Introduction
Current psychiatric classification systems (e.g. DSM IV, ICD-10) have attempted to sub cateagrise children with PDD/autism into subcategories of autism and Asperger Syndrome. Research suggests that this strategy may be meeting with little success, and clinicians are confused about how to diagnose these conditions in a reliable and valid way, based on both current presentation and developmental history. The re-surgence of interest in Asperger Syndrome which has taken place over the last few years has generated considerable research focused on trying to identify similarities and differences between Asperger syndrome and autism.
A number of studies (e.g. Gillberg and Gillberg,
1989 ; Szatmari, Tuff, Finlayson, & Bartolucci, 1990; Waterhouse
et al. , 1996; Ozonoff, Pennington & Rogers 1991) have compared
children diagnosed with autism or Aspergers on symptom patterns,
behavioural manifestations, cognitive profiles, and theory of
mind tasks, to try to find more objective ways of differentiating
between the disorders. It is probably fair to say that none of
these attempts have led to generalisable conclusions. Generally,
it seems that Asperger children are those who are at the higher
levels of functioning, amongst a group of children with autistic
disorder. They are not easily differentiated from high functioning
autistic individuals.
One method of investigating whether there are in fact any empirically 'true' diagnostic differences is to use statistical approaches, to look at factors or clusters of symptoms which characterise putative distinguishable sub groups. There have been some previous attempts at sub categorising autistic children using cluster analytic or taxonomic techniques, most commonly using young children. Recent studies of this genre have included that of Waterhouse and colleagues (Waterhouse et al 1993, 1995; Golden & Mayer, in press) who identified two taxa or sub groups of young ( 3- 7 year old) children, primarily differentiated by developmental status (rather than by symptoms), including chronological age, and verbal IQ.
Szatmari (1992), Castelloe & Dawson (1993), Eaves
et al (1994), and Siegel et al (1986) have also reported sub classification
attempts. It is probably fair to say that in most of this work,
differences related to severity of impairment and level of functioning
rather than distinctive patterns of symptoms are found.
A further issue of interest concerns the relationship between diagnosis and classification and other marker variables for autism. Strong claims have been made for the importance of 'mentalising ability' or Theory of Mind (ToM) as a distinguishing feature of autism, and for likely connections between specific brain systems or modules and capacities of this kind. In the case of autism it is suggested that there is a basic dysfunction in those systems serving mentalising functions (e.g. Baron-Cohen, 1989; Frith, 1989 ). Originally it was believed that this ability was independent of verbal and cognitive capacities, but most recent research has suggested that this is a 'false belief'. Studies of higher functioning children including AS cases have shown some capacity for mentalising even if this is limited by comparison with that of normally developing children.
It is possible that individuals who show ToM ability may belong to a different category or sub type of autistic disorder from those who do not; hence the role of ToM as a diagnostic marker may be relevant within an autistic spectrum. This research was designed to investigate this possibility. In brief, our study was an attempt to discover whether;
The first step in this project was to gather a large sample of children with these clinical diagnoses, and to attempt to separate them into empirically derived clusters, based on symptom patterns and developmental history. We could then look at theory of mind and other cognitive and neuropsychological characteristics to see whether they might validate clusters and also provide some clues relevant to theories of aetiology (see Manjiviona and Prior, 1995, for data on the clumsiness marker and their 1996, accompanying paper at this conference).
This paper reports on the cluster analyses undertaken
to meet this aim, and the relationships between performance on
Theory of Mind experimental tasks and cluster group membership.
METHOD
Subjects: A group of children diagnosed with AS or high functioning autism (HFA) or related PDD, from Britain and from the South Eastern states of Australia was initially enrolled (N = 135).
Added to this a sample is a group of 22 normal British children to provide a comparison group.
Only high functioning children were included since the study would involve theory of mind testing, which required a minimum mental age of approximately 3-4 years. In any case, the diagnostic confusion which drives this research pertains primarily to higher functioning cases.
Children had been independently diagnosed by (nineteen)
clinicians from various agencies as HFA, AS, PDD-NOS, DAMP, 'autistic
features'. Their ages ranged from 3 to 21 years, with a mean
age of 10.22. Verbal Mental Age as assessed with the PPVT or the
BPVS ranged from 2.5 to 34 years, with a mean of 9.8 years. There
were 114 boys and 21 girls. (Table 1)
Measures:
The major measure used to provide a detailed account
of the developmental history and behavioural characteristics of
the subjects was a Diagnostic Check List which was developed by
the authors and which covers the symptoms required for diagnosing
according to DSM III R, DSM IV, and ICD 10 systems. Empirically
derived diagnostic clusters can thus be compared with the clinical
diagnosis by any one of these systems.
The checklist version used in this study was based on an earlier instrument devised by Wing (published in Rapin 1996), plus some items from the Diagnostic Interview for Social and Communicative Disorders). It covers the domains of : impairments in social interaction (including use of body language, greeting behaviour, comfort seeking, and giving, awareness of feelings of others, friendships, and awareness of social rules; imitation and play , including joint referencing and interactive play, pretend play, and imitation;
impairments in communication and imagination, including comprehension and use of language, speech characteristics, non verbal communication, imagination and pretence;
restrictions and repetition in self chosen behaviour, inciuding stereotyped movements, pre-occupations with objects and with patterns of interests, maintenance of sameness.
In addition, where possible, data is gathered on
pregnancy and birth history, developmental milestones, health
problems, famiiy history of any disorders, onset of the disorder
(or when first noticed), treatment, and current educational milieu.
The Peabody Picture Vocabulary test - Revised
is administered to the child to provide a measure of verbal mental
age and standard score.
Theory of Mind measures included the now famous "Sally/Anne" task ; and the "Box of Smarties" task, as tests of 'first order' theory of mind, in this case 'false belief'. Performance on these tasks illustrates whether children have the ability to recognise that other people may have false beliefs about a situation which lead them to behave in certain ways.
We also administered a test of 'second order' theory
of mind, i.e., understanding related to a belief about another
person's belief. In Australia a version of Bowler's (1992) shopping
story was used, and in Britain, Baron-Cohen's (1989) Ice Cream
story was used. Both stories involve two characters who want
to buy something, and through a series of events they develop
differing knowledge states. The subject is asked to solve the
problem of the kind " X thinks Y thinks that........"
and to predict a character's behaviour on the basis of his/her
false belief.
The checklist was completed during interview with parents in the famliy home, or (occasionally) in the clinic. In some cases questions are only relevant to either early (e.g. babbling) or current history (e.g. maintaining friendships at school or work), hence developmentally relevant data are incorporated. Children were also observed before, during, or after the interview.
The interview and assessment takes on average about
two hours for each child and family.
RESULTS
Methods of Analysis
A K Means Cluster Analysis was the major statistical approach taken, to search for sub categories of children.
This approach uses an iterative partitioning method in which data are successively partitioned into the specified number of clusters.
We tried 2,3,4,5,6,7,and 8 cluster solutions, and the 3 cluster solution seemed to best agree with clinical presentation as described in the literature. It is this solution which is reported here.
(Note, 5 variables for which <10% of yes
responses occurred were eliminated; these included no babbling,
no spoken language, no response to communication, lack of spontaneous
activity, and smearing faeces).
Three cluster solution
Cluster 1 contained 36
cases (plus all of the controls who will not be referred to again
in the comparisons to follow but who serve as a kind of validation
of a less problematic group), and was termed the 'Mild' group
.
Cluster 2 contained 42 cases and was called the Asperger group.
Cluster 3 contained 57
cases and was termed the Autistic group.
Table 2 shows the relationships between cluster diagnosis
and clinicians' diagnosis. It demonstrates that most of the clinically
diagnosed AS children clustered in the AS group; children with
autism clinical diagnoses were evenly divided into AS and Autism
clusters, and the Mild cluster comprised most of the PDD and 'other'
children, as well as a small and equivalent number of AS and autism
diagnosed children.
The important question is how did these children differ from each other, i.e. what was the basis for the separation into the three clusters.
The next section provides information on discriminating
variables.
1. Background variables
There were no group differences on:
age, gender, history of language delay or deviance,
age of first walking, age when problems were first recognised,
or family history of disorder.
The Mild group was more likely to have delayed sitting than the other two groups (p= .016)
The AS group was more likely to have delay in crawling (p = .04).
The Mild group was least likely and the Autistic group the most likely to have had a history of birth difficulties (p = .01).
The Asperger group was significantly older than the Mild group (1 1.6 vs 8.5 years; p < .05)
The verbal MA of the Autistic cluster was significantly lower at 8 years; with the AS and Mild groups equivalent at 11.7 and 10.9 years (p <.05). See Table 1.
The higher functioning children tended to cluster
in the AS or mild groups.
2. Diagnostic variables (Tables 3.4.5.)
Logistic regression analyses, which allows the use of dichotomous variables (e.g. symptom or behaviour present or absent) were used as a means of identifying discriminating symptoms between the groups. This technique chooses the best (but not the only) variables to identify group differences, on the grounds of parsimony and significance.
They were run firstly comparing the AS and Autistic groups, secondly comparing the AS and Mild groups, and thirdly comparing the Autistic and Mild groups .
Prediction rates for group membership were all high
across all domains, with group comparisons varying between 74%
and 99% correct prediction
For most of the symptoms on which a difference was
found, the direction was consistently that the autistic group
was more severely impaired. This was true for comparisons with
both AS and Mild clusters . There are some exceptions to this
which are underlined in the tables.
Comparisons between the AS and Mild group were consistent
in showing the Mild group as less impaired.
The highlights of these analyses which are relevant to our current diagnostic dilemmas include:
AS children were more likely than other groups to have a friend with similar circumscribed interests;
were more likely than autistic children to use long winded pedantic speech;
were more likely to show joint attention behaviour;
and were more likely than the Mild (but not the autistic)
group to engage in one sided repetitive conversation, and to interpret
language literally.
3. Theory of Mind comparisons.
Tables 6 and 7 show that there were significant group differences on each ToM task.
In each case the AS group was more likely to give responses demonstrating ToM, or mentalising ability.
However, it is important to remember that this group is also older and has a higher Verbal Mental Age by comparison with the Autistic (but not the Mild) group. Almost all AS children passed the first order tasks.
It should also be noted that well over half the AS group passed second order ToM, compared with less than a third of the Mild group (who were two or more years younger than the other two groups), and one third of the Autistic group.
Results concerning the relationships between diagnostic
group membership and ToM performance are thus moderated by the
age and Verbal MA associations.
They confirm that the AS cluster is less handicapped by comparison with the Autistic group in all domains.
However, symptomatically AS children are more handicapped
in many domains than the Mild group; their superior performance
on the ToM tasks may be explicable in terms of age and IQ differences.
This does suggest that these may be more powerful influencial
variables in sub group discrimination than autism related social
and communicative impairments, and is consistent with arguments
which have suggested that ToM may be another attribute of intellectual/cognitive
development rather than a specific 'modular' cognitive factor.
SUMMARY AND DISCUSSION
1. Empirical clustering showed that it was possible to differentiate three groups or clusters roughly corresponding to those familiar to us through clinical experience. But since the group comparisons suggested that severity of symptoms was a major underlying factor, rather than particular distinctive symptom patterns, this argues for a spectrum concept of autistic type disorders, rather than for distinctive categories.
In other words there seems little evidence other
than severity of symptoms and levels of cognitive functioning
(which are no doubt related factors), that AS is a separable group.
These children were mostly distinguished by their less impaired
joint attention skills, their pedantic and long winded style of
language and for their tendency to be able to have a 'friendship'
based on mutual 'circumscribed' interests.
The fact that there were no cluster group differences
on history of language delay or deviance needs to be highlighted.
It suggests that this is not a differentiating feature
which can reliably be used in differential diagnosis.
2. Comparison of clinician diagnosis with cluster group membership suggests that AS cluster children are more likely to have received an AS diagnosis, but that Autistic cluster children are as likely to have received an AS diagnosis as an Autism one. This could be related particularly to the fact that there has been a marked increase in clinician's use of the AS diagnosis for higher functioning children over the years of this study.
Chi squared comparisons showed that there were no systematic differences between clinicians (nor between Australian and British children) in the likelihood of an AS rather than an autism diagnosis.
It is also worth recalling that the sample in this study was a relatively high functioning one, excluding the substancial proportion of children who are resistant to assessment on standard tests; hence we sampled the upper part of the spectrum.
All of our AS children also met conventional diagnostic
criteria (DSM III-R, ICD-10) for autism. The cluster analysis
showed that despite this, they could be differentiated on some
key items of behaviour.
3. Performance on ToM tasks confirmed the importance of ability /age variables which has been increasingly emerging in studies of this genre. Few AS children failed first order tasks, and more than half could also pass the second order task. Comparison with the Bowler (1992) study (Table 7), indicates that a similar proportion of his AS subjects passed the second order task.
It is also noteworthy that more than half of the Autistic group passed first order ToM, and of that sub sample presented with the second order task, more than half passed . This suggests that when verbal ability is not too far from an average level these deficits are less likely to be evident.
Again we would argue that this impairment may be primarily associated with developmental cognitive and language delay rather than simply autistic disorders.
In general, we believe that the results of this research argue for a continuum of autistic disorders in which severity of social and communicative impairments underlie individual differences. The same conclusion almost certainly applies to interpretations of theory of mind performance.
In their study comparing two taxonomically derived groups ( which broadly comprised a core autistic cluster and an 'other PDD' cluster) with APA and ICD systems, and with Wing and Gould's categorisation of 'active but odd', 'passive' and 'aloof', Waterhouse et al (1 996) considered whether the grouping reflected differentiation on the basis of level of functioning, based on the existence of a 'core' autistic group plus 'others', or based on a severity continuum or spectrum. This question is also of importance to our interpretations.
Bearing in mind that we had an older sample, and three groups rather than two, (compared with Waterhouse et al, our non autistic cluster sub divided into AS and 'Mild'), our data generally support an interpretation based on what Waterhouse called 'severity of developmental compromise' . We would argue that this is not inconsistent with the spectrum conceptualisation of autism. The ToM results demonstrated that verbal comprehension abilities were important in assisting children with considerable social impairments to pass ToM tasks. Our comparisons between AS and Mild groups highlight the importance of this variable. We have gone one step further than Waterhouse et al. in showing that a PDD group can be subdivided according to selected social impairments and ToM ability and that this latter is related to verbal capacity.
We would argue that the distinguishing symptoms of
our AS group (limited friendships, pedantic speech and circumscribed
interests) might also be related to level of cognitive functioning
(see Tsai 1992, and Prior & McMillan 1973) suggesting that
it is this which is often primary in influencing an AS diagnosis.
The fact that the developmental history variables
did not discriminate between the sub groups in any diagnostically
meaningful way (particularly noting the failure of the language
development variable in discriminating AS children), suggests
the need for caution in using such data for differential diagnosis.
It suggests that etiological factors too, may support a continuum
concept of autistic disorders.
REFERENCES
Baron-Cohen, S. (1991b). The theory of mind deficit
in autism: How spcific is it? Brilish Journal
of Developmental Psychology, 9, 301-314.
Bowier, D. M. (1992). "Theory of mind"
in Asperger's syndrome. Journal of Child Psychology and Psychiatry,
33, 5, 877-893.
Castello, P., & Dawson, G. (1993). Subclassification
of children with autism and Pervasive developmental Disorders:
A questionnaire based on Wing's subgrouping scheme. Journal
of Autism and Developmental Disorders, 23, 225-237.
Eaves, L.C., Ho, H.H., & Eaves, D.M. (1994).
Subtypes of autism by cluster analysis. Journal of Autism
and Developmental Disorders, 24, 3-22.
Frith, U. (1989). Autism: Explaining the enigma.
Oxford: Basil Blackwell.
Ghaziuddin, M., Butler, E., Tsai, L., & Ghaziuddin,
N. (1994). Is clumsiness a marker for Asperger syndrome? Journal
of Intellectual Disability Research, 38, 5, 519-527.
Gillberg, I. C., & Gillberg, C. (1989). Asperger
syndrome-Some epidemiological considerations: A research note.
Journal of Child Psychology and Psychiatry, 30, 4, 631-638.
Manjiviona, J., & Prior, M. (1995). Comparison
of Asperger syndrome and high-functioning autistic children on
a test of motor impairment. Journal of Autism and Developmental
Disorders, 25, 23-39.
Ozonoff, S., Rogers, S.J., & Pennington, B.F.
(1991). Asperger's syndrome: evidence of an empirical distinction
from high-functioning autism. Journal of Child Psychology and
Psychiatry, 32, 1107-1122.
Prior, M.R., & MacMillan, M.B. (1973). Maintenance
of sameness in children with Kanner's Syndrome. Journal of
Autism & Childhood Schizophrenia, 3, 154-167.
Rapin, I. (1996). Preschool children with inadequate
communication: developmental language disorders, autism or low
IQ. Clinics in Developmental Medicine, No. 139, Mac Keith
Press: London.
Siegal, B., Anders, T.F., Ciaranello, R.D., Bienenstc>ck,
B., & Kramer, H.C. (1986). Empirically derived subclassification
of the autistic syndrome. Journal of Autism and Developmental
Disorders, 16, 275-294.
Szatmari, P. (1992). The validty of autistic spectrum
disorders: A literature review. Journal of Autism and Developmental
Disorders, 22, 583-600.
Szatmari, P., Archer, L., Fisman, S., Streiner, D.L.,
& Wilson, F. (in press). Asperger's syndrome and autism:
Differences in behaviour, cognition, and adaptive functioning.
Journal of the American Academy of Child and Adolescent Psychiatry.
Szatmari, P., Tuff, L., Finlayson, M.A.J., &
Bartolucci, G. (1990). Asperger's syndrome and autism: Neurocognitive
aspects. Journal of the American Academy of Child and Adolescent
Psychiat-y, 29, 130-136.
Tsai, L. (1992). Diagnostic issues in high functioning
autism. In E. Schopler & G. Mesibov (Eds), (underline) High
Functioning Individuals with Autism, pp. 11-40. New York:
Plenum.
Wing, L. (1981). Asperger's syndrome: A clinical
account. Psychological Medicine, 11, 115-129.
Wing, L., & Gould, J. (1979). Severe impairments
of social interaction and associated abnormalities in children:
Epidemiology and classification. Journal of Autism and Developmental
Disorders, 9,11-29.
| N=157 | (22 controls) |
| M=114
F=21 | PDD |
| _
X CA= | Asp 11-6 (N=42)
Aut 10-3 (N=57) "Mild" 8-5 (N=36) Asp > Mild p<.05 |
| _
X VMA= (PPVT) | Asp = 11-7
Aut= 8.0 "Mild"= 10-9 Aut < Asp p<.05 |
| Cluster
diagnosis | CLINICIANS DIAGNOSIS | |||
| Autism | Asperger | PPDNOS | "other" | |
| "Mild" | 11 | 12 | 5 | 8 |
| "Asperger" | 11 | 30 | 0 | 1 |
| "Autism" | 26 | 27 | 2 | 2 |
p=0.000 Highly sig diff b/w groups
FIRST ANALYSIS- which variables predict Autistic
and Asperger group N=99
Variable
mild/not likely ---- severe/more likely
Social domain
A1BC- dislikes physical affection asp-aut
A3DC- stereotyped comfort seeking asp-aut
A4CC- is distressed by others pain- no offer
of comfort or sympathy asp-aut
A5EC- has 1 friend with same circumscribed interest
aut-asp
A5EC is moderately correlated with A7HC (O=0.5) (see
below)
A6AC- unaware of need for personal modesty asp-aut
A6AC is moderately correlated with
A6BC (O=0.5) "unaware of psychological barriers"
A7AC- no interest in simple games asp-aut
A7AC is highly correlated with
A3AC (O=1.0) "never seeks comfort/ignore pain,heat,
cold"
A7BC- does not point to objects to show interest/pleasure asp-aut
A7EC- involves other children only as mechanical aids in play aut-asp
A7HC- engages with other person who has same
circumscribed interest (correlated with
A5EC above) aut-asp
Prediction rate of groups 97.98% with these soccial variables
Communication domain
B1CC- has speech but does not initiate conversation asp<aut
B2BC- no response to instructions asp<aut
B2CC- frequently respond to word/phrase out of context asp<aut
B3AC- use of echolalia asp<aut
B3BC- reverse pronouns asp>aut
B3CC- use of idiosyncratic words or pphrases asp<aut
B3DC- use of long winded pedantic speech aut<asp
B3DC is highly correlated to A6DC (O=1.0) "
Make embarrassing remarks in public"
Prediction rate of groups 86.9%
Repetitive / Stereotyped Behaviours domain
C3BC- does collect objects for no apparent purpose aut<asp
C3DC- shows interest in parts of objects aut<asp
C3FC- unusually interested in abstract properties of objects asp<aut
C5AC- insist on maintaining same routines asp<aut
C7AC- severe impairment of spontaneous activities
asp<aut
Prediction rate 74%
SECOND ANALYSIS - which variables predict Asperger
and Milds group N=78
Social Domain
A2BC - does not spontaneously wave good bye mild<asp
A2DC - does not say or sigh "hello" mild<asp
A3CC - shows distress if hurt-does not come for comfort mild<asp
A3DC - stereotyped comfort seeking mild<asp
A5CC - wants friends but poor grasp of friendship
mild<asp
A5CC is moderately correlated with
A4AC (O=0.52) "unaware of others personal space"
A5EC - has 1 friend with same circumscribed interest
mild < asp
A5EC is highly correlated with
A2EC (O=1.0) "shows innappropriate affection", and moderately correlated with
A7HC (O=0.58) "engage with one other specific
person who has same interest"
A6DC - makes embarrassing remarks in public mild<asp
A8EC - tries to imitate social behaviour-looks bizarre mild<asp
A8EC is highly correlated with
B6BC (O=1.0) "shows appropriate use of miniature
objects but play is mechanical"
A9AC - fails to animate toy animals or dolls etc
mild<asp
A9AC is moderately correlated with
A9BC (O=0.52) "animates few toys but done in
a limited repetitive way"
Predicition rate 98,72%
Communication domain
BlDC - engages in one sided repetitive conversations mild<asp
B2DC - interprets language literally mild<asp
B3AC - echolalia mild<asp
B3CC - idiosyncratic use of language mild<asp
B4AC - unusual tone of voice mild<asp
Predicition rate 89.74%
Repetitive stercotyped behaviours domain
C1DC - use of complex finger and hand movements mild<asp
C1DC is highly correlated to
A2EC (O=1.0) "innapropriate show of affection",
A8DC (O=I.0) "imitates person, animal or object", and
C2CC (O=I.0) "engages in self injury"
THIRD ANALYSIS- which variables predict Autistic
and Mild group N=93
Social domain
A2DC - say hello to greet mild<aut
A6AC - unaware of need for personal modesty mild<aut
A6AC is highly correlated with
ALCC (O=1.0) "no look or smile at social approach",
A3EC (O= 1.0 "comfort seeking is bizarre and repetitive",
A5EC (O= 1.0) "one friend with same interest",
A6BC (O=0.5) "unaware of psycholocical barriers"
A9AC (O=1.0) "fails to animate toys"
C5BC (O=0.5) "has limited self chosen activities", and
C5CC (O= 1.0) "prefer to cling to home or familiar
place"
A6DC - makes embarrassing, remarks in public mild<aut
A6EC - lack of awareness/innapp. response to other's emotions mild<aut
A7BC - does not point to objects mild<aut
A9BC - animates few toys but done in a limited repetitive
way
Prediction rate 97.9%
Communication domain
B1CC - does not initiate conversations mild<aut
B lDC - encages in one sided repetitive conversations mild<aut
B2DC - understands language in a literal manner mild<aut
B2DC is highly correlated with
A3EC (O=1.0) "comfort seeking is bizarre and repetitive" and
A9AC (O=1.0) "does not animate toys, dolls"
B3AC - echolalia mild<aut
B5AC - no use of non-verbal communication mild<aut
Prediction rate 88.2%
Stereotyped/repetitive behaviours domain
C2DC - sensory disturbances mild<aut
C5AC - interested in maintaining same routines mild <aut
C5BC- limited pattern of self -chosen activities mild<aut
C6BC - act out roles of a person, animal etc mild<aut
C6CC - special skill mild<aut
Prediction rate 83.9%
C3BC - collects objects for no apparent use mild<asp
C5BC - has a limited pattern of self-chosen activities mild<asp
C6BC - acts out roles of person etc in repetitive
way mild<asp
C6CC - has a special skill mild<asp
C6CC is moderately correlated with
A5EC "has one friend with same circumscribed interest", and
A7HC "engages with one friend with same interest"
Prediction rate 92.3%
THEORY OF MIND
There are sig diffs on both 1st order ToM games.
Asp group more likely to pass.
ToM Sally /Anne
p=0.000
| Pass | |||
| Fail |
ToM SMARTIES p=0.000
| Pass | |||
| Fail |
| Pass | |||
| Fail |
Figures in brackets = % of total sample
Note 1 Asp= older and have significantly higher VMA
Note 2 Passers could not explain their answers
Note 3 Similarity of results c.f.
Bowlers Asperger group who were older