Using Point-of-View Video Modeling to Teach Play to Preschoolers With Autism

Abstract

This study evaluated the effectiveness of point-of-view video modeling in teaching selected toy-play skills to two preschoolers with autism. This type of modeling involved the experimenters carrying or holding the video camera at eye level (from the child's perspective) and without recording models (persons]) to show the environment as a child would see it when he or she was performing the targeted skills. The researchers used a multiple-probe design across two children and two behaviors to evaluate the effect of the point-of-view modeling on the children's acquisition and maintenance of play actions. They used generalization probes to assess the degree to which the participants used the new skills across novel toys and during classroom activities. The results indicated that point-of-view modeling was an effective tool for teaching toy-play actions to preschoolers with autism. The authors discuss the extension of current video modeling research and implications for home and school interventions.

Full text

TECSE 26:2 83–93 (2006) 83
Using Point-of-View Video Modeling to Teach
Play to Preschoolers With Autism
T
his study evaluated the effectiveness of point-of-view video modeling in teaching
selected toy-play skills to two preschoolers with autism. This type of modeling
involved the experimenters carrying or holding the video camera at eye level
(from the child’s perspective) and without recording models (persons]) to show the en-
vironment as a child would see it when he or she was performing the targeted skills.
The researchers used a multiple-probe design across two children and two behaviors to
evaluate the effect of the point-of-view modeling on the children’s acquisition and
maintenance of play actions. They used generalization probes to assess the degree to
which the participants used the new skills across novel toys and during classroom ac-
tivities. The results indicated that point-of-view modeling was an effective tool for
teaching toy-play actions to preschoolers with autism. The authors discuss the extension
of current video modeling research and implications for home and school interventions.
Jeffrey F. Hine
and
Mark Wolery
Peabody College of
Vanderbilt University
Address: Jeffrey F. Hine, 27730 Lake Jem Rd., Mount Dora, FL 32757; e-mail: jeffhine@hotmail.com
Play involves a number of diverse and complex behav-
iors and is viewed as central to children’s development
(Jordan & Libby, 1997). Young children with autism of-
ten are described as having impairments in play (Jarrold,
2003; Jarrold, Boucher, & Smith, 1993; Jordan & Libby,
1997; Ungerer & Sigman, 1981; Van Berckelaer-Onnes,
2003), and they often do not develop spontaneous play
(Van Berckelaer-Onnes, 2003). Libby, Powell, Messer,
and Jordan (1988) showed that children with autism had
similar levels of functional play as children with Down
syndrome and young children with typical development,
but they had impairments in symbolic (pretend) play.
Toy play is an essential forerunner for symbolic play
(Van Berckelaer-Onnes, 2003) and perhaps for social
play (Pierce-Jordan & Lifter, 2005).
Learning through observation and imitation of others
can account for acquisition of some behaviors, and sim-
ply watching another child receive reinforcement (vicar-
ious reinforcement) for a behavior often increases rates
of that behavior (Bandura, 1965). In vivo modeling (using
live models) is effective in teaching various developmental
skills to children with autism (Charlop, Schreibman, &
Tryon, 1983), including appropriate play (Stahmer, In-
gersoll, & Carter, 2003). For modeling strategies to be
effective, the child with autism must attend to another
person, must attend to that person’s actions, and must be
imitative. The developmental course of imitation in chil-
dren with autism is unclear. Dawson and Osterling (1997)
suggested that imitation is a fundamental mechanism for
learning and, in particular, a precursor for learning so-
cial skills. Young children with autism seem to imitate
actions on toys more readily if sensory feedback occurs
than if it is absent (Ingersoll, Shreibman, & Tran, 2003),
which suggests a lack of motivation for acting on toys;
thus, studies of interventions featuring supplementary
reinforcement contingencies are needed.
Technology, particularly video modeling, provides
another way to promote observational learning. Video
modeling interventions can include any number of com-
ponents, but they generally involve the following:
1. edited images of appropriate or new be-
havior shown on a monitor to a child;
2. repeated clips of the same behavior or mul-
tiple exemplars of the behavior shown to
the participant;
3. discrete practice sessions or role-playing of
the new skills;
4. assessment of skill generalization (e.g.,
probes across settings, people, or materi-
als); and
5. periodic review of tapes, if needed.

84 Topics in Early Childhood Special Education 26:2
Stahmer et al. (2003) suggested that video modeling
may succeed when other methods have failed, especially
for children who do not learn when social interactions
are involved. Researchers have used video modeling to
teach individuals with autism such skills as purchasing
(Alcantara, 1994; Haring, Kennedy, Adams, & Pitts-
Conway, 1987), conversation (Charlop & Milstein, 1989;
Sherer et al., 2001), perspective taking (Charlop-Christy
& Daneshvar, 2003; LeBlanc et al., 2003), spelling (Kin-
ney, Vedora, & Stromer, 2003), and daily living (Shipley-
Benamou, Lutzker, & Taubman, 2002). Video modeling
also can be used to promote play skills in children with
autism. Taylor, Levin, and Jasper (1999) used video mod-
eling to teach two children with autism to engage in
play-related statements with their siblings. D’Ateno,
Mangiapanello, and Taylor (2003) used video modeling
to increase children’s use of complex play sequences.
Video modeling has several advantages over in vivo
modeling, including (a) being more time and cost effi-
cient; (b) depicting a variety of naturalistic settings that
would be difficult to create in vivo in a clinic or class-
room; (c) having more control over the model, because
the tape can be recreated until the desired scene is ob-
tained; (d) presenting the model repeatedly without the
person modeling the desired behavior being present; and
(e) reusing videotapes with other individuals, which means
that more children can be treated (Charlop-Christy, Le,
& Freeman, 2000). Zihni and Zihni (2005) pointed out
that video modeling takes advantage of the visual skills
of children with autism, eliminates the social context
that occurs during in vivo modeling, increases the pre-
dictability and controllability of the model, and allows
extraneous features to be filtered out, which makes learn-
ing easier. Video modeling also can promote generaliza-
tion by exposing participants to a sufficient number of
exemplars and response variations. Haring et al. (1987)
demonstrated generalization of purchasing skills to three
community settings using video modeling with adults with
autism and noted that video modeling was an effective
way to promote generalization. Charlop and Milstein
(1989) demonstrated generalization of conversational
speech involving untrained topics (new toys, unfamiliar
persons, siblings, peers, and other settings).
Video modeling interventions have used other-as-
model (adults or peers doing target behaviors) and self-
as-model (participants act as their own models) formats.
In the latter instance, the video is edited to show the
child performing only the desired behaviors. Self-as-
model videos have been effective in increasing respond-
ing behaviors (Buggey, Toombs, Gardener, & Cervetti,
1999), requesting (Hepting & Goldstein, 1996; Wert &
Neisworth, 2003), and conversation skills (Sherer et al.,
2001) in children with autism. Sherer et al. (2001) com-
pared the efficacy of self versus other video modeling and
found no overall differences in the rate of acquisition.
An alternative to other- and self-models is observer point-
of-view modeling, which involves recording tapes using
the perspective of the person who is the target of the in-
tervention and showing him or her the target behavior
from the viewer’s vantage point but without showing the
entire person who is modeling the behavior. Schreibman,
Whalen, and Stahmer (2000) used point-of-view model-
ing to reduce problematic behaviors in difficult transi-
tions by taping what the transition would look like as
the child did it (e.g., taking different routes through a
mall). Shipley-Benamou et al. (2002) taught daily living
skills to children with autism via this procedure. They
suggested that point-of-view modeling eliminates the
need for staging performances and editing tapes, which
can be time-consuming in self-as-model videotapes. In
addition, using the child’s point of view avoids having to
identify which model characteristics (e.g., peer or adult,
familiar or unfamiliar) are most effective when teaching
new skills.
To date, researchers have had success in teaching
new skills to participants using other persons or the child
as models in video modeling studies, and positive ef-
fects have been most evident for participants older than
preschool age. In this study, we attempted to extend the
literature by teaching symbolic (pretend) play to preschool-
age children with autism and evaluated the use of point-
of-view video modeling. Only two studies (Schreibman
et al., 2000; Shipley-Benamou et al., 2002) used point-
of-view modeling, and only five studies included children
with autism age 5 years of age or younger. Although
some studies (e.g., Nikopoulos & Keenan, 2004) focused
on play skills, only one study addressed symbolic (pre-
tend) play (D’Ateno et al., 2003). In this study, our goal
was to evaluate the effectiveness of point-of-view video
modeling on the play of preschoolers with autism using
materials during sensory bin activities. We posted the fol-
lowing research questions:
1. Will preschoolers with autism readily imi-
tate actions seen through point-of-view
video modeling?
2. Will any acquired skills generalize to chil-
dren’s classroom sensory activities and
across untrained materials?
METHOD
Participants
Our participants were two girls. Christine was 30 months
old at the beginning of the study, and Kaci was 43 months
old. Both girls were verbal, but their communication, so-
cial, and play skills had been targeted as intervention
goals. Both participants met the criteria for autism of the
Diagnostic and Statistical Manual of Mental Disorders–

Point-of-View Video Modeling 85
Fourth Edition (DSM-IV; American Psychiatric Asso-
ciation, 1994). According to teacher reports, both chil-
dren participated regularly during sensory bin activities
and consistently contacted materials. Christine’s sensory
bin play often involved repetitive movements, usually
with one object. Kaci’s play involved repetitive manipu-
lation of toys and stereotypic behavior. Both girls needed
instruction on how to use the toys. From teacher and
parent reports, both children enjoyed watching videos at
home and school. Both girls were accustomed to leaving
the classroom with another adult. We administered the
Motor Imitation Scale (Stone, Ousley, & Littleford, 1997).
Christine scored 29 and Kaci scored 30 (32 was the ceil-
ing), indicating that they readily imitated simple adult
actions with materials.
Settings and Materials
The study occurred at an inclusive, full-day, university-
based preschool program accredited by the National
Association for the Education of Young Children. The
participants attended for the full day and were in sepa-
rate classrooms. Each classroom contained 10 to 14 chil-
dren (5 to 7 with disabilities) and 2 to 4 adults. Video
modeling sessions occurred in the preschool therapy
room. The therapy room (3 m × 4.5 m) contained an of-
fice desk (2 m × 1 m) with the experimental materials on
it. During all sessions, the first author, the participant,
and a graduate student observer were present. Materials
generalization probes occurred in the same therapy
room. Settings generalization measurement occurred in
the child’s classroom and during sensory table activities
at a large table with the sensory bin on it. Other class-
mates and adults were present during the settings gener-
alization probes.
We chose sensory activities as the focus of this study
for several reasons. First, sensory bin activities occurred
daily in the children’s classrooms, as well as in many other
early childhood classrooms, making generalization of
play skills relevant. Second, the sensory bin materials al-
lowed the children to do a variety of appropriate actions
that were not dictated by the materials themselves. This
allowed us to evaluate precisely the effects of the video
modeling on the target children’s behavior. Third, use of
sensory bin materials in a nonrepetitive (stereotypic)
manner allowed the children to appear more like their
typically developing classmates. Fourth, these types of
skills also provided new opportunities for object explo-
ration, play expansions, social initiations and responses,
and promotion of engagement (e.g., participating appro-
priately in an activity for more than 10 min). The same
sensory material (e.g., potting soil) was used for the base-
line, treatment, and maintenance phases for Christine;
the only stimuli we changed were the toys used to ma-
nipulate the sensory material. In Kaci’s case, the sensory
material was changed to colored rice in Session 48.
During material generalization probes, the toys used to
manipulate the materials were unfamiliar to both partic-
ipants. A list of toys used for all conditions is provided
in Table 1. We chose toys based on their availability in
most early childhood classrooms (e.g., shovels and pots)
and because they could be categorized as a pretend play
set (i.e., gardening toys or cooking toys). At the start of
each session, the sensory bin was placed on the floor or
on another storage bin so the materials were at the
child’s height. A laptop computer used for video viewing
was placed on a storage bin, and the child could watch it
while sitting or kneeling on the floor.
Video Production
We presented the modeling video clips, which we created
using a digital video camera, on a laptop computer.
These videos showed a pair of adult hands performing
the actions with the toys to manipulate the sensory ma-
terial. A screen shot of the video clips viewed by the par-
ticipants is shown in Figure 1. The hands on the video
performed the specific actions with the toys, manipulat-
ing the sensory materials discretely. The video clip de-
picted exactly what the child’s viewpoint should look
like: two hands, a few toys (e.g., shovels and buckets),
and a tub containing the sensory materials (e.g., potting
TABLE 1. Materials Used in Both Tasks and During Materials Generalization
Probes
Gardening set materials Cooking set materials
3 toy shovels (1 shovel) 3 cooking spoons (1 cooking spoon)
3 small planter pots (1 plastic bucket) 3 plastic toy plates (1 toy plate)
3 synthetic aquarium plants (2 red plants) 3 measuring scoops (1 scoop)
10 large flower bulbs (5 flower bulbs) 3 plastic toy bowls (1 toy bowl)
2 small tin pots (1 tin pot)
Note. Materials used during materials generalization probes are shown in parentheses.

86 Topics in Early Childhood Special Education 26:2
soil). We chose this viewpoint so the camera could focus
on manipulations being made to the materials, thus con-
trolling the model characteristics and other irrelevant
cues to which the children might have attended. We iden-
tified the actions shown in the video modeling through
observing the play actions of typical children during sen-
sory time.
During intervention, each participant viewed two
separate videos containing one behavior set: gardening
play and cooking play. Each video clip containing the de-
sired actions was less than 2 min in length and contained
three examples of each action to expose the participant
to a variety of exemplars of the desired actions with a
variety of objects belonging to the same stimulus class.
Dependent Measures
The number of different types of modeled actions per-
formed during each probe was the dependent measure. A
research assistant used a digital video recorder to video-
tape each daily probe and practice session, which the
first author then coded. We trained graduate student ob-
servers via the use of a training video and actual clips of
the participants during practice sessions. The measure-
ment system used a coding sheet listing the modeled ac-
tions the participant could imitate from the video clip
(six for the gardening task and five for the cooking task).
When the child exhibited one of the modeled actions at
any time during the probe, the action was marked as
completed. A list of each measured action type is shown
in Table 2.
We selected increasing the use of modeled actions as
a dependent measure for four reasons:
1. Studies of pretend (symbolic) play often
measure actions on materials from which
a theme, routine, or narrative can be in-
ferred—in this case, gardening and cooking.
2. Increasing children’s appropriate actions
may reduce repetitive actions, which in
turn may promote opportunities for social
exchanges with peers at the sensory bin.
3. Increasing children’s actions on materials
would increase the number of behaviors on
which adults could comment (addressing
communicative goals) and expand (pro-
moting play goals).
4. Increasing actions on sensory bin materials
would reduce the apparent differences be-
tween the participants and their classmates.
Design
We used a multiple-probe design (Horner & Baer, 1978)
across two behaviors (gardening task and cooking task)
and across two participants to evaluate the effectiveness
of the point-of-view modeling. The intermittent probes
provided an alternative to continuous baseline measure-
FIGURE 1. Screen shot of a video clip viewed by participants.

Point-of-View Video Modeling 87
ment to reduce (a) time involved in measurement and
(b) reactivity (Horner & Baer, 1978). Each target child
participated in three phases for each of the two behaviors:
a baseline phase, a treatment phase, and a posttreatment
maintenance phase. For each child, we collected data
during baseline probes, daily treatment probes, daily
treatment practice sessions, and posttreatment mainte-
nance probes. Daily treatment probes occurred prior to
each daily video viewing to assess acquisition in the ab-
sence of immediate video viewing. Setting and material
generalization probes were taken at least once in both
the baseline and maintenance conditions.
Baseline Procedures. We conducted baseline mea-
surement for each of the two sequences of behavior. For
the baseline probes, each participant knelt in front of the
bin in the therapy room. The investigator placed the first
toy set (e.g., gardening toys) in the bin and verbally cued
the child to play with the toys. No further instructions
on how to use the toys or what to do with the materials
were given. After an interval of 2 min, the investigator
removed the first toy set, and the child watched a 2-min
cartoon. The investigator then gave the child the second
toy set (e.g., cooking toys). Only the toys to which the
child was to be exposed during the video modeling were
used for the probes. The order in which the toys sets were
given was randomized for each session.
Video Modeling Treatment Procedures. The treat-
ment sessions consisted of three segments: a daily probe of
the behavior being modeled, a video modeling/viewing
segment, and a daily practice segment. Each instructional
session lasted a maximum of 15 min (2-min for the daily
probe, 3 to 5 min for viewing the video, 3 min for prac-
tice, and 3 min were reserved for transitioning between
these segments). Video modeling sessions occurred at the
same time of day for each participant. We conducted
probes each day prior to viewing the videotape to assess
acquisition in the absence of immediate video viewing.
Treatment probes were identical to baseline probes ex-
cept only the materials for the toy set were available. The
investigator gave the child the set of materials to be
viewed in the video and then cued her to play with the
materials for 2 min. At the end of each probe, the inves-
tigator provided verbal praise and tangible rewards (e.g.,
edibles) for on-task behavior (i.e., staying at the sensory
bin). Reinforcement was contingent on contacting the
toys and staying at the bin; it was not delivered for per-
forming specific actions (including modeled actions)
with the toys.
After the probe, the investigator turned on the lap-
top, and a clip of the child’s favorite cartoon prefaced
the video modeling clip. We chose the cartoons through
informal interviews with the participants’ parents. The
cartoon introduction helped the child attend to the com-
puter screen, but it did not overly upset the child when
turned off. The cartoon played for 2 min, after which the
video modeling clip began. The clip showed the first set
of materials (e.g., a bin of potting soil, three toy shovels,
three pots). A recorded female voice unfamiliar to the
participants said, “Play with your toys!” A pair of hands
entered the screen and modeled the appropriate actions.
The video showed the hands picking up each object, per-
forming a specific action, putting the object down, pick-
ing up the next object, performing the same action, and
so forth, until the action had been performed once with
each exemplar in the set. After the hands modeled the ac-
tion with each object, the same female voice said, “Great
job playing with your toys!” and the cartoon played for
another 60 s. The investigator provided the child with
the materials used on the video and told her to play with
the toys. No further instructions on how to use the toys
or what to do with the materials were given during these
practice sessions. The practice sessions lasted 3 min, and
the child only received prompts for standing at the bin
and playing with the toys, not for performing the mod-
eled actions. The investigator did not deliver reinforce-
ment for imitating the modeled actions at any time
during the baseline, treatment, or maintenance condi-
tions.
During the maintenance phase, we withdrew the
video treatment and the practice sessions. The mainte-
nance probe procedure was identical to the baseline
probe procedure in that the child was given 2 min. to
play with each set of materials separately. No instruc-
tions on how to use the toys were given, and reinforce-
ment was limited to staying at the sensory bin and
making contact with the materials.
Adapted Procedure. We adapted the procedure for
Kaci beginning in the eighth session of the treatment
phase during the cooking behavior set. We decided that
TABLE 2. Play Actions Modeled During the Video
Modeling Sessions
Gardening Cooking
behavior set behavior set
1. Uses shovel to dig hole 1. Uses scoop to scoop soil
2. Puts soil into empty pot 2. Scoops soil into pot using
only the scoop
3. Puts flower in pot 3. Uses spoon to mix in pot
4. Puts seed in the ground 4. Pours soil from pot to
plate only
5. Covers seed with soil 5. Uses spoon to dish soil
from pot to bowl
6. Covers plant with soil

88 Topics in Early Childhood Special Education 26:2
adaptation was necessary due to Kaci’s apparent satia-
tion with the materials, which had resulted in decreased
responding during both the treatment probes and the
practice sessions. First, we changed the sensory material
(potting soil) to colored rice to introduce novelty to the
task. The toys used to manipulate the materials were the
same. We then provided a specific prompt (e.g., “Do
what you saw on the video”) to Kaci prior to each treat-
ment probe and practice session. Third, during only the
practice sessions, we reinforced the modeled actions on a
fixed-ratio-1 schedule. Reinforcement consisted of spe-
cific praise (e.g., “Good job! You did what you saw on
the video!”) and a small edible reward.
Settings Generalization Probes. We conducted set-
tings generalization probes once during each baseline
and maintenance phase, using the same stimuli from the
treatment sessions. The child was in the classroom and
standing at the sensory table. We gave the child each set
of materials separately for 3 min and told her to play
with the toys. Measurement was identical for both the
baseline and treatment phases. No rewards were pro-
vided for imitating the modeled actions.
Materials Generalization Probes. Generalization
probes occurred at least once during the baseline and
maintenance conditions. Probes occurred in the therapy
room and were procedurally identical to baseline probes,
with the only difference being the availability of un-
trained toys. Stimulus generalization materials were not
shown in the videos. Probes involved objects and toys
similar enough to the trained stimuli so that the child
could discriminate which action to perform with them.
The directions provided to the child were the same as in
the baseline condition, and we delivered reinforcement
only for attending to the materials.
Social Validity
Upon completion of the study, 20 graduate students in
special education rated videotapes of the pretraining and
posttraining sessions. We randomly assigned these grad-
uate students to view four video clips. Ten of the gradu-
ate students viewed two pretraining tapes for Christine,
and two posttraining tapes for Kaci. Ten of the graduate
students viewed two pretraining tapes for Christine, and
two posttraining tapes for Kaci. They completed a five-
item questionnaire for each segment in which they rated
the child’s engagement, manipulation of materials, ap-
propriate use of materials, enjoyment of the activity, and
need for help using the materials on a 5-point Likert-type
scale (1 = strongly disagree, 2 = disagree, 3 = neither agree
nor disagree, 4 = agree, 5 = strongly agree). Taped ses-
sions were selected based on the degree to which the child
exhibited the modeled actions (i.e., no modeled actions
for preintervention tapes and five to six modeled actions
for postintervention tapes). The rater was not informed
as to whether the video clip was from before or after im-
plementation of the video modeling intervention.
Interobserver Agreement
We collected interobserver agreement (IOA) measures
for at least 25% of the baseline, treatment, and mainte-
nance sessions in each condition for each child. We cal-
culated IOA by dividing the number of agreements by
the number of agreements plus disagreements and multi-
plying by 100. We rated a response as an agreement if
both observers marked the same action as being per-
formed by the participant within the session. The IOA
for this study was 97.7% for the gardening task and
100% for the cooking task. For Kaci’s measured behav-
iors, IOA was 100% agreement; for Christine, it was
98.0% agreement.
Procedural Fidelity
To assess the degree to which all intervention sessions
were executed according to protocol, all observers and
the experimenter completed a procedural fidelity check-
list outlining the procedure for each session. Procedural
fidelity was assessed for at least 25% of all sessions in
each condition for each child. Observers completed a
task list containing all of the behaviors that were to oc-
cur in each phase and condition of the study, including
placement of toys, placement of sensory bin, verbal cues
for the child to play with the toys, reinforcement for con-
tacting materials, absence of experimenter-implemented
reinforcement for correct responding, and order of inter-
vention steps (e.g., probe, video viewing, practice session).
Procedural fidelity results showed that the baseline,
treatment, maintenance, and generalization session pro-
tocol was followed with 95% accuracy (range = 83.3%–
100%) throughout the study.
RESULTS
Probe and Practice Session Data
The number of action types performed on the gardening
and cooking tasks for Christine and Kaci are shown in
Figure 2. The highest number of action types possible
was six for the gardening task and five for the cooking
task. The open circles represent the number of actions
from the probe sessions across the three conditions: base-
line, video modeling, and maintenance (no video).
During the video modeling condition, these probe ses-
sions occurred each day before the children watched the
video. Figure 2 also contains data from the video model-
ing condition on the number of actions performed dur-

Point-of-View Video Modeling 89
FIGURE 2. Multiple probe data across two participants and two behaviors, with number of types of play actions per-
formed during baseline, treatment, and posttreatment maintenance sessions. Probes are marked as open circles
connected by a solid line. Practice sessions are marked as an “x” connected with a dotted line.

90 Topics in Early Childhood Special Education 26:2
ing practice sessions (depicted by an ×). These sessions
followed the daily viewing of the video; thus, they oc-
curred only during the video modeling condition.
The top panel of Figure 2 represents the number of
action types used by Christine in the gardening task.
During the baseline condition in the probe sessions for
the gardening task, Christine had between none and two
action types per session, with one action in 3 of 5 ses-
sions. With the introduction of the video modeling con-
dition, Christine displayed four action types in the first
session, dropped to one action in the next 2 sessions, and
then increased to three or four action types for the re-
mainder of the condition (with one exception on the
10th datum point of the condition, which had two ac-
tion types). In 11 of 14 probe sessions during the video
modeling condition, she performed three or four action
types per session, exceeding the highest level (i.e., two)
obtained during the baseline condition. During the 12
maintenance assessments with no video modeling,
Christine performed at least four action types in all ses-
sions except for 1. In 4 of 12 maintenance assessments,
she performed five action types per session, which was a
greater number than occurred during video modeling.
The practice sessions (depicted by an “×” in Figure 2) oc-
curred only during the video modeling condition, imme-
diately after she watched the video. Christine performed
between two and five action types each session, with 3 of
the last 4 sessions resulting in five action types each.
The second panel of Figure 2 shows Kaci’s data on
the gardening task. During baseline, Kaci performed one
or two action types per probe session, with four of six
probe sessions being one action type. With the introduc-
tion of the video modeling, she performed one action in
the first session and then four action types during each
of the next four sessions. This dropped to one action per
session for the next two sessions, increased to four ac-
tion types for two sessions, and then dropped to one ac-
tion type for the last session of the condition. During the
maintenance condition, Kaci performed between one ac-
tion type and six types per session, with four of seven
maintenance probe sessions containing four or more ac-
tion types per session. On the practice sessions in the
video modeling condition, Kaci performed from one ac-
tion type to six action types per session. She tended to do
more action types per practice session early in the condi-
tion as compared to later in the condition.
The third panel of Figure 2 shows the number of ac-
tion types performed by Christine on the cooking task.
During 10 baseline probe sessions, Christine exhibited
none of the action types. After introduction of the video
modeling, Christine’s performance of modeled actions
during treatment probes increased to between one action
type and five action types per probe session. In 3 of the
last 4 sessions of the video modeling condition, she per-
formed five action types per session. In 5 maintenance
probes, she performed between three and five action
types. Christine performed four action types in the last 2
maintenance probe sessions. During the practice sessions
of the video modeling condition, she performed one to
five action types per session. In the last six sessions, she
performed five action types per session.
The fourth panel of Figure 2 represents Kaci’s data
on the cooking task. In 10 baseline probe sessions, Kaci
ranged between no actions and two action types per ses-
sion. In the first baseline session, she performed two ac-
tion types; in the 9th probe session, she performed one
action, but in all other baseline probe sessions no actions
occurred. Introduction of the video modeling did not
change Kaci’s behavior. In 6 probe sessions, she had no
actions in 5 sessions and one action in 1 session. This re-
sulted in the adaptation described previously: the investi-
gator changed the sensory materials to colored rice, told
Kaci before each probe and practice session to play with
the toys like she saw on the video, and reinforced actions
on the toys on a fixed-ratio-1 schedule during the prac-
tice sessions. The investigator did not deliver any rein-
forcement during the probe sessions. In the probe
sessions, the adaptation resulted in an increase in the
number of actions performed, but then a drop occurred
to two actions or one action per session. For the practice
sessions before the adaptation, most sessions resulted in
no actions being performed. After the adaptation, the
number of actions increased sharply, with the last 6 prac-
tice sessions containing five action types per session.
Generalization
Generalization Across Materials. Generalization data
across materials for the gardening and cooking tasks are
shown in Table 3. As shown, Christine and Kaci dis-
played few actions during the baseline materials general-
ization probes for the gardening task and increased the
number of actions performed in the maintenance condi-
tion probes. This pattern was replicated for Christine in
the materials generalization probes for the cooking task.
Kaci performed no actions in the baseline probe sessions
on the cooking task, and she performed only one action
in the maintenance condition probe session.
Generalization Across Settings. The generalization
across settings occurred in the girls’ classrooms. As
shown in Table 3, for the gardening task, Christine and
Kaci displayed fewer actions in the baseline generaliza-
tion probes than in the maintenance probe sessions. For
the cooking task, however, neither girl displayed an in-
crease in the number of actions performed in the setting
generalization probes.
Social Validity Results
The results for the social validity assessment are shown
in Table 4. For the engagement, use of multiple actions,

Point-of-View Video Modeling 91
appropriate use of materials, and enjoyment of the activ-
ity items, posttreatment means were higher than pre-
treatment means. For the needing help with materials
item, the mean decreased. Children in the posttreatment
clip were rated higher on their engagement in the activ-
ity, their use of multiple actions while manipulating ma-
terials, using materials appropriately, and enjoying the
activity than children viewed in the pretreatment clip.
Children viewed in the posttreatment videos were also
rated as needing help in using materials less often. These
results suggest that the raters viewed the treatment as so-
cially valid.
DISCUSSION
This study evaluated the effectiveness of point-of-view
video modeling on preschoolers’ performance of play ac-
tions to determine if these skills generalized to novel ma-
terials and to the classroom. The results indicated that
the video modeling was effective in teaching specific ac-
tions on toys and sensory materials to two girls with
autism. In three of four behavior sets (gardening for both
girls, and cooking for Christine), the children acquired
new play behaviors in the absence of experimenter-im-
plemented reinforcement for performing the video-mod-
eled actions and without instructional cues on what to
do with the materials. The addition of reinforcement
during practice sessions and instructions to imitate the
actions from the video allowed Kaci to acquire the be-
haviors of the cooking set. These findings extend those
of D’Ateno et al. (2003) and Taylor et al. (1999), sug-
gesting that viewing videotapes may change not only ver-
bal behavior but also motor movements on toys in the
absence of programmed reinforcers. Whether the in-
crease in new actions constitutes symbolic or pretend
play deals in part with the level of evidence required to
infer pretense; however, this increase clearly provided the
children with additional behaviors from which pretend
routines and narratives could be displayed.
The children also generalized the skills to untrained
materials for both tasks and generalized skills into the
classroom for the gardening task but not for the cooking
task. This finding indicates that generalization can be
promoted through point-of-view video modeling involv-
ing multiple exemplars of materials. It confirms the find-
ings of Haring et al. (1987) and Charlop and Milstein
(1989) showing that multiple exemplars presented in
videos result in generalized responding to novel stimuli.
In this study, children generalized their actions across
settings for the gardening task but not for the cooking
task. The videotape did not contain multiple exemplars
of location, and this reason—among other reasons—may
account for the lack of generalization to the classroom
for the cooking task. Other reasons may be (a) the lim-
ited number of classroom probes, (b) activities in the
classroom competing for the target children’s attention,
and (c) satiation with the materials.
TABLE 3. Performance of Participants in Probes of Generalization Across Materials and Across Settings for Baseline
and Maintenance Conditions
Baseline Maintenance
Participant # of assessments M Lowest–highest # of assessments M Lowest–highest
Generalization across materials—Gardening
Christine 1 1 — 4 4 (2–5)
Kaci 2 1.5 (1–2) 3 4 (4–4)
Generalization across materials—Cooking
Christine 2 1 (1–2) 3 3 (0–5)
Kaci 3 0 — 1 1 —
Generalization across settings—Gardening
Christine 1 2 — 2 2.5 (1–4)
Kaci 2 0.5 (0–1) 2 3.5 (2–5)
Generalization across settings—Cooking
Christine 2 0 — 1 0 —
Kaci 3 0.67 0–2 1 0 —

92 Topics in Early Childhood Special Education 26:2
These findings suggest that point-of-view video mod-
eling may be useful in helping young children with autism
learn to perform actions on toys from which pretense can
be inferred. We should note, however, that both partici-
pants readily imitated in vivo actions of adults with ma-
terials, contacted materials without adult prompting,
and were verbal. Future research should explore the ex-
tent to which point-of-view modeling is successful with
children who do not display these skills or who have
variations of them.
This study has a number of notable limitations. First,
acquisition of skills did not occur in the cooking task for
Kaci until a verbal instruction to imitate the video and
contingent reinforcement for doing so were used. Thus,
watching the video by itself may not be sufficient to
promote skill acquisition in some instances; other inter-
ventions may be needed. In some cases, a cue telling par-
ticipants to do what they saw in the video was necessary
for them to apply those skills in the training situation as
well as across settings and materials. Previous studies
have used reinforcement of correct imitative responding
during the video modeling phase (Charlop & Milstein,
1989, Shipley-Benamou et al., 2002). Video modeling
plus specific cues and reinforcement of correct respond-
ing may increase the rate at which skills are acquired,
and in conjunction with the use of multiple exemplars
may be more effective in promoting skill generalization.
This assertion could be a focus of future studies.
Second, performance of modeled actions may have
been more positively affected with a reduced number of
probes. In the current study, we exposed the participants
to the same materials once before the video (probe), dur-
ing the video, and after the video (practice sessions) dur-
ing each treatment session. Limiting sessions to just the
video and practice, while probing once every few ses-
sions, would decrease the amount of treatment time and
may inhibit satiation. Also, alternating materials across
sessions and thereby decreasing exposure may increase
novelty, interest, engagement, and subsequent perfor-
mance of the modeled actions. Procedures also could be
altered so the child views the video clip in the classroom
before the day’s sensory bin activity. This may promote
faster acquisition and more generalized responding due
to more functional practice. Future studies could mea-
sure the diversity of play and the effects of using toys
functionally on children’s play sequences, repetitive ac-
tions, and social exchanges with peers.
Third, the study had relatively limited measures of
generalization, especially for generalization to the class-
room. Classroom measurement was made difficult by
competing activities, which in conjunction with apparent
satiation on trained materials made other activities and
materials more interesting to the participants. In addi-
tion, the same adult was involved in implementation,
observation, and reinforcer delivery, suggesting some
stimulus control was possible over participants’ perfor-
mance. Evaluating children’s behaviors with different
adults as well as teachers’ use of the procedures need to
be studied.
Video modeling interventions capitalize on the vi-
sual strengths of children with autism, and teachers can
use them in the classroom. This study indicates that pos-
itive effects occurred on play skills after introduction of
point-of-view video modeling, which was efficient in terms
of intervention preparation. Video modeling interven-
tions can be included as resourceful methods to foster
play in preschoolers with autism. ◆
AUTHORS’ NOTES
1. This study was conducted as a thesis study by the first author in
partial fulfillment of the master’s of education degree from the
Department of Special Education, Peabody College, Vanderbilt Uni-
versity.
2. The study was supported by a U.S. Department of Education,
Personnel Preparation, PR award (H325A030093). The opinions
expressed, however, do not necessarily reflect those of the U.S.
Department of Education.
3. The authors are grateful to the parents who allowed their children to
participate. We also are grateful to teachers Bri Klibbe and Kyle Alm-
gren of the Susan Gray School, Peabody College, and to Dr. Ruth
Wolery, director of the Susan Gray School.
TABLE 4. Mean Responses and Standard Deviations for the Social Validation of Pre- and Posttreatment Video Clips
Pretreatment clip Posttreatment clip
Questionnaire item MSD MSD
1. The child is engaged in the activity 3.73 1.04 4.78 0.42
2. The child uses multiple actions while manipulating materials 2.53 1.15 4.48 0.68
3. The child uses the materials appropriately 3.15 1.25 4.73 0.51
4. The child is enjoying the activity 3.43 0.68 4.18 0.75
5. The child needs help using the materials 3.45 1.24 1.50 0.60
Note. Scale: 1 = strongly disagree, 5 = strongly agree.

Point-of-View Video Modeling 93
REFERENCES
Alcantara, P. R. (1994). Effects of videotape instructional package on
purchasing skills of children with autism. Exceptional Children, 61,
40–55.
American Psychiatric Association. (1994). Diagnostic and statistical
manual of mental disorders (4th ed.). Washington, DC: Author.
Bandura, A. (1965). Influence of models’ reinforcement contingencies
on the acquisition of imitative responses. Journal of Personality and
Social Psychology, 6, 589–595.
Buggey, T., Toombs, K., Gardener, P., & Cervetti, M. (1999). Training
responding behaviors in students with autism: Using videotaped self-
modeling. Journal of Positive Behavior Interventions, 4, 205– 214.
Charlop, M. H., & Milstein, J. P. (1989). Teaching autistic children
conversational speech using video modeling. Journal of Applied Be-
havior Analysis, 22, 275–285.
Charlop, M. H., Schreibman, L., & Tryon, A. S. (1983). Learning
through observation: The effects of peer modeling on acquisition
and generalization in autistic children. Journal of Abnormal Child
Psychology, 11, 355–366.
Charlop-Christy, M. H., & Daneshvar, S. (2003). Using video model-
ing to teach perspective taking in children with autism. Journal of
Positive Behavior Interventions, 5, 12–21.
Charlop-Christy, M. H., Le, L., & Freeman, K. A. (2000). A comparison
of video modeling with in vivo modeling for teaching children with
autism. Journal of Autism and Developmental Disorders, 30, 537–
552.
D’Ateno, P., Mangiapanello, K., & Taylor, B. A. (2003). Using video
modeling to teach complex play sequences to a preschooler with
autism. Journal of Positive Behavior Interventions, 5, 5–11.
Dawson, G., & Osterling, J. (1997). Early intervention in autism. In
M. J. Guralnick (Ed.), The effectiveness of early intervention (pp. 307–
326). Baltimore: Brookes.
Haring, T. G., Kennedy, C. H., Adams, M. J., & Pitts-Conway, V.
(1987). Teaching generalization of purchasing skills across commu-
nity settings to autistic youth using videotape modeling. Journal of
Applied Behavior Analysis, 20, 89–96.
Hepting, N. H., & Goldstein, H. (1996). Requesting by preschoolers
with developmental disabilities: Videotaped self-modeling and
learning of new linguistic structures. Topics in Early Childhood
Special Education, 16, 407–427.
Horner, D. R., & Baer, D. M. (1978). Multiple probe technique: A vari-
ation of the multiple baseline. Journal of Applied Behavior Analy-
sis, 11, 189–196.
Ingersoll, B., Schreibman, L., & Tran, Q. H. (2003). Effect of sensory
feedback on immediate object imitation in children with autism.
Journal of Autism and Developmental Disorders, 6, 673–682.
Jarrold, C. (2003). A review of research into pretend play in autism.
Autism, 7, 379–390.
Jarrold, C., Boucher, J., & Smith, P. (1993). Symbolic play in autism: A
review. Journal of Autism and Developmental Disorders, 23, 281–
307.
Jordan, R., & Libby, S. (1997). Developing and using play in the cur-
riculum. In S. Powell & R. Jordan (Eds.), Autism and learning: A
guide to good practice. London: David Fulton.
Kinney, E. M., Vedora, J., & Stromer, R. (2003). Computer-presented
video models to teach generative spelling to a child with an autism
spectrum disorder. Journal of Positive Behavior Interventions, 5,
22–29.
LeBlanc, L. A., Coates, A. M., Daneshvar, S., Charlop-Christy, M. H.,
Morris, C., & Lancaster, B. M. (2003). Using video modeling and
reinforcement to teach perspective-taking skills to children with
autism. Journal of Applied Behavior Analysis, 36, 253–257.
Libby, S., Powell, S., Messer, D., & Jordan, R. (1998). Spontaneous
play in children with autism: A reappraisal. Journal of Autism and
Developmental Disorders, 28, 487–497.
Nikopoulos, C. K., & Keenan, M. (2004). Effects of video modeling
on social initiations by children with autism. Journal of Applied
Behavior Analysis, 37, 93–96.
Pierce-Jordan, S., & Lifter, K. (2005). Interaction of social and play be-
haviors in preschoolers with and without pervasive developmental
disorder. Topics in Early Childhood Special Education, 25, 34–47.
Schreibman, L., Whalen, C., & Stahmer, A. C. (2000). The use of video
priming to reduce disruptive transition behavior in children with
autism. Journal of Positive Behavior Interventions, 2, 3–11.
Sherer, M., Pierce, K. L., Paredes, S., Kisacky, K. L., Ingersoll, B., &
Schreibman, L. (2001). Enhancing conversation skills in children
with autism. Behavior Modification, 25, 140–158.
Shipley-Benamou, R. S., Lutzker, J. R., & Taubman, M. (2002).
Teaching daily living skills to children with autism through instruc-
tional video modeling. Journal of Positive Behavior Interventions, 4,
165–175, 188.
Stahmer, A. C., Ingersoll, B., & Carter, C. (2003). Behavioral ap-
proaches to promoting play. Autism, 7, 401–413.
Stone, W. S., Ousley, O. Y., & Littleford, C. D. (1997). Motor imita-
tion in young children with autism: What’s the object? Journal of
Abnormal Child Psychology, 25, 475–485.
Taylor, B. A., Levin, L., & Jasper, S. (1999). Increasing play-related
statements in children with autism toward their siblings: Effects of
video modeling. Journal of Developmental and Physical Disabili-
ties, 11, 253–264.
Ungerer, J. A., & Sigman, M. (1981). Symbolic play and language com-
prehension in autistic children. Journal of the American Academy
of Child Psychiatry, 20, 318–337.
Van Berckelaer-Onnes, I. A. (2003). Promoting early play. Autism, 7,
415–423.
Wert, B. Y., & Neisworth, J. T. (2003). Effects of video self-modeling
on spontaneous requesting in children with autism. Journal of Posi-
tive Behavior Interventions, 5, 30–34.
Zihni, F., & Zihni, F. (2005).
AZ method: The use of video techniques
to develop language skills in autistic children. Retrieved March 10,
2005, from http://www.autismuk.com/index13.htm