• E-mail
• Print
• Comment
• Font Size
• Digg
• del.icio.us
• Discuss article

Using Enrichment Reading Practices to Increase Reading Fluency, Comprehension, and Attitudes

Posted on: Thursday, 22 May 2008, 03:00 CDT

By Reis, Sally M Eckert, Rebecca D; McCoach, D Betsy; Jacobs, Joan K; Coyne, Michael

ABSTRACT. The authors used a randomized design to investigate the effects of an enriched reading program, Schoolwide Enrichment Reading Model (SEM-R), with blocks of independent reading on oral reading fluency (ORF), reading comprehension, and attitudes toward reading for students in Grades 3-5. The authors used SEM-R in comparison with basal reading programs to investigate the addition of planned enrichment strategies and independent reading on students' reading achievement. Students and teachers were randomly assigned to either the SEM-R treatment with 1 hr of SEM-R and 1 hr of basal instruction or to the control group with 2 hr of basal instruction. The authors found statistically significant differences favoring the SEM-R treatment group in ORF. These results suggest that an enriched reading program that challenges and engages students may produce higher ORF than a standard basal program. Keywords: attitudes, comprehension, enrichment reading practices, increased reading fluency, independent reading

For many years, educators and researchers have noted the relation between improved reading comprehension and scholastic achievement (Anderson, Hiebert, Scott, & Wilkinson, 1985; Chall & Conard, 1991; Snow, Burns, & Griffin, 1998). Research has suggested that several factors-including classroom environment, socioeconomic status (SES), and teacher training-play a major role in the lower rates of literacy and achievement often experienced by at-risk students (Chall & Conard, 1990; Duke, 2000; Snow et al.). Despite years of reform and research, few researchers have specifically addressed the diverse needs of readers in Grades 3-5 (Erlbaum, Vaughn, Hughes, & Moody, 2000). Duke suggested that to promote successful literacy development and continued academic achievement, students must experience large amounts of print representing a variety of genres and print types and a sense of student agency (i.e., self- determination and choice).

To achieve growth in student reading skills and ensure later school success, teachers must provide all students with appropriately challenging instructional materials (Anderson et al., 1985; Chall & Conard, 1991; National Reading Panel [NRP], 2000). In many elementary reading classes in U.S. schools, teachers must plan to meet the needs of a diverse group of readers, from students with learning disabilities to those who read above grade level (Durkin, 1990; Rickford, 2001).

A recent study at The National Research Center on the Gifted and Talented (NRC/GT; Reis et al., 2004) of instructional strategies in Grade 3 and Grade 7 reading classrooms demonstrated that when classroom teachers had advanced books or resources available in their classrooms, these materials remained on the shelf. Talented readers received little differentiation for reading in most of the 12 classrooms observed. Rather, students who read well above grade level usually received instruction and curricular materials identical to those of students who read at grade level and those who were several years below grade level.

In response, researchers at the NRC/GT developed a reading intervention, the Schoolwide Enrichment Reading Model (SEM-R; Reis et al., 2005), which uses the pedagogy of gifted education to enhance reading instruction for all students. The SEM-R provides many students with a proper match between abilities, course content, and instruction to increase their challenge level. Specifically, we designed the present study to examine the belief that an increase in the amount of time spent independently reading self-selected, appropriately challenging books in Grades 3-5 would positively affect student achievement.

Schoolwide Enrichment Model

The SEM-R is an enrichment approach to reading that evolved from the Enrichment Triad Model (Renzulli, 1977) and the Schoolwide Enrichment Model (SEM; Renzulli, 1977; Renzulli & Reis, 1985, 1997), an approach to talent development that is widely used in districts across the United States.

The SEM-R includes three general dynamic categories of reading instruction that were designed to enable flexibility of implementation and content in response to the needs of teachers and students. This approach is based on Renzulli's (1977) Enrichment Triad Model with three levels of enrichment: (a) Type 1 enrichment includes broad exposure to areas in which students might have interests, (b) Type 2 enrichment is training and methods instruction, and (c) Type 3 enrichment provides students opportunities to pursue self-selected topics of interest. The background of the SEM-R intervention evolved from the SEM, whose goals include improved academic performance of all students in all areas.

In the SEM, students receive three services. First, interest and learning styles assessments create or identify students' interests and encourage students to further develop and pursue these interests. Learning-style preferences that are assessed in the SEM include projects, independent study, teaching games, simulations, peer teaching, programmed instruction, lecture, drill and recitation, and discussion. This information, which focuses on strengths rather than deficits, is compiled in a total talent portfolio that teachers use to decide on talent development opportunities for students in the talent pool or, in many schools, for all students. This expanded approach to identifying talent potentials is essential for educators to make genuine efforts to include underrepresented students in a plan for total talent development. This approach is also consistent with the flexible conception of developing gifts and talents that has been a cornerstone of the SEM approach promoting equity in enrichment or gifted programs (Renzulli & Reis, 1985, 1997).

Second, curriculum modification and compacting is provided in the SEM to all eligible students for whom the regular curriculum is modified by eliminating portions of previously mastered content. This elimination or streamlining of curriculum enables above- average students to avoid repetition of previously mastered work and guarantees mastery while finding time for more appropriately challenging activities (Reis, Burns, & Renzulli, 1992). The SEM uses a form titled the Compactor (Renzulli & Smith, 1978) to document content areas that have been compacted and alternative work that has been substituted. In the SEM, Types 1, 2, and 3 enrichment are offered to all students; however, Type 3 enrichment is usually more appropriate for students with higher levels of ability, interest, and task commitment.

The SEM encourages enjoyment in learning and the opportunity to pursue creative, stimulating work by the three components that are also integrated into the SEM-R (Renzulli, 1977). In separate studies on the SEM and the Triad Model, researchers have demonstrated the SEM's effectiveness in schools with widely differing socioeconomic levels and program organization patterns (Burns, 1998; Olenchak, 1988; Olenchak & Renzulli, 1989). Researchers have studied and field tested the effectiveness of the model for more than 20 years regarding (a) student creative productivity (Delcourt, 1993; H?bert, 1993; Westberg, 1999), (b) personal and social development (Olenchak, 1991), (c) use of SEM with culturally diverse or special needs populations (Baum, Renzulli, & H?bert, 1999; Olenchak, 1991), (d) student self-efficacy (Schack, Starko, & Burns, 1991; Starko, 1988), (e) the SEM as a curricular framework (Karafelis, 1986; Reis, Gentry, & Park, 1995), (f) learning styles and curriculum compacting (Reis et al., 1993), and (g) longitudinal research on the SEM (Delcourt; H?bert; Westberg). This research on the SEM suggests that the model is effectively serving high-ability students and providing enrichment in a variety of educational settings, including schools serving culturally diverse and lowsocioeconomic populations.

Oral Reading Fluency

Oral reading fluency (ORF) is the ability to read aloud with speed, accuracy, and expression (NRP, 2000; Snow et al., 1998). Fluent reading has long been considered a hallmark of well- developed reading skills. According to Adams (1990), the "most salient characteristic of skillful readers is the speed and effortlessness with which they seem able to breeze through text" (p. 409). Fluency is also closely related to reading comprehension (Fuchs, Fuchs, Hosp, & Jenkins, 2001). This relation is explained by several prominent theories with similar underlying reasoning (Stanovich, 2000). These theories assert that the efficient processing of lower level reading skills (i.e., word recognition) characterized by fluent reading frees up cognitive resources for higher level reading skills (i.e., comprehension). Thus, if a reader has to spend too much time and energy figuring out what the words are, she or he will be unable to concentrate on what the words mean (Coyne, Kame'enui, & Simmons, 2001). For the nonfluent reader, "reading becomes a slow, laborintensive process that only fitfully results in understanding" (NRP, p. 3). Because reading fluency reflects the complex orchestration of both lower and higher level processes, researchers can consider it a reliable indicator of overall reading proficiency (Fuchs et al.). Research on Independent Reading

A relation seems to exist between reading fluency and time engaged in independent reading (Cunningham & Stanovich, 1998), but little experimental research has been conducted to explore that relation. A recent study of the reading attitudes of almost 2,000 middle school students revealed that their first choice of reading activity was independent reading (Ivey & Broaddus, 2001), but in studies with random assignment of students and teachers, few researchers have investigated the effects of independent reading on reading fluency. Results are mixed in those studies that have been conducted (NRP, 2000).

Review of Related Research

Little recent experimental research has been conducted on increasing reading comprehension and fluency; we review some of the studies selected for the NRP (2000) report in this section. One difficulty in reviewing the research that examines the relation between independent reading and increased reading fluency and comprehension is that researchers often define independent reading differently, and they use dissimilar units of analysis, observations, and statistical procedures for data analysis. In the few studies in which researchers have reported using random assignment, different results have emerged across various grade levels. For example, Holt and O'Tuel (1989) studied seventh- and eighth-grade students who read two or more grade levels below average in reading; those researchers randomly assigned participants to a 10-week program of sustained silent reading for 20 min at three times each week. Increases in vocabulary use resulted, and increases in reading comprehension occurred only for seventh-grade students. Burley (1980) studied the use of sustained silent reading as one of four interventions to which economically disadvantaged high school students who were enrolled in a 6-week summer program were randomly assigned. In that study, sustained silent reading had more positive effects on reading achievement for these students than did programmed textbooks, programmed cassette tapes, and programmed skill development kits.

Manning and Manning (1984) used random assignment with 415 students who participated in a sustained silent reading program with peer-interaction groups that had higher reading achievement scores than a group who used sustained silent reading with individual teacher-student conferences. Collins (1980), who used random assignment to study the effects of sustained silent reading on reading achievement, found that students in Grades 3-5 who were randomly assigned to experimental groups and read for 20-30 min daily scored significantly higher on the Iowa Test of Basic Skills Reading Comprehension tests than did students in the control group. In a quasi-experimental study involving the use of sustained silent reading, Summers and McClelland (1982) found no significant achievement differences in 65 intact classrooms in which half of the classrooms participated in sustained silent reading while the other half served as control classrooms.

Other researchers have studied student reports of independent reading at home and in school. Anderson, Wilson, and Fielding (1988), for example, studied the relation between the amount of student reading outside school and reading achievement for 155 fifth- grade students, and found that the reading of books was the best predictor of reading achievement. Taylor, Frye, and Maruyama (1990) conducted a study of 195 fifth- and sixth-grade students who kept daily reading logs, and those researchers found that time spent in reading in school contributed to growth in reading achievement.

Debate still exists among reading researchers about the level of text difficulty in independent reading that must be present for growth to occur. For example, Carver and Leibert (1995) implemented a summer reading program in which students were randomly assigned to read books that were either relatively easy or matched to be somewhat challenging. The researchers found that students who "engaged in reading relatively easy library books for 6 weeks gained in their reading level, vocabulary, rate, or efficiency" (p. 46). The level of challenge in the text is discussed in some, but not in most, studies of independent reading. Chall and Conard (1991) described an optimal text as one that is slightly above a student's reading level, thus requiring the student to make an effort to read the text and occasionally to ask for assistance. To achieve optimal challenge, a reader must encounter new concepts and language.

In his zone of proximal development theory, Vygotsky (1962) suggested that children's language development is determined by shared meanings of the adult language that the child internalizes. From the beginning of a child's life, knowledge is socially influenced and constructed (Vygotsky), and in one model for instruction, a zone that Vygotsky established as a proximal level of difficulty suggests a need for adult assistance in the form of guided practice with a clear goal, followed by an evaluation of independent performance by someone who is more experienced. Vygotsky theorized that studying challenging material with assistance enabled students to internalize more complex material. In other words, students must experience a supportive struggle to learn.

Because of the wide range of results and the variation in sample size, definition of silent reading, length of intervention, and type of intervention, it is understandable that the NRP (2000) concluded that "based on the existing evidence, the NRP can only indicate that while encouraging students to read might be beneficial, research has not yet demonstrated this in a clear and convincing manner" (p. 3). The panel called for research on the effectiveness of encouraging students to engage in independent reading and emphasized the need for rigorous experimental studies that measure a range of reading outcomes, including reading fluency and comprehension. The present study is aligned with this goal.

Method

We used an experimental design with multilevel modeling to evaluate the effectiveness of the SEM-R intervention in two elementary schools. Students in Grades 3-5 were randomly assigned either to participate in SEM-R reading instruction or to continue with the regular reading program control groups. Teachers were also randomly assigned to teach the SEM-R curriculum or to continue with the regular reading program. SEM-R treatmentgroup students participated in 1 hr of the school's basal reading program and 1 hr of the SEM-R intervention. Control-group students participated in 2 hr of the regularly scheduled basal program.

Participants

The number of teachers and students who participated in the study from both schools is summarized in Table 1, and brief descriptions of each school and corresponding demographic information are provided. Seventeen treatment teachers participated from the two schools as did 14 control teachers, 313 treatment students, and 245 control students from Grades 3-5, with an approximately even distribution of boys and girls (see Table 1).

Meadowbrook Elementary School. For the last 30 years, the population of this small New England town has been stable, especially among school-aged children. Meadowbrook is off the town's main thoroughfare in the center of a suburban neighborhood 20 miles from a major urban area. Meadowbrook Elementary School housed all Grade 3-5 students in the school district and had a full complement of faculty and staff, including classroom teachers, reading and math specialists, a school psychologist, and other ancillary personnel to work with the 475 students in the school (see Table 2). At Meadowbrook, 17.1% of the students received compensatory education services, such as remedial reading, and although many were included in both treatment and control classrooms, students who spent the entire language arts block in special education pull-out programs were not included in the present study. The teaching staff included 3-4 classroom teachers at each grade level and specialists in special education, music, art, physical education, Spanish, technology, reading, speech, and gifted education. The support staff included a full-time nurse, media specialist, social worker, part- time school psychologist, occupational therapist, physical therapist, and 15 paraprofessionals. Test scores at the school have shown gradual improvement in the 5 years prior to the study; and at the time of the study, 62% of students achieved the state goal of mastery on the fourth-grade state reading assessment. This performance was comparable to the average mastery rate in the state (57.9%).

Northwest School. Northwest School is an elementary school in a New England mill city of approximately 23,000 residents. Although the area was once known for its thriving textile manufacturing, the town and its inhabitants now struggle for economic viability; median household income is approximately 50% below the state average. Northwest has 22 classrooms, including one portable classroom, housing 414 students in Grades K-4, with approximately 3-4 classrooms per grade level and similar specialists as at the other school, including reading and math consultants, a school psychologist, and other ancillary personnel, such as aides (see Table 3). The school also hosts the district's dual-language education program for students in Grades 1-4. At each of these grade levels, there are two classes in the two-way bilingual program, spending half of each day with instruction in English and the other half with instruction in Spanish, and several other classes in which English is the only language spoken. Students in the dual-language program are bilingual, bicultural, and biliterate. In addition, the school has a new-arrivals program to help ease the transition of students in Grades 3 and 4 whose first language is Spanish. Satisfactory reading progress in the school had not been demonstrated on the statewide mastery test because only 31% of fourth-grade students in this school had achieved the state goal for mastery in reading in the year prior to our intervention study. Sampling Procedures

The first author recruited elementary schools the year before the implementation of the research study. Several principals from school districts in professional development partnerships with university researchers attended a meeting describing this study, and subsequent discussions were held with interested principals about participation and the need for random assignment to treatment and control group. Two principals made a commitment to participate. We held meetings with the district superintendents and teachers of both schools to seek permission for participation and explain the process of random assignment, as well as explaining to all teachers that professional development for implementing the SEM-R would be extended to all teachers at the end of the intervention, ensuring that all control- group teachers could be trained in the intervention. We received permission to conduct the study from the institutional review board of our university; also, all treatment and control-group students were sent a permission form for participation. We randomly assigned all students to treatment or control groups and randomly assigned all teachers to either the SEM-R intervention or to serve as control teachers.

The principals in both elementary schools identified a research liaison to facilitate the implementation of the study. Once this process was complete, researchers observed language arts lessons in a sample of three or four classrooms in each school to collect baseline data about the preexisting reading program and school climate.

The next stage of the reading study included a halfday professional development session for the treatmentgroup teachers, scheduled approximately 2 weeks prior to the start of the intervention. The research team provided teachers with initial training in the SEM-R; teachers met the team's professional development coach who would be working with them throughout the intervention. The teachers received written information about the SEM-R, an initial collection of 75 classroom books at various instructional levels, and a comprehensive reading list specifically developed for the intervention. At the end of this staff development session, classroom teachers began to plan the implementation of the SEM-R in their classrooms. School liaisons worked cooperatively with the research team coaches to help with data collection and implementation of the SEM-R. Research team members were available daily during the intervention to provide support and to monitor both intervention and control classrooms.

Teachers who were randomly assigned to the treatment group used the basal program to teach 1 hr of language arts to students in their general heterogeneous classroom and 1 hr of SEM-R instruction to a different randomly assigned group of students. Schedules were changed because of the random assignment of all Grade 3-5 students in both schools for inclusion in this study. All students remained with their regular classroom teacher for the 1st hr of language arts each day; instruction focused on writing and spelling activities. Then, all students in both groups were reassigned to another classroom to participate in the regular reading program or the SEM- R for the 2nd hr of the literacy block. Teachers randomly assigned to teach the control group continued instruction for 2 hr of the regular language arts program and were trained in the SEM-R at the conclusion of the academic year. The daily class schedules of all teachers and students in both schools changed for 14 weeks.

SEM-R Intervention

The emphasis of the SEM-R is on enjoyment in the process of learning, with a focus on planned, systematic enrichment experiences. In Phase 1 of the SEM-R, teachers selected literature- interspersed with higher order questioning and thinking skills instruction-to read aloud to students. Because the purpose of these "book hooks" was to increase student exposure to literature and build student interest, teachers introduced a new book (or books) each day rather than the more traditional reading of a novel from cover to cover. The initial length of these sessions, 15-20 min, decreased over the course of the intervention, enabling students to spend more time on the other two phases of the SEM-R.

Researchers provided teachers with a collection of 75 high- interest books to begin this Phase 1 read-aloud component. As the intervention continued, all teachers augmented their classroom collections, selecting literature based on the interests and needs of their students. All students and teachers received a set of more than 60 bookmarks that featured higher order questions focusing on skills such as synthesis and evaluation. Teachers also received suggestions for engaging students' interests and exposing them to a variety of literary genres, including mystery, poetry, historical and science fiction, biography, autobiography, and other nonfiction.

Phase 2 of the SEM-R emphasized the development of students' self- regulation skills within supported, independent reading of self- selected books, coupled with individualized, differentiated reading conferences. This phase, which was the longest, directly concerned independent reading. We defined supported independent reading (SIR) as students' reading silently from self-selected literature while teachers coached students to select books that were slightly above their current reading level. The appropriate match was assessed through conferences with each student two to three times each week. Students recorded their daily progress in a reading log featuring the book title and number of minutes spent in reading, and once each week all students wrote a response to a teacher-generated higher order question (often taken from the provided bookmarks). Teachers tracked individual student progress by monitoring the reading logs, reading and responding to students' writing, and keeping records of individual conferences. Initially, teachers and research team members observed that the majority of students selected books that were too easy. Students were told that they could bring easier books home to read but that during class they had to select books that had some challenge, defined as containing unfamiliar words and ideas that made them reflect on the text.

In the beginning of the intervention with SEM-R, our classroom observations indicated that most students in both schools could read appropriately challenging books for 5-15 min a day without losing concentration. Teachers added 1-2 min each day during the SEM-R intervention, extending that time within 3-4 weeks to 35-45-min sessions daily across all treatment classes in both schools. During this in-class reading time, teachers and research-team members circulated to provide individual, differentiated instruction and support. For struggling and below-grade readers, teachers helped to decode difficult words and modeled strategies for inferring challenging vocabulary from context. With advanced readers, teachers discussed higher order themes and asked critical questions focusing on synthesis, evaluation and discussion, and advanced reading strategy use.

In the third component of SEM-R, teachers and team members encouraged students to move from teacher-directed opportunities to self-choice enrichment activities during the intervention. Among the phases, Phase 3 is the most variable in length and content because of the range of literature-based options and the developmental variability across Grades 3-5, resulting in differing needs for structure and scaffolding within and between classrooms. Activities included (but were not limited to) (a) exploring new technology and engaging in discussion groups, (b) creative writing, (c) buddy reading, (d) creativity training in language arts, (e) learning centers, (f) interest-based projects, (g) continuation of self- selected reading, and (h) book discussion groups. The intent of these experiences was to provide time for developing and exploring student interest in reading. In addition, students engaged in creative and critical thinking training and in Internet training to locate information about various literary genres, such as biography and autobiography. Teachers and team members also focused on enabling students to learn to read critically and to locate enjoyable and challenging reading materials so that they might develop additional positive associations with texts and reading (Alexander & Wade, 2000). Options for independent study were also available for students during Phase 3. We developed each component of the SEM-R to help students to increase their enjoyment of reading and we coached differentiated reading strategies and efforts to increase self-regulation in reading.

Implementation of the SEM-R Intervention

In both schools, a half-day of staff development was used to train teachers, classroom aides, and administrators on the use of the SEM-R in the classroom. The researchers presented the background and purpose of the study and an introduction to the SEM-R, focusing on a thorough explanation of each of the three phases. Before the training, all intervention teachers and administrators received an informational prereading packet, including a brief overview of the SEM, an introduction to the SEM-R, and a journal article about the NRP's (2000) recommendations for reading instruction. When teachers (including special education teachers) arrived at the training, they received an implementation manual including detailed information and suggestions for all aspects of the SEM-R. To build enthusiasm and supplement classroom libraries, the research team gave each SEM-R teacher 75 high-interest books selected to challenge students reading across several grade levels in their classes. Throughout the intervention, team members gave approximately 200 additional books to the school libraries and classrooms to augment the selection of nonfiction materials and reading selections for students interested in reading a book about a particular topic. Teachers also received books on tape and compact disk (CD) and portable CD players to augment the tape and CD players that they had in their classrooms. To help treatment teachers implement the SEM-R and maintain treatment fidelity, research team members were in the schools daily during the 14-week intervention to monitor treatment, observe control-group classrooms, and discuss questions and concerns about the SEM-R. Over the course of the intervention, from the beginning of February and continuing for 14 weeks, the length and activities of the three phases of SEM-R varied to meet the needs of teachers and students and to follow the suggested delivery of the SEM-R. The approximate time of Phase 1 varied but averaged between 10-25 min daily from the beginning of the intervention to the end. However, as the students' ability to maintain focus on SIR increased, teachers shortened Phase 1 to accommodate a desire for more independent reading time. In Phase 2, all the treatment classrooms began with 10- 20 min of SIR time per class, but by the end of the intervention, the majority of the students in the SEM-R sustained increased amounts of independent reading, averaging 35-40 min each day. As the amount of SIR time increased, teachers were able to have conferences more frequently with students, especially struggling readers who needed more coaching in reading skills.

In most intervention classrooms, the special education teacher or a classroom aide provided support for learningdisabled students and second-language learners included in the regular reading class. Special education teachers and classroom aides were available in approximately the same number of control classrooms as treatment classrooms because of the random distribution of the identified special education population between treatment and control groups. All professionals and paraprofessionals working in the treatment classrooms received materials and training in the use of SEM-R techniques, with a special emphasis on conferencing strategies.

During the first week of implementation, Phase 3 activities included the introduction of creativity training activities, literary thinking skills, the study of genres, and Webbased training. As students became more comfortable with independent work, Phase 3 activities expanded to involve more independent opportunities, including book discussion groups and explorations of brief biographies. By the 9th week of the intervention, almost all the intervention teachers at Meadowbrook adopted a schedule that included 4 days of Phases 1 and 2 activities with lengthy SIR and 1 day per week devoted solely to Phase 3 activities, allowing students to work in-depth on their projects. At Northwest, teachers provided 10 min per day for Phase 3 activities so that students could engage in SIR every day. In both schools, the research team frequently observed the treatment teachers implementing the SEM-R, usually 2 or 3 times each week. Each teacher had to complete a weekly plan documenting the activities and progress made in each phase of SEM- R.

Instrumentation

Iowa Tests of Basic Skills (ITBS). We measured reading comprehension with the ITBS reading comprehension subtest (Form A). The ITBS measures achievement in 15 subject areas for students in Grades K-8. The reading subtests of the ITBS "measure how students derive meaning from what they read" (Hoover et al., 2003, p. 32). The reading comprehension subtest consists of a variety of reading passages representing narrative, poetry, and nonfiction material from science and social studies. After students read each passage, comprehension is assessed through the use of four to seven multiple- choice questions that ask students to recall facts, make generalizations, and draw inferences (Hoover et al.).

Reading teachers administered the ITBS to all the students on 2 consecutive days during the last week of the SEM-R study as a postassessment only, because of the random assignment of students to reading groups and the time necessary to administer the test during a relatively short intervention period. Kuder-Richardson reliability coefficients for the ITBS reading comprehension subtests (Form A, 2001, spring norms) ranged from .91 for students in Grade 3 to .90 for students in Grades 4 and 5. The ITBS has been a standard measure of reading comprehension in many research studies reviewed in this article. Ceiling effects were not an issue, and the ITBS was sensitive enough to measure changes in reading comprehension. This is a limitation of the study, however, because reading comprehension may be defined more broadly than that measured by the ITBS.

Elementary Reading Attitude Survey (ERAS). Students' attitudes toward or feelings about reading influence their willingness to actively participate in activities such as school that involve reading (Wigfield, 1997). We assessed students' preattitudes and postattitudes toward reading with the 20-item ERAS developed by McKenna and Kear (1990), who asserted that the survey can "serve as a means of monitoring the attitudinal impact of instructional programs" (p. 627). Of the ERAS items, 10 measure recreational reading and 10 measure academic reading. The scores on the recreational and academic subscales can be combined to create a composite measure of reading attitudes (McKenna & Kear). The ERAS uses a 4-point, Likert-type scale with pictorial anchors of a smiling and scowling Garfield cat. McKenna and Kear's psychometric evaluation of the ERAS indicated satisfactory Cronbach's alpha internal reliability consistency coefficients of .88 for students in Grade 3 and .89 for students in Grades 4-6. The reported estimates for Cronbach's alpha internal reliability consistency coefficients ranged from .80 to .87 for the recreational subscale in Grades 3-6, and from .81 to .83 for the academic subscale in Grades 3-6. McKenna and Kear also conducted a factor analysis of the instrument and collected evidence of criterion-related validity. Kazelskis, Thames, and Reeves (2004) conducted a study of the factorial invariance of the ERAS across gender and race, using confirmatory factor analysis. They found that the instrument displayed full factorial invariance across gender and partial factorial invariance across race.

We also examined the internal consistency and test- retest reliability estimates for the present sample. Cronbach's alpha internal consistency reliability estimates for the total scale were .90 at pretest and .90 at posttest. The correlation between the pretest score and the posttest score was .64.

ORF. Measures of ORF assess the speed, accuracy, and efficiency with which a student reads a particular text. Current research indicates that "fluency is a critical component of skilled reading" (NRP, 2000, p. 1). We assessed prereading and postreading fluency with procedures described by Hasbrouck and Tindal (2005). We selected reading passages to represent all grade levels (3-5) in the study. To facilitate comparisons across grade levels, all students read from the three increasingly difficult, 250-word passages for three separate 1-min reading trials. The research team recorded the number of words read correctly for each passage and calculated and recorded a mean ORF score for each student. Because ORF reflects the complex orchestration of lower and higher level processes, researchers can consider it an elegant and reliable indicator of overall reading proficiency (Fuchs et al., 2001). Test-retest and alternate-form reliability of ORF measures are consistently above .90, in general, and criterion-related validity with other standardized measures of reading decoding and comprehension average .80 or higher (Hasbrouck & Tindal).

A team of researchers, including professors and graduate students who did not know whether students were in the treatment or control group, administered the prereading and postreading fluency assessments in each school media center. To ensure reliability of scores, each test administrator participated in a 30-min training session conducted by one of the primary investigators of this study. On the day of the assessment, the research team reviewed the procedures and observed each test administrator to ensure that the assessments were standardized.

Study implementation and treatment fidelity. The length of the intervention at both schools extended from February to May, providing students assigned to the treatment group with 1 hr of instruction in SEM-R each day for 14 weeks. At least two of the research team members regularly monitored treatment fidelity in all classrooms. For the first 6 weeks of the intervention, observations and instructional support occurred almost daily in all treatment classes; control classes were observed at least twice each week. As the intervention progressed and teachers' levels of comfort and efficacy increased, the number of observations in treatment classrooms decreased to three times each week. Researchers summarized field notes after each day and integrated the summaries into weekly written reports for both treatment and control classes. During observations, researchers took detailed notes on the specific features of each phase of the SEM-R, guided by the items summarized in the protocol in Table 4. In Phase 1, field notes detailed (a) the type of read-aloud used, (b) teachers' use of planning to read from different genres of books, and (c) the integration of reading strategies or higher level thinking skills that accompanied the read- aloud. In Phase 2, field notes documented (a) amount of SIR, (b) classroom environment, (c) completion of student reading logs, and (d) use of conferences and differentiated reading strategies. In Phase 3, field notes showed (a) time available for independent and small-group choice activities and (b) types of choices that students made for this activity time each day. Researchers also recorded approximate time spent in the three phases. Careful systematic analysis of the protocol in field notes indicated that all treatment teachers at both schools implemented the SEM-R with moderate to good treatment fidelity. An analysis of field notes, documented teacher lesson plans, and teacher interviews showed that treatment teachers expressed confidence in their ability to implement the program and perceived that their students were reading more and benefiting from differentiated reading skill instruction. The two observers met weekly to document treatment fidelity according to their notes; they agreed on high-, medium-, and low-treatment fidelity 90% of the time. Accordingly, interrater reliability was judged to be high. In future research on this approach, we plan to expand the protocol summarized in Table 4 to include quantitative measures of interrater reliability.

An analysis of field notes for control classes confirmed that instruction was based on the basal reading program used at both schools. A consistent set of activities occurred each day, including whole-group instruction, small-group reading instruction, and 3-5 min of reading, followed by workbook and other activities. Regular, repeated observations of both treatment and control classes were necessary to ensure that the process maintained the fidelity of the SEM-R throughout the intervention and to carefully monitor the instruction in control classrooms to ensure that it was based on the basal reading program and on whole- and small-group instruction.

Existing reading program in the control classrooms. Observations conducted in both schools before and during implementation of the SEM-R showed an organized literacy program to improve students' reading achievement in place in both schools. During the year before the intervention, the faculty at Meadowbrook began using the Houghton-Mifflin reading program with all students. The faculty purchased supplementary trade books and all the teaching materials to accompany the basal reading program. Also, all teachers participated in intensive training in how to use the program. As a supplement to the basal reading program, school personnel had purchased a large selection of high-interest level books for classroom use and stored them in the reading specialist's office. However, the reading consultant reported that these books were rarely requested or used by classroom teachers.

During the intervention, students and teachers who were assigned to the control group continued with the reading curriculum already in place. In both schools, in addition to the varied activities from the basal reading series, students participated in a wide variety of learning activities, including vocabulary games and lessons drawn from class reading assignments, literature circles, class novel studies, journal writing, author studies, oral presentations, and infrequent teacher read-alouds.

Observations in Northwest prior to the implementation of SEM-R suggested that teachers were using a basal language arts program despite low scores that had members of the community calling for the adoption of a more direct instructional program in reading. The main emphasis of the language arts program was the integration of these basal materials with whole-language strategies. Students who failed to meet mastery on the state test were offered remedial programs during school, after school, and in the summer. Observations in classrooms before and during implementation of SEM-R suggested that teachers used a variety of instructional approaches at Northwest. These approaches consisted of a combination of daily reading groups and small-group and whole-group instruction in the reading groups, weekly writing assignments, discussion, and basal reading activities. Learning environments varied among classrooms; each had at least one bookcase filled with popular paperbacks and one to three computers with Internet access. Inspirational posters, educational bulletin boards, art and drawing supplies, and Spanish vocabulary words were also in every room. Before the SEM-R intervention, students remained in their regular heterogeneously grouped classroom for a 2-hr reading and language arts instruction block, but the experimental nature of the study required students to change classes for 1 hr of the block each day.

Results

To investigate differences in reading fluency, comprehension, and attitudes across instructional groups at Northwest and Meadowbrook, we conducted a series of three 2-level multilevel models using hierarchical linear modeling (HLM) version 6.04 (Raudenbush, Bryk, Cheong, Congdon, & du Toit, 2004). The dependent variables of interest were student postassessment scores on ORF measures, posttest scores on the reading comprehension section of the ITBS, and postassessment scores on the ERAS. We used preassessment reading fluency scores as a covariate for the analyses of reading fluency and reading comprehension and entered pretest ERAS scores as a covariate in the postreading attitudes analysis. Table 5 contains descriptive statistics for the pretest and posttest measures used in this study.

Level 1 contained student-level information; Level 2 contained classroom and school information. Because we focused on only two schools, we treated school as a fixed effect and entered it at Level 2 of the model. We entered prereading fluency, a student level covariate, at Level 1. The independent variable of greatest interest, exposure to SEM-R instruction (SEM-R), included two conditions- the SEM-R treatment group (coded 1) and control (coded 0)-and this variable, condition, was entered at Level 2. In addition, school was a Level 2 variable and coded as Northwest = 0, Meadowbrook = 1. Last, we modeled the interaction between school and treatment group by creating the cross-product of the two Level 2 variables, SEM-R and school. Given the small Level 2 sample size, we used restricted maximum likelihood estimation for all HLM analyses (Raudenbush & Bryk, 2002).

Standard assumptions of HLM included the normality of Level 1 residuals and the homogeneity of Level 1 variances; therefore, we empirically examined these model assumptions. Because the assumption of homogeneity of variance is powerful and sensitive to even slight departures from normality, we set the alpha level of this test to .02. We also conducted basic residual analyses, including the examination of the normality of the Level 1 residuals. Fluency and comprehension were normally distributed (skewness and kurtosis values were less than .25 for each of these variables) and had normally distributed Level 1 residuals. However, posttest attitude scores were slightly negatively skewed (skewness = -.54), as were the Level 1 residuals. Fluency and comprehension exhibited relatively homogeneous Level 1 variance (p > .02); posttest attitudes exhibited heterogeneous Level 1 variances (p < .001). The attitudes scale was somewhat skewed. This departure from normality seemed a plausible explanation for the rejection of the assumption of homogeneity of variances. Because we had a small number of Level 2 units, and explicitly modeling the heterogeneity of variances requires using full information maximum likelihood estimation techniques, which provide biased estimates of variance components when the number of Level 2 units is small (Raudenbush & Bryk, 2002), we did not explicitly model the heterogeneity of variances in the posttest attitudes scale. Instead, we used the robust standard errors for the attitudes analyses. These standard errors are considered be more robust to violations of normality and homogeneity than are the conventional standard errors.

Reading Fluency

First, to estimate the intraclass correlation (ICC), we estimated a baseline model with no predictors at Level 1 or Level 2. The ICC, which measures proportion of variance at the school level in relation to the total variance, is the between-class variance (tau) divided by the total variance (sigma^sup 2^ + tau). The ICC for reading fluency was .212. In other words, 21.2% of the variance in reading fluency lay between classes. Table 6 shows the results of this baseline model.

Second, we estimated a Level 1 model that included prereading fluency as a predictor of postreading fluency but did not include any predictors at Level 2. The addition of prereading fluency as a Level 1 covariate resulted in a large reduction in variance at Levels 1 and 2. Prereading fluency accounted for close to 89% of the residual variance at Level 1 and more than 99.5% of the residual variance at Level 2. After we controlled pretest reading fluency, only 0.4% of the variance in postintervention reading fluency lay between classes. After we accounted for prefluency, significant variability remained to be explained in the prefluency slope, chi^sup 2^(30) = 45, p = .038, but not in the intercept, chi^sup 2^(30) = 39.19, p = .12 These results demonstrate that prereading fluency is a powerful covariate, explaining most of the variability in postfluency scores.

Last, we estimated the full Level 2 model, which included prereading fluency at Level 1 and school, SEM-R, and the SEM-R School interaction at Level 2. Because of the coding system, the intercept (gamma^sub 00^ = 116.78) represented the predicted postfluency score for a student at Northwest (school = 0) who was in the control group (SEM-R = 0) and who scored at the grand mean on prereading fluency (grand M centered prefluency = 0). The coefficient for school (gamma^sub 01^ = 7.04) represented the differential between control scores at Meadowbrook and control scores at Northwest when prefluency = 0. Therefore, the average score for a control student of average prefluency at Meadowbrook was 123.82 (116.78 + 7.04). The coefficient for the SEM-R (gamma^sub 02^ = 9.79) represented the differential for a student of average prefluency who received the treatment at Northwest (school = 0). Therefore, the predicted fluency score for a student of average fluency in the SEM-R group at Northwest was 126.57 (116.78 + 9.79). The coefficient for the interaction term (gamma^sub 03^ = -8.78) represented the differential effect of the SEM-R treatment at Meadowbrook (as compared with Northwest). Therefore, the predicted score for a student of average prefluency who received SEM-R at Meadowbrook was 116.78 + 7.04 + 9.79 - 8.78 = 124.83. After we controlled prefluency scores, the main effect of school was statistically significant, as Meadowbrook outperformed Northwest. Also, the main effect of SEM-R was statistically significant. On average, SEM-R students outperformed control students. However, the interaction between SEM-R and school was also statistically significant, suggesting that the SEM-R treatment had a much larger positive impact at Northwest and only a small positive impact at Meadowbrook. Although the predicted differential between SEM-R and control students at Northwest was large (9.79 points), the differential was far more modest at Meadowbrook School (1.01 points). After we controlled prefluency, the model-predicted difference between the SEM-R and control groups at Northwest was approximately 10 points, or 0.25 SDs. In contrast, the model- predicted difference between the SEM-R and control groups at Meadowbrook was only 1 point, or less than 0.03 SDs after we controlled prefluency. Overall, although the SEM-R treatment had a positive effect on reading fluency at Northwest, the effect of the SEM-R treatment on reading fluency at Meadowbrook was more ambiguous.

We also modeled the impact of the Level 2 variables: school, SEM- R, and School SEM-R on the prefluency slope. The intercept for prefluency (gamma^sub 10^ = 0.90) represented model-predicted slope for prefluency in the control condition (SEM-R = 0) at Northwest. For control students at Northwest, for every point increase in pretest reading fluency, we predicted that posttest reading fluency would increase 0.90 points. The coefficient for school (gamma^sub 11^ = 0.15) represented the differential between the prefluency slope for Meadowbrook control students and the prefluency slope for Northwest students. Therefore, the model-predicted prefluency slope for a control student at Meadowbrook was 1.05 (0.90 + 0.15). For every point increase in prefluency for control students at Meadowbrook, there was a corresponding predicted increase in postfluency of 1.05 points. This means that the relation between prefluency and posttest fluency was stronger for control students at Meadowbrook than it was for control students at Northwest.

The coefficient for the SEM-R (gamma^sub 12^ = 0.19) represented predicted differential slope effect for the SEM-R group at Northwest (school = 0). The prefluency slope in the SEM-R group at Northwest was 1.09 (0.90 + 0.19). For every point increase in prefluency for treatment students at Northwest, there was a corresponding predicted increase in postfluency of 1.09 points. The coefficient for the interaction term (gamma^sub 13^ = -0.15) represented the differential slope effect for SEM-R students at Meadowbrook (compared with Northwest). Therefore, the predicted prefluency slope in the treatment group at Meadowbrook equaled 0.90 + 0.15 + 0.19 - 0.15 = 1.09. For every point increase in prefluency for SEM-R students at Meadowbrook, there was a corresponding predicted increase in postfluency of 1.09 points.

The Level 2 model resulted in an 85.6% reduction in unexplained variance at this level. After accounting for the variables at Levels 1 and 2, no significant variability remained to be explained in the prefluency slope, chi^sup 2^(27) = 35.27, p = .13, or the intercept, chi^sup 2^(27) = 17.14, p > .50.

Reading Comprehension (ITBS)

We estimated the impact of school, treatment, and the interaction between treatment and school on reading comprehension scores, using reading fluency scores at pretest as a covariate. First, we estimated the ICC using the baseline model with no predictors at Level 1 or 2. The ICC for reading comprehension was .268. In other words, 26.8% of the variance in reading comprehension lay between classes. Table 7 shows the results of this baseline model.

Second, we estimated a Level 1 model that included prereading fluency as a predictor of postreading comprehension, but did not include any predictors at Level 2. The addition of prereading fluency as a Level 1 covariate resulted in a moderate reduction in variance at Levels 1 and 2. Prereading fluency accounted for almost 28% of the within-classroom residual variance and approximately 67.5% of the between-classrooms residual variance. After accounting for prefluency, significant variability remained to be explained in the ITBS intercept, chi^sup 2^(30) = 105.85, p < .0001, but not in the prefluency slope, chi^sup 2^(30) = 33.79, p = .29. Using the deviances of two nested models, we compared the fit of the Level 1 model with a random effect for the prefluency slope (deviance = 4,903.79 with four random effects) to a more parsimonious model that fixed the prefluency slope (deviance = 4,903.95 with two random effects). The chi-square difference test that compared the variance covariance components was statistically nonsignificant, chi^sup 2^(2) = .17, p > .50. The more parsimonious model provided no worse a fit than did the more parameterized model, so we favored the more parsimonious model. Thus, for the Level 2 model, we fixed the prefluency slope. Furthermore, we estimated the effects of the Level 2 variables on the intercept but not on the Level 1 slope.

Last, the full Level 2 model included prereading fluency at Level 1 and school, SEM-R, and the SEM-R School interaction at Level 2. We used the Level 2 variables as predictors of the reading comprehension intercept but not the prefluency slope. Because of the coding system, the intercept (gamma^sub 00^ = 201.10) represented the predicted posttest reading comprehension score for a student at Northwest (school = 0) who was in the control group (SEM-R = 0) and who scored at the grand mean on prereading fluency (grand-mean- centered prefluency = 0). The coefficient for school (gamma^sub 01^ = 5.39) represented the differential between control scores at Meadowbrook and control scores at Northwest when prefluency = 0. Therefore, the average score for a control student of average prefluency at Meadowbrook was 206.49 (201.10 + 5.39).

The coefficient for the SEM-R (gamma^sub 02^ = 7.44) represented the differential for a student of average prefluency who received the SEM-R treatment at Northwest (school = 0). The predicted fluency score for a student of average fluency in the SEM-R group at Northwest was 208.54 (201.10 + 7.44). The coefficient for the interaction term (gamma^sub 03^ = -7.61) represented the differential effect of SEM-R at Meadowbrook (compared with Northwest). Therefore, the predicted score for a student of average prefluency who received SEM-R instruction at Meadowbrook was 201.10 + 5.39 + 7.44 - 7.61, or 206.32. After we controlled prefluency scores, none of the Level 2 variables were statistically significant. However, the reading comprehension data followed the same trend as did the reading fluency data. Although the predicted differential between treatment and control students was more pronounced at Northwest (7.44 points), the differential was essentially zero at Meadowbrook (-0.18 points).

The Level 2 model did not reduce the unexplained between- classrooms variance when compared with the model that included only the Level 1 variable (prefluency). After accounting for the variables at Levels 1 and 2, significant variability remained to be explained in the intercept, chi^sup 2^(30) = 115.22, p < .001.

Reading Attitudes (ERAS)

We estimated the impact of school, SEM-R treatment, and the interaction between treatment and school on posttest reading attitudes, using pretest reading attitude as a covariate. First, we estimated the ICC by using a baseline model with no predictors at Levels 1 or 2. The ICC for reading attitudes was .088. In other words, 8.8% of the variance in reading attitudes lay between classes. This percentage is much lower than that of variance in reading achievement measures (reading fluency and comprehension) that lay between schools. Table 8 shows the results of this baseline model.

Second, we estimated a Level 1 model that included prereading attitudes as a predictor of postreading attitudes but not predictors at Level 2. The addition of pretest reading attitudes as a Level 1 covariate resulted in a moderate reduction in variance at Levels 1 and 2. Pretest reading attitudes accounted for approximately 36% of the withinclassroom residual variance and 51% of the betweenclassroom residual variance. After accounting for pretest reading attitudes, significant variability remained to be explained in the ERAS intercept, chi^sup 2^(30), = 66.48, p < .001, and in the prefluency slope, chi^sup 2^(30) = 44.89, p = .04.

Last, the full Level 2 model included prereading fluency at Level 1 and school, SEM-R, and the SEM-R School interaction at Level 2. We used the Level 2 variables as predictors of the reading comprehension intercept, as well as the pretest reading attitudes slope. With the coding system, the intercept (gamma^sub 00^ = 60.59) represented the predicted posttest reading comprehension score for a student at Northwest (school = 0) who was in the control group (SEM- R = 0) and who scored at the grand mean on pretest reading attitudes (grand-M-centered preattitudes = 0). The coefficient for school (gamma^sub 01^ = -1.78) represented the differential between control scores at Meadowbrook and control scores at Northwest when preattitudes = 0. Therefore, the average score for a control student of average preattitudes at Meadowbrook was 58.81 (60.59 - 1.78). The coefficient for the SEM-R (gamma^sub 02^ = 0.83) represented the differential for a student of average preattitudes who received the treatment at Northwest (school = 0). The predicted postattitudes score for a student of average preattitudes in the SEM-R group at Northwest was 61.42 (60.59 + 0.83). The coefficient for the interaction term (gamma^sub 03^ = -1.43) represented the differential effect of the SEM-R treatment at Meadowbrook (as compared with Northwest). The predicted score for a student of average attitudes who was in the SEM-R treatment group at Meadowbrook was 60.59 - 1.78 + 0.83 - 1.43, or 58.21. After we controlled pretest ERAS scores, none of the Level 2 variables was statistically significant. We also modeled the impact of the Level 2 variables, school, SEM-R, and School SEM-R on the pretest attitudes slope. The intercept for preattitudes (gamma^sub 10^ = 0.75) represented model-predicted slope for pretest attitudes in the control condition (SEM-R = 0) at Northwest. For control students at Northwest, we predicted that for every point increase in pretest reading fluency, posttest reading fluency would increase 0.75 points. The coefficient for school (gamma^sub 11^ = -0.13) represented the differential between the pretest attitudes slope for Meadowbrook control students and the pretest attitudes slope for Northwest students. The model-predicted pretest attitudes slope for a control student at Meadowbrook was 0.62 (0.75 - 0.13). For every point increase in pretest attitudes for control students at Meadowbrook, a corresponding predicted increase of 0.62 points occurred in postfluency. The coefficient for SEM-R (gamma^sub 12^ = - 0.05) represented the differential in the pretest attitudes slope for a student who received the treatment at Northwest (school = 0). The pretest attitudes slope in the SEM-R group at Northwest was 0.65 (0.75 - 0.05). The coefficient for the interaction term (gamma^sub 13^ = -0.06) represented the differential slope effect for SEM-R treatment students at Meadowbrook (as compared with Northwest). The predicted prefluency slope in the treatment group at Meadowbrook was 0.75 - 0.13 - 0.05 - 0.06, or 0.51. However, none of these cross- level interaction terms were statistically significant.

The Level 2 model resulted in a small reduction in the unexplained between-classes variance in the intercept (2.3%) and no reduction in the unexplained betweenclasses variance in the pretest attitudes slope. After we accounted for the variables at Levels 1 and 2, statistically significant variability remained to be explained in the prefluency slope, chi^sup 2^(27) = 61.21, p < .001, and the intercept, chi^sup 2^(27) = 40.79, p = .04.

Discussion

Debate and considerable research continue to be associated with reading instruction and the development of literacy in children of diverse achievement in urban and suburban schools. We did not discover any prior research on the effects of an enriched reading program that focused on interest-based reading and differentiated instruction for a block of time each day. Furthermore, we found no studies that used experimental research methods to study enriched reading instruction and daily independent reading for urban and suburban students who achieve at differing levels.

We examined different instructional practices affecting reading fluency and achievement. After 14 weeks, the SEM-R treatment group scored statistically significantly higher than did the control group in reading fluency; however, this impact was much stronger at Northwest than it was at Meadowbrook, perhaps suggesting that an enrichment approach to reading was more effective in an urban school than a suburban school. These results indicate that a student- centered, enrichment-based reading program with independent reading was at least as effective as the use of the basal program alone in both schools at significantly increasing reading fluency for students. There were no significant differences in reading comprehension or attitudes toward reading during the 14-week intervention, suggesting that a good deal of grouped instruction can be eliminated without any negative effects to students' reading comprehension scores and without altering their attitudes toward reading.

The NRP (2000) concluded that insufficient empirical evidence exists to determine whether encouraging students to read for longer periods results in increased reading fluency. A major component of the SEM-R intervention was providing students with opportunities to engage in extended periods of structured silent reading of self- selected challenging books, accompanied by supported, individualized reading instruction. The findings of this study provide experimental evidence that enriched independent reading may be a promising way to increase reading fluency.

Prefluency was a strong covariate for postfluency and explained most of the between-classes variability in postfluency assessments. Northwest, the urban school, had lower fluency scores overall, but the treatment effect was stronger at Northwest. Thus, focusing on enriched independent reading of choice with an emphasis on finding enjoyable reading experiences for students might have had more of an impact on urban students than on suburban students, who had more opportunities for enjoyable reading. This effect size, although small (0.25 SDs), was achieved after only 14 weeks. Future research will address the implementation of the SEM-R approach for a longer period of time. Many school reform reading models suggest direct instruction for struggling urban readers, yet the present research shows that gains in fluency can occur when students select books to read independently each day for 25-35 min with a focus on enjoying what they are reading. Research about culturally diverse, urban students such as those who attend Northwest suggests the importance of any research showing differences in either reading fluency or comprehension in reading. A r

Source: Journal of Educational Research, The

More News in this Category