Name: Sidni Justus

Dissertation Proposal Defense Meeting

Date: Tuesday, July 3, 2018

Time: 11:15am

Location: J.S. Coon bldg. room 150

Advisor: 

Jenny Singleton, Ph.D. (Georgia Tech)

Dissertation Committee Members:

Christopher Hertzog, Ph.D. (Georgia Tech)

Lizanne DeStefano, Ph.D. (Georgia Tech)

Christopher Stanzione, Ph.D. (Georgia Tech)

Agata Rozga, Ph.D. (Georgia Tech)

Title:  Early predictors of autism: Exploring variable-centered and person-centered analysis approaches

Summary: Autism Spectrum Disorder (ASD) is now considered one of the most common developmental disabilities (Newschaffer et al., 2007). Over the past 20+ years, researchers have worked towards identifying early behavioral or physiological predictors of ASD so that early treatment and intervention can be implemented. These efforts include the development of rapid, behavior based screeners (e.g., Rapid-ABC by Ousley et al., 2013) to supplement or replace the commonly used parent-report methods (e.g., Modified Checklist for Autism in Toddlers) and lengthy behavioral and interview assessments (Autism Diagnostic Observation Schedule; Autism Diagnostic Interview) that are considered gold-standard for ASD diagnosis. The present study explores which means of measuring early infant and toddler social communication and language behavior (i.e., parent-report data vs. Rapid-ABC screener) are the most predictive of autism spectrum disorder diagnosis in childhood or early adolescence using both variable-centered and person-centered analysis approaches. Parents of children who previously participated in a study evaluating early infant ASD-risk behaviors when their children were 15-34 months of age will now participate in a follow-up interview about their child’s social communicative development and medical updates over the last 3-7 years. This follow-up measure will be used to create a longitudinal dataset that will be used to first compare composite scores on the behavioral assessment (Rapid-ABC) and parent-report screeners (M-CHAT; CBCL; CSBS-DP) and how they each predict later clinical outcomes. I hypothesize that both types of measures will predict ASD outcome, but that one or more of these measures will predict only part(s) of the spectrum of ASD. Therefore, I hypothesize that latent class analysis will reveal meaningful risk subgroups that further relate to diagnostic outcomes.