Abstract
The purpose of this study was to examine the effectiveness of video-based mathematics instruction for seven middle school students with specific learning disability (SLD), using an augmented reality-based training package. The dependent variable was the percentage of steps students performed correctly to solve each type of mathematics problem. The independent variable was the augmented reality video-based intervention, which used video to model the individual steps for solving four types of multistep mathematics problems: (a) addition and subtraction of integers, (b) multiplication and division of integers, (c) using ratio reasoning to convert measurement units, and (d) using multiplication and division to calculate rate of change. Results indicated a functional relation between the video-based mathematics intervention and the percentage of steps completed correctly for each type of problem. All seven participants showed significant gains immediately after receiving the intervention and maintained improved problem-solving skills in at least three out of the four problem categories.
Educators are continually developing and implementing evidence-based practices to prepare individuals with specific learning disability (SLD) to live productive and fulfilling lives—refining effective methodologies and creating new ones. Basic proficiency in numeracy is necessary to function in both academic and daily living contexts (Beswick, 2008). But unfortunately, a significant disparity exists between the mathematics proficiency levels of individuals with SLD and those of their peers without disabilities (The Nation’s Report Card, 2015).
The National Assessment of Educational Progress conducted by the U.S. Department of Education in 2015, commonly referred to as the Nation’s Report Card, found that only 8% of eighth grade students with disabilities were at or above the proficient level in math, compared with 36% of students without disabilities (The Nation’s Report Card, 2015). With 92% of students with disabilities and 64% of those without disabilities performing below proficiency in math, interventions are critically needed.
Due to current pressure for academic progress at federal, state, and local levels, public as well as professional emphasis on building student proficiency in mathematics has drastically increased. As the demand for math proficiency has increased, so too has technology in education. Video-based technology such as augmented reality (AR) paired with video modeling (VM) and video prompting (VP) has shown promise for increasing the percentage of students with and without disabilities who are proficient in mathematics (Kellems et al., 2019).
Augmented Reality (AR)
AR technology enables blending of the physical and digital worlds, enhancing a user’s immediate environment by providing additional information with camera-generated pictures, videos, and audio on mobile devices (Sommerauer & Müller, 2014). The popular game Pokemon Go is an example of AR. AR utilizes “trigger” objects or images from the physical world to cause the videos to play, not a generic button to activate them in strict sequence as other forms of digital/video platforms do, such as VM and VP. A “trigger” image can be either a static picture or a physical object that when scanned by an AR-enabled device utilizes image recognition to activate a preselected video or other digital material on the device. The AR system is characterized by (a) combining real and virtual worlds, (b) providing an interaction (e.g., trigger selection) that happens in real time, and (c) superimposing real objects or places and three-dimensional (3D) digital information (Sommerauer & Müller, 2014).
The term AR was introduced in 1992; however, examples of AR have existed since 1968, even though the technology available at the time was immovable, large, and expensive—much different from the affordable mobile devices, like smart phones and tablets, available today (Walker et al., 2017). The cognitive theory of multimedia learning is based on the principle that words and pictures are more effective for learning than words only (Mayer, 1997); AR overlays printed text, audios, videos, and other virtual content (Sommerauer & Müller, 2014). AR can also function as an assistive technology and/or instructional technology; in accord with the principles of the universal design for learning, it can support students with disabilities in their learning process (Walker et al., 2017).
Several studies have asserted that AR has high relevance to 21st-century learners. However, limited research has been conducted utilizing AR interventions for students with disabilities. One of the few studies focusing on elementary students with intellectual disabilities (ID) utilized AR to successfully teach basic matching skills (Richard et al., 2007). Two studies were conducted using AR to teach navigation skills to students with ID (Smith et al., 2016). In an innovative study working with postsecondary students with ID, researchers used AR to teach science vocabulary (McMahon et al., 2016). Other areas where AR has been used with students with disabilities include teaching work-related skills (Chang et al., 2013) and demonstrating ways to recognize emotions (Chen et al., 2015). Although these previous studies have begun to explore the potential for AR as an instructional method, particularly among individuals with ID, little to date has been done to examine its effectiveness in students with SLD.
Video Modeling and Video Prompting
Extensive research of the past two decades has established VM as an evidence-based practice (Bellini & Akullian, 2007) for teaching a variety of skills to students with disabilities (Kellems & Morningstar, 2012). VM involves filming a model—a peer, an educator, a family member, or one of the learners—correctly performing a task (Kellems et al., 2017). The video is shown to the individuals with disabilities, who learn how to perform the task as they watch the video and memorize, imitate, and generalize the behavior (Kellems & Morningstar, 2012). VM presents the entire task sequence without interruption, which helps the learners develop a sense of integration among all the steps. After watching the video in its entirety, the student is required to perform the task from beginning to end (Cannella-Malone et al., 2006).
VM has been determined to be more effective than face-to-face modeling for both acquisition and generalization of skills and behaviors, in addition to its time efficiency for teachers, who are freed to maximize their classroom availability to help individual students (Kellems et al., 2016). Furthermore, VM has been demonstrated to be more stimulating than face-to-face methodologies for some learners as it varies from learners’ typical format of instruction (Gardner & Wolfe, 2013).
VP is a variation of VM which the educator separates video clips into the individual steps of a task or behavior and shows them one at a time. The learner performs each step before the next clip is shown (Kellems et al., 2016). Each clip can be presented repeatedly to suit learners’ individual pacing, until they become proficient in that aspect of the task. A review of 18 studies involving VP indicated that people with disabilities who have difficulty watching lengthy videos benefit particularly from this step-by-step approach, as it facilitates and accelerates acquisition, maintenance, and generalization of academic skills (Banda et al., 2011; Cannella-Malone et al., 2006). Many learners with disabilities have challenges associated with attention and memory; thus, since a model needs to be attended to in its entirety to be useful, breaking a video into smaller parts (VP) has proven to be more effective than showing the entire video (VM) in supporting the learning process and increasing independence of individuals with disabilities (Cannella-Malone et al., 2011).
Research Questions
This study sought to explore methods of teaching students with SLD mathematics skill-based problem-solving skills based on the Common Core State Standards Initiative (2010). Three research questions guided the study:
Method
Setting and Participants
This study took place at a western U.S. urban middle school that serves students in Grades 7 and 8. During the study (which took place during the school year), 822 students were enrolled in the school, of whom 45% were from ethnic minority backgrounds, 12% were English language learners, and 15.1% were identified as having a disability; 50.2% were eligible for free and reduced-price lunch. The school was in a district that used the regression discrepancy method (multiple regressions between ability and achievement scores) for identifying students with SLD.
Participants were students who (a) were classified as having an SLD, (b) had Individualized Education Program (IEP) mathematics goals, (c) were enrolled in both a general education mathematics class and a resource mathematics support class, and (d) had the visual and auditory capability to attend to the intervention package (videos, checklists, and social validity questionnaire). The school included one Grade 7 and one Grade 8 mathematics support class; permission forms to participate in the study were distributed in both classes. The students who returned signed permission forms were given a pretest (to be described), with a predetermined cutoff score of 50% or lower for participation in the study.
Seven students met the eligibility criteria, all of whom were eighth graders who had been diagnosed by the school as having SLD. Table 1 outlines demographic information for the individual participants. As evident in Table 1, one of the students, Ben, had a Full Scale Intelligence Quotient (FSIQ) of 73. The researchers were wary of his SLD diagnosis with the discrepancy method as the sole criterion. Ben’s mathematics achievement scores were commensurate with scores of the other participants. The intervention was delivered in a private study room in the school’s media center during each participant’s scheduled resource mathematics class instruction. The study was implemented by four interventionists: one with a bachelor’s degree in special education, one with a master’s degree in special education, and two undergraduate students majoring in special education.
Participant Information for Study on Video-Based Instruction.
Note. IDEA: Individuals with Disabilities Education Act; FSIQ: Full Scale Intelligence Quotient; LD = learning disability; CTONI = Comprehensive Test of Nonverbal Intelligence; KTEA-2 = Kaufman Test of Educational Achievement; WJ = Woodcock–John Test of Achievement; SBIT = Stanford–Binet Intelligence Scales; WISC = Wechsler Intelligence Scale for Children.
Target Mathematics Skills
Target problems and all intervention questions were determined before selecting participants. Researchers reviewed the common core standards for mathematics and selected specific standards after conferring with the school’s Grades 7 and 8 general education mathematics teachers and with the resource mathematics teacher (Common Core State Standards Initiative, 2010). The Grade 7 teachers identified four standards (outlined below) as consistently difficult for their students to understand but important as basal skills upon which higher order skills in subsequent grades would be built. Each of the selected mathematic skills was presented as a stand-alone problem and designed to be completed independently, not functionally related to the other problems.
All of the identified problems fit within an instructional unit based on integers, defined as being whole numbers, not fractions—all integers are rational numbers, but not all rational numbers are integers. The Grade 8 teachers considered the selected standards foundational to many eighth grade mathematics standards these students would need to learn. They also identified many of their struggling Grade 8 students as not having mastered these particular concepts. The difficulty and importance of these standards were confirmed later by an outside expert. The study participants, all Grade 8 students, received a score of 50% or lower on the pretest. Selected skills were designed to teach the following concepts: (a) addition and subtraction of integers, (b) multiplication and division of integers, (c) using ratio reasoning to convert measurement units, and (d) using multiplication and division to calculate rate of change.
After the standards had been selected, the researchers developed one task for each standard. For each selected task, 24 different problems were selected. An explanation of each standard and example problems can be found in Table A1 in the appendix.
Materials and Equipment
Task analysis and data collection sheets
Before beginning baseline, the researchers created a task analysis for each of the targeted skills, outlining every individual step in the problem-solving process for the target skill (see Table A2). An expert with a PhD in mathematics instruction who had not participated in developing the task analyses was asked to perform the skills by strictly following each task analysis to verify that all the necessary steps were addressed clearly. No revisions were suggested. Data collection sheets were then created based on the developed task analyses.
Videos
Videos were developed for each of the selected target skills. Using the task analysis as the script, a female researcher aged approximately 25 years, serving as the model, was filmed performing the skills exactly as the students would be taught and expected to perform them. The videos were edited with iMovie on the computer, and voiceover instructions were added to clarify every step. Each video that modeled an entire task was split into segments: some consisting of a single step of the task analysis and others including multiple steps that the researchers considered a logical grouping. Videos length varied: The integer addition and subtraction video was 2:25, integer multiplication and division was 1:33, operations was 2:11, and rate of change was 2:37. Videos were then divided into separate clips for each of the steps on the task analysis. This way if a student only needed instruction on one part of the problem he or she could then trigger that specific clip instead of having to watch the entire video.
AURASMA app
Both the full videos and the segments were uploaded to the AURASMA app on the iPad and individually connected to trigger images specifically created for this project. To activate a video via AR, a student had to independently turn the iPad on, slide the iPad screen to unlock it, open the AURASMA app, and hover the iPad over the corresponding trigger image. The video would then automatically start playing, and double tapping the screen would enlarge it to full screen size.
Booklets and checklists
The images necessary to trigger the videos through the AURASMA app were organized in booklets. The image triggering the entire video was placed first, with the separate videos for all the segments following. Each segment trigger image was associated with a number and color coded to match checklists outlining the steps of the task analysis. Thus, the students could choose (a) to watch the full VM the entire task, (b) to watch all the segments in succession, pausing between them to perform a viewed step or series of steps, or (c) to watch only the segments they needed additional assistance with to perform the task correctly. The checklist served as the students’ guide to determine which trigger image, thus video, was associated with each step.
Problems and other materials
The researchers created 24 different problems for each of the four targeted skills. The order in which the problems were presented to the students was randomized and none of the 24 problems was the same as the problem modeled in the video. Example problems can be found in Table A1. All of developed problems and task analyses were vetted by an external expert with a PhD in mathematics education. Depending on the task, other materials would be needed, such as a black dry erase marker, colored dry erase markers, a list of mathematics vocabulary words, a calculator, and a number line. Participants had access to all of these materials during all phases of the study including baseline, intervention, and maintenance. The only item students did not have access to during baseline and maintenance was the AR-based instructional package.
As the skills were being performed, the problem card was placed in front of the student together with other needed materials (e.g., calculator, number line, markers); the checklist was located on the table above the problem card; the booklet with the trigger images was available to the left of the materials; and the iPad was accessed on the right.
Measures and Data Collection
Pretest, posttest, and social validity questionnaires
A pretest and a posttest were created, equivalent in difficulty, addressing all the target skills. The tests were Curriculum Based Assessments (CBAs), both consisting of 10 problems: two each for the skills of rate of change, addition and subtraction of integers, and multiplication and division of integers, with four for the task of operations. Each question was scored as one point. The pretest was administered as part of the selection process before baseline began. The actual administration of the pretest and posttest was on a paper copy of the assessment on which students independently wrote their answers. A question bank of 24 questions was developed and was used to create both the pretest and posttest (see Table A3). Questions from each group were task randomized to create both the pretest and posttest for each participant. The pretest and posttest were both reviewed by an outside expert with a PhD in mathematics instruction to ensure they were appropriate for the material to be tested and to make sure they were both equivalent in difficulty. The posttest was administered after the participants’ final maintenance data point. A social validity questionnaire was distributed for the students to indicate whether they perceived the intervention as useful and functional.
Dependent variable
The dependent variable in this study was the percentage of steps completed correctly according to the task analysis. Although the dependent variable was procedural in nature, varying the procedure was designed to help the students translate the procedural understanding into conceptual understanding. Students should have an understanding of procedural knowledge prior to being able to grasp conceptual understanding. The independent variable was the AR-based intervention package. Data were collected with an observational data checklist that documented and charted the percentage of steps the participant performed correctly for each task. Each step in the task analysis was listed next to a column used to indicate if the step had been performed correctly or incorrectly. If a step was performed correctly, with or without VM and VP, the researcher would mark a plus sign in the corresponding box. If a step was not performed correctly, with or without VM and VP, or was not performed at all, the researcher would mark a minus sign in the corresponding box. The last step for each task was write down the correct answer.
All data collection sessions across all conditions were video recorded. As each participant performed the skills, data on the percentage of steps completed correctly were collected and live recorded by one of the researchers using the previously created data sheets. Recordings of the sessions were later reviewed and scored by a different independent researcher to establish interobserver reliability and calculate kappa.
Interobserver agreement and procedural reliability
Interobserver agreement data were collected for 40% of the sessions in each phase of each problem area for all participants. An observer who had not recorded the first set of data scored the participants’ performance by watching recordings of the different sessions using the same data collection sheets. An agreement index was used to compute reliability assessments by dividing the number of agreements by the total number of agreements plus disagreements. The mean scores indicated an agreement index of 96.9% or higher for each skill for each participant across all phases with a range of 92.3% to 100% across all phases for all participants. Because each step was scored as correct or incorrect there was a 50% chance of rater agreement. To account for the possibility of chance agreement Kappa was also calculated (McHugh, 2012). Kappa was calculated according to the formula provided by McHugh (2012) and across all participants and settings was found to be .89 (95% confidence interval (CI) = [.84, .94]).
Experimental Design
The researchers chose a single-case design replicated across participants due to the nature of the research questions and the intervention. The single-case research design implemented for the current study was consistent with the standards set by the What Works Clearinghouse, which are consistent with the quality indicators for single-case research stated by Horner et al. (2005). Specifically, a multiple probe design was adopted to examine the percentage of steps completed correctly for each task. This design also allowed the researchers to implement the intervention for the steps being completed while baseline data were still being collected on the remaining skills through probes. When a student acquired a targeted skill before the following skills were introduced, a probe was conducted on the remaining skills to confirm the baseline prediction. Then the intervention was introduced for the following skill, while the others remained in baseline phase (Barlow et al., 2009).
The process was repeated until the intervention had been implemented for all skills. Acquisition was defined as completing the task with at least 80% accuracy (80% of the steps completed correctly) for at least three consecutive data collection sessions with a minimum total of five sessions. After acquisition, maintenance data were collected at intervals of approximately 1 week with a minimum of 1 week. Maintenance conditions were similar to baseline, as the individual completed the task without access to any of the materials available during the intervention.
The sequence in which participants received the intervention was randomized by pulling one identifying item blindly out of a cup containing indicators of all four skills. The sequence in which the participants received the problems for each task was also randomized. Each participant’s treatment schedule can be ascertained by examining the graphs of individual results.
Experimental Procedures
Baseline
During baseline each participant was brought individually into the room, where he sat at a table containing all of the materials needed to complete the target task: question, paper, pencil, and a calculator or number line as applicable. The researcher asked the participant to complete the mathematics problem, with no access to the iPad or any other component of the intervention, to create intervention-free baseline conditions. Data on the percentage of steps completed correctly were collected separately for every task based on the specific task analyses. Baseline data were collected for a minimum of five data points, or until stable for at least three consecutive data points.
Training
After initial baseline data had been collected and before the intervention was introduced, a sample task, writing the student’s own name, was used to teach participants how to use the iPad with the AURASMA app. Writing the students name was chosen to teach and test operation of the iPad and Aurasma app. because this task was unrelated to the mathematic content to be taught and required no thought processing that would distract from focusing on the technology. A researcher modeled the necessary steps for turning on the iPad, opening the correct app, triggering the videos through the images, and enlarging the videos to full screen. Like the task, the videos used during the training were not related to any of the targeted mathematic content. Using a task analysis and data sheets related to the sample task, the researchers monitored the participants’ performance of the steps. This process continued until he or she was able to independently perform all the steps in the task analysis.
Intervention
After pretraining, the AR-based intervention was introduced. Similar to baseline, participants had access to all the materials needed to complete the skills, but the iPad with the intervention videos, the booklets with the trigger images, and the checklists. The participants accessed the videos independently, as they had been taught during pretraining. They decided for themselves if they preferred watching the full video, the individual segments, all the videos, just some of them, or none at all. After having watched the full video or after each segment the student would perform the steps to complete the task. Data were collected on the percentage of steps performed correctly for each task. Intervention data were collected for a minimum of five sessions with at least three consecutive sessions attaining at least 80% accuracy in reaching the correct solution. Anecdotal data collected found that at the beginning of the intervention students watched both the complete video and each individual component video as they completed the problems. However, as the intervention progressed students would watch the complete video but would only view the individual component videos for the steps they found challenging.
Maintenance
After criterion had been reached in the intervention data, maintenance data were collected. The maintenance phase returned to baseline conditions, with the participants having no access to the intervention. Data were collected in 1-week increments, starting a week after the conclusion of the intervention phase. Because of the nature of the multiple probe design, quantity of maintenance data varied according to task.
Results
Researchers used visual analysis to analyze individual participant performance data using Ledford’s six data characteristics of visual analysis (level, trend, variability, overlap, consistency, and immediacy; Ledford et al., 2018). Visual analysis supported a functional relation between the introduction of the intervention and the percentage of steps performed correctly for the target skills.
Acquisition of Target Skills
Research Question 1 asked if there was a functional relationship between the percentage of steps completed correctly on multi-step math problems and the use of an AR-based intervention package. As shown in Figures 1 through 4, all seven participants acquired all four skills with 28 demonstrations of a positive effect. With regard to level, trend, and variability, six of the seven participants had stable baseline data, all with zero-celerating trend. As seen in Figure 2, Ben had two skills that started high in baseline and then stabilized at a lower level after four data points. Once the intervention was introduced, five of the participants showed immediate substantial, gains. Molly’s first intervention data point for the multiplication and division of integers skill was the same as her last baseline data point. However, subsequent data points showed an increase. Ben’s first intervention data points for the skills of operations and multiplication and division of integers were equal to or lower than his last baseline data points. They then increased with the second intervention data point. With regard to consistency, data patterns were consistent across all baseline conditions for six of the participants. Ben showed some inconsistency in the data for the rate of change and addition and subtraction of integers skills during the first three baseline data points. Data patterns were consistent across intervention conditions for six of the seven participants for all skills during intervention conditions. For example, Carl showed immediate substantial gains across all skills needing four data points to reach 100% mastery. Molly’s data showed some inconsistency with a dip in performance on the operations skill during the fourth intervention data point; however, her performance improved following two additional intervention sessions. Similarly, changes in data were generally consistent and in the expected direction for six of the seven participants across all skills with the exception of the dip in the operations task for Molly. All seven participants’ data showed a large positive increase in level between baseline and intervention conditions across all four skills. With regard to overlap, we found no overlapping data between baseline and intervention conditions for four of seven participants. Molly and Maddie had one overlapping data point and Ben had three overlapping data points. With regard to immediacy, all condition changes from baseline to intervention resulted in immediate changes in level for five of the seven participants, with the first data point in each condition being different in level from the data point in the preceding condition in the expected direction. Molly’s first intervention data point for multiplication and division of integers was the same as the last baseline data point, and Ben’s first intervention data point for addition and subtraction of integers was level with his last baseline data point but increased with the second intervention data point. On the multiplication and division of integers, his first intervention data point was significantly lower than his last baseline data point, but his second intervention data point reached 100%.
Percentage of steps completed correctly by Carl and Molly.
Percentage of steps completed correctly by Ben and Josh.
Percentage of steps completed correctly by Kelly and Maddie.
Percentage of steps completed correctly by Maggie.
Maintenance of Target Skills
Research Question 2 examined whether participants maintained the target skills and the ability to independently generalize the problem-solving skills correctly to other problems without the use of the AR-based intervention. As shown in Figures 1 through 4, all seven participants maintained at least one of the four skills at intervention levels. Overall, 16 of 28 skills were maintained at or above intervention levels. With regard to level, trend, and variability, level was above baseline levels for all participants and equal to or higher than the final intervention data point for 16 of the 28 skills. For the other 12 skills, the maintenance level was below intervention levels. The trend for 16 of the 28 skills was consistent, with five of the skills showing increasing trend. Seven skills had just one maintenance data point, preventing an analysis of trend. Maintenance data showed more variability with trend compared with baseline and intervention conditions across skills and individuals. With regard to consistency, data patterns were less consistent during maintenance than other phases for six of the seven participants. However, Maggie’s maintenance data were higher or equal to her intervention data. With regard to overlap, there was overlapping data across all participants and skills. In total, 17 of the 28 skills had overlapping intervention and maintenance points at 100%. With regard to immediacy, there was an immediate decrease of 40% in nine of the 28 skills across six of the participants. For example, Figure 2 shows that Ben went from 100% on the final intervention data point to 60% on the first maintenance data point for both the addition and subtraction of integers and multiplication and division of integers skills. All of these initial decreases during maintenance ended with an upward trend.
Social Validity
The third research question looked at the social validity of the intervention. We used a social validity questionnaire, requesting that participants respond to nine open-ended items inquiring about their experience and success during the study and asking their opinion about its usefulness. All participants who completed the survey responded that they had enjoyed watching the videos, that the videos had helped them learn new skills, and that after the study they felt more comfortable about the skills they had learned. All participants indicated that the AURASMA app delivering the videos was easy to use and had helped them apply the step-by-step instruction without getting stuck. All participants also expressed their interest in receiving instruction in other subjects (science, reading, social skills) and other mathematics skills (functions). Overall, the participants expressed satisfaction about the study and the way it had helped them learn to perform certain mathematics skills; several indicated they had talked about it and discussed their improvement with other people, including school teachers, close friends, and family.
Discussion
This study has contributed to the literature supporting the effectiveness of AR in the field of education as a means of prompting VM and VP intervention videos, to increase the percentage of mathematics steps completed for individuals with SLD. Results indicated a functional relationship between the dependent variable, the percentage of steps performed correctly for each task, and the independent variable, the AR intervention package.
Acquisition of Target Skills
In response to RQ1, all seven of the participants showed a substantial improvement in the percentage of steps they completed correctly in each task immediately after the intervention package was introduced; all participants eventually reached 100% accuracy in all skills. The results of this study are consistent with results from similar studies addressing the effectiveness and validity of AR in teaching various skills. The results of this study support those of the study conducted by Kellems et al. (2016) on the effectiveness of VP in teaching multistep mathematics skills to young adults with various disabilities. The current study and that by Kellems et al. (2016) demonstrated successful delivery of mathematics instruction, measured by percentage of problem-solving steps completed correctly, through VP to individuals with disabilities including those with SLD.
The results of the present study additionally support the limited but increasing literature supporting use of AR in the field of education, in particular with individuals with SLD. Specifically, this study affirms the findings of Sommerauer & Müller’s (2014) large field experiment determining the effectiveness of AR in delivering instruction about formal content (i.e., mathematics concepts at a mathematics exhibit) in both formal and informal environments. The findings of both this study and the Sommerauer study reinforce that procedural accuracy when completing mathematics can be enhanced through the use of AR. The present study also demonstrates effectiveness of AR in teaching procedural steps in mathematics to individuals with SLD.
Maintenance of Target Skills
Results from RQ2, in which we examined whether participants could maintain the skills, were inconclusive. Although all participants did maintain some of the previously acquired skills, one pattern that emerged was that six of the seven participants showed an immediate decrease in the percentage of steps completed correctly for at least one task at the beginning of the maintenance phase. It should be noted that all participants’ maintenance data that was initially low ended on an increasing trend. These data suggest that some students may need continued prompting and review of the procedures as they complete new problems. Another possibility is that subsequent prompting after initial skill acquisition can be more brief, but consistent.
Social Validity
Data collected regarding the social validity of the intervention was overwhelmingly positive consistent with the findings of other video and AR-based interventions such as those conducted by Kellems et al. (2016) and McMahon et al. (2016). The positive feelings toward the intervention may have to do with the individuals’ preference for technology-based instruction technology or the way the content was presented. It should be noted that no social validity data was collected from teachers or parents.
Study Limitations
A number of limitations of this study should be noted and addressed in future research. One limitation is that in the current study the dependent variable did not have any measures for students’ conceptual understanding of the problem or problem-solving accuracy, only the number of steps completed correctly was recorded. As the technology used in this study was developed in 2011 and is still quite new, relatively little research has been done to date exploring its potential as an instructional tool—as well as its availability.
An additional limitation was that most of the mathematic problems were written following the same format, creating a possible question of whether students understood the difference between the various components of each type of problem or had just memorized the format of the problem, which would present a problem for generalization of the different skills. A final limitation is that no formal reliability data was collected on the pre/posttest CBA was collected other than their validation by a content expert. Another point for consideration is the possibility that one of the participants, Ben potentially did not have SLD based on the assessment information provided.
Suggested Future Research and Implications
Future research should explicitly measure the accuracy of responding in addition to measuring the steps completed correctly. Future research should also focus on delivering the intervention to participants with a wider range of disabilities and different levels of mathematical ability to determine what groups the intervention is effective in helping. Future research should also explore how the use of this or a similar intervention might be used to teach a wider range of mathematics skills (geometry, etc.) once it has been determined in future studies if the current intervention is effective at improving math accuracy. It would also be valuable to conduct further research on how a similar intervention could be used to teach functional mathematic skills to individuals with disabilities.
Future studies should incorporate a measure for conceptual understanding of the problems or skills; for example, whether the students can explain why they used a particular method to solve a problem or whether another method might have resulted in the same solution. Furthermore, in the present study the intervention was delivered only via iPads; further research might consider whether different devices, such as phones, mini iPads, and different tablets, could be used to achieve the same result. Additional research should also be conducted on best methods for practitioners implementing AR-based interventions with their student such as if the intervention is something that needs to be teacher supervised or if it can be used independently by students,
Conclusion
The present study examined the use of an AR intervention package for teaching procedural steps to solving mathematics problems under the common core to middle school students with SLD. We identified a functional relation between the percentage of steps completed correctly on selected math problems and the AR-based intervention package All students showed a substantial improvement in their performance after receiving the intervention. However, not all of the gains were maintained. Further research is needed to provide evidence of improving mathematics accuracy and conceptual understanding to students with disabilities.
Footnotes
Appendix
Average Individual and Overall Accuracy in Completing Math Problem Solving Steps.
| Participants | Operations (CCSS.MATH.CONTENT.MP1) |
Integers: addition and subtraction (CCSS.MATH.CONTENT.7.NS.A.1) |
Integers: multiplication and division (CCSS.MATH.CONTENT.7.NS.A.2) |
Rate of change (CCSS.MATH.CONTENT.6.RP.A.3.D) |
||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| BL (%) | INT (%) | MT (%) | BL (%) | INT (%) | MT (%) | BL (%) | INT (%) | MT (%) | BL (%) | INT (%) | MT (%) | |
| Carl | 24 | 89 | 72 | 45 | 94 | 70 | 60 | 100 | 100 | 33 | 100 | 100 |
| Molly | 27 | 87 | 100 | 13 | 96 | 43 | 20 | 78 | 100 | 12 | 100 | 100 |
| Ben | 17 | 100 | 100 | 45 | 87 | 50 | 53 | 85 | 60 | 17 | 87 | 100 |
| Josh | 30 | 88 | 97 | 16 | 95 | 60 | 31 | 100 | 20 | 13 | 98 | 100 |
| Kelly | 24 | 93 | 83 | 14 | 90 | 80 | 20 | 100 | 20 | 16 | 100 | 100 |
| Maddie | 17 | 100 | 100 | 18 | 84 | 67 | 43 | 95 | 20 | 11 | 100 | 100 |
| Maggie | 17 | 92 | 92 | 14 | 81 | 100 | 28 | 100 | 100 | 11 | 100 | 100 |
| Overall | 22 | 91 | 92 | 23 | 90 | 67 | 36 | 94 | 57 | 16 | 98 | 100 |
Note. BL = baseline; INT = intervention; MT = maintenance.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
