CHAPTER 3: EVIDENCE FOR STIMULUS CONTROL-BASED
3.2 Methods
This study was an extant data analysis of a larger NIH funded study (R01- HD42168; PI: Baranek) investigating the sensory experiences of children with ASD. Recruitment occurred through a state-wide research subject registry, schools, developmental evaluation clinics, social media, university emails, and parent support groups. Families were compensated $20-$75 based on their level of participation in the larger project.
The full sample included 119 children with ASD [M(SD) = 5.08 (2.15)]; see Table 3.1 for a full description of sample characteristics. Participants were given a clinical diagnosis of ASD by an independent licensed psychologist or physician. Diagnosis was confirmed using the Autism Diagnostic Interview-Revised (ADI-R; Rutter, LeCouteur, & Lord, 2003) and Autism Diagnostic Observation Schedule (ADOS; Lord, Rutter, DiLavore, & Risi, 1999). General exclusion criteria included: Fragile X syndrome, uncontrolled seizure disorder, cerebral palsy, mental age <6 months, or failure on vision or hearing screenings.
Assessments
Cognitive measures. To estimate each participant’s non-verbal mental age, one of two assessments of cognitive functioning were used based on the child’s chronological age and developmental level.
Mullen Scales of Early Learning. The Mullen Scales of Early Learning is a standardized, examiner-administered measure of cognitive functioning for children from birth to 68 months of age. An age-equivalent score on the Visual Reception scale (tests of visual discrimination and visual memory) was used as an estimate of non-verbal mental age. This was converted to a developmental quotient, or IQ proxy score (MA/CA x 100).
Stanford-Binet Intelligence Scales. The Stanford-Binet Intelligence Scales, Fifth Edition (Roid, 2003) is a standardized, examiner-administered cognitive assessment for individuals from 2 to 85 years of age. To estimate non-verbal mental age, the score on the Nonverbal IQ domain (a combination of five non-verbal subtests) was used. This was converted to a developmental quotient, or IQ proxy score (MA/CA x 100).
The Tactile Defensiveness and Discrimination Test-Revised. The Tactile
Defensiveness and Discrimination Test-Revised (TDDT-R; Baranek, 2009) is a 15-20 minute child-friendly, semi-structured, play-based assessment of tactile processing. Multiple studies have demonstrated reliability and validity of the TDDT-R for use with children with ASD aged 3-14 years (e.g., Baranek, Foster, & Berkson, 1997; Foss-Feig, Heacock, & Cascio, 2012). Administration is performed by a trained examiner knowledgeable about sensory processing and experienced in working with young children. Items are administered in a series of six subtests, referred to as ‘games’ in order to maintain the child’s interest as much as possible.
The assessment is divided into two sections based on the method of administration. In the first section, the child experiences tactile stimulation (light touch via stickers or a finger
puppet/Q-tip) administered by the examiner to arms, hands, or face, with vision occluded; thus, the child cannot predict the location or timing of the stimulus. Two ‘games’ are included in this section (Fuzzy Puppet Game and Sticker Game) and had eight items each (for a total of sixteen items). After administering each stimulus, the examiner determines 1) whether the child was able to accurately discriminate its location, and 2) a numerical score indicating the child’s level defensiveness. Defensiveness scores were on a scale of zero to three (0 = no negative reaction, 1 = slight reaction, 2 = definite reaction, 3 = severe reaction) and were used in this study as
measured variables predicting the latent construct of hyperresponsiveness to externally- controlled stimuli (HYPER-external).
In the second section, tactile items (sensory materials/toys) are divided into three ‘games’ (Treasure Hunt Game I, Treasure Hunt Game II, Feely and Gooey Games). Each item (e.g., dried pasta, sand, putty) is presented on a table in front of the child and the child is provided an opportunity to 1) manipulate the items, and 2) locate other items intentionally placed within the stimuli. The examiner determines a score for 1) the amount of approach/avoidance the child displays towards the stimulus on a scale from zero to two (0= child engages spontaneously, 1= child engages/continues with prompts or encouragement, 2= child refuses to manipulate item); 2) the time required for the child to locate other items; and 3) presence/absence of sensory seeking behaviors, aversive behaviors, or unusual behaviors (all on a scale of 0= absence or 1=presence). The child’s approach/avoidance score is combined with their aversive score to indicate
defensiveness for these items. This score was used in this study as the measured variable predicting hyperresponsiveness to internally-controlled stimuli (HYPER-internal).
Repetitive Behavior Scale-Revised. The Repetitive Behavior Scale-Revised (RBS-R; Bodfish, Symons, & Lewis, 1999) is a caregiver questionnaire consisting of 43 items separated into six conceptually derived subscales: stereotyped behavior, self-injurious behavior,
compulsive behavior, ritualistic behavior, sameness behavior, and restricted behavior. Items are scored on a four-point scale (0= behavior does not occur, 1= behavior occurs and is a mild problem, 2= behavior occurs and is a moderate problem, and 3= behavior occurs and is a severe problem). This measure has been well-validated for use with children with ASD (Mirenda et al., 2010). An empirically-derived factor analysis resulted in five factors with insistence on sameness and ritualized behaviors loading onto the same factor and other slight difference in loadings of
individual items (Lam & Aman, 2007). This model had good internal consistency and good inter- rater agreement. Therefore, these five factors were used in the analyses for this study.
Child Behavior Checklist. The Child Behavior Checklist (CBCL) is an informant-based measure consisting of the emotional, behavioral, and social assessment of children. Parents completed one of two versions based on the participant’s age: the CBCL for ages 1.5-5 years, and the CBCL for ages 6-18 years (Achenbach & Rescorla, 2001). A subset of items on these two assessments was used to comprise the Anxiety Problems subscale (Achenbach, Dumenci, & Rescorla, 2003; Read et al., 2015). Scales were the same for both versions, indicating the
accuracy of a series of statements about the child’s behavior (0= not true, 1= somewhat or sometimes true, 2= very true or often true). Standardized T-scores from each respective version (eleven items on the CBCL for ages 1.5-5 and six items on the CBCL for ages 6-18 years) were averaged into a single anxiety score, which was used as the outcome variable.
Data Analysis
All measures used in this study were completed concurrently. To answer the first research question, a confirmatory factor analysis (CFA) was conducted on items related to tactile
defensiveness on the TDDT-R. The CFA evaluated associations between item-level scores and underlying latent constructs of hyperresponsiveness due to examiner-controlled tactile items (noted as HYPER-external) and hyperresponsiveness due to tactile items under the child’s control (noted as HYPER-internal). Items were divided a priori into these two categories based on the nature of their administration. The HYPER-internal and HYPER-external categories were independently validated by three professionals with knowledge of the assessment protocol (two individuals with doctorates in related fields and one licensed occupational therapist). Based on their feedback, one item (toy massager from Feely and Gooey Games)was excluded from the
Due to the grouped nature of items administered on the TDDT-R, some variance between these items may be due to the measurement itself and not their shared contribution to the
underlying constructs of HYPER-internal or HYPER-external (Kenny & Kashy, 1992).
Therefore, a bi-factor approach was used in the confirmatory factor analysis. Three measurement factors were included in the model based on the grouping of items on the Fuzzy Puppet Game,
Sticker Game, and Feely and Gooey Games. A factor for the group of items included in the
Treasure Hunt Games I and II was not included, as there were not enough items in this category. The three measurement factors were fixed to be uncorrelated with all other factors in the model in order to remove this shared variance from the latent constructs of interest. For a diagram representation, see Figure 3.1.
Data were entered into Mplus (Muthén & Muthén, 2010) and error checked prior to analysis. A low number of data were missing, which was addressed via list-wise deletions. The model was standardized (such that factor means were fixed to zero and factor variances were fixed to one) and the hypothesized loadings between factor correlations were estimated freely. The model was analyzed using the weighted least squares estimator (i.e., WLSMV, in Mplus). To answer the second and third research questions, a subset of the total sample that had scores on either version of the CBCL was used (n=54). A series of regression analyses were conducted in Mplus to estimate associations between HYPER-internal and HYPER-external with multiple dependent variables: 1) the five empirically-derived categories of RRBs (i.e., insistence on sameness/ritualized behavior, stereotyped behavior, self-injurious behavior, compulsive behavior, and restricted behavior) and 2) an overall anxiety score from the CBCL. Covariates were entered in the regression models to control for the effects of age, race, gender, and maternal education. As a follow-up, any outcomes with a difference in statistical significance between
HYPER-internal and HYPER-external were evaluated using the Wald Test of Parameter Constraints to determine whether this difference itself was also statistically significant.
The model with anxiety score as the dependent variable was used to address the final research question. A moderation term was introduced by independently multiplying each of the five RRB factors by HYPER-internal (one series of analyses) and HYPER-external (a separate series of analyses). The same set of covariates was also included.