What range of trait levels can the the Autism Spectrum Quotient (AQ) measure reliably? An item response theory analysis

Aja Louise Murray, Tom Booth, Karen McKenzie, Renate Kuenssberg

Research output: Contribution to journalArticlepeer-review

29 Citations (Scopus)
2 Downloads (Pure)

Abstract

It has previously been noted that inventories measuring traits that originated in a psychopathological paradigm can often reliably measure only a very narrow range of trait levels that are near and above clinical cutoffs. Much recent work has, however, suggested that autism spectrum disorder traits are on a continuum of severity that extends well into the nonclinical range. This implies a need for inventories that can capture individual differences in autistic traits from very high levels all the way to the opposite end of the continuum. The Autism-Spectrum Quotient (AQ) was developed based on a closely related rationale, but there has, to date, been no direct test of the range of trait levels that the AQ can reliably measure. To assess this, we fit a bifactor item response theory model to the AQ. Results suggested that AQ measures moderately low to moderately high levels of a general autistic trait with good measurement precision. The reliable range of measurement was significantly improved by scoring the instrument using its 4-point response scale, rather than dichotomizing responses. These results support the use of the AQ in nonclinical samples, but suggest that items measuring very low and very high levels of autistic traits would be beneficial additions to the inventory.
Original languageEnglish
Pages (from-to)673-683
JournalPsychological Assessment
Volume28
Issue number6
Early online date24 Aug 2015
DOIs
Publication statusPublished - Jun 2016

Fingerprint

Dive into the research topics of 'What range of trait levels can the the Autism Spectrum Quotient (AQ) measure reliably? An item response theory analysis'. Together they form a unique fingerprint.

Cite this