An Evaluation of Parent-Produced Video Self-Modeling to Improve Independence in an Adolescent With Intellectual Developmental Disorder and an Autism Spectrum Disorder

Method

Measurement

A number of possible target responses were initially identified by Holly and her parent (e.g., crossing the street independently), but Holly and her parent prioritized (a) requesting help from a quick shop employee, (b) checking out from a store, and (c) ordering food at a fast-food restaurant because these were more commonly encountered routines. A task analysis was conducted by one of the investigators (second author) to identify the critical steps in each task. Next, target responses were operationally defined (Table 1).

OPEN IN VIEWER

Table 1.

Targeted Responses and Response Steps.

Targeted response	Number of response steps	Response step examples
Requesting help	10	1) approach and face the clerk, 2) say “excuse me, miss/sir,” 3) wait for response, 4) establish eye contact, 5) ask, “Can you help me find (name the item),” 6) wait for a response, 7) follow the clerk if necessary, 8) establish eye contact, 9) say, “thank you,” and 10) take the item
Checking out	11	1) approach the counter, 2) establish eye contact, 3) say, “I would like to buy these,” 4) wait for clerk to tell the total amount owed, 5) count out enough money, 6) establish eye contact, 7) say, “here you go,” 8) hand the money to the clerk, 9) wait for approval of payment, 10) say, “thank you,” and 11) say, “good bye.”
Ordering food	12	1) approach the counter, 2) establish eye contact, 3) say, “hello” or “hi,” 4) Say, “I would like to order (name item),” 5) wait for response, 6) establish eye contact, 7) name additional item desired or say, “that is all,” 8) wait for clerk to name total amount owed, 9) count out correct payment, 10) say, “here you go,” 11) hand money to clerk, and 12) say, “thank you.”

Measurement involved paper and pencil event recording by a primary observer (second author) and a trained secondary observer (undergraduate psychology student) of the presence of each step on the task list for each routine. During all conditions, the parent was asked to allow the participant about 5 s to respond before verbally prompting completion of the step if necessary. Observers independently recorded whether each step was performed and whether there was any verbal or physical prompting by the parent. The primary dependent measure was percentage of steps completed without prompting in each routine.

Results

Measures of treatment integrity confirm that the parent did implement treatment as requested. Direct observations indicated that the parent did produce VSM videos for each of the three routines and did reliably require the participant to watch the relevant VSM video immediately prior to performing each routine during intervention. In addition, the scores on the VMRS indicated that the parent did produce videos of acceptable quality. In general, the videos showed the key steps in each routine, omitted irrelevant steps, included appropriate narrations, and provided relevant feedback without significant distractions. The Requesting Help routine video was scored at 89%, with deductions for not including performance feedback. The Checking Out routine video was scored slightly lower at 83% because of some visual distractions as well as an absence of performance feedback. Finally, the Ordering Food routine video was scored at 94%.

Results are depicted in Figure 1, which shows the percentage of unprompted correct responding on each of the three target routines. In baseline, Holly was only able to perform the Requesting Items routine unprompted about 26% of the time. Holly was only able to perform the Checking Out routine unprompted about 36%. There were slight increasing trends in her performance on the Checking Out and Order Food routines across the baseline sessions, but the data began to show stability before the VSM intervention was introduced in each case.

OPEN IN VIEWER

Figure 1.

Percentage of correct responding per session for each of three different routines.

When VSM was introduced, Holly’s performance showed marked improvement after just one session of viewing the Requesting Items Video. Her performance maintained a level consistently near 80% correct responding, ranging from 60% to 100% correct. More important, this pattern was replicated with the Checking Out routine, where performance showed marked and immediate improvement following just one viewing of the VSM intervention. Finally, this pattern was replicated a third time, when the Ordering Food performance showed a marked increase from less than 40% correct performance in baseline to greater than 70% correct responding in the intervention phase.

Figure 1 also shows results of the generalization probes, depicted by the arrows on each panel of the figure. Two probes were conducted in the natural setting for each target skill, where Holly was observed to exhibit between 80% and 90% correct responding.

Finally, the parent rated the parent-created VSM approach as highly acceptable. She indicated an overall score of 39 out of 45. She strongly endorsed statements indicating that she liked the approach and that she found it to be effective and beneficial to Holly.

Discussion

The results of this investigation suggest that a parent can, with minimal training, effectively use available tablet-based technology to record and implement VSM to enhance independence in the community for an individual with ASD and intellectual disabilities. This parent was able to independently create videos of three different community-based routines that resulted in marked improvements in independent responding after the videos were produced and watched prior to each routine. That is, the amount of prompting required by the parent was markedly reduced in each routine. Perhaps most importantly, the performances by Holly were found to generalize to the natural community setting where these routines would commonly be most useful.

It is interesting that the responses learned in the analogue training setting generalized as well as they did to the natural setting. Although Holly was permitted to watch the self-modeling videos prior to each generalization probe, there were some concerns that videos produced in the home setting might not produce generalization to the natural environment because the settings were so different. However, the parent did include props (i.e., toys, puzzles, shelves, cash register, menu) in the analogue setting that were similar to those that the participant would likely encounter in the natural setting, essentially programming common stimuli across settings (Stokes & Baer, 1977).

In general, there was a reasonable demonstration of experimental control. First, although both the Checking out and the Ordering Food baseline routines showed upward trends in performance when VSM was introduced with the first routine, these trends did not continue and baseline responding was relatively stable by the time intervention was introduced with the second and third routines. Second, the performance in each routine showed marked changes in level when and only when the VSM was introduced, considered to be a hallmark demonstration of experimental control in multiple baseline designs (Barlow, Nock, & Hersen, 2009). Third, the confidence in this demonstration of experimental control is bolstered by the fact that the response to the VSM intervention was immediate, with clear changes in level occurring after only a single exposure to VSM. Finally, this immediate response was replicated across the three targeted routines.

In spite of these improvements in performance, the participant had not completely mastered each routine and she did still require some parental prompting. There are a number of possibilities that might account for this. First, the videos that were produced by the parent included the parent herself in the video (role-playing a clerk or restaurant employee) where she sometimes prompted the participant if the participant became stuck. Thus, the participant sometimes watched herself performing with prompts rather than watching unprompted mastery. It may be that watching unprompted mastery in the video would have resulted in better performance later on. If that were true, it would be important to modify the VideoTote tutorial to recommend that caregivers rehearse until prompts are not needed. However, prompting was only observed in the Ordering Food and Checking Out videos (Step 10), and other studies have suggested that watching a model who is learning is as good or better than watching a model who is perfect (e.g., Cumming & Ramsey, 2011; Schunk & Hanson, 1985). Thus, whether the prompting contained within the video diminished the treatment effects remains an empirical question.

It is also possible that the participant’s performance might have been improved had the parent zoomed in more closely on some critical aspects of performance during filming. These are not features that are known to be critical to success, but neither have they been empirically evaluated. Note that the video that scored the lowest quality rating (i.e., Checking Out) was associated with the least overall improvement (i.e., required more prompting). The parent lost credit, in part, because there were visual distractions that may have detracted from the value of the video and also because there was no performance feedback included in the narration.

It is important to note that because the VSM intervention involved both the production and the watching of a video, it is unclear whether one component alone or the combination of components may have been critical to the observed improvements in performance. For example, the participant may have benefitted solely from her involvement in the production of the video itself. However, prior to the study, the participant had made little progress despite months of practice with her parents, so it seems unlikely that the benefits of VSM were the result of a practice effect alone. Finally, in a recent evaluation of VSM, the investigators specifically tested whether performance improved simply as a result of making a VSM video, and they found that it did not (Goh & Bambara, 2013). Nevertheless, this study was not intended to be a component analysis, so any conclusions about the mechanism of change are unwarranted.

One promising aspect of this study was that a parent was able to produce an effective VM intervention for her daughter using available technology with little guidance from the investigators. This seems critical to the overall successful dissemination of video modeling. Indeed, evidence-based treatments such as video modeling are more likely to be disseminated if they can be embedded in a family’s everyday life (Rotheram-Borus, Swendeman, & Chorpita, 2012). Nevertheless, there remains much to learn about how to produce and use a video most effectively. For example, it is often considered important to include models who are similar and/or familiar to the observer; however, the evidence for this is considered by some to be weak (e.g., Dowrick, 2012). In addition, there is limited evidence of the value of voice over narration (Mechling & Collins, 2012) and only preliminary evidence of the value of incorporating positive performance feedback (e.g., Hawkins & Heflin, 2010). Furthermore, we do not know how often a video needs to be watched, and whether watching more often (e.g., multiple times before each performance) would produce better outcomes.

Although the results do show good internal validity, the external validity remains unknown. That is, we do not know whether other parents could produce VSM videos as easily or as effectively with their care receivers, and whether these results would be found in other types of community, vocational, or daily living settings. Replication is required. In addition, we do not know how well these results would have maintained had the VSM been withdrawn. For example, it is possible that the original dependency on prompting was simply shifted from the parent to the video. Future research will need to look both at dose–response relations and whether the benefits of VSM persist after withdrawal. Finally, there may be other benefits derived from exploring other technologies for video modeling, including videos produced with smaller, mini-tablet devices, phone-based cameras, and portable GoPro® devices. These may help extend the reach of video modeling into even more parents’ hands and further improve the dissemination of evidence-based technology.