Date of Award
5-3-2013
Document Type
Dissertation
Degree Name
Educational Leadership, Ed.D.
First Advisor
George Foldesy
Committee Members
Amany Saleh; David Cox; Joan Henley; Myleea Hill
Call Number
LD 251 .A566d 2013 P34
Abstract
The purpose of this study was to derive one or more single indicators or sets of correlated indicators that could predict a heightened probability for an autism spectrum disorder (ASD) diagnosis, a critical first step in the development of a risk-model. Educators need a risk-model that will identify infants and young children with ASD before their symptoms are fully manifested. Such a model would help educators to fulfill their federal mandate to provide timely identification and early-intervention services, improving long-term outcomes for affected children. Fortunately, advanced statistical modeling techniques and extant early-intervention data suggest a solution: Exploratory latent class factor analysis (ELCFA) and logistic regression analysis can be used to analyze early-intervention case data and elucidate a set of indicators that predict an ASD diagnosis. Recognizing the aforementioned, this investigator used ELCFA to analyze early-intervention data for 30 risk-indicators for children diagnosed with ASD (n = 167). Use of ELCFA elucidated three sets of correlated indicators that appeared to contribute to a tendency for an ASD diagnosis: Maternal Psychiatric Illness, Gestational Maturity, and Advanced Parental Age. The three constructs and 16 remaining single indicators of interest were juxtaposed against ASD and non-ASD case data, forming the empirical foundation for the logistic regression model. Statistical diagnostics confirmed the adequacy and usefulness of the model [Hosmer and Lemeshow Test, ÷2 (8) = 13.23, p = .104; Omnibus Tests of Model Coefficients table, ÷2 (19) = 35.20, p = .013] and indicated that two predictors contributed to the model significantly. The single indicator Family Member with ASD predicted an ASD diagnosis [OR = 5.182; 95% CI (1.997, 13.445); p = .001] and the multiple-indicator construct Maternal Psychiatric Illness predicted an ASD diagnosis [OR = 1.275; 95% CI (1.003, 1.621); p = .047]. Ultimately, the Gestational Maturity and Advanced Parental Age constructs failed to demonstrate statistical significance. The results of this study suggest that extant early-intervention data may be used to derive a model that can predict a future ASD diagnosis, establishing a warrant for refinement of the model.
Rights Management
This work is licensed under a Creative Commons Attribution 4.0 International License.
Recommended Citation
Phillips, Cristy S., "Predictors For an Autism Spectrum Disorder Diagnosis: A Risk Modeling Endeavor" (2013). Student Theses and Dissertations. 798.
https://arch.astate.edu/all-etd/798
