2.2 Method
2.2.3 Measure – Developing the Observation Coding Scheme
The Repetitive Behaviour Coding Scheme (RBCS) was developed using the methods
described below. The final version of the Repetitive Behaviour Coding Scheme is in Table
2.2.
2.2.3.1 Pre-pilot and pilot observations. I conducted pre-pilot observations on 30
hours of video records (this was 15 infants). These preliminary observations were conducted
by watching one infant at a time, recording all movements made by each infant. I transcribed
a continuous narrative record of all repetitive behaviours made by each infant and this
narrative approach was used for the preliminary sub-sample of 15 infants. I then read over the
narrative records and the behaviours that resembled the motor stereotypies and repetitive
actions on objects from previous studies were extracted. Whilst focusing on Thelen’s
behavioural categories I read my transcripts and focused on extracting any behaviours that
resembled her previous work, whilst also noting any behaviours that were not recorded by
Thelen. This method resulted in a list of 8 categorical items (flap, bounce, rock, head
movements, bang toy against another toy, bang toy against another object, clap and arm
banging a surface [e.g. wall, sofa]). A repetition was operationally defined as a movement of
a part of the body that is repeated in the same way three times (or more) within a two second
period. Should flap, bounce, rock or head movements occur whilst the infant was holding or
70 distinction between motor stereotypies and repetitive actions with objects discussed in
previous sections on the thesis
In general, the onset of the behaviour occurred at the moment the particular movement
started and then the offset occurred the moment the child ceased the behaviours, engaged in a
different action with the same objects of a different object, or paused between repetitive
actions for a period longer than 5 seconds (e.g. if an infant flapped and then stopped flapping
for five seconds and then flapped again, the second instance is a separate behavioural event).
Different repetitive behaviours could be coded simultaneously (i.e. when an infant flapped
and bounced at the same time).
The initial observation coding scheme was then used in pilot observations, where a
different 10 participants were observed and coded by three independent observers (these 10
infants were randomly selected from the 100 in the study. The second coder was trained to
use the coding scheme and the third coder was not). The 10 participants were different from
the original 15 that were used to develop the scheme. Operational definitions for each item
were edited to ensure clarification and ease of use of the coding scheme, these can be seen in
Table 2.2. In Table 2.3 I have provided examples of behaviours occurring without object
(motor stereotypies) and with object (repetitive action with objects).
2.2.3.2 Applying the RBCS to the sample. For the purpose of the formal coding, I
started coding again and coded all 100 participants (including the ones used in the
preliminary observations). Each video was observed two times, each viewing recorded the
behaviours of one individual participant on an event-based coding. Initially observers
transcribed and coded the type of behavioural repetition exhibited by the infant (e.g. flapping,
71 behavioural repetition and were also instructed to count the number of behavioural repetitions
displayed per event. Subsequently, the RBCS allows measurement of the frequency of the
behavioural bouts of repetition, the duration of the behaviour and the number of repeats
within each behavioural event. A randomly selected transcript can be seen in Appendix II.
2.2.3.3 Establishing reliability. I coded all of the videos. Because the RBCS was a
newly developed measure of repetitive behaviours, a second observer coded 33% of the
videos. To measure reliability, intra-class correlation coefficients (ICC) were calculated for
the total number of behavioural repetitions per participant, ICC inter-rater agreement for the
total number of repetitive motor behaviour bouts per participant was .95. On the item level,
for the behavioural category label provided for the bout exhibited, Kappa coefficient
agreement was .91.
In order to ensure that coding had remained consistent throughout the thesis, I coded
5% of the sample (n=5 children) again, 36 months after the initial coding had been
completed. Test-retest reliability was established with the number of behavioural bouts per
participant (ICC = .98) and the type of repetitive behaviour exhibited (ICC = .99).
2.2.3.4 Creating composite variables. The total number of repetitive behaviours
exhibited was calculated for each participant. The total number of repetitive actions with
objects was calculated (flapping, bouncing and rocking transcribed with object, as well as
banging toy against toy and banging toy against another object). Similarly, the total number
of bouts of motor stereotypies was calculated (flapping, bouncing, rocking and head
movement transcribed without objects as well as arm banging against surfaces and clapping).
Composite variables were therefore calculated to provide a simple measure of whether or not
the participant engaged in repetitive behaviour at all, whether they engaged in repetitive
72 time of all behaviours were noted and therefore I was able to calculate the total time infants
engaged in repetitive behaviour.
Participants’ free play session lasted for 25 minutes. For ease of comparison with other
research and in order to compare the descriptive information with those presented throughout
the thesis, a rate per hour was calculated for total repetitive behaviour observed, sum of
motor stereotypies observed and sum of object based repetition observed. A rate per hour was
also calculated for the individual behavioural categories (flap, bounce, rock, head
movements, clap and banging categories). A rate per hour therefore gives a consistent time
frame over which data can be compared across the studies reported in this thesis and allows
other researchers to compare their data to mine simply.