The findings o f this study are circumscribed by consideration regarding the validity o f the measures used. Campbell & Fiske (1959) describe four criteria which need to be fulfilled in order for a test to be considered valid. One o f these is convergent validity, which they describe as a “confirmation by independent measurement procedures.” Firstly considering the theory o f mind measures, there were no significant correlations between the Happe
stories, the Eyes task or the Dewey stories. This is especially consequential for the Stories and the Eyes as they are both described as measures of theory o f mind. There were no ceiling or floor effects on either task, so even if the tests were o f a different complexity, or sensitivity, one would expect a significant relationship. One explanation would be that emotion recognition as assessed by the eyes task, is not necessarily a measure o f theory o f mind.
There were similar findings for the executive function measures, where there were no significant correlations between measures. The two measures which are described as measuring planning, did not correlate significantly with each other, Six Elements and Key Search. One possible explanation for this could be that the groups may be failing the two tasks for different reasons. Indeed, the instructions on the Six Elements task are long and complex, whilst the instructions for Key Search are relatively simple. Ozonoff (1995) found improved performance on the WCST in autistic participants when verbal and social demands were removed by using a computerised version o f the WSCT. She recommends future research needs to implement paradigms which remove any extraneous measurement variance, by removing unintentional task demands.
There were also concerns regarding Predicaments. In this study, the DEX questionnaire which did significantly differentiate the groups, did not correlate significantly with any other measures. The DEX is designed to be a measure o f the difficulties experienced in everyday life. The Predicaments is a newly developed assessment and has not been used with adolescents, therefore it is disappointing that it shows no significant relationship with the DEX. One possible explanation for the lack o f relationship would be the status of the rater. The
concern around their child’s behaviours. Their responses may have then reflected the difficulties they were experiencing in managing difficult behaviours, or the difficulties o f the developmental tasks at hand in adolescence, for example beginning secondary school. This could mean that the DEX was not a truly objective account o f actual everyday behaviour. Partial support for this comes from comparison o f the DEX ratings o f adults with frontal damage, Channon & Crawford (1999) found that adults with anterior damage scored a mean o f 25.08 and adults with posterior damage scored a mean o f 18.67 when a relative rated their behaviour. In our group the Asperger’s group scored a mean o f 40.71, much higher than the patients with frontal damage, and the controls scored 16.64. Thus it would have been useful to get a second rating by another rater such as a teacher. Some support for the ecological validity of Predicaments is found in Channon & Crawford, (1999). They found that the group o f patients they assessed with most real life difficulties, (those with anterior lesions), showed the greatest level o f impairment on Predicaments. The Predicaments task had been developed and devised for adults. The predicaments involved were derived at on the basis o f interviewing adults. This may reduce their ecological validity. However, whilst the predicaments were not generated specifically by adolescents a wide range of people of varying ages were interviewed in order to develop the situations. Further, there is no evidence to suggest than the participants had any difficulties understanding these predicaments, and most o f the predicaments appeared on the face o f it, to be relevant to adults and young people alike. However there would be an increased personal relevance to this group, if the actors in the situations had been o f a similar age, and the predicaments had taken place in more familiar places such as colleges and schools, for example, rather than the workplace.
mirrors more real life social problem solving. These experimental tasks can underestimate difficulties. In these experimental tasks there are cues which structure performance, although Predicaments minimised these by using open ended problem situations and restricting prompting to the factual account o f the predicament. There was however an inevitable task structure. The participant was asked to think o f as many solutions as they could, and then to select one, a structure which is not necessarily in place in real life. There is also the possibility, that there will be discrepancies between how one would actually act, especially in situations which require immediate reaction, and how one describes verbally one’s solution. Similarly difficulties could be overestimated if participants have difficulties in imagining themselves in these situations, or are less motivated by a wish to resolve the predicaments successfully.
The attempt to try and compare Social and Non Social Predicaments was not entirely satisfactory. One issue is the different methods o f presentation used in each task. However, when Channon & Crawford (1999) looked at the Social Predicaments and presented half as videos and half as stories, they concluded there was little evidence to suggest that method o f presentation significantly altered the nature o f the results. A further problem with the Non Social Predicaments is that their description as Non Social is not entirely justified. Although the situations were devised to remove the social element, as far as possible, by only having one character and attempting to use practical predicaments, the solutions to these predicaments immediately introduced a social element. Thus the lack o f group differences here may have been to factors such as reduced complexity rather than simply the removal of the social element.