Psychometric characteristics of the aberrant behavior checklist in a well-defined sample of youth with autism Spectrum disorder

https://doi.org/10.1016/j.rasd.2019.02.001Get rights and content

Highlights

  • Results provide further support for use of the ABC in individuals with ASD.

  • The structural, convergent, and divergent validity of the ABC were supported.

  • Normative analyses were very comparable to the only previously published norms for youth with ASD.

Abstract

Background

Behavior and emotional difficulties often occur in children and adolescents with Autism Spectrum Disorder (ASD), yet there are few instruments available to assess such problems in this population. The Aberrant Behavior Checklist (ABC), one option for this, is widely used and has substantial psychometric support. Despite this, only two studies to date have examined its structural validity in samples of individuals diagnosed exclusively with Autism Spectrum Disorder (ASD). This study sought to further examine the ABC’s validity for use with children and adolescents with ASD.

Method

Data from 470 individuals aged 2–14 years were submitted to confirmatory factor analysis (CFA). Correlations with other measures were examined. MANOVA was conducted to examine effects of subject characteristics on subscale scores of the ABC and assist in developing norms.

Results

Results supported the original factor structure of the ABC. Convergent and divergent validity analyses indicated correspondence with analogous measures and lack of relationship for dissimilar constructs. Results of normative analyses were very comparable to the only previously published norms for youth with ASD.

Conclusions

Overall, results provide further psychometric support for use of the ABC in individuals with ASD and better understanding the ABC’s clinical usefulness, particularly given the similarity between current and previous normative data. In a broader context, these results lend further support to the ABC as an instrument with wide applications.

Introduction

Assessment of behavioral and emotional difficulties in individuals with Autism Spectrum Disorder (ASD) is an important area of study due to the increased risk for such difficulties in this population. Recent data suggest a prevalence of ASD of 6 per 1000 children, 30% of whom function within the range of intellectual disability (ID) (Elsabbagh et al., 2012). The prevalence of behavioral and emotional difficulties in those with ASD is much higher than in individuals without an ASD diagnosis (e.g., Buck et al., 2014; Leyfer et al., 2006; Siminoff et al., 2008). While many instruments exist to assess for ASD symptoms (e.g., the Modified-Checklist for Autism in Toddlers, the Social Responsiveness Scale) relatively few instruments have been developed to assess behavioral and emotional difficulties in those with ASD. In the absence of such instruments, researchers and clinicians often use instruments developed for assessing emotional and behavior problems in people with other developmental disabilities. One such instrument that has demonstrated promise is the Aberrant Behavior Checklist (ABC; Aman & Singh, 2017).

The ABC is a widely used instrument originally designed to assess treatment effects in individuals with developmental disabilities. It has been used in hundreds of studies ranging from medication efficacy research to behavioral phenotyping of specific genetic conditions, such as Fragile X, Prader-Willi, Cri-du-chat, Angleman’s syndrome, and Smith Megenis Syndrome (e.g., Aman & Singh, 2017; Baumgardner, Reiss, Freund, & Abrams, 1995; Clarke, Boer, Chung, Sturmey, & Webb, 1996; Clarke & Dykens, 1997; Summers & Feldman, 1999). It has been translated from English into at least 35 other languages (Aman & Singh, 2017). It was originally developed for people living in residential facilities, but has been adapted for use in the community (Brown, Aman, & Havercamp, 2002). It is well-studied and has demonstrated strong psychometric properties across a range of samples, including adults, adolescents, and children with developmental disabilities (e.g., Hill, Powlitch, & Furniss, 2008; Marshburn & Aman, 1992). For instance, in a meta-analysis of 14 replication studies, Aman and Singh (2017) found that 85% of ABC items, across studies, loaded on the originally reported factors (Aman & Singh, 1985). However, to date only two studies (Brinkley et al., 2007; Kaat, Lecavalier, & Aman, 2014) have examined the ABC’s psychometric properties within samples comprising ASD alone. This is a valuable area of inquiry given the increasing prevalence rates of ASD, high risk of behavioral and emotional problems in those with ASD, and the stability of ABC psychometric properties across other populations.

One of the studies examining the validity of the ABC exclusively in a sample of those with ASD conducted exploratory and confirmatory factor analysis (EFA and CFA, respectively) with 275 individuals aged three to 21 years (Brinkley et al., 2007). Results overall suggested the ABC factor structure was robust in this sample, though some discrepancies from the original factor structure were noted. In EFA, both four- and five- factor solutions emerged. In the five-factor solution, the majority of items (over 70%) loaded onto original factors and 76% of the variability was accounted for by the solution. The Irritability factor did not emerge, and instead many items from the original Irritability factor loaded with items on the original Hyperactivity/Noncompliance subscale, creating a “general disruptive behavior” factor. Additionally, a few items formed a new factor representing self-injury, a finding which has sometimes occurred in other samples, such as those with Fragile X (Sansone et al., 2012). CFA indices of fit for this model suggested moderate fit. The four-factor solution accounted for 71% of the variance in the data. This solution was compared to the four-factor solution obtained with teacher-completed data in a sample of special education students (Marshburn & Aman, 1992) and 78% of the items were found to load onto the anticipated factor. Again an Irritability factor failed to emerge and a Self-injury factor (with three items) did emerge.

More recently, Kaat et al. (2014) submitted ABC data from 1893 youth with ASD aged two to 18 years to EFA and CFA; the large overall sample size permitted for both a calibration and a validation sample. Overall, Kaat et al. found that the original factor structure performed well in exploratory and confirmatory analyses, and convergent and divergent validity was supported. In the five-factor solution which emerged from EFA, 90% of the items loaded onto the original factors. Five different models (including both the 4- and 5-factor models reported by Brinkley et al., 2007 and others) were submitted to CFA with marginal-to-poor fits indicated for all. Possible effects of participant characteristics on subscale scores were examined. Age, sex, and IQ were found to be largely unrelated to scores, though to a mild extent increased age was associated with decreased hyperactivity and irritability, and lower IQ was associated with increased stereotypic behavior. Autism symptom severity (as measured by the ADOS comparison score) was generally unrelated to the ABC subscale scores, and scores of the Child Behavior Checklist (CBCL; Achenbach & Rescorla, 2000) were generally correlated with analogous ABC subscales, as anticipated. Normative data also were presented based on age group and IQ, as analyses did not support gender-related differences; these norms are the only published set for individuals with ASD to date.

Of the two previous studies examining the ABC in ASD-specific samples, Kaat et al. drew participants from a large registry (the Autism Treatment Network) which permitted for a large sample size. The other study conducted by Brinkley et al. had a well-characterized though smaller sample, which did not permit for splitting participants into separate subsamples for the EFA and CFA; this can be problematic as it may capitalize on sample-specific error. Both studies provided support for use of the ABC in ASD samples, but there also were differences in results across these studies which call for further investigation. This study aimed to provide such examination, an important activity given relative lack of instruments validated for measuring behavioral and emotional problems in those with ASD (Hanratty et al., 2015). The aims of this study were to: (1) examine the ABC factor structure in a well-characterized sample of youth with ASD; (2) examine convergent and divergent validity of the ABC; and (3) provide normative data for the ABC as a basis for comparison to the only set of norms in this population that has been described.

Section snippets

Participants

The sample included 470 youth ranging in age from 2 through 14 years inclusive, and their guardians/caregivers. Additional inclusion criteria were that children have a clinical diagnosis of ASD (confirmed by an ADOS-2 score above the Autism Spectrum cut-off, plus meeting diagnostic criteria by DSM-IV-TR/5 per documentation), and that participants be receiving one of the several types of intervention relevant to the larger study’s goals (as noted below). Eight-two percent of the sample was male

Factor analysis

Results of CFAs are comparable to previous studies and lend additional support to the ABC’s factor validity. Results suggest similar fits for the original 5-factor structure and both the 4- and 5-factor structures reported by Brinkley et al. (fit estimates found in Table 1). The 4-factor models found by Brown et al. and Marshburn et al. had poorer fits than these other models. The CFI and GFI did not indicate acceptable fit for any model. The model which emerged from EFA in the Kaat et al.

Discussion

Results of this study provide further support for use of the ABC in youth diagnosed with ASD. Regarding the first goal of examining factor validity, we chose to retain the original factor structure for theoretical and practical reasons. In this sample, fit indices were most supportive of this structure compared to other tested models. Fit indices for the four- and five-factor solutions proposed by Brinkley et al. were similar to indices for the original structure, but the original factor

Conflict of interest

The authors have no conflicts of interest to disclose.

Acknowledgement

This work was supported by the National Institutes of Health (grant number R01 MH 097726).

References (30)

  • T.R. Buck et al.

    Psychiatric comorbidity and medication use in adults with autism spectrum disorder

    Journal of Autism and Developmental Disorders

    (2014)
  • E. Butter et al.

    The Ohio Autism Clinical Impressions Scale (OACIS)

    (2006)
  • D.J. Clarke et al.

    Correlates of maladaptive behavior in individuals with 5p-(cri du chat) syndrome

    Developmental Medicine and Child Neurology

    (1997)
  • D.J. Clarke et al.

    Maladaptive behavior in Prader-Willi syndrome in adult life

    Journal of Intellectual Disability Research

    (1996)
  • J. Cohen

    A power primer

    Psychological Bulletin

    (1992)
  • Cited by (0)

    View full text