Journal of Autism and Developmental Disorders

https://doi.org/10.1007/s10803-017-3424-6

BRIEF REPORT

Brief Report: Factor Analysis of the Brazilian Version of the Adult

Autism Spectrum Quotient
Julia Horta Tabosa do Egito1,2 · Gabrielle Martins Ribeiro Ferreira1 · Manoela Izidoro Gonçalves1 ·
Ana Alexandra Caldas Osório1,2,3

© Springer Science+Business Media, LLC, part of Springer Nature 2017

Abstract
This study examined the factor structure of the Brazilian version of the Autism-Spectrum Quotient. This is a self-report
questionnaire for continuous and quantitative assessment of autistic spectrum traits in adults. Confirmatory factor analysis
was performed on the five-factor model (social skill, attention switching, attention to detail, communication and imagination)
proposed by the original authors, support not being found for this model in our sample. An exploratory factor analysis was
then performed that resulted in an alternative three-factor model (social skills, details/patterns and imagination). Confirma-
tory factor analysis of the latter model revealed adequate psychometric indexes. The Brazilian version of the AQ was shown
to be an adequate instrument for the evaluation of signs compatible with the autism spectrum in adults.

Keywords  Autism · Autism Spectrum Quotient · Factor analysis · Validation

Introduction in the typical population (Constantino and Todd 2003; Rut-

ter 2011), the so-called “broader phenotype” (Bailey et al.
According to the American Psychiatric Association (2013), 1995). People with autistic features may be able to adjust
Autistic Spectrum Disorder (ASD) is characterized by to the environment in which they live, without presenting
impairments in communication and social interaction, as symptoms with the necessary severity for a diagnosis of
well as by restricted and repetitive patterns of behavior, ASD.
activities and interests. As the name implies, the signs and In order to quantify the autistic traits in typically-devel-
symptoms of each individual with ASD are distributed along oping adults with preserved intelligence, Baron-Cohen et al.
a continuum of lesser or greater severity. This dimensional (2001) developed the Autism Spectrum Quotient (AQ), a
approach, alongside the categorical (presence vs. absence), self-report questionnaire with 50 statements that explores
thus allows the evaluation of the levels of function or dys- five areas of the functioning of the individual: social skill,
function of individuals (Volkmar & MacPartland 2014). attention to detail, attention switching, communication and
This approach also supports the view that milder versions of imagination. According to Baron-Cohen et al. (2001), it
the characteristics observed in ASD are normally distributed is an instrument capable of quantifying where the person
is placed in the continuum that includes ASD and typical
development. The AQ is not intended to be a diagnostic tool,
* Ana Alexandra Caldas Osório and the cutoff score proposed by the authors (32 + points)
[email protected] indicates that the individual has the potential for a diagnosis
1
Social and Cognitive Neuroscience Laboratory, Mackenzie if an in-depth investigation is carried out. Given the potential
Presbyterian University, São Paulo, Brazil demonstrated by the instrument for the screening of autistic
2
Developmental Disorders Graduate Program, Mackenzie traits in an English sample, researchers around the world
Presbyterian University, São Paulo, Brazil began to focus on its validation for other cultural contexts.
3
Social and Cognitive Neuroscience Laboratory For example, in Japan, by Wakabayashi et al. (2006); in
and Developmental Disorders Program, Center for Health Scotland, by Stewart & Austin (2009); in a French–Cana-
and Biological Science, Mackenzie Presbyterian University, dian sample by Lepage et al. (2009); and in Australia by
Rua Piauí, 181, 10 floor, Consolação, São Paulo 01241‑001, Broadbent et al. (2013).
Brazil

13
Vol.:(0123456789)
 Journal of Autism and Developmental Disorders

However, the five-factor structure proposed by its original was initially conducted by a researcher fluent in English,
authors has not found consistent support in the various inter- and then back-translated into English by a second researcher
national validation studies. Austin (2005) proposed a three- fluent in English. Both versions have been compared, and
factor alternative solution, comprising 26 items. The three any inconsistencies have been resolved (the AQ version in
domains produced were identified as: social skill, details/ Brazilian Portuguese is available at https://www.autismre-
patterns and communication/mindreading. Hurst, Mitchell searchcentre.com/arc_tests). The original version of the AQ
et al. (2007), following this alternative three-factor model, comprises 50 statements exploring five areas of functioning
applied the AQ in a sample of American college students, of the individual assessed, each with ten items: social skill
also finding a better consistency for this structure than for (items 1, 11, 13, 15, 22, 36, 44, 45, 47 and 48), attention
the original. In a study by Hoekstra et al. (2008) with Dutch switching (items 2, 4, 10, 16, 25, 32, 34, 37, 43 and 46),
university students, the authors reported better psychometric attention to detail (items 5, 6, 9, 12, 19, 23, 28, 29, 30 and
adequacy with a two-factor model, as they found that four of 49), communication (items 7, 17, 18, 26, 27, 31, 33, 35, 38
the originally proposed domains presented high correlation and 39) and Imagination (items 3, 8, 14, 20, 21, 24, 40, 41,
values among themselves, and when grouped improved the 42 and 50). The total AQ score and those of its domains were
model. Thus, this study proposed two factors: social interac- based on the original Likert scale of four scores (1 = “defi-
tion (social interaction (composed of social skill, attention nitely agree”, 2 = “slightly agree”, 3 = “slightly disagree”,
switching, communication and imagination) and attention 4 = “definitely disagree”). Reverse statements were those in
to detail. which the “agree” option actually characterized the autism
Given the usefulness and importance of a self-report phenotype, requiring recoding (“I definitely agree” = 4,
instrument for assessing autistic traits in adults in the gen- “definitely disagree” = 1). This procedure was required for
eral population, the present study aimed to evaluate the fac- items: 2, 4, 5, 6, 7, 9, 12, 13, 16, 18, 19, 20, 21, 22, 23, 26,
tor structure of the Brazilian version of the AQ applied in 33, 35, 39, 41, 42, 43, 45, and 46. The total score had a pos-
a non-clinical sample of university students. Since previ- sible range of 50–200 points obtained from the sum of the
ous studies evaluating the psychometric properties of AQ items (recoded as necessary), the higher the score, the more
(Austin 2005; Hoekstra et al. 2008) have suggested alterna- autistic traits the individual presented. The application of the
tive models to the original five-factor model proposed by instrument took about 20 min.
Baron-Cohen et al. (2001), factor analyses (confirmatory and
exploratory) were performed to evaluate the specificities of Procedure
the data distribution in the present sample.
To obtain a sample as heterogeneous as possible, stu-
dents from various courses at a private university in São
Method Paulo, Brazil, were invited to participate in the research.
Data collection was done by three researchers who applied
Participants the AQ individually to students recruited on campus. The
free and informed consent was obtained from all individ-
Two hundred and twenty five college students (M = 22.76 ual participants included in the study, and the study was
years, SD = 4.94, age range = 18–55) from the fields of natu- approved by the Institutional Ethics Committee (CAAE:
ral sciences (31.6%), social sciences (45.3%) and humani- 25874414.0.0000.0084).
ties (23.1%) participated. The classification criteria for the
different degrees were based on Hoekstra et al. 2008, in Data Analyses
which students from Social Sciences included degrees such
as psychology, education and marketing; Sciences included SPSS for Windows, version 21.0 was used for descriptive
mathematics, physics, computer science and engineering; and inferential data analyses. Gender differences were tested
and Humanities included history and law. The sample was using independent samples t-tests and differences between
male (47.1%) and female (52.9%), and of a non-probabilistic students from different degrees were explored with a multi-
type, since only those who volunteered to take part in the variate analysis of variance (MANOVA). We also reported
study answered the questionnaire. Participants received effect sizes for the t-tests (Cohen´s d: small, from 0.20;
school credits for participation. medium from 0.50; and large from 0.80) and MANOVA (ƞ:
small, from 0.01; medium from 0.06; and large from 0.14).
Materials The same software was used to conduct the exploratory fac-
tor analysis (EFA). In the principal component analysis,
The translation of the AQ took place after receiving the per- varimax rotation was used, aiming to explain the correlation
mission of the author, Professor Simon Baron-Cohen, and between the data with the fewest possible factors.

13
Journal of Autism and Developmental Disorders

The statistical package AMOS GRAPHICS 21.0 was correlated with the subscales social skill (r = .29), commu-
used to perform the confirmatory factor analyses (CFA) on nication (r = .34), and imagination (r = .18). Additionally,
the original five-factor model (proposed by Baron-Cohen Communication was significantly correlated with social
et al. 2001) and our proposed three-factor model (Table 1). skill (r = .54), attention to detail (r = .15), and imagination
The CFA allowed the evaluation of the fit of the proposed (r = .22). This pattern of associations reinforced our decision
models based on the following indicators: the ratio of Chi to retain the three-factor solution. Those items in the EFA
square to degrees of freedom (χ2/gl), to assess the adequacy with loads above 0.30 were selected (Hurst et al. 2007), and
of the fit (the lower the value, the better the fit, values up to the resulting 30 items were grouped into the following theo-
3 being acceptable Tolentino et al. (2015); the Comparative retical categories: social skills (SS): 1, 11, 13, 15, 17, 22, 26,
Fit Index (CFI), to compare the model estimated to a null 34, 38, 44, 47; Details/patterns (DP): 6, 9, 16, 19, 23, 29, 39,
value (satisfactory values are close to one) (Hair et al. 2005); 41, 46; imagination (I): 3, 8, 10, 14, 20, 21, 27, 31, 40, 50.
the Root-Mean-Square Error of Approximation (RMSEA), In the confirmatory factor analysis (CFA), it was found
which assesses the model fit (ideal values 0.05–0.08) (Hair that five additional items presented below acceptable loads
et al. 2005); the Tucker–Lewis Index (TLI), which may have (SS: item 1 = 0.29, DP: item 46 = 0.13, I: item 27 = 0.30, item
a different interpretation of the pattern applied to similar 31 = 0.24, item 10 = 0.21) and therefore were excluded in the
indexes, in that the TLI may exceed the desired range (0–1, new test (see Fig. 1). We found that the Kaiser–Meyer–Olkin
with optimal value above 0.95) (Bentler 1990); and finally, (KMO) index of the 25 item three-factor solution showed
the Consistent Akaike Information Criterion (CAIC) and the best result (0.76, comparing with 0.65 of the 5-factor
Expected Cross-Validation Index (ECVI), indicators to solution), as well as an improvement in all other indicators.
assess the adjustment of a model in comparison to others, Lower values for CAIC and ECVI demonstrated that this
which lower values represent a better fit (Bilich et al. 2006). model showed the best fit solution for our Brazilian sample.
Regarding analyses of factor structure, the following
steps were taken: Firstly, we conducted a CFA to explore Gender and Degree Comparisons
the structure of the original five-factor model (Baron-Cohen
et al. 2001). However, given the poor fit we conducted an Independent samples t-tests revealed that men and women
EFA which ultimately yielded a novel three-factor solution. differed significantly on all factors, except SS (p = .163).
In turn, this solution was submitted to a CFA, which evi- Males reported higher interest in DP, more difficulties in I
denced indices of good fit. and overall more autistic traits (Table 2).
Regarding the differences in AQ scores in students
studying different degrees, a MANOVA showed a sig-
Results nificant effect of degree on AQ scores, F(6, 442) = 11.26,
p < .001. Separate univariate ANOVAS revealed group
All questionnaires collected were included in the analysis, differences in DP, F(2, 222) = 25.58, p < .001, ƞ = 0.187,
as no total AQ values higher than four standard deviations I, F(2,222) = 10.03, p < .001, ƞ = 0.083, and overall score,
above the mean were observed. The observed asymmetry F(2,222) = 15.28, p < .001, ƞ = 0.121. No significant differ-
and kurtosis values for all factors and total score indicate ences were found in SS, F(2,222) = 0.46, p = .63, ƞ = 0.004.
that the data from the final version of the AQ score has an Overall, the Sciences students attained the highest scores
approximately normal distribution (Kline 1994). (more ASD-like traits), followed by the Humanities students
and the Social Sciences students (Table 3).
Factor Structure of the AQ

The CFA conducted indicated that the Baron-Cohen five- Discussion

factor model (2001) presented statistical indicators that do
not corroborate the adequacy of the original structure of the Autism Spectrum Disorder (ASD) is characterized by a set
AQ scale for the Brazilian sample. Psychometric indicators of behavioral and socio-emotional changes that arise early in
are not considered acceptable according to the literature, development, with consequences and costs for the affected
with χ2/gl being high and TLI/CFI below ideal. individuals, their families and society as a whole. The cur-
Because of the previous result, we performed an EFA. An rent view considers ASD to be a spectrum with varying
analysis of the inflexion point of the scree plot was sugges- levels of symptom severity, and it is increasingly accepted
tive of three-factors (however, up to 17 factors had eigenval- that the milder end of this spectrum will encompass a non-
ues above 1). Therefore, we explored the intercorrelations clinical population (Bailey et al. 1995; Constantino and
between the five factors originally proposed to investigate Todd 2003; Rutter 2011). It is therefore possible that adults
possible overlaps. Attention Switching was significantly with preserved intelligence and considered to have typical

13
 Journal of Autism and Developmental Disorders

Table 1  Factor structure of the original five-factor model and our

three-factor model of the AQ

5-Factor model (50 items) 3-Factor model (25 items)

χ2/gl 2.13 1.74

TLI 0.36 0.79
CFI 0.38 0.81
RMSEA 0.07 0.06
CAIC 3146.13 799.13
ECVI 12.07 (11.45–12.73) 2.58 (2.33–2.87)

Best fit model in bold

present adequate indicators. Therefore an EFA, followed by

a new CFA were carried out, which indicated support for a
three-factor model with 25 items, grouped into three sub-
scales named: social skill (SS), details/patterns (DP) and
imagination (I). The titles were given based on the names
initially proposed by Baron-Cohen and collaborators, as well
as by other authors who carried out international validation
studies. In what concerns the items belonging to each of
the three factors of the proposed Brazilian version, several
aspects are noteworthy. Firstly, the factor SS was the first to
emerge and shared a majority of items (6 out of 10—items
11, 13, 15, 22, 44 and 47) with the original factor by Baron-
Cohen and colleagues. Indeed, several previous studies
conducted in both Western and Eastern cultures found that
a social factor accounted for a large proportion of the vari-
ance (e.g., Austin 2005 UK; Hoekstra et al. 2008 The Neth-
erlands; Freeth et al. 2013 UK, Malaysia and India). Fur-
thermore, our SS factor also incorporated three items from
Baron-Cohen’s communication scale (items 17, 26 and 38)
and one from attention to detail (item 34). The fact that our
social factor congregated mostly items concerning SS and
communication is in line with the most recent diagnostic cri-
teria for ASD proposed on the latest version of the diagnos-
tic and statistical manual of mental disorders (DSM). While
in the DSM-IV-TR, persistent deficits in social interaction
and social communication constituted separate criteria, the
Fig. 1  CFA of alternative three-factor model (25 items) for the total DSM-5 combined them into a single criterion, reflecting an
sample. Values and weights of the standard regression per item in the understanding that the social deficits evidenced by people
AQ; SS Social Skills, DP Details/Patterns, I Imagination with ASD involve atypical interaction and communication.
As for our DP factor, again most of the items are shared
with the original scale’s attention to detail (5 out of 8—
development (including being well-adapted to diverse social items 6, 9, 19, 23 and 29). The remaining three items belong
and professional situations), present characteristics compat- to attention switching (item 16 —“I tend to have very strong
ible with an ASD. In order to quantify autistic traits in these interests, which I get upset about if I can’t pursue”), commu-
individuals, Baron-Cohen et al. (2001) developed the AQ, nication (item 39—“People often tell me that I keep going
which in the present study was validated for the Brazilian on and on about the same thing”) and imagination (item
context in a non-clinical sample of university students. 41—“I like to collect information about categories of things
In the process of analyzing the factor structure of the AQ, (e.g. types of car, types of bird, types of train, types of plant,
the CFA performed showed that the five-factor model (with etc.)”). These three specific items do seem to have some
50 items) proposed by Baron-Cohen et al. (2001) did not affinity to the DSM-5 diagnostic criterion of restricted and

13
Journal of Autism and Developmental Disorders

Table 2  Means and standard Male (n = 106) Female (n = 119) t (223) p Cohen’s d

deviations of the AQ total score mean (SD) mean (SD)
and subscales comparing gender
SS 19.43 (5.05) 18.47 (5.25) 1.40 0.163 0.186
DP 19.85 (4.29) 18.46 (4.29) 2.44 0.016 0.303
I 14.24 (3.38) 13.07 (3.54) 2.55 0.012 0.338
Total score (min.: 53.54 (7.60) 50.00 (7.03) 3.63 < 0.001 0.483
25-max.: 96)

Bold values indicate p < 0.05

Table 3  Means and standard deviations of the AQ total score and subscales comparing the different degrees
Total sample (N = 225) Sciences—S (n = 71) Humanities—H Social sciences—SS Significant between-
mean (SD) mean (SD) (n = 52) mean (SD) (n = 102) mean (SD) group differences
(p-value)

SS 18.92 (5.17) 18.69 (5.10) 18.54 (4.61) 19.28 (5.50) S = H = SS (p = n.s.)
DP 19.12 (4.33) 21.58 (4.34) 19.42 (3.19) 17.25 (3.96) S > H (p = .003)
H > SS (p = .001)
S > SS (p < .001)
I 13.62(3.51) 14.87 (3.71) 13.96 (3.29) 12.58 (3.17) S > SS (p < .001)
H > SS (p = .017)
Total score 51.67 (7.50) 55.14 (6.82) 51.92 (6.06) 49.12 (7.66) S > H (p = .013)
H > SS (p = .021)
S > SS (p < .001)

repetitive patterns of behavior, interests or activities (hence failed to include attention switching in their AQ structure,
the name DP ). Finally, in what concerns our I factor, all of suggesting that this factor may be particularly susceptible to
the items are shared with the original scale (7 out of 7— contextual influences (e.g., Austin 2005; Stewart and Austin
items 3, 8, 14, 20, 21, 40, 50). 2009).
Overall, comparing the obtained factor structure with the Gender differences were consistent with previous reports
original structure, I is the most similar factor, followed by SS (e.g., Austin 2005; Baron-Cohen et al. 2001; Hoekstra et al.
(integrating social skill and communication) and DP (inte- 2008) as males scored higher than females on the overall
grating mostly details). It is noteworthy that only two items AQ as well as in the DP and I factors. Interestingly, no
from the original version’s attention switching were retained significant gender differences were found in the SS factor.
(item 16 belongs to our DP factor and item 34 to our SS fac- Furthermore, we also replicated previous reports of sig-
tor). Therefore, difficulties in switching one’s focus of atten- nificant differences depending on the field of study, as the
tion do not seem to characterize ASD-like traits in our Bra- Sciences students attained significantly higher scores that
zilian sample. Cultural differences in ASD symptom profiles the Humanities and Social Sciences students (Austin 2005;
have been widely reported in the literature (e.g., Freeth et al. Baron-Cohen et al. 2001; Hoekstra et al. 2008). Similar to
2013; Ashley et al. 2017; Matson et al. 2017). In this specific Hoekstra et al. (2008), our Social Sciences group scored sig-
case, we can see that many items from attention switching nificantly lower than the Humanities group on some factors
regard attentional difficulties such as problems in perform- (DP, I) and in total AQ. As reported in the Dutch study, our
ing several tasks simultaneously (item 32) or in resuming Social Sciences group was mainly composed of Psychology
an activity when interrupted (item 37), which also occur in students, who may be more prone to engage in and enjoy
psychological conditions other than ASD, such as depression social interactions.
or anxiety. Interestingly, cross-cultural data on the Big Five Some limitations of this study should be highlighted
traits indicate that South American countries (these analyses in relation to the other studies in the area. One of these
included data from Brazil) were among the cultures with concerns the fact that the data were not collected from
the highest levels of Neuroticism—a tendency for feeling individuals with ASD, therefore, whether the Brazil-
anxious, depressed or irritable (Schmitt et al. 2007). Nev- ian version of this instrument was able to discriminate
ertheless, it is important to note that other studies have also between individuals with clinical levels of autistic traits

13
 Journal of Autism and Developmental Disorders

and individuals with typical development was not tested. References

Given this, it is possible that if the data had been collected
from people with ASD, we may have found higher psy- American Psychiatric Association (2013). Diagnostic and statistical
chometric indicators, as found by Broadbent et al. (2013), manual of mental disorders; (DSM-V). Washington, DC: APA.
Ashley, J., Harrison, A. J., Long, K. A., Tommet, D. C., & Jones,
and Baron-Cohen et al. (2001). However, it is important to R. N. (2017). Examining the role of race, ethnicity, and gender
consider that the AQ aims at the rapid quantitative assess- on social and behavioral ratings within the Autism Diagnostic
ment of autistic spectrum traits in typical samples, not the Observation Schedule. Journal of Autism and Developmental
diagnosis of ASD. Another limitation concerns the size Disorders, 47(9), 2770–2782.
Austin, E. J. (2005). Personality correlates of the broader autism
and characteristics of the study sample. Although using phenotype as assessed by the Autism Spectrum Quotient (AQ).
university samples for validation studies is very common, Personality and Individual Differences, 38(2), 451–460.
the absence of data from the general (non-university) pop- Bailey, A., Le Couteur, A., Gottesman, I., Bolton, P., Simonoff, E.,
ulation may raise some doubts regarding the generalization Yuzda, E., et al. (1995). Autism as a strongly genetic disorder:
Evidence from a British twin study. Psychological Medicine,
of the factor structure for this wider group. Furthermore, 25(1), 63–77.
future studies may study the psychometric properties of Baron-Cohen, S., Wheelwright, S., Skinner, R., Martin, J., & Club-
this scale in larger and more representative samples. ley, E. (2001). The autism-spectrum quotient (AQ): Evidence
It is worth emphasizing that AQ is not an instrument from asperger syndrome/high-functioning autism, males and
females, scientists and mathematicians. Journal of Autism and
for the diagnosis of ASD, and for the purposes of detailed Developmental Disorders, 31(1), 5–17.
clinical analysis the methods considered as gold standard Bentler, P. M. (1990). Comparative fit indexes in structural models.
should be used. However, in several research situations Psychological Bulletin, 107(2), 238.
(e.g. large-scale epidemiological studies) and clinical stud- Bilich, F., Silva, R., & Ramos, P. (2006). Análise de flaxibilidade em
economia da informação: modelagem de equações estruturais.
ies (e.g. broad-spectrum baseline assessments) where there Revista de Gestão da Tecnologia e Sistemas de Informação,
are time and availability limitations for participants or 3(2), 93–122.
patients, this screening instrument may be of great utility. Broadbent, J., Galic, I., & Stokes, M. A. (2013). Validation of autism
Our results suggest that the Brazilian version of the spectrum quotient adult version in an Australian sample. Autism
research and treatment, pp. 1–7.
three-factor AQ with 25 items has adequate psychometric Constantino, J. N., & Todd, R. D. (2003). Autistic traits in the gen-
properties. Therefore, the possibility of using the AQ in eral population: A twin study. Archives of General Psychiatry,
the Brazilian context should generate relevant contribu- 60(5), 524–530.
tions to research and clinical areas, allowing quick and Freeth, M., Sheppard, E., & Ramachandran, R. (2013). A cross-
cultural comparison of autistic traits in the UK, India and
easy tracking of autistic traits in large samples. Malaysia. Journal of Autism and Developmental Disorders,
43, 2569–2583.
Acknowledgments  Julia Horta Egito is supported by a PhD grant Hair, J. F., Anderson, R. E., Tatham, R. L., & Black, W. (2005).
(Fundação de Amparo à Pesquisa do Estado de São Paulo, FAPESP Análise Multivariada de Dados. Porto Alegre: Bookman.
-2016/17242-5). Ana Alexandra Caldas Osório is supported by research Hoekstra, R. A., Bartels, M., Cath, D. C., & Boomsma, D. I. (2008).
Grant No. 161001 (Fundo Mackenzie de Pesquisa—MackPesquisa). Factor structure, reliability and criterion validity of the Autism-
The authors wish to thank Beatriz Marques Sanchez for her assistance Spectrum Quotient (AQ): A study in Dutch population and
in the translation process, Gabriel Gaudêncio do Rêgo for his assistance patient groups. Journal of Autism and Developmental Disor-
with graphic edition and Ana Karla Silva Soares with her assistance ders, 38(8), 1555–1566.
in data analysis. Hurst, R. M., Mitchell, J. T., Kimbrel, N. A., Kwapil, T. K., &
Nelson-Gray, R. O. (2007). Examination of the reliability and
Authors Contributions  JHE participated in study design, data collec- factor structure of the Autism Spectrum Quotient (AQ) in a
tion, statistical analyses and manuscript preparation; GMRF and MIG non-clinical sample. Personality and Individual Differences,
participated in study design, data collection and data analyses; AACO 43(7), 1938–1949.
conceived and designed the study, coordinated the statistical data analy- Kline, P. (1994). An easy guide to factor analysis. New York, NY:
ses, and participated in manuscript preparation and review. All authors Routledge.
read and approved the final manuscript. Lepage, J. F., Lortie, M., Taschereau-Dumouchel, V., & Théoret, H.
(2009). Validation of French–Canadian versions of the Empathy
Quotient and Autism Spectrum Quotient. Canadian Journal of
Compliance with Ethical Standards  Behavioural Science, 41(4), 272.
Matson, J. L., Matheis, M., Burns, C. O., Esposito, G., Venuti, P.,
Conflict of interest  The authors declare that they have no conflict of Pisula, E., et al. (2017). Examining cross-cultural differences
interest. in autism spectrum disorder: A multinational comparison from
Greece, Italy, Japan, Poland, and the United States. European
Ethical Approval  All procedures performed in studies involving human Psychiatry, 42, 70–76.
participants were in accordance with the ethical standards of the insti- Rutter, M. (2011). Research review: Child psychiatric diagnosis and
tutional and/or national research committee and with the 1964 Helsinki classification: Concepts, findings, challenges and potential.
declaration and its later amendments or comparable ethical standards. Journal of Child Psychology and Psychiatry, 52(6), 647–660.
Schmitt, D. P., Allik, J., McCrae, R. R., Benet-Martínez, V., Alca-
lay, L., Ault, L., et al. (2007). The geographic distribution of

13
Journal of Autism and Developmental Disorders

Big Five personality traits: Patterns and profiles of human self Volkmar, F. R., & McPartland, J. C. (2014). From Kanner to DSM-
description across 56 nations. Journal of Cross-Cultural Psy- 5: Autism as an evolving diagnostic concept. Annual Review of
chology, 38, 173–212. Clinical Psychology, 10, 193–212.
Stewart, M. E., & Austin, E. J. (2009). The structure of the Autism- Wakabayashi, A., Baron-Cohen, S., Wheelwright, S., & Tojo, Y.
Spectrum Quotient (AQ): Evidence from a student sample in (2006). The Autism-Spectrum Quotient (AQ) in Japan: A cross-
Scotland. Personality and Individual Differences, 47(3), 224–228. cultural comparison. Journal of Autism and Developmental Dis-
Tolentino, T. M., Maia, M. D. F. D. M., Formiga, N. S., Sousa, B. V. orders, 36(2), 263–270.
D. O., & Melo, G. F. (2015). Modelagem estrutural fatorial e
consistência interna da escala de autoestima de Rosenberg em
adolescentes brasileiros. Revista de Psicologia, 6(2), 40–49.

13