Personality correlates of the broader autism phenotype as assessed by the Autism Spectrum Quotient (AQ)

https://doi.org/10.1016/j.paid.2004.04.022Get rights and content

Abstract

The main objective of this study was to characterise the five-factor personality model profile of the broader autism phenotype as assessed by the Autism Spectrum Quotient (AQ). Comparison data were also obtained for a brief Asperger syndrome screening measure. The psychometric properties and factor structure of the AQ were also assessed and group differences in AQ scores examined. The AQ and personality scale were completed by 201 undergraduates. A second group of 136 adults completed the personality scale and the Asperger screening measure. Examination of correlations showed that high scores on both `autism' measures were associated with high neuroticism and low extraversion and agreeableness. Three of the five proposed sub-scales of the AQ emerged from the factor analysis. The AQ and sub-scales had satisfactory or near-satisfactory reliabilities. Group differences were generally as expected: Males had higher AQ scores than females, `hard' science students had higher scores than other students and students with parent(s) in a scientific occupation had higher scores.

Introduction

The clinical view of autism requires an individual to meet specified criteria in order to receive a positive diagnosis (American Psychiatric Association, 1994). There is however considerable evidence for the existence of a broader autism phenotype; within this framework clinical cases can be regarded as having extreme scores on an underlying continuous dimension. Studies of the non-autistic relatives of autistic probands have indeed shown that a phenotype with similarities to that of autism cases can be found in this group. Results include findings of social, communication and language difficulties in relatives (Piven, Palmer, Jacobi, Childress, & Arndt, 1997; Piven, Palmer, Landa, et al., 1997). Family studies have also shown differences in personality scores compared to the general population, with relatives being found to have higher than average scores on the traits of aloof, tactless, undemonstrative, hypersensitive, anxious, impulsive, shy, irritable, eccentric and rigid (Murphy et al., 2000; Piven, Palmer, Jacobi, et al., 1997; Piven et al., 1994). A related line of research has examined the familial patterning of vocational interests, with associations to relatives having occupations in engineering, physics or mathematics being found (Baron-Cohen, 2002; Baron-Cohen et al., 1998; Baron-Cohen, Wheelwright, Stott, Bolton, & Goodyer, 1997). The possibility of a characteristic cognitive style (weak central coherence) acting as a marker for the broader phenotype has also been proposed (Happé, 1999). Observations on the familiality of a range of traits linked to autism provide indirect evidence for the existence of a genetic liability for this condition; these findings are convergent with results from twin and family studies (Lauritsen & Ewald, 2001; Rutter, 2000) which provide evidence for a genetic contribution to autism and Asperger syndrome.

The dimensional approach to autistic traits has led to the development of the Autism Spectrum Quotient (AQ; Baron-Cohen, Wheelwright, Skinner, Martin, & Clubley, 2001). The AQ is a 50-item questionnaire which is designed to assess autistic spectrum traits in the general population. Baron-Cohen et al. (2001) reported a series of validity studies on this instrument, showing for example that Asperger syndrome and high-functioning autism cases have higher scores than controls on the AQ, and that science students score higher than arts and social science students. Two aspects of the AQ however merit further study: its associations with major personality dimensions and its factor structure.

It is of interest to assess associations between the AQ and major personality dimensions, because aligning this instrument with personality descriptions widely used by individual difference researchers will assist in the characterisation of the broader autism phenotype. There is a well-established consensus among personality researchers that the five-factor model of personality provides a meaningful account of the dimensions of human personality variation (e.g. McCrae & John, 1992). In a clinical context considerable progress has been made in relating personality disorders to the theoretical framework of the five-factor model (e.g. O'Connor & Dyce, 2001; Schroeder, Wormworth, & Livesley, 1992), and similarly establishing the five factor model correlates of the broader autism phenotype is of considerable interest.

The AQ was designed to assess five different areas of functioning: social skill, attention switching, attention to detail, communication and imagination. Factor analysis can be used to test whether these five groupings provide the most appropriate reduction of the item set, or whether a different structure exists. Factor-analytic studies have on occasion (e.g. Deary, Bedford, & Fowkes, 1995) shown that instruments derived from clinical criteria may contain a different number of underlying dimensions than originally proposed, so it is of interest to assess the AQ in this context.

The objectives of the present study were

  • (1)

    to examine the correlations between AQ full-scale and sub-scale scores and personality traits, allowing the five-factor model personality profile of the broader autism phenotype to be characterised. On the basis of previous work (Murphy et al., 2000; Piven et al., 1994; Piven, Palmer, Landa, et al., 1997) it was hypothesised that AQ scores would be associated with high scores on Neuroticism and possibly Conscientiousness and low scores on Extraversion. The basis of these predictions was the associations of aloofness and shyness with low Extraversion, anxiety, hypersensitivity and irritability with Neuroticism and, more tentatively, rigidity with Conscientiousness;

  • (2)

    to examine the factor structure of the AQ using data from a non-clinical student population;

  • (3)

    to examine group differences in AQ scores. Here it was predicted that

    • (i)

      in agreement with previous work on non-clinical groups and with the large male:female ratio amongst autism cases (Baron-Cohen, 2002; Baron-Cohen et al., 2001), males would score higher than females;

    • (ii)

      as found by Baron-Cohen et al. (2001), science students would have higher scores;

    • (iii)

      students from a `scientific' family background (as indexed by parental occupation) would have higher scores. This result would be expected given the findings on familial occupational associations to autism (Baron-Cohen et al., 1997, Baron-Cohen et al., 1998);

  • (4)

    to compare the findings from the student group with results from a short Asperger syndrome screening instrument which provides coverage of the same domains of functioning as the AQ (Else & Baron-Cohen, 2001). Results were available for this instrument for a group of older adults. It was hypothesised that this scale would show similar personality correlations to the AQ full-scale score in the student sample, that males would score higher than females and that participants reporting scientific occupations would score higher than non-scientists.

Section snippets

Participants

The majority of the data presented in this paper involved a sample of 201 undergraduates, 79 males and 122 females. The mean age of the group was 20.9 years, standard deviation 2.1 years. Some results were also available for a group of 136 adult volunteer panel members who completed measures relevant to this study as part of a larger survey. This group consisted of 103 females and 33 males, mean age 53.9 years, standard deviation 14.8 years.

The Autism Spectrum Quotient (AQ)

This 50-item questionnaire (Baron-Cohen et al., 2001)

Results

For the student sample degree subject was initially coded into three groups: non-science (e.g. divinity, law), biological science (e.g. marine biology, genetics, psychology) and `hard' science/maths (subjects with substantial mathematical content, e.g. physics, engineering). Father's and mother's occupations were coded as no mathematical or scientific content (including cases where it was not possible to assess mathematical/science content from the information given) or substantial mathematical

Discussion

This study examined the psychometric properties of the AQ (Baron-Cohen et al., 2001) and group differences in scores on this instrument and an Asperger screening instrument (Else & Baron-Cohen, 2001) in a non-clinical population. The personality characteristics of the broader autism phenotype were examined using both the AQ and the Asperger screening instrument.

The AQ was found to have good psychometric properties, with the full-scale score and one sub-scale showing satisfactory internal

Acknowledgements

Professor Simon Baron-Cohen kindly supplied a pre-publication copy of the AQ to the author.

References (19)

  • S. Baron-Cohen

    The extreme male brain theory of autism

    Trends in Cognitive Sciences

    (2002)
  • I.J. Deary et al.

    The Personality Deviance Scales––their development, associations, factor structure and restructuring

    Personality and Individual Differences

    (1995)
  • F. Happé

    Autism: Cognitive deficit or cognitive style?

    Trends in Cognitive Sciences

    (1999)
  • American Psychiatric Association (1994). DSM IV. Diagnostic and statistical manual of mental disorders. Washington, DC:...
  • S. Baron-Cohen et al.

    Autism occurs more often in the families of physicists, engineers and mathematicians

    Autism

    (1998)
  • S. Baron-Cohen et al.

    Is there a link between engineering and autism?

    Autism

    (1997)
  • S. Baron-Cohen et al.

    The Autism-Spectrum Quotient (AQ): Evidence from Asperger syndrome/high-functioning autism, males and females, scientists and mathematicians

    Journal of Autism and Developmental Disorders

    (2001)
  • P.T. Costa et al.

    Revised NEO personality inventory and NEO five factor inventory professional manual

    (1992)
  • L. Else et al.

    In a different world

    New Scientist

    (2001)
There are more references available in the full text version of this article.

Cited by (328)

View all citing articles on Scopus
View full text