Modified COSMIC 21
This study reports on the modification of a naturalistic observation measure of social
communication in young children with ASD. Data were collected during social interactions with
a teacher and peers at school, and included two contexts: a teacher-led activity and an
unstructured free play session. Concurrent validity of the M-COSMIC codes was evaluated
through comparison of rates of behaviour at school with scores on an autism diagnostic measure
and standardised language assessments. Additionally, evaluation of child behaviour across the
two contexts of group activity and free-play informed the extent to which the specific behaviour
sampling context might influence the results obtained by such a naturalistic observational
measure.
6.1 M-COSMIC and ADOS-G
A number of strong associations were apparent between the M-COSMIC and ADOS-G
algorithm total scores for communication and social interaction ability. Both verbal and non-
verbal forms of communication were associated with the ADOS-G algorithm total scores for
communication and/or social impairments. Inspection at the item level yielded a more modest
pattern of associations, with some non-verbal forms such as gesturing, pointing, showing and
giving being related to relevant ADOS-G items. Additionally, the ADOS-G item Overall Level
of Non-Echoed Language was associated with M-COSMIC rates of initiated vocalization/single
words. However, there was no such association between the former and M-COSMIC rates of
initiated phrase speech.
With respect to functions and roles of communication, associations with ADOS-G were
less clear. M-COSMIC rates of act for the function of behaviour regulation were highly
associated with ADOS-G algorithm scores, as were rates of initiated communication. However,
rates of act for the function of social interaction and joint attention, as well as rates of overall