A Meta-Analysis of Writing Interventions for Students With Learning Disabilities

Abstract

In this meta-analysis, the impact of writing interventions on the quality of writing produced by students with learning disabilities (LD) was assessed. Ancestral and electronic searches were used to locate experimental, quasi-experimental, and within-subjects design studies with participants in Grades 1–12 with documented LD. The effects of writing interventions on participants’ writing quality were averaged across 43 eligible studies to calculate an average weighted mean effect size (ES). Average weighted mean ES were also calculated for six writing treatment subgroups that contained four or more studies each (i.e., strategy instruction, dictation, procedural facilitation, prewriting, goal setting, and process writing). Overall, writing interventions had a statistically significant positive impact on the writing quality of students with LD (ES = 0.74). All writing treatment subgroups also had positive average weighted ES, but only four were statistically different from zero (i.e., strategy instruction ES = 1.09, dictation ES = 0.55, goal setting ES = 0.57, and process writing ES = 0.43). In addition, treatments designed specifically to enhance writing processes (e.g., planning, revising) were only effective when instruction was provided. Implications for the types of writing treatments and the types of instruction that may be most beneficial to students with LD are discussed and directions for future research are provided.

Full text

Vol. 80, No. 4, pp. 454 –473.
© 2014 Council for Exceptional Children.
DOI: 10.1177/0014402914527238
Exceptional Children
454 Summer 2014
Students with learning disabilities
(LD) have considerable difficulty
learning to write (Graham &
Harris, 2003). Their difficulties
with writing are complex and
manifested in multiple ways. Compared to
their typically achieving peers, they spend less time
planning, less time generating coherent ideas, and
less time revising for meaning and content (Gersten
& Baker, 2001; Graham, Harris, & McKeown, in
press). Further, students with LD typically approach
writing as if it involves a single process: content gen-
eration. When given a topic to write about, they
search long-term memory for relevant information,
generate text based on what they know about the
topic and the genre of writing, and compose each
subsequent phrase or sentence in response to the one
before it. This process continues until they have
exhausted their knowledge of the topic
or met the length requirement for the assignment
(e.g., write one page); the resulting text often lacks
coherence, clarity, and purpose (Graham, 1990).
A Meta-Analysis of Writing
Interventions for Students With
Learning Disabilities
Amy GilleSpie
Southern Methodist University
Steve GrAhAm
Arizona State University
AbStrAct: In this meta-analysis, the impact of writing interventions on the quality of writing
produced by students with learning disabilities (LD) was assessed. Ancestral and electronic
searches were used to locate experimental, quasi-experimental, and within-subjects design studies
with participants in Grades 1–12 with documented LD. The effects of writing interventions on
participants’ writing quality were averaged across 43 eligible studies to calculate an average weighted
mean effect size (ES). Average weighted mean ES were also calculated for six writing treatment
subgroups that contained four or more studies each (i.e., strategy instruction, dictation, procedural
facilitation, prewriting, goal setting, and process writing). Overall, writing interventions had a
statistically significant positive impact on the writing quality of students with LD (ES = 0.74). All
writing treatment subgroups also had positive average weighted ES, but only four were statistically
different from zero (i.e., strategy instruction ES = 1.09, dictation ES = 0.55, goal setting ES = 0.57,
and process writing ES = 0.43). In addition, treatments designed specifically to enhance writing
processes (e.g., planning, revising) were only effective when instruction was provided. Implications
for the types of writing treatments and the types of instruction that may be most beneficial to students
with LD are discussed and directions for future research are provided.
527238ECXXXX10.1177/0014402914527238Exceptional ChildrenSpring 2014
research-article2014

Exceptional Children 455
In addition, students with LD often approach
revising as if it were proofreading. Most of their
revisions involve making changes in spelling,
grammar, and mechanics (Graham, MacArthur, &
Schwartz, 1995). They experience difficulty coor-
dinating the processes involved in skilled revision,
including evaluating text, making decisions about
what to change, and executing a plan for the
proposed changes (Graham, 1997).
Another challenge for students with LD
involves difficulties with handwriting, typing, and
spelling (Graham et al., in press). These transcrip-
tion problems impede their writing in several ways.
When students must focus on forming letters
properly and spelling words correctly, they have
little room left in working memory to devote to
content, meaning, and written coherence (Baker,
Gersten, & Graham, 2003). Difficulties with tran-
scription skills make it more likely students with
LD will forget ideas or writing plans being held in
working memory. Also, they may terminate the
writing process prematurely due to fatigue.
Because writing can be cognitively overwhelm-
ing, physically exhausting, and time consuming
for students with LD, they often develop negative
attitudes about writing (Graham et al., in press).
As a result, many students with LD put forth mini-
mal effort when writing and avoid writing when
possible (Baker et al., 2003; Garcia-Sanchez &
Fidalgo-Redondo, 2006).
To address the writing difficulties experienced
by students with LD, it is important to identify
instructional practices that enhance the quality of
their writing. There are many practices that are
potentially effective. For instance, students’ tran-
scription difficulties might be addressed through
dictation or by teaching them handwriting and typ-
ing skills (Portilla-Revollar, 1994). Unsophisticated
approaches to planning and revising might be ame-
liorated by explicitly teaching students with LD to
use strategies for these tasks (e.g., Garcia-Sanchez &
Fidalgo-Redondo, 2006) or by providing proce-
dural supports (e.g., prompts) that help them carry
out these processes (e.g., Graham, 1990, 1997).
The identification of effective writing practices
provides more than just practical information. It
also provides important theoretical insights into
the writing challenges faced by students with LD.
For example, if procedures designed to enhance
students’ planning or revising processes result in
improvement in writing quality, this provides
evidence of the role of such processes in the
writing difficulties experienced by students with
LD. Likewise, the presumed impact of weak tran-
scription skills on writing is further verified if
writing improves with the removal of these skills
via dictation.
One approach for identifying effective writing
practices for students with LD is to conduct a sys-
tematic review of writing interventions with these
students. To do this, we chose meta-analysis. In
contrast to a narrative review of the literature,
meta-analysis permits analysis of the magnitude,
direction, and consistency of writing intervention
effects obtained across all relevant empirical studies
(Borenstein, Hedges, Higgins, & Rothstein, 2009).
Researchers have reported positive effects of spe-
cific writing treatments for struggling writers and
students with LD in several meta-analyses (Gra-
ham, 2006: Graham & Harris, 2003; Graham,
McKeown, Kiuhara, & Harris, 2012; Graham &
Perin, 2007; Rogers & Graham, 2008). However,
none of these reviews focused specifically on stu-
dents with LD; only one provided results disaggre-
gated for students with LD (Graham & Harris,
2003), but this was only for strategy instruction.
Researchers have reported positive
effects of specific writing treatments for
struggling writers and students with
LD in several meta-analyses.
To date, only Gersten and Baker (2001) have
conducted a meta-analysis synthesizing the
research on writing interventions for students with
LD. They examined the impact of 13 true- and
quasi-experiments on the writing of students with
LD in Grades 1 to 9. Overall, writing interven-
tions had a positive impact (ES = 0.81) on stu-
dents’ writing. After a post hoc analysis of specific
features of the writing interventions, Gersten and
Baker indicated three instructional components
were essential for effective writing instruction for
students with LD: (a) explicit instruction in the
steps of the writing process, (b) explicit instruction
in text structures of various writing genres, and (c)
guided feedback from the instructor or peers dur-
ing the writing process.

456 Summer 2014
The present meta-analysis extends Gersten and
Baker’s (2001) review in four ways. First, only
studies that included a measure of writing quality
were reviewed, as this measure provides a broad
index of writing performance. Gersten and Baker
computed an ES for each study using all writing
measures assessed. This is problematic, as studies
varied considerably in the writing measures used to
compute the aggregate ES. Although measures of
writing quality vary, they provide a more consis-
tent index than aggregating all possible outcomes.
Second, our search for studies was more extensive
than Gersten and Baker’s search, as we searched
more databases; hand searched more journals;
included grey literatures, dissertations, and theses;
and conducted ancestral searches of all obtained
documents. Third, the current review incorpo-
rated 12 additional years of research beyond the
Gersten and Baker paper. As a result, we found
more than four times as many writing comparisons
as were located by Gersten and Baker. Fourth, we
tested if explicit instruction in writing processes
was related to variability in effects.
The present review was guided by two research
questions: (a) Are writing interventions, in gen-
eral, effective for students with LD? and (b) Which
specific writing interventions are effective with
students with LD? Moderator analyses were also
applied to determine if specific study-level charac-
teristics (e.g., study quality, explicit instruction)
accounted for excess variability in effects.
method
Study Eligibility CritEria
To be eligible for this review, each study had to
meet these criteria: (a) included students in Grades
1 to 12 identified as LD with appropriate support-
ing information (e.g., test scores, Individualized
Education Plans, special education placement);
(b) tested a writing intervention (this ranged from
teaching writing to activities where students car-
ried out a procedure designed to improve writing
performance); (c) assessed the quality of students’
writing (quality measures included holistic scales,
analytic scales, and norm-referenced assessments);
(d) involved a true-experiment with randomization,
a quasi-experiment with pretest data, or a within-
subjects group design (i.e., students participated in
both the treatment and comparison conditions);
(e) provided data for calculating an ES and an aver-
age weighted ES (authors were contacted to obtain
missing information for studies published in the
last 10 years); and (f) was published in English. We
did not include studies examining the effectiveness
of word processing with students with LD, as this
was examined in a recent meta-analysis by Morphy
and Graham (2012).
SEarCh S tratEgiES
We conducted electronic searches, ending in
December 2011, of Education Abstracts, ERIC,
PsycInfo, ProQuest, and Dissertation Abstracts
International. Search terms describing writing
(writing skills or writing instruction or written lan-
guage), the population (special education or strug-
gling writers or learning disabilit* or learning
disabled or writing problems), and writing out-
comes (writing quality or composition or assessment)
were combined with these keywords: revising,
editing, prewriting, peer writing, summary writing,
writer’s workshop, process writing, strategy instruc-
tion, dictation, sentence combining, genre, imagery,
rubrics, goal setting, inquiry, self-assessment, self-
monitor, self-evaluation, grammar, spelling, mechan-
ics, motivation, planning, peer collaboration, free
writing, models, 6 traits, evaluative scales, and cre-
ativity. These searches generated 26,716 items.
The first author read abstracts for all items, and
full-text documents were obtained for promising
items (confirmed by the second author).
We also searched the National Technical Infor-
mation Service (NTIS) to identify technical
reports that met eligibility criteria, and we searched
the reference lists of prior meta-analyses (e.g., Ger-
sten & Baker, 2001; Graham et al., 2012; Graham
& Perin, 2007; Hillocks, 1984; Sandmel & Gra-
ham, 2011). We searched these journals by hand:
Exceptional Children, Journal of Learning Disabili-
ties, Learning Disability Quarterly, Learning Dis-
abilities Research & Practice, Reading & Writing,
Remedial & Special Education, and The Journal of
Special Education. The reference lists of all col-
lected documents were further examined to find
additional studies. Of the 281 documents identi-
fied, 43 studies met inclusion criteria. Some stud-
ies had more than one writing intervention, so 53
different ESs were calculated.

Exceptional Children 457
Coding o f S tudy f EaturES
Each study was coded for (a) year of publication,
(b) type of publication (journal article, book chap-
ter, dissertation, thesis, conference presentation, or
technical report), (c) grade level of participants, (d)
learning environment (general education class-
room, special education classroom, resource room/
pull-out, or after school program), (e) who deliv-
ered the intervention (teacher, researcher, or peer),
(f) subject in which the treatment was delivered
(English/Language Arts, Math, Science, History),
(g) genre of writing emphasized during the inter-
vention (narrative, persuasive, or expository), and
(h) treatment duration.
Studies were further coded and scored for seven
quality indicators: (a) design (true-experiment = 1
point; quasi-experiment = .5 point; within-subject
design = zero points), (b) treatment fidelity (1
point if fidelity was .80 or greater), (c) control for
instructor effects (1 point if methods were used
to control instructor effects; e.g., instructors ran-
domly assigned to conditions, instructors taught
both conditions), (d) number of instructors (1
point if two or more instructors were assigned to
each condition), (e) study attrition (1 point if at
least 80% of students completed the study), (f)
equal attrition across conditions (1 point if there
was no more than 10% difference in attrition
between groups), and (g) reliability of writing qual-
ity measure (1 point if reliability was .80 or higher).
Two quality indicators, control for instructor
effects and more than one instructor per condi-
tion, were not scored when writing was not directly
taught (e.g., studies involving dictation or studies
where students were asked to use a graphic orga-
nizer). Likewise, the equal attrition indicator was
not applicable to within-subjects group designs. As
a result, we calculated a total quality score for each
study as points earned divided by points possible,
multiplied by 100%.
The first author coded all 53 writing compari-
sons. Thirty percent of the studies were randomly
chosen and coded by the second author. Interrater
reliability was 99%.
CatEgorizing S tudiES b y t rEatmEnt
ConditionS
After reading the collected papers, the first author
placed all studies into the following 15 categories:
(a) strategy instruction, which involved modeling
how to use specific strategies for planning, writ-
ing, revising, and/or editing text and incorporated
student practice of the strategies in at least two
sessions with the goal of independent use over
time; (b) process writing, which consisted of stu-
dents engaging in cycles of planning, drafting,
revising, editing, and publishing their writing,
sustained time for writing for authentic purposes
and authentic audiences, and instruction con-
ducted in mini-lessons to target students’ writing
needs as they arose; (c) prewriting, which involved
students participating in activities such as brain-
storming or using a graphic organizer to help gen-
erate and organize ideas for their writing; (d)
procedural facilitation, which included supports
for students such as verbal prompts or cue cards
that facilitated planning, writing, or revising com-
positions; (e) goal setting, which involved provid-
ing students with a goal for their writing (e.g.,
include elements of a persuasive essay) or students
choosing their own goals for writing; (f) peer
tutoring, which consisted of peers instructing
each other and assisting each other with writing
tasks; (g) creativity training, which included activ-
ities designed to foster creative thinking abilities;
(h) the addition of self-regulation procedures to
strategy or skill instruction; (i) dictation to a
scribe or into a tape recorder; (j) adding instruc-
tion to process writing (students were taught writ-
ing strategies while participating in process
writing); (k) activities to increase writing motiva-
tion; (l) instruction on how to write sentences;
(m) self-evaluation using a rubric; (n) collabora-
tive writing where students created compositions
with peers; and (o) comprehensive writing pro-
grams where multiple writing treatments were
combined (e.g., strategy instruction, process writ-
ing, and prewriting were combined).
The second author confirmed the placement of
all but four studies. These studies were discussed
and reassigned until 100% consensus was reached
on their placement.
data a nalySiS
Effect sizes (ESs). ES was calculated by subtract-
ing the control or comparison group’s mean writ-
ing quality score at posttest from the treatment
group’s mean writing quality score at posttest, and

458 Summer 2014
then dividing by the pooled standard deviation of
the two groups. When pretest scores were available
(all quasi-experiments and some true-experiments
and within-subject designs), the difference between
treatment and control groups was first adjusted by
subtracting the mean pretest score for each group
from its mean posttest score. We also calculated ESs
for maintenance and generalization scores when
possible (i.e., in 7 of 43 studies).
The following rules were applied when comput-
ing ESs for writing quality. If a holistic measure
(i.e., a single score that encompasses factors such as
ideation, organization, vocabulary, mechanics, and
voice) was reported, it was used to calculate the ES
for a study. If a study included a holistic as well as
an analytic measure (i.e., separate scores for factors
such as ideation, organization, coherence) or norm-
referenced outcome, an ES was calculated for only
the holistic measure. In five instances only an ana-
lytic measure was reported, so we computed sepa-
rate ESs for each writing attribute and averaged
them to produce a single ES. In essence, this
reduced the analytic scores to a holistic score. There
were also nine instances where only a norm-refer-
enced measure was available for computing an ES.
For some ESs, it was necessary to aggregate the
performance of two or more groups (e.g., boys and
girls) using procedures developed by Nouri and
Greenberg (Cortina & Nouri, 2000). Missing stan-
dard deviations also had to be estimated from the
statistics presented in the publication for two stud-
ies (Clippard & Nicaise, 1998; Croes, 1990). We
corrected all ESs for small sample size using proce-
dures recommended by Hedges and Olkin (1985).
Average weighted effects across studies. A ran-
dom effects model was applied when calculating
average weighted ESs, statistical significance, and
confidence intervals. We also computed two mea-
sures of heterogeneity (Q and I2) to determine if
the variability in ESs was greater than what would
be expected from sampling error alone (Lipsey &
Wilson, 2001).
To answer Research Question 1 (Are writing
interventions, in general, effective for students
with LD?), we calculated a single ES for all 43
studies in this review. If a study had multiple writ-
ing comparisons, we aggregated scores (see Cortina
& Nouri, 2000) for all writing treatment groups
and compared the aggregated score to the score for
the control condition.
To answer Research Question 2 (Which
specific writing interventions are effective?), we
calculated an average weighted ES for each writing
treatment containing four or more ESs (i.e., strat-
egy instruction, dictation, procedural facilitation,
prewriting, goal setting, and process writing). This
was consistent with other writing meta-analyses
(e.g., Graham & Perin, 2007).
Nine studies included more than one writing
treatment comparison. To avoid violating the
assumption of independence of ESs (Lipsey &
Wilson, 2001), only one ES was used per study
when computing average weighted ESs and when
conducting moderator analyses.
Outliers. Prior to computing average weighted
ESs, we examined the ESs and sample sizes of all
studies for extreme outliers (any ES or sample size
falling three times the interquartile range above
the 75th percentile or three times the interquartile
range below the 25th percentile). One ES (Eissa,
2009) met this definition, and it was winsorized to
a less extreme value (i.e., an ES equivalent to the
definition of an extreme outlier above).
Moderator analysis. To determine if specific
study characteristics were related to excess variabil-
ity in ESs for Question 1 (Are writing interven-
tions, in general, effective for students with LD?),
we conducted a meta-regression (Borenstein et al.,
2009), which is similar to a multiple regression.
We examined if four moderators accounted for
excess variability in ESs. Study quality was
selected as a moderator because studies of higher
quality may produce smaller effects, as method-
ological rigor may prevent inflated outcomes or
erroneous findings. Publication date was chosen
as a moderator, because more recent publications
may produce larger effects, as researchers may have
improved writing interventions over time. Type of
publication was selected because journal articles
may have larger effects than dissertations, which
are generally conducted by less skilled researchers.
Who delivered the intervention was also chosen
as a moderator, but it was unclear if teacher- or
researcher-delivered interventions would have
a greater effect. Students may work harder for
their teachers, resulting in greater writing gains.
In contrast, teachers may be less likely to admin-
ister interventions with high fidelity due to class-
rooms demands, resulting in less effective writing
interventions.

Exceptional Children 459
For Question 2 (Which specific writing inter-
ventions are effective?), we used the ANOVA ana-
log (Hedges & Olkin, 1985) to assess excess
variability in ESs. Consistent with our decision
about the calculation of average weighted ESs, we
only conducted such an analysis when there were
at least four effects in each category of a moderator
variable. The only intervention with enough ESs
was strategy instruction. Three of the same mod-
erators used in the meta-regression (study quality,
publication year, and instructor) were used with
this analysis. Additionally, we examined differ-
ences between studies using Self-Regulated Strat-
egy Development (SRSD) and those not using
SRSD. SRSD involves explicit instruction in
task-specific writing strategies through a series of
criterion-based lessons. The instructor models the
strategies and scaffolds students’ strategy use until
students can apply the writing strategies indepen-
dently. Students also learn to self-regulate and
monitor their progress in learning and using the
writing strategies. SRSD differs from other strat-
egy approaches in its emphasis on self-regulation
and mastery learning (Harris, Graham, Mason, &
Friedlander, 2008).
Finally, we made an a priori decision to use
the ANOVA analog to examine the differences
between studies with writing process treatments
that involved instruction (n = 16) and those that
involved minimal to no instruction (n = 10). Dic-
tation and goal setting were excluded from this
analysis because they were not focused on writing
processes specifically. Studies of the process writ-
ing approach were also excluded because the type
and extent of instruction in these studies was
unclear. For the purposes of this analysis, a study
was deemed to include instruction if it had a min-
imum of two sessions and included instructor
modeling, student practice applying the writing
process or strategy, and a goal for independent
student application. Studies that did not meet
these requirements were considered to have mini-
mal or no instruction (e.g., students used cue
cards to remind them to include specific elements
in their writing).
reSultS
Table 1 presents individual descriptions for
all writing comparisons. Descriptive information
(publication type, design, grade level, writing
genre emphasized during the intervention, sum-
mary of treatment and comparison conditions,
and sample size) as well as Hedge’s g and the qual-
ity score are reported for each study. Writing treat-
ments are ordered from the treatment containing
the most studies to the treatment with the least
studies, with treatments with an equal number of
studies ordered alphabetically. Studies in a cate-
gory with fewer than four ESs are presented at the
end of the table.
Quality o f r ESEarCh
Overall, 60% of the quality indicators were met
for all studies (see Table 2). Of the studies report-
ing information about attrition, most (93%) had
little attrition; all had equal attrition across treat-
ment and comparison groups. Only 44% of stud-
ies employed a true experimental design. Another
42% were quasi-experimental and 14% were
within-subjects designs. More than two thirds of
studies (67%) reported reliability of writing qual-
ity outcome measures. Of these, 79% reported
reliability at .80 or higher. Few studies reported
treatment fidelity or controlled for instructor
effects. Less than half of studies involved multiple
instructors in treatment and comparison condi-
tions. As seen in Tables 1 and 2, quality varied
across studies and treatments.
QuEStion 1: ar E W riting
intErvEntionS, in g EnEral,
EffECtivE for S tudEntS W ith ld?
In 43 studies, researchers evaluated the effective-
ness of writing interventions on the writing quality
of students with LD. Most studies (n = 35)
involved students in upper elementary and middle
grades (i.e., Grades 4-8). The remaining studies
included students in primary grades (n = 3) and
high school (n = 5). Writing interventions were
delivered in a variety of settings, including (a)
resource room/pullout (n = 15), special education
(n = 9), general education (n = 6), and after school
programs (n = 3). A majority of studies (n = 22)
did not include information about the content
area in which a writing intervention was imple-
mented; 17 studies involved writing interventions

460 Summer 2014
tAble 1
Descriptions of Individual Writing Comparisons
Study
Type of
publication Design
Grade
level
Writing
genre
Treatment and
comparison
conditions N
Hedge’s
g ES
Quality
scorea
Writing treatments that included four or more effect sizes
Strategy instruction
Bryson & Scardamalia
(1996)
J E 10 P Inquiry strategies vs.
genre elements
15 1.22 57%
Curcic (2009) D E 7-8 EX Big 6 Skills strategy
vs. BAU
20 0.80 71%
Curry (1997) D Q 4 N Plan/write strategies
vs. writing skills
48 0.57 50%
De La Paz & Graham
(1997)
J E 5-7 P Plan/write strategies
vs. genre elements
42 0.91
1.08 M
100%
Eissa (2009) J E 9 P Planning strategy vs.
BAU
67 3.50b 43%
Englert, Raphael, Anderson,
Anthony, & Stevens
(1991)
J Q 4-5 EX Plan/write/revise
strategies vs. BAU
55 0.55
0.85 G
64%
Garcia & de Caso (2004) J E 5-6 MG Plan/write strategies
vs. BAU
127 0.96 57%
Garcia & de Caso-Fuertes
(2007)
J Q 5-6 MG Plan/write strategies
vs. BAU
100 0.71 50%
Garcia-Sanchez &
Fidalgo-Redondo (2006)
J E 5-6 EX Plan/write/revise
strategies vs.
writing skills
121 2.21 71%
MacArthur, Schwartz,
& Graham (1991)
J Q 4-6 N Revise/edit strategies
vs. BAU
29 1.42 64%
Reynolds, Hill, Swassing,
& Ward (1988)
J Q 6-8 CD Revising strategies
vs. BAU
53 0.15 50%
Sawyer, Graham, & Harris
(1992)
J E 5-6 N Plan/write strategies
vs. BAU
21 1.14 86%
Therrien, Hughes, Kapelski,
& Mokhtari (2009)
J E 7-8 EX Writing prompt
strategies vs. BAU
40 0.32 71%
Troia & Graham (2002) J E 4-5 N Planning strategy vs.
BAU
20 0.821
.73 M
–0.48 G
86%
Welch (1992) J Q 6 EX Plan/write strategies
vs. BAU
18 1.79 36%
Dictation
De La Paz & Graham
(1997)
J E 5-7 P Dictate to scribe vs.
write by hand
42 0.64
0.43 M
100%
Graham (1990) J WS 4, 6 P Dictate to recorder
vs. write by hand
23 0.75 50%
Lane & Lewandowski
(1994)
J WS 7-8 N Dictate to recorder
vs. write by hand
19 1.19 50%
MacArthur & Graham
(1987)
J WS 5-6 N Dictate to recorder
vs. write by hand
11 0.61 50%
Montague, Graves,
& Leavell (1991)
J E 7-8 N Dictate to recorder
vs. write by hand
40 –0.20 80%
Portilla-Revollar (1994) D E 2-4 N Dictate to recorder
vs. write by hand
24 0.36 80%
continues

Exceptional Children 461
Study
Type of
publication Design
Grade
level
Writing
genre
Treatment and
comparison
conditions N
Hedge’s
g ES
Quality
scorea
Procedural facilitation
Graham (1997) J WS 5-6 N Index cards to revise
vs. BAU revising
12 0.00 75%
Graham (1990) J WS 4, 6 P Prompt to write
more vs. no
prompt
23 1.15 50%
Graham et al. (1995) J E 4-6 N Procedure to revise
vs. BAU revising
43 0.02 80%
Graves, Montague, & Wong
(1990)
J E 5-6 N Cue cards for
elements vs. no
cue cards
30 1.22 60%
Montague et al. (1991) J WS 7-8 N Cue cards for
elements vs. no
cue cards
40 –0.43 50%
Page-Voth & Graham
(1999)
J E 7-8 P Procedure to
include elements
vs. no procedure
20 –0.52 100%
Prewriting
Bahr, Nelson, & Van Meter
(1996)
J WS 5, 8 N Prewriting questions
vs. no prewriting
6 0.00 50%
Blair (2003) D Q 7-8 EX Prewriting story
webs vs. no
prewriting
18 0.01 70%
Bulgren, Marquis, Lenz,
Schumaker, & Deshler
(2009)
J E 9-10,
12
EX Prewriting graphic
organizer vs.
BAU
18 1.37 80%
Kurtz (1987) D Q 4-6 N Graphic organizer
vs. no prewriting
12 0.97 50%
Sturm & Rankin-Erickson
(2002)
J WS 8 EX Concept map vs. no
prewriting
12 –0.46 33%
Goal setting
Ferretti, Lewis, & Andrews-
Weckerly (2009)
J E 4, 6 P Genre elements goal
vs. general goal
48 0.36 100%
Ferretti, MacArthur, &
Dowdy (2000)
J E 4, 6 P Genre elements goal
vs. general goal
62 0.34 60%
Graham et al. (1995) J E 4-6 N Goal to add text vs.
general goal
43 0.75 80%
Page-Voth & Graham
(1999)
J E 7-8 P Genre elements goal
vs. no goal
20 1.59 100%
Process writing
Clippard & Nicaise (1998) J Q 4-5 CD Writer’s workshop
vs. writing related
to content area
curriculum
27 0.36 64%
Croes (1990) D Q 1-5 MG Process writing vs.
BAU
157 0.34 36%
Curry (1997) D Q 4 N Writer’s workshop
vs. writing skills
45 0.45 50%
Weiss (1992) D Q 2-3 MG Process writing vs.
BAU
24 1.11 36%
tAble 1. Continued
continues

462 Summer 2014
Study
Type of
publication Design
Grade
level
Writing
genre
Treatment and
comparison
conditions N
Hedge’s
g ES
Quality
scorea
Writing treatments that did not include four or more effect sizes
Brantley & Small (1991) T Q Junior
high
CD Evaluated own
writing with rubric
vs. no rubric
37 0.23 36%
Bui, Schumaker, & Deshler
(2006)
J Q 5 N 6 Traits; genre
elements, plan/
sentence/
paragraph
strategies vs. BAU
14 0.18 50%
Curry (1997) D Q 4 N Prewrite/genre
strategies and
Writer’s workshop
vs. Writer’s
workshop
51 0.27 50%
de Caso, Garcia, Diez,
Robledo, & Alvarez
(2010)
J E 5-6 N Self-efficacy and
motivation vs.
BAU
60 1.10 43%
Fewell (1985) D Q 10-12 CD Intrinsic motivation
vs. no motivation
30 0.45 50%
Fewell (1985) D Q 10-12 CD Sentence
construction
skills vs. BAU
23 –0.79 50%
Fortner (1986) J Q 3-6 CD Creative thinking
skills vs. BAU
49 1.03 64%
Garcia and de Caso (2006) J Q 5-6 MG Self-efficacy and
motivation vs.
BAU
60 1.59
2.04 M
36%
Jolly (1988) D Q 9-11 N Writing with a
partner vs. BAU
20 0.55
–0.26 M
70%
Kurtz (1987) D Q 3-6 N Prewriting with
self-regulation
vs. prewriting
with instructor
guidance
12 1.21 50%
MacArthur, Graham,
Schwartz, & Schafer
(1995)
J Q 5 MG Writer’s workshop;
word processor;
plan/revise
strategies vs. BAU
166 0.43 36%
Sawyer et al. (1992) J E 5, 6 N Strategies with goal
setting and self
monitoring vs.
strategies
22 –0.02
–0.13 M
0.19 G
86%
Utay & Utay (1997) J E 2-6 MG Partners taught
writing skills vs.
BAU
71 0.00 43%
Note. Several studies were included in multiple categories because they had multiple treatment-control group
comparisons. J = journal; D = dissertation; E = experimental; Q = quasi-experimental; WS = within-subjects;
P = persuasive; EX = expository; N = narrative; CD = cannot determine; MG = multiple genres; BAU = business as
usual; M = maintenance; G = generalization.
aEffect size was winsorized because it was an extreme outlier.
bQuality score is a percentage of the points possible for each study.
tAble 1. Continued

Exceptional Children 463
conducted in English/Language Arts classrooms.
Writing treatments lasted from 1 day to 1 school
year, and comparison conditions varied as well
(see Table 1).
Overall, writing interventions had a statisti-
cally significant positive impact (see Table 3) on
the writing quality of students with LD (average
weighted ES = 0.74, p < .001), with 86% of
tAble 2
Percent of Quality Indicators Met for Writing Treatments That Included Four or More Effect Sizes
Writing
treatment
Quality indicators
Design Fidelity
Instructor
effects
N > 1
Instructor per
condition Attrition
Equal
attrition Reliability
All studies
(N = 43)
65%
(28/43)
19%
(8/43)
33%
(9/27)
48%
(13/27)
93%
(40/43)
100%
(37/37)
53%
(23/43)
Strategy
instruction
(n = 15)
80%
(12/15)
27%
(4/15)
40%
(6/15)
53%
(8/15)
100%
(15/15)
100%
(15/15)
47%
(7/15)
Dictation
(n = 6)
50%
(3/6)
17%
(1/6)
NA NA 100%
(6/6)
100%
(3/3)
100%
(6/6)
Procedural
facilitation
(n = 6)
50%
(3/6)
33%
(2/6)
NA NA 100%
(6/6)
100%
(3/3)
83%
(5/6)
Prewriting
(n = 5)
40%
(2/5)
0%
(0/5)
100%
(1/1)
0%
(0/1)
100%
(5/5)
100%
(3/3)
60%
(3/5)
Goal setting
(n = 4)
100%
(4/4)
50%
(2/4)
NA NA 100%
(4/4)
100%
(4/4)
75%
(3/4)
Process writing
(n = 4)
50%
(2/4)
0%
(0/4)
0%
(0/4)
75%
(3/4)
75%
(3/4)
100%
(4/4)
25%
(1/4)
Note. NA = not applicable.
tAble 3
Average Weighted Effect Sizes and Heterogeneity Statistics for Writing Treatments That Included Four or
More Effect Sizes
Writing
treatment
Number of
ESs
Average
weighted ES SE
Confidence
interval pa
Heterogeneity
QI2
All studies 43 0.74 0.10 (0.55, 0.94) <.001 170.13*** 75.30
Strategy
instruction
15 1.09 0.19 (0.72, 1.47) <.001 73.58*** 81.00
Dictation 6 0.55 0.20 (0.17, 0.94) <.01 10.00 50.00
Procedural
facilitation
6 0.24 0.31 (–0.37, 0.84) 0.45 26.13*** 80.90
Prewriting 5 0.33 0.35 (–0.35, 1.01) 0.34 10.20* 60.80
Goal setting 4 0.57 0.22 (0.14, 0.99) <.01 5.46 45.10
Process
writing
4 0.43 0.13 (0.18, 0.68) <.01 2.99 0.00
aTest of the null hypothesis that ES = zero.
*p < .05. **p < .01. ***p < .001.

464 Summer 2014
studies yielding a positive ES. The Q-statistic for
the test of heterogeneity was statistically signifi-
cant (Q = 170.13, df = 42, p < .001), and I2 indi-
cated that 75% of the observed variance between
ESs was due to true variance between studies.
Therefore, we conducted a meta-regression to
determine if study quality score, year of publica-
tion, type of publication, or who delivered the
intervention (i.e., researcher or teacher) explained
any of the excess variance.
Table 4 displays the results from the multivari-
ate regression model predicting the average
weighted ES for all studies with the four study-
level moderators. There was no evidence of multi-
collinearity between the moderator variables and
the average weighted ES. Together, the four
moderators explained 9% of the variance between
studies. The Q-residual statistic (Q = 137.91, df =
38, p < .001) indicated a statistically significant
amount of between-study variance was left unex-
plained with the moderators in the model. Thus, it
is not surprising that none of the moderators (i.e.,
study quality score, year of publication, type of
publication, and who delivered the intervention)
had statistically significant relationships with the
average weighted ES. I2 indicated that 72% of
the residual variance was due to between-study
factors.
QuEStion 2 : WhiCh S pECifiC W riting
intErvEntionS arE EffECtivE for
improving thE W riting Quality o f
StudEntS With ld?
The average weighted ESs for the six writing treat-
ments containing four or more ESs were all posi-
tive (see Table 3). Four of the writing treatments
(i.e., strategy instruction, dictation, goal setting,
and process writing) had statistically significant
effects on the writing quality of students with LD.
Strategy instruction. Fifteen studies examined
the effects of strategy instruction. Most studies
(n = 13) involved students in Grades 4 to 8, with
a majority taking place in resource room/pullout
(n = 7) or self-contained special education classes
(n = 3). In 11 studies, students learned strate-
gies for planning and writing texts. Two studies
involved strategies for revising and editing texts,
and two studies involved strategies for planning,
writing, and revising. Comparison conditions
varied, with most studies (n = 11) using a busi-
ness-as-usual comparison condition. Students
in the other study comparison conditions either
practiced writing skills (e.g., grammar, spelling)
or learned and practiced using the elements of
specific writing genres. Approximately half of the
studies involving strategy instruction were deliv-
ered by researchers, whereas strategy instruction
was delivered by teachers in six studies and by
educational psychologists in one study. Length
of strategy instruction varied from 7 days to
7 months.
Teaching writing strategies to students with LD
had a statistically significant (ES = 1.09, p < .001)
impact on writing quality (see Table 3). The Q-
statistic was also statistically significant (Q =
73.58, df = 14, p < .001), indicating variance
between ESs was greater than expected from sam-
pling error alone. I2 indicated that 81% of the
variance between ESs was due to variance between
studies. Thus, we proceeded with moderator anal-
yses to determine if study quality, year of publica-
tion, type of instructor, or the use of SRSD could
explain excess variability.
The average weighted ES for studies with qual-
ity scores of 80% and above (n = 4) was 0.73; it
was 1.05 if study quality was below 80% (n = 11).
Both ESs were statistically significant (p < .001),
but they were not statistically different from each
other (Q = 2.38, df = 1, p = .12). All of the variance
tAble 4
Multivariate Regression Model Predicting Average Weighted ES for All Studies (Question 1)
Moderator B SE t
Study quality score –0.00 0.01 –0.37
Year of publication –0.01 0.02 –0.99
Type of publication –0.27 0.23 –1.19
Who delivered the intervention –0.19 0.17 –1.28

Exceptional Children 465
in ESs was accounted for by sampling error alone
(Q = 2.89, df = 3, p = .41; I2 = 0%) in higher qual-
ity studies. A statistically significant amount of
heterogeneity between studies remained for lower
quality investigations (Q = 68.31, df = 10, p <
.001; I2 = 85%).
To determine if year of publication accounted
for excess variance, we examined differences
between studies published in the last 10 years,
from 2002 to the present (n = 7), and studies pub-
lished before 2002 (n = 8). Both average weighted
ESs were statistically significant (p < .001). The
average weighted ES for studies published from
2002 to the present (ES = 1.14) was statistically
larger (Q = 5.51, df = 1, p = .02) than the average
weighted ES for studies published before 2002
(ES = 0.76). Although variation in ESs was
accounted for by sampling error alone for studies
published before 2002 (Q = 13.30, df = 7, p = .07;
I2 = 47%), this was not the case for studies pub-
lished in the last 10 years (Q = 54.77, df = 6,
p < .001; I2 = 89%).
The average weighted ES for strategy interven-
tions delivered by teachers (n = 6) was 0.85,
whereas the average weighted ES for interventions
delivered by researchers (n = 7) was 0.63. Although
both average weighted ESs were significantly dif-
ferent from zero (p < .001), there was not a statis-
tically significant difference between them (Q =
1.46, df = 1, p = .23). The variation in the ESs was
consistent with what would be expected from
sampling error alone for studies with interven-
tions delivered by teachers (Q = 7.65, df = 5, p =
.18; I2 = 35%) and for studies with interventions
delivered by researchers (Q = 7.42, df = 6, p = .28;
I2 = 19%).
Seven of the studies involved strategy instruc-
tion using SRSD (Harris et al., 2008). The average
weighted ES for studies using SRSD was 1.33,
which was statistically larger (Q = 12.06, df = 1,
p < .01) than the average weighted ES for studies
that did not use SRSD (ES = 0.76). Both ESs were
statistically significant (p < .001). The variation in
the ESs was consistent with what would be
expected from sampling error alone for studies that
did not use SRSD (Q = 10.70, df = 7, p = .15; I2 =
35%). However, statistically significant heteroge-
neity remained between studies that used SRSD
for teaching writing strategies (Q = 50.82, df = 6,
p < .001; I2 = 88%).
Maintenance and generalization were assessed
in two studies each. Students who received strategy
instruction continued to outperform students in
comparison conditions at 2 weeks (ES = 1.08;
De La Paz & Graham, 1997) and 4 weeks (ES =
1.73; Troia & Graham, 2002) postintervention.
Generalization effects were mixed. In Englert et al.
(1991), students in the strategy condition outper-
formed comparison students (ES = 0.85) when
generalizing to a different genre, whereas in Troia
and Graham (2002), students in the strategy
condition did not outperform comparison stu-
dents (ES = –0.48) when generalizing a strategy to
a different genre.
Dictation. The impact of dictating into a tape
recorder (n = 5) or to a scribe (n = 1) was tested
in six studies. In all studies, students in compari-
son conditions composed by hand. Five studies
involved students in Grades 4 to 8. One study
involved students in Grades 2 to 4. No studies
reported data about the content area/subject mat-
ter used with dictation, but three reported that
dictation was conducted in resource room/pullout
settings. Most dictation interventions (n = 5) were
delivered by researchers and lasted between 1 and
3 days (see Table 1).
Dictation resulted in statistically significant
improvements in writing quality (average weighted
ES = 0.55, p < .01; see Table 3), and 83% of studies
produced positive effects. The Q-statistic was not
statistically significant (Q = 10.00, df = 5, p = .08),
indicating variation in the ESs was consistent with
sampling error or chance alone. Nonetheless, I2
indicated 50% of the observed variance was due to
between-study factors. One study assessed mainte-
nance effects. In this study, students in the dicta-
tion group continued to outperform students in
the comparison condition 2 weeks postinterven-
tion (ES = 0.43; De La Paz & Graham, 1997).
Procedural facilitation. Researchers evaluated
the effects of procedural facilitation in six studies
with students in Grades 4 through 8. Half of the
studies used procedures (e.g., cue cards) to prompt
students to include specific genre elements in their
written compositions. Two studies incorporated
procedures to assist students in making decisions
about revising their texts. In the remaining study,
students were given verbal prompts to add more
information to their texts. Students in comparison
conditions wrote (n = 4) or revised (n = 2) texts

466 Summer 2014
without extended procedures or prompts to assist
them while writing. Researchers delivered proce-
dural facilitation interventions in all studies and
interventions lasted 2 to 6 days. Although no stud-
ies provided information about the content areas
where the interventions were implemented, four
studies reported students participated in the inter-
ventions in resource room/pullout classrooms.
Procedural facilitation interventions had a posi-
tive impact on students’ writing quality, with an
average weighted ES of 0.24 (see Table 3), but this
was not statistically significant (p = .45). Half of
the procedural facilitation studies produced posi-
tive effects, whereas two studies produced effects
favoring the comparison treatment, and one study
produced no effect (ES = 0.00). Although there
was statistically significant variability between
studies (Q = 26.13, df = 5, p < .001) and I2 indi-
cated 81% of the observed variance was due to
between-study factors, we did not conduct follow-
up moderator analyses because there were only six
studies in this subgroup.
Prewriting. In five studies, researchers inves-
tigated the impact of prewriting activities. The
majority of prewriting studies (n = 4) involved
students completing graphic organizers (e.g., con-
cept maps, webs) to plan and organize text before
writing. In one study, students answered prewrit-
ing questions to generate and organize ideas before
composing. In all studies, comparison conditions
involved students composing texts without spe-
cific prewriting tasks; in one study, students had
the option to take notes before writing, but no
directions for how to do so were provided. Stud-
ies involved students in Grades 4 through 12. Pre-
writing interventions were delivered in English/
Language Arts (n = 2), Science (n = 1), and mul-
tiple content areas (n = 1). Researchers (n = 3) and
teachers (n = 2) delivered prewriting treatments:
(a) after school (n = 2), (b) in general education
(n = 1), (c) in resource rooms/pullout settings (n =
1), and in study hall (n = 1). Interventions lasted
between 2 and 9 days, except for one study lasting
39 days.
Four of the five prewriting studies produced
positive effects and the average weighted ES was
0.33 (see Table 3), but this effect was not statisti-
cally significant (p = 0.34). Variability between
studies was statistically significant (Q = 10.20, df =
4, p = 0.04) and I2 indicated 61% of the variance
was between studies. However, we did not conduct
moderator analyses because there were only five
studies testing the effectiveness of this writing
treatment.
Goal setting. The effectiveness of goal setting
was examined in four studies. In one study, stu-
dents selected a goal for their compositions from
a set of goals provided by the instructor. In other
studies, instructors gave students specific goals for
revising (n = 1) or for including genre elements
in their writing (n = 2). In two studies, control
students responded to writing prompts but did
not set or select writing goals. Control students in
other studies were told to make their papers bet-
ter (n = 1), or they would share their essays with
an instructor who would provide feedback (n = 1).
All studies involved students in Grades 4 through
8. Goal setting interventions were delivered by
researchers and lasted from 2 to 6 days. The con-
tent area for goal setting was not provided in any
studies. Three studies incorporated goal setting
into resource room/pullout classes and one used
goal setting in a general education setting.
Goal setting had a statistically significant (p <
.01) effect on the writing quality of students with
LD (see Table 3). All goal setting studies produced
positive effects, and the average weighted ES for
goal setting interventions was 0.57. The variability
between studies was not statistically significant
(Q = 5.46, df = 3, p = 0.41), but I2 indicated 45%
of the observed variance was due to differences
between studies within this writing treatment
subgroup.
Process writing. Process writing was examined
in four studies. All studies involved students in
Grades 1 to 5, in English/Language Arts classes.
In three studies, students in comparison condi-
tions learned writing skills through worksheets/
textbook activities. One study involved a writing
comparison group who practiced writing texts
related to themes in content area instruction.
Although both teachers and researchers delivered
process writing in one study, teachers delivered
the interventions in the remaining studies. The
duration of process writing varied from 2 months
to 10 months. Treatment place varied as well. Pro-
cess writing was carried out in general education
classrooms (n = 1), resource room/pullout class-
rooms (n = 1), and special education classrooms
(n = 1). Students in one study participated in

Exceptional Children 467
process writing in both general education and
resource/room pullout settings.
Involving students in process writing had a sta-
tistically significant (p < .01) effect on writing qual-
ity (see Table 3). All studies in this subgroup
produced positive effects, in favor of process writing
over comparison conditions. The average weighted
ES was 0.43. There was no variation in ESs (I2 =
0%) attributed to true study differences; all variance
was what would be expected from sampling error, or
chance, alone (Q = 2.99, df = 3, p = 0.39).
Instruction versus minimal to no instruction. Six-
teen studies examined the effects of writing process
treatments that involved instruction. This included
all strategy instruction studies (n = 15) and one
study (Sturm & Rankin-Erickson, 2002) involv-
ing prewriting. Ten studies examined the effects
of writing process treatments with minimal to no
instruction. These studies included four studies
involving prewriting and all procedural facilitation
studies (n = 6).
The average weighted ES for treatments with
instruction (ES = 0.93) was statistically larger (Q =
25.80, df = 1, p < .001) than the average weighted
ES for treatments with minimal/no instruction
(ES = 0.22). Although the average weighted ES for
treatments with instruction was statistically signifi-
cant (p < .001), the average weighted ES for treat-
ments with minimal to no instruction was not (p =
0.06). Both sets of studies had considerable hetero-
geneity. For studies with instruction (Q = 86.24, df
= 15, p < .001), I2 indicated 83% of observed vari-
ance between ESs was due to variance between
studies. For studies with treatments with minimal/
no instruction (Q = 33.83, df = 9, p < .001), 73%
of variance was due to between-study factors.
diScuSSion
arE W riting intErvEntionS, in gEnEral,
EffECtivE for StudEntS With ld?
The writing quality of students with LD was
improved through intervention. Writing interven-
tions had a positive impact on the writing quality
of students with LD, resulting in an average
weighted ES of 0.74 across 43 studies. Thirty-eight
of the studies produced positive effects favoring
the writing treatment. These effects were found
across all grades, with most involving students in
Grades 4 to 8. Although they reviewed fewer stud-
ies with less variety in participant and treatment
characteristics, Gersten and Baker (2001) reported
similar findings, with an overall ES of 0.81 for 13
writing intervention studies conducted with stu-
dents with LD in Grades 1 through 9.
Despite these promising results, the overall
findings must be tempered by the quality of stud-
ies (although study quality was not related to vari-
ability in effects). It is especially important that
future writing intervention studies are true experi-
ments that control for instructor effects, report
reliability of outcome measures, and provide treat-
ment fidelity data, as these were weaknesses in the
studies reviewed here. Unfortunately, the weak-
nesses observed in this review are common in edu-
cational research (Gersten, Baker, Smith-Johnson,
Flojo, & Hagan-Burke, 2004; Pressley & Harris,
1994). Obvious means for correcting these short-
comings are better preparation for researchers and
more funding to reduce the compromises research-
ers often make when they have limited resources.
It is especially important that future
writing intervention studies are true
experiments that control for instructor
effects, report reliability of outcome
measures, and provide treatment
fidelity data, as these were weaknesses
in the studies reviewed here.
With a corpus of just 43 studies, there is con-
siderable need for additional research. The search
we conducted was extensive, with no date restric-
tions, and yet only 43 studies were found. Only
five involved students beyond Grade 8. Similarly
low numbers of studies for older students were
found by Gersten and Baker (2001) and Graham
and Harris (2003). More research with older stu-
dents is needed.
All of the writing interventions assessed here
should be the subject of additional research, as most
involved six or fewer studies. Other writing inter-
ventions tested even less frequently should also be
the subject of future research. Emphasis should be
placed on assessing maintenance and generalization
effects too, as few studies in this review did so.

468 Summer 2014
WhiCh S pECifiC W riting
intErvEntionS arE E ffECtivE With
StudEntS With ld?
Of the six writing treatments including four or
more studies, strategy instruction, dictation, goal
setting, and process writing had positive and statis-
tically significant effects on the writing of students
with LD. The practical and theoretical implica-
tions of findings for each writing treatment are
discussed below.
Strategy instruction. Strategy instruction signif-
icantly improved the quality of writing of students
with LD. More recent strategy instructional studies
had a greater impact on writing quality than stud-
ies published before 2002, and SRSD studies pro-
duced greater effects than studies that did not use
SRSD. The practical implications of these findings
are that teaching students with LD to plan, write,
and revise using strategy instruction is an effective
method for improving their writing. These effects
are most pronounced if strategies are taught via
SRSD. This approach emphasizes criterion-based
instruction and teaches students the background
strategies and self-regulation procedures needed to
use the target strategies effectively (Harris et al.,
2008). These findings are similar to those obtained
in other meta-analyses examining strategy instruc-
tion in studies conducted with a broad range of
students (Graham, 2006; Graham & Harris, 2003;
Graham et al., in press).
Most of the positive effects for strategy instruc-
tion were for students in Grades 4 to 8, but two
studies also produced positive effects for high
school students (Bryson & Scardamalia, 1996;
Eissa, 2009), so this method appears promising for
older students too. The effects of strategy instruc-
tion also support Graham’s (1997) theoretical con-
tention that the writing difficulties experienced by
students with LD are due to strategic difficulties
with planning, revising, and editing. When they
are taught strategies for carrying out these pro-
cesses, students with LD show considerable
improvement in the quality of their writing.
Although strategy instruction treatments were
better represented in the literature than any other
writing treatments, additional research is still
needed. Few studies assessed strategy instruction
for revising and editing. In addition, strategy
instruction studies need to be conducted with a
wider grade range of students with LD, with a vari-
ety of writing genres, and with different writing
strategies.
Dictation. Students with LD who dictated
their compositions into a tape recorder or to a
scribe showed greater writing improvements than
students who composed by hand. In terms of class-
room practice, short dictation interventions (1 to
3 days) were effective in improving the writing
of students with LD in elementary and middle
grades. It appears dictation into a tape recorder
provides a relatively low cost option for assisting
students. However, it must be noted that in each
study reviewed here adults transcribed oral com-
positions, reducing the practicality of dictation in
the classroom.
Theoretically, the positive impact of dictation
interventions supports the contention that prob-
lems with transcription skills (e.g., handwriting,
spelling) contribute to the writing difficulties of
students with LD (Graham & Harris, 2003).
When the demands of text transcription were
removed via dictation, students with LD produced
texts of higher quality, presumably because inter-
ference from mechanical concerns was lessened
(Baker et al., 2003).
Future studies should test dictation beyond
Grade 8. High school students with LD are less
likely to receive instruction in handwriting and
spelling; thus, transcription difficulties that still
exist are likely to impact their writing. Although
the practicality of using dictation procedures
remains an open question, high school students
may be ideal candidates for dictation interven-
tions, including speech to text synthesis.
Goal setting. Setting goals for writing was also
an effective intervention for improving the writ-
ing of students with LD. Similar to the findings
reported by Graham and Perin (2007), relatively
short goal setting interventions, involving goals
for revising or including specific genre elements,
were effective for students in upper elementary
and middle grades. Only four studies of goal set-
ting interventions with students with LD were
located, so more research testing this interven-
tion is needed. Despite this limitation, providing
students with goals for their writing is a relatively
easy process, which can be translated into class-
room practice with minimal preparation or use of
instructional time.

Exceptional Children 469
Theoretically, the positive impact of goal set-
ting in the studies reviewed here suggests students
with LD possess greater capabilities for carrying
out writing processes than they apply spontane-
ously. The writing and revising behaviors of these
students were positively changed by directing
their attention to what needed to be done while
writing. Their failure to typically apply capabili-
ties they possess is likely a consequence of multi-
ple factors. For example, many students with LD
view writing as a process of simply generating text
(Graham & Harris, 2003), potentially limiting
the knowledge, skills, and writing processes they
apply when writing. Because students with LD
often develop negative beliefs about their writing
capabilities, they may not be motivated to exert
the effort needed to apply their capabilities fully.
Additional research is needed to examine why stu-
dents with LD fail to use relevant resources when
writing. Future studies should also test the effects
of student-created goals with students with LD, as
the studies reviewed here involved instructor-
developed goals. Such studies may help to learn
more about the types of students (e.g., grade level)
and types of goals students with LD can set and
achieve on their own.
Process writing. Process writing was effective
in improving the writing quality of students with
LD in the elementary grades. Our finding differed
from the conclusion drawn in a meta-analysis by
Sandmel and Graham (2011). They did not find
process writing to be effective with struggling
writers. The primary difference between the two
reviews was that Sandmel and Graham included
studies involving English language learners.
Theoretically, process writing may be effective
for improving the writing of students with LD
because it addresses a number of the difficulties they
experience. With process writing, students learn to
follow the stages involved in writing, addressing dif-
ficulties students with LD may have understanding
how to approach writing and what writing entails
beyond surface-level components such as text gen-
eration or spelling (Graham et al., in press). Process
writing also incorporates writing for authentic pur-
poses and audiences, which may provide incentives
for students with LD who lack motivation to write
(Baker et al., 2003). It further involves direct
instruction in writing skills as the need arises, which
may serve to bolster the weaknesses students with
LD have with specific aspects of writing (Graham,
2006; Graham & Harris, 2003).
As with the other writing treatments examined
in this review, additional research is needed.
Future studies should examine the impact of this
treatment with older students, as the studies
reviewed here only involved students in the ele-
mentary grades. The long-term impact of process
writing also needs to be assessed. Additionally,
component analysis studies need to be undertaken
to determine which instructional elements of pro-
cess writing, or combinations of instructional ele-
ments, have the greatest positive impacts on
students’ writing quality.
One caveat for implementing process writing in
the classroom is that it may require considerable
changes in how some educators teach writing (e.g.,
instruction is targeted to students’ needs as they
arise, sustained time for writing is provided). Thus,
teachers interested in implementing this approach
should be prepared for the time and effort involved
in setting up and running an effective process writ-
ing classroom.
Treatments designed to enhance a
specific writing process were only
effective when time was devoted to
teaching the writing skill or process.
Instruction versus minimal to no instruction.
Treatments designed to enhance a specific writing
process were only effective when time was devoted
to teaching the writing skill or process. Thus, sim-
ply providing students with a graphic organizer or
procedure to use while writing, without providing
explicit instruction, modeling, and guided prac-
tice, is likely insufficient for students with LD. To
learn to use these writing processes independently,
students with LD need systematic instruction,
teacher support, and scaffolding. This finding
seems obvious, given what we know about the
writing weaknesses of students with LD, as well as
theories about the processing and working mem-
ory difficulties of these students (Swanson, Harris,
& Graham, 2003). Not only does instruction and
scaffolding provide students with LD the writing
skills they lack, but it likely provides the practice
required to internalize these skills. Although other

470 Summer 2014
reviews have suggested that explicit instruction
is important to the writing success of students
with LD (Gersten & Baker, 2001), this is the first
study to test this proposition.
limitationS
Although the findings from this meta-analysis sup-
port and extend what is known about effective
writing treatments for students with LD, five fac-
tors limit interpretation of the results. One, we
tried to locate all possible studies using a variety of
search methods, but it is possible that a publication
bias exists, as studies with smaller ESs are less likely
to be publically available. Two, conclusions from
our analysis must be tempered due to variability in
comparison conditions across studies reviewed.
Three, as noted earlier, conclusions presented in
this article are limited by the quality of the studies
available. Four, a majority of studies reviewed
involved upper elementary and middle school stu-
dents, limiting the generalizability of our conclu-
sions. Five, we limited our review to studies where
writing quality was the outcome measure. Not all
studies, however, applied the same quality measure,
and a sole focus on writing quality excluded some
types of writing interventions. For example, we
located no handwriting or spelling intervention
studies where writing quality was assessed.
futurE dirECtionS
Taken together, the findings from this review and
their theoretical implications lead us to an over-
arching recommendation for future research.
Good writing involves mastery of and simultane-
ous use of multiple skills (e.g., planning, transcrib-
ing, revising) (Gersten & Baker, 2001; Graham &
Harris, 2003), and students with LD tend to
struggle with many of them. Thus, perhaps most
importantly, future research should evaluate com-
prehensive writing programs and multicomponent
interventions that involve teaching a wide range of
writing skills to students with LD.
In fact, there is some evidence that programs
targeting a range of writing skills are effective for
students with LD. In Bui, Schumaker, and Deshler
(2006), a program combining several writing strat-
egies, genre elements instruction, and the process
approach to writing had a positive impact on the
writing quality of students with LD in Grade 5.
Similarly, in MacArthur et al. (1995), fifth-grad-
ers with LD who participated in a writer’s work-
shop, composed on a word processor, and learned
planning and revising strategies outperformed
students in a control group who participated in
business-as-usual writing instruction. More
research is needed to replicate findings for these
types of multicomponent programs with other
grade levels and with other combinations of evi-
dence-based writing instruction. With more mul-
tifaceted evidenced-based approaches to writing
instruction, researchers and teachers may be able
to better meet the complex writing needs of stu-
dents with LD.
referenceS
References marked with an asterisk indicate studies
included in the meta-analysis.
*Bahr, C. M., Nelson, N. W., & Van Meter, A. M.
(1996). The effects of text-based and graphics-based
software tools on planning and organizing of sto-
ries. Journal of Learning Disabilities, 29, 355–370.
doi:10.1177/002221949602900404
Baker, S., Gersten, R., & Graham, S. (2003).
Teaching expressive writing to students with learning
disabilities: Research-based applications and exam-
ples. Journal of Learning Disabilities, 36, 109–123.
doi:10.1177/002221940303600204
*Blair, R. B. (2003). The effects of story webbing and visual
thinking software on the written language performance of
students with mild disabilities. Unpublished doctoral dis-
sertation, University of Oklahoma, Norman, OK.
Borenstein, M., Hedges, L. V., Higgins, J. P. T., &
Rothstein, H. R. (2009). Introduction to meta-analysis.
Chichester, UK: Wiley. doi:10.1002/9780470743386
*Brantley, H., & Small., D. (1991). Effects of self-
evaluation on written composition skill in learning dis-
abled children (ERIC 345404). Washington, DC: U.S.
Department of Education.
*Bryson, M., & Scardamalia, M. (1996). Fostering reflec-
tivity in the argumentative thinking of students with dif-
ferent learning histories. Reading & Writing Quarterly,
12, 351-384. doi:10.1080/1057356960120402
*Bui, Y. N., Schumaker, J. B., & Deshler, D. D. (2006).
The effects of a strategic writing program for students
with and without learning disabilities in inclusive fifth-
grade classes. Learning Disabilities Research & Practice,
21, 244–260. doi:10.1111/j.1540-5826.2006.00221.x

Exceptional Children 471
*Bulgren, J. A., Marquis, J. G., Lenz, B. K., Schumaker,
J. B., & Deshler, D. D. (2009). Effectiveness of
question exploration to enhance students’ written
expression of content knowledge and comprehen-
sion. Reading & Writing Quarterly, 25, 271–289.
doi:10.1080/10573560903120813
*Clippard, D., & Nicaise, M. (1998). Efficacy of writ-
ers’ workshop for students with significant writing deficits.
Journal of Research in Childhood Education, 13(1), 7–26.
doi:10.1080/02568549809594723
Cortina, J. M., & Nouri, H. (2000). Effect size for
ANOVA designs (Vol. 129). Thousand Oaks, CA: Sage.
*Croes, M. J. (1990). The efficacy of employing a writing
process approach for the instruction of language arts with
learning disabled elementary students. Unpublished doc-
toral dissertation, Baylor University, Waco, TX.
*Curcic, S. (2009). Students with learning disabilities
and the world wide web: Teaching information problem
solving to improve expository report writing. Unpublished
doctoral dissertation, Michigan State University, East
Lansing, MI.
*Curry, K. A. (1997). A comparison of the writing prod-
ucts of students with learning disabilities in inclusive and
resource room settings using different writing instruction
approaches. Unpublished doctoral dissertation, Florida
Atlantic University, Boca Raton, FL.
*de Caso, A. M., Garcia, J. N., Diez, C., Robledo, P., &
Alvarez, L. M. (2010). Enhancing writing self-efficacy
beliefs of students with learning disabilities improves
their writing processes and products. Electronic Journal
of Research in Educational Psychology, 8, 195–206.
Retrieved from http://repositorio.ual.es:8080/jspui/
handle/10835/797
*de la Paz, S., & Graham, S. (1997). Effects of dic-
tation and advanced planning instruction on the
composing of students with writing and learning prob-
lems. Journal of Educational Psychology, 89, 203–222.
doi:10.1037/0022-0663.89.2.203
*Eissa, M. A. (2009). The effectiveness of a program
based on self-regulated strategy development on the
writing skills of writing-disabled secondary school
students. Electronic Journal of Research in Educational
Psychology, 7, 5–24.
*Englert, C. S., Raphael, T. E., Anderson, L. M.,
Anthony, H. M., & Stevens, D. D. (1991). Making
strategies and self-talk visible: Writing instruction in
regular and special education classrooms. American
Educational Research Journal, 28, 337–372.
*Ferretti, R. P., Lewis, W. E., & Andrews-Weckerly,
S. (2009). Do goals affect the structure of students’
argumentative writing strategies? Journal of Educational
Psychology, 101, 577-589. doi:10.1037/a0014702
*Ferretti, R. P., MacArthur, C. A., & Dowdy, N. S.
(2000). The effects of an elaborated goal on the persuasive
writing of students with learning disabilities and their nor-
mally achieving peers. Journal of Educational Psychology,
92, 694–702. doi:10.1037/0022-0663.92.4.694
*Fewell, D. H. (1985). Intrinsic motivation and
learning strategies: A combined approach to educating
learning disabled adolescents. Unpublished doctoral
dissertation, University of North Carolina, Chapel
Hill, NC.
*Fortner, V. L. (1986). Generalization of creative
productive-thinking training to LD students’ written
expression. Learning Disability Quarterly, 9, 274–284.
doi:10.2307/1510381
*Garcia, J. N., & de Caso, A. M. (2004). Effects of a
motivational intervention for improving the writing of
children with learning disabilities. Learning Disability
Quarterly, 27, 141–159. doi:10.2307/1593665
*Garcia, J. N., & de Caso, A. M. (2006). Changes in
writing self-efficacy and writing products and processes
through specific training in the self-efficacy beliefs of
students with learning disabilities. Learning Disabilities:
A Contemporary Journal, 4, 1–27.
*Garcia, J. N., & de Caso-Fuertes, A. M. (2007).
Effectiveness of an improvement writing program
according to students’ reflexivity levels. The Spanish
Journal of Psychology, 10, 303–313.
*Garcia-Sanchez, J.- N., & Fidalgo-Redondo, R. (2006).
Effects of two types of self-regulatory instruction pro-
grams on students with learning disabilities in writing
products, processes, and self-efficacy. Learning Disability
Quarterly, 29, 181-211. doi:10.2307/30035506
Gersten, R., & Baker, S. (2001). Teaching expressive
writing to students with learning disabilities: A meta-
analysis. The Elementary School Journal, 101, 251–272.
doi:10.1086/499668
Gersten, R., Baker, S. K., Smith-Johnson, J., Flojo, J.
R., & Hagan-Burke, S. (2004). A tale of two decades:
Trends in support of federally funded experimental
research in special education. Exceptional Children, 70,
323–332.
*Graham, S. (1990). The role of production fac-
tors in learning disabled students’ compositions.
Journal of Educational Psychology, 82, 781–791.
doi:1037/0022-0663.82.4.781
*Graham, S. (1997). Executive control in the revis-
ing of students with learning and writing difficul-
ties. Journal of Educational Psychology, 89, 223–234.
doi:10.1037/0022-0663.89.2.223
Graham, S. (2006). Strategy instruction and the
teaching of writing. In C. MacArthur, S. Graham,

472 Summer 2014
& J. Fitzgerald (Eds.), Handbook of writing research
(pp. 187–207). New York, NY: Guilford.
Graham, S., & Harris, K. R. (2003). Students with
learning disabilities and the process of writing: A meta-
analysis of SRSD studies. In L. Swanson, K. R. Harris,
& S. Graham (Eds.), Handbook of learning disabilities
(pp. 323–344). New York, NY: Guilford.
Graham, S., Harris, K. R., & McKeown, D. (in press).
The writing of students with LD, meta-analysis of
SRSD writing intervention studies, and future direc-
tions: Redux. In L. Swanson, K. Harris, & S. Graham
(Eds.), Handbook of learning disabilities (2nd ed.). New
York, NY: Guilford.
*Graham, S., MacArthur, C., & Schwartz, S. (1995).
Effects of goal setting and procedural facilitation
on the revising behavior and writing performance
of students with writing and learning problems.
Journal of Educational Psychology, 87, 230–240.
doi:10.1037/0022-0663.87.2.230
Graham, S., McKeown, D., Kiuhara, S. A., & Harris,
K. R. (2012). A meta-analysis of writing instruction for
students in the elementary grades. Journal of Educational
Psychology, 104, 879-896. doi:10.1037/a0029185
Graham, S., & Perin, D. (2007). A meta-analysis
of writing instruction for adolescent students.
Journal of Educational Psychology, 99, 445–476.
doi:10.1037/0022-0663.99.3.445
*Graves, A., Montague, M., & Wong, Y. (1990). The
effects of procedural facilitation on the story composi-
tion of learning disabled students. Learning Disabilities
Research, 5, 88–93.
Harris, K. R., Graham, S., Mason, L. H., & Friedlander,
B. (2008). Powerful writing strategies for all students.
Baltimore: Brookes.
Hedges, L., & Olkin, I. (1985). Statistical methods for
meta-analysis. Orlando, FL: Academic Press.
Hillocks, G. (1984). What works in teaching com-
position: A meta-analysis of experimental treatment
studies. American Journal of Education, 93, 133–170.
doi:10.1086/443789
*Jolly, D. V. (1988). Effects of word processing, synthesized
word processing, and collaborative freewriting on the writ-
ten language of learning disabled students. Unpublished
doctoral dissertation, Southern Illinois University at
Edwardsville, Edwardsville, IL.
*Kurtz, H. (1987). Effects of a specific strategy training
on the written expression of learning disabled students
grades 3-6. Unpublished doctoral dissertation, Drake
University, Des Moines, IA.
*Lane, S. E., & Lewandowski, L. (1994). Oral and writ-
ten compositions of students with and without learning
disabilities. Journal of Psychoeducational Assessment, 12,
142–153. doi:10.1177/073428299401200204
Lipsey, M. W., & Wilson, D. B. (2001). Practical meta-
analysis. Thousand Oaks, CA: SAGE.
*MacArthur, C., & Graham, S. (1987). Learning dis-
abled students’ composing under three methods of
text production: Handwriting, word processing, and
dictation. The Journal of Special Education, 21, 22–42.
doi:10.1177/002246698702100304
*MacArthur, C. A., Graham, S., Schwartz, S., &
Schafer, W. D. (1995). Evaluation of a writing instruc-
tion model that integrated a process approach, strategy
instruction, and word processing. Learning Disability
Quarterly, 18, 278–291. doi:10.2307/1511234
*MacArthur, C. A., Schwartz, S., & Graham, S. (1991).
Effects of a reciprocal peer revision strategy in special
education classrooms. Learning Disabilities Research and
Practice, 6, 201–210.
*Montague, M., Graves, A., & Leavell, A. (1991).
Planning, procedural facilitation, and narrative compo-
sition of junior high students with learning disabilities.
Learning Disabilities Research and Practice, 6, 219–224.
Morphy, P., & Graham, S. (2012). Word processing
programs and weaker writers/readers: A meta-analysis
of research findings. Reading and Writing, 25, 641–678.
doi:10.1007/s11145-010-9292-5
*Page-Voth, V., & Graham, S. (1999). Effects of goal
setting and strategy use on the writing performance and
self-efficacy of students with writing and learning prob-
lems. Journal of Educational Psychology, 91, 230–240.
doi:10.1037/0022-0663.91.2.230
*Portilla-Revollar, C. (1994). Differences in story produc-
tion between students with learning disabilities and nor-
mally achieving students under two modes of production.
Unpublished doctoral dissertation, University of Texas
at Austin, Austin, TX.
Pressley, M., & Harris, K. R. (1994). Increasing the
quality of educational intervention research. Educational
Psychology Review, 6, 191–208. doi:10.1007/BF0221
3181
*Reynolds, C. J., Hill, D. S., Swassing, R. H., & Ward,
M. E. (1988). The effects of revision strategy instruction
on the writing performance of students with learning
disabilities. Journal of Learning Disabilities, 21, 540–
545. doi:10.1177/002221948802100904
Rogers, L., & Graham, S. (2008). A meta-analysis of
single subject design writing intervention research.
Journal of Educational Psychology, 100, 879–906.
doi:10.1037/0022-0663.100.4.879
Sandmel, K., & Graham, S. (2011). The process writ-
ing approach: A meta-analysis. Journal of Educational

Exceptional Children 473
Research, 104, 396-407. doi:10.1080/00220671.2010.
488703
*Sawyer, R., Graham, S., & Harris, K. (1992). Direct
teaching, strategy instruction, and strategy instruction
with explicit self-regulation: Effects on the composition
skills and self-efficacy of students with learning dis-
abilities. Journal of Educational Psychology, 84, 340–352.
doi:10.1037/0022-0663.84.3.340
*Sturm, J. M., & Rankin-Erickson, J. L. (2002). Effects
of hand-drawn and Computer generated concept
mapping on the expository writing of middle school
students with learning disabilities. Learning Disabilities
Research & Practice, 17, 124–139. doi:10.1111/
1540-5826.00039
Swanson, L., Harris, K. R., & Graham, S. (2003).
Handbook of learning disabilities. New York: Guilford.
*Therrien, W., Hughes, C., Kapelski, C., & Mokhtari,
K. (2009). Effectiveness of a test-taking strategy on
achievement in essay tests for students with learning
disabilities. Journal of Learning Disabilities, 42, 14–23.
doi:10.1177/0022219408326218
*Troia, G., & Graham, S. (2002). The effectiveness of a
highly explicit, teacher directed strategy instruction rou-
tine. Journal of Learning Disabilities, 35, 290–305. doi:1
0.1177/00222194020350040101
*Utay, C., & Utay, J. (1997). Peer-assisted learning:
The effects of cooperative learning and cross-age peer
tutoring with word processing on writing skills of stu-
dents with learning disabilities. Journal of Computing in
Childhood Education, 8, 165–185.
*Weiss, D. H. (1992). The effects of process writing
instruction on the writing and reading performance of stu-
dents with learning disabilities. Unpublished doctoral dis-
sertation, Florida International University, Miami, FL.
*Welch, M. (1992). The PLEASE strategy: A
metacognitive learning strategy for improving the
paragraph writing of students with mild learning dis-
abilities. Learning Disability Quarterly, 15, 119-128.
doi:10.2307/1511013
About the AuthorS
Amy GilleSpie, Department of Teaching and
Learning, Simmons School of Education and
Human Development, Southern Methodist Uni-
versity. Steve GrAhAm, Mary Emily Warner
Professor, Mary Lou Fulton Teachers College,
Arizona State University.
Address correspondence concerning this article
to Amy Gillespie, Department of Teaching and
Learning, Southern Methodist University, 3101
University Blvd, Ste 345 Dallas, TX 75205
(e-mail: amy.gillespie@vanderbilt.edu).
Manuscript received July 2012; accepted January
2013.

CopyrightofExceptionalChildrenisthepropertyofSagePublicationsInc.anditscontent
maynotbecopiedoremailedtomultiplesitesorpostedtoalistservwithoutthecopyright
holder'sexpresswrittenpermission.However,usersmayprint,download,oremailarticlesfor
individualuse.