Research'Based Implications From Extensive Early Reading ...

School Psychology Review, 2007, Volume 36, No. 4, pp. 541-561

Research'Based Implications From Extensive Early Reading Interventions

Jeanne Wanzek Florida State University

Sharon Vaughn The University of Texas at Austin

Abstract. A synthesis of the extant research on extensive early readitig interventions for students with reading difficulties and disabilities is provided. Findings from 18 studies published between 1995 and 2005 revealed positive outcomes for students participating in extensive interventions. Results indicated higher effects for studies providing intervention to students in the smallest group sizes as well as providing intervention early (grades K-1). No differences in overall outcomes were revealed between studies implementing highly standardized interventions or interventions with less standardized implementation. Implications for practice and future research are discussed.

Response to intervention (RTI) provides inadequate or inappropriate instruction a framework for accomplishing several highly (Fletcher, Coulter, Reschly, & Vaughn. 2004). valued goals: (a) early identification of stu- Although school psychologists embrace these dents at risk for academic difficulties through goals conceptually, decisions related to effecuniversal screening practices; (b) early and tive implementation of these goals are often targeted intervention for students at risk; (c) more problematic. ongoing progress monitoring--more firequently Perhaps the most challenging aspects of for students most at risk and less frequently for implementation in RTI models are the decitypical achieving students; (d) use of increas- sions about what types of interventions should ingly more intensive tiers of research-based be implemented; the amount of time interveninstruction to meet students' needs; and (e) tions should be implemented; who should proimproved confidence that students referred for vide the interventions; and whether a uniform, special education who participated in evi- more standardized approach to instruction dence-based RTI models are less likely to be should be implemented or whether schools students who are academic casualties from should use a less standardized, or more indi-

Special thanks to Kim Kayser for her assistance with the selection of articles for the synthesis. We also thank Nancy Scammacca and Kathryn Tackett for their competent assistance in organizing data from the studies. Correspondence regarding this article should be addressed to Jeanne Wanzek. Florida State University and Florida Center for Reading Research. 227 North Bronough. Suite 7250, Tallahassee, FL 32301; E-mail: jwanzek@fsu.edu Copyright 2007 by the National Association of School Psychologists, ISSN 0279-6015

541

School Psychology Review, 2007. Volume 36, No. 4

vidualized. approach (Bums, Appleton, & Stehouwer. 2005; Fuchs, Mock. Morgan, & Young, 2003; Reschly. 2005; Vaughn & Fuchs, 2003). Fundamentally, an evidencebased approach to addressing these questions would be aligned with the guidelines as well as recommended practice {see, for review, Kratochwill & Stoiber, 2002).

Purpose and Rationale for Synthesis

Extensive Interventions

For the purpose of this synthesis, we were interested in reading interventions that would be considered more extensive and that provided for a significant enough period of time so students' RTI and overall intervention effectiveness could be determined. We defined extensive interventions as occurring for 100 sessions or longer, which is the equivalent of 20 weeks of daily intervention.

This article provides a synthesis of research aimed at exploring some of the fundamental questions related to effective implementation of RTI. We synthesized the extant research on extensive early reading interventions lo examine several issues related to RTI impiementation: (a) outcomes for students with reading difficulties or disabilities after participation in extensive early reading interventions, and (b) features of interventions associated with high effect sizes, including instructional group size, duralion, and level of standardization.

We recognized that to adequately address these questions would require a relatively large population of studies that addressed a common academic problem. For this reason, we selected studies that represented early reading interventions (kindergarten through Grade 3). Reading interventions were selected because ofthe following: (a) most students are referred for leaming disabilities that are attributed to difficulties with reading (Lyon, 1995); (b) most students identified as leaming dis-

Interventions differ from typical reading instruction in that they are designed to address the instructional needs of students who are experiencing difficulties in leaming to read or who have reading disabilities. There are several reasons for selecting extensive interventions for this synthesis. First, as Vellutino and colleagues have specified (Vellutino, Scanlon, & Jaceard. 2003). we can understand whether students have true reading difficulties only if we can control for inadequacies in instruction--best accomplished by providing extensive interventions to students at risk to determine relative progress. Second, previous syntheses of the reading intervention research have not specifically addressed the effectiveness of extensive interventions (Foorman. 2003; McCardle & Chhabra. 2004; Pressley. 2006). Third, within an RTI model, understanding the efficacy of interventions provided to students after classroom instmction and less extensive interventions (depending on criteria used by the district) is essential and less well understood (Vaughn, Wanzek. Linan-Thompson, & Murray. 2007).

abled have reading as an Individualized Education Program goal (Lemer, 2000); (c) there is a substantial data base of studies on early reading interventions (e.g.. McCardle & Chhabra, 2004; Rayner. Foorman. Perfetti, Pesetsky, & Seidenberg, 2001); and (d) many of the recent initiatives at the state level (e.g., Alabama Reading Initiative, Florida Reading Initiative. Texas Reading Initiative) and policies at the federal level (Reading First and Early Reading First programs within the No Child Left Behind Act, 2001) were based on findings from these early reading studies.

Types of Extensive Interventions

A simple way to conceptualize interventions might be to consider that they could be implemented in two primary ways: standardized or individualized. Standardized interventions specify, a priori, the elements of reading instruction that will be implemented. The elements selected are (a) associated with improved outcomes in previous studies, (b) well defined in a curriculum, and (c) implemented by personnel who are trained specifically in ihe implementation of the curricula. Although adjustments may be made to address students'

542

Implications From Extensive Inlerventions

levels, fundamental to applyitig a standardized

Method

approach is using a research-based standard curriculum and assuring fidelity ot implementation.

A second type of intervention is a more individualized approach. In school psychology, this approach may also he referred to as a prohlem-solvuig approach (Bergen & Kratochwill. 1990). Typically, these approaches are directed

Studies were identified through a twostep process. First, we conducted a computer search of ERIC and PsycINFO to locate studies pubiisbed between 1995 and 2005. We selected the decade between 1995 and 2005 because a large number of early intervention studies were reporied in the research literature

at defining the student's problem in behavioral during this period and the studies during this

terms, measuring performance in the natural set- period were the ones that greatly influenced

ting, determining the specific goals to address policy related to RTI. Key disability descrip-

the problem, designing and/or selecting an inter- tors or root forms of descriptors (reading diff*,

vention to meet those goals, monitoring the stu- leaming disab*, delay, disorder*, at-risk. high

dent's progress toward those goals to determine the effectiveness of the intervention, adjusting the intervention if needed, and then making decisions about future interventions (Ikeda et al.. 2002; National Association of State Directors of Special Education. 2005). In our review of the literature on extensive interventions, we were unable to locate any interventions implemented with an individualized approach. So, although it would be valuable to contrast more standardized and individualized intervention, the lack of research in this area prevented us from addressing

risk, disab*, dyslex*) were used in combination with key reading descriptors or root forms of descriptors (reading, interven*. instruction, reading intervention, reading strategies, supplemental instruction, special educ.*, inclus*, integrat*, phonological awareness, phonemic awareness, phon*, fluency, vocab*. comp*) to identify possible articles. The initial electronic search yielded 26.062 articles. Second, we calculated a hand search of 7 major journals for the years 2003-2005 {Exceptiottal Children. Journal of Educational Psychology, Journal of Leaming Disabilities, Journal of Special

this question directly.

Education, Leaming Disabilities Research &

For the purpose of this synthesis, we Practice, Reading Research Quarterly, and

examined student outcomes after participation in extensive interventions implemented with varying levels of standardization. The studies reviewed provided a range in their descriptions of implementation standardization, with

Scfwol Psychology Review). We selected the years 2005-2005 because they were tiie most recent years and therefore studies during that time would not be located through citation searching and other syntheses.

some categorized as high on standardization (curricula provided to all participants are well specified with few or no modifications) to low on standardization (curricula and instructional practices are less well specified and more re-

Studies were selected if they met the following criteria: (a) The study was reported in a peer-reviewed journal and printed in English; (b) The participants included students with leaming disabilities or students identified

sponsive to individual students' needs).

as at risk for reading difficulties (e.g., students

In addition to documenting the effects of with low ability, low phonemic awareness,

interventions that were high and low on stan- low income, language disorders); studies with

dardization, we were interested in describing additional participants were included if disag-

the components of interventions (e.g., instruc- gregated data were provided for the students

tional elements, personnel) associated with high effect sizes. This is the type of evidencebased information frequentiy requested by school psychologists and special educators (National Association of State Directors of Special Education, 2005).

with leaming disabilities or the students were identified as at risk; (c) The participants were enrolled in grade levels between kindergarten and third grade inclusive; (d) Interventions targeted early literacy in an alphabetic language, were provided for 100 sessions or

543

School Psychology Review, 2007, Volume 36, No. 4

more, and were not part of the general educa- and al! relevant pairs of groups for instances in

tion curriculum provided to all students; (e) which the study involved more than two

Interventions were provided as pan of the groups that were of interest. In cases where

school programming (does not include home, means, standard deviations, and sample sizes

clinic, or camp programs); and (f) Dependent were provided for two or more independent

variables addressed reading outcomes.

groups, these data were used to compute effect

sizes. For Mathes et al. (2005), mean differ-

Data Analysis

ences, sample size, and independent t statistics

were used to compute effect sizes and standard Only studies that met the above-stated errors. criteria were coded. The vast majority of stud-

ies were eliminated based on their abstracts, which provided information that assured us

Results

the study would not meet criteria. When abstracts were ambiguous or led us to believe the study met criteria, we located and reviewed the study to ensure it met our criteria.

A total of 18 studies, reported in 20 journal articles, met criteria for inclusion in the synthesis. Fourteen studies used a treatment or comparison group design, with 5 of these studies

Coding procedures. An extensive coding sheet was developed to organize pertinent information about each of the studies. The code sheet was based on code sheets used in previous research (Vaughn, Kim et al., 2003; Wanzek et al.. 2006) as well as the elements specified in the What Works Clearinghouse Design and Implementation Assessment Device (Institute of Education Sciences, 2003). Data were collected on (a) participants (e.g., age, gender, exceptionality); (b) methodology (e.g., research design, assignment); (c) intervention and comparison descriptions; (d) measures; and (f) findings. There were three coders for the articles. Interrater agreement was calculated separately for each codesheet category (e.g., participants, design, and so on) and reached 91% or ahove for all categories. Interrater agreement was calculated as the number of agreements divided by the number of agreements plus the number of disagreements.

randomly assigning students to conditions (Gunn, Biglan, Smolkowski, & Ary, 2000: Jenkins, Peyton. Sanders, & Vadasy, 2004; Mathes et al., 2005; Torgesen, Wagner, & Rashotte, 1997; Vadasy, Jenkins, Antil, Wayne. & O'Connor, 1997). Two of the studies with random assignment of students also provided additional data in joumal articles published at a later date (Gunn, Smolkowski. Biglan, & Black, 2002; Torgesen et al., 1999). Random assignment is the most critical element of an experimental design providing the greatest evidence of causal effects. These studies, then, may provide the most reliable evidence of intervention effects. All of these studies also measured intervention effects on standardized, norm-referenced measures, increasing conHdence that student gains are generalizable to the broad skills measured (e.g.. comprehension, word recognition) and not specific to the intervention skills taught.

All code sheets were reviewed by the first

Three studies examined student re-

author for comprehetisiveness and accuracy. sponse over time in a single treatment group

(Dev, Doyle, & Valente. 2002; Englert,

Effect size calculation. Standardized Mariage, Garmon, & Tarrant, 1998; Vaughn,

mean difference effect sizes and standard er- Li nan-Thompson, & Hickman, 2003) and one

rors were calculated using the data reported in study implemented a single-subject design

each study. For all studies, the procedure for (Snider, 1997). These studies were not de-

calculating unbiased effect size estimates for signed to provide causal information, but did

Cohen's d provided by Hedges (1981) was offer key information for researchers and ed-

used (this statistic is also known as Hedges's ucators related to interventions and instruction

g). Effect size estimates and standard errors with potential for affecting student outcomes.

were computed for all dependent measures Replication of fitidings and future studies test-

544

Implications From Extensive Interventions

ing these interventions with treatment or com- ing the intervention students to a group of stu-

parison designs are needed to confirm the find- dents receiving no additional intervention.

ings and generalize outcomes. A description

Alternatively, smaller effects were cal-

of the key features for each study is provided in Table 1.

Sufficient data for computing effect sizes was included for 13 of the studies. We summarize these study results by examining effects by the following: (a) duration of intervention, (b) instructional group size, (c) grade level of intervention, and (d) degree of standardization. It should be noted that the studies synthesized were not designed to specifically answer questions related to duration, group size, or grade level, and only one study held the degree of intervention standardization constant to examine related effects (Mathes et al., 2005). As a result, these factors and the reported findings may not be causally related. Without the experimental manipulation of these factors of interest, no causal inferences should be made about the factors' individual contributions. Rather, we examine the effects of the individual studies in these areas to describe the extant literature available to school

culated for two studies investigating a phonological awareness (PA) intervention for kindergarten students over a period of 5-6 months (O'Connor. Notari-Syverson. & Vadasy, 1996; Schneider et a!., 2000). A notreatment comparison group was not available in the O'Connor et al. study, but students with disabilities in self-contained classrooms were compared to students with disabilities in integrated classrooms who also received the same intervention. The mean effect for students receiving the intervention in the self-contained classrooms was 0.18 (range = -0.43 to 0.75). However, when compared to students without disabilities receiving the same intervention, the students with disabilities instructed in the self-contained classrooms lagged behind on a number of prereading skills. Schneider et al. also compared students receiving the intervention to not-at-risk students receiving instruction in typical classrooms. Students receiving a PA intervention over the 5-month period significantly outperformed students not at risk

psychologists and identify areas for additional at post-test on measures of phonological

research.

awareness. However, 1-2 years later, at the

Descriptive information is provided for end of first and second grade, the comparison

the five studies in which effect sizes could not group of not-at-risk students maintained sig-

be computed (Dev et al., 2002; Bnglert et al.. nificantly higher outcomes on decoding and

1998; Schneider. Roth, & Ennemoser, 2000: comprehension. A second group of students in

Snider, 1997; Vellutino & Scanlon, 2002). A the Schneider et al. study received the PA

summary of study findings and the dependent intervention along with intervention in letter

measures included in the effect sizes is pro- sounds and achieved outcomes equivalent to

vided in Table 2.

the not-at-risk students on the measures of decoding and comprehension at the end of first

Effects by Duration of Intervention

and second grade. Student IQ was used as a

Interventions implemented for 5-7 months. Two studies implemented a first-grade, phonics-based intervention (Sound Partners) with reading of text (Jenkins et al., 2004; Va-

covariate in the analyses. The students in the typical classrooms participated in social events and games and did not receive any formal reading instruction.

dasy, Jenkins, Antil, Wayne, & O'Connor,

Vaughn. Linan-Thompson, and Hick-

1997). Jenkins and colleagues reported similar man (2003) implemented intervention over 30

effects on measures of prereading and reading weeks but allowed students who obtained pre-

whether students read more decodable (range, determined fluency levels to exit intervention

effect size [ES] = 0.35 to 0.99) or less decodable after 20 weeks. Pre- to post-test standardized

(range, ES = 0.41 to 1.11) text during the inter- mean change effect sizes ranged from 0.53

vention. Mean effects for Vadasy et al. to 6.06 on measures of word attack, passage

were 0.50 (range = 0.31 to 0.78) when compar- comprehension, fluency, and phonological

545

School Psychology Review, 2007, Volume 36, No. 4

I

a.

-J C/3

S ?-

00 OX) Ml i 3

SS? -

13

g E3

o 2

'^ .1 o

i^ ?s^ iI5I e bg-i*B a, .c ? k, .c

2g ?

1c 1c

CO u u a u

0- H t- H

u o H OS

a, I

u

a. cS

. Ml

M -- -- r ) 00

(fl en ?S 6

5

u E nj

I" t/i ^

^

C C

O

a E

o ^ '0

17

f S CT\ ( S

(N 00 m (N

X

I?I

X

X X

Vi

?* m

i'7

c c 'E E

-- icns o o

X X

7

O =3

^

1/3

-a

3s

?a

.,, ,,

CTJ ^ ^

^

"c r "c

T-> I--? , j _ ^

D

c

re

iai/i

r- '?>

o

?oS

a.

X o^

Is

o u u o

546

Implications From Extensive Interventions

awareness. However, effect sizes calculated school instruction, the mean effect size for

for pre- to post-test gains were generally students in the intervention with text was 0.64

higher than effect sizes calculated with a treat- (range = 0.17 to 0.99) and the mean effect

ment and comparison group. The pre- to posttest effects in the Vaughn et al. study provide an indication of positive outcomes for students, but these effects cannot be accurately compared to the treatment versus comparison group effects in the other studies.

size for students in the intervention without text reading was 0.65 (range = 0.13 to 1.33). The same intervention with decodable text reading was implemented over 35 weeks in a previous study (Vadasy, Sanders, Peyton. & Jenkins, 2002). In this study, students increased standard scores into the average range

Interventions implemented over 8-9 months. A full range of effects were demonstrated for interventions implemented over approximately one school year. No effects were found for students with reading disabilities receiving an intervention of analytic phonics (onset-rime word patterns) incorporated with writing and reading phonetic readers when compared to students receiving whole-word instruction on measures of word reading, spelling, and phonemic awareness

on measures of reading, spelling, and word identification by the end of the intervention. The 2002 study also examined a second intervention (Thinking Partners) for second-graders that was implemented for 35 weeks. The Thinking Partners intervention incorporated grade-level trade hooks and comprehension strategy instruction and yielded a mean effect of 0.23 (range = 0.09 to 0.38) over students in typical school instruction (Vadasy et al., 2002).

(mean ES = -0.05; [range = -0.23 to 0.27];

There were three studies yielding larger

Foorman et al., 1997). However, a mean effect effects following intervention for 8-9 months.

of 0.27 (range = 0.05 to 0.59) was demon- Two of the studies examined the same interven-

strated for a synthetic phonics intervention in tion (Early Steps) with first-grade students. The

comparison to the whole-word instruction on students were provided with supported reading

these same measures. Mathes et al. (2005) and rereading of leveled books as well as phon-

reported on two interventions provided for ics instruction with a focus on word patterns.

approximately 8 months in addition to en- These interventions were compared to students

hanced classroom instruction. One interven- receiving a small-group, pull-out intervention

tion provided a standardized protocol includ- provided by the school. On measures of word

ing explicit phonics instruction with decodahle recognition, word attack, spelling, and compns-

text and comprehension strategies (Proactive), hension. the mean effect sizes were 0.76

whereas the other intervention provided word (range = 0.68 to 0.83; Morris, Tyner. & Pemey,

work and reading in leveled books based on 2000) and 0.74 (range = 0.59 to 0.91; Santa &

individual student needs (Responsive). When Hoien, 1999). A similar intervention reported by

compared to the outcomes of students receiv- Miller (2003) also yielded high effects on infor-

ing only the enhanced classroom instruction, mal measures of word recognition and spelling

hoth interventions yielded similar mean ef- in comparison to students who did not receive

fects (mean ES = 0.33 [range = 0.00 to 0.63] intervention (mean ES = 0.84 [range = 0.71

for Proactive; mean ES = 0.31 [range = 0.22 to 1.09]). In addition, a single-subject study ex-

to 0.53] for Responsive). Effects specifically amining seven students with leaming disabilities

for comprehension were 0.21 and 0.30 for demonstrated gains in fluency, with all but one

the Proactive and Responsive interventions, snident reading at least 70-105 words per

respectively.

minute after 9 months of intervention. Vellutino

Higher mean effects were revealed for two intervention studies implementing Sound Partners with and without text reading for 8 months (Vadasy, Sanders, & Peyton. 2005). In comparison to students receiving typical

and Scanlon (2002) provided intensive intervention (two semesters) in first and second grade to students demonstrating either low growth or very low growth and reported 75% of low growth and 67% of very low growth students

547

School Psychology Review, 2007, Volume 36, No. 4

u

o oo

-- gi -- 00--? --

-- ooooooooooooooooooo

n

u

U

OOOOOOOOOOOOCJOOOOOOOOOOO

?>a 'a3

Ul

Of .

 ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download