On the Misapplication of the Broad Autism Phenotype Questionnaire in a Study of Autism

Abstract

Nishiyama et al. (2013) recently examined the reliability and construct validity of the Broad Autism Phenotype Questionnaire (BAPQ) and three other instruments with the goal of measuring quantitative autistic traits in 64 individuals with autism spectrum disorder (ASD) and two population samples. As the authors of the BAPQ (Hurley et al. 2007; Sasson et al. 2013),we write to point out that although the study appeared to be carefully and thoughtfully conducted in almost every way, the BAPQ was not designed to be administered to individuals with ASD, nor measure the defining characteristics of autism. Thus, we question the appropriateness of the validation process employed by the authors as well as their interpretation of the study’s results. In contrast to the other questionnaires examined in Nishiyama et al. (2013), the BAPQ does not measure autistic traits but rather is, to our knowledge, the only instrument currently available in the field that was specifically designed to measure the broad autism phenotype. As described in our two papers that detail the development of the instrument, examine its psychometric properties and discuss its extensive validation (Hurley et al. 2007; Sasson et al. 2013), the BAPQ was designed on the basis of the senior author’s experience conducting direct assessments of the parents of autistic individuals, who themselves specifically did not have an ASD (Piven et al. 1997). This distinction is important, and results in three primary ways in which the constructs underlying the BAP and ASD differ. First, while the BAP is often referred to as representing characteristics that are milder but qualitatively similar to ASD, there are also areas in which the two differ qualitatively. For instance, the BAPQ purports to measure the personality construct ‘rigidity’, defined primarily as a preference for routine. An example might include a father who prefers to go to the same restaurant every Saturday night or gets uncomfortable when asked to vary his route to work by two blocks in order to mail a letter for his wife. This differs vastly from characteristics like immediate and delayed echolalia, which is an example of a repetitive and stereotyped characteristic specific to ASD that is not addressed in the BAPQ. Second, there is a difference between the level of severity in the construct of the BAP as assessed by the BAPQ and the level of severity inherent in a diagnosis of ASD. Specifically the BAP is not associated with clinical impairment in our studies (Piven 2002), whereas impairment is an implicit aspect of the diagnosis of ASD. Finally, while the BAPQ is both a selfand informant-report questionnaire, the validity of its use as a self-report questionnaire in individuals with ASD, who by definition have impairments in language and appreciation of social nuances, is suspect. In summary, we believe that Nishiyama et al. (2013) inappropriately evaluated the BAPQ by assessing its ability to differentiate adults with ASD from a population sample, a task it was not designed to address. Although the BAPQ performed well in this context and we applaud the conceptualization of the characteristics inherent in the genetic liability for autism as quantitative traits, it is critical to keep in mind that the measurement of constructs at different ends of such a dimension are not equivalent. To us, this invalidates the results of this paper with respect to analyses involving the BAPQ. J. Piven (&) University of North Carolina, Chapel Hill, NC, USA e-mail: joe_piven@med.unc.edu