The Literacy Trust, Inc. 

Reading Rescue: An Effective Tutoring Intervention Model

for First-Grade Struggling Readers

Linnea C. Ehri, Lois G. Dreyer, Bert Flugman, and Alan Gross

Graduate Center, City University of New York

Author Note

            Linnea C. Ehri, Department of Educational Psychology, Graduate Center of the City University of New York; Lois G. Dreyer, Center for Advanced Study in Education, Graduate Center of the City University of New York; Bert Flugman, Center for Advanced Study in Education, Graduate Center of the City University of New York; Alan Gross, Department of Educational Psychology, Graduate Center of the City University of New York, 365 Fifth Avenue, New York, NY 10016

                  Reading Rescue was developed in cooperation with the University of Florida and is sponsored by a charitable, not-for-profit organization, The Literacy Trust, Inc. This study was an independent evaluation conducted with grants from The Literacy Trust, Inc. and Pearson Learning. All aspects of the research were under the control of investigators who were neither employed by nor affiliated with the funding agencies. We express gratitude to the following colleagues for their assistance: Nora Hoover, Kathy Kaufman, Alba Langenthal, Tara Mastrorilli, Belinda Nix, Henry Park, and Benedict Silverman. We thank the district and school personnel including principals, teachers and paraprofessionals for their cooperation and assistance in the conduct of our study. For more detailed information about the Reading Rescue program, contact The Literacy Trust, www.literacytrust.org.

Correspondence concerning this article should be addressed to Linnea C. Ehri, Program in Educational Psychology, CUNY Graduate Center, 365 Fifth Avenue, New York, NY 10016. E-mail: LEhri@gc.cuny.edu. Phone: 212-817-8294. Fax: 212-817-1516.           

Abstract

A tutoring intervention model (RES) was investigated with 64 low SES first grade students at risk for reading difficulties. School staff including reading specialists, credentialed teachers, and paraprofessionals, provided a mean of 50 tutoring sessions to teach phonological awareness, systematic phonics, vocabulary, fluency, and reading comprehension. Tutored students made significantly greater gains in word reading and text comprehension than controls who received a small group intervention or no intervention. In fact, the majority of tutored students reached normal reading levels. Paraprofessionals tutored as well as the other tutors except in skills benefiting nonword decoding. Also they were less efficient than reading specialists. Contrary to conventional wisdom advocating text reading practice at an instructional level, regression analyses indicated greater benefits at an independent level.

Reading Rescue: An Effective Tutoring Intervention Model

for First-Grade Struggling Readers

            Currently there is much interest in developing effective ways to intervene early with young children who are at risk of difficulty in learning to read. Studies indicate that children who fail to acquire adequate reading skills in first grade often continue to have difficulties and may never catch up (Foorman, 1997; Juel, 1988; Stanovich, 1986). Children who are at risk can be identified early (Vellutino et al.,1996). Results of the National Reading Panel Report (2000) showed that early intervention was more effective than later remediation.  There is much interest in evaluating intervention models to provide schools with evidence as the basis for choosing programs to teach reading. In fact, some government initiatives have made funding contingent upon schools’ choosing reading programs that are evidence based. One purpose of the present study was to obtain evidence regarding the effectiveness of Reading Rescue, a comprehensive tutoring intervention model with struggling readers in first grade.

            The best known tutoring intervention model aimed at struggling readers in the early grades is Reading Recovery developed by Clay (1985, 1993). Various reviews of the effectiveness of Reading Recovery have been conducted (e.g., Shanahan & Barr, 1995; Elbaum, Vaughn, Hugh, & Moody, 2000). In the majority of studies, students who received this tutoring made greater gains in reading achievement during first grade than students who were not tutored. Studies have also shown that if Reading Recovery instruction is enriched with more phonemic awareness or systematic phonics instruction, its impact on reading is enhanced (Greaney, Tunmer & Chapman, 1997; Hatcher, Hulme, & Ellis, 1994; Iversen & Tunmer, 1993; Santa & Hoien, 1999). A question of interest in the present study was whether the Reading Rescue model would prove effective as well, and whether this model would benefit students in high poverty schools from families whose first language is not English.

            In developing effective reading intervention models, several features require specification including the following: the reading instructors, their qualifications; and the specific training they need to provide instruction; the instructional materials and procedures including the texts to be read; the students targeted for the intervention, their entering reading capabilities, and the match between these capabilities and the instruction provided; the observational tools and tests to monitor students’ progress and make instructional decisions; the costs to schools.

            The Reading Rescue model was implemented with the following features in the present study. The instructors were personnel already employed by schools but not necessarily credentialed teachers. Paraprofessionals as well as reading specialists and credentialed teachers received on-site training that included workshops, supervision and coaching. The students were low SES first graders reading near the bottom of their class as indicated by an inability to read primer level text and low scores on other literacy tests. However, students needed to have substantial letter name knowledge (i.e., at least 17 letters) considered a prerequisite for making progress in the program. This is because the program focused on teaching letter-sound correspondences and their use to read and write.

            The instruction involved one-to-one tutoring for 30 minutes daily or as frequently as possible during a week. It was comprehensive in covering all five components identified by the National Reading Panel (2000) as evidence based and essential for learning to read (i.e., phonemic awareness, systematic phonics, fluency, vocabulary, and reading comprehension). The instructional materials and procedures engaged students in word study and sentence writing that taught phonemic awareness, letter-sound relations, decoding, reading high frequency words, and spelling. Students reread little books to acquire oral text reading strategies, sight words, vocabulary, fluency, and comprehension. The texts were leveled, they contained both high frequency and decodable words, and they were tied to a scope and sequence chart used for phonics instruction. Observational tools and informal tests were used. Tutors kept records of students’ performance during each session, and they tested students to determine when they were ready to progress to higher-level lessons and texts. The costs were relatively modest because existing school staff members rather than outside teachers provided the tutoring.

            Several features of the Reading Rescue intervention model were expected to contribute to its effectiveness as shown by previous research. Some of the activities were drawn from the Reading Recovery model, such as sentence writing and the rereading of little books.  The inclusion of systematic, sequential phonics instruction was expected to enhance its impact (Ehri, Nunes, Stahl, & Willows, 2001). The use of many decodable books at various reading levels was expected to help students apply the letter-sound correspondences and decoding skills they had been taught (Juel & Roper-Schneider, 1985; Torgesen, et al., 1999). The inclusion of vocabulary instruction was expected to make the program especially appropriate for low SES students from bilingual families (Biemiller, 1999). Questioning students in various ways about the books they read and distinguishing between text explicit and text implicit questions were expected to foster text comprehension (Block & Pressley, 2002).

            There are many reasons why tutoring in first grade has been especially attractive to educators as a means of preventing early reading failure (Clay, 1985). Instruction can be tailored to the difficulties of individual students. Tutors can easily monitor students’ progress and the need for review. Compared to group instruction, tutoring allows more time for students to read, and their errors and misconceptions receive immediate feedback. Also, having the tutor’s undivided attention is motivating.

            Although one-to-one tutoring has many advantages over small group instruction for teaching reading, not all studies have reported positive results. Matthes & Fuchs (1994) detected a wide range of effect sizes across studies. In a meta-analysis evaluating phonemic awareness instruction, Ehri, Nunes, Willows, Schuster, Yaghoub-Zadeh, & Shanahan (2001) found significantly smaller effect sizes for tutoring than for small groups. In a meta-analysis of systematic phonics instruction, Ehri, et al. (2001) found that effect sizes for small groups did not differ significantly from effect sizes for individual tutoring. In their meta-analysis of tutoring studies with struggling readers, Elbaum et al. (2000) also found no difference. One purpose of the present study was to compare the effects of tutoring to the effects of small group instruction.

            Another issue addressed here was whether paraprofessionals are as effective at tutoring struggling readers as reading specialists and other credentialed teachers. The use of paraprofessionals as instructors is especially valuable in schools serving low SES populations. Such schools may have larger numbers of children reading below grade level, and their limited budgets may preclude hiring enough certified reading specialists to work with all the students who are needy. If trained paraprofessionals can deliver effective supplementary reading instruction, this provides a more affordable means of addressing the need.

             Findings regarding the effectiveness of paras, however, are mixed. Wasik and Slavin (1993) reviewed studies of programs that provided tutoring for struggling readers in first grade, including Reading Recovery and Success for All. All programs yielded significant positive effects, but those utilizing certified teachers showed larger effects than those using paraprofessionals. However, paras and certified teachers were not compared teaching the same programs. Moreover, the programs used by paras were more focused and less comprehensive in the reading components taught than the programs used by teachers. Thus, these findings are not definitive.

            In a recent study, Brown, Morris and Fields (2005) compared the effectiveness of certified teachers and paraprofessionals. Both used the same reading program, Next Steps (Morris, 1999), to tutor second and third grade struggling readers. The program included guided oral reading of instructional level text, word study, fluency training through rereadings of text, and listening comprehension. Results revealed that students tutored by paraprofessionals read as well as students tutored by certified teachers on most measures of reading.

            Although Brown et al. (2005) documented the effectiveness of paras with older students, it is not clear whether the same results would hold for first graders. In the Wasik and Slavin (1993) review, paras were less effective at tutoring at risk first grade readers than credentialed teachers. Perhaps teachers possess a greater knowledge base that gives them an advantage over paras in teaching reading to beginners. One limitation of the Brown et al. (2005) study was that paras were randomly assigned to students whereas classroom teachers tutored students from their own classes. The present study used complete random assignment to compare trained paraprofessionals to credentialed teachers and reading specialists. Apart from Brown et al., few researchers have conducted such comparisons systematically.

            The Reading Rescue (RES) intervention model has been found effective in previous studies reported at conferences or in technical reports (Britt, 2002; Hoover, 1994, 1995; Hoover & Lane, 2001; Hoover & Sullivan, 1996). These studies found that RES tutoring raised students’ reading achievement from the bottom quartile to grade-level performance. Results of a recent experiment (Muller & Davies, 2004) showed that high-poverty, urban, predominantly Hispanic students who received RES tutoring outperformed a control group of students who received an alternative intervention. Results of the present study were intended to replicate and extend findings of these studies.

            Several questions were addressed:

1. Will struggling first-grade readers who receive Reading Rescue tutoring make greater gains in their reading achievement than struggling readers who do not receive tutoring but are enrolled in the same schools, and greater gains than struggling readers who are enrolled in comparable schools not implementing the Reading Rescue intervention? Both types of control groups were included to take account of contamination effects on reading instruction possibly resulting from the presence of the RES program in the schools.

2. Will struggling readers who receive RES tutoring make greater gains in reading than struggling readers who receive a small group intervention program adopted by the district?

3. Will RES tutoring increase students’ reading achievement to expected levels of performance based on nationally normed tests?

4. Will entry-level abilities of students and features of RES tutoring predict how much first graders improve in their reading as a result of tutoring?

5. Will all types of tutors, including reading specialists, credentialed teachers, and trained paraprofessionals, be equally effective at improving the reading achievement of their students?

Method

Participants
            Participants were enrolled in five low-SES urban public schools that had implemented the RES tutoring intervention for 2-3 years and three comparable schools that had not implemented RES but hoped to do so in the future. All schools were located in the same district in a large metropolitan city. The majority of students, above 90%, were from homes where Spanish was the first language, and 95% qualified for free or reduced-price lunch. Students judged by teachers to have no knowledge of English were excluded from the study. Most students had sufficient English to qualify. Students in special education at the beginning of first grade (i.e., those with speech, language, or physical disabilities or behavioral problems) were excluded by schools because they were already receiving special services.

            First graders in the five RES schools (N=497) were group-administered the Reading Rescue Classwide Screening Assessments to identify lower performing students (N=203) who were tested individually on the RES pretests. Students who could name at least 17 letters¹ but who were unable to read a preprimer passage on an informal reading inventory formed the pool of students who qualified for the intervention (N = 190). Within each school, these candidates were rank ordered on the RES individual pretests, and adjacent scores were used to form 64 matched pairs. Between the time that pairs were formed and the intervention began, four of the five schools assigned some of these students to a small group intervention mandated by the district, Voyager/Passport (2004). In these cases, the other member of the pair was assigned to receive RES. For the remaining pairs, members were randomly assigned to the RES and control groups. During the year, a few students left school (N = 8 RES and 9 controls). Replacements (i.e., 8 RES and 7 controls) were students drawn from the larger pool with the most similar pretest rankings.

            To address the possibility that the presence of RES in schools would affect the reading instruction in those schools, a second control group of first graders was drawn from three comparable schools not implementing the RES intervention. This group consisted of struggling readers who had similar scores to students in the RES group on the fall GMRT4 pretest, and who completed both fall and spring GMRT4 pre- and posttests.

There were 64 students who received the RES intervention, 62 control students (C1) enrolled in schools offering the RES program, and 60 control students (C2) enrolled in non-RES schools. Students in the two control groups were reconfigured to distinguish those who received the district-mandated small group intervention (N=52) from those who did not receive this intervention (N=70). As evident in Table 1, none of the groups differed statistically in age or gender (all ps > .05).

            All students received regular classroom instruction in reading and writing. The same instructional programs were used in all the schools: Month-by-Month Phonics for First Grade (Cunningham & Hall, 1997) and readers/writers workshop.

             Tutoring was provided by 59 adults who were members of the school staff and had received formal RES training. The tutors included certified reading specialists with graduate degrees (N=17), adults certified in other areas (guidance counseling, math and social work) (N=15), and paraprofessionals (N=27).  Those certified in counseling and math had completed at least 6 credits of coursework in reading and language arts. All paras were high school graduates, many had bachelor’s degrees or were working toward BAs and teacher certification, and a few had MAs in other fields. Half of the tutors were bilingual. All tutors had additional responsibilities within the school. For many, this included teaching the small group intervention program. Tutors were randomly assigned to tutees within each school and remained with those students throughout the intervention. All but three tutors worked with only one RES student.

Materials and Procedures

            The pretests, posttests, and tutoring intervention were administered by school personnel. Researchers had contact with Reading Rescue trainers and with school coordinators who were liaisons to the study but no direct contact with other tutors or their students.

            Pretests. Group administered and individually administered pretests were given to assess entry-level reading ability and to identify potential candidates for tutoring. Group tests were given in all 8 schools whereas individual tests were given only in the 5 RES schools.

            1. Gates-MacGinitie Reading Test. The GMRT4, Level BR (Beginning Reading) test (MacGinitie, MacGinitie, Maria & Dreyer, 2000) was administered to classrooms of first graders in all eight schools in November. This multiple-choice test measured students’ knowledge of letter-sound correspondences and their ability to read high-frequency words. Teachers read aloud each question and answer choice and gave students sufficient time to respond. Raw scores were transformed to normal curve equivalent (NCE) scores based on national norms. NCEs are percentiles that have been transformed into a scale of equal units, and are therefore appropriate for computing averages. The K-R 20 reliability is reported to be 0.95. Scores were used to match students from the non-RES control schools to students in the RES group. Testing occurred before the interventions began.

            2. ITBS Vocabulary Test. The Iowa Tests of Basic Skills, Level 6, Vocabulary subtest (Hoover et al., 2001) was administered to classrooms of students in November. The teacher read the questions and answer choices aloud, and students recorded their answers. The K-R 20 reliability is reported to be 0.75.

            3. Reading Rescue Classwide Screening Assessments. The RES Screen consisted of four group-administered tests adapted from Clay (1993a, 2002) that were given by classroom teachers in mid-September to students in RES schools in a print-free testing environment. Scores were used to identify low-performing candidates for the study. The tests assessed alphabetic knowledge (i.e., students wrote all the lower- and upper-case letters they knew), word writing (i.e., students had 10 minutes to write all the words they knew how to spell), and developmental spelling (i.e., 10 words were dictated and students wrote letters for the sounds they heard).

            4. Reading Rescue Individual Pretests. Tutors administered 7 RES pretests individually to confirm students’ candidacy for tutoring in early November.  Students (1) named lower case letters, (2) named upper case letters, (3) gave sounds of the letters or examples of words beginning with the letter-sounds, (4) blended spoken sounds to form words, (5) segmented words into phonemes, (6) read aloud nonwords, and (7) read sight words from graded lists of the informal reading inventory. Scores in these tasks were summed to yield an overall score. Students needed to know at least 17 letter names to qualify for RES tutoring.

            5. Informal Reading Inventory. The Ekwall/Shanker Reading Inventory, 4^th Edition (Shanker & Ekwall, 2000) was administered in early November. Scores on the sight word task above determined the proper starting point for reading the graded passages.  Each passage was followed by comprehension questions. Two grade-level scores were calculated: word reading accuracy, and comprehension. Students began with the preprimer passage and continued until their word reading accuracy dropped below 91% or their comprehension below 60%. Any student who read the primer or Grade 1 passage at a minimum of 95% accuracy and at least 60% comprehension was disqualified.

            The Tutoring Intervention. The program was not guided by a teacher’s manual with predetermined content specifying each day’s lesson. Rather tutors made instructional decisions that were guided by the RES scope and sequence of skills as well as by their analyses of students’ performance on assessments, the previous written record of text reading, and observations of students’ response to instructional activities. Tutors were trained to teach explicitly and systematically rather than incidentally, while being responsive to individual needs.

              The intervention began in December, occurred during school hours, and consisted of “easing in” sessions followed by regular instructional sessions, and an “easing out” period. During easing in sessions, tutors administered the individual assessments and established rapport with students through such activities as drawing pictures and reading easy books together. Regular instruction consisted of a sequence of lessons organized into phases that gradually presented more challenging concepts and material as students made progress. Lessons taught fluency, phonological awareness and phonics, comprehension and vocabulary development, with the amount of time for each dependent on the phase of the program.

            To apply the strategies and skills acquired, students read fiction and nonfiction small illustrated books from the Ready Readers series (Englebretson et al., 2000). These books were used but without the supplementary materials accompanying the books. The books were organized by difficulty, and they systematically introduced and provided practice with high frequency sight words as well as sound-symbol relationships in decodable words. The books provided practice in the phonics instruction that students received, and they were read and reread to build fluency. During the easing out sessions, a final review of the application of learned skills and strategies was conducted and students’ skills were assessed.

            The following components were taught in this order during each session:

1. Fluency

            Students started each lesson by rereading one or more familiar books. Beginning in Phase 1, oral reading accuracy, sight word knowledge, and confidence as a reader were stressed. In Phase 2, students practiced the application of blending, segmenting and syllabication skills. Timed readings of familiar text were used to increase reading rate and reduce errors. In Phase 3 and beyond, reading speed and expression were emphasized.

2. Word Analysis and Comprehension Strategies

            Students reread orally a book that had been introduced to them during the previous lesson (see component 5 below). A written record was taken indicating the student’s word reading accuracy and use of word and text reading strategies. This served as a basis for deciding when to move to a higher level of difficulty. In selecting texts, tutors were advised to avoid books that were too easy or too hard and to maintain reading accuracy levels between 90% and 97%. In deciding which features of words to correct, tutors consulted a phonics elements chart. Errors on letter-sound relations that were less regular or were too advanced were not corrected.

3. Phonological Awareness and Word Study

            Multisensory approaches were used. A phonics elements chart guided instruction and was supported by text reading practice in the Ready Readers. The phonics chart grouped instruction in letter-sound correspondences by phases. Phase 1 covered basic consonants, short vowels, and consonant-H digraphs. Phase 2 covered the short-long vowel distinction, additional consonants, and ER and ED endings. Phase 3 covered contractions, R-controlled vowels, additional vowel patterns, hard and soft G and C, and compound words. In addition, 18 to 23 phonograms were taught in each phase. Phonics was taught systematically and sequentially using both analytic and synthetic approaches.  For example, students saw and heard a phonetically regular word, they repeated the word slowly saying its separate sounds while pointing to its letters, and then they spelled the word by moving plastic letters to segment and represent the sounds. Additional activities were employed for students having difficulty, such as tracing letters while saying their sounds, and using a mirror to monitor articulatory movements in words.

            Phonemic awareness was taught systematically. During Phase 1, Elkonin boxes were used to teach phonemic segmentation. Students progressed from two-phoneme words to four-phoneme words, each illustrated by pictures so that pronunciations were recognized. Students kept a personal book, the Letters and Words I Know Book. The names and sounds associated with unfamiliar letters were taught and practiced. Segmenting and blending were taught using fingers and magnetic letters. In Phase 2 students were taught to decode by analogy using key words. They focused on high frequency words containing common phonograms. In late Phase 2 and beyond, tutors taught syllabication and other ways of dividing words to decode them, for example, common content words like carrot, morning, talking, bathtub.

4. Writing to Develop Phonological Awareness, Phonics, and Comprehension

            In Phases 1 through 3, tutors assisted students in writing one or more sentences that the students had produced in conversation about the books they read.  Attention was directed at spelling words correctly as well as comprehending the texts. When students were deemed ready in Phase 3, tutors taught students about text structure and how to organize information in texts they had read, for example, through the creation of semantic maps for expository text.

5. Comprehension and Vocabulary Development with the New Book

            In the initial phase, tutors introduced a new book by talking through it with students, page by page. As the books became longer, introductions focused more on making predictions based on the title and cover page. One purpose was to develop oral language by encouraging students to talk about the books and  by explaining the meanings of new vocabulary words. These words were written in students’ personal books, and the meanings were reviewed each time the book was read. Another goal was to spark the student’s interest in the book.  The tutor coached the student through the reading of the book and prompted the application of the reading skills and strategies that had been taught. Students were encouraged to decode unknown words by relying on their letter-sound knowledge and then cross checking with meaning and pictures to confirm the identities of the words. Students were asked literal, inferential, and evaluative questions after the text was read. In Phase 3, students were taught the difference between text explicit and implicit questions.

            Graduation from Reading Rescue. The RES program specifies multiple criteria for determining when students have met the goals of the program. Data on graduation criteria and decisions were not collected in the present study. Rather the reading achievement of all the students who received tutoring was assessed at the end of first grade.

Training of Tutors and Coordinators. The tutors were taught how to provide the intervention by Reading Rescue program staff. Training was guided by the Reading Rescue Tutor Handbook (Hoover, 2001). The initial phase consisted of five days of workshops delivered at the schools. The first two days at the beginning of the school year developed tutors’ understanding of the reading process, the components of the intervention model, the purpose, administration and scoring of the Classwide Screening Assessments, and how to begin working with students. The next three days of training took place after tutoring had begun and focused on the lessons.

A school coordinator who was a member of the school’s staff served as an on-site resource whose role was to provide modeling, feedback and support to tutors, to help tutors form a Peer Coaching Team, and to monitor implementation of the program. The coordinator had other literacy-related responsibilities within the school, but did not have classroom teaching duties. Coordinators were in their 2^nd or 3^rd year of implementation. They had attended advanced training institutes. They were given all the materials necessary, including video-taped lessons, to continue the training of tutors in their school. They were expected to schedule regular staff development meetings. RES program staff returned to each school several times to provide continued training and technical assistance for tutors and coordinators, and for quality assurance visits.

            Assessment of Tutoring. Tutors completed lesson record sheets as they tutored during each session. The sheets reminded tutors of assessments, lesson components, activities, and reading strategies to be taught and mastered in each phase. Tutors listed dates, materials used to teach components, and how students responded.

            To assess characteristics of tutoring and tutors’ adherence to procedures, two professionals thoroughly familiar with the program scored the records kept by tutors whose names had been removed. The scorers analyzed various characteristics including number of easing in, easing out, and regular tutoring sessions, the number of readings or rereadings of books, and reading accuracy levels which consisted of the proportions of readings that were completed at a 98-100% accuracy level, at a 90-97% accuracy level, and below 90%. These reading measures were taken from students’ second exposure to the books when they read them independently while tutors kept a running record. The first exposure to the books had occurred during the previous session when tutors coached students through the book (see Components 2 and 5 of the intervention above).

            To provide an index of tutors’ adherence to the intervention model, judges rated the following six dimensions of the tutoring records on a four-point scale to indicate whether expectations were met: the number of books read; the difficulty of the books; selection of elements for word study; features of the sentences written; instructional decisions made during sentence writing; provision of tutoring on a regular basis on the days available. Scores were summed to provide an adherence measure with a maximum score of 24. To illustrate, the criteria used to rate tutors’ selection of sentences was as follows:

            Rating of 4: Sentences chosen by the tutor for the student to write conformed to the

                        recommended criteria with regard to their length, complexity, and lexical features.

            Rating of 3: With few exceptions, sentences chosen by the tutor conformed to the

                        recommended criteria.

            Rating of 2: Many sentences chosen by the tutor conformed to the recommended criteria.

            Rating of 1: For the majority of lessons, sentences chosen by the tutor did not reflect program

                        recommendations, or many sentences were not recorded.

The two judges rated all of the records independently and were in agreement on 90 percent of their ratings. Disagreements were discussed and resolved. Non-conformance to the program corresponded to a rating of 1 on five dimensions and 2 on one dimension (total score of 7). Conformance to the program corresponded to ratings of 2 or above on five dimensions and 3 or above on one dimension for a minimum adherence score of 13.

            Small Group Intervention. Some of the struggling readers in the control groups (C1 and C2) across the 8 schools received instruction in a small group (SG) intervention program, Voyager Passport (2004). The SG program was adopted by the school district as a supplemental intervention for first-grade struggling readers with the goal of bringing their reading up to grade level. This was the first year of implementation. The program provided intensive, explicit, scripted instruction consisting of 130 lessons scheduled to be taught daily for 26 weeks. Tutor training in the program was provided by the district and consisted of one session plus on-site assistance by a literacy coach. Because instruction was strictly controlled by a script, more extensive training was not regarded as necessary.

            Sessions were conducted in groups (3-6 students) and lasted for 30-40-minutes. The program taught the same components as the RES program. These were phonemic awareness, phonics, daily story reading, vocabulary instruction that included multiple exposures to new words, reading and listening comprehension with both narrative and expository passages, and fluency. Phonics instruction was emphasized and included teaching letters and their sounds, decoding regular words, reading high frequency irregular words, teaching various word types, word building exercises, and spelling. Fluency included practice in speeded letter naming and word reading as well as text reading in a limited number of decodable books. The program included a Vital Indicators of Progress Guide for assessment which was scheduled to occur every fifth lesson. Parents were expected to provide home support with take-home readers. Once students entered the program, they remained for the duration of the school year. The small group intervention was taught by many of the same staff members who also tutored in the RES program in the five RES schools. However, it was also taught to some of the C2 students in the non-RES schools by teachers and paraprofessionals who were not trained as RES tutors.

            Posttests. Students in the RES and C1 groups completed several individually administered posttests in May: two WRMT-R (Woodcock, l987) subtests given by trained reading specialists, and the RES Posttests given by tutors. Classroom teachers administered the GMRT4 to first graders in all eight schools and the RES Classwide Screen to first graders in the RES schools.

            1.  WRMT-R Word Identification Test. Form G of the Woodcock Reading Mastery Tests-Revised (Woodcock, 1987) required students to read a list of real words that increased in difficulty.  The median split-half reliability of this measure is reported to be 0.98.

            2.  WRMT-R Word Attack Test. Students read a list of decodable nonwords. The median split-half reliability of this measure is reported to be 0.94 (Woodcock, l987).

            3. Reading Rescue Individual Posttests. The seven subtests of the RES Individual Pretests were repeated as posttests along with the Ekwall/Shanker Reading Inventory, 4^th Edition (Shanker & Ekwall, 2000). (See descriptions above.) The maximum score on the RES posttests was higher, 223, because the sight word test included 110 words. The highest passage given to any student in the posttest phase was the third grade passage.

            4. Gates-MacGinitie Reading Test. The GMRT4, Level 1 (MacGinitie et al., 2000) was group administered by classroom teachers. Questions were read and answered by students independently. The test consisted of two subtests whose scores were treated separately as well as combined into a total score. The K-R 20 reliability of the total score is reported as 0.96 (MacGinitie et al., 2002).

            a. Word Decoding. There were 43 items. On each, students identified which of four words corresponded to a picture.  The foils resembled the correct words in appearance or sound. Knowledge of letter-sound correspondences and letter sequences was required to make correct choices. K-R 20 reliability of this subtest is reported as 0.94 for the spring of Grade 1 (MacGinitie, et al., 2002).

            b. Reading Comprehension. Students read passages covering both fiction and nonfiction and answered each comprehension question by choosing from a panel of pictures the one that illustrated or answered a question about a segment of the text. K-R 20 reliability of this subtest is reported as 0.93 for spring of Grade 1 (MacGinitie et al., 2002).

 Design and Analyses

            Performance of the group receiving RES tutoring was compared to performance of two different configurations of the same control students. The first configuration divided the students into a control group drawn from the same five schools as the RES students, referred to as Control 1 (C1), and a control group drawn from the three schools that did not offer RES, Control 2 (C2). The purpose of including C2 was to check on the possibility that the presence of RES and the professional development it offered to schools might contaminate instruction provided to control students in those schools. The second configuration divided the students into those receiving the small group intervention (SG) and those who did not receive this intervention (NI). Students in SG and NI groups came from all 8 schools.

            Analyses of variance (ANOVA) were applied to fall pretest scores to determine whether the groups differed in entry-level reading skills. To assess effects of the RES and SG interventions, analyses of covariance (ANCOVA) were applied to spring posttest scores with GMRT4 pretest scores as the covariate.  The dependent variables included word decoding and reading comprehension scores on the GMRT4, word reading and nonword reading scores on the WRMT-R, and performance on the RES program tests. In some analyses, data were missing for a few students. Statistical tests were conducted on actual scores.

            To compare the three types of tutors--reading specialists, credentialed staff, and paraprofessionals--the performance and tutoring experiences of their students were subjected to ANOVAs. To determine whether features of the tutoring instruction predicted students’ reading achievement at the end of first grade, regression analyses were conducted.

Results

             Equivalence of Reading Rescue Treatment and Control Groups on the Pretests. Table 1 shows the mean pretest performance of students who received RES tutoring and control students. From mean percentiles on the fall GMRT4 reading test and ITBS vocabulary test in Table 1, it is apparent that participants were struggling readers at risk for future reading difficulties and had English language vocabularies well below average compared to national norms for first graders.

            The ANOVA of pretest scores with treatment group as the independent variable revealed that the three groups did not differ on the vocabulary test but they did differ on the GMRT4 test. A post hoc Bonferroni test indicated that the RES and C2 groups performed equivalently and significantly better than the C1 group. This was unexpected because RES and C1 students had been matched on their performance on the RES individual tests. One possible reason for poorer performance of the C1 group is that four out of five RES schools undermined random assignment by assigning students they deemed struggling readers to the small group intervention.

            To check on whether C1 students were poorer readers than RES students on the other literacy pretests, mean scores were compared. ANOVAs revealed that the groups did not differ on any of the three group-administered RES Classwide Screen tests (all ps > .05) or on any of the 7 individually administered RES pretests (all ps > .05). Correlations between the RES pretests and the GMRT4 were as high as r = .69 (developmental spelling) and r = .67 (sight word reading) indicating that the RES tests were not lacking in sensitivity. These findings raise doubt that the RES and C1 groups actually differed in their reading level in the fall.

            Nevertheless, to take account of the possibility of initial differences in ability favoring the RES and C2 groups, fall scores on the GMRT4 were used as a covariate in ANCOVAs of spring posttest scores. This analysis and choice of a covariate to adjust for initial differences in reading ability were considered legitimate for several reasons: the GMRT4 test has high reliability; it was administered to all participants at the same time of year before tutoring began; it was strongly correlated with the next level in the GMRT4 test series given as a posttest (r = .68, p < .001, N=844).

            Comparison of Reading Rescue Treatment to C1 and C2 Control Groups on Posttests. Results of the ANCOVA applied to posttest scores revealed significant main effects of treatment group. Table 1 reports means and test statistics on the nationally normed tests. Bonferroni post hoc comparisons  on GMRT4 performance revealed that RES tutored students decoded significantly more words and comprehended text significantly better than both of the control groups. In turn, the C1 group, located in RES schools, decoded significantly more words than the C2 group, located in non-RES schools, but comprehension of the two groups did not differ. This finding of a difference suggests that the presence of the RES program in schools may have improved the decoding instruction that struggling readers in these schools received.

            On the WRMT-R tests that were individually administered, the RES tutored group read significantly more words and decoded significantly more nonwords than the C1 group. Mean grade-equivalent scores on the WRMT-R tests in Table 1 show that RES students were equally skilled at reading real and nonwords, an indication of good phonics skills. In contrast, C1 students were weaker in reading nonwords than real words.

            Effect sizes were calculated by dividing the difference between the RES and C1 adjusted means by the pooled standard deviation on each measure. According to Cohen (1988), an effect size of 0.20 is considered small, an effect size of 0.50 is moderate, and 0.80 and above is large. Results on the GMRT4 were 0.53 (decoding), 0.43 (comprehension), and 0.50 (total score). Results on the WMRT-R were 0.72 (word reading) and 0.65 (word attack). Thus, effect sizes favoring the tutoring treatment over no tutoring were in the moderate range.

            Performance of the RES tutored and C1 control groups on posttests from the RES batteries are shown in Table 2. ANCOVAs were conducted with fall GMRT4 scores as the covariate. The groups did not differ on four of the five tests of letter knowledge where means were close to ceiling indicating that students in both groups had learned most of the letter names and sounds. However, significant main effects were observed on other posttests. RES tutored students outperformed control students in writing lower case letters, segmenting and blending phonemes, reading sight words and pseudowords, generating plausible spellings of words, and reading and comprehending text. Effect sizes ranged from moderate to large on these measures. These findings provided additional evidence that struggling readers who received RES tutoring read significantly better at the end of first grade than struggling readers who did not receive this program.

              Another question of interest was whether RES tutoring would raise students’ reading achievement from below average in the fall to an average level in the spring based on national test norms. We regarded the minimum average level as performance at the 40^th percentile. From Table 1, it is apparent that on the GMRT4, RES students performed at the 50^th percentile in decoding words, thus exceeding the minimum average level. RES students performed at the 38^th percentile on the reading comprehension test. Although this is somewhat below the national average, reaching this level of achievement on a grade-appropriate test is impressive given that these were primarily low SES students with low vocabulary scores from homes where English was not the first language. On the word and nonword reading subtests of the WRMT-R, RES tutored students’ mean grade-equivalent scores placed them at the 1.9 grade level which is the level expected at the end of first grade. These findings show that RES tutoring did indeed raise the means of the group from below average to average levels based on national norms. In contrast, both control groups performed well below the normal range (see Table 1).  

            To determine what proportion of the struggling readers in each group reached average levels by the end of the year on the GMRT4 posttests, scores of individual students were examined. As shown in Table 3, before tutoring began, very few struggling readers scored at or above the 40^th percentile across the groups, only 5% to 9%. However, at the end of the year, a much greater proportion of RES-tutored students than C1 or C2 students’ scores reached average levels. When RES students were compared to the entire sample of first graders (i.e., the WS group in Table 3), a greater proportion of RES tutored students reached average levels on the decoding test, and about the same proportions of RES and WS students reached average levels on the comprehension test. These findings show that the RES intervention brought many more of the struggling readers to average levels than would have occurred without this program. In fact, the record of success for RES students was comparable if not better than the record for the entire sample of first graders.

            Supplementary Comparison of RES Treatment Group to a Larger Control Group. Because the RES and C1 groups differed on one of the pretests, this left open the possibility that the groups reflected different populations of struggling readers. To address this possibility, a larger control group was created from the five schools implementing the tutoring program. This group included not only the 62 C1 struggling readers but also 62 struggling readers who qualified to receive tutoring but were not selected for the RES or C1 group. Comparison of entry-level performance of the RES group to this larger control group revealed no statistically significant differences on any of the pretests, including the GMRT4 (all ps > .05). The only pretest that approached significance was the RES individually administered phoneme segmentation test (p < .07), with means favoring the control group. Thus, entry-level reading capabilities did not distinguished the RES group from this larger control group. ANCOVAs of posttest scores with the fall GMRT4 as the covariate revealed significant main effects of treatment, with the RES group outperforming the control group on all three GMRT4 measures (decoding, reading comprehension, total score, all ps < .001). These findings replicate those above and show that the positive effects of the RES tutoring intervention held when compared to all struggling readers who were eligible for the program but did not receive it.

            Comparison of Students Receiving the Tutoring Intervention, Small Group Intervention, and Neither Intervention. Students in the C1 and C2 control groups were reconfigured to distinguish those who received a small group intervention (SG) from those who received neither intervention. Performance of these groups was compared to that of students receiving RES tutoring. ANOVAs of pretest means reported in Table  4 revealed a significant main effect of treatment group on the GMRT4 but no significant effect on the ITBS vocabulary test. Bonferroni post hoc comparisons on the GMRT4 showed that the RES group did not differ significantly from the SG or NI groups, but the NI group performed significantly better than the SG group. These findings indicate that even though some of the schools assigned struggling readers to the SG group and thus undermined random assignment of matched pairs, the reading level of students receiving the small group intervention was not significantly lower than that of students receiving the tutoring intervention.

            To adjust for initial pretest differences statistically, ANCOVAs were conducted on outcomes with fall GMRT4 scores as the covariate. The groups were compared only on the GMRT4 posttest because of incomplete data for the SG and NI groups on the other measures. Results revealed main effects of treatment (see Table 4). Bonferroni post hoc pairwise comparisons showed that in all three analyses, RES tutored students significantly outperformed the SG and NI groups who did not differ. These findings indicate that struggling readers who received RES tutoring decoded words and comprehended text more effectively than struggling readers who received the small group intervention as well as readers who received neither intervention.

            The small group intervention was not effective in raising performance above that of students who did not receive  this intervention.  Effect sizes comparing mean performance of the SG and NI groups were very small and nonsignificant, ranging  from -.08 on the word decoding measure to +.11 on the reading comprehension measure. Moreover, the mean performance of both groups was well below average based on national norms (see Table 4). These findings were somewhat surprising. The small group intervention program was comprehensive in its coverage of beginning reading skills with a strong emphasis on systematic phonics instruction.

            We also assessed the proportion of RES, SG, and NI students whose scores rose to average levels (i.e., at or above the 40^th percentile) on the GMRT4 at the end of first grade. As shown in Table 3, although proportions rose dramatically from fall to spring, many fewer SG and NI children than RES tutored children scored within the normal range at the end of the school year.

              In sum, these findings show that RES tutoring helped struggling first-grade readers progress in their reading more than the small group intervention program and more than receipt of neither intervention. In contrast, the small group intervention did not improve the reading achievement of students over that of students not receiving this intervention.

             Tutors and Characteristics of Tutoring. During each tutoring session, the tutors kept records. These were used to assess the occurrence of several features of tutoring. Mean values are reported in Table 5. The mean number of total tutoring sessions was 49.6, with a standard deviation of 18, indicating that individual students differed substantially in the amount of tutoring they received. The amount of text reading and rereading completed by students was extensive, M = 164 texts, with a large standard deviation also indicating much variability. Not surprisingly, the correlation between the number of sessions and the number of books read was high, r = .79, p < .01.

            The program recommends that books be selected to enable students to practice reading at their instructional level, indicated by accuracy scores ranging from 90% to 97%. To examine text accuracy levels, running records were evaluated. These were recorded during the first independent reading of newly introduced books after the books had been previewed and coached during the previous session. As indicated by means in Table 5, only 37% of the books were read at the recommended instructional level. More books, 50%, were read at an independent level (98-100% accuracy), and only 15% of the readings were at a frustration level (below 90% accuracy). This reveals that book reading accuracy was even higher than that recommended by the program.

            The records that tutors kept were used to assess their adherence to the program. Raters evaluated adherence on six dimensions, such as the number of books read in each session, the difficulty level of the books, the appropriateness of the sentences written, and whether tutoring was provided on a regular basis. Ratings were summed to provide an overall score (maximum of 24). A total score of 13 was considered the minimum level indicating full adherence to recommended practice. As seen in Table 5, the mean value, 13.5, shows that tutors on average were performing slightly above this level. The small standard deviation (3.2) indicates that there was limited variation in scores. Inspection of scores showed that 63 percent of the tutors displayed full adherence to program procedures. Two dimensions displayed the strongest adherence, indicating that tutors were best at selecting the appropriate number of books to read, and having students write appropriate sentences that contained sight words and words with the letter-sound relations being learned. These findings indicate that the majority of the tutors did conform to expected instructional procedures.

            To determine whether tutors’ adherence to the program influenced how well their students performed on outcome measures, correlations were examined between adherence scores and the various posttests. Results revealed little relationship, with values ranging from r = 0.06 on the Woodcock word identification test to r = 0.20 on the GMRT4 total score, all ps > .05. Because tutors did not differ much in their adherence to the program, little relationship to outcomes was observed.

            Three types of tutors worked with students: reading specialists, other credentialed personnel, and paraprofessionals. To determine whether the three types differed in their tutoring effectiveness, ANOVAs were conducted with type of tutor as the independent variable and various measures taken on their students as the dependent variables. All but three of the 59 tutors taught only one of the 64 RES students, so most data points emerged from different tutors.

            The three types of tutors did not differ in the reading level of their students when tutoring began, as indicated by a non-significant main effect of tutor type on the GMRT4 fall pretest (F < 1, p > .05). This is as expected because students were randomly assigned to tutors. ANCOVAs of reading outcomes with fall GMRT4 as the covariate revealed no main effects of tutor type on the GMRT4 posttests, the WMRT-R word identification posttest, or the RES program posttests (all Fs < 1). However, a difference was detected on the WRMT-R nonword test assessing students’ ability to decode unfamiliar words. As shown in Table 5, Bonferroni post hoc analyses revealed that reading specialists were significantly better at teaching students to decode than paraprofessionals (p < .003).

            Measures of events that occurred during tutoring were subjected to ANOVAs with tutor type as the independent variable. From Table 5, it is apparent that tutors differed significantly in the number of regular sessions they conducted with students. Post hoc analyses revealed that the paraprofessionals conducted significantly more sessions than the other two types of tutors. However, the tutors did not differ in the number of easing in or easing out sessions. In addition, the tutors differed significantly in the number of books that they had students read and reread. Paraprofessionals exceeded reading specialists in this regard, not surprisingly because they provided more tutoring sessions. Other tutoring events were examined as well, but as shown in Table 5, no other effects of tutor type were detected (all ps > .05). The fact that reading specialists required fewer sessions to achieve effects than paras suggests that reading specialists were more efficient in their tutoring.

            From these findings we conclude that paraprofessionals tutored as effectively as the credentialed teachers who were not reading specialists. Moreover, they tutored as well as the reading specialists in strengthening students’ word reading and text comprehension skills, but not as well in teaching the skills that benefit nonword decoding. In addition, paras took more sessions to work with students than reading specialists, indicating they were less efficient in delivering the tutoring. Perhaps decoding skill was more effectively taught by reading specialists, because it requires special knowledge about the alphabetic system (Moats, 2000) or because reading specialists regarded it as more important to teach than paras did.

            Prediction of Reading Achievement for Students Receiving RES Tutoring. The next question of interest was whether entry-level abilities of tutored students and characteristics of tutoring might explain why some students improved more in reading than others. The outcome measure was the total GMRT4 spring score, which was highly correlated with the other reading outcome measures (i.e., r’s ranged from .62 to .93) indicating that it was indicative of general reading performance. Correlation coefficients between this outcome and each of the other variables involving students’ entry-level skills and tutoring characteristics revealed several statistically significant values reported in Column 1 of Table 6. The significant predictors included performance on the fall GMRT4 pretest, letter writing on the pretests, RES program individual pretests (total score), the proportions of books read at the three accuracy levels, and the number of easing in sessions. The two accuracy levels, 90-97% and < 90%, were each strongly and negatively correlated with the 98-100% accuracy measure (see Table 6), indicating that all three measures reflect one factor, text reading accuracy. This factor is best represented by the 98-100% accuracy measure. The number of easing in sessions correlated negatively with the GMRT4. One reason may be that students with more severe reading problems often took longer to assess because tutors spaced out the tests to avert a sense of failure. None of the other student or tutoring variables was correlated significantly with reading achievement (all ps > .05).

            Regression analyses were conducted with the spring GMRT4 total score as the reading outcome. The predictors were the four pretest measures, the 98-100% text accuracy measure, and the easing in measure. When all of the predictors were entered as a single set, only the 98-100% text accuracy variable explained significant variance in the model, with intercept = 21.3, slope = 25.5, p < .001. When the accuracy variable was entered first in a stepwise regression, none of the other variables explained any significant additional variance (all p’s > .05). When the accuracy variable was entered last in a stepwise regression, it explained significant additional variance (p < .001), with r² increasing from .36 to .51, and contributions of each of the other predictors becoming non-significant (all ps > .05).  These findings indicate that high text reading accuracy during tutoring was the strongest predictor and the only unique predictor of students’ reading achievement at the end of first grade.   

            The strong relationship between high text reading accuracy during tutoring and reading is open to at least two interpretations. One is that the relationship may reflect the effects of tutoring. Children whose tutors enabled them to read a greater proportion of texts at high accuracy levels became better readers by the end of the year than children whose tutors did not do this. Higher levels might have been achieved either by tutors’ selecting easier texts or by tutors previewing and coaching students more effectively through the texts during the previous session when the books were introduced. The other interpretation is that students who entered the study as the better readers were able to read texts at higher levels of accuracy during tutoring than the weaker readers and also achieved higher reading scores at the end of first grade. Findings of the regression analyses appear to rule out the second interpretation. Pretest reading skills of students taken when they began tutoring did not explain variance on the reading posttest once the impact of the text accuracy tutoring variable was considered. Also these students had very limited reading skills when tutoring began, so their growth in reading resulted primarily from the instruction that they received, not from entry-level reading skill. These findings favor the explanation that students’ experiences learning to read texts at high accuracy levels during tutoring influenced how well they read as a result of the tutoring.

Discussion

            Several important findings emerged in this study. First grade struggling readers who received tutoring according to the Reading Rescue intervention model made significantly greater improvement in reading than struggling readers who did not receive this tutoring but were enrolled in the same schools, and also greater improvement than struggling readers who were enrolled in comparable schools not using the program. Also students receiving RES tutoring outperformed students receiving a commercially produced, small group intervention program mandated by the school district for struggling readers. An advantage occurred despite the fact that both intervention programs were comprehensive and provided instruction in the same components. Greater gains in reading by tutored students were observed on standardized measures of word and pseudoword reading and reading comprehension. In fact, RES tutoring raised the majority of students’ reading from below average to average levels. This contrasted with many fewer students reaching average levels in the comparison groups.

            The reading achievement of students who received RES tutoring appeared to be explained primarily by one aspect of their tutoring experience, reading texts at a high level of accuracy, between 98 and 100%. This finding needs more careful consideration because it appears to contradict a commonly accepted view about the quality of text reading that is most effective for building fluency. Rasinski (2003) stated this shared wisdom in his book, The Fluent Reader (p. 63), “As with any direct-teaching activity, the greatest gain will occur when the difficulty of the material is at the student’s instructional level (i.e., 90-95 percent accuracy in word recognition)—neither too hard nor too easy, just right!” In fact, the developer of the RES tutoring model adopted this view and recommended it to tutors. However, present findings suggest otherwise.

            Following program guidelines, tutors in the present study coached students through a book during one session and then in the next session had students reread the book, this time independently, and recorded their accuracy. Results revealed that the proportion of texts read at an instructional level (90-97% accuracy) was strongly and negatively correlated with positive growth in reading, whereas the proportion of texts read at a higher, independent level (98-100% accuracy) was strongly and positively correlated with reading growth. In fact, the latter, high accuracy measure figured more prominently than any other variable in explaining variance on the reading posttest. Even when students’ entry-level reading skill was controlled statistically, this tutoring variable still explained significant variance. Before concluding that reading text at high accuracy levels is important for promoting growth in struggling readers, we need to consider alternative interpretations.

            In the present study, the high accuracy levels that proved important occurred on text that was read independently after prior coaching. It may be that conventional wisdom regarding recommended accuracy levels (i.e., 90-97%) applies to text read without prior coaching.  Reading text cold is certainly susceptible to more errors. What difference this makes deserves further research.

            Monitoring high text accuracy levels may have proven important for reading achievement because it reflected the impact of the preview session for improving students’ word reading. According to theories of word learning (Ehri, 1992; Perfetti, 1992; Rack, Hulme, Snowling, & Wightman, 1994; Share, 1999, 2004a), readers remember how to read words when they possess knowledge of grapheme-phoneme relations that provide the connections securing the written words to their pronunciations and meanings in memory. Perhaps more thorough coaching by tutors to help students use decoding skill to secure words in memory as they previewed the decodable texts throughout tutoring was an important factor strengthening students’ word and text reading accuracy and their general reading ability. This possibility needs further study.

            Another possible factor contributing to the effectiveness of the RES model merits consideration, specifically, the criteria used to qualify struggling readers for the RES program. Rather than selecting all first graders at the bottom of the class, the program selected those who knew at least 17 letters yet who received low scores on other literacy tasks such as phonemic awareness and who lacked any text reading ability. Many studies have shown that letter name knowledge is the best single kindergarten predictor of success in learning to read (Scarborough, 2001). If students have not learned the shapes and names or sounds of letters, it is impossible for them to acquire word decoding skill. Share (2004b) showed that if students already knew the names of letters it was much easier for them to learn their sounds because most names contain their sounds. Letter names or sounds are typically taught in kindergarten, so first graders are expected to know them. Selecting students who already possess substantial letter knowledge reduces the need for RES tutors to spend time teaching letters and allows instruction to focus primarily on the application of letter knowledge to acquire reading and spelling skills. According to the model, low performing readers who lack sufficient letter knowledge at the beginning of first grade are admitted for RES tutoring later in the year once they qualify. This more refined tailoring of program characteristics to suit student capabilities very likely enhanced its effectiveness.

            Although having letter knowledge when they began RES tutoring may have enabled students to make better progress, it is important to note that this factor does not explain why RES students outperformed control students at the end of the year. In the present study, control students met the letter knowledge criteria as well when they were selected for the study. This suggests that the nature of the instruction as it capitalized on entry-level letter knowledge was of primary importance in explaining the greater success of RES students in learning to read.

            Another purpose of the study was to assess the effectiveness of various types of tutors. Present findings indicated that paraprofessionals delivered RES tutoring as effectively as reading specialists and credentialed teachers except in two respects. Paras were less successful in impacting students’ ability to decode pseudowords than reading specialists were. Paras provided more tutoring sessions than the other tutors and, as a result, had students read and reread more books. The fact that paras’ students were just as good at reading real words and comprehending text as students taught by the other tutors suggests that paras were less efficient in their tutoring.

             Interestingly, one of these findings replicates and extends that reported by Brown et al. (2005) who found that students tutored by paras did not differ from students tutored by teachers on tests of word and text reading but only on the WMRT-R pseudoword reading test, the same test used in the present study. Although both studies show agreement, confidence in the present finding is greater because tutors here were randomly assigned to students whereas tutors in the Brown et al. study were not.

            What might explain the shortcoming in paras’ tutoring that was observed in both studies? Perhaps paras spent less time teaching decoding than the reading specialists did. In the present study, paras had their students read more books than other tutors. Perhaps paras lacked the background knowledge to teach phonics effectively, or perhaps they did not consider decoding as important to teach. According to Moats (2000), a high level of knowledge about the alphabetic system is needed to teach systematic phonics programs effectively. These findings and their explanation deserve further study to determine whether more extensive staff development in how to teach decoding is needed to prepare paraprofessional tutors adequately.

             Comparison of the RES tutoring intervention to the small group intervention suggested that tutoring was more effective. Differences in reading outcomes were observed despite the fact that the reading instruction components taught were similar in the two programs. Tutoring may have proven especially effective because of the greater ease of adapting instruction to the needs of individuals, and the greater amount of reading practice with feedback that was possible. Alternatively, extenuating circumstances may have created an advantage for tutoring in the present study. The RES program had been in use for two to three years in the five schools, whereas the SG program was in its first year of implementation. RES tutors and staff were more experienced delivering RES than SG, so this may explain the difference. On the other hand, new programs can benefit from a novelty effect in the first year, and this would be expected to benefit the SG program.

            The author of the Reading Rescue Program suggests that implementing the program in schools may raise the reading achievement of even those first graders who are not tutored (Hoover, 1996). This general benefit may result from the professional development and tutoring experience that teachers and paras on the staff receive. Some evidence for this was detected in the present study. Control students enrolled in RES schools outperformed control students in non-RES schools on the GMRT4 word decoding posttest. The possibility that participation in the RES program improved teachers’ ability to teach decoding outside of the RES program in their classrooms merits systematic study.

            Possible limitations of the present study need to be considered. Because we were unable to assign all the students randomly to the tutored and control groups, our design was quasi-experimental. However, we took steps to match students in assigning them to the groups, and we adjusted for any entry-level differences in our statistical analyses of outcomes. In addition we compared the tutored group to several different comparison groups and found an advantage for tutoring in all cases.

            A second possible limitation is that our paraprofessionals were relatively well educated, with many holding college degrees. Our findings may not generalize to paraprofessionals with less than a high school diploma. This possibility awaits study.

            A third limitation is that only short-term effects of the RES tutoring model were tested, at the end of the year when instruction ended. Of equal interest is whether the effects of tutoring in first grade are lasting and maintain students’ reading at average levels in subsequent grades. This matter awaits study.

            A fourth limitation bears on the comparison of tutoring to small group intervention. These two forms of instruction were confounded with other variables. The two systematic phonics programs were similar but not identical. RES tutors received more training and were more experienced using their program than SG tutors were. The time that students spent receiving SG instruction was not monitored so comparability to RES time is unknown. As a result, it is not clear that the instructional delivery unit was the critical factor explaining the difference in students’ performance following the interventions, so any conclusion about the greater effectiveness of tutoring over small group instruction remains tentative.

            Present findings do make it clear, however, that the small group intervention was not effective, whatever the cause. Struggling readers who received this intervention performed no better than struggling readers who did not receive it. This came as a surprise. The program held a strong reputation and was mandated by the district. The program was scripted and explicit regarding the steps for delivering instruction. Reading specialists as well as paraprofessionals served as the teachers. Sessions were conducted on a regular basis at all the schools. Because this program is used widely, uncertainty about its effectiveness merits further study. One possibility to explore is that admission of students without letter knowledge to the small groups necessitated attention to letter instruction and as a result limited the progress of group members who already knew letters (i.e., students in the SG group of the present study).

            Despite some limitations, the present study has many strengths. This was a well designed, field-based research study, not all that common in the literature, and never perfectly executed due to the capriciousness of school environments. The study was conducted in several typical urban, low SES elementary schools. Procedures of the intervention programs were implemented by school personnel with few if any modifications introduced by researchers. As a result, the study has high external validity. Random assignment was employed with tutors so the internal validity of conclusions regarding their effectiveness is high. Although random placement of students in treatment and control groups was not fully implemented, statistical adjustments applied to multiple outcome measures maintained the internal validity of the study.

            Findings advance our understanding of the nature and impact of early intervention, and they carry important implications for practice. Results offer evidence for the effectiveness of a specific tutoring intervention model providing comprehensive instruction in multiple components of reading needed by struggling beginning readers. The model proved effective with the neediest students, those in urban, high-poverty schools with large numbers of minority students and students from bilingual homes. The core of the intervention model is not a commercial program but rather involves a service offered by a nonprofit foundation whose staff helps schools institute a tutoring program by using existing teachers and paraprofessionals already employed at the school. As a result, implementation costs substantially less than programs requiring the hiring of credentialed, specially trained teachers.  The present study contributes by documenting the effectiveness of this approach to reading intervention. In addition, findings show how other interventions programs for struggling readers in first grade might be structured to maximize their effectiveness.

References

            Biemiller, A. (1999). Language and reading success: From reading research to practice: A series for teachers (Volume 5). Cambridge, MA: Brookline Books.

            Block, C. & Pressley, M. (Eds.) (2002). Comprehension instruction: Research-based best practices. New York: Guilford Press.

            Britt, L.A. (2002) Reading Rescueâ in District Six Community Schools of the New York City Public School System. (Technical Report). Gainesville, FL: The Literacy Trust, Inc.

            Brown, K., Morris, D., & Fields, M. (2005). Intervention after Grade 1: Serving increased numbers of struggling readers effectively. Journal of Literacy Research, 37, 61-94.

            Clay, M. (1985). The early detection of reading difficulties. (3^rd ed.) Portsmouth, NH: Heinemann.

            Clay, M. M. (1993a). An observation survey of early literacy achievement. Portsmouth, NH:  Heinemann.

            Clay, M. (1993b). Reading Recovery: A guidebook for teachers in training. Portsmouth, NH: Heinemann.

            Clay, M. M. (2002). An observation survey of early literacy achievement (2^nd ed). Portsmouth, NH:  Heinemann.

            Cohen, J. (1988).  Statistical power analysis for the behavior sciences (2nd ed.).  Hillsdale,

NJ:  Erlbaum.

Cunningham, P. & Hall, D. (1997). Month-by-Month Phonics for First Grade: Systematic, Multilevel Instruction. Greensboro, NC: Carson-Dellosa.

Ehri, L. C. (1992).  Reconceptualizing the development of sight word reading and its relationship to recoding.  In P. Gough, L. Ehri, & R. Treiman (Eds.), Reading Acquisition, (pp. 107-143).  Hillsdale, NJ:  Erlbaum.

            Ehri, L., Nunes, S., Willows, D., Schuster, B., Yaghoub-Zadeh, Z., & Shanahan, T. (2001). Phonemic awareness instruction helps children learn to read:  Evidence from the National Reading Panel's meta-analysis. Reading Research Quarterly, 36, 250-287.

            Ehri, L., Nunes, S., Stahl, S., & Willows, D. (2001). Systematic phonics instruction helps students learn to read: Evidence from the National Reading Panel's meta-analysis. Review of Educational Research, 71, 393-447.

            Elbaum, B., Vaughn, S., Hughs, M., & Moody, S. (2000). How effective are one-to-one tutoring programs in reading for elementary students at risk for reading failure? A meta-analysis of the intervention research. Journal of Educational Psychology, 92, 605-619.

            Englebretson, R., Hiebert, E., & Juel, C. (2000). Ready Readers. Parsippany, NJ: Modern Curriculum Press.

            Foorman, B., Francis, D., Shaywitz, S., Shaywitz, B., & Fletcher, J. (1997). The case for early reading intervention. In B. Blachman (Ed.), Foundations of reading acquisition and dyslexia: Implications for early intervention (pp. 243-264). Mahwah, NJ: Erlbaum.

            Greany, K., Tunmer, W., & Chapman, J. (1997). Effects of rime-based orthographic analogy training on the word recognition skills of children with reading disability. Journal of Educational Psychology, 89, 645-651.

            Hatcher, P., Hulme, C., & Ellis, A. (1994).  Ameliorating early reading failure by integrating the teaching of reading and phonological skills:  The phonological linkage hypothesis.  Child Development, 65, 41‑57.

            Hoover, H., Dunbar, S., Frisbie, D., Oberley, K., Bray, G., Naylor, R., et al. (2003). Iowa Test of Basic Skills Complete/ Core Battery Norms and Score Conversions. Itasca, IL: Riverside.

            Hoover, H., Dunbar, S., Frisbie, D., Oberley, K., Ordman, V., Naylor, R., et al. (2001). The Iowa Tests of Basic Skills. Itasca, IL: Riverside.

            Hoover, N. (1994). Reading Rescue for Elementary Readers:  An Early Literacy Intervention Program 1993-1994 End-of- Pilot Year Report. (Tech. Report No. 1). Gainesville, FL: University of Florida.

            Hoover, N. (1995).  Reading Rescue for Elementary Readers:  An Early Literacy Intervention Program 1994-1995 End-of-Second Year Report on Student Progress at Interlachen Elementary School, Putnam County, Florida. Gainesville, Florida. (Tech. Report No. 2). Gainesville, FL: University of Florida.

            Hoover, N., & Sullivan, L. (1996, April). Teaching at-risk children to read: One school’s successful implementation of cost-effective tutoring.  Paper presented at the annual meeting of the American Educational Research Association, New York, N.Y.

            Hoover, N. (1996, April). Primary teachers learn to teach at-risk students to read: Empowerment and improvement through a school-based inservice model.  Paper presented at the annual meeting of the American Educational Research Association, New York, NY.

            Hoover, N. (2001). Reading Rescueâ: A Handbook for Tutors. Gainesville, FL: The Literacy Trust. Inc.

            Hoover, N., & Lane, H. (2001, May). Preventing Literacy Failure through Cost-effective Tutoring: A Report on Outcomes in High-poverty and Rural Schools.  Paper presented at the annual meeting of the International Reading Association, San Francisco, CA.

            Iversen, S., & Tunmer, W. (1993) Iversen, S., & Tunmer, W. (1993).  Phonological processing skills and the Reading Recovery Program.  Journal of Educational Psychology , 85, 112‑126.

            Juel, C. (1988). Learning to read and write: A longitudinal study of 54 children from first through fourth grades. Journal of Educational Psychology, 80, 437-447.

            Juel, C. & Roper-Schneider, D. (1985). The influence of basal readers on first grade reading. Reading Research Quarterly, 20, 134-152.

            MacGinitie, W., MacGinitie, R., Maria, K., & Dreyer, L. (2000).  Gates-MacGinitie Reading Tests, 4^th Edition.  Itasca, IL:  Riverside.

            MacGinitie, W., MacGinitie, R., Maria, K., & Dreyer, L. (2002).  Gates-MacGinitie Reading Tests, 4^th Edition Technical Report.  Itasca, IL:  Riverside.

            Matthes, P. & Fuchs, L. (1994). The efficacy of peer tutoring in reading for students with mild disabilities: A best-evidence synthesis. School Psychology Review, 23, 59-80.

            Moats, L. (2000). Speech to print: Language essentials for teachers. Baltimore, MD: Brookes.

            Morris, D. (1999). The Howard Street tutoring manual: Teaching at-risk readers in the primary grades. New York: Guilford.

            Muller, P. & Davies, R. (2004). Reading Rescue® in Inner City Schools: An Experimental Study Examining Reading Outcomes of a One-on-One Tutoring Intervention. Bloomington, IN: Indiana University, The Indiana Evaluation Center.

National Reading Panel, (2000). Report of the National Reading Panel: Teaching children to read: An evidence-based assessment of the scientific research literature on reading and its implications for reading instruction: Reports of the subgroups. Rockville, MD: NICHD Clearinghouse.

            Perfetti, C. (1992).  The representation problem in reading acquisition.  In P. Gough, L. Ehri, & R. Treiman (Eds.), Reading acquisition.  (pp. 107-143).  Hillsdale, NJ:  Erlbaum.

            Rack, J., Hulme, C., Snowling, M., & Wightman, J. (1994).  The role of phonology in young children learning to read words: The direct-mapping hypothesis. Journal of Experimental Child Psychology, 57, 42-71.

            Rasinski, T. (2003). The fluent reader: Oral reading strategies for building word recognition, fluency, and comprehension. New York: Scholastic.

            Santa, C., & Hoien, T. (1999). An assessment of Early Steps: A program for early intervention of reading problems. Reading Research Quarterly, 34, 54-79.

            Scarborough, H. (2001). Connecting early language and literacy to later reading (dis)abilities: Evidence, theory, and practice. In S. Neuman & D. Dickinson (Eds.), Handbook of early literacy research (pp. 97-110). New York: Guilford.

            Shanahan, T. (1998). On the effectiveness and limitations of tutoring in reading. Review of Research in Education, 23, 217-234.

            Shanahan, T. & Barr, R. (1995). Reading Recovery: An independent evaluation of the effects of an early instructional intervention for at-risk learners. Reading Research Quarterly, 30, 958-996.

            Shanker, J., & Ekwall, E.(2003). Ekwall/Shanker Reading Inventory (4th ed).  Needham Heights, MA: Allyn and Bacon.

            Share, D. (1999). Phonological recoding and orthographic learning: A direct test of the self-teaching hypothesis. Journal of Experimental Child Psychology, 72, 95-129.

            Share, D. (2004a). Orthographic learning at a glance: On the time course and developmental onset of self-teaching. Journal of Experimental Child Psychology, 87, 267-298.

            Share, D. (2004b). Knowing letter names and learning letter sounds: A causal connection. Journal of Experimental Child Psychology, 88, 213-233.

            Stanovich, K. (1986). Matthew effects in reading: Some consequences of individual differences in the acquisition of literacy. Reading Research Quarterly, 21, 360-406.

            Torgesen, J., Wagner, R., Rashotte, C., Rose, E., Lindamood, P., Conway, T., et al. (1999). Preventing reading failure in young children with phonological processing disabilities: Group and individual responses to instruction. Journal of Educational Psychology, 91, 579-593.

            Vellutino, F., Scanlon, D., Sipay, E., Small, S., Pratt, A., Chen, R. et al. (1996). Cognitive profiles of difficult-to-remediate and readily remediated poor readers: Early intervention as a vehicle for distinguishing between cognitive and experiential deficits as basic causes of specific reading disability. Journal of Educational Psychology, 88, 601-638.

            Voyager Expanded Learning (2004).  Voyager Passport.  Dallas, TX: Voyager Expanded Learning, Inc.

            Wasik, B., & Slavin, R. (1993). Preventing early reading failure with one-to-one tutoring: A review of five programs. Reading Research Quarterly, 28, 178-200.

            Woodcock, R. (1987). Woodcock Reading Mastery Tests-Revised. Circle Pines, MN: American Guidance Service.

            Woodcock, R. (l998).  Woodcock Reading Mastery Tests-Revised Examiner’s Manual Normative Update.  Circle Pines, MN:  American Guidance Service.

Footnotes

                  ¹Substantial letter knowledge is required to benefit from the RES program. Low-performing students who do not know at least 17 letters in the fall are typically enrolled in the program later in the year after they have learned letters in their classrooms. Because tutoring began late in this study, it was not possible to tutor a second wave of students.