23 vues

Transféré par api-400727133

- Colorectal cancer
- 2015 Strengths Business Unit Meta-Analysis
- How NOT to Write a Medical Paper
- 2016 - Beyond Born versus Made A New Look at Expertise - Hambrick y otros.pdf
- Statistics for Business and Economics: bab 7
- Antioxidants_Review of Evidence on All-cause Mortality (Chocrane, 2012)
- description: tags: techappendix01 359
- Strategies to Prevent Death by Suicide- Meta-Analysis of Randomised Controlled Trials
- Instructional Scaffolding in STEM Education
- Hippotherapy—an Intervention To
- The Clock Drawing Test
- Research on the Effectiveness of Online Learning.pdf
- Assistive technology interventions for adolescents and adults with learning disabilities: An evidence-based systematic review and meta-analysis
- 1 Predictors of Persistent Pain After Breast Cancer Surgery
- Development, Problem Behavior, And Quality of Life in Population Based Sample of Eight Year Old Children With Down Syndrome
- 12.full
- 10 Commitment Across Cultures
- EGZR Erelera ofbeko olug
- jurnal penggunaan antibiotik topikal sebelum penutupan insisi luka primer operasi
- Cassey and Mcardle

Vous êtes sur la page 1sur 21

research-article2017

ECXXXX10.1177/0014402917737467Exceptional ChildrenJitendra et al.

Article

Exceptional Children

2018, Vol. 84(2) 177–196

Mathematical Interventions for © The Author(s) 2017

DOI: 10.1177/0014402917737467

https://doi.org/10.1177/0014402917737467

Secondary Students With Learning journals.sagepub.com/home/ecx

Difficulties: A Meta-Analysis

Ahmed A. Alghamdi1, Scott B. Hefte1,

and John Mouanoutoua1

Abstract

This meta-analysis is the first to provide a quantitative synthesis of empirical evaluations of

mathematical intervention programs implemented in secondary schools for students with

learning disabilities and mathematics difficulties. Included studies used a treatment-control group

design. A total of 19 experimental and quasi-experimental studies containing 20 independent

samples met study inclusion criteria. Results of a random effects model analysis indicated that

mathematical interventions influence mathematics outcomes (g = 0.37, 95% confidence interval

[0.18, 0.56]) for students with learning disabilities and mathematics difficulties. In addition,

instructional time moderated the relation between mathematics interventions and student

learning. Limitations of the study, future directions for research, and implications for practice

are discussed.

In an increasingly competitive job market, Progress data indicated that the goal of the No

where the demand for mathematics-intensive Child Left Behind legislation to eradicate

science and engineering jobs is outpacing achievement differences among student sub-

overall job growth three to one (National groups had not yet been met (Dossey,

Mathematics Advisory Panel, 2008), the abil- McCrone, & Halvorsen, 2016). For example,

ity to think mathematically is a crucial skill. 45% of fourth graders with disabilities scored

Although the percentage of students reaching below the basic level on the 2015 National

proficient levels in mathematics has increased Assessment of Educational Progress, com-

over the past decade, it is disconcerting that, as pared with 14% without disabilities; further,

children progress from the elementary grades 68% of eighth graders with disabilities scored

into secondary grades, the mathematical gains below the basic level versus 23% without dis-

made in elementary school are, on average, not abilities. In addition, the National Longitudi-

matched in later years (National Mathematics nal Transition Study-2 (Wagner, Newman,

Advisory Panel, 2008). For example, on the

2015 National Assessment of Educational 1

University of Minnesota

Progress, only 33% of U.S. eighth-grade stu- 2

Bellarmine University

dents, compared with 40% of fourth-grade stu-

dents, were proficient or advanced in their Corresponding Author:

Asha K. Jitendra, University of Minnesota, 245 Education

knowledge of mathematics (National Center Sciences Building, 56 E. River Road, Minneapolis, MN

for Education Statistics, 2015). Further, the 55455.

2015 National Assessment of Educational E-mail: jiten001@umn.edu

178 Exceptional Children 84(2)

Cameto, Levine, & Garza, 2006) found that Several meta-analyses and narrative

secondary students with learning disabilities reviews already exist regarding the effective-

(LD) perform worse than their peers without ness of mathematics interventions for students

disabilities on mathematics subtests. with LD or MD. However, these reviews tar-

There is evidence that secondary students geted elementary students with MD (Dennis

with disabilities score significantly below et al., 2016; Kroesbergen & Van Luit, 2003)

grade level in mathematics and that their and K–12 students with LD or MD (Baker,

growth rate in mathematics slows consider- Gersten, & Lee, 2002; Gersten et al., 2009;

ably in secondary schools (Cortiella, 2011; Swanson & Hoskyn, 1998), or they focused on

Wei, Lenz, & Blackorby, 2013). The mathe- specific mathematical content instruction,

matics underachievement of these students as such as word problem solving (Xin & Jitendra,

they enter middle and high school grades may 1999; Zhang & Xin, 2012; Zheng, Flynn, &

be due to several reasons, such as increasingly Swanson, 2013), fractions (Misquitta, 2011;

difficult mathematics content as the mathe- Shin & Bryant, 2015), and algebra (Hughes,

matical ideas become more abstract (Witzel, Witzel, Riccomini, Fries, & Kanyongo, 2014;

2016) or limited access to higher-level math- Maccini, McNaughton, & Ruhl, 1999; Watt,

ematics instruction given the focus on lower- Watkins, & Abbitt, 2016). The previous meta-

level skills, including fact recall and mastery analyses for K–12 students with LD or MD

of procedures (Miller & Hudson, 2007). Many reported positive effects on test scores for stu-

secondary students with disabilities continue dents with LD (Gersten et al., 2009; Swanson

to experience difficulty with early numeracy & Hoskyn, 1998) or MD (Baker et al., 2002)

concepts, which are crucial for more advanced with regard to several components of mathe-

mathematics (National Mathematics Advisory matics instruction. Baker and colleagues

Panel, 2008). Further, difficulties in acquiring (2002) found support for peer-assisted learn-

higher-order thinking skills such as reasoning ing, explicit instruction, and providing stu-

(e.g., Hunt & Vasquez, 2014) are major dents with data and feedback. Gersten et al.

impediments and inhibit opportunities to pur- (2009) examined four components of mathe-

sue mathematics-related careers. matics instruction and found strong effects for

approaches to instruction (e.g., explicit instruc-

tion, providing visual models, heuristics), cur-

There is evidence that secondary

riculum design, and providing feedback to

students with disabilities score teachers based on formative assessment data.

significantly below grade level in Swanson and Hoskyn (1998) reported that

mathematics. cognitive strategies and direct instruction were

effective instructional models in mathematics

If the goal is to help secondary students and other domains. However, these reviews

with disabilities meet the high expectations did not examine grade level as a factor that

of, for example, the Common Core State Stan- might influence the direction and magnitude

dards (National Governors Association Center of intervention effects.

for Best Practices & Council of Chief State Previous reviews have included students

School Officers, 2010) and Every Student with LD, MD, or both. However, the authors

Succeeds Act of 2015, effective interventions did not operationalize the terms LD (Maccini

need to be identified. Much research on the et al., 1999) or MD (Baker et al., 2002;

effectiveness of mathematics interventions for Kroesbergen & Van Luit, 2003), or they used

secondary students with disabilities or those different criteria for inclusion of students with

identified as having mathematics difficulties MD. For example, students with MD were

(MD) has accumulated over the last three identified as those scoring <25th percentile on

decades to justify a meta-analysis of the effec- a mathematics test (Dennis et al., 2016; Shin &

tiveness of such interventions on students’ Bryant, 2015; Zheng et al., 2013), were con-

mathematics learning. sidered low achieving or referred by classroom

Jitendra et al. 179

teachers as being at risk for failure in mathe- lem solving and found that the efficacy of

matics (Zhang & Xin, 2012), were receiving intervention approaches (heuristic, semantic,

remedial mathematics instruction (Shin & or authentic) varied and that these interven-

Bryant, 2015; Xin & Jitendra, 1999), or were tions differed with regard to real-world con-

classified as struggling learners (Hughes et al., nections. Marita and Hord (2017) focused on

2014; Misquitta, 2011). interventions directed to students with LD and

Another limitation of the previous reviews is found that a variety of interventions, such as

the inconsistencies in calculating effect sizes systematic instruction, problem-based learn-

that impede directly comparing effect sizes ing, and visual representations, can improve

among reviews. For example, in two meta-anal- the problem-solving skills of students with

yses (Xin & Jitendra, 1999; Zhang & Xin, 2012), LD. Finally, Myers and colleagues (2015)

the authors included several single-group pre- detailed several instructional approaches and

test-posttest design studies and measured noted the benefits of cognitive and metacogni-

changes in scores from pre- to postintervention. tive strategies and enhanced anchored instruc-

There are concerns with the use of change scores tion for secondary students with LD.

and the lack of a control group because matura- The question of whether and to what extent

tion effects cannot be detected and standardized mathematics interventions are effective for sec-

mean change effects would be inflated (Borman, ondary students with LD and MD has not been

Hewes, Overman, & Brown, 2003). considered in the previous reviews. As such,

Our main objective was to conduct a meta- there is a need to integrate the empirical research

analysis of the effects of mathematics inter- on mathematics interventions for secondary

ventions aimed at improving the mathematics students with LD and MD and answer questions

performance of secondary students with LD regarding their effectiveness. This can be

and MD. Given the difficulties that secondary addressed by a meta-analysis, as it can estimate

students with LD and MD experience with the average effectiveness of mathematics inter-

mathematics, there is a need to improve the ventions on students’ mathematical learning. In

understanding of research related to mathe- addition, a meta-analysis can explore the types

matics interventions across mathematical con- of settings or student populations for which the

tent areas for this group of students. Although intervention works best. The present review

evidence suggests that students with mathe- sought to answer the following questions:

matics LD and those with MD (i.e., scored

below average to low average on tests of math •• What is the average effect of mathe-

achievement) are characteristically different matical interventions for secondary

(Mazzocco, 2007), they often encounter simi- students with LD and MD?

lar challenges with mathematics that justify •• Does it vary by student status (LD vs.

the need for high-quality instruction (Fuchs, MD)?

Fuchs, & Compton, 2013). We found three •• What instructional characteristics (e.g.,

recent narrative reviews (Hwang & Riccomini, setting, instructional time) or method-

2016; Marita & Hord, 2017; Myers, Wang, ological characteristics (e.g., research

Brownell, & Gagnon, 2015) of mathematics design, implementer, fidelity of imple-

intervention research for secondary students mentation [FOI]) of mathematical inter-

with or at risk for LD, but we did not find a ventions moderate student outcomes?

meta-analysis to date on this topic. These nar-

rative reviews provide a summary of existing

research (group and single-case studies) by Method

taking a qualitative rather than quantitative Literature Search and Criteria for

approach. The three reviews extended the ear-

lier review of the topic by Maccini, Mulcahy,

Inclusion

and Wilson (2007). Hwang and Riccomini We conducted a literature search of mathemat-

(2016) reviewed studies of mathematical prob- ics intervention studies ending in May 2017.

180 Exceptional Children 84(2)

First, we included the online databases ERIC, be identified with a learning disability (school

PsycINFO, and ProQuest Dissertation and or researcher identified) or mathematics diffi-

Theses. We searched abstracts for key popula- culty. We operationally defined students as

tion search terms and roots (learning dis- having MD when they scored ≤35th percen-

abilit*, learning difficult*, at-risk, struggling, tile on a standardized mathematics test. This

learning problem, underperforming, below criterion was selected to ensure broad repre-

average, underachieving, disadvantaged) in sentation of students who may be at low to

conjunction with mathematics search terms high risk of developing MD (see Fuchs et al.,

and roots (math, mathematic*, intervention, 2014; Mazzocco, 2007). We included studies

instruction, treatment, training, teaching when >50% of the participants were students

method*) and grade search terms (middle with LD or MD or disaggregated data were

school, high school, secondary, adolescent*) provided for students identified as such.

in various combinations. Second, we scanned Second, the studies had to involve students

the reference sections of each study to locate with LD or MD in Grades 6 to 12. The third cri-

other possible studies, as well as the reference terion was that studies were based on a random-

sections of existing meta-analyses and review ized controlled trial (RCT) or quasi-experimental

studies on this topic. Third, we hand-searched design and that they compared students in treat-

key journals in special education from 2016 to ment groups taught via a specific mathematics

2017: Exceptional Children, Journal of Learn- intervention with those in control groups who

ing Disabilities, The Journal of Special Educa- were taught via standard methods or an alterna-

tion, Learning Disability Quarterly, Learning tive intervention. Fourth, studies could have

Disabilities Research & Practice, and Reme- been conducted in any country, but the report

dial and Special Education. had to be available in English. Fifth, to be

The primary literature search yielded 3,355 included, studies had to evaluate the effects of

abstracts for screening, and after removal of mathematics interventions or programs on learn-

duplicate abstracts, 3,175 abstracts remained. ing of mathematical subjects. The sixth criterion

Of these, 2,970 were not considered per title is based on studies providing the necessary

and abstract review. The remaining 205 stud- quantitative information to estimate effect sizes.

ies retrieved in the primary search appeared to Last, the dependent measures had to include

match the selection criteria based on their titles quantitative measures of mathematics perfor-

and abstracts. However, most of the selected mance. Studies were excluded if they examined

studies were removed after retrieval of their only measures of student attitudes toward math-

full-text versions, either because studies were ematics or strategy use.

outside the age/grade range of this meta-analy-

sis (n = 46) or did not have a suitable research Coding Procedure and Intercoder

design (no control group, correlational studies;

n = 83). Moreover, 12 reports were reviews,

Agreement

editorials, or commentaries, and 19 studies Two authors developed a coding form to

included all students but did not contain a suf- extract and record relevant information from

ficient sample of students identified as having each study. We evaluated the form multiple

LD or did not identify students as MD based times in repetitive cycles in consultation with

on standardized test scores. For 26 studies, the coders. Then we coded studies for instruc-

there were other reasons for exclusion, such as tional and methodological characteristics.

dissertations that were published and studies

that did not focus on mathematical interven-

Instructional Characteristics

tions or did not provide sufficient information

to compute effect sizes. The primary search Participants. We coded how students with LD

resulted in a total of 19 relevant studies. or MD were identified, using open-ended

We applied seven criteria for inclusion in items (LD or MD as described in text, criteria

this meta-analysis. First, participants had to for classifying students with MD). We also

Jitendra et al. 181

coded participants’ grade as middle school concepts, such as operations and algebraic

(Grades 6–8), high school (Grades 9–12), or thinking (e.g., arithmetic word problem solv-

combined. ing) and number system (e.g., fractions, deci-

mals), and (b) higher-level mathematics topics

Intervention approaches. We examined written (i.e., ratio and proportional relationships,

descriptions of interventions used in the stud- expressions and equations). The higher-level

ies. Interventions varied in their approaches mathematics category also included studies

and sometimes included a combination of that encompassed several mathematical topics

approaches; however, most studies included or a combination of foundational and higher-

explicit and systematic instruction. We coded level topics (e.g., number system, ratio and

intervention approaches as a categorical vari- proportional relationships, measurement and

able: (a) visual models only (e.g., fraction data).

circles or squares, manipulative objects, dia-

grams, mental representations); (b) visual

models combined with other strategies, such

Methodological Characteristics

as priming the underlying problem structure, We coded the type of report or publication

cognitive strategy, or contextualized instruc- source as a dichotomous variable (journal

tion (problems used real-world scenarios); (c) article, unpublished report). We coded the

computer-based modules (e.g., videos) com- year of publication as either 2000 and before

bined with visual models or contextualized or 2001 and later. We created a dichotomous

instruction; and (d) other (i.e., problem-based variable to code research design as RCT ver-

learning, peer-assisted learning strategies and sus quasi-experimental. RCTs referred to

curriculum-based instruction, question-and- studies in which students, teachers or class-

answer strategy). rooms, or schools were randomly assigned to

conditions and the level of random assign-

Intervention features. We gathered information ment determined the unit of analysis. In con-

on intervention features using three open- trast, quasi-experiments may also have used

ended items (i.e., setting, implementation of random assignment at the teacher or class-

the intervention, instructional time). We coded room and school levels, but the analysis was

instructional setting as general education done at the student level because of limited

classroom, special education classroom, or samples of classrooms or schools. We coded

combined. We coded implementation of the studies to identify who delivered the interven-

intervention as a dichotomous variable, with tion to students: researcher or school person-

instruction conducted in small groups (<10 nel. In addition, we created a dichotomous

students) or large groups (≥10 students). We variable to code FOI as either reported (yes)

coded the time of treatment implementation in or not reported (no). We coded FOI when

hours and then categorized it as ≤10 hr or >10 studies reported this information, and we

hr. Specifically, we calculated instructional recorded the percentage of FOI. We catego-

time as the hours of intervention across the rized the types of measures as being researcher

duration of the intervention (i.e., number of developed or standardized. We examined the

sessions multiplied by the length of each type of control group as a proxy variable for

session). the strength of the control condition or the

counterfactual with regard to its effect on

Mathematical topic. We coded the mathemati- observed treatment effects. We coded this

cal topic of instruction as a categorical variable variable as either business as usual (BAU; tra-

with the following categories, based on the ditional textbook method) or an alternative

Common Core State Standards for mathemat- intervention.

ics (National Governors Association Center for Four coders received training from a

Best Practices & Council of Chief State School researcher with experience in coding on the use

Officers, 2010): (a) foundational skills and and interpretation of each item, with several

182 Exceptional Children 84(2)

examples provided. Coders practiced by inde- each based on a different sample. In addition, we

pendently coding the same article, and they dis- extracted effect sizes for each outcome, and

cussed discrepancies and checked reliability by when studies included more than one outcome

comparing their ratings with that of the measure that qualified for inclusion in the meta-

researcher. The training continued until ≥90% analysis, we calculated separate effect sizes for

agreement was reached. The four coders inde- each outcome. However, when estimating the

pendently coded the studies; all studies and overall effect of mathematics interventions, we

effect sizes were double-coded by two coders. averaged the effect sizes from all measures so

The coders discussed any discrepancies in cod- that the sample contributed only one effect size

ing and resolved differences by either reexam- to the analysis.

ining the studies to settle on the most

appropriate coding or consulting with the first

author when the disagreement could not be

Data Analysis

resolved. The intercoder agreement was 88% We used Comprehensive Meta-Analysis soft-

for all studies and 98% for all effect sizes. ware (Borenstein, Hedges, Higgins, & Rothstein,

2011) for data analysis. We selected a random

effects model to analyze the effect sizes and com-

Effect-Size Calculation pute estimates of mean effects and standard

We used Hedges’s g, a measure of effect size, errors, as this model allows for generalizations to

for each study in this meta-analysis because be made beyond the studies included in the anal-

the samples in many of the studies were small ysis to the population of studies from which they

and Hedges’s g correction was used to reduce come (Card, 2012).

this small sample size bias. We first calculated We also conducted an analysis of publica-

effect sizes as the difference between treat- tion bias, especially given that our unpub-

ment and control groups’ posttests after adjust- lished studies were primarily dissertations,

ments for pretests and other covariates, divided meaning that we cannot rule out the possibil-

by the pooled standard deviation; then, we ity that other studies with nonsignificant

applied the Hedges’s g correction to all effect findings may have been excluded. Therefore,

sizes to adjust for the small sample size we first conducted an analysis of publication

(Borenstein, Hedges, Higgins, & Rothstein, bias by visually inspecting a funnel plot

2009). We extracted adjusted effect sizes, illustrating the relation between effect size

which were either based on means of gain and study size or precision (standard error of

scores (i.e., posttests – pretests) and standard the effect estimate). This approach would

deviations of pretest scores or based on covari- suggest a possible presence of publication

ance-adjusted means and unadjusted standard bias if there is a gap indicating missing stud-

deviations (see What Works Clearinghouse, ies on the left side of the distribution. Sec-

2014). For three studies that reported only the ond, to further assess for publication bias, we

treatment and comparison groups’ mean post- used Duval and Tweedie’s (2000) trim-and-

test scores, standard deviation, and sample fill procedure, which is available as part of

sizes, we calculated unadjusted effect sizes, the Comprehensive Meta-Analysis software.

which do not take into account other variables We evaluated heterogeneity of variance

that might have had an influence on the out- using the Q statistic and conducted testing for

comes. In one case (Ives, 2007), no means and moderators only when statistically significant

standard deviations were available, and we variance was found under the mixed effects

estimated g from the F statistics (Lipsey & model (Lipsey & Wilson, 2001). When stud-

Wilson, 2001). ies failed to provide sufficient data to code all

We used independent samples as the unit of moderator variables, we dropped those stud-

analysis to address effect size dependency ies from the moderator analysis for the vari-

issues. For example, we coded Manalo, Bunnell, ables but included them in the overall estimate

and Stillman (2000) as two independent studies, of the mean effect size. In addition, we

Jitendra et al. 183

included levels of each moderator in the mod- effect sizes was significant, Qb(19) = 58.7, p <

erator analysis only if the sample was ade- .001, suggesting that between-study variation

quately powered (five or more effect sizes; in effect sizes is larger than what can be

Borenstein et al., 2009). As such, our conclu- explained by sampling error alone. The mean

sions based on the moderator analyses are ten- effect size estimate for students with LD was

tative because of low power. 0.50, suggesting that the average student in

the treatment group performed at about the

69th percentile of the control group; the mean

Results effect size of 0.14 for students with MD sug-

Characteristics of the Selected gests that the average student in the treatment

Studies group performed at about the 55th percentile

of the control group. However, the moderator

Table 1 summarizes the key features of each analysis comparing students with LD and

study. The literature search located 19 reports those with MD was not statistically signifi-

that met our study inclusion criteria, with the cant, Qb(1) = 2.98, p = .08; therefore, student

19 reports providing 20 effect sizes. These status (LD or MD) did not moderate effect

studies spanned 27 years (1990–2017). The size. Thus, in what follows, we describe the

sample sizes in the reported studies ranged aggregated results for students with LD and

from 17 to 806, with a total sample size of MD with regard to the effects of interventions

1,959. Of those with available ethnicity data on mathematics outcomes.

(n = 1,416), 65% were White, and 35% were

minorities (i.e., Black, Hispanic, Asian,

American Indian, multiracial). Nine studies Examining Publication Bias

included middle school students; six involved We first conducted a visual inspection of the

high school students; and four included both. funnel plot (Rothstein, Sutton, & Borenstein,

Fourteen studies (74%) involved students 2005) to determine initial evidence of publi-

with LD. There were important differences cation bias based on asymmetrical distribu-

among the studies that examined the efficacy tion of studies around the mean effect sizes

of mathematics interventions for secondary (Kepes, Banks, & Oh, 2014). Our funnel plot

school students with LD and MD: 79% were showed a symmetrical inverted funnel with

published in a peer-reviewed journal; 68% larger studies toward the top and smaller stud-

were RCTs; 63% were conducted after the ies scattered at the bottom, suggesting the

articulation of the National Council of Teach- absence of bias. Next, we examined whether

ers of Mathematics (2000) standards; in our meta-analytic effect size point estimate

68%, school personnel implemented the differed much from Duval and Tweedie’s

intervention; 63% included BAU as the con- (2000) trim-and-fill imputed effect size point

trol group; 68% reported FOI; 63% were estimate. Results showed no difference

conducted in special education classrooms; between the trim-and-fill imputed point esti-

and 63% addressed higher-level mathemati- mate (0.37) and the random effects model

cal content. point estimate (0.37). This suggests that there

was an absence of publication bias, providing

Overall Effects validity for our results.

positive direction, three in a negative direc-

tion, and one exactly 0. The mean effect size,

Overall Effect Sizes

g = 0.37, was statistically significant (p < Table 2 summarizes the findings from the

.001, 95% confidence interval [0.18, 0.56]), analysis of the moderator variables. Here we

with the 20 effect sizes ranging from −1.10 to summarize the findings by key instructional

1.83. In addition, the homogeneity test for and methodology characteristics and discuss

Table 1. Study Characteristics Included in the Meta-Analysis.

184

Research Sample,

Studya design grade Setting IOI (n) Intervention (n) Instruction time, hr Implementer Math topic FOI, % ES

Bottge & RCT MD, 9 Spec ed Small group T (15): Contextualized 3.3 (5 sessions, 40 Teacher Fractions Yes (98.0) 0.14

Hasselbring (7–8) problem instruction via min)

(1993) videodisc presentation

C (14): BAU (keyword

method)

Bottge et al. (2015) RCT MD, 6–8 Gen ed Large group T (62): Enhanced anchored 49.5~82.5 (66 Teacher Ratio and Yes (87.0) 0.35

(M = 20) instruction (computer- sessions, 45–75 proportional

based, video-based, and min) relationships,

hands-on activities) number system,

C (72): BAU statistics and

probability,

geometry

Butler, Miller, QEX LD, 6–8 Spec ed Large group T (26): Concrete- 7.5 (10 sessions, Teacher Fractions Yes (100) 0.22

Crehan, Babbitt, (12–13) representational-abstract 45 min)

& Pierce (2003) C (24): Representational-

abstract

Calhoon & Fuchs QEX LD, 9–12 Spec ed Small group T (45): PALS and CBM 15.0 (30 sessions, Teacher Operations and Yes (90.0– 0.04

(2003) (M = 9.2) 30 min) algebraic thinking, 96.2)

C (47): BAU

measurement,

geometry

Hutchinson (1993) RCT LD, 8–10 Spec ed Small group T (12): Cognitive strategy 26.7 (40 sessions, Researcher Algebra No 1.22

(one-to-one) instruction and visuals 40 min)

C (8): BAU

Ives (2007) RCT LD, 7–12 Spec ed Large group T (14): Graphic organizers 3.3 (4 sessions, 50 Researcher Algebra No 0.58

(~10) and strategy instruction min)

C (16): Strategy instruction

Jitendra, Dupuis, RCT MD, 7 Gen ed Large group T (149): Schema-based 24.2 (30 sessions, Teacher Ratio and Yes (87.0– 0.40

Star, & Rodriguez (M = 27.6) instruction 45–50 min) proportional 97.0)

(2016) C (111): BAU relationships

Jitendra, Harwell, RCT MD, 7 Gen ed Large group T (399): Schema-based 24.2 (30 sessions, Teacher Ratio and Yes (68.0) 0.26

Dupuis, & Karl (M = 24.0) instruction 45–50 min) proportional

(2017) C (407): BAU relationships

(continued)

Table 1. (continued)

Research Sample,

Studya design grade Setting IOI (n) Intervention (n) Instruction time, hr Implementer Math topic FOI, % ES

Kelly, Gersten, & RCT LD, 9–11 Spec ed Small group T (12): Direct instruction 7.5 (10 sessions, Researcher Fractions Yes (93.0– 0.88

Carnine (1990) (7–8) with Mastering Decimals and 45 min) 94.0)

Percents videodisc program

C (16): BAU

Konold (2004)b RCT LD, 6–12 Gen ed Large group T (37): Concrete- 5.5 (11 sessions, Teacher Expressions and Yes (97.0) 0.02

+ spec representational-abstract 30 min) equations

ed and cognitive strategy

instruction

C (24): BAU

Lambert (1996)b RCT LD, 9–12 Spec ed Small group T (38): Cognitive strategy 7.3 (8 sessions, 55 Teacher Whole number No 0.11

(M = 9) instruction and visuals min) operations,

fractions, decimals,

C (38): BAU

percent

Manalo, Bunnell, & RCT LD, 8 NA Small group T (9): Process mnemonics 4.2 (10 sessions, Researcher Whole number No − 0.03

Stillman (2000) (3–5) 25 min) computations

C (8): Demonstration

(1) involving all four

imitation

operations

Manalo et al. RCT LD, 8 NA Small group T (9): Process mnemonics 4.2 (10 sessions, Researcher Whole number No −0.22

(2000) (2) (3–5) 25 min) computations

C (8): Demonstration

involving all four

imitation

operations

Muoneke (2001)b QEX LD, 9–12 Spec ed Small group T (26): Question-and-answer 24.0~36.0 (16–24 Teacher Whole number Yes (85.0) 1.35

(M = 8) strategy sessions, 90 min) one-step additive

word problems;

C (21): BAU

two-step word

problems involving

all four operations

Rodgers (2011)b QEX MD, 11 Spec ed Small group T (18): Problem-based 3.0 (4 sessions, 45 NA Geometry— No −1.10

(8–10) learning and traditional min) perimeter, area,

math surface area,

volume

C (17): BAU

(continued)

185

186

Table 1. (continued)

Research Sample,

Studya design grade Setting IOI (n) Intervention (n) Instruction time, hr Implementer Math topic FOI, % ES

Walker & Poteet QEX LD, 6–8 Spec ed Small group T (33): Diagrammatic 8.5 (17 sessions, Teacher One-step additive Yes (100) 0.31

(1989–1990) (M = 8) 30 min) word problems

C (37): BAU (keyword

method)

Witzel, Mercer, & RCT LD, 6–7 Gen ed Large group T (34): Concrete- 15.8 (19 sessions, Teacher Expressions and Yes (100) 0.69

Miller (2003) (M = 30) representational-abstract 50 min) equations

C (34): Representational-

abstract

Woodward, QEX LD, 8–9 Spec ed Large group T (21): Mathematics in the 16.7 (20 sessions, Teacher Decimals No 0.00

Baxter, & (21–23) Mind’s Eye: Modeling 50 min)

Robinson (1999) Rationals

C (23): Mastering Decimals

and Percents videodisc

program

Woodward & QEX LD, 6 Gen ed Large group T (25): Transitional One school year Teacher Operations and Yes (NA) 1.13

Brown (2006) (NA) mathematics (55 min daily) algebraic thinking,

number system,

C (28): Connected

measurement and

mathematics program

data, statistics

and probability,

geometry

Xin, Jitendra, RCT LD, 6–8 Spec ed Small group T (11): Schema-based 12.0 (12 sessions, Researcher Ratio and proportion Yes (94.0) 1.83

& Deatline- (4–7) instruction 60 min) word problems

Buchman (2005) C (11): General strategy

instruction

Note. IOI = implementation of intervention; FOI = fidelity of implementation; ES = effect size; RCT = randomized controlled trial; MD = mathematics difficulties; spec ed = special

education classroom; T = treatment; C = control; BAU = business as usual; gen ed = general education classroom; QEX = quasi-experimental; LD = learning disability; PALS and

CBM = peer-assisted learning strategies and curriculum-based measurement; NA = not available.

a

Independent sample. bDissertation.

Jitendra et al. 187

Table 2. Testing for Moderators of Effect Sizes Based on Random Effects Model.

Variable k g SE 95% CI Qb p

Instructional characteristics

Grade level 0.49 .48

Middle school 10 0.44 0.11 [0.23, 0.66]

High school 6 0.23 0.29 [–0.34, 0.79]

Combined 4 0.36 0.25 [–0.13, 0.85]

Intervention approach 1.53 .22

Visuals only 6 0.28 0.13 [0.04, 0.53]

Visuals + other strategies 8 0.52 0.14 [0.24, 0.80]

CBM + other strategies 3 0.39 0.16 [0.09, 0.70]

Other 3 0.10 0.61 [–1.09, 1.28]

Instructional setting 0.07 .79

General education 5 0.46 0.12 [0.24, 0.69]

Special education 12 0.41 0.18 [0.05, 0.77]

General and special education 1 0.02 0.26 [–0.49, 0.53]

Not specified 2 –0.13 0.34 [–0.79, 0.53]

Implementation of intervention 0.00 .99

Large group 9 0.37 0.09 [0.20, 0.55]

Small group 11 0.38 0.21 [–0.04, 0.79]

Instructional time 5.74* .02

≤10 hr 10 0.11 0.15 [–0.18, 0.39]

>10 hr 10 0.58 0.13 [0.32, 0.84]

Mathematics topic 0.01 .92

Foundational content 8 0.36 0.17 [0.02, 0.70]

Higher-level content 12 0.38 0.12 [0.13, 0.62]

Methodological characteristics

Year of publication 0.64 .43

1990–2000 8 0.27 0.14 [0.00, 0.54]

2001–2017 12 0.42 0.13 [0.16, 0.67]

Research design 0.15 .70

RCT 13 0.39 0.09 [0.20, 0.57]

QEX 7 0.28 0.26 [–0.23, 0.79]

Implementer 1.37 .24

Researcher 6 0.70 0.29 [0.14, 1.27]

School personnel 13 0.35 0.08 [0.19, 0.52]

Fidelity of implementation 2.43 .12

Yes 13 0.47 0.10 [0.27, 0.68]

No 7 0.06 0.25 [–0.43, 0.54]

Type of measure 0.12 .73

Researcher developed 19 0.39 0.11 [0.18, 0.59]

Standardized 6 0.48 0.24 [0.01, 0.94]

Type of control group 1.05 .30

Business as usual 12 0.29 0.11 [0.07, 0.50]

Alternative intervention 8 0.53 0.21 [0.12, 0.93]

Note. k = number of effect sizes; CI = confidence interval; CBM = computer-based modules; RCT = randomized

controlled trial; QEX = quasi-experimental.

*p < .05. **p < .01. ***p < .001.

188 Exceptional Children 84(2)

outcomes that are based on five or more out- 1990 and 2000; this difference between means

comes. was not statistically significant (p = .43). In

addition, the study design variable was not

Instructional characteristics. We examined six statistically significant (p = .70): The mean

key instructional characteristics: grade level, effect size estimate for RCT studies was 0.39,

instructional approach, instructional setting, and that for quasi-experimental studies was

implementation of intervention, instructional 0.28. Researcher-implemented interventions

time, and mathematical topic. The results by yielded a large effect size, 0.70, whereas the

grade level showed no statistically significant mean effect size estimate for interventions

difference (p = .48) between the mean effect implemented by school personnel was 0.35,

size estimate for middle school studies (g = indicating a small to moderate effect. How-

0.44) and that for high school studies (g = ever, the contrast between means was not sta-

0.23). With regard to instructional approaches, tistically significant (p = .24), suggesting that

the eight studies of visual models combined treatment implementer did not moderate

with other strategies (e.g., priming the under- effect size. Results indicated a moderate effect

lying problem structure, cognitive strategy) size, 0.47, for studies reporting fidelity of

produced the largest effect size, 0.52, and the intervention. In contrast, the mean effect size

effect size for studies of only visual models estimate for studies not reporting fidelity of

was 0.28; however, the difference was not intervention was 0.06, a small to negligible

statistically significant (p = .22). The advan- effect. However, the difference between

tage of using these approaches, as compared means for the instructional variable of FOI

with regular classroom instruction or alterna- was not significant.

tive approaches, was significant for both Similarly, the type of measure was not sig-

approaches, indicated by the fact that the con- nificant (p = .73): Researcher-developed

fidence intervals of the effect sizes did not assessments and standardized assessments

contain 0. yielded small to moderate effect sizes (g =

Effect sizes were similar for studies con- 0.39 and 0.48, respectively). With regard to

ducted in general education (g = 0.46) or spe- type of control group, studies using BAU

cial education settings (g = 0.41), and the instruction yielded a small effect size, 0.29,

difference was not statistically significant (p = whereas studies using alternative interven-

.79). We also found no significant differences tions indicated a moderate effect size, 0.53;

(p = .99) between large group (g = 0.37) and however, the contrast between means was not

small group (g = 0.38) implementation of statistically significant (p = .30).

intervention. The effect of instructional time

was found to be significant, Qb(1) = 5.74, p =

Discussion

.02. The mean effect size for instruction pro-

vided for >10 hr (g = 0.58) was greater than The purpose of this meta-analytic review was

that for instructional time ≤10 hr (g = 0.11). to extend prior evaluations of mathematical

The effect of mathematical topic was not sig- interventions for secondary students with LD

nificant (p = .92), with similar effect sizes for and MD by quantifying intervention effec-

studies addressing foundational content (g = tiveness in terms of effect sizes. The finding

0.36) and higher-level content (g = 0.38). from this review indicates a small to moder-

ate effect size (g = 0.37). An effect size of

Methodology characteristics. We examined six 0.37 can be interpreted as approximately 65%

key methodology characteristics: year of pub- of students in the treatment group performing

lication, research design, implementer, FOI, above the mean of students in the control

type of measure, and type of control group. group (Lipsey et al., 2012). Our finding that

By year of publication, the mean effect size the interventions in these studies were effec-

estimate for studies conducted after 2000 was tive is largely congruent with the conclusion

0.42 and 0.27 for studies conducted between of prior reviews of mathematics interventions

Jitendra et al. 189

for K–12 students with or at risk for MD 2008). Thus, one could argue that interven-

(Baker et al., 2002; Dennis et al., 2016; Ger- tions would be expected to produce stronger

sten et al., 2009; Kroesbergen & Van Luit, effects for middle school than high school stu-

2003; Swanson & Hoskyn, 1998). In addition dents. Although average effects for the two

to the overall findings, this review examined grade levels were aligned with this expecta-

the differential effects of mathematics inter- tion (i.e., middle school, g = 0.44, vs. high

ventions on student mathematics outcomes school, g = 0.23) and the advantage for middle

by various instructional and methodological school was significant (i.e., the confidence

features. interval of the effect sizes did not contain 0),

this contrast was not statistically significant.

The most likely explanation for this finding

Instructional Characteristics

has to do with the inability to include all stud-

The effect sizes for the following instructional ies in these two levels of the variable for this

characteristics were nearly identical: (a) gen- moderator analysis, which could have

eral education and special education settings decreased the statistical power of the meta-

(g = 0.46 and 0.41, respectively), (b) large and analysis.

small group implementation (g = 0.37 and In terms of intervention approaches, there

0.38, respectively), and (c) foundational con- are contrasting hypotheses to consider in exam-

tent and higher-level mathematical content (g ining the results of the meta-analysis. On one

= 0.36 and 0.38, respectively). These findings hand, single-component interventions (visuals

for setting, grouping, and mathematical topic only) may by parsimonious and work best for

are somewhat surprising given the different students with LD and MD because of limited

profiles of students with LD and MD served cognitive, behavioral, and attentional capaci-

in general education and special education ties (e.g., Compton, Fuchs, Fuchs, Lambert, &

classrooms (Wagner, Marder, Blackorby, & Hamlett, 2011; Fuchs et al., 2014; Geary,

Cardoso, 2002) and the different kinds of Hoard, Byrd-Craven, Nugent, & Numtee,

structures and mathematical topics covered in 2007). On the other, multicomponent interven-

the two environments. However, the compa- tions (i.e., visual models combined with other

rable findings would be expected if research- strategies) may produce stronger effects; meta-

ers took these differences (e.g., large group analysis of reading interventions support this

instruction in general education classrooms, expectation (see Wanzek, Wexler, Vaughn, &

adding prerequisite knowledge to be learned) Ciullo, 2010). The results revealed a tendency

into account and designed appropriate inter- toward this latter hypothesis (i.e., visuals only,

ventions. g = 0.28, vs. visuals combined with other strat-

With respect to the grade-level feature egies, g = 0.52), but the contrast was not statis-

examined in this meta-analysis, research indi- tically significant. However, it is encouraging

cates that the mathematical skills of children that the effect size for visuals combined with

with LD progress about 1 year for every 2 strategies is somewhat higher than the effect

years in school but plateau by age 10 or 12 size (g = 0.47) found in Gersten et al. (2009).

(Cawley, Parmar, Yan, & Miller, 1998; Wei A key finding of this meta-analysis was that

et al., 2013). Further, the mathematics content the effect of these interventions appeared to be

targeted in high schools is more complex and greater when the intervention lasted >10 hr than

advanced than the content in middle school. when it was ≤10 hr. This finding is consistent

Note also that the average annual growth with research regarding the length of instruc-

effect sizes for typically achieving students tional time provided to students being an essen-

ranged from 0.30 in Grades 6 and 7 to 0.01 in tial variable to learning (Rosenshine & Berliner,

Grades 11 and 12 on nationally normed mea- 1978). Further, within the context of a multi-

sures of mathematics achievement, indicating tiered system of support model, a framework

a decrease in effect sizes as grade level for providing high-quality mathematical con-

increases (see Bloom, Hill, Black, & Lipsey, tent and instruction to all students (core or Tier

190 Exceptional Children 84(2)

1 instruction) and additional intervention sup- studies and was generally high when reported

port for some students (secondary and intensive (see Table 1), which might have accounted for the

levels of intervention), the length of instruc- lack of differences in implementation by research-

tional time is one of the features for intensifying ers and school personnel. Our finding that inter-

instruction for students who do not adequately ventions appeared to be feasibly implemented by

respond to Tier 1 instruction. Results from the researchers and school personnel is consistent

moderator analysis of the present study support with recent research on reading interventions in

the need for engaging students in learning for an the early elementary grades (Wanzek et al., 2016).

extended period, especially when learning

higher-level mathematical content. A key finding of this meta-analysis

was that the effect of these

Methodological Characteristics interventions appeared to be

With regard to year of publication, it is not greater when the intervention lasted

clear whether the overall effectiveness of >10 hr than when it was ≤10 hr.

mathematical interventions would be higher

for studies published on or before 2000 or Research suggests that measuring treat-

published after 2000, when the National ment fidelity is crucial to assessing internal

Council of Teachers of Mathematics stan- validity—that is, the extent to which “observ-

dards were in effect. The finding of no sig- able differences in outcomes—or lack

nificant difference between studies published thereof—were due to the intervention and not

on or before 2000 (g = 0.27) and those pub- extraneous factors” (Woolley, Rose, Mercado,

lished after 2000 (g = 0.42) is probably due & Orthner, 2013, p. 59). Failure to consider or

to low power for the moderator analysis. check the FOI of the treatment has been iden-

One possible source of variation may tified as a potential deterrent to reaching accu-

relate to research design of the studies. In rate conclusions about the effectiveness of the

this review, 65% of the studies were RCTs. treatment (Bellg et al., 2004). We hypothe-

Because quasi-experimental studies gener- sized that studies that reported and assessed

ally have larger variances, we expected FOI would yield larger effect sizes than stud-

these studies to produce larger effects than ies that did not report and, presumably, did not

RCTs, even though such results could be assess treatment fidelity. However, the con-

misleading (Hedges, 1982). However, the trast was not significant (p = .12) between

evidence does not support this expectation. studies assessing treatment fidelity (g = 0.47)

The results in this review revealed similar and those that did not assess it (g = 0.06).

effects for RCTs (g = 0.39) and quasi-exper- Contrary to expectation, researcher-

iments (g = 0.28). developed measures designed to match the

With respect to the implementer variable, content of the interventions produced a some-

there are several possible reasons to expect that what smaller effect size (g = 0.39) than stan-

researchers as implementers would produce dardized measures (g = 0.48). However, the

stronger effects than school personnel as imple- contrast was not significant, possibly because

menters. One explanation is that researchers tend of the small number of studies (n = 6) using

to implement interventions as planned more so standardized measures. To evaluate the rela-

than teachers. Previous meta-analyses have indi- tive effectiveness of their interventions con-

cated implementer effects (e.g., Berkeley, sistently, it would be useful for researchers to

Scruggs, & Mastropieri, 2010; Dennis et al., utilize common measures.

2016; Wanzek et al., 2010). Our results revealed a Finally, research suggests that the use of dif-

tendency toward this hypothesis (i.e., researchers, ferent types of control groups yields different

g = 0.70, vs. school personnel, g = 0.35), but this effect size estimates (Karlsson & Bergmark,

difference was not statistically significant. It is 2015). Research methods textbooks make a

worth noting that FOI was reported in 65% of the distinction between inactive control groups

Jitendra et al. 191

(e.g., no treatment, BAU or standard treatment, with LD and MD to develop the necessary

or a waiting-list control) and active control mathematics skills to enter or advance in

groups (e.g., an alternative intervention), which STEM-related careers (see Hughes et al., 2014),

produce absolute and relative effect estimates, we urge researchers to continue to conduct stud-

respectively. An additional distinction is that, ies that will provide evidence about effective

compared with no-treatment controls, BAU interventions in mathematics for these students.

often includes “some potentially effective com- A second limitation is that we found only

ponents that may be absent in no-treatment five studies of students with MD that met our

controls” (p. 422). Therefore, and as expected, inclusion criteria. Further, how students were

the more effective the comparison group com- identified as LD may account for the variabil-

ponents, the smaller the effect size estimates. ity in effect sizes because most authors studied

However, we found no significant difference school-identified students, and, although they

between studies using BAU controls (g = 0.29) provided some description of the identification

and alternative intervention controls (g = 0.53). process, most did not provide sufficient infor-

A possible explanation is that the BAU control mation about these students’ MD. It is clear

condition in the included studies taught the that in order to identify how to better meets the

same content and for the same duration as the needs of students with LD and MD and

treatment group. improve intervention outcomes, more studies

need to be conducted with these populations.

Another limitation is posed by the lack of

Limitations and Future Directions availability of sufficient studies to inform the

Similar to other meta-analyses and primary field about what intervention approaches are

studies, there are at least three limitations that most effective. We found that the most com-

are relevant to the current meta-analysis. First, monly used method in the studies included in

there is the possible limitation of generalizing this meta-analysis was visual representa-

results from the small set of RCT and quasi- tions. Although not statistically significant,

experimental studies. Despite our exhaustive visuals combined with other strategies

search process, including published and unpub- resulted in the largest effect size (g = 0.52),

lished studies, we found only 19 studies in the which is practically significant and can be

past 27 years that focused on the outcome of interpreted as showing that approximately

secondary mathematics interventions for stu- 70% of students in the treatment group per-

dents with LD and MD. This number is rela- formed above the mean of students in the

tively small compared to intervention studies of control group (Lipsey et al., 2012). However,

reading and other content areas. For example, the strategies were varied (i.e., priming the

prior meta-analyses identified 43 reading inter- underlying structure, cognitive strategy, or

vention studies and 28 science intervention contextualized instruction), with few in each

studies for secondary students with disabilities category. Additional testing of these inter-

(Scammacca, Roberts, Vaughn, & Stuebing, ventions needs to be addressed in future

2015; Scruggs, Mastropieri, Berkeley, & research.

Graetz, 2010). Further, the effect of the small

number of studies influenced results of the

Implications for Practice

moderator analyses, which are lower powered

than the main effect analysis (Hedges & Pigott, The results of this meta-analysis demonstrate

2004). Future studies may reveal that our find- that secondary students with LD and MD ben-

ing of no significant difference between the MD efit from well-designed mathematics interven-

and LD groups is a product of having too few tions. Although the majority of interventions in

studies with too few participants. Given the this meta-analysis lasted for less than a school

increasing growth in jobs related to STEM (sci- year, the mean effect size reflected an average

ence, technology, engineering, and mathemat- gain of a little more than one third a standard

ics) and how crucial it is for secondary students deviation, which on average was about a year’s

192 Exceptional Children 84(2)

growth when compared with annual growth majority of studies in this meta-analysis that

rates for typically achieving students in second- provided >10 hr of instruction focused on

ary grades. Further, the studies in this review advanced mathematics (e.g., ratios and propor-

compared treatment to a BAU or alternative tional relationships, algebra) and resulted in

treatment comparison group receiving interven- enhanced mathematics performance. In sum, if

tion, which implies the added benefits of treat- the achievement gap between students with

ment over another intervention rather than the LD and MD and their peers without MD is to

benefit of treatment over no intervention. Based be reduced, it is important to continue to eval-

on this information, one implication that we can uate the effect of interventions to understand

draw from our finding is that practitioners can how to support these students effectively.

support the learning of struggling students in

Grades 6 to 12 by using well-designed mathe- References

matics interventions (e.g., visual models com-

*References marked with an asterisk indicate stud-

bined with other strategies). Given that two

ies included in the meta-analysis.

thirds of eighth graders with disabilities per- Baker, S., Gersten, R., & Lee, D. S. (2002). A synthe-

form below the basic level on national assess- sis of empirical research on teaching mathemat-

ments (National Center for Education Statistics, ics to low-achieving students. Elementary School

2015), it is crucial that students with LD and Journal, 103, 51–73. doi:10.1086/499715

MD receive mathematics instruction that Bellg, A. J., Resnick, B., Minicucci, D. S.,

addresses their needs and improves mathemat- Ogedegbe, G., Ernst, D., Borrelli, B., . . .

ics performance. Czajkowski, S. (2004). Enhancing treatment

fidelity in health behavior change studies:

Best practices and recommendations from

Although the majority of

the NIH behavior change consortium. Health

interventions in this meta-analysis Psychology, 23, 443–451. doi:10.1037/0278-

lasted for less than a school year, 6133.23.5.443

the mean effect size reflected an Berkeley, S., Scruggs, T. E., & Mastropieri, M.

A. (2010). Reading comprehension instruc-

average gain of a little more than tion for students with learning disabilities,

one third a standard deviation, 1995–2006: A meta-analysis. Remedial and

which on average was about a Special Education, 31, 423–436. doi:10.1177/

year’s growth. 0741932509355988

Bloom, H. S., Hill, C. J., Black, A. R., & Lipsey, M.

W. (2008). Performance trajectories and perfor-

The findings from this review also high- mance gaps as achievement effect-size bench-

light the importance of optimizing instruc- marks for educational interventions. Journal

tional time if students with LD and MD are to of Research on Educational Effectiveness, 1,

become proficient in mathematics. For exam- 289–328. doi:10.1080/1934574080240072

ple, our results show that when students with Borenstein, M., Hedges, L. V., Higgins, J. P. T., &

LD and MD receive >10 hr of well-designed Rothstein, H. R. (2009). Introduction to meta-

mathematics instruction, one can expect an analysis. West Sussex, England: Wiley.

increase of 22 percentile points, compared Borenstein, M., Hedges, L. V., Higgins, J. P. T.,

with a gain of only 4 percentile points when & Rothstein, H. R. (2011). Comprehensive

instruction occurs for ≤10 hr. As such, it is Meta-Analysis (Version 2.2.064) [Computer

software]. Englewood, NJ: Biostat.

vital that students with LD and MD receive

Borman, G. D., Hewes, G. M., Overman, L. T.,

sufficient instructional time that addresses & Brown, S. (2003). Comprehensive school

their specific needs. Many students with LD reform and achievement: A meta-analysis.

and MD often receive very little focused Review of Educational Research, 73, 125–230.

instructional time on higher-level mathematics doi:10.3102/00346543073002125

despite its educational and occupational *Bottge, B. A., & Hasselbring, T. S. (1993). A com-

importance (Miller & Hudson, 2007). The parison of two approaches for teaching complex,

Jitendra et al. 193

remedial math classes. Exceptional Children, 59, doi:10.1111/j.0006-341X.2000.00455.x

556–566. doi:10.1177/001440299305900608 Every Student Succeeds Act of 2015, Pub. L. No.

*Bottge, B. A., Toland, M. D., Gassaway, L., 114-95, § 1177. Stat. (2015).

Butler, M., Choo, S., Griffen, A. K., & Ma, X. Fuchs, L. S., Fuchs, D., & Compton, D. L. (2013).

(2015). Impact of enhanced anchored instruc- Intervention effects for students with comor-

tion in inclusive math classrooms. Exceptional bid forms of learning disability understand-

Children, 81, 158–175. doi:10.1177/001 ing the needs of nonresponders. Journal

4402914551742 of Learning Disabilities, 46, 534–548.

*Butler, F. M., Miller, S. P., Crehan, K., Babbitt, doi:10.1177/0022219412468889

B., & Pierce, T. (2003). Fraction instruction Fuchs, L. S., Schumacher, R. F., Sterba, S. K.,

for students with mathematics disabilities: Long, J., Namkung, J., Malone, A., . . .

Comparing two teaching sequences. Learning Changas, P. (2014). Does working memory

Disabilities Research & Practice, 18, 99–111. moderate the effects of fraction intervention?

doi:10.1111/1540-5826.00066 An aptitude-treatment interaction. Journal

*Calhoon, M. B., & Fuchs, L. S. (2003). The of Educational Psychology, 106, 499–514.

effects of peer-assisted learning strategies doi:10.1037/a0034341

and curriculum-based measurement on the Geary, D. C, Hoard, M. K., Byrd-Craven, J.,

mathematics performance of secondary stu- Nugent, L, & Numtee, C. (2007). Cognitive

dents with disabilities. Remedial and Special mechanisms underlying achievement deficits

Education, 24, 235–245. doi:10.1177/074193 in children with mathematics learning dis-

25030240040601 ability. Child Development, 78, 1343–1359.

Card, N. (2012). Applied meta-analysis for social doi:10.1111/j.1467-8624.2007.01069.x

science research. New York, NY: Guilford. Gersten, R., Chard, D. J., Jayanthi, M., Baker,

Cawley, J. F., Parmar, R. S., Yan, W., & Miller, S. K., Morphy, P., & Flojo, J. (2009).

J. H. (1998). Arithmetic computation per- Mathematics instruction for students with

formance of students with learning disabili- learning disabilities: A meta-analysis

ties: Implications for curriculum. Learning of instructional components. Review of

Disabilities Research & Practice, 13, 68–74. Educational Research, 79, 1202–1242.

Compton, D. L., Fuchs, L. S., Fuchs, D., doi:10.3102/0034654309334431

Lambert, W., & Hamlett, C. (2011). The Hedges, L. V. (1982). Estimation of effect

cognitive and academic profiles of read- size from a series of independent experi-

ing and mathematics learning disabilities. ments. Psychological Bulletin, 92, 490–499.

Journal of Learning Disabilities, 45, 79–95. doi:10.1037/0033-2909.92.2.490

doi:10.1177/0022219410393012 Hedges, L. V., & Pigott, T. D. (2004). The power

Cortiella, C. (2011). The state of learning dis- of statistical tests for moderators in meta-

abilities. New York, NY: National Center for analysis. Psychological Methods, 9, 426–445.

Learning Disabilities. doi:10.1037/1082-989X.9.4.426

Dennis, M. S., Sharp, E., Chovanes, J., Thomas, A., Hughes, E. M., Witzel, B. S., Riccomini, P. J.,

Burns, R. M., Custer, B., & Park, J. (2016). A Fries, K. M., & Kanyongo, G. Y. (2014). A

meta-analysis of empirical research on teaching meta-analysis of algebra interventions for

students with mathematics learning difficulties. learners with disabilities and struggling learn-

Learning Disabilities Research & Practice, 31, ers. Journal of the International Association of

156–168. doi:10.1111/ldrp.12107 Special Education, 15, 36–47.

Dossey, J., McCrone, S., & Halvorsen, K. (2016). Hunt, J. H., & Vasquez, E., III. (2014). Effects of

Mathematics education in the United States ratio strategies intervention on knowledge of

2016: A capsule summary book written ratio equivalence for students with learning

for the Thirteenth International Congress disability. The Journal of Special Education,

on Mathematical Education (ICME-13). 48, 180–190. doi:10.1177/0022466912474102

Reston, VA: National Council of Teachers of *Hutchinson, N. L. (1993). Effects of cognitive

Mathematics. strategy instruction on algebra problem solv-

Duval, S., & Tweedie, R. (2000). Trim and fill: ing of adolescents with learning disabilities.

A simple funnel-plot–based method of test- Learning Disability Quarterly, 16, 34–63.

ing and adjusting for publication bias in doi:10.2307/1511158

194 Exceptional Children 84(2)

Hwang, J., & Riccomini, P. J. (2016). Enhancing *Lambert, M. A. (1996). Teaching students with

mathematical problem solving for second- learning disabilities to solve word-problems:

ary students with or at risk of learning dis- A comparison of a cognitive strategy and a

abilities: A literature review. Learning traditional textbook method (Doctoral disser-

Disabilities Research & Practice, 31, 169– tation). Available from ProQuest Dissertations

181. doi:10.1111/ldrp.12105 and Theses. (UMI No. 9639550)

*Ives, B. (2007). Graphic organizers applied to Lipsey, M. W., Puzio, K., Yun, C., Hebert, M.

secondary algebra instruction for students A., Steinka-Fry, K., Cole, M. W., . . . Busick,

with learning disorders. Learning Disabilities M. D. (2012). Translating the statistical rep-

Research & Practice, 22, 110–118. resentation of the effects of education inter-

doi:10.1111/j.1540-5826.2007.00235.x ventions into more readily interpretable

*Jitendra, A. K., Dupuis, D. N., Star, J. R., & forms (Publication No. NCSER 2013-3000).

Rodriguez, M. C. (2016). The effects of Washington, DC: National Center for Special

schema-based instruction on the proportional Education Research. Retrieved from https://ies.

thinking of students with mathematics diffi- ed.gov/ncser/pubs/20133000/pdf/20133000.

culties with and without reading difficulties. pdf

Journal of Learning Disabilities, 49, 354–367. Lipsey, M. W., & Wilson, D. B. (2001). Practical

doi:10.1177/0022219414554228 meta-analysis. Thousand Oaks, CA: SAGE.

*Jitendra, A. K., Harwell, M. R., Dupuis, D. N., & Maccini, P., McNaughton, D., & Ruhl, K. L. (1999).

Karl, S. R. (2017). A randomized trial of the Algebra instruction for students with learn-

effects of schema-based instruction on pro- ing disabilities: Implications from a research

portional problem solving for students with review. Learning Disability Quarterly, 22,

mathematics problem-solving difficulties. 113–126. doi:10.2307/1511270

Journal of Learning Disabilities. 50, 322–336. Maccini, P., Mulcahy, C. A., & Wilson, M. G.

doi:10.1177/0022219416629646 (2007). A follow-up of mathematics inter-

Karlsson, P., & Bergmark, A. (2015). Compared ventions for secondary students with learning

with what? An analysis of control-group disabilities. Learning Disabilities Research

types in Cochrane and Campbell reviews of & Practice, 22, 58–74. doi:10.1111/j.1540-

psychosocial treatment efficacy with sub- 5826.2007.00231.x

stance use disorders. Addiction, 110, 420–428. *Manalo, E., Bunnell, J. K., & Stillman, J. A.

doi:10.1111/add.12799 (2000). The use of process mnemonics in

*Kelly, B., Gersten, R., & Carnine, D. (1990). teaching students with mathematics learning

Student error patterns as a function of cur- disabilities. Learning Disability Quarterly, 23,

riculum design: Teaching fractions to 137–156. doi:10.2307/1511142

remedial high school students and high Marita, S., & Hord, C. (2017). Review of math-

school students with learning disabilities. ematics interventions for secondary stu-

Journal of Learning Disabilities, 23, 23–29. dents with learning disabilities. Learning

doi:10.1177/002221949002300108 Disability Quarterly, 40, 29–40. doi:10.1177/

Kepes, S., Banks, G. C., & Oh, I. S. (2014). 0731948716657495

Avoiding bias in publication bias research: The Mazzocco, M. M. M. (2007). Defining and dif-

value of “null” findings. Journal of Business ferentiating mathematical learning disabilities

and Psychology, 29, 183–203. doi:10.1007/ and difficulties. In D. B. Berch & M. M. M.

s10869-012-9279-0 Mazzocco (Eds.), Why is math so hard for

*Konold, K. B. (2004). Using the concrete-rep- some children? The nature and origins of

resentational-abstract teaching sequence mathematical learning difficulties and disabil-

to increase algebra problem-solving skills ities (pp. 29–47). Baltimore, MD: Brookes.

(Doctoral dissertation). Available from Miller, S. P., & Hudson, P. J. (2007). Using

ProQuest Dissertations and Theses. (Accession evidence-based practices to build mathe-

Order No. AAT 3143383) matics competence related to conceptual, pro-

Kroesbergen, E. H., & Van Luit, J. E. (2003). cedural, and declarative knowledge. Learning

Mathematics interventions for children with Disabilities Research & Practice, 22, 47–57.

special educational needs a meta-analysis. doi:10.1111/j.1540-5826.2007.0023.x

Remedial and Special Education, 24, 97–114. Misquitta, R. (2011). A review of the literature:

doi:10.1177/07419325030240020501 Fraction instruction for struggling learners in

Jitendra et al. 195

mathematics. Learning Disabilities Research Scruggs, T. E., Mastropieri, M. A., Berkeley, S.,

& Practice, 26, 109–119. doi:10.1111/j.1540- & Graetz, J. E. (2010). Do special educa-

5826.2011.00330.x tion interventions improve learning of sec-

*Muoneke, A. F. (2001). The effects of a question ondary content? A meta-analysis. Remedial

and action strategy on the mathematical word and Special Education, 31, 437–449. doi.

problem-solving skills of students with learn- org/10.1177/0741932508327465

ing problems in mathematics (Doctoral disser- Shin, M., & Bryant, D. P. (2015). Fraction inter-

tation). Available from ProQuest Dissertations ventions for students struggling to learn math-

and Theses. (Accession Order No. AAT ematics: A research synthesis. Remedial and

3008402) Special Education, 36, 374–387. doi:10.1177/

Myers, J. A., Wang, J., Brownell, M. T., & Gagnon, 0741932515572910

C. J. (2015). Mathematics interventions for Swanson, H. L., & Hoskyn, M. (1998).

students with learning disabilities (LD) in Experimental intervention research on stu-

secondary school: A review of the litera- dents with learning disabilities: A meta-

ture. Learning Disabilities: A Contemporary analysis of treatment outcomes. Review

Journal, 13, 207–235. of Educational Research, 68, 277–321.

National Center for Education Statistics. (2015). doi:10.3102/00346543068003277

National Assessment of Educational Progress. Wagner, M., Marder, C., Blackorby, J., & Cardoso,

Washington, DC: U.S. Department of D. (2002). The children we serve: The demo-

Education. Retrieved from http://nces.ed.gov/ graphic characteristics of elementary and

nationsreportcard/ middle school students with disabilities and

National Council of Teachers of Mathematics. their households. Menlo Park, CA: SRI

(2000). Principles and standards for school International.

mathematics. Reston, VA: Author. Wagner, M., Newman, L., Cameto, R., Levine,

National Governors Association Center for Best P., & Garza, N. (2006). An overview of find-

Practices & Council of Chief State School ings from Wave 2 of the National Longitudinal

Officers (2010). Common Core State Standard Transition Study-2 (NLTS2). Retrieved from

in mathematics. Washington, DC: Author. https://ies.ed.gov/ncser/pdf/20063004.pdf

National Mathematics Advisory Panel. (2008). *Walker, D. W., & Poteet, J. A. (1989–1990). A

Foundations for success: Final report of comparison of two methods of teaching math-

the National Mathematics Advisory Panel. ematics story problem-solving with learning

Washington, DC: U.S. Department of disabled students. National Forum of Special

Education. Education Journal, 1, 44–51.

*Rodgers, L. D. (2011). Examining the imple- Wanzek, J., Vaughn, S., Scammacca, N., Gatlin,

mentation of a problem-based learning and B., Walker, M., & Capin, P. (2016). Meta-

traditional hybrid model of instruction in analyses of the effects of Tier 2 type reading

remedial mathematics classes designed for interventions in Grades K–3. Educational

state testing preparation of eleventh grade Psychology Review, 28, 551–576. doi:10.1007/

students (Doctoral dissertation). Available s10648-015-9321-7

from ProQuest Dissertations and Theses. Wanzek, J., Wexler, J., Vaughn, S., & Ciullo, S.

(Accession Order No. AAT 3532763) (2010). Reading interventions for struggling

Rosenshine, B. V., & Berliner, D. C. (1978). readers in the upper elementary grades: A

Academic engaged time. British Journal of synthesis of 20 years of research. Reading and

Teacher Education, 4, 3–16. doi:10.1080/ Writing, 23, 889–912. doi:10.1007/s11145-

0260747780040102 009-9179-5

Rothstein, H. R., Sutton, A. J., & Borenstein, M. Watt, S. J., Watkins, J. R., & Abbitt, J. (2016).

(Eds.). (2005). Publication bias in meta-anal- Teaching algebra to students with learning

ysis: Prevention, assessment, and adjustment. disabilities: Where have we come and where

New York, NY: Wiley. should we go? Journal of Learning Disabilities,

Scammacca, N. K., Roberts, G., Vaughn, S., & 49, 437–447. doi:10.1177/0022219414564220

Stuebing, K. K. (2015). A meta-analysis of Wei, X., Lenz, K. B., & Blackorby, J. (2013). Math

interventions for struggling readers in Grades growth trajectories of students with disabili-

4–12. Journal of Learning Disabilities, 48, ties: Disability category, gender, racial, and

369–390. doi:10.1177/0022219413504995 socioeconomic status differences from ages

196 Exceptional Children 84(2)

7 to 17. Remedial and Special Education, 34, tion fidelity to professional development.

154–165. doi:10.1177/0741932512448253 Journal of Education and Training Studies, 1,

What Works Clearinghouse. (2014). WWC pro- 55–68. doi:10.11114/jets.v1i1.48

cedures and standards handbook. Retrieved Xin, Y. P., & Jitendra, A. K. (1999). The effects

from https://ies.ed.gov/ncee/wwc/Docs/refer- of instruction in solving mathematical word

enceresources/wwc_procedures_v3_0_stan- problems for students with learning prob-

dards_handbook.pdf lems: A meta-analysis. The Journal of

Witzel, B. S. (Ed.) (2016). Bridging the gap Special Education, 32, 207–225. doi:10.1177/

between arithmetic and algebra. Arlington, 002246699903200402

VA: Council for Exceptional Children. *Xin, Y. P., Jitendra, A. K., & Deatline-Buchman,

*Witzel, B. S., Mercer, C. D., & Miller, M. D. A. (2005). The effects of mathematical word

(2003). Teaching algebra to students with problem-solving instruction on middle school

learning difficulties: An investigation of students with learning problems. The Journal

an explicit instruction model. Learning of Special Education, 39, 181–192. doi:10.117

Disabilities Research & Practice, 18, 121– 7/00224669050390030501

131. doi:10.1111/1540-5826.00068 Zhang, D., & Xin, Y. P. (2012). A follow-up

*Woodward, J., Baxter, J., & Robinson, R. (1999). meta-analysis for word-problem-solving inter-

Rules and reasons: Decimal instruction for aca- ventions for students with mathematics diffi-

demically low achieving students. Learning culties. Journal of Educational Research, 105,

Disabilities Research & Practice, 14, 15–24. 303–318. doi:10.1080/00220671.2011.627397

doi:10.1207/sldrp1401_2 Zheng, X., Flynn, L. J., & Swanson, H. L.

*Woodward, J., & Brown, C. (2006). Meeting the (2013). Experimental intervention stud-

curricular needs of academically low-achieving ies on word problem solving and math dis-

students in middle grade mathematics. The abilities a selective analysis of the literature.

Journal of Special Education, 40, 151–159. doi: Learning Disability Quarterly, 36, 97–111.

10.1177/00224669060400030301 doi:10.1177/0731948712444277

Woolley, M. E., Rose, R. A., Mercado, M., &

Orthner, D. K. (2013). Teachers teaching dif- Manuscript received August 2017; accepted Sep-

ferently: A qualitative study of implementa- tember 2017.

Copyright of Exceptional Children is the property of Sage Publications Inc. and its content

may not be copied or emailed to multiple sites or posted to a listserv without the copyright

holder's express written permission. However, users may print, download, or email articles for

individual use.

- Colorectal cancerTransféré parregina
- 2015 Strengths Business Unit Meta-AnalysisTransféré parMalou Oosterholt
- How NOT to Write a Medical PaperTransféré parKaustubh Keskar
- 2016 - Beyond Born versus Made A New Look at Expertise - Hambrick y otros.pdfTransféré parmad10000
- Statistics for Business and Economics: bab 7Transféré parbalo
- Antioxidants_Review of Evidence on All-cause Mortality (Chocrane, 2012)Transféré parWilson
- description: tags: techappendix01 359Transféré paranon-812524
- Strategies to Prevent Death by Suicide- Meta-Analysis of Randomised Controlled TrialsTransféré parManuelCárdenas
- Instructional Scaffolding in STEM EducationTransféré paralfonso lopez alquisirez
- Hippotherapy—an Intervention ToTransféré parec16043
- The Clock Drawing TestTransféré parIcaro
- Research on the Effectiveness of Online Learning.pdfTransféré parMohammedElGohary
- Assistive technology interventions for adolescents and adults with learning disabilities: An evidence-based systematic review and meta-analysisTransféré parMarcela Portela
- 1 Predictors of Persistent Pain After Breast Cancer SurgeryTransféré parFirman Dariyansyah
- Development, Problem Behavior, And Quality of Life in Population Based Sample of Eight Year Old Children With Down SyndromeTransféré paratika sgrt
- 12.fullTransféré parDarmawan
- 10 Commitment Across CulturesTransféré parFakher Khalili
- EGZR Erelera ofbeko olugTransféré parthclaess
- jurnal penggunaan antibiotik topikal sebelum penutupan insisi luka primer operasiTransféré parAnonymous 9DDZ54
- Cassey and McardleTransféré parJoseLuisTaczaPonce
- Standard Deviation and VarianceTransféré parJannela yernaidu
- Jasso Response, 1986Transféré parLawrence D. LaPlue
- tmpCCD5.tmpTransféré parFrontiers
- Handouts in StatTransféré parKathleen Dando
- Disagreement and Biases in Inﬂation Expectations∗Transféré parapi-26691457
- mle.tools==An R Package for Maximum Likelihood Bias CorrectionTransféré parsss
- Pediatrics 2011 Finnell Peds.2011 1332Transféré parGustavo Uriondo
- 0000542-201102000-00015Transféré par'-dooublleaiienn Itouehh Iin
- ArticoleTransféré parVisovan Alexandra
- 2014 PA in SpA - A Systematic ReviewTransféré par21flybaby

- pardonTransféré parapi-400727133
- part aTransféré parapi-400727133
- ebdTransféré parapi-400727133
- intellecTransféré parapi-400727133
- intelleTransféré parapi-400727133
- intellTransféré parapi-400727133
- graphTransféré parapi-400727133
- learninggTransféré parapi-400727133
- evidenceTransféré parapi-400727133
- fishTransféré parapi-400727133

- ch06Transféré parAqib Shahzad
- robbins_mgmt11_ppt10_GE.pptTransféré parPrince Husshen Sotelo
- Investor Presentation ArticleTransféré parpravesh.dudani9077
- Asia Pacific Office Market Overview Q4 2009Transféré parColliers International Thailand
- pompeii.pdfTransféré parIvona Petite
- Chopin s Polish BalladeTransféré parSoni Petrovski
- investor preference to derivative marketsTransféré parSharanya Sv
- physical regions of canada 10Transféré parapi-193600542
- 4413_Ver_4_0.pdfTransféré parZaryab Nisar
- Med Baleros Patient2Transféré parDessa Baleros
- Rexesas_RX_MCUs.pdfTransféré parAbdelhadi Sdedeke
- Regulatory on Derivatives (1) (from shodhganga)Transféré parsharath
- 168 Pulmonary Hypertension Review of the New WHO ClassificationTransféré parsofiah
- My Experience With GodTransféré parJason Raquin Roque
- Dasgupta - Jammu & KashmirTransféré parapi-3720996
- Pro-se Prepresentation ArticleTransféré parrabbi_josiah
- CIV2FAQTransféré parcbillingsworth
- NS12-1loadsExTransféré parJizelle Jumaquio
- A Workaholic EconomyTransféré parNGUYÊN PHẠM NGỌC KHÔI
- Star Trek Roleplaying GameTransféré parLorkhon
- 06 Paras v Comelec GR 123169 110496Transféré parWinfred Tan
- SAP BPC 10 MS Product Availability MatrixTransféré parJohnny Jiang
- Fencing News - August 2009Transféré parFencingNews
- Peters Creek Community Initiative - ResolutionTransféré parCP Tew
- Zurich Edition 17 FullTransféré parsyedman
- COMPETENCY BASED JOB ANALYSIS AND PERFORMANCE MANAGEMENT OF NON-MEDICAL STAFF AT APOLLO HOSPITALS, DHAKATransféré parChayan Ard
- KX-TEB308 User ManualTransféré pardenckav
- PSE New Trading RulesTransféré parknightshalo
- Critical Care and the Global Burden of Critical Illness in AdultsTransféré parPaulo Hibernon
- NOKIA PROJECT FINALTransféré partanu2011