International Journal of Educational Research 87 (2018) 100–109

Contents lists available at ScienceDirect

International Journal of Educational Research

journal homepage: www.elsevier.com/locate/ijedures

A comparison of video modelling techniques to enhance

social-communication skills of elementary school children
Angelika Anderson* , Brett Furlonger, Dennis W. Moore, Veronica D. Sullivan,
Maximilian P. White
Faculty of Education, Monash University, 57 Scenic Boulevard, Clayton Campus, Wellington Road, Melbourne, Victoria, Australia

A R T I C L E I N F O A B S T R A C T

Article history:
Received 15 February 2016 This study compared the effect of video modelling with and without embedded narration
Received in revised form 1 April 2016 on the social communication skills of four typical school children. Target behaviours were
Accepted 27 May 2016 appropriate conversational turn taking and correct conversational body posture. The
Available online 16 June 2016 differential effects of the two video modelling procedures were assessed using Alternating
Treatments Designs with target behaviours and treatment conditions counterbalanced.
Keywords: Responding on both target behaviours was below normative levels during baseline
Video modelling assessment but improved across the alternating treatments phase, and remained at or
Schools
above normative levels across an optimal treatments phase and in follow-up. Video
Communication skills
modelling with narration was more efficient than video modelling without narration for all
four participants and more effective for two. Implications of the findings and suggestions
for future research are discussed.
ã 2016 Elsevier Ltd. All rights reserved.

1. Introduction

How children conduct themselves during conversations is an important aspect of social competence as this is the primary
means by which we make interpersonal contact, form relationships and regulate our interactions (Gallagher, 1993). Children
who do not adhere to common social-communicative rules, such as displaying appropriate body language, and allowing
their companion to speak without interrupting, will find it difficult to successfully engage and develop good relationships
with peers and adults. The lack of these skills can have a significant impact on a child’s ability to make and maintain
friendships, and can detrimentally impact their academic achievement (Dwyer, Osher, & Warger, 1998; Elliott, Malecki, &
Demaray, 2001).
Studies have also indicated that poor social-communication skills can lead to behavioural and self-regulatory problems,
and adverse academic, social, and relationship outcomes (McClelland, Morrison, & Holmes, 2000; Osofsky & Osofsky, 2001;
Segrin & Flora, 2000; Wentzel, 1991). In contrast, students who have well developed social- communication skills are
typically happier, better adjusted and achieve greater academic success (Elliott et al., 2001; Spence, 2003).
School based teaching of social-communication skills has proven problematic with interventions varying in success.
Indeed, only small positive effects have generally been reported, whether studies involved children with ASD (Bellini, Peters,
Benner, & Hopf, 2007) children with emotional or behavioural disorders (Quinn, Kavale, Mathur, Rutherford, & Forness,

* Corresponding author.
E-mail addresses: Angelika.anderson@monash.edu (A. Anderson), Brett.Furlonger@monash.edu (B. Furlonger), Dennis.Moore@monash.edu
(D.W. Moore), vsul5@student.monash.edu (V.D. Sullivan), maximilian.white@gmail.com (M.P. White).

http://dx.doi.org/10.1016/j.ijer.2016.05.016
0883-0355/ã 2016 Elsevier Ltd. All rights reserved.
 A. Anderson et al. / International Journal of Educational Research 87 (2018) 100–109 101

1999), or typically developing individuals in regular education classrooms (January, Casey, & Paulson, 2011). Early
intervention occurring in the regular classroom rather than in some pull-out facility (Bellini et al., 2007), and targeting
specific rather than generic skills (Quinn et al., 1999) have each been shown to result in more effective outcomes though
clearly more research is needed in this field. Video modelling offers a potentially powerful contribution to social-
communication interventions. It has been trialled extensively in intervention research with children with Autism Spectrum
Disorder (ASD) (Bellini & Akullian, 2007; Moore et al., 2013; Nikopoulos & Keenan, 2004). Video modelling is a procedure in
which (a) a participant is asked to watch a video prior to performing the target skill, (b) the target skill is modelled on screen
by an adult or peer usually in the authentic context of the activity, (c) the instructor provides prompts and reinforcers to the
participant for attending to relevant stimuli, and (d) the participant is then provided with opportunities to perform the skills
displayed in the video, prompted if necessary, and reinforced for imitating the behaviour of the model (Nikopoulos & Keenan,
2006).
Charlop-Christy, Le, and Freeman (2000) suggest that video modelling may result in faster acquisition of skills, and better
generalisation than in vivo modelling. They posit that it is effective because screen based media playing devices are engaging
for children and associated with sustained attention. Video modelling also permits accentuation of certain stimulus features
and minimization of distracting or irrelevant features (Shukla-Mehta, Miller, & Callahan, 2010). Other strategies can be
combined with video modelling to teach a range of social skills to children (Apple, Billingsley, & Schwartz, 2005; Charlop &
Milstein, 1989; Litras, Moore, & Anderson, 2010). Indeed, the incorporation of instructional prompts and positive
reinforcement (both verbal and tangible) appear more effective than video modelling alone for promoting the acquisition,
maintenance, and generalisation of target skills (Shukla-Mehta et al.).
To date little research exists demonstrating the usefulness of video modelling in educational contexts, although video
modelling might be a useful instructional technique to improve the social and communication skills of all children identified
as having deficits in these areas relative to their peers. In many cases, the deficits in social communication exhibited by some
typically developing children are similar to those in children with ASD (Donno, Parker, Gilmour, & Skuse, 2010).
The specific behavioural markers that characterise individuals with ASD exist on a continuum, and individuals without a
diagnosis can still display deficits in social communication skills similar to those in individuals with ASD. The reasons for
these deficits may be common to both populations, such as a diminished capacity to learn via imitation. This is likely given
that typically developing children engaged in mainstream settings, have plenty of opportunity to experience appropriate
behavioural (in vivo) models in their peers. Modelling interventions involving screen media might be effective for such
populations for all the reasons why they work well for individuals with ASD.
Interestingly, many VM intervention studies have included narration overlaying the video model providing additional
cues or directions for the targeted behaviour. The benefits of such voice-overs are unclear leading some to question whether
narration overlaying VM might hinder learning rather than enhance it (Rayner, Denholm, & Sigafoos, 2009). Alternatively
narration might make learning the requisite discriminations simpler. Skinner (1974) first drew a distinction between
contingency-shaped and rule-governed behaviour and pointed out that rules can exert rapid control over behaviour, and that
a person following instructions may behave differently from a person who has been exposed to the contingencies described
by the instructions. Ayllon and Azrin (1964) provide a graphic demonstration of the differential effects of contingencies and
vocal instructions. Working with 18 psychiatric patients they were unable to effect behavioural change by provision of
reinforcement contingencies until they gave instructions about the desired response. The relationship was reciprocal
however; the change in behaviour resulting from instructions was short-lived unless combined with reinforcement. Ayllon
and Azrin concluded that “It would seem that for humans, instructions and reinforcement are complementary” (p 330),
suggesting that utilizing existing verbal repertoires in humans’ instructions may “ . . . eliminate the necessity for arduous
and impractical . . . procedure(s)” (p. 330). Similarly Baron, Kaufman, and Stauber (1969) found that instructions had a
major influence on the establishment and maintenance of behaviour of college students.
Only a few studies have examined this empirically, comparing the effects of video with and without narration, and
producing some evidence for video plus narration being superior to video modelling without narration. Mechling and Collins
(2012), working with participants with moderate intellectual disabilities, reported that video with narration was more
effective for three of their four participants. In 2013, Smith, Ayres, Mechling, and Smith (2013) reported a systematic
replication of the earlier study working with four participants, all with diagnoses of ASD. They reported that video modelling
with narration was more effective for one of these participants the two procedures being equally effective with the others,
and more efficient (errors to criterion) for two of the four. All of these participants were reported to have enjoyed learning
new skills using the video models, but indicated a preference for videos with narration. Similarly their teachers opined that
though it would be easier to develop video models than video models with narration, they would rather use videos with
narration to teach other complex tasks. Mechling and Collins (2012) were working with young adults with intellectual
disabilities enrolled in a school transition program, and Smith et al. (2013) with adolescents with ASD. Further research
exploring the differential effects of narration combined with video modelling for other populations is clearly warranted.
Consequently, with a view to examining the relative effectiveness of video modelling alone and video modelling with
embedded narration the present study examined the effectiveness of a video modelling-based intervention to improve the
social-communication skills of four typically developing elementary school children who had been identified as having
social-communication skill deficits relative to their peers. Based on the results of previous studies it was expected that video
modelling with embedded narration would be faster, or more efficient, and produce a greater effect than video modelling
without such narration.
102 A. Anderson et al. / International Journal of Educational Research 87 (2018) 100–109

2. Methods

2.1. Participants and setting

Four students, Chris, Grant, Fred, and John, all pseudonyms, participated in the study. The participants’ teachers identified
each as having social-communication skill deficits relative to classroom peers. All four boys attended an open plan classroom
in an inner-city government primary school. The class was termed a learning community as it included three teachers and a
composite of approximately 60 students from Grades 1–3.
Chris was a Grade 2 student aged 7 years. Grant (8 yrs), Fred (9 yrs), and John (9 yrs) were all in Grade 3. Even though all
were competent native speakers of English they were described by their teacher as not following the conventions of
conversational turn taking and maintaining eye contact with conversational partners.
It should be noted that none of these participants had previously received targeted interventions to address their social-
communication skill deficits. However, in each case their teachers considered their social skill deficits to be interfering with
peer relationships, academic performance and, with Grant and Fred, their participation in, and happiness at school.
Two other students in the class, identified by their teachers as having age appropriate social-communication skills were
observed to provide a normative sample against which the target students’ behaviour was compared. Finally, seven Grade
6 students from the same school, selected by teachers as having expressed an interest in drama, participated as peer models
and coaches in the creation of the videos to be used in the intervention. Other than acting as models/coaches in the videos
these students had no other involvement in the study.

2.2. Materials

Materials used in this study included; a purpose-developed student profile form; a parent interview schedule;
intervention videos; digital video recorder; intervals, a time signalling app for smart phones; coding sheets; and post-
intervention questionnaire and interview schedule for teachers and students (copies of all forms available from the
corresponding author on request).
A teacher-completed student profile consisted of six yes/no questions relating to the student’s functioning in the school
environment e.g. “In your opinion, does this deficit negatively impact on relationships with peers?” and their past history of
intervention e.g. “Has this student ever had intervention to address this deficit? Was it successful?” Room for comments was
provided including space for teachers to identify any behaviour that they considered to be a result of a deficit in the identified
skills.

2.2.1. Interview schedules

Interview schedules for parents of intervention participants consisted of four open ended questions. The interview
focussed on parents’ perceptions of their child’s social ability e.g. “Have you ever been concerned about your child’s social
interactions with yourself or others?” Parents were also asked about any diagnosed conditions that may impact on their
child’s ability to gain age appropriate social skills.

2.2.2. Videos
Videos used in the interventions were four, 2–3 min films featuring the peer models demonstrating the targeted
behaviours, with and without narrated instructions as per the two treatment conditions. Class laptops were used to present
the video models to the participants during the treatment phases.

2.2.3. Digital video camera

A digital video camera, the intervals app, and data recording coding sheets were used during both the baseline and
treatment phases to record the behaviour of the target and normative-sample participants. Intervals is designed for interval
recording and produces auditory signals at set intervals for a pre-determined period of time.

2.2.4. The post-intervention questionnaire and interview

The post-intervention questionnaire and interview schedule for teachers included a questionnaire containing five yes/no/
maybe questions with space for comments. The questionnaire rated how successful the teacher believed the intervention to
have been, and whether they thought it would be a practical method for use in a regular classroom. The interview schedule
contained 11 open-ended questions to further investigate the teachers’ perception of the intervention. Questions included:
“Was there any significant difference that you noticed in the students?” and “Are there barriers that you think might prevent
an intervention such as video modelling being used in schools? If so what are they?”
The post-intervention questionnaire and interview schedules for intervention participants included a five item
questionnaire, using a Likert scale consisting of four cartoon faces indicating yes/maybe/no/don’t know. Participants rated
how they felt about the intervention and whether they were now finding their social interactions more fluid and enjoyable.
The interview schedule contained six questions including: “What did you learn?” and “Do you think it was helpful? Why/
why not?”
 A. Anderson et al. / International Journal of Educational Research 87 (2018) 100–109 103

2.3. Target behaviours

Two target skills were identified by the school principal in consultation with the classroom teachers independent of the
researchers; appropriate conversational turn taking, and correct body orientation while in a conversation (conversational
posture). Appropriate conversational turn taking was operationalised as two target behaviours: (a) Waiting for others to
finish talking, and; (b) pausing to allow the other person to speak as evidenced by switches in conversation. Conversational
posture was operationalised as two target behaviours: (a) orienting the body toward the other person (students were coded
as correct if they had full or partial orientation towards the person they were talking to; e.g.: if sitting next to each other and
the student had their shoulder back so that they were angled towards the person they were talking to this was coded as a
demonstration of correct body language), and; (b) making eye contact for any length of time at least once during the five
second recording interval.
With both target skills the two identified behaviours had to occur in order for a student’s behaviour to be coded as a
correct demonstration of the skill within that observation interval. For example if a participant was looking at the floor but
was directly in front of the person speaking, this would be coded as incorrect as only one of the behaviours was being
exhibited. Daily total intervals coded as correct were converted to a percentage of intervals in which conversational turn
taking, correct conversational posture occurred.

2.4. Treatment conditions

The two treatments were video modelling (VM) and video modelling with embedded narration (VM-N). VM consisted of a
video featuring a brief narrated definition of the skill, followed by five clips featuring models performing the desired
behaviour, without any rules or instructions provided. VM-N consisted of a video featuring one of the models who presented
clearly defined rules for each behaviour. For appropriate conversational turn taking the rules were: 1) “take turns by letting
the other person talk”; 2) “listen to what the other person has to say”; 3) “do not interrupt the other person”; 4) “do not talk
all the time and not let the other person join in”. For conversational posture the rules were; 1) “look at the person”; 2) “face
our bodies toward that person”; 3) don’t turn away or start fiddling with something else”; 4) “don’t get really close to the
person or stand a long way away". This narration was followed by two of the VM clips showing models performing the
desired behaviour. The instructions were reiterated after each clip, with the final presentation including the instruction in
written format on the screen.
Allocation of target behaviours to treatment conditions was counterbalanced across participants. Chris and Fred were
assigned VM for appropriate conversational turn taking and VM-N for Conversational Posture, and Grant and John were given
VM for Conversational Posture and VM-N for appropriate conversational turn taking. Participants flipped a coin to determine
which treatment they undertook in each session. The videos showed older students from the school performing the desired
skills across several settings (in the playground and in the class room). Videos were produced in the term prior to the study
commencing.

2.5. Social validity

Social validity was assessed in three ways; (1) teachers’ perception of the viability of the intervention; (2) rating of overall
enjoyment of school for participants, and; (3) participants’ observed social competencies relative to that of normative peers.
Perceived viability of the intervention, and students’ enjoyment of school following intervention was assessed with a survey
and interview conducted by the researchers after the completion of the intervention with both participants and teachers.
Participants’ observed social competencies were compared with those of two normative peers observed during baseline and
intervention phases (see Finn & Sladeczek, 2001).

2.6. Research design

An alternating treatments design was used to assess the relative effectiveness of video modelling (VM) and video
modelling with embedded narration (VM-N). The study consisted of three phases, a baseline, an intervention phase in which
the alternating treatment design was implemented, and a subsequent phase in which a single intervention was implemented
to control for multi-treatment effects. Follow up observations were also obtained three weeks later.

2.7. Observation method

Partial interval recording was used to measure appropriate conversational turn taking and conversational posture. Five
recordings were taken each observation session, each recording was one minute in duration and began once the student
successfully initiated or was engaged in a conversation. Observation intervals were 5 s long, yielding 12 observation intervals
per recording.
All observation sessions were video recorded and coding was performed subsequently from these recordings. Two
observers independently coded 31% of observation sessions, distributed evenly across the three phases of the study. Inter-
104 A. Anderson et al. / International Journal of Educational Research 87 (2018) 100–109

observer Agreement (IOA) was calculated interval by interval (occur plus non-occur). An average agreement of 87% was
attained.

2.8. Procedure

Ethics approval for this study was gained from the appropriate University and Government Departmental Institutional
Review Boards prior to the commencement of the study. Schools were contacted via email with information relating to the
study and expressions of interest to participate in the study were sought. Written informed consent for the study was
obtained from the participating school, teachers, the guardians or parents of all participating children, and the students.
The school and participating teachers assisted in the identification of potential participants and, on completion of their
recruitment, the teachers completed the student profile information for each participant. Parallel VM and VM-N videos were
constructed for each of the two target skills (appropriate conversational turn taking and conversational posture).
Baseline observational data was collected for all four participants in class while they were engaged in typical
conversations with teachers and classmates. Throughout baseline and intervention phases, normative data were also
obtained by observing conversations of two control students selected as representative of classroom norms. The same
observation procedures were applied and the average of their rates of responding on the target behaviours served as
normative data for comparison with the intervention group.
Prior to the intervention phase each student was allocated a treatment condition for each skill (see Table 1). Students
were shown the intervention video immediately prior to a lesson in which they would have opportunity to exhibit the skills
being taught. Students watched the intervention video on their laptop computers. Reinforcement for attending to the video
was provided in the form of 3 mins playing a video game of their choice before returning to the classroom activity. An
observation session was conducted in the lesson immediately following. Students’ behaviour was video recorded in the
classroom and on completion of the session Chris and Grant were praised for correctly displaying the target behaviours.
Students underwent three intervention sessions per week for five weeks including two final weeks of the optimal treatment
condition. Follow up observations were also obtained 3 weeks after the optimal treatment condition ended. On completion
of the study students and teachers were interviewed about the study and asked to complete the post-intervention
questionnaire.

3. Results

Figs. 1–4 show the proportion of observation intervals in which, during conversations between pairs of students in the
class, correct conversational posture and appropriate conversational turn taking occurred across baseline, alternating
treatments, optimum treatment and follow up phases for Chris, Grant, Fred, and John respectively. In addition the average
level of occurrence of these two dependent variables (DV) for the two normative students is depicted. Across the entire study
these control students on average displayed correct conversational posture on 69% of observational intervals and appropriate
conversational turn taking exchanges on 66% of these five sec intervals.
The figures show that for all participants, baseline rates of responding on target behaviours was consistently below
normative levels with average scores for correct conversational posture of 33% (Chris), 51.2% (Grant), 25.5% (Fred), and 55%
(John); and for appropriate conversational turn taking: 38.9% (Chris), 40.5% (Grant), 32.2% (Fred), and 36.7% (John). The
within baseline data paths for both DVs are relatively stable for Chris and John, though Fred showed greater variability for
appropriate conversational turn taking and Grant had considerable within phase variability with both turn taking and
conversational posture
Once the alternating treatments phase was introduced, responding on both behaviours increased for all participants such
that by the second half of this phase all were performing at or about normative levels, and continued or improved upon these
performances in the following optimal treatment phase and follow up observations.
Possible differential effects of video modelling with and without embedded narration may be ascertained by inspecting
the separate data paths across the Alternating Treatment phase in each of the figures together with the associated baseline
rates of occurrence of the target behaviours. The degree of separation in the data paths is generally understood to reflect
differential treatment effectiveness (DTE) (Cooper, Heron, & Heward, 2007), though in our case differences in speed of
impact reflected in a steep learning gradient early in the phase, within phase variability, and attained levels of performance
at the end of the intervention phase may all reflect on DTE.

Table 1
Allocation of target behaviours by intervention type.

VM VM-N
Conversational Posture Granta Chrisa
John Fred
Appropriate conversational turn taking Chrisa Granta
Fred John
a
Praised for correctly displaying target behaviours.
 A. Anderson et al. / International Journal of Educational Research 87 (2018) 100–109 105

Baseline Alternating Treatment VM Only Follow-up

100
Conversational Posture

% Displayed Target Behaviour

Conversational
Posture (VM-N)
80 Normm

60
Taking turns
Talking Norm
40
Taking turns
talking
20 (VM)

0
0 5 10 15 20 25 30

Sessions
Fig. 1. Incidents of desired behaviour for body language when talking and taking turns talking for Chris across baseline, alternating treatments, VM only and
follow up phase.

Chris (Fig. 1) was exposed to the VM procedure (white circles in the Figures) for taking turns talking while receiving VM-N
(black triangles) for conversational posture. Some separation between the data paths is evident with only one of the VM-N
data points being below the VM trend line. The initial four data points in both paths particularly when considered together
with their associated baselines suggest a more rapid initial effect for VM-N, with conversational posture performance
being at class normative levels by the third intervention session while appropriate conversational turn taking (with the VM
intervention) continued at baseline levels. Though both data trajectories are upwards the final three data points for
conversational posture (VM-N) are separate to and above the conversational turn taking data path.
Grant (Fig. 2) was given the VM procedure for conversational posture while receiving VM-N for appropriate
conversational turn taking. A similar pattern to that seen with Chris is evident in these data. A separation between the two
data paths is evident VN-N outperforming VM in seven of nine sessions. Initial benefits of VM-N are also apparent with a
greater shift from baseline levels in the first four data points. Both trajectories are positive such that by the final three
observations both were again at or above the established normative level, though greater variability is evident in the taking
turns talking (VM-N) data at this stage.
Fred (Fig. 3) received VM for appropriate conversational turn taking and VM-N for conversational posture. Both data paths
trend upwards across the ATD phase and overlap substantially. Conversational turn taking (VM-N) shows greater initial
treatment gains relative to baseline though both show similar trajectories, and both end at about the target normative levels.
The data paths for John (Fig. 4) also both trend positively and overlap markedly, starting and ending at similar points
relative to the normative data. However conversational turn taking (VM-N) was substantially below conversational posture

Baseline Alternating Treatment VM Only Follow-up

100
% Displayed Target Behaviour

Conversational
Posture
80 Norm

60
Taking turns
Taking turns talking Norm
40 talking
(VM-N)
Conversational Posture
20 (VM)

0
0 5 10 15 20 25 30

Sessions
Fig. 2. Incidents of desired behaviour for body language when talking and taking turns talking for Grant across baseline, alternating treatments, VM only
and follow up phase.
106 A. Anderson et al. / International Journal of Educational Research 87 (2018) 100–109

Fig. 3. Incidents of desired behaviour for body language when talking and taking turns talking for Fred across baseline, alternating treatments, VM-N only
and follow up phase.

in baseline and a stronger initial impact for this VM-N condition is again evident. As with Grant (Fig. 2) considerable within
phase variability is evident in the taking turns talking (VM-N) data path.
As observed DTEs were relatively small, logistics largely determined which intervention was allocated to each participant
in the final single intervention phase. Chris and Grant watched VM only videos for both target behaviours through this phase,
Fred and John being presented with the VM-N condition. Regardless of intervention type, the data for all participants across
this phase show for both behaviours increased stability, with little within phase variability, all at or above the normative
benchmarks, and with the exception of Fred and taking turns, the subsequent follow up observation. Though the effect of the
provision of praise for two participants (Chris and Grant) was not assessed experimentally it is worth noting that the addition
of praise to the intervention does not appear to have had any differential effect.
Chris, Grant, John and Fred all completed the social validity interview and questionnaire. Grant, John and Fred all
indicated that they found the videos helpful, though they did indicate, perhaps not unexpectedly, that by the end of the
process they were finding the video viewing boring. Chris reported he did not learn anything from the videos, therefore he
did not find them helpful. Fred, Chris and Grant reported that they now found school more enjoyable, that their friends were
being nicer. Chris and Fred also felt that their relationship with their teachers had improved. Unexpectedly, when asked Chris
and John both said they preferred the video models without narration over the video models with narration. John said ‘I like
lots of examples, I’m a visual learner, I prefer to watch examples’.

Fig. 4. Incidents of desired behaviour for body language when talking and taking turns talking for John across baseline, alternating treatments, VM-N only
and follow up phase.
 A. Anderson et al. / International Journal of Educational Research 87 (2018) 100–109 107

The teachers agreed that the program was not overly intrusive to general classroom processes but suggested that
resourcing and implementing such interventions in the context of their regular classroom activities could be problematic.
Interestingly, while teachers had nominated these students because of their skill deficits, rather than to reduce disruptive
behaviour, all thought the intervention was not very successful because the participants’ behaviours had not reduced
sufficiently. One teacher commented that she thought that the intervention may be more beneficial with younger students;
that for VM to be useful in the mainstream setting the target skills needed to be more complex and integrated into the
classroom environment, that perhaps the skills were too simplistic to amount to any significant behaviour change for Grade
3 boys.

4. Discussion

The purpose of the study was to investigate the effects of a video modelling based interventions on the social-
communication skills of four typically developing elementary school children and to identify whether video modelling was
more effective with or without embedded narration. Narration was expected to be associated with faster and greater effects
on the target behaviours. Results in the main confirmed expectations. In baseline the participants were all shown to be
performing substantially below normative levels within the class with reference to both appropriate conversational posture
and conversational turn taking. Consistent with previous research which has shown video modelling to be an effective
strategy for teaching social skills (Charlop & Milstein, 1989; Charlop-Christy & Daneshvar, 2003; Nikopoulos & Keenan, 2004,
2007; Shukla-Mehta et al., 2010), the introduction of both video modelling interventions in the ATD phase was associated
with gains such that after eight or nine intervention sessions all four participants were performing at or above the normative
levels on both dependent variables regardless of which intervention variant they were exposed to. Despite the positive
trajectories evident in all cases through the ATD phase, considerable within phase variability was evident, particularly with
Fred and John. It is noteworthy therefore that with all four participants both behaviour trajectories stabilised at above
normative levels in the final single-intervention phase regardless of which intervention was utilised. Alternative
explanations for this pattern include maturation (the introduction of this phase was not staggered across participants) or the
achievement of some optimal performance level/ceiling effect, and the possibility of multi-treatment effects in the ATD
phase impacting the target behaviours in unknown ways. Further research exploring this is clearly justified.
Gains in the participants’ conversational posture and appropriate conversational turn taking were maintained at three-
week follow-up with performances in excess of class norms, except for Fred’s turn taking, though even this was a substantial
improvement over his baseline performance. The maintenance of the target behaviour in conjunction with a lack of a
differential effect of praise versus no praise suggests that the improved performance of the target behaviours came into
contact with natural contingencies of reinforcement.
In partial confirmation of our expectations, there is also evidence of narration embedded into the video differentially
boosting both the efficiency and, to a lesser extent, the overall effectiveness of the video modelling intervention on our
participants’ social-communication skills. The efficiency, or speed of effect, of the two procedures was determined by
analysing the rate of change evident in the first half of the ATD intervention phase. In all cases, and irrespective of the target
behaviour allocated to the different treatments (experimentally counterbalanced) the version of the video with narration
included was associated with more rapid initial gains than was the behaviour which had been exposed to the video model
without narration. This finding therefore confirms that of Smith et al. (2013) who, working with young adults with ASD,
found that for two of their four participants videos with narration were more efficient while for the other two there was no
apparent advantage.
Effectiveness, or overall size of the gains made, was determined by analysing the separation of the two data paths in the
ATD phase of the study as well as examining the performance levels achieved by the end of this phase. Chris and Grant, with
intervention type and target behaviours counterbalanced, both had clear separation evident in their data paths with the
video model with narration outperforming the video without narration. In Chris’ case a higher overall level of performance
was achieved in the VM-N condition at phase end. With both Fred and John no such separation in the data paths is evident,
both data paths show substantial overlap and no differences in level are apparent at the end of this phase. Thus for two of our
four participants there is evidence that narration embedded into the videos was associated with enhanced intervention
effectiveness while for the other two there were no apparent differential effects. This cannot be explained by the addition of
praise for the first two participants, for whom VM and VM-N had a differential effect as their performance of the target
behaviour met with praise in both intervention conditions and for both target behaviours. These findings substantiate
conclusions by both Mechling and Collins (2012), and Smith et al. (2013) who both reported similar effectiveness benefits
from narrations in video modelling, while extending their findings to typically developing elementary school children who
had been identified as having social-communication skill deficits relative to their peers.
During the exit interview three of the four participants said they had enjoyed learning through video modelling, less so
with the repeated viewings the fourth stated that he had not learned anything. Three indicated that they were finding their
relationships with friends and teachers had improved possible verification of the social validity of the target behaviours.
Unexpectedly, two of our participants indicated a preference for the video models without narration. The explanation they
provided alluded to learning style preferences. Future research exploring individual specific preferences, optimal frequency
and distribution (massed vs spaced learning) of video modelling interventions may well increase our understanding of these
dimensions.
108 A. Anderson et al. / International Journal of Educational Research 87 (2018) 100–109

Interestingly, and contrary to expectations, the participants’ teachers all thought the interventions to be ineffective. There
are a number of possible explanations for this. Perhaps it indicates a misunderstanding of the goals of the intervention. Even
though they nominated participants based on perceived deficits in social skills (and hence an intervention that targeted skills
increase), their nomination might have reflected an actual expectation that the intervention would result in a reduction of
disruptive behaviour. Nonetheless, the data suggest that the study had high social validity for the target students in terms of
the outcomes, given the normative data, which suggests that all participants indeed had skill deficits in the chosen target
behaviours, and the literature, which suggests that addressing these skill deficits would be of benefit to the students.

5. Conclusion

Deficits in social communication skills are associated with poor developmental, educational and behavioural outcomes.
Addressing such deficits early and in an efficient manner is important. Video-modelling interventions are promising in this
regard. However to date few studies have investigated the relative effectiveness or efficiency of different video modelling
variants. The current paper compared the effectiveness and efficiency of video modelling with and without embedded
narration. Both intervention types were effective in achieving meaningful levels of behaviour change. For two of the four
participants the video modelling with narration condition was associated with a more rapid skill acquisition, making it a
more efficient intervention. Given that no detrimental effects of the video modelling with narration variant were observed it
can be concluded that where a rapid behaviour change is desired this condition would be the intervention of choice.

References

Apple, A. L., Billingsley, F., & Schwartz, I. S. (2005). Effects of video modeling alone and with self-management on compliment-giving behaviors of children
with high-functioning ASD. Journal of Positive Behavior Interventions, 7(1), 33–46. http://dx.doi.org/10.1177/10983007050070010401.
Ayllon, T., & Azrin, N. H. (1964). Reinforcement and instructions with mental patients. Journal of the Experimental Analysis of Behavior, 7, 327–331. http://dx.
doi.org/10.1901/jeab.1964.7-327331.
Baron, A., Kaufman, A., & Stauber, K. A. (1969). Effects of instructions and reinforcement-feedback on human operant behaviour maintained by fixed-interval
reinforcement. Journal of the Experimental Analysis of Behavior, 12, 701–712. http://dx.doi.org/10.1901/jeab.1969.12-701.
Bellini, S., & Akullian, J. (2007). A meta-analysis of video modeling and video self-modeling interventions for children and adolescents with autism spectrum
disorders. Exceptional Children, 73, 264–287. http://dx.doi.org/10.1177/001440290707300301.
Bellini, S., Peters, J. K., Benner, L., & Hopf, A. (2007). A meta-analysis of school-based social skills interventions for children with autism spectrum disorders.
Remedial and Special Education, 28(3), 153–162. http://dx.doi.org/10.1177/07419325070280030401.
Charlop, M. H., & Milstein, J. P. (1989). Teaching autistic children conversational speech using video modeling. Journal of Applied Behavior Analysis, 22(3), 275–
285. http://dx.doi.org/10.1901/jaba.1989.22275.
Charlop-Christy, M., & Daneshvar, S. (2003). Using video modelling to teach perspective taking to children with autism. Journal of Positive Behavior
Interventions, 5(1), 12–21. http://dx.doi.org/10.1177/10983007030050010101.
Charlop-Christy, M., Le, L., & Freeman, K. (2000). A comparison of video modelling with in vivo modelling for teaching children with autism. Autism and
Developmental Disorders, 30, 535–552. http://dx.doi.org/10.1023/A:1005635326276.
Cooper, J. O., Heron, T. E., & Heward, W. L. (2007). Applied behavior analysis. Essex, England: Pearson Education Limited.
Donno, R., Parker, G., Gilmour, J., & Skuse, D. H. (2010). Social communication deficits in disruptive primary-school children. The British Journal of Psychiatry,
196(4), 282–289. http://dx.doi.org/10.1192/bjp.bp.108.061341.
Dwyer, K., Osher, D., & Warger, C. (1998). Early warning, timely response: a guide to safe schools. Washington, DC: U.S. Department of Education.
Elliott, S. N., Malecki, C. K., & Demaray, M. K. (2001). New directions in social skills assessment and intervention for elementary and middle school students.
Exceptionality, 9(1–2), 19–32. http://dx.doi.org/10.1207/s15327035ex091&2_3.
Finn, C. A., & Sladeczek, I. E. (2001). Assessing the social validity of behavioural interventions: a review of treatment acceptability measures. School
Psychology Quarterly, 16(2), 176. http://dx.doi.org/10.1521/scpq.16.2.176.18703.
Gallagher, T. M. (1993). Language skill and the development of social competence in school-age children. Language, Speech, and Hearing Services in Schools, 24
(4), 199–205. http://dx.doi.org/10.1044/0161-1461.2404.199.
January, A. M., Casey, R. J., & Paulson, D. (2011). A meta-analysis of classroom-wide interventions to build social skills: do they work? School Psychology
Review 40, 242–256. http://go.galegroup.com.ezproxy.lib.monash.edu.au/ps/i.do?
ty=as&v=2.1&u=monash&it=search&s=RELEVANCE&p=AONE&qt=TIA%20meta-analysis%20of%20classroom-wideSP242IU2SN0279-
6015VO40&lm=&sw=w&authCount=1.
Litras, S., Dennis, W., Moore, D. W., & Anderson, A. (2010). Using video self-modelled social stories to teach social skills to a young child with autism. Autism
Research and Treatment. http://dx.doi.org/10.1155/2010/834979.
McClelland, M. M., Morrison, F. J., & Holmes, D. L. (2000). Children at risk for early academic problems: the role of learning-related social skills. Early
Childhood Research Quarterly, 15(3), 307–329. http://dx.doi.org/10.1016/s0885-2006(00)00069-7.
Mechling, L. C., & Collins, S. (2012). Comparison of the effects of video models with and without verbal cuing on task completion by young adults with
moderate intellectual disabilities. Education and Training in Autism and Developmental Disabilities, 47, 223–235.
Moore, D. W., Anderson, A., Treccase, F., Deppeler, J., Furlonger, B., & Didden, R. (2013). A video-based package to teach a child with autism spectrum disorder
to write her name. Journal of Developmental and Physical Disabilities., 25(5), 493–503. http://dx.doi.org/10.1007/s10882-012-9325-x.
Nikopoulos, C. K., & Keenan, M. (2004). Effects of video modeling on social initiations by children with autism. Journal of Applied Behavior Analysis, 37(1), 93–
96. http://dx.doi.org/10.1901/jaba.2004.37-93.
Nikopoulos, C. K., & Keenan, M. (2006). Video modelling and behaviour analysis: a guide for teaching social skills to children with autism. London: Jessica
Kingsley Publishers.
Osofsky, H. J., & Osofsky, J. D. (2001). Violent and aggressive behaviors in youth: a mental health and prevention perspective. Psychiatry, 64(4), 285–295.
http://dx.doi.org/10.1521/psyc.64.4.285.18603.
Quinn, M. M., Kavale, K. A., Mathur, S. R., Rutherford, R. B., & Forness, S. R. (1999). A meta-analysis of social skill interventions for students with emotional or
behavioral disorders. Journal of Emotional and Behavioral Disorders, 7(1), 54–64. http://dx.doi.org/10.1177/106342669900700106.
Rayner, C., Denholm, C., & Sigafoos, J. (2009). Video-based intervention for individuals with autism: key questions that remain unanswered. Research in
Autism Spectrum Disorders, 3, 291–303. http://dx.doi.org/10.1016/j.rasd.2008.09.001.
Segrin, C., & Flora, J. (2000). Poor social skills are a vulnerability factor in the development of psychosocial problems. Human Communication Research, 26(3),
489–514. http://dx.doi.org/10.1093/hcr/26.3.489.
 A. Anderson et al. / International Journal of Educational Research 87 (2018) 100–109 109

Shukla-Mehta, S., Miller, T., & Callahan, K. J. (2010). Evaluating the effectiveness of video instruction on social and communication skills training for children
with autism spectrum disorders: a review of the literature. Focus on Autism and Other Developmental Disabilities, 25, 23–36. http://dx.doi.org/10.1177/
1088357609352901.
Skinner, B. F. (1974). About behaviorism. New York: Alfred A. Knopf.
Smith, M., Ayres, K., Mechling, L., & Smith, K. (2013). Comparison of the effects of video modeling with narration vs. video modeling on the functional skill
acquisition of adolescents with autism. Education and Training in Autism and Developmental Disabilities, 48(2), 164–178.
Spence, S. H. (2003). Social skills training with children and young people: theory, evidence and practice. Child and Adolescent Mental Health, 8(2), 84–96.
http://dx.doi.org/10.1111/1475-3588.00051.
Wentzel, K. R. (1991). Relations between social competence and academic achievement in early adolescence. Child Development, 62(5), 1066–1078. http://dx.
doi.org/10.1111/j.14678624.1991.tb01589.x.