Conflict of interest statement: No conflicts declared.
Measurement equivalence of the autism symptom phenotype in children and youth
Article first published online: 24 AUG 2013
© 2013 The Authors. Journal of Child Psychology and Psychiatry © 2013 Association for Child and Adolescent Mental Health.
Journal of Child Psychology and Psychiatry
Volume 54, Issue 12, pages 1346–1355, December 2013
How to Cite
Duku, E., Szatmari, P., Vaillancourt, T., Georgiades, S., Thompson, A., Liu, X.-Q., Paterson, A. D. and Bennett, T. (2013), Measurement equivalence of the autism symptom phenotype in children and youth. Journal of Child Psychology and Psychiatry, 54: 1346–1355. doi: 10.1111/jcpp.12103
- Issue published online: 16 NOV 2013
- Article first published online: 24 AUG 2013
- Manuscript Accepted: 12 APR 2013
- Autism Speaks
- Health Research Board
- Genome Canada
- Medical Research Council
- Hilibrand Foundation
- Autism Diagnostic Interview-Revised;
- measurement equivalence;
- autism spectrum disorder;
- confirmatory factor analysis;
- structural equation modelling
The Autism Diagnostic Interview-Revised (ADI-R) is a gold standard assessment of Autism Spectrum Disorder (ASD) symptoms and behaviours. A key underlying assumption of studies using the ADI-R is that it measures the same phenotypic constructs across different populations (i.e. males/females, younger/older, verbal/nonverbal). The objectives of this study were to evaluate alternative measurement models for the autism symptom phenotype based on the ADI-R algorithm items and to examine the measurement equivalence of the most parsimonious and best fitting model across subgroups of interest.
Data came from the Autism Genome Project consortium and consisted of 3,628 children aged 4–18 years (84.2% boys and 75% verbal). Twenty-eight algorithm items applicable to both verbal and nonverbal participants were used in the analysis. Measurement equivalence of the autism phenotype was examined using categorical confirmatory factor analysis.
A second-order model resembling the proposed DSM-5 two-factor structure of the phenotype showed good overall fit, but not for all the subgroups. The autism symptom phenotype was best indexed by the first-order, six-factor measurement model proposed by Liu et al. (2011). This model was well fitting and measurement equivalent across subgroups of participants (age, verbal ability and sex).
The autism symptom phenotype is adequately characterized by a six-factor measurement model; this model appears to be measurement equivalent across subgroups of children and youth with ASD that differ in age, sex and verbal ability. The two-factor model provides equally good fit for the sample as a whole, but comparison of these two dimensions between subgroups that might differ in terms of age, sex or verbal ability is challenged by lack of measurement equivalence.