729508

research-article2017
RSEXXX10.1177/0741932517729508Remedial and Special EducationBarton et al.

Literature Reviews
Remedial and Special Education

Technology-Aided Instruction and

1­–16
© Hammill Institute on Disabilities 2017
Reprints and permissions:
Intervention for Students With ASD: A sagepub.com/journalsPermissions.nav
DOI: 10.1177/0741932517729508
https://doi.org/10.1177/0741932517729508

Meta-Analysis Using Novel Methods of rase.sagepub.com

Estimating Effect Sizes for Single-Case

Research

Erin E. Barton, PhD1, James E. Pustejovsky, PhD2,

Daniel M. Maggin, PhD3, and Brian Reichow, PhD4

Abstract
The adoption of methods and strategies validated through rigorous, experimentally oriented research is a core professional
value of special education. We conducted a systematic review and meta-analysis examining the experimental literature on
Technology-Aided Instruction and Intervention (TAII) using research identified as part of the National Autism Professional
Development Project. We applied novel between-case effect size methods to the TAII single-case research base. In addition,
we used meta-analytic methodologies to examine the methodological quality of the research, calculate average effect sizes
to quantify the level of evidence for TAII, and compare effect sizes across single-case and group-based experimental
research. Results identified one category of TAII—computer-assisted instruction—as an evidence-based practice across
both single-case and group studies. The remaining two categories of TAII—augmentative and alternative communication
and virtual reality—were not identified as evidence-based using What Works Clearinghouse summary ratings.

Keywords
meta-analysis, single-case design, effect size, technology-aided instruction and intervention

The development, identification, and dissemination of evi-
dence-based interventions have been dominant themes in
special education for more than a decade (Odom et al.,
2005). Following the logic of evidence-based practice,
experimental research methods are used to determine which
strategies and practices are most likely to produce desired
outcomes (Council for Exceptional Children [CEC], 2014;
Horner et al., 2005; Odom et al., 2005). A foundational
tenet of evidence-based practice is the use of transparent,
replicable, and objective methods for reviewing studies to
distinguish those practices with sufficient empirical support
to warrant their use in schools and classrooms. These prin-
ciples are applicable regardless of the particular type of
research being reviewed and should be extended to all
aspects of the review process (What Works Clearinghouse
[WWC], 2014).
Despite the need for transparent and replicable proce-
dures, there remain challenges with available methods for
assessing, analyzing, and synthesizing results from single-
case research as part of a systematic review and meta-anal-
ysis (Shadish, Hedges, Horner, & Odom, 2015). The aim of
this article is to demonstrate the use of a combination of
novel methods for conducting systematic reviews and meta-
analyses in a synthesis of the research base on Technology-
Aided Intervention and Instruction (TAII) for students with
autism spectrum disorders (ASD).
Technology-Aided Intervention and
Instruction
TAII refers to a range of interventions in which technology
is used as the primary method to deliver instruction (Wong
et al., 2015). Technology is defined as any electronic appa-
ratus or virtual network that is used to target a particular
academic, social, or behavioral skill of the student (Odom
1
Vanderbilt University, Nashville, TN, USA
2
The University of Texas at Austin, USA
3
The University of Illinois at Chicago, USA
4
University of Florida, Gainesville, USA
Corresponding Author:
Erin E. Barton, Vanderbilt University, 230 Appleton Place, Peabody 228,
Nashville, TN 37203, USA.
Email: erin.e.barton@vanderbilt.edu
2 Remedial and Special Education 00(0)
et al., 2015). TAII includes computer-assisted instructional
(CAI) programs and speech generating devices that are
more specifically focused on communication outcomes
(Lofland, 2016; Odom et al., 2015). The communicative,
social, and behavioral challenges that confront students
with ASD require the development and implementation of
empirically supported practices to target their idiosyncratic
symptomatology and functional learning outcomes
(Reichow, Doehring, Cicchetti, & Volkmar, 2011). TAII has
been used to teach students with ASD facial and emotional
recognition, safety skills, and reading and vocabulary out-
comes (Jones, Wilcox, & Simons, 2016). A variety of inter-
ventions and outcomes can be addressed through TAII, with
tools that range in sophistication and modality from com-
plex virtual reality (VR) programs to basic touch screen and
icon communication systems. Furthermore, TAII includes
an assortment of devices and delivery formats such as lap-
top computers, smart phones and tablets, and virtual net-
works and applications (Wong et al., 2014).
In an effort to provide school personnel and related ser-
vice providers with guidance on which practices were most
effective for a range of academic, behavioral, and social
outcomes for students with ASD, Wong and colleagues
(2015) conducted a large-scale systematic review of the
ASD intervention literature. TAII was among the interven-
tion categories identified as evidence-based by Wong and
colleagues (2015). The broad scope of the project did not
allow for analysis of specific details regarding procedures,
risk of bias, or outcomes, or the relative strengths of the
interventions, which might have important practical impli-
cations. Other extant systematic reviews and meta-analyses
examining TAII have supported their use on a range of aca-
demic, communication, and social outcomes for students
with ASD (Grynszpan, Weiss, Perez-Diaz, & Gal, 2014;
Odom et al., 2015; Pennington, 2010; Ramdoss et al., 2011;
Weng, Maeda, & Bouck, 2014). Collectively, these reviews
indicated that TAII has the potential to be an effective class
of interventions, though additional research is needed to
refine technologies that have the greatest potential for spe-
cific outcomes. A common shortcoming of these reviews is
that they treated single-case and group design research
using separate, incomparable methodologies. For instance,
Ramdoss and colleagues (2011) used standardized mean
difference effect sizes to describe effects from group
designs, but a nonoverlap index to describe effects from
single-case designs. Grynszpan and colleagues (2014)
excluded single-case studies from their review. Knight,
McKissick, and Saunders (2013) compared contemporary
quality indicators across group and single-case studies,
respectively; however, they did not conduct meta-analyses
or examine the relative strengths of the interventions. Thus,
there remains a need to develop and test risk of bias assess-
ments and effect size indices that can be compared across
both types of research designs.
Risk of Bias Assessment
Part of conducting a systematic review involves the critical
appraisal of primary research studies. The Risk of Bias
(RoB) assessment tool of the Cochrane Collaboration
(Higgins, Altman, & Sterne, 2008) is often considered one
of the best tools for the appraisal of randomized controlled
trials (Sterne, 2009). Group RoB tools assess selection bias
(systematic differences between baseline characteristics of
the groups), performance bias (systematic differences
between groups in the care that is provided or in exposure to
factors other than the interventions of interest), attrition bias
(systematic differences between groups in withdrawals
from a study), detection bias (systematic differences
between groups in how outcomes are determined), and
reporting bias (systematic differences between reported and
unreported findings). The Cochrane RoB tool has been
adapted for use in reviews of nonrandomized studies (B. C.
Reeves et al., 2013), and recently for use with single-case
designs (SCDRoB; Reichow, Barton, & Maggin, 2017).
The SCDRoB was developed based on current conceptual-
izations of the types of bias that could affect empirical sin-
gle-case research (Sterne, 2009). Although Tate and
colleagues (2013) also developed a risk of bias tool for
single-case research, their criteria and bias domains do not
align with the Cochrane RoB tool, thus precluding compari-
sons across group and single-case research methodologies.
Using RoB tools with consistent domains allows for a com-
prehensive comparison of internal validity of a body of lit-
erature including both group and single-case research.
Between-Case Effect Sizes
Hedges, Pustejovsky, and Shadish (2012, 2013) recently
introduced a distinct conceptualization of effect sizes for
single-case research by constructing an effect size index
that is directly comparable to the standardized mean differ-
ence effect size commonly used with between-groups
research designs. This novel effect size index—the between-
case standardized mean difference (BC-SMD)—differs
from other effect sizes commonly used with single-case
designs because it characterizes average treatment effects
at the level of the study, rather than at the level of the indi-
vidual case. It is defined based on a hierarchical model for
the data from a single-case design, which captures both
within-case and between-case variation in the dependent
variable. Using such a hierarchical model provides a way to
estimate the same standardized mean difference parameter
that would be estimated by a hypothetical between-groups
experimental design on the same population (Pustejovsky,
Hedges, & Shadish, 2014). Effect size estimates from each
type of design therefore share a common scale.
Shadish et al. (2015) suggested that the advantages of
BC-SMDs are that (a) they allow researchers to compare the
Barton et al. 3
results of single-case research with the results of group
designs and (b) they are in a metric that is familiar to educa-
tional researchers who work primarily with between-subjects
designs. Furthermore, because BC-SMDs are on the same
scale as standardized mean differences from between-groups
designs, results from both types of designs can be compared
or even combined in a single meta-analysis, at least in prin-
ciple. In practice, of course, single-case studies and between-
group studies might differ in dimensions other than just the
research design, such as participant inclusion criteria, class of
outcome measures, or length of follow-up. Thus, there
remains a need to consider the extent to which evidence from
single-case studies aligns with, or diverges from, evidence
from between-groups studies on a common topic.
In addition to questions of conceptual comparability,
methods for estimating BC-SMDs are currently available
only for certain types of single-case designs. Specifically,
because the BC-SMD involves between-case variation, it
can only be estimated for studies that include multiple indi-
vidual cases, such as across-participant multiple baseline
designs, across-participant multiple probe designs, or with-
drawal (A-B-A-B) designs that are replicated across partici-
pants. Therefore, it is important to examine the extent to
which this technical limitation has consequences for appli-
cation of the BC-SMD in practice. The present review con-
siders the implications of these limitations by calculating
BC-SMD estimates for single-case studies—and comparing
the results with standardized mean difference estimates
from group design studies—on the effects of TAII for stu-
dents with ASD.
Purpose and Research Questions
The purpose of this article is to demonstrate the use of a
combination of novel methods for conducting systematic
reviews and meta-analyses in a synthesis of the research
base on TAII for students with ASD. In doing so, we
expanded and updated the review conducted by Wong and
colleagues (2015) regarding the use of TAII for students
with ASD. Specifically, we evaluated studies for risk of bias
using a framework that can be applied to both types of
designs, we quantified the magnitude of effects using
BC-SMDs, and we synthesized the effect sizes using formal
meta-analytic models. In demonstrating the application of
these recently developed methods to the research base on
TAII, we hoped to stimulate further discussion about their
strengths and limitations as tools for synthesizing single-
case research and thereby informing evidence-based prac-
tice in special education. This included examining the
quality of the research, calculating effect sizes to quantify
the level of evidence for TAII, and comparing single-case
and group-based experimental research effect sizes. Given
that TAII can be implemented in a variety of ways and can
focus on a range of outcomes, we disaggregated the studies
included in the recommendations and closely evaluated the
contributing studies to provide precise guidance and recom-
mendations for research. Two research questions guided our
work:
Research Question 1: To what extent is TAII more
effective than business as usual or as compared with
other interventions in improving targeted communica-
tion, academic, engagement, social, emotion recogni-
tion, and adaptive outcomes for students with ASD?
Research Question 2: Do the group and single-case
experimental research bases provide sufficient empirical
support for particular independent-dependent variable
combinations (i.e., augmentative and alternative com-
munication [AAC], CAI, VR) to be classified as evi-
dence-based practices for students with ASD?
Method
The application and evaluation of novel methods for the
calculation of single-case design data included an update of
the review of TAII by Wong and colleagues (2015). We
used methods consistent with recommendations of Cochrane
for updating systematic reviews and meta-analyses and
adhered to the guidelines for systematic reviews and meta-
analyses proposed by Preferred Reporting Items for
Systematic Reviews and Meta-Analyses (PRISMA; Moher,
Liberati, Tetzlaff, Altman, & PRISMA Group, 2009).
Eligibility Criteria and Study Identification
To be eligible for inclusion in this review, studies were
required to meet several criteria. First, the study evaluated
the effects of TAII as compared with a control condition or
non-TAII-based instruction (as defined by Odom et al.,
2015). Second, the majority of participants in the study were
reported to have ASD. Third, the study used an eligible
between-group design or single-case design. Group designs
(i.e., those involving comparisons across participants) were
eligible if they used a nonequivalent groups quasi-experi-
mental design or a randomization process for allocating par-
ticipants to conditions. Single-case studies were eligible if
they were experimental (i.e., at least three opportunities to
demonstrate behavior change at three different points in
time). Studies had to meet further eligibility criteria to be
included in the effect size calculations and meta-analysis, as
explained in more detail in the next sections.
We started with the original pool of studies identified by
Wong and colleagues (2015) and screened all for eligibility.
Studies included in their review were identified through
electronic searches of education and social science research
databases and appraisal of relevant literature reviews, sup-
plemented by ancestral searches of identified studies. We
also screened for eligibility studies identified in previous
4 Remedial and Special Education 00(0)
reviews by Ganz and colleagues (2012), Irish (2013),
Kagohara and colleagues (2013), Knight et al. (2013),
Ploog, Scharf, Nelson, and Brooks (2013), Odom and col-
leagues (2015), and Ramdoss and colleagues (2012). In
addition, the first author conducted an electronic search of
the following databases during February of 2015 and lim-
ited the date of publication from 2012 to 2015: Educational
Resources Information Center (ERIC), ProQuest,
PsycINFO, and PubMed. Key terms used in the electronic
search included ([autism or Asperger or pervasive develop-
mental disorder] AND [intervention or treatment or prac-
tice or strategy or therapy or program or procedure or
instruction] AND [technology or computer or iPod or iPad
or computer assisted instruction or computer aided instruc-
tion or device or personal digital assistant or app or virtual
or electronic]). The lower limit was set at 2012 because
Wong and colleagues (2015) searched through 2011, and
the search was intended to supplement their findings. A
PRISMA flow diagram (Moher et al., 2009) of study inclu-
sion, incorporating recommendations for the display of
updated reviews (Stovold, Beecher, Foxlee, & Noel-Storr,
2014) is provided in Online Appendix A.
Study Coding Procedures
Four concurrent coding procedures were used. First, two
special education graduate students extracted descriptive
information from all studies related to the following vari-
ables: participant and setting characteristics, target outcomes
and measures, intervention procedures and dosage, technol-
ogy characteristic and platform (using categories described
by Odom et al., 2015), study design, and study-reported
results. Second, the first, third, and fourth authors assessed
the methodological rigor of all studies using two frame-
works: (a) WWC study design standards as outlined in the
WWC Procedures and Standards Handbook 3.0 (2014) and
(b) an adaptation of Cochrane’s risk of bias tool (Higgins
et al., 2008) for group research design studies and an adapta-
tion of the tool for single-case studies (Reichow et al., 2017).
Third, a graduate student extracted all data from the single-
case studies. Fourth, the second and third authors extracted
summary statistics and effect size information from the
group design studies. The first author contacted authors of
four studies where effect sizes could not be computed for all
variables based on published information. Two of the four
authors responded but none could provide the needed data.
Descriptive characteristics. Descriptive characteristics were
coded for all identified group design and single-case studies
using a systematic coding protocol. Given the heteroge-
neous nature of TAII, this descriptive information was
important for understanding conditions under which TAII
was effective. All entries were independently coded by the
special education graduate students and discrepancies were
discussed with the first author until consensus was reached.
There were discrepancies on fewer than 5% of the entries
across coded variables.
Participants and setting characteristics.  The graduate stu-
dents extracted information regarding the age, ethnicities,
disability status, confirmation of ASD diagnosis, descrip-
tions of participants’ functional repertoires, and the mea-
surement of technology skill levels. This information was
extracted and summarized to provide further understanding
of the populations for whom TAII has been investigated.
The country of origin of the study and the physical location
of the intervention sessions were also extracted.
Target outcomes and measures.  Three items related spe-
cifically to the dependent variables. These items were (a)
primary skills targeted, (b) categorical domain of the pri-
mary target skills, and (c) the functional relevance of the tar-
get skills for the participants. The categories for the primary
skills were mutually exclusive and exhaustive and included
communication, academic, engagement/task completion,
social, emotion recognition, and adaptive. The functional
relevance of the skill was rated as yes if the skills were
needed for participants in most or all situations to facili-
tate independence, used across most daily activities and
routines, or occasioned learning more sophisticated or com-
plex functional skills; if two of the previous three clauses
were not true of the primary target skill, the relevance was
rated as no. Several items related specifically to the mea-
surement of the dependent variable. First, the measure-
ment procedures for the primary dependent variable were
coded as direct observation or standardized assessments (or
both). Studies using direct observation were further coded
as using event recording, percentage correct (i.e., accuracy),
or an interval system (e.g., partial interval recording). Sec-
ond, measurement of generalization was coded as occurring
across people, settings, materials, activities, or measures.
Third, the length of time between the posttreatment data
collection and maintenance data collection for the primary
dependent variable was coded.
Intervention procedures and dosage.  Several items related
specifically to the intervention (i.e., independent variable).
The intervention agent was coded as researcher, classroom
staff, external therapist, peer, or parent. Regarding inter-
vention dosage, the total number of instructional sessions
and the number of trials per session were coded. Additional
dosage information was extracted when provided (e.g.,
duration of sessions, rate of sessions per week or month).
The name of the instructional procedure was extracted
along with the authors’ brief description. The exact TAII
materials used were listed. The technology characteristics
and platforms were coded using categories identified from
previous reviews (Odom et al., 2015). These were speech
Barton et al. 5
generating device, personal computer, Internet/web, mobile
device, shared active surface, VR, sensory/wearable tech-
nology, robotics, and natural user interface; multiple cat-
egories could be used in each study. The interventions were
further coded into three mutually exclusive categories: (a)
AAC, (b) CAI, and (c) VR.
Study design and methodological features. The exact
designs used across studies were listed. The graduate stu-
dents coded the single-case studies as a withdrawal (A-B-
A-B, coding the exact notation); multiple baseline across
participants, behaviors, or stimuli; multiple probe across
participants, behaviors, or stimuli; or alternating treatment
design. For group designs, the unit of randomization and the
group comparison categories were listed.
Risk of Bias
We evaluated study-level RoB using an adaptation of the
Cochrane’s RoB tool (Higgins et al., 2008) for group
research design studies, which incorporated concerns for
inclusion of nonrandomized studies (e.g., L. M. Reeves,
Umbreit, Ferro, & Liaupsin, 2013), and a single-case adap-
tation of the tool: SCDRoB (Reichow et al., 2017). The
RoB tool uses a domain-based evaluation system so that
critical assessments are made separately for different
domains relevant to the internal validity of the study. The
third and fourth authors independently coded each study
using the respective RoB tools and reviewed and discussed
disagreements until consensus was reached.
Visual Analysis
The graduate student coders used a multistep process for
coding single-case study outcomes. First, coders used a sys-
tematic visual analysis protocol (adapted from Gast &
Spriggs, 2014; Kratochwill et al., 2013) to assess the pres-
ence or absence of a functional relation for each case. A
case was conceptualized as at least three opportunities to
demonstrate behavior change at three different points in
time within or across participants, target behaviors, settings,
or materials. Second, coders reviewed and entered the
results as described by the study authors. Third, the first
author reviewed the coders’ assessment of functional rela-
tions and the authors’ description of results and coded as
agreement or disagreement. This multistep coding process
was established to reduce the likelihood that author bias
affected their assessment of functional relations.
Statistical Analysis
Group design and single-case studies were meta-analyzed
both separately and jointly, using effect size indices
designed to be comparable across both types of designs.
Group design effect sizes.  Of the 12 group design studies,
data for calculating effect sizes were available from 10.
Several studies reported the results of more than one experi-
ment or for more than one sample; in these cases, each sam-
ple was treated as a separate, independent study. Estimates
of the SMD were calculated for each outcome in each sam-
ple based on the posttest difference in means and the pooled
posttest standard deviation. Estimates were corrected for
small-sample bias using Hedges’ g (Hedges, 1981).
Single-case effect sizes. To calculate effect sizes for the
included single-case research, raw data from each study
were digitally extracted from published single-subject
graphs using WebPlotDigitizer (Rohatgi, 2014), which has
high intercoder reliability (Drevon, Fursa, & Malcolm,
2017). One study was excluded from the statistical analyses
because the graphs did not provide discernible data and data
points could not be accurately extracted (Ke & Im, 2013).
The remaining 22 single-case studies had a total of 43
graphs. All 43 graphs were digitized using the WebPlotDig-
itizer program.
The BC-SMD (Hedges, Pustejovsky, & Shadish, 2012,
2013) was calculated for purposes of quantifying the mag-
nitude of treatment effects in the single-case research. The
BC-SMD effect size is premised on a statistical model with
the following assumptions: (a) the baseline is stable (e.g.,
no trend), (b) the intervention leads to an immediate change
in level (e.g., no intervention-phase trend), (c) the interven-
tion effect is constant across cases, (d) the outcome is nor-
mally distributed about case- and phase-specific mean
levels, and (e) deviations from mean levels follow a first-
order autoregressive process. The final assumption accounts
for potential autocorrelation arising from repeated measure-
ment of cases.
A limitation of the BC-SMD is that it can only be cal-
culated for studies that include at least three participants
and that use an across-participant multiple baseline,
across-participant multiple probe, or withdrawal (i.e.,
A-B-A-B) design. Analysis of the BC-SMD was therefore
limited to 10 studies that used an eligible design and
included at least three cases. Some studies with complex
designs were also included because a subset of cases could
be extracted to create an eligible design. For example,
Cihak, Wright, and Ayres (2010) used a multiple probe
design across settings with an embedded A-B-A-B design,
all replicated across three cases; for this study, BC-SMD
estimate was calculated based on the embedded A-B-A-B
series across the three participants. Two studies reported
data on multiple outcomes for the same set of cases (Choi,
O’Reilly, Sigafoos, & Lancioni, 2010; King et al., 2014);
here, separate effect size estimates were calculated for
each outcome. Calculations were carried out using the
scdhlm package (Pustejovsky, 2016) for the R statistical
computing environment.
6 Remedial and Special Education 00(0)
Meta-analysis. Effect size estimates from the single-case
studies and group design studies were synthesized using
robust variance estimation methods (Hedges, Tipton, &
Johnson, 2010; Pustejovsky & Ferron, 2017) to address the
problem that most studies contributed multiple, noninde-
pendent effect size estimates. Most of the group design
studies included effect size estimates for multiple outcome
measures based on a common set of participants. It is not
generally reasonable to assume that effect size estimates
based on a common sample are independent, which pre-
cludes use of standard random effects models for meta-
analysis. Similarly, some of the single-case studies reported
multiple outcomes, for which separate BC-SMD estimates
were calculated. Thus, robust variance estimation with a
“correlated effects” working model (Hedges et al., 2010;
Tanner-Smith, Tipton, & Polanin, 2016) was used to syn-
thesize the SMDs for group design and the BC-SMDs for
single-case research. Given the relatively small number of
included samples, small-sample adjustments for hypothesis
tests and confidence intervals (CIs; Tipton, 2015; Tipton &
Pustejovsky, 2015) were used for all analyses. The meta-
analysis was conducted using the robumeta package (Fisher
& Tipton, 2015) and clubSandwich package (Pustejovsky,
2016) for the R statistical computing environment.
Outcome reporting bias analysis.  Outcome reporting biases,
including publication bias and other forms of outcome cen-
soring, are an important threat to the validity of meta-anal-
yses of group designs (Rothstein, Sutton, & Borenstein,
2005). With group designs, it is typically assumed that out-
come reporting biases arise because statistically significant
results are more likely to appear in the published literature.
In the present review, outcome reporting bias for the group
design studies was assessed using visual inspection of fun-
nel plots, the trim-and-fill procedure (Duval & Tweedie,
2000), and a version of Egger’s regression test (Egger,
Smith, Schneider, & Minder, 1997), adapted for use with d
statistics and robust variance estimation. The trim-and-fill
procedure and Egger’s regression test use a modified stan-
dard error of the effect size estimate to avoid induced cor-
relation between the d estimate and its standard error.
Publication and outcome reporting biases are also
potential threats in syntheses of single-case research.
Recent initial evidence indicates that published single-case
studies might not be representative of the full research base
(Shadish, Zelinsky, Vevea, & Kratochwill, 2016; Sham &
Smith, 2014). However, the process that leads to publica-
tion bias in the single-case literature is likely to be driven
by factors such as visual determinations of experimental
control and functional relationships, rather than by the sta-
tistical significance of results, as in the group design litera-
ture (Shadish et al., 2015). Consequently, tools for
analyzing publication bias in group designs are not ideally
suited for application to single-case research, yet neither
are alternative methods available. Lacking methods that
are better suited for single-case research, the present review
employed conventional publication bias analyses for group
designs (i.e., funnel plots, trim and fill, and robust Egger’s
regression test), but interpreted the results cautiously. This
approach is consistent with at least one previous meta-
analysis using the BC-SMD (Shadish, Hedges, &
Pustejovsky, 2014).
Summary Identification of Evidence-Based
Practices
The final WWC design standard ratings for all single-case
and group design studies were applied to the summative
analyses here but are not reported individually. For single-
case studies, we applied the WWC (2014) and Horner and
colleagues (2005) criterion of five studies with 20 partici-
pants across three research groups to the single-case stud-
ies. We included single-case studies that met the following
criteria in our analysis: (a) visual analysts identified a func-
tional relation, (b) raters indicated the study had three or
fewer domains at high risk of bias, and (c) the study met
WWC design standards with or without reservations. For
group design studies, we applied the WWC criteria for rat-
ing the effectiveness of an intervention (WWC, 2014).
These criteria specify that, to receive a positive effect rat-
ing, an intervention needed to have two studies showing
statistically significant positive effects with at least one
meeting group design standards without reservations and no
negative effects.
Results
Study Selection
Searches resulted in identification of 51 eligible studies
based on title and abstract review, including 20 studies that
were included in the previous review by Wong and col-
leagues (2015), 59 studies identified through electronic
searches, and two studies identified through ancestral search
and appraisal of other literature reviews. These 81 records
were screened and 30 were excluded. Full-text evaluation
of the 51 identified studies resulted in exclusion of 16 addi-
tional studies, including 1 group design that did not use ran-
dom assignment (Williams, Wright, Callaghan, & Coughlan,
2002) and 15 single-case studies that were not experimen-
tal—including one that was included in Wong and col-
leagues’ review (Stromer, Mackay, Howell, McVay, &
Flusser, 1996). The fourth author conducted a reliability
check using the inclusion and exclusion criteria and inter-
rater reliability was 100% for identified studies. The 35
studies retained in the current analysis included 12 group
designs and 23 single-case studies. Of these, 10 group
designs and 10 single-case were eligible for meta-analysis.
Barton et al. 7
Descriptive Characteristics
Participants and setting characteristics.  A total of 540 partici-
pants were included across the 35 studies. This included
321 participants receiving a TAII and 219 in a control or
comparison group. The mean age of participants was 11.5
years (range = 2–52 years). Thirty-three studies included
only participants younger than 19 years old; two studies
included children older than 19 years (Golan & Baron-
Cohen, 2006; Smith et al., 2014). Seventy-one female and
320 male participants were included; three studies did not
report participant gender. A total of 482 participants were
reported to have ASD; 21 had other disabilities and 37 were
typically developing. Researchers in five of the 35 studies
conducted an independent diagnostic evaluation to confirm
ASD diagnoses. Race/ethnicity was reported in nine of the
35 studies for 169 participants. For particpants for whom
race/ethnicity were reported, 58% (n = 98) were White,
24% were Black (n = 40), 8% were Asian (n = 13), 8% were
multiracial (n = 13), and 3% were Latino/a (n = 5). Research-
ers in 22 of the 35 studies provided descriptions of the func-
tional repertoires of participants. Researchers in two of the
35 studies assessed participant technology skill level prior
to commencing their study. Interventions were primarily
conducted in classrooms (n = 16 studies) and separate
school rooms (n = 11 studies); other intervention settings
included homes (n = 5), clinics (n = 4), and community cen-
ters (n = 1). Over half of the studies occurred in the United
States (n = 26). Three were conducted in the United King-
dom, one in Taiwan, and one in Australia. Researchers in
four studies did not report the country of origin.
Dependent variable characteristics.  Table 1 lists the skills tar-
geted and measurement procedures used in the group design
and single-case studies. The target skills in 32 of the 35
studies were assessed to be functionally relevant and
researchers in 30 studies provided a specific rationale for
the identified targets. Researchers in eight of the 35 studies
measured generalization of skills across settings (n = 3),
materials (n = 2), skills (n = 1), activities (n = 1), or people
(n = 1). Researchers in 13 of the 35 studies measured main-
tenance of target behaviors after the intervention ended. The
length of time between the end of the intervention and
maintenance assessments ranged from 1 week to 12 weeks.
Independent variable characteristics and dosage.  Table 1 pro-
vides summaries of intervention agents, dosage, and tech-
nology devices. Across all 35 studies, the intervention was
most likely to be implemented by researchers (n = 22);
however, intervention fidelity was not reported for all stud-
ies. Intervention fidelity was reported for at least 20% of
sessions in 19 studies and averaged at least 80% in all
of these studies. The number of sessions was reported in 30
of the 35 studies and ranged from three to 105 sessions.
However, the number of trials per sessions was reported in
just 11 of the 35 studies; when reported, it ranged from one
to 96 trials. Given the heterogeneity of technology devices
used across this literature, we further classified the TAII
interventions into three categories based on the primary
technology device used in the study. The intervention types
included CAI (n = 23), AAC (n = 10), and VR (n = 3).
Study design and methodological features.  The 23 single-case
studies used the following methodologies: A-B-A-B (n =
6), multiple probe (n = 9), multiple baseline (n = 4), alter-
nating treatments (n = 2), and combination designs (n = 2;
multiple baseline with embedded alternating treatments and
multiple probe with embedded A-B-A-B). Of these, BC-
SMD effect sizes could be computed for 10 studies, includ-
ing two A-B-A-B designs, six multiple probe, and two
multiple baseline designs. Twelve studies used group design
methodologies. Nine of these 12 used a randomized experi-
mental design. Across the group studies, the comparison
group received either business as usual (BAU; n = 7 stud-
ies); an active, non-technology-based intervention (n = 4);
or an active, different TAII intervention (n = 1; see Table 1).
Risk of Bias
Results from the risk of bias evaluation summaries are pro-
vided in Online Appendix B; they were created using the
RevMan computer program (The Cochrane Collaboration,
2014). In general, the single-case studies were unlikely to
use random assignment (n = 21 were rated as unclear or
high in sequence generation), blinding of participants (n =
23 were rated as unclear or high in blinding of participants),
or outcome assessors (n = 22 were rated as unclear or high
in blinding of outcome assessors). However, most studies
adequately measured procedural fidelity (n = 15 were rated
as low risk of bias) and dependent variable reliability (n =
20 were rated as low). Furthermore, most studies used
appropriate participant selection processes.
The group studies were generally mixed across risk of bias
domains. Most studies did not adequately measure procedural
fidelity (n = 11 were rated as unclear or high), used blind cod-
ing for outcome assessors (n = 10 were rated as unclear or
high), or blind participants or treatment personnel (n = 12 were
rated as unclear or high). Areas of methodological strength
across the studies included issues of selective reporting (n =
12) where all studies did not appear to selectively report the
results of outcomes (n = 12), contamination bias with most
studies using procedures that would limit intervention spill-
over, and attrition bias (n = 9) with a majority of studies having
relatively low rates of participants leaving the study.
Visual Analyses
For the single-case studies, the coders (i.e., trained graduate
students) identified functional relations using visual analy-
sis (i.e., three demonstrations of consistent behavior change)
8 Remedial and Special Education 00(0)

Table 1.  Summary of Dependent and Independent Variable Characteristics of Included Studies.

Group design studies Single-case design studies

Study characteristics (n = 12) (n = 23) Total (n = 35)
Dependent variable
•• Communication 4 8 12
•• Academic skills 1 6 7
•• Engagement/task completion 1 6 7
•• Social skills 2 2
•• Emotion recognition 6 6
•• Adaptive skills 1 1
Outcome measurement
•• Standardized measures 8 1 9
•• Direct observation 5 23 28
 Accuracy 3 12 15
  Event recording 1 8 9
  Interval recording 2 2
  Duration recording 1 2 3
Intervention agent, dosage, fidelity
•• Researcher 7 15 22
•• Classroom staff 3 8 11
•• Parent 2 1 3
•• Intervention sessions (range) 3–25 5–105 3–105
•• Intervention fidelity reported 2 18 20
Technology devices
•• Mobile devices 4 15 19
•• Personal computers 7 9 16
•• Speech generating devices 1 12 13
•• Virtual reality 3 1 4
•• Internet 1 1 2
•• Shared active surface 1 1
Comparison condition (group only)
•• Business as usual 7  
•• Active, nontechnology intervention 4  
•• Active, other technology intervention 1  

for at least one participant or behavior in 16 of the 23 stud-
ies. These results were used for all summative analyses of
evidence-based practices listed in Table 2.
Meta-Analysis of Group Design
For the group design studies, the analytic sample included
52 effect size estimates from 12 unique studies, with between
two and eight effect sizes per studies. Online Appendix C
provides a forest plot of these effect size estimates. Table 3
reports the overall effect size estimates for the group design
studies, as well as the weighted average effect sizes by inter-
vention type and outcome domain. The overall weighted
average effect size across all 52 effect size estimates was d =
0.66 (95% CI = [0.41, 0.91], p < .001). Controlling for inter-
vention type, the effects were still heterogeneous, with an
estimated between-study SD of 0.51 (I2 = 61%). Differences
in effect size between types of intervention could not be sta-
tistically distinguished, F(2, 1.14) = 0.89, p = .59, due to the
small number of studies evaluating AAC and VR interven-
tions. Differences in effect sizes across the outcome domains
of emotion recognition, communication, and social skills
outcomes could not be statistically distinguished, F(2, 4.6) =
0.13, p = .88.
Of the three types of intervention, only CAI had more
than two group design studies contributing to the average
effect estimate; it is the only type of intervention for which
average effects could be statistically distinguished from
zero (95% CI = [0.39, 1.23], p = .003). For this category of
interventions, effects compared with a business-as-usual
control condition were only slightly smaller than those
compared with an active control condition and were not sta-
tistically distinguishable (95% CI for difference = [–1.27,
0.90], p = .66).
Outcome reporting bias.  Online Appendix D provides a fun-
nel plot of the effect size estimates from the group design
studies. Asymmetry in the distribution of effect size
Barton et al. 9

Table 2.  Summary of Evidence-Based Practices for Select Single-Case Studies.

AAC CAI

No. of No. of Primary outcome

Reference participantsa Primary outcome domain participantsa domain
Choi, O’Reilly, Sigafoos, and Lancioni (2010) 4 Language/communication  
Cihak, Wright, and Ayres (2010) 3 Engagement/task
completion
Kagohara et al. (2010) 1 Language/communication  
(social communication)
Kodak, Fisher, Clements, and Bouxsein (2011) 1 Academic
Lorah, Parnell, and Speight (2014) 3 Language/communication  
(tacts)
Mechling and Savidge (2011) 3 Engagement/task
completion
Mechling, Gast, and Seid (2009) 3 Engagement/task
completion
Mechling, Gast, and Cronin (2006) 2 Engagement/task
completion
Neely, Rispoli, Camargo, Davis, and Boles (2013) 2 Engagement/task
completion
Soares, Vannest, and Harrison (2009) 1 Academic
Taylor, Hughes, Richard, Hoch, and Coello (2004) 3 Social
van der Meer et al. (2013) 2 Language/communication  
Yakubova and Taber-Doughty (2013) 3 Adaptive
Participant total 10 21  

Note. Single-case studies included in this summative analysis met the following criteria: (a) visual analysts identified a functional relation, (b) raters
indicated the study had three or fewer domains at high risk of bias, and (c) the study met WWC design standards with or without reservations.
AAC = augmentative and alternative communication; CAI = computer-assisted instruction; WWC = What Works Clearinghouse.
a
This refers to the numbers of participants in studies with functional relations identified.

Table 3.  Average Effect Size Estimates by Intervention Type, Outcome Domain, and Study Design.

Group designs Single-case designs

Study Category Studies (effects) Est. (SE) 95% CI Studies (effects) Est. (SE) 95% CI
Overall 12 (52) 0.66 (0.10) [0.41, 0.91] 10 (13) 1.97 (0.48) [0.73, 3.21]
Intervention type
 AAC 2 (5) 0.67 (0.40) [−4.44, 5.79] 5 (8) 1.61 (0.43) [0.41, 2.80]
 CAI 8 (35) 0.81 (0.18) [0.39, 1.23] 5 (5) 2.41 (0.97) [−0.32, 5.15]
  Virtual reality 2 (12) 0.37 (0.16) [−1.71, 2.46] 0 (0)  
Outcome domain
  Academic skills 2 (3) 1.20 (0.39) [−3.79, 6.18] 2 (2) 1.08 (0.45) [−4.59, 6.75]
  Adaptive skills 0 (0) 1 (1)  
 Communication 4 (11) 0.66 (0.22) [−0.10, 1.42] 5 (8) 1.58 (0.42) [0.40, 2.76]
  Emotion recognition 7 (25) 0.67 (0.22) [0.09, 1.26] 0 (0)  
 Engagement 2 (3) 0.50 (1.07) [−13.09, 14.08] 1 (1)  
  Social skills 4 (10) 0.79 (0.17) [0.25, 1.33] 1 (1)  

Note. Est. = point estimate; CI = confidence interval; AAC = augmentative and alternative communication; CAI = computer-assisted instruction.

estimates indicates the possibility of outcome reporting funnel plot appears to be symmetric. The robust variant of
bias. Effect size estimates falling outside of the dashed lines Egger’s test did not indicate asymmetry, t(3.9) = 1.04, p =
of the funnel are indicative of heterogeneity, which makes .355, although this test does not have strong power. Results
interpretation of the funnel plot more ambiguous. The of trim-and-fill analysis estimated that there were no
10 Remedial and Special Education 00(0)
censored effect sizes, and thus no outcome reporting bias in
the average effect size estimates.
Meta-Analysis of Single-Case Research
Effect size estimates.  In total, BC-SMDs could be calculated
for 10 out of the 23 identified single-case studies. Online
Appendix C displays a forest plot of the BC-SMD effect
size estimates. Table 2 reports the overall average effect
size estimates for the single-case studies, as well as the
average effect sizes by intervention type and outcome
domain. Unlike the group design studies, no single-case
studies examined VR. Using robust random effects estima-
tion, the average effect size across all single-case studies
and outcomes was 1.97 (95% CI = [0.73, 3.21], p = .010).
Controlling for intervention type, the effects were heteroge-
neous, with an estimated between-study SD of 1.26 (I2 =
87%). Differences in effect size between the two interven-
tion types could not be statistically distinguished, F(1, 7.77)
= 0.58, p = .47. The distribution of effect size estimates was
concentrated on communication-related outcomes, with
few effect sizes available in other domains. For communi-
cation-related outcomes, the average effect was statistically
distinguishable from zero (95% CI = [0.40, 2.76]). Average
effect size estimates were omitted from Table 3 for domains
that included only a single study.
Two of the effect size estimates were larger than 5.00
and imprecisely estimated. Due to the possibility that these
effects are outliers, average effect sizes were recomputed
after excluding them. Excluding these studies, the average
effect size for the remaining 11 outcomes from nine studies
was 1.38 (95% CI = [0.68, 2.08], p = .005), with an esti-
mated between-study SD of 0.71 (I2 = 72%). Thus, although
average effects were still statistically distinguishable from
zero, the estimates of average magnitude and variability of
effects were both substantially influenced by the two outly-
ing effect sizes.
Although the average effect size estimate from the sin-
gle-case studies was larger than the corresponding estimate
from the group studies, the difference was not statistically
distinguishable from zero. Controlling for intervention
type, the difference in effects between the single-case and
group studies was 1.30 (95% CI = [–0.23, 2.83], p = .085);
excluding two outlying effects from single-case studies, the
difference was 0.70 (95% CI = [–0.27, 1.67], p = .130).
Outcome reporting bias. The funnel plot of the BC-SMD
effect size estimates is in Online Appendix D. The distribu-
tion of effect size estimates appears to be somewhat asym-
metrical, with smaller effects tending to be more precisely
estimated and larger effects tending to be less precise. This
visual assessment of asymmetry is consistent with robust
Egger’s regression tests, t(4.9) = 4.05, p = .010. In meta-
analysis of between-groups studies, such asymmetry can be
indicative of outcome reporting bias. Results of trim-and-
fill analysis were ambiguous. When based on the “R0” vari-
ant, it estimated that there were six censored effect sizes,
inclusion of which reduced the overall average effect size
estimate to 0.83. However, when based on the “L0” variant,
it estimated that there were no censored effect sizes, and
thus no outcome reporting bias in the average effect size
estimates.
Summary Identification of Evidence-Based
Practices
WWC results are provided in Online Appendix E. No group
design studies met WWC design standards without reserva-
tions. Table 2 reports results from the summative analyses
identifying evidence-based practices based on single-case
studies. Overall, the visual analysis results for the single-
case studies indicated that CAI is an evidence-based prac-
tice for students with ASD as there were nine studies with
21 participants across seven research groups; however, the
target outcomes varied across the studies (i.e., engagement/
task completion, academic, social, and adaptive skills). In
addition, only three of these nine CAI studies were able to
be included in the BC-SMD analyses (Cihak et al., 2010;
Taylor et al., 2004; Yakubova & Taber-Doughty, 2013).
Positive effects were noted for all three of these studies
although one was larger than five and imprecisely estimated
(Cihak et al., 2010). These findings were aligned with visual
analyses as functional relations were identified for each of
these three studies. Functional relations were also identified
for the remaining six studies not included in the BC-SMD
analyses. Three methodological areas were identified across
the CAI studies as sources of possible bias based on the risk
of bias evaluation; these included issues with selection bias,
and blinding of participants and outcome assessors. Using
visual analysis summaries, there were insufficient studies
and participants with functional relations—fewer than five
and 20, respectively. Neither AAC nor VR would be consid-
ered evidence-based practice.
Discussion
The primary purpose of this review was to apply novel meta-
analytic techniques to synthesize the single-case and group
design research on TAII for students with ASD and to iden-
tify the strengths, limitations, and external validity of this
research. On the basis of the single-case studies, we con-
cluded that CAI was an evidence-based practice. However,
the primary outcome domains varied across these studies,
which limits interpretations of the overall summary. The
other categories of TAII—AAC and VR—did not include a
sufficient research base to classify as an evidence-based
practice. The group research produced similar results with
additional support for CAI, whereas there was insufficient
Barton et al. 11
support for AAC and VR interventions. However, only two
AAC and two VR group studies met the eligibility criteria,
indicating a need for additional research with these TAII
modalities. Regarding CAI, there were 35 outcomes mea-
sures identified across the eight group studies. Results indi-
cated that these CAI group studies collectively produced
meaningful differences between the groups. As with the
single-case results, however, there are important qualifica-
tions to consider. For instance, the CAI group research con-
sisted of a range of dependent measures although the
majority of the studies reported measures of emotion recog-
nition (n = 6 of 8 studies). The research could be strength-
ened with greater consideration of methodological issues
that introduce bias.
Meta-Analytic Findings
Several findings emerged from the meta-analytic synthesis
of single-case research and group designs. First, for both
group designs and single-case research, there was marked
heterogeneity in effect sizes, with between-study SDs in
excess of 0.5. Such a high degree of heterogeneity indicates
that many unexplained factors contribute to the efficacy of
TAII, even after controlling for the type of technology. This
should serve as a caution in generalizing about the overall
effects of a given type of technology. That is, there are
likely to be effective and ineffective ways to use TAII,
which should be explored in future research.
Second, average effects from single-case research were
estimated to be larger than those from group designs
although the difference was not statistically distinguishable
from zero due to the high degree of unexplained heteroge-
neity. Still, this trend is consistent with theoretical expecta-
tions that effect sizes from single-case research may
generally be larger than those from group designs, even
when assessed on comparable scales. For example, visual
analysis of single-case research can reliably detect large
effects better than small or moderate effects. Therefore,
within the single-case research community, there is strong
publication preference for studies with larger effects
(Shadish et al., 2015; Shadish et al., 2016). This inadver-
tently limits the pool of studies for which to draw effect
sizes and might artificially inflate average effects.
Finally, outcome reporting bias analyses indicated the
possibility that the set of studies included in this synthesis
may not be fully representative of the true distribution of
effects. In particular, trim-and-fill and Egger’s regression
test results applied to the single-case research suggested
that we may be missing small studies with ambiguous or
null results; identification and inclusion of such studies
would tend to reduce the estimated effects. As noted in a
previous section, use of these outcome reporting bias analy-
ses with single-case research is exploratory, given that the
techniques were developed for group designs and may not
have a sound rationale when used with single-case research.
Nonetheless, we believe that outcome reporting biases are
an important concern if single-case research is to be used as
a basis for establishing evidence-based practices (Shadish,
Sham, & Smith, 2014; Shadish et al., 2016). There is an
outstanding need for further research and development of
better methodology in this area.
Systematic Review Findings
The results of the current review demonstrated that there is
a need for additional research on the use of AAC and VR
interventions for students with ASD. The body of available
experimental research on these technologies is currently too
small to make any definitive statements regarding their
overall potential for addressing the array of academic, com-
municative, and social outcomes that students with ASD
confront. As such, we encourage the pursuit of additional,
rigorous research and innovation to continue to expand the
portfolio of interventions available for addressing the
diverse needs of students with ASD.
Our review provided support for the use of CAI to address
the academic, communicative, and social needs of students
with ASD. CAI was the most frequently used TAII interven-
tion category in which skills were taught using mobile
devices or personal computers. In fact, TAII was generally
used to teach socially significant skills. Researchers across a
majority of the studies used TAII to teach functionally rele-
vant skills and were likely to provide a rationale for their
targets. Not surprisingly, the most frequently reported target
behaviors were communication, academic, and social skills.
Although communication and social skills might be particu-
larly important for students with ASD, given that their symp-
tomatology specifies known deficits in these areas (American
Psychiatric Association, 2013), additional research is needed
examining the use of TAII to teach other functional skills.
For example, several researchers in the current review effec-
tively used TAII to teach emotion recognition (Golan et al.,
2006; Hopkins et al., 2011). Smith and colleagues used a VR
software program to teach job interviewing skills. These
studies provide examples of the innovative uses of TAII
across the broad spectrum of functional skills for students
with ASD. Future research should continue to examine inno-
vations in the devices used and the target behaviors taught.
The current research on TAII could also be strengthened
through improved reporting on sample characteristics.
Although all studies in this review included students with
ASD, comparisons across studies are limited because
researchers in only five of the 35 studies independently con-
firmed ASD diagnoses. Also, nearly 11% of the sample were
not reported to have ASD. Race and ethnicity also were
rarely reported. Perhaps most surprisingly, researchers in
only two of the 35 studies reported participants’ prestudy
skill levels with technology. For example, it might be critical
12 Remedial and Special Education 00(0)
to identify the specific motor or attending skills required to
use specific TAII devices (Kagohara et al., 2013).
Researchers also might advance the science of TAII by
minimizing risk of bias in future studies. Overall, sequence
generation and blinding of participants and outcome assessors
were the domains with most studies with high risk of bias rat-
ings. In single-case studies, sequence generation refers to the
condition ordering, which when controlled by the researcher
can introduce bias. Blinding of participants and key personnel
occurs when the participants and the researchers making deci-
sions about condition changes are blind to study condition
changes. Although the dynamic nature of single-case research
allows for response guided decisions using baseline logic
(Barton et al., 2016), there are acceptable randomization pro-
cedures that can be introduced within specific single-case
designs (i.e., alternating treatments designs, multielement
designs) that might minimize bias due to the ordering and
iteration of conditions. Likewise, although there are generally
accepted standards for the quality of data collection and the
amount and degree of reliability data in single-case research
(Ayres & Ledford, 2014), standardized procedures under
which data should be collected are not ubiquitous. To mini-
mize threats to internal validity, outcome assessors should be
blind to study outcomes and conditions for both group and
single-case research and reliability data should be formatively
analyzed to monitor the presence of systematic bias (Chazin,
Ledford, Barton, & Osbourne, 2017).
The present review identified two challenges for the con-
tinued development of TAII for students with ASD. The
first is that there remains considerable variation in the defi-
nitions of TAII and the specific categories used across the
literature (cf. Grynszpan et al., 2014; Kagohara et al., 2013;
Odom et al., 2015). Although some variation might be
expected, major definitional differences might indicate a
need for developing a consistent framework or categories of
TAII to support the synthesis of results across reviews. For
example, Grynszpan and colleagues (2014) identified 21
articles for their review on TAII with students with ASD,
yet only five of those 21 were included in the current review.
Part of the reason for the lack of overlap can be traced back
to the definition of technology. Specifically, Grynszpan and
colleagues focused their review primarily on interventions
that were delivered through a computerized system, whereas
the review by Odom and colleagues—and the present
review—adopted a broader definition of technology. The
second challenge confronting the TAII literature is the need
for the use of more rigorous research designs and methods.
This issue is exemplified again in the overlap of included
studies across Grynszpan and colleagues and Odom and
colleagues. Specifically, there were a number of studies
excluded in the present review because they did not use an
eligible single-case or group-based experimental design.
These experimental methods are essential for drawing valid
conclusions regarding overall effectiveness.
Limitations
There are at least three limitations to note. First, we started
with an existing review (i.e., Wong et al., 2015) and only
searched for articles after the end date of their review. Thus,
we might have inadvertently missed studies. Furthermore,
several studies have been published since the date of our
search that would have met our inclusion criteria. Second,
because our review examined a specific type of indepen-
dent variable, the range of dependent variables included
across studies limited interpretations of the results. Even
similar types of TAII might have been applied in using dis-
parate instructional techniques. Furthermore, given students
with ASD are known to be heterogeneous, the user or par-
ticipant characteristics might have varied across studies to a
degree that makes syntheses difficult to interpret. Third, the
BC-SMD effect size measure used to meta-analyze data
from the single-case designs could be applied only to a sub-
set of the studies (10 out of 23) that otherwise met inclusion
criteria. To the extent that the studies meeting technical
inclusion criteria systematically differ from studies that
include fewer cases or use other designs, the meta-analytic
findings from included studies might not generalize. Future
syntheses should consider using multiple effect sizes,
including both between-case and within-case measures, to
characterize the magnitude of the effects of TAII.
Conclusion
A primary purpose of this review was to apply new methods
for calculating between-case effect sizes for single-case
data and comparing results with effect sizes from group
design research. There is a history of disagreement in the
field regarding appropriate methods for calculating effect
sizes for single-case research (Shadish et al., 2015; Wolery,
Busick, Reichow, & Barton, 2010). While we present a sin-
gle method, we do not believe this method is uniformly
superior to other methods or the most valid method of cal-
culating effect sizes for all single-case data. Rather, the
results of this systematic review highlighted many limita-
tions that still must be addressed through careful study.
Systematic comparisons of novel methods will be needed.
However, our initial demonstration of the utility of between-
case effect sizes for single-case research and the potential
for use alongside more conventional and accepted methods
for calculating between-group effect sizes is noteworthy
and warrants additional attention.
A secondary purpose of the review was to summarize the
TAII research. The results of our review support the use of
TAII for students with ASD. This is consistent with the
findings from other reviews although reviews varied in
their scopes and foci (cf. Grynszpan et al., 2014; Knight
et al., 2013). Overall, TAII was used to improve student out-
comes using a variety of formats, devices, and software
within authentic educational settings. Furthermore, the
Barton et al. 13
research in this area is rapidly growing as an increased
number of studies were identified each subsequent year.
Given the heterogeneity of the outcomes and the TAII tools,
researchers should continue to examine disaggregated com-
ponents or characteristics of TAII that are effective for spe-
cific outcomes.
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 authors received no financial support for the research, author-
ship, and/or publication of this article.
Supplemental Material
The online appendices are available at http://journals.sagepub.
com/doi/suppl/10.1177/0741932517729508.
References
References marked with an asterisk indicate studies included in
the meta-analyses.
*Achamadi, D., Kagohara, D. M., van der Meer, L., O’Reilly,
M., Lancioni, G., Sutherland, D., . . . Sigafoos, J. (2012).
Teaching advanced operation of an iPod-based speech-gener-
ating device to two students with Autism Spectrum Disorders.
Research in Autism Spectrum Disorders, 6, 1258–1264.
American Psychiatric Association. (2013). Diagnostic and sta-
tistical manual of mental disorders (5th ed.). Arlington, VA:
American Psychiatric Publishing.
Ayres, K., & Ledford, J. R. (2014). Dependent measures and
measurement systems. In D. L. Gast & J. R. Ledford (Eds.),
Single case research design in behavioral sciences (2nd ed.,
pp. 124–153). New York, NY: Routledge.
Barton, E. E., Ledford, J. R., Lane, J. D., Decker, J., Germansky, S.
E., Hemmeter, M. L., & Kaiser, A. (2016). The iterative use of
single case research designs to advance the science of EI/ECSE.
Topics in Early Childhood Special Education, 36, 4–14.
*Beaumont, R., & Sofronoff, K. (2008). A multi-component
social skills intervention for children with Asperger syn-
drome: The junior detective training program. Journal of
Child Psychology and Psychiatry, 49, 743–753.
*Boesch, M. C., Wendt, O., Subramanian, A., & Hsu, N.
(2013). Comparative efficacy of the Picture Exchange
Communication System (PECS) versus a speech-generating
device: Effects on social-communicative skills and speech
development. Augmentative and Alternative Communication,
29, 197–209.
Chazin, K. T., Ledford, J. R., Barton, E. E., & Osbourne, K.
(2017). Antecedent exercise for young children: An examina-
tion of bias on experimental outcomes. Manuscript Submitted
for Publication.
*Choi, H., O’Reilly, M., Sigafoos, J., & Lancioni, G. (2010).
Teaching requesting and rejecting sequences to four chil-
dren with developmental disabilities using augmentative
and alternative communication. Research in Developmental
Disabilities: A Multidisciplinary Journal, 31, 560–567.
*Cihak, D. F., Wright, R., & Ayres, K. M. (2010). Use of self-
modeling static-picture prompts via a handheld computer to
facilitate self-monitoring in the general education classroom.
Education and Training in Developmental Disabilities, 45,
136–149.
The Cochrane Collaboration. (2014). Review Manager (RevMan)
(Version 5.3) [Computer program]. Copenhagen, Denmark:
The Nordic Cochrane Centre, The Cochrane Collaboration.
Council for Exceptional Children. (2014). Council for Exceptional
Children standards for evidence-based practices in special
education. Retrieved from https://www.cec.sped.org/~/media/
Images/Standards/CEC%20EBP%20Standards%20cover/
CECs%20Evidence%20Based%20Practice%20Standards.pdf
*DeThorne, L., Betancourt, M. A., Karahalios, K., Halle, J., &
Bogue, E. (2015). Visualizing syllables: Real-time computer-
ized feedback within a speech-language intervention. Journal
of Autism and Developmental Disorders, 45, 3756–3763.
Drevon, D., Fursa, S. R., & Malcolm, A. L. (2017). Intercoder reli-
ability and validity of WebPlotDigitizer in extracting graphed
data. Behavior Modification, 41, 323–339.
Duval, S., & Tweedie, R. (2000). A nonparametric “trim and fill”
method of accounting for publication bias in meta-analysis.
Journal of the American Statistical Association, 95, 89–98.
Egger, M., Smith, G., Schneider, M., & Minder, C. (1997). Bias
in meta-analysis detected by a simple, graphical test. BMJ:
British Medical Journal, 315, 629–634.
*Faja, S., Aylward, E., Bernier, R., & Dawson, G. (2007).
Becoming a face expert: A computerized face-training pro-
gram for high-functioning individuals with Autism Spectrum
Disorders. Developmental Neuropsychology, 33, 1–24.
Fisher, Z., & Tipton, E. (2015). robumeta: An R-package for robust
variance estimation in meta-analysis. (arXiv:1503.02220).
Retrieved from https://www.rdocumentation.org/packages/
robumeta
*Flores, M., Musgrove, K., Renner, S., Hinton, V., Strozier, S.,
Franklin, S., & Hil, D. (2012). A comparison of communi-
cation using the Apple iPad and a picture-based system.
Augmentative and Alternative Communication, 28, 74–84.
Ganz, J. B., Earles-Vollrath, T. L., Heath, A. K., Parker, R. I.,
Rispoli, M. J., & Duran, J. B. (2012). A meta-analysis of
single-case research studies on aided augmentative and alter-
native communication systems with individuals with Autism
Spectrum Disorders. Journal of Autism and Developmental
Disorders, 42, 60–74.
Gast, D. L., & Spriggs, A. (2014). Visual analysis of graphic data.
In D. Gast & J. R. Ledford (Eds.), Single case research meth-
odology: Applications in special education and behavioral
sciences (pp. 234–375). New York, NY: Routledge.
*Golan, O., Ashwin, E., Granader, Y., McClintock, S., Day, K.,
Leggett, V., & Baron-Cohen, S. (2010). Enhancing emotion
recognition in children with autism spectrum conditions: An
intervention using animated vehicles with real emotional faces.
Journal of Autism and Developmental Disorders, 40, 269–279.
*Golan, O., & Baron-Cohen, S. (2006). Systemizing empathy:
Teaching adults with Asperger syndrome or high-functioning
autism to recognize complex emotions using interactive mul-
timedia. Development and Psychopathology, 18, 591–617.
14 Remedial and Special Education 00(0)
Grynszpan, O., Weiss, P. L., Perez-Diaz, F., & Gal, E. (2014).
Innovative technology-based interventions for autism spec-
trum disorders: A meta-analysis. Autism, 18, 346–361.
Hedges, L. V. (1981). Distribution theory for Glass’s estimator
of effect size and related estimators. Journal of Educational
Statistics, 6, 107–128.
Hedges, L. V., Pustejovsky, J. E., & Shadish, W. R. (2012). A
standardized mean difference effect size for single case
designs. Research Synthesis Methods, 3, 224–239.
Hedges, L. V., Pustejovsky, J. E., & Shadish, W. R. (2013). A
standardized mean difference effect size for multiple baseline
designs across individuals. Research Synthesis Methods, 4,
324–341.
Hedges, L. V., Tipton, E., & Johnson, M. C. (2010). Robust vari-
ance estimation in meta-regression with dependent effect size
estimates. Research Synthesis Methods, 1, 39–65.
Higgins, J. P. T., Altman, D. G., & Sterne, J. A. C. (2008).
Assessing risk of bias in included studies. In J. P. T. Higgins
& S. Green (Eds.), Cochrane handbook for systematic reviews
of interventions (pp. 187–241). Chichester, UK: John Wiley.
*Hopkins, I. M., Gower, M. W., Perez, T. A., Smith, D. S.,
Amthor, F. R., Wimsatt, F. C., & Biasini, F. J. (2011). Avatar
assistant: Improving social skills in students with an ASD
through a computer-based intervention. Journal of Autism
and Developmental Disorders, 41, 1543–1555.
Horner, R. H., Carr, E. G., Halle, J., McGee, G., Odom, S., &
Wolery, M. (2005). The use of single-subject research to iden-
tify evidence-based practice in special education. Exceptional
Children, 71, 165–179.
Irish, J. E. (2013). Can I sit here? A review of the literature sup-
porting the use of single-user virtual environments to help
adolescents with autism learn appropriate social communica-
tion skills. Computers in Human Behavior, 29, A17–A24.
Jones, P., Wilcox, C., & Simons, J. (2016). Evidence-based
instruction for students with Autism Spectrum Disorder:
TeachTown Basics. In T. A. Cardon (Ed.), Technology and
the treatment of children with Autism Spectrum Disorder (pp.
113–130). New York, NY: Springer.
*Kagohara, D. M., van der Meer, L., Achmadi, D., Green, V. A.,
O’Reilly, M. F., Mulloy, A., . . . Sigafoos, J. (2010). Behavioral
intervention promotes successful use of an iPod-based com-
munication device by an adolescent with autism. Clinical
Case Studies, 9, 328–338. doi:10.1177/1534650110379633
Kagohara, D. M., van der Meer, L., Ramdoss, S., O’Reilly, M. F.,
Lancioni, G. E., Davis, T. N., . . . Green, V. A. (2013). Using
iPods® and iPads® in teaching programs for individuals with
developmental disabilities: A systematic review. Research in
Developmental Disabilities, 34, 147–156.
*Kasari, C., Kaiser, A., Goods, K., Nietfeld, J., Mathy, P., Landa,
R., . . . Almirall, D. (2014). Communication interventions for
minimally verbal children with autism: A sequential mul-
tiple assignment randomized trial. Journal of the American
Academy of Child & Adolescent Psychiatry, 53, 635–646.
*Ke, F., & Im, T. (2013). Virtual-reality-based social interac-
tion training for children with high-functioning autism. The
Journal of Educational Research, 106, 441–461.
*King, M. L., Takeguchi, K., Barry, S. E., Rehfeldt, R. A., Boyer,
V. E., & Mathews, T. L. (2014). Evaluation of the iPad in
the acquisition of requesting skills for children with Autism
Spectrum Disorder. Research in Autism Spectrum Disorders,
8, 1107–1120.
Knight, V., McKissick, B. R., & Saunders, A. (2013). A review
of technology-based interventions to teach academic skills
to students with autism spectrum disorder. Journal of Autism
and Developmental Disorders, 43, 2628–2648.
*Kodak, T., Fisher, W. W., Clements, A., & Bouxsein, K. J.
(2011). Effects of computer-assisted instruction on cor-
rect responding and procedural integrity during early inten-
sive behavioral intervention. Research in Autism Spectrum
Disorders, 5, 640–647.
Kratochwill, T. R., Hitchcock, J. H., Horner, R. H., Levin, J. R.,
Odom, S. L., Rindskopf, D. M., & Shadish, W. R. (2013).
Single-case intervention research design standards. Remedial
and Special Education, 34, 26–38.
Lofland, K. B. (2016). The use of technology in the treatment of
autism. In T. A. Cardon (Ed.), Technology and the treatment
of children with Autism Spectrum Disorder (pp. 27–35). New
York, NY: Springer.
*Lorah, E. R., Parnell, A., & Speight, D. R. (2014). Acquisition
of sentence frame discrimination using the iPad™ as a speech
generating device in young children with developmental dis-
abilities. Research in Autism Spectrum Disorders, 8, 1734–
1740.
*McKissick, B. R., Spooner, F., Wood, C. L., & Diegelmann, K.
M. (2013). Effects of computer-assisted explicit instruction
on map-reading skills for students with autism. Research in
Autism Spectrum Disorders, 7, 1653–1662.
*Mechling, L. C., Gast, D. L., & Cronin, B. A. (2006). The effects
of presenting high-preference items, paired with choice,
via computer-based video programming on task comple-
tion of students with autism. Focus on Autism and Other
Developmental Disabilities, 21, 7–13.
*Mechling, L. C., Gast, D. L., & Seid, N. H. (2009). Using a per-
sonal digital assistant to increase independent task comple-
tion by students with Autism Spectrum Disorder. Journal of
Autism and Developmental Disorders, 39, 1420–1434.
*Mechling, L. C., & Savidge, E. J. (2011). Using a personal digi-
tal assistant to increase completion of novel tasks and inde-
pendent transitioning by students with Autism Spectrum
Disorder. Journal of Autism and Developmental Disorders,
41, 687–704.
*Mineo, B. A., Ziegler, W., Gill, S., & Salkin, D. (2009).
Engagement with electronic screen media among students
with Autism Spectrum Disorders. Journal of Autism and
Developmental Disorders, 39, 172–187.
Moher, D., Liberati, A., Tetzlaff, J., Altman, D. G., & PRISMA
Group. (2009). Preferred reporting items for systematic
reviews and meta-analyses: the PRISMA statement. PLoS
Medicine, 6(7), e1000097.
*Moore, M., & Calvert, S. (2000). Brief report: Vocabulary acqui-
sition for children with autism: Teacher or computer instruc-
tion. Journal of Autism and Developmental Disorders, 30,
359–362.
*Myles, B. S., Ferguson, H., & Hagiwara, T. (2007). Using a
personal digital assistant to improve the recording of home-
work assignments by an adolescent with Asperger syndrome.
Focus on Autism and Other Developmental Disabilities, 22,
96–99.
Barton et al. 15
*Neely, L., Rispoli, M., Camargo, S., Davis, H., & Boles, M.
(2013). The effect of instructional use of an iPad® on chal-
lenging behavior and academic engagement for two students
with autism. Research in Autism Spectrum Disorders, 7,
509–516.
Odom, S. L., Brantlinger, E., Gersten, R., Horner, R. H.,
Thompson, B., & Harris, K. R. (2005). Research in special
education: Scientific methods and evidence-based practices.
Exceptional Children, 71, 137–148.
Odom, S. L., Thompson, J. L., Hedges, S., Boyd, B. A., Dykstra,
J. R., Duda, M. A., . . . Bord, A. (2015). Technology-aided
interventions and instruction for adolescents with Autism
Spectrum Disorder. Journal of Autism and Developmental
Disorders, 45, 3805–3819.
Pennington, R. C. (2010). Computer-assisted instruction for
teaching academic skills to students with Autism Spectrum
Disorders: A review of literature. Focus on Autism and Other
Developmental Disabilities, 25, 239–248.
*Pennington, R. C., Stenhoff, D. M., Gibson, J., & Ballou, K.
(2012). Using simultaneous prompting to teach computer-
based story writing to a student with autism. Education and
Treatment of Children, 35, 389–406.
Ploog, B. O., Scharf, A., Nelson, D., & Brooks, P. J. (2013).
Use of computer-assisted technologies (CAT) to enhance
social, communicative, and language development in chil-
dren with autism spectrum disorders. Journal of Autism and
Developmental Disorders, 43, 301–322.
Pustejovsky, J. E. (2016). scdhlm: Estimating hierarchical linear
models for single-case designs. Retrieved from https://cran.r-
project.org/web/packages/scdhlm
Pustejovsky, J. E., & Ferron, J. M. (2017). Research synthesis and
meta-analysis of single-case designs. In J. M. Kaufmann, D.
P. Hallahan & P. C. Pullen (Eds.), Handbook of special edu-
cation (2nd ed., pp. 168–186). New York, NY: Routledge.
Pustejovsky, J. E., Hedges, L. V., & Shadish, W. R. (2014).
Design-comparable effect sizes in multiple baseline designs:
A general modeling framework. Journal of Educational and
Behavioral Statistics, 39, 368–393.
Ramdoss, S., Lang, R., Mulloy, A., Franco, J., O’Reilly, M.,
Didden, R., & Lancioni, G. (2011). Use of computer-based
interventions to teach communication skills to children with
Autism Spectrum Disorders: A systematic review. Journal of
Behavioral Education, 20, 55–76.
Ramdoss, S., Machalicek, W., Rispoli, M., Mulloy, A., Lang,
R., & O’Reilly, M. (2012). Computer-based interventions to
improve social and emotional skills in individuals with autism
spectrum disorders: A systematic review. Developmental
Neurorehabilitation, 15, 119–135.
Reeves, B. C., Higgins, J., Ramsay, C., Shea, B., Tugwell, P., &
Wells, G. A. (2013). An introduction to methodological issues
when including non-randomised studies in systematic reviews
on the effects of interventions. Research Synthesis Methods,
4, 1–11.
Reeves, L. M., Umbreit, J., Ferro, J. B., & Liaupsin, C. J. (2013).
Function-based intervention to support the inclusion of stu-
dents with autism. Education and Training in Autism and
Developmental Disabilities, 48, 379–391.
Reichow, B., Barton, E. E., & Maggin, D. (2017). Risk of bias
assessment for single case designs. Unpublished manuscript,
Anita Zucker Center for Excellence in Early Childhood
Studies, University of Florida, Gainesville.
Reichow, B., Doehring, P., Cicchetti, D. V., & Volkmar, F. R.
(2011). Evidence-based practices in autism: Where we
started. In B. Reichow, P. Doehring, D. V. Cicchetti & F. R.
Volkmar (Eds.), Evidence-based practices and treatments for
children with autism (pp. 3–24). New York, NY: Springer.
*Richter, S., & Test, D. (2011). Effects of multimedia social sto-
ries on knowledge of adult outcomes and opportunities among
transition-aged youth with significant disabilities. Education
and Training in Autism and Developmental Disabilities, 46,
410–424.
Rohatgi, A. (2014). WebPlotDigitizer user manual (Version 3.4).
Retrieved from http://arohatgi.info/WebPlotDigitizer/user-
Manual.pdf
Rothstein, H. R., Sutton, A. J., & Borenstein, M. (2005).
Publication bias in meta-analysis. In H. R. Rothstein, A. J.
Sutton & M. Borenstein (Eds.), Publication bias in meta-
analysis: Prevention, assessment, and adjustments (pp. 1–7).
Chichester, UK: John Wiley.
Shadish, W. R., Hedges, L. V., Horner, R. H., & Odom, S. L.
(2015). The role of between-case effect size in conducting,
interpreting, and summarizing single-case research (NCER
2015-002). Washington, DC: National Center for Education
Research, Institute of Education Sciences, U.S. Department
of Education. Retrieved from http://ies.ed.gov/ncser/
pubs/2015002/
Shadish, W. R., Hedges, L. V., & Pustejovsky, J. E. (2014).
Analysis and meta-analysis of single-case designs with a stan-
dardized mean difference statistic: A primer and applications.
Journal of School Psychology, 52, 123–147.
Shadish, W. R., Sham, E., & Smith, T. (2014). Publication bias in
studies of an applied behavior-analytic intervention: An initial
analysis. Journal of Applied Behavior Analysis, 47, 663–678.
Shadish, W. R., Zelinsky, N. A. M., Vevea, J. L., & Kratochwill,
T. R. (2016). A survey of publication practices of single-
case design researchers when treatments have small or large
effects. Journal of Applied Behavior Analysis, 49, 1–18.
Sham, E., & Smith, T. (2014). Publication bias in studies of an
applied behavior-analytic intervention: An initial analysis.
Journal of Applied Behavior Analysis, 47, 663–678.
*Shih, C. H., Chiang, M. S., & Shih, C. T. (2015). Assisting
students with autism to cooperate with their peers to per-
form computer mouse collaborative pointing operation on a
single display simultaneously. Research in Autism Spectrum
Disorders, 10, 15–21.
*Silver, M., & Oakes, P. (2001). Evaluation of a new computer
intervention to teach people with autism or Asperger syn-
drome to recognize and predict emotions in others. Autism,
5, 299–316.
*Smith, M. J., Ginger, E. J., Wright, K., Wright, M. A., Taylor,
J. L., Humm, L. B., . . . Fleming, M. F. (2014). Virtual real-
ity job interview training in adults with Autism Spectrum
Disorder. Journal of Autism and Developmental Disorders,
44, 2450–2463.
*Soares, D. A., Vannest, K. J., & Harrison, J. (2009). Computer
aided self-monitoring to increase academic production
and reduce self-injurious behavior in a child with autism.
Behavioral Interventions, 24, 171–183.
16 Remedial and Special Education 00(0)
Sterne, J. A. C. (Ed.). (2009). Meta-analysis in Stata: An updated
collection from the Stata journal. College Station, TX: Stata
Press.
Stovold, E., Beecher, D., Foxlee, R., & Noel-Storr, A. (2014). Study
flow diagrams in Cochrane systematic review updates: an
adapted PRISMA flow diagram. Systematic Reviews, 3, 54–59.
Stromer, R., Mackay, H. A., Howell, S. R., McVay, A. A., &
Flusser, D. (1996). Teaching computer-based spelling to indi-
viduals with developmental and hearing disabilities: Transfer
of stimulus control to writing tasks. Journal of Applied
Behavior Analysis, 29, 25–42.
Tanner-Smith, E. E., Tipton, E., & Polanin, J. R. (2016). Handling
complex meta-analytic data structures using robust variance
estimates: A tutorial in R. Journal of Developmental and Life-
Course Criminology, 2, 85–112.
Tate, R. L., Perdices, M., Rosenkoetter, U., Wakim, D., Godbee, K.,
Togher, L., & McDonald, S. (2013). Revision of a method qual-
ity rating scale for single-case experimental designs and n-of-1
trials: The 15-item Risk of Bias in N-of-1 Trials (RoBiNT)
Scale. Neuropsychological Rehabilitation, 23, 619–638.
*Taylor, B. A., Hughes, C. E., Richard, E., Hoch, H., & Coello, A.
R. (2004). Teaching teenagers with autism to seek assistance
when lost. Journal of Applied Behavior Analysis, 37, 79–82.
Tipton, E. (2015). Small sample adjustments for robust variance
estimation with meta-regression. Psychological Methods, 20,
375–393.
Tipton, E., & Pustejovsky, J. E. (2015). Small-sample adjustments
for tests of moderators and model fit using robust variance
estimation in meta-regression. Journal of Educational and
Behavioral Statistics, 40, 604–634.
*van der Meer, L., Kagohara, D., Roche, L., Sutherland, D.,
Balandin, S., Green, V. A., . . . Sigafoos, J. (2013). Teaching
multi-step requesting and social communication to two
children with Autism Spectrum Disorders with three AAC
options. Augmentative and Alternative Communication, 29,
222–234.
Weng, P. L., Maeda, Y., & Bouck, E. C. (2014). Effectiveness
of cognitive skills-based computer-assisted instruction for
students with disabilities: A synthesis. Remedial and Special
Education, 35, 167–180.
*Whalen, C., Moss, D., Ilan, A. B., Vaupel, M., Fielding, P.,
Macdonald, K., . . . Symon, J. (2010). Efficacy of TeachTown:
Basics computer-assisted intervention for the intensive com-
prehensive autism program in Los Angeles Unified School
District. Autism, 14, 179–197.
What Works Clearinghouse. (2014). What Works Clearinghouse
procedures and standards handbook (Version 3.0).
Washington, DC: Institute for Education Sciences.
Williams, C., Wright, B., Callaghan, G., & Coughlan, B. (2002).
Do children with autism learn to read more readily by com-
puter assisted instruction or traditional book methods? A pilot
study. Autism, 6, 71–91.
Wolery, M., Busick, M., Reichow, B., & Barton, E. E. (2010).
Comparison of overlap methods for quantitatively synthesiz-
ing single-subject data. The Journal of Special Education,
44(1), 18–28.
Wong, C., Odom, S. L., Hume, K. A., Cox, A. W., Fettig, A.,
Kucharczyk, S., . . . Schultz, T. R. (2014). Evidence-
based practices for children, youth, and young adults with
autism spectrum disorder. Chapel Hill: Autism Evidence-
Based Practice Review Group, Frank Porter Graham Child
Development Institute, The University of North Carolina.
Wong, C., Odom, S. L., Hume, K. A., Cox, A. W., Fettig, A.,
Kucharczyk, S., . . . Schultz, T. R. (2015). Evidence-based
practices for children, youth, and young adults with autism
spectrum disorder: A comprehensive review. Journal of
Autism and Developmental Disorders, 45, 1951–1966.
*Yakubova, G., & Taber-Doughty, T. (2013). Brief report:
Learning via the electronic interactive whiteboard for two
students with autism and a student with moderate intellectual
disability. Journal of Autism and Developmental Disorders,
43, 1465–1472.