated using paired samples t tests for continuous variables (i.e., age, parent education, and IQ) and Fisher’s exact tests for categorical variables (i.e., sex and ethnicity). Differences between the two conditions (i.e., HFASD vs. TD) and two rater types (i.e., parent report vs. child self-report) in terms of symptoms of depression, anxiety, and hyperactivity were analyzed using 2 × 2 within-subjects ANOVAs. In addition, two further 2 × 2 ANOVAs were conducted, as secondary analyses for two other BASC-2 scales (i.e., atypicality and attention problems) that were available across both the parent rating and self-report forms. Finally, inter-variable correlations within each condition were calculated and reported.

Demographics Comparisons

To help ensure that demographic differences did not account for the results, comparisons were made between the HFASD and TD control groups for age, parent education, IQ, sex, and ethnicity. Potential differences in age, years of parent education, and IQ were assessed using dependent samples t tests. Differences in age (t[43] = .304, p = .763) and parent education (t[43] = .221, p = .826) were non-significant; while a statistically significant difference was found for IQ (t[43] = − 3.97, p < .001; HFASD M = 101.66, SD = 15.59; TD M = 113.25, SD = 10.93). (Examination of IQ differences as a potential confound is addressed in later analyses.) Potential sex and ethnicity differences across groups were assed using the Fisher’s exact test. Differences were non-significant for both (sex, p = 1.00; ethnicity, p = .484).

Primary Analyses: Depression, Anxiety, and Hyperactivity

Primary analyses consisted of five 2 × 2 within-subjects ANOVAs with the alpha level constrained to .01 per ANOVA in order to keep the overall alpha across the five ANOVAs at .05. For each 2 × 2 ANOVA, one factor consisted of condition (i.e., HFASD vs. TD) and the second factor consisted of rater type (i.e., child self-report vs. parent report of child). Both factors were considered to be within-subjects variables, because cross-condition cases consisted of matched pairs (i.e., matched by age, sex, and parent education) and rater pairs consisted of parent and child where both rated the child. Thus, treating both factors as within-subjects variables allowed for the explicit modeling of dependency within the analyses.

Results for the five ANOVAs are listed in Table 1 and visually depicted in Fig. 1. The first ANOVA involved depression ratings using the BASC-2 PRS and SRP as the dependent variable. The 2 × 2 interaction was significant (F [1, 43] = 40.049, p < .001, \(\eta _{G}^{2}\) = .09). Though main effects were also significant, they were likely confounded by the interaction, as such, interpretation focused on the interaction alone. Bonferroni corrected simple comparisons were conducted to probe the interaction. These indicated that within the HFASD condition, parent ratings of child depression were significantly higher than the child’s self-ratings of depression (t [43] = 6.894, p < .001, \(d_{z}^{\prime }\) = − 1.04; PRS M = 65.98 [SD = 15.19], SRP M = 50.73 [SD = 9.56]). Among parent raters, parents of children with HFASDs rated depression significantly higher for their children than did parents of TD children (t [43] = 7.688, p < .001, \(d_{z}^{\prime }\) = 1.16; HFASD PRS M = 65.98 [SD = 15.19], TD PRS M = 48.82 [SD = 9.10]). Within the TD group, parent and child ratings of depression were not significantly different from each other (t [43] = 1.600, p = .117, \(d_{z}^{\prime }\) = − .241). Among child self-ratings, differences between the HFASD and TD conditions were not significantly different following the Bonferroni correction (t [43] = 2.639, p = .012, \(d_{z}^{\prime }\) = .398). Thus, Hypothesis 1 was supported by the pattern of ANOVA results.

Table 1 Results of Primary 2 × 2 ANOVAs with Measures of Depression, Anxiety, and Hyperactivity as Dependent Variables
Full size table
Fig. 1
figure 1

Graphical representation of 2 × 2 ANOVA analyses

Full size image

The second ANOVA involved anxiety ratings using the BASC-2 PRS and SRP as the dependent variable. The 2 × 2 interaction was significant (F [1,43] = 10.170, p = .003, \(\eta _{G}^{2}\) = .04), as was the main effect for condition—but only the interaction was interpreted. Bonferroni corrected simple comparisons showed that within the HFASD condition, parent ratings of child anxiety were significantly higher than child self-report of anxiety (t [43] = 3.544, p = .001, \(d_{z}^{\prime }\) = − 6.66; PRS M = 57.66 [SD = 9.62], SRP M = 51.00 [SD = 10.06]). Among parent raters, parents of children with HFASDs rated their children’s anxiety significantly higher than did parents of TD children (t [43] = 5.416, p < .001, \(d_{z}^{\prime }\) = .816; HFASD PRS M = 57.66 [SD = 9.62], TD PRS M = 48.50 [SD = 8.29]). Within the TD group, parent and child ratings of anxiety were not significantly different (t [43] = 0.593, p = .556, \(d_{z}^{\prime }\) = .090). Among child self-ratings, ratings of anxiety were not significantly different between the HFASD and TD conditions (t [43] = 0.964, p = .340, \(d_{z}^{\prime }\) = .145). Thus, Hypothesis 2 was supported by the pattern of ANOVA results.

In the third ANOVA, depression was the outcome of interest; however, child self-ratings were conducted using the CDI while parent ratings of the child’s depression continued to rely on the BASC-2 PRS depression score. The 2 × 2 interaction (F [1, 43] = 37.370, p < .001, \(\eta _{G}^{2}\) = .09) and both main effects were significant; however, due to confounding of main effects by the interaction, interpretation focused on the interaction alone. Bonferroni corrected simple comparisons indicated that within the HFASD condition, parent ratings of child depression measured by the BASC-2 PRS were significantly higher than the child’s self-ratings of depression measured by the CDI (t [43] = 7.668, p < .001, \(d_{z}^{\prime }\) = − 1.16; PRS M = 65.98 [SD = 15.19], CDI M = 49.05 [SD = 10.29]). Within the TD group, parent and child ratings of depression were not significantly different from each other after the alpha correction (t [43] = 2.452, p = .018, \(d_{z}^{\prime }\) = − .370). Among child self-ratings, differences between the HFASD and TD conditions were not significantly different following the alpha correction (t [43] = 2.144, p = .038, \(d_{z}^{\prime }\) = .323). (Differences between the parent ratings of depression on the PRS, across the two conditions, were reported in the earlier paragraph for the first ANOVA.) Hypothesis 3 was supported by the non-significant difference between the two conditions on CDI self-reports.

In the fourth ANOVA, anxiety ratings were the outcome of interest; however, child self-ratings were measured using the MASC while parent ratings continued to be measured using the BASC-2 PRS anxiety score. This 2 × 2 interaction was also significant (F [1, 43] = 8.416, p = .006, \(\eta _{G}^{2}\) = .05). Probing the interaction using Bonferroni corrected simple comparisons indicated that within the HFASD condition, parent ratings of child anxiety measured by the BASC-2 PRS were not significantly different than the child’s self-report of anxiety measured by the MASC (t [43] = 1.755, p = .086, \(d_{z}^{\prime }\) = − .265). Within the TD group, child ratings of anxiety were significantly higher than parent ratings of anxiety (t [43] = 2.894, p = .006, \(d_{z}^{\prime }\) = .436; PRS M = 48.50 [SD = 8.29], MASC M = 53.18 [SD = 8.86]). Among child self-ratings, MASC ratings of anxiety between the HFASD and TD conditions were not significantly different (t [43] = 0.129, p = .898, \(d_{z}^{\prime }\) = .019). (Differences between the parent ratings of anxiety on the PRS, across the two conditions, were reported in a prior paragraph for the second ANOVA.) Hypothesis 4 was supported by the non-significant difference between the two conditions on MASC self-reports.

In the fifth ANOVA, hyperactivity ratings from the BASC-2 PRS and SRP were the dependent variable. The 2 × 2 interaction (F [1, 43] = 60.308, p < .001, \(\eta _{G}^{2}\) = .18) and both main effects were significant, but only the interaction was interpreted. As indicated by Bonferroni corrected simple comparisons, within the HFASD condition, parent ratings of child hyperactivity were significantly higher than the child’s self-ratings of hyperactivity (t [43] = 8.300, p < .001, \(d_{z}^{\prime }\) = − 1.25; PRS M = 66.34 [SD = 11.90], SRP M = 52.11 [SD = 9.58]). Among parent raters, parents of children with HFASD rated hyperactivity significantly higher for their children than did parents of TD children (t [43] = 8.667, p < .001, \(d_{z}^{\prime }\) = 1.31; HFASD PRS M = 66.34 [SD = 11.90], TD PRS M = 47.86 [SD = 9.11]). Within the TD group, differences between parent and child ratings of hyperactivity were not significant after the Bonferroni correction (t [43] = 2.214, p = .032, \(d_{z}^{\prime }\) = .333). Among child self-ratings, differences between the HFASD and TD conditions were not significantly different than each other (t [43] = 0.257, p = .799, \(d_{z}^{\prime }\) = .039). Hypothesis 5 was supported by the pattern of ANOVA results.

Secondary Analyses

Two additional secondary 2 × 2 ANOVA analyses were available across the BASC-2 PRS and BASC-2 SRP (see Table 2; Fig. 1). Though specific hypotheses were not made regarding these subscales, potential cross-condition by cross-rater differences were likely given the nature of the constructs involved and the samples used. Thus, the results of these analyses are included for interested readers. Though the 2 × 2 ANOVA results for these two dependent variables will be assessed at an alpha level of .01, the Bonferroni correction only fully protects the primary analyses and not the secondary.

Table 2 Results of Secondary 2 × 2 ANOVAs with Atypicality and Attention Problems as Dependent Variables
Full size table

In the first secondary ANOVA, ratings of atypicality from the BASC-2 PRS and BASC-2 SRP were the dependent variable. Both main effects and the interaction were significant. The interaction was probed using Bonferroni simple comparisons and revealed that, within the HFASD condition, parents rated child atypicality significantly higher than the children’s self-ratings. Among parent raters, parents of children with HFASDs rated atypicality significantly higher for their children than did parents of TD children. Within the TD group, parent and child ratings of atypicality were not significantly different than each other. Among child self-ratings, atypicality ratings between the HFASD and TD conditions were not significantly different following the Bonferroni correction (see Table 2).

In the second secondary ANOVA, attention problems measured using the BASC-2 PRS and BASC-2 SRP was the dependent variable. The interaction was significant. Bonferroni corrected simple comparisons indicated that, within the HFASD condition, parent ratings of child attention problems were significantly higher than child self-ratings of attention problems. Among parent raters, parents of children with HFASD rated their children’s attention problems significantly higher than parents of TD children rated their children’s attention problems. Within the TD group, parent and child ratings of attention problems were not significantly different following the Bonferroni correction. Among child self-ratings, attention problems ratings were not significantly different between the HFASD and TD conditions (see Table 2).

Inter-Variable Correlations within Each Condition

Inter-variable correlations within each condition are reported in Table 3. As expected, child IQ and years of parent education were positively and significantly correlated within the TD condition; however, it is noteworthy that the child’s IQ and years of parent education were not significantly related within the HFASD condition. For both conditions, correlations between the BASC-2 PRS and BASC-2 SRP ratings of the same construct were generally low and non-significant—with occasional exceptions. Self-report cross-measure correlations between the BASC-2 SRP depression and CDI total score, and the BASC-2 SRP anxiety and MASC total anxiety score, were generally moderately and significantly correlated within both conditions.

Table 3 Inter-variable correlation matrices for high-functioning autism spectrum disorder and typically developing conditions
Full size table

Controlling for IQ Differences

In order to assess the potential influence of IQ differences across the groups on the outcome variables (e.g., anxiety, depression, hyperactivity, atypicality, and attention problems), cases were re-matched by age, sex, and IQ. This yielded a subset of 23 matched pairs (22 male pairs; 1 female pair). For this subset, the average difference in IQ was less than one point (HFASD M = 110.38; TD M = 111.32). Mean age in years for the HFASD sample was 9.88 and 10.13 for the TD sample. Mean parent education was 15.61 for the HFASD sample and 15.65 for the TD sample. Using these 23 matched pairs, the prior seven 2 × 2 within-subjects ANOVAs were re-run. The interaction terms for six of the seven prior ANOVAs continued to be statistically significant at an alpha level of p < .01. The interaction term for the ANOVA involving the BASC-2-PRS anxiety rating and the MASC anxiety child self-rating yielded a p = .064; however, the pattern of effect size estimates replicated for all seven ANOVAs (i.e., there were no substantive changes in effect size between the original seven ANOVAs run with 44 matched pairs and the replicated analyses with 23 matched pairs).

Discussion

While evidence suggests higher rates of co-morbid depression, anxiety, and ADHD among those with ASD/HFASD, estimates vary considerably across studies (Anderson et al. 2015; Jang et al. 2013; van Steensel et al. 2011). Assessment source differences, operational and measurement differences, and sample characteristics likely play significant roles. The relative value of self-report ratings vs. parent ratings of these issues, especially depression and anxiety, has been an area of concern among those who research ASD/HFASD—due to concerns that this population may have more difficulties with self-awareness, emotional understanding, perception of internal states, etc. (Baron-Cohen 2002; Lopata et al. 2010; Mazefsky et al. 2011). Theoretical concerns regarding delayed development in theory of mind (Baron-Cohen et al. 1985; Spek et al. 2010) or a higher prevalence of alexithymia (Bird and Cook 2013) in ASD contribute to these concerns, as they could compromise one’s ability to perceive and express internal states. It is noteworthy, that though emphasis on use of multiple assessment methods and multiple sources is standard practice in applied assessment (Sattler 2008; Whitcomb 2018), self-report is often relied upon more heavily in assessing internalizing issues while direct observations and parent/teacher ratings tend to play a greater role in the assessment of externalizing issues (Whitcomb 2018). Thus, impaired or biased self-report is a critical concern, especially for internalizing issues—where external observers and raters have much less access to important indicators, which can lead to delayed identification and intervention efforts.

Research comparing self-report ratings with parent ratings of externalizing issues, like hyperactivity, with the ASD/HFASD population has been limited. Those who research and assess ADHD, outside of the autism spectrum, tend to rely more heavily on parent/teacher ratings and observations (Pelham et al. 2005). Studies examining parent report and youth self-report in ADHD indicate that parent ratings tend to be significantly higher than youth self-report ratings (Cantwell et al. 1997; Danckaerts et al. 1999; Hartung et al. 2005; Rietz et al. 2016), while self-report ratings of those with ADHD tend to be higher than self-report ratings of TD controls (Fischer et al. 1993; Klimkeit et al. 2006; Lufi and Parish-Plass 1995).

Contextualizing Results

The present case-control study replicated and extended the findings of Lopata et al. (2010). Consistent with Lopata et al., the present results indicated that when depression and anxiety were assessed using the BASC-2-PRS and BASC-2-SRP, mean parent ratings were significantly different for the HFASD condition relative to self-report in the HFASD condition, and relative to parent and self-report in the TD condition. It is noteworthy that the mean parent-rated depression score was in the at-risk range for the HFASD condition, while the three other depression ratings (HFASD self-report, parent TD report, and TD self-report) were well within the average range. However, though the parent-rated anxiety score for the HFASD condition was significantly higher than the other three anxiety ratings (HFASD self-report, parent TD report, and TD self-report), all four mean ratings fell within the average range.

One extension of the study, beyond that conducted by Lopata et al. (2010), was the incorporation of an additional depression self-report measure (CDI) and an additional anxiety self-report measure (MASC). Though hypotheses were only made at the self-report level across HFASD and TD condition for these measures, they were each evaluated in separate 2 × 2 ANOVAs using the BASC-2-PRS depression or anxiety score as the parent estimate. (Note that the parent vs. self-report ratings in this case are across different measures with more differences in items and different normative samples, but also reflect a combination of measures that might be used by a practicing clinician.)

The CDI total score did not significantly differ across HFASD and TD conditions following the alpha correction. It is noteworthy that the comparison p-value would have reached significance prior to the alpha correction. Thus, the failure of the mean 4.05-point difference to achieve significance could have been a beta error due to inadequate statistical power. However, even if this was the case, the means themselves for both groups are well within the average range and not indicative of concern. It is interesting that approximately the same difference (4.12 points) was found between mean depression BASC-2-SRP scores across the two conditions. Therefore, even if a small difference is present, it is not likely to reflect a clinically meaningful difference.

The MASC total score did not significantly differ across the HFASD and TD conditions either before or after the alpha correction. The same was true for the BASC-2-SRP anxiety findings. For both self-report measures, the average difference in anxiety scores across the HFASD and TD conditions was less than one point. Thus, the effect sizes for self-reported anxiety across the two conditions were negligible.

When the MASC replaced the BASC-2-SRP anxiety score in the 2 × 2 ANOVA, the difference between the child self-report (MASC total score) and parent rating (BASC-2-PRS) of anxiety became non-significant. The MASC total score appeared to yield slightly higher (on average 2.5–3.84 points) T scores relative to the BASC-2-SRP anxiety score, which moved the MASC mean closer to the BASC-2-PRS anxiety mean in the HFASD group and moved the MASC mean further above the BASC-2-PRS anxiety mean in the TD group. Such discrepancies could be due to such issues as differences in item content sampled or differences in the normative samples for the two instruments. Regardless of the reason, clinicians should be mindful that such systematic differences are more likely to occur across instruments that were not designed and normed together.

A second extension of the study, beyond that conducted by Lopata et al. (2010), was the examination of three additional BASC-2-PRS and BASC-2-SRP scales (i.e., hyperactivity, atypicality, and attention problems). For all three of these scales, the BASC-2-PRS mean for the HFASD condition fell clearly in the at-risk range, while means for HFASD BASC-2-SRP, TD BASC-2-PRS, and TD BASC-2-SRP fell clearly within the average range.

BASC-2-SRP comparisons across HFASD and TD conditions for hyperactivity, atypicality, and attention problems were all non-significant following the alpha correction. In the case of hyperactivity and attention problems, the comparisons were non-significant even before the application of the alpha correction. However, the mean BASC-2-SRP atypicality score for the HFASD condition was significantly higher than that for the TD condition prior to the alpha correction, although both were well within the average range. Thus, it is still difficult to make an argument for this difference being potentially meaningful in a clinical sense—even if a study with greater statistical power were to render it statistically significant.

Within the TD condition, the differences between the BASC-2-PRS and BASC-2-SRP were not significantly different following the alpha correction for the hyperactivity and attention problems scales. However, the mean BASC-2-SRP score was significantly higher than the mean BASC-2-PRS score for both scales prior to the alpha correction. These average T score point differences were not necessarily meaningful for the TD condition itself, as all means (as expected) were well within the average range.

As reported in the “Results” section, the self-report cross-measure correlations between the BASC-2-SRP depression and CDI total score, and BASC-2-SRP anxiety and MASC total score, were generally in the moderate range within both conditions. These sample estimates were generally consistent with concurrent validity coefficients reported for BASC-2-SRP depression and anxiety scales in the test manual (see Reynolds and Kamphaus 2004). The cross-informant correlations for the BASC-2-PRS and BASC-2-SRP ratings of the same construct were variable, but generally considerably lower for both conditions.

An examination of the correlations between various demographic variables and scores on study measures indicated that, within each condition, the demographic variables (i.e., IQ, age, sex, and parent education) show few substantive correlations with other variables. It was noteworthy that IQ and parent education were correlated in the TD condition (as expected), but not in the HFASD condition. Beyond this, IQ was correlated with the BASC-2-SRP anxiety score in the TD condition, and with sex in the HFASD condition. It was not correlated with any other variables. We highlight this, because of the significant difference in IQ scores across the two conditions. Initially, we had expected the IQ scores between the two conditions to be more similar following the matching based on parent education. When this did not occur, we examined case records more closely and discovered that the majority of cases where difficulties with testing occurred (e.g., need for frequent refocusing, significant off task behavior, perseveration, etc.) were concentrated in the HFASD condition. These behavior issues likely adversely impacted IQ scores in some of these cases. This may have contributed to the attenuated relationship between IQ and parent education in the HFASD condition. Analyses on the subset of 23 case pairs that resulted from rematching by IQ, age, and sex were consistent with and supported the findings from the larger set of 44 pairs matched by parent education, age, and sex.

Theoretical Considerations

Although delayed development in theory of mind (Baron-Cohen et al. 1985; Happe 2003; Spek et al. 2010) or alexithymia (Bird and Cook 2013; Heaton et al. 2012) are possible contributors to reporting lower levels of internalizing issues in self-reports within HFASD relative parental reports, these may be less adequate explanations for why youths with HFASD may report lower levels of externally observable symptoms, like hyperactivity, about which the youth is likely to get more feedback from the environment. Perhaps executive function deficits, suspected in both ASD (Hill 2004) and ADHD (Craig et al. 2016), might contribute to reporting lower levels of these symptoms. It is also noteworthy that current evidence suggests that executive functions likely support the development of theory of mind (Pellicano 2012). Thus, the relationships among these potential explanatory constructs may complicate interpretations, and require more thorough measures and more sophisticated designs to control rival explanations.

Study Strengths

The present study involved a number of important strengths, including the systematic replication, in youth between 8 and 16 years old, of prior results from Lopata et al. (2010); expanded constructs examined beyond internalizing issues to include hyperactivity, attention problems, and atypical behavior; the BASC-2-PRS and BASC-2-SRP allowed for the use of co-designed and co-normed instruments for parent and self-report in five of the seven 2 × 2 ANOVAs; and self-report results for anxiety and depression were extended to an additional instrument for each construct (i.e., CDI for depression and MASC for anxiety). Methodological strengths included a relatively large sample that was adequate to test major hypotheses according to an a priori power analysis; rigorous screening procedure to confirm ASD diagnosis and assure minimum functional level requirements were met for HFASD; use of a case–control design with matched typically developing control group (reasonably ruling out age, sex, parent education, and IQ as confounds); and within-subjects analyses that appropriately modeled dependency between rater pairs and across matched cases.

Study Limitations

Despite its many strengths, the study involved limitations. First, though diagnostic and functional level confirmation processes were strong, they did not include the consistent use of a gold standard ASD diagnostic instrument [e.g., Autism Diagnostic Interview-Revised (ADI-R; Rutter et al. 2003) or Autism Diagnostic Observation Schedule-Second Edition (ADOS-2; Lord et al. 2012)]. Similarly, not all participants were surveyed for a previous diagnosis of anxiety, depression, or ADHD; although the measures used in this study provide good initial insight on mental health symptoms, they are intended to be used as screeners or as part of a more comprehensive evaluation and are not intended for diagnosis in isolation. With regards to instrumentation, parent versions of the MASC and CDI were not administered and the instruments used to measure the dependent variables are not ASD specific. Further, though the sample was reasonably large and adequate for statistical power, it was not very diverse. The sample was predominantly Caucasian, male, and with parent education averaging between 3 and 4 years of college. Also, the HFASD sample was selected according to criteria for an intervention study, which may have impacted the generalizability of the results to more broadly defined HFASD samples. Though source differences were assessed between parents and youth self-report, other potentially useful raters were not included (e.g., teachers or clinicians). Similarly, the study only focused on source differences using rating scales. It is possible that other methods, such as clinical interviews of the child vs. the parent would yield different results. Studies like this address the existence of similarities and differences across ratings, but do not address the question of whose ratings are more valid when compared to other outside criteria.

Recommendations for Future Studies

There is a lot that we do not yet understand about the relative conditions under which self-report, parent report, or other reports may be better or poorer. A fundamental question is how reliable and valid self-report ratings are within typically developing youth and youth with various clinical conditions, at different ages, across different item content/types, at different levels of functioning, etc. Thus, the comparison of those with HFASD to matched non-ASD controls diagnosed with anxiety disorders, depression, ADHD, etc. could be very informative.

More cross-informant and cross-method work would be helpful in determining which informants and methods are most useful. In addition, comparing ratings to other assessment methods, like structured clinical interviews, could reveal if there are differences in the quantity and quality of information gained across methods. Furthermore, it is critical that more attention be paid to assessing various rater or interview sources against less biased or more independent external criterion measures [e.g., direct behavior observations, neurocognitive performance measures, biological indicators of stress (e.g., cortisol), etc.]. Such comparisons could address the relative validity of data from different rating or interview sources.

Clinical Implications

Though the methods of this study cannot clearly and definitively determine whether the parent ratings or the youth self-report ratings were more accurate or useful for those with HFASD, the tentative evidence suggests that the youth self-reports yield relatively lower levels of both internalizing and externalizing issues compared to parent reports. Therefore, clinicians should be cautious when using self-report ratings from youth with HFASD and certainly not rely on them in isolation. Screening methods that rely primarily or exclusively on self-report may be more likely to generate false negatives and miss cases. That being said, in general, clinicians should take any elevated self-report scores seriously, and should not assume that lack of a self-reported elevation rules out concern. A multi-method and multi-source assessment, that includes information from others who know the youth well, should be standard practice for these issues.