2010 Outcome Evaluation Report: Virginia Reading First November 30, 2010 2010 Outcome Evaluation Report: Virginia Reading First Submitted to: Virginia Department of Education James Monroe Building 101 N. 14th Street Richmond, VA 23219 804.786.3925 Submitted by: Sonia Jurich, M.D., Ed.D., Project Director Michael Frye, M.S., Research Associate RMC Research Corporation 1501 Wilson Boulevard, Suite 1250 Arlington, VA 22209 Phone: 703.558.4000 Fax: 703.558.4823 TABLE OF CONTENTS List of Tables .................................................................................................................................. ii List of Figures ................................................................................................................................ iii Executive Summary ....................................................................................................................... iv Introduction ..................................................................................................................................... 1 2010 Assessment Results ................................................................................................................ 3 Stanford Reading First .................................................................................................................4 Aggregated Results (All Students) ...........................................................................................5 Disaggregated Results ..............................................................................................................9 Virginia Standards of Learning ..................................................................................................16 Analysis Over Time ...................................................................................................................... 19 Stanford Reading First ...............................................................................................................19 Virginia Standards of Learning ..................................................................................................23 Conclusions ................................................................................................................................... 28 Addressing the Research Questions ...........................................................................................28 Next Steps ..................................................................................................................................33 Virginia Reading First Outcome Evaluation – i LIST OF TABLES Table 1: Number of students participating in the 2010 Stanford Reading First by grade ........ 4 Table 2: Demographics of students participating in the 2010 Stanford Reading First ............. 5 Table 3: Percentage of students at grade level on the 2010 Stanford Reading First disaggregated by grade and gender ......................................................................................... 10 Table 4: Percentage of students at grade level on the 2010 Stanford Reading First disaggregated by grade and socioeconomic status.................................................................. 11 Table 5: Percentage of students at grade level in the 2010 Stanford Reading First disaggregated by grade and race/ethnicity .............................................................................. 13 Table 6: Percentage of students at grade level in the 2010 Stanford Reading First by grade and participation in special education ..................................................................................... 14 Table 7: Percentage of students at grade level in the 2010 Stanford Reading First by grade and LEP classification............................................................................................................. 15 Table 8: Demographics of the Virginia Reading First students taking the SOL English/ reading assessment for grade three ......................................................................................... 17 Table 9: Results of the 2010 SOL English/reading assessment for grade three disaggregated by student subgroups............................................................................................................... 18 Table 10: Effect Size for change in SOL English/reading assessment between 2004 and 2009 ................................................................................................................................................. 25 Virginia Reading First Outcome Evaluation – ii LIST OF FIGURES Figure 1: Results of the 2010 Stanford Reading First – multiple choice subtest ...................... 6 Figure 2: Results of the 2010 Stanford Reading First – kindergarten ...................................... 6 Figure 3: Results of the 2010 Stanford Reading First – first grade .......................................... 7 Figure 4: Results of the 2010 Stanford Reading First – second grade...................................... 8 Figure 5: Percentage of students at grade level in the 2010 Stanford Reading First ................ 8 Figure 6: Percentile ranks on the 2010 Stanford Reading First – multiple choice subtest ....... 9 Figure 7: Percentile ranks for the multiple choice subtest of the 2010 Stanford Reading First Assessment by grade and student subgroups .......................................................................... 16 Figure 8: Percentage of kindergarten students at grade level on the Stanford Reading First (from 2004 to 2010) ................................................................................................................ 20 Figure 9: Percentage of first-grade students at grade level on the Stanford Reading First (from 2004 to 2010) ................................................................................................................ 20 Figure 10: Percentage of second-grade students at grade level in the Stanford Reading First (from 2004 to 2010) ................................................................................................................ 21 Figure 11: Stanford Reading First average NCE Scores for students in Cohort 1 schools (between 2004 and 2009) ........................................................................................................ 22 Figure 12: Stanford Reading First average NCE Scores for students in Cohort 2 schools (between 2007 and 2010) ........................................................................................................ 22 Figure 13: Percentage of third-grade students passing the SOL English/Reading ................. 23 Figure 14: Percentage of third-grade students passing the SOL English/Reading - Cohort 2 schools..................................................................................................................................... 24 Figure 15: Average scaled scores for the 2010 SOL English/reading assessment third grade by years of attendance in Virginia Reading First Schools ...................................................... 26 Figure 16: Proficiency levels on the SOL English/reading assessment for grade three by years in schools ................................................................................................................................ 27 Virginia Reading First Outcome Evaluation – iii EXECUTIVE SUMMARY Reading First is a program from the U.S. Department of Education, authorized under the No Child Left Behind Act of 2001. The program provides grants to states to improve reading instruction for students from kindergarten to grade three through the use of scientifically-based reading programs. The Virginia Department of Education (VDOE) received a Reading First grant to begin implementing the program with the 2003-2004 academic year. Grant funds were used to hire a reading coach at each participant school, provide intensive professional development for the coaches and teachers, and purchase scientifically-based reading programs and interventions. VDOE Reading First reading specialists provided additional technical assistance, professional development, and monitoring. In 2008, RMC Research Corporation competed for, and was awarded a contract to conduct the evaluation of the program’s impact on student performance as measured by standardized reading assessments – the Stanford Reading First (SRF) for students from kindergarten to grade two, and the Virginia Standards of Learning (SOL), for students in grade three. This report reflects the analyses of student performance on the tests for the 2009-2010 academic year, the final year of the program. It also provides a longitudinal perspective of the impact of Reading First on student academic performance from the 2003-2004 to 2009-2010 academic years. The report complements the two previous evaluation reports that RMC provided to VDOE in October 2008 and November 2009. Accomplishments of the Virginia Reading First program during the 2009-2010 academic year included: A total of 21 schools in 17 school divisions served 5,339 students from kindergarten to grade three. Of these students, 60 percent were classified as economically disadvantaged (SIED), 47 percent were minorities, seven percent were students with limited English proficiency (LEP), and nine percent were students with disabilities (SWD). Virginia Reading First Outcome Evaluation –iv Students scoring at grade level in the Stanford Reading First (SRF), that is conducted with students from kindergarten to grade two, included 73 percent of kindergarteners, 66 percent of first-grade students, and 65 percent of second-grade students. 73 percent of the students in grade three in the Virginia Reading First schools passed the SOL English/Reading, with 44 percent scoring at the proficient level and 31 percent attaining advanced proficiency. The percentage of students passing the SOL English/reading assessment disaggregated by student subgroup was as follows: Girls –78 percent White – 81 percent Boys – 71 percent LEP – 56 percent African Americans – 67 percent SWD – 50 percent Hispanics – 64 percent SIED – 68 percent. In addition, a longitudinal perspective across the seven years of the program highlights areas of accomplishments and areas that require ongoing attention. Celebrating success Between spring 2004 and spring 2010, kindergarten results from the different components of the SRF kindergarten assessment increased between 13 and 29 percent. In the last two test years (2009 and 2010), Virginia Reading First kindergarten students scored on average at the 70th percentile on the SRF. Results on the SOL English/reading for students in grade three at Virginia Reading First schools also showed improvements. In 2004, 58 percent of third graders in Virginia Reading First schools passed the assessment, compared to 72 percent statewide, for a difference of 14 percentage points. In 2010, 75 percent of the third graders in Reading First schools passed the SOL English/reading assessment, compared to 83 percent statewide, an eight percentage point difference. Between spring 2004 and spring 2010, the percentage of students in third grade passing the SOL English/reading assessment grew by 29 percent in Reading First schools, compared to 15 percent statewide. Improved academic achievement was observed for all student subgroups, and the academic gap between minority and non-minority students declined, as gains in test results for minority students outpaced the gains for non-minority students. Virginia Reading First Outcome Evaluation –v Considering the schools that benefitted from six years in the program (cohort 1 schools), the effect size of Reading First on results for the SOL English/reading assessment was above 0.5 - a strong result for education programs that are not implemented under controlled conditions. The longer the students stayed in a Virginia Reading First school, the greater their likelihood to score at the advanced level on the grade 3 SOL English/reading. Looking forward More attention should be given to the teaching of vocabulary development and reading comprehension, the two components of the SRF that showed the weakest results across all three grade levels and all student subgroups. Achievement gaps on the grade 3 SOL English/reading decreased by 20 to 30 percent for students from minority and SIED backgrounds between 2004 and 2009. Although students from traditionally under-performing subgroups are showing improvement, they still lag behind their peers and the effort to help these students should not decline. Time enrolled in a Reading First school proved to have a direct positive relationship on student performance in the SOL. However, one third of students in third grade in a given year are new to the schools. The impact of mobility on student performance poses a significant challenge for schools in high poverty areas in Virginia as well as nationwide. Finding ways to minimize this effect is essential to sustain and expand the progress made by Virginia Reading First on the reading skills of Virginia students. In summary, Virginia Reading First schools showed strong improvements throughout the grant years. For these schools and school divisions, the challenge will be to maintain and expand these gains as grant funds expire. For the remaining schools, the challenge is to attain the success of the Reading First schools. At this time, the Virginia Department of Education, as many other state agencies across the nation, is actively involved in the development of a comprehensive literacy plan from birth to grade twelve. This is an important time to revisit the findings from the Reading First program, and re-examine these findings from the perspective of new advances in reading research. This merging of evidence derived from practice and research findings creates the perfect foundation for a plan that will ensure that all students graduate from Virginia schools with strong literacy skills. Virginia Reading First Outcome Evaluation –vi INTRODUCTION Reading First is a program from the U.S. Department of Education (USED) authorized by Title I, Part B, Subpart 1 of the Elementary and Secondary Education Act, as amended by the No Child Left Behind Act of 2001, Public Law 107-110. The objective of Reading First is to promote the use of scientifically-based reading programs and practices, instructional tools and assessments that are supported by research, with the ultimate goal of ensuring “that all children learn to read well by the end of third grade”.1 The Virginia Department of Education (VDOE) received a Reading First grant that spanned seven years, from 2003-2004 to 2009-2010 academic year. During this period, the Virginia Reading First program served more than 90 schools in 43 school divisions statewide. A total of 132,038 students participated in the program. Seventy percent of the Virginia Reading First schools were located in small towns or rural areas, 20 percent were located in middle-sized towns (population less than 250,000), and 10 percent in suburban (urban fringe) areas.2 The evaluation of the Virginia Reading First program was initially conducted by the University of Virginia. In fall 2007, VDOE issued a Request for Proposals for the evaluation of the program’s impact on student academic achievement. Two assessments were used to measure student performance in reading for students from kindergarten to second grade: the Stanford Reading First (SRF), and the Virginia Standards of Learning (SOL) English/reading assessment for students in grade three. RMC Research Corporation competed for and was awarded the contract. This is the last of three reports that RMC prepared for VDOE on the evaluation of Virginia Reading First. The first report presented findings from the analysis of the reading assessments conducted during spring 2008. The second report summarized findings from the 2009 assessment and discussed changes in assessment results from spring 2004 to spring 2009. This report includes an analysis of the reading assessments conducted in spring 2010 1 U.S. Department of Education, Office of Communications and Outreach, Guide to U.S. Department of Education Programs, Washington, D.C., 2009, p. 171. 2 Data retrieved on September 29, 2010, from http://readingfirstdataonline.org/awards.aspx Virginia Reading First Outcome Evaluation – 1 and revisits the longitudinal analysis presented in the 2009 report to integrate the most recent findings. The report comprises three sections: Section 1, 2010 Assessments, presents the results from the assessments conducted in spring 2010 with students from kindergarten through third grade. Section 2, Longitudinal Perspective, provides a descriptive analysis of changes in academic performance for Virginia Reading First students across the grant years. Section 3, Conclusions, addresses the evaluation questions and discusses next steps. The evaluators’ objective was to write a report that provides state, school division, and school staff with useful evidence regarding the impact of the Virginia Reading First program on student performance on standardized reading assessments. Graphics are used frequently to facilitate the explanation of findings. Results from the statistical tests are mentioned in the body of the report and included only when essential to explain a statement. Footnotes are used for further clarification or information on data sources without interrupting the flow of the narrative. The evaluators hope that the information in this report will be helpful to inform decisions on how to sustain the program gains and improve on its shortcomings as the Reading First students move to upper grades and new students enter Virginia schools eager to learn and succeed. Virginia Reading First Outcome Evaluation – 2 2010 ASSESSMENT RESULTS In the 2009-2010 academic year, 21 schools participated in the Virginia Reading First Program. This section provides information about results for the reading assessments that these schools conducted in spring 2010. Virginia Reading First schools used two assessments to test the reading skills of its students: the Stanford Reading First (SRF) and the Virginia SOL English/reading assessment. Stanford Reading First: The SRF is a norm-referenced test published by Pearson. The Virginia Reading First program used the test for students in kindergarten through second grade. The test comprises two subtests: multiple choice and oral fluency. The multiple choice subtest includes five components: phonemic awareness, phonics, vocabulary development, reading fluency, and reading comprehension. The oral fluency subtest comprises two components: speaking vocabulary and oral reading fluency. In addition to the multiple choice subtest, the students also responded to the speaking vocabulary component of the oral fluency subtest. The oral reading fluency was voluntary in Virginia and most schools did not use the test. Results for the multiple choice subtest are provided in scaled scores, normal curve equivalent, stanines, and percentiles. Based on the percentiles, three proficiency levels are determined: At Grade Level: students who score at or above the 40th percentile, Needs Additional Intervention: students who score between the 20th and the 40th percentiles, and Needs Substantial Intervention: students who score below the 20th percentile. Standards of Learning: The SOL English/reading assessment is a criterionreferenced achievement test that is conducted with students from grade three to high school. Results are provided in scaled scores and proficiency levels determined by the number of correct items. For the third-grade SOL English/reading assessment, the proficiency levels are determined as follows: Fail/Basic: 13 correct items out of 35 items; Pass/Proficient: 23 correct items out of 35 items; and Pass/Advanced: 31 correct items out of 35 items. Virginia Reading First Outcome Evaluation – 3 This section is divided into two subsections. The first subsection discusses the results for the SRF tests, and the second subsection presents the SOL English/reading assessment results. Stanford Reading First Description of the population The 21 schools receiving Reading First grants in the 2009-2010 academic year served a total of 4,059 students in kindergarten through grade two. Of these 4,059 students, 4,007 (98.7 percent) took the Stanford Reading First (SRF) multiple choice test and the speaking vocabulary part of the oral fluency test. Because the percentage of students who did not take the SRF was about one percent of student enrollment, an analysis of absentees’ demographics was not needed. Table 1 displays the number of students who took the SRF assessment in spring 2010 disaggregated by grade. Table 2 disaggregates the group of test-takers by grade and subgroups. As seen in the tables, the students taking the SRF in spring of 2010 were equally distributed by grade and had similar gender representation (about half males and half females). Regarding race/ethnicity, the majority of test-takers were white (52 percent), followed by students from African American (35 percent) and Hispanic (10 percent) origins. Students classified as Asian and “other” comprised about 3 percent of the test-taking population. The majority (59 percent) of the students were identified as economically disadvantaged (SIED); 7 percent were classified as having Limited English Proficiency (LEP), and 8 percent were students with disabilities (SWD). Table 1: Number of students participating in the 2010 Stanford Reading First by grade Grades (N) Total Categories K First Second Number Percent Total students per grade 1,330 1,369 1,360 4,059 100.0 Students taking the 2010 SRF 1,310 1,362 1,335 4,007 98.7 Virginia Reading First Outcome Evaluation – 4 Table 2: Demographics of students participating in the 2010 Stanford Reading First Grades (N) Total Subgroups K First Second Number Percent Female 664 677 656 1,997 49.8 Male 646 685 679 2,010 50.2 Asian 10 13 8 31 0.8 African American 433 462 499 1,394 34.8 Hispanic 123 130 140 393 9.8 White 705 727 663 2,095 52.3 Other 39 30 25 94 2.3 762 795 823 2,380 59.4 Students with disabilities (SWD) 76 118 140 334 8.3 Students with Limited English Proficiency (LEP) 69 102 109 280 7.0 1,310 1,362 1,335 4,007 Gender Race/ Ethnicity Students identified as economically disadvantaged (SIED) Total test-takers 100.0 Aggregated Results (All Students) Figure 1 displays the percentages of students who scored at the different performance levels in the multiple choice part of the SRF. The multiple choice subtest comprises five components: phonemic awareness, phonics, vocabulary development, reading fluency, and reading comprehension. As seen in the chart, the majority of students scored at grade level for all three grades tested with SRF. The percentage in need of intervention (additional or substantial) was greater for students in grades one and two when compared to kindergarten students. Virginia Reading First Outcome Evaluation – 5 Figure 1: Results of the 2010 Stanford Reading First – multiple choice subtest 100% 8% 15% 20% 21% 66% 65% Grade 1 (N=1,362) Grade 2 (N=1,335) 19% 80% Students (%) 14% 60% 40% 73% 20% 0% Kindergarten (N=1,310) At Grade Level Needs Additional Intervention Needs Substantial Intervention Figure 2 breaks down the components of the multiple choice test for kindergarten students, and includes results for the speaking vocabulary test, which is part of the oral fluency component. As shown in the graphic, the majority of students were at grade level for phonemic awareness (88 percent), phonics (76 percent), reading fluency (67 percent), and speaking vocabulary (64 percent). About half of the kindergarteners were at grade level for reading comprehension, while fewer than half (46 percent) were at grade level for vocabulary development. It must be noted that more than 20 percent of the kindergarten students are in need of substantial intervention for reading fluency and 40 percent are in need of additional intervention for vocabulary development. Figure 2: Results of the 2010 Stanford Reading First – kindergarten 100% 8% 14% 40% 11% 40% 60% 88% 13% 23% 14% 80% Students (%) 10% 76% 37% 14% 64% 67% 51% 46% 20% 18% 0% Phonemic Awareness At Grade Level Phonics Vocabulary Development Reading Fluency Needs Additional Intervention Reading Comprehension Speaking Vocabulary Needs Substantial Intervention Virginia Reading First Outcome Evaluation – 6 Figure 3 displays assessment results for students in first grade. As explained in the 2009 evaluation report, the weak results in the phonics test have been attributed to the lack of alignment between the content of the test and the Virginia English Standards of Learning, rather than a lack of phonics instruction within the schools. For the other components of the SRF, 60 percent or more of the first-grade students are scoring at grade level. It must be noted the percentages of first-grade students in need of substantial intervention in vocabulary development (21 percent) and reading fluency (19 percent). Figure 3: Results of the 2010 Stanford Reading First – first grade Students (%) 100% 6% 11% 25% 19% 17% 20% 61% 61% Vocabulary Development Reading Fluency 15% 80% 60% 40% 9% 12% 21% 29% 38% 82% 20% 73% 62% 37% 0% Phonemic Awareness Phonics At Grade Level Needs Additional Intervention Reading Comprehension Speaking Vocabulary Needs Substantial Intervention Figure 4 displays results for students in second grade. Except for phonemic awareness, where almost all students were at grade level, results for the other components were not as strong as for students in kindergarten and grade one. A little more than half of the students were at grade level in vocabulary development (54 percent), reading fluency (55 percent) and speaking vocabulary (55 percent). Additionally, the percentage of students in need of substantial intervention is high, and varies from 21 percent, in the speaking vocabulary component, to 46 percent in phonics. Virginia Reading First Outcome Evaluation – 7 Figure 4: Results of the 2010 Stanford Reading First – second grade Students (%) 100% 80% 28% 26% 18% 19% 54% 55% Vocabulary Development Reading Fluency 46% 60% 21% 31% 24% 23% 11% 96% 40% 43% 20% 55% 46% 0% Phonemic Awareness Phonics At Grade Level Needs Additional Intervention Reading Comprehension Speaking Vocabulary Needs Substantial Intervention Figure 5 compares results across grades, focusing on the percentage of students who were classified at grade level in each component of the test. Figure 5: Percentage of students at grade level in the 2010 Stanford Reading First 64% 62% Speaking Vocabulary 55% 51% Reading Comprehension 73% 46% 67% Reading Fluency 61% 55% 46% Vocabulary Development 61% 54% 76% Phonics 37% 43% 88% Phonemic Awareness 82% 96% 0% 20% Kindergarten 40% Grade 1 60% 80% 100% Grade 2 As suggested by the graphic, 82 percent or more of Virginia Reading First students, regardless of grade, were at grade level on the phonemic awareness component of the test. Virginia Reading First Outcome Evaluation – 8 Results are particularly strong for students in second grade. First-grade students had strong results in reading comprehension (73 percent, compared to 51 percent for kindergarten and 46 percent for second-grade students) and vocabulary development (61 percent, compared to 54 percent for kindergarteners and 46 percent for second-grade students). Students at grade level in phonics comprised the majority (76 percent) of kindergarteners, but fewer than half of the students in first and second grade. The mean scaled scores in the multiple choice section of the SRF were averaged and converted into percentiles to facilitate a comparison between results for Virginia Reading First students, and those who took the test nationwide. As shown in Figure 6, kindergarten students in Virginia Reading First schools scored, on average, at or above 72 percent of testtakers nationwide at the same grade level. Virginia Reading First first-grade students, on average, scored at or above 58 percent of their peers nationwide, while second-grade students scored about midway on the national sample (53th percentile). As a reminder, the 50th percentile is the expected average if the results follow a normal distribution. Therefore, the results in the multiple choice sub-test of the SRF suggest that Virginia Reading First kindergarteners are doing, on average, better in the test than peers nationwide, while students in first and second grades are doing about the same, on average, as their peers across the nation. These percentile ranks were similar to those found in the 2009 tests (73rd, 58th, and 54th percentiles, respectively). Figure 6: Percentile ranks on the 2010 Stanford Reading First – multiple choice subtest 72nd 58th 53th Kindergarten Grade 1 Grade 2 Disaggregated Results Analyses of the 2010 SRF results were conducted with three student subgroups that have greater representation among Virginia Reading First students: gender, economically disadvantaged students, and minority students. The numbers of students with disabilities and those classified as having Limited English Proficiency were too small to ensure reliable results. Virginia Reading First Outcome Evaluation – 9 Gender: A one-way analysis of variance (ANOVA) was conducted to examine the mean scores for boys and girls in the components of the SRF.3 The results of the test indicate that in kindergarten, girls and boys perform at similar levels in the SRF assessments for all components, except reading comprehension. For this component, girls outperformed boys. At first grade, girls outperformed boys in four of the six components, including phonemic awareness, reading fluency, reading comprehension, and speaking vocabulary. At second grade, girls outperformed boys in all components, except vocabulary development. Results in the vocabulary development component of the SRF were similar for boys and girls in all three grade levels. Table 3 displays the percentage of girls and boys who scored at grade level in the different components of the SRF by grades. A brief examination of the table confirms the results of the comparison of means test. In general, more girls than boys scored at grade levels in all the components of the test and at all grade levels. Table 3: Percentage of students at grade level on the 2010 Stanford Reading First disaggregated by grade and gender 52.4 65.8 664 Male 88.4 74.1 44.4 66.7 48.5 62.4 646 Total (N) Speaking Vocabulary 66.3 Reading Fluency 48.8 Vocabulary Development 77.6 Phonics 87.8 Phonemic Awareness Female Subgroup Reading Comprehension Percentage At Grade Level Kindergarten First Grade Female 83.9 37.5 61.6 63.8 77.0 62.2 677 Male 80.9 36.5 61.2 57.7 69.3 58.5 685 Second Grade Female 97.1 47.9 55.6 58.1 48.5 60.4 656 Male 95.3 39.0 52.3 52.7 44.2 49.8 679 3 All comparison of means tests for results on the six components of SRF used raw scores, as scale scores are provided only for the multiple choice subtest. Statistical significance was established at α= 0.05. The term “outperforms” indicates differences that are statistically significant between the groups being compared. Virginia Reading First Outcome Evaluation – 10 Students identified as economically disadvantaged: The comparison of means analysis found that the students who were not economically disadvantaged (Not-SIED) students outperformed their SIED peers in all components of the test and at all grade levels. Results were statistically significant even with α = 0.01.4 Table 4 presents the percentage of students who scored at grade level disaggregated by socioeconomic status. Two groups were included: students identified as economically disadvantaged (SIED) and students who were not economically disadvantaged (Not SIED). Table 4: Percentage of students at grade level on the 2010 Stanford Reading First disaggregated by grade and socioeconomic status Speaking Vocabulary 36.0 58.5 42.5 58.4 762 94.0 83.9 60.6 77.6 61.5 72.1 548 Vocabulary Development 70.1 Phonics 83.9 Phonemic Awareness Reading Comprehension Reading Fluency Percentage At Grade Level Total (N) Kindergarten SIED Not SIED First Grade SIED Not SIED 77.5 30.1 52.8 54.1 67.4 57.2 795 89.2 46.7 73.4 70.0 81.1 69.5 567 Second Grade SIED Not SIED 94.8 36.8 46.3 48.2 38.2 50.1 823 98.4 53.9 66.2 66.8 59.4 62.9 512 Race/ethnicity: A comparison of means analysis was conducted using an independent sample t-test to examine the differences in mean scores across the race/ethnic subgroup. Only three categories had sufficient student numbers for a reliable comparison: 4 A lower alpha indicates a greater probability that the same results will be obtained no matter how many times the test is repeated; in non-statistical terms, the lower alpha indicates that the gap in performance between the two groups is real and requires attention. Virginia Reading First Outcome Evaluation – 11 whites, African Americans, and Hispanics. Results of the test indicated that mean scores for students identified as white were significantly higher than mean scores for African American students across all components and grades, except for phonemic awareness and reading fluency at kindergarten, and speaking vocabulary at first grade. In these three cases, mean scores for the African American students were similar to mean scores for white students. The comparison between scores for Hispanics and white students showed that mean scores for white students tended to be significantly higher than mean scores for Hispanic students, except for phonemic awareness in first and second grades. For these two components, the average scores for Hispanic students were similar to the average scores for white students. Table 5 displays the percentages of students scoring at grade level disaggregated by grade and race/ethnicity. The percentages in the table indicate how many students from a specific race/ethnicity scored at grade level compared to the total enrollment in grade for that same race/ethnicity. For instance, 86 percent of the African American students in kindergarten scored at grade level in phonemic awareness; 85 percent of all Hispanic students in first grade scored at grade level in phonemic awareness, and so on. Virginia Reading First Outcome Evaluation – 12 Table 5: Percentage of students at grade level in the 2010 Stanford Reading First disaggregated by grade and race/ethnicity Speaking Vocabulary 90.0 60.6 100.0 80.8 70.0 10 86.1 71.6 39.0 64.4 47.3 60.7 433 78.0 72.4 24.4 58.5 29.3 50.4 123 White 91.9 79.3 55.2 69.5 56.7 69.2 705 Other 74.4 69.2 30.8 51.3 30.8 51.3 39 Reading Fluency Vocabulary Development 80.0 Phonics Asian African American Hispanic Phonemic Awareness Reading Comprehension Percentage At Grade Level Total (N) Kindergarten First Grade Asian African American Hispanic 92.3 69.2 76.9 69.2 84.6 61.5 13 76.4 27.7 49.4 53.7 67.6 59.7 462 85.4 33.1 46.2 56.2 63.1 53.8 130 White 85.4 43.3 71.3 65.7 77.6 65.3 727 Other 83.3 33.3 66.7 63.3 86.7 66.7 30 Second Grade Asian African American Hispanic 100.0 50.0 75.0 62.5 62.5 100.0 8 95.4 33.5 46.5 44.3 33.9 47.3 499 95.7 39.3 21.4 44.3 37.9 50.7 140 White 97.3 51.3 66.1 66.1 57.2 62.0 663 Other 84.0 52.0 56.0 52.0 48.0 32.0 25 Students with disabilities: Table 6 displays the percentage of students with disabilities (SWD) and students who do not receive special education services (Not SWD). As seen in the table, SWD students are less likely to score at grade level, when compared to Not SWD students. As a group, the SWD students seem to struggle particularly in reading comprehension and speaking vocabulary beginning in kindergarten, and their difficulties in these two areas appear to remain as they move to upper grades. Virginia Reading First Outcome Evaluation – 13 Table 6: Percentage of students at grade level in the 2010 Stanford Reading First by grade and participation in special education Speaking Vocabulary 64.5 32.9 61.8 28.9 48.7 76 Not SWD 88.8 76.6 47.1 66.8 51.8 65.1 1,234 Vocabulary Development 76.3 Phonics SWD Phonemic Awareness Reading Comprehension Reading Fluency Percentage At Grade Level Total (N) Kindergarten First Grade SWD 58.5 20.3 35.6 39.0 55.9 39.0 118 Not SWD 84.6 38.6 63.8 62.8 74.8 64.5 1,244 Second Grade SWD 88.6 23.6 45.7 42.1 30.0 43.6 140 Not SWD 97.1 45.7 54.9 56.9 48.2 56.3 1,195 When analyzing results for students with disabilities, one must take into account the fact that these students differ considerably in the type and intensity of their disabilities, and how the disability affects their cognitive abilities. Comparisons between such a diverse group and another group that is more homogenous are not reliable or helpful. For this reason, a comparison of means test was not conducted. Limited English Proficient Students: Table 7 displays the percentages of students who scored at grade level in the SRF test disaggregated in two subgroups: students with limited English proficiency (LEP) and students proficient in English (Not LEP). As the table shows, LEP and Not LEP did well in the phonemic awareness test at all three grade levels. In all other components, LEP students were less likely to score at grade level when compared to their Not LEP peers. Vocabulary development was the component of the SRF where LEP students had the greatest difficulty in all three grade levels, but particularly in the second grade. Caution must be exercised when interpreting results for LEP students. First, the small number of these students does not allow for reliable comparisons with results from Not LEP Virginia Reading First Outcome Evaluation – 14 students. Second, students under this classification represent a range of levels of proficiency in the English language. Table 7: Percentage of students at grade level in the 2010 Stanford Reading First by grade and LEP classification Speaking Vocabulary 23.2 58.0 29.0 37.7 69 Not LEP 89.3 76.5 47.7 67.1 51.6 65.9 1,241 Reading Fluency 66.7 Vocabulary Development 75.4 Phonics LEP Phonemic Awareness Reading Comprehension Percentage At Grade Level Total (N) Kindergarten First Grade LEP 82.4 32.4 40.2 52.9 60.8 48.0 102 Not LEP 82.4 37.7 63.7 61.6 74.2 63.9 1,260 Second Grade LEP 94.5 31.2 18.3 34.9 29.4 48.6 109 Not LEP 96.3 44.8 57.7 57.6 48.2 55.9 1,226 Summary: Figure 7 displays the results of percentile ranks [converted from the mean scaled score] for the multiple choice test by grade and student subgroups. As seen in the table, kindergarten students scored above the 50th percentile on the multiple choice subtest for all student subgroups. For first-grade students, only students receiving special education services (SWD) scored below the 50th percentile. For second-grade students, three subgroups scored at or below the 40th percentile: Hispanics (40th percentile), LEP (33rd percentile), and SWD (37th percentile). Virginia Reading First Outcome Evaluation – 15 Figure 7: Percentile ranks for the multiple choice subtest of the 2010 Stanford Reading First Assessment by grade and student subgroups 100 90 80 Percentile 70 60 50 40 30 20 10 0 Kindergarten Grade 1 Grade 2 Note: The dotted line indicates the 40th percentile – the cut score for “at grade level.” Virginia Standards of Learning In the 2009-2010 academic year, Virginia Reading First schools served 1,346 thirdgrade students. Of these, 1,332 (99 percent) took the Standards of Learning (SOL) English/reading assessment in spring 2010. Table 8 displays the demographics of the students who participated in the test in spring 2010. Virginia Reading First Outcome Evaluation – 16 Table 8: Demographics of the Virginia Reading First students taking the SOL English/ reading assessment for grade three Subgroups Number Percentage Female 680 51.1 Male 652 48.9 African American 462 34.7 13 1.0 Hispanic 114 8.6 White 726 54.5 Other 17 1.3 LEP 105 7.9 SWD 140 10.5 SIED 807 60.6 1,332 100.0 Gender Asian Race/ ethnicity Total (all test takers) Of the 1,332 third-grade students who took the 2010 Standards of Learning (SOL) English/reading assessment, 992 (75 percent) passed the test, with 44 percent demonstrating proficiency and 31 percent demonstrating advanced proficiency. Statewide, 83 percent of the students in third grade passed the Standards of Learning (SOL) English/reading assessment, with 43 percent demonstrating proficiency, and 41 percent demonstrating advanced proficiency. Table 9 displays the assessment results disaggregated by student demographics. As seen in the table, half or more of the students in all subgroups passed the test. The majority of students who failed the test scored at the basic level, with fewer than 10 percent scoring below basic, except for students with disabilities (13 percent below basic). It is of note that more than one third of the girls and white students attained advanced proficiency. Students identified as economically disadvantaged did equally well, with close to 70 percent passing the test. Virginia Reading First Outcome Evaluation – 17 Table 9: Results of the 2010 SOL English/reading assessment for grade three disaggregated by student subgroups Percentages1 Pass Subgroups Gender Race/ Ethnicity2 LEP SWD SIED Female Male Black Hispanic White Adv. Prof. Total Basic 34.7 27.0 19.7 21.9 39.4 19.0 14.3 23.3 42.8 44.3 47.0 42.1 41.6 37.1 35.7 45.0 77.5 71.3 66.7 64.0 81.0 56.1 50.0 68.3 20.3 22.9 27.9 27.2 16.9 34.3 37.1 26.5 Fail Below Basic 2.2 5.8 5.4 8.8 2.1 9.5 12.9 5.2 Total Total (N) 22.5 28.7 33.3 36.0 19.0 32.8 50.0 31.7 680 652 462 114 726 105 140 807 1 Percentages calculated as the number of students within subgroups at a specific performance level relative to the total number of students in that same subgroup (e.g., number of girls scoring advanced divided by total number of girls). 2 Information is provided for groups with 30 or more students. Virginia Reading First Outcome Evaluation – 18 ANALYSIS OVER TIME This section is divided into two subsections. The first focuses on the impact of Reading First on students in grades K-2, as measured by the Stanford Reading First (SRF) assessments. The second focuses on the impact of the program on students in grade three, as measured by the Virginia Standards of Learning English/Reading assessment. An explanation about the types of analyses used and their limitations is included in the 2009 evaluation report. The 2009 report also included an analysis of the association between scores on SRF and SOL. That analysis found a weak relationship between the two tests. A new analysis was done this year with similar results. Since no new information can be presented, the analysis was omitted in this report. Stanford Reading First Figures 8 to 10 display the percentages of Virginia Reading First students at grade level in the different components of the SRF across the grant years. The numbers in parentheses at the horizontal axis represent the number of students who took the SRF, while the vertical axis reflects the percentage of students who scored at grade level in the different components of the test. Only the five components that comprise the multiple choice subtest are included. Figure 8 focuses on students in kindergarten. Kindergarten students showed steady progress in all components of the test across the years, with the greatest increase occurring in the first year. Between 2004 and 2010, the percentage of kindergarten students at grade level in the SRF assessment increased by 13 percent in phonemic awareness, 22 percent in phonics, 24 percent in vocabulary development, 19 percent in reading fluency, and 29 percent in reading comprehension. Vocabulary development and reading comprehension, which were the components with the lowest percentages of students at grade level in all of the years, showed the greatest improvements. The phonemic awareness component had the strongest results, with 80 percent or more of the students scoring at grade level across the years. Virginia Reading First Outcome Evaluation – 19 Figure 8: Percentage of kindergarten students at grade level on the Stanford Reading First (from 2004 to 2010) 100% 80% 60% 40% 20% 0% 2004 (4,398) 2005 (3,471) 2006 (4,436) 2007 (5,704) 2008 (5,727) 2009 (5,200) 2010 (1,310) Phonemic awareness Reading fluency Phonics Reading comprehension Vocabulary development Figure 9 displays the percentage of students in first grade who scored at grade level on the SRF between 2004 and 2010. Growth in test performance for first-grade students was negligible, except for phonics. Between 2004 and 2010, the percentage of first-grade students who scored at grade level on the SRF test increased by two percent in vocabulary development, three percent in phonemic awareness, four percent in reading comprehension, six percent in reading fluency, but 25 percent in phonics. Figure 9: Percentage of first-grade students at grade level on the Stanford Reading First (from 2004 to 2010) 100% 80% 60% 40% 20% 0% 2004 (4,150) 2005 (3,852) 2006 (4,562) 2007 (5,723) 2008 (5,719) 2009 (5,303) 2010 (1,362) Phonemic awareness Reading fluency Phonics Reading comprehension Vocabulary development Virginia Reading First Outcome Evaluation – 20 Figure 10 displays the percentage of second-grade students who scored at grade level on the SRF. It is of note the difference in scores for the phonemic awareness component, when compared to the other components. While 90 percent or more of the students scored at grade level in the phonemics awareness component across the years, about 60 percent or fewer scored at grade level on the other components. Between 2004 and 2010, growth was nonexistent or minimal for reading fluency (no growth), phonemic awareness (three percent), and vocabulary development (six percent). At reading comprehension, the scores showed a growth of nine percent between the beginning and end of the Virginia Reading First program. Although phonics was the component with the weakest performance, it showed the greatest growth (24 percent). Figure 10: Percentage of second-grade students at grade level in the Stanford Reading First (from 2004 to 2010) 100% 80% 60% 40% 20% 0% 2004 (4,294) 2005 (3,214) 2006 (4,411) 2007 (5,552) 2008 (5,558) 2009 (5,375) 2010 (1,335) Phonemic awareness Reading fluency Phonics Reading comprehension Vocabulary development Figure 11 displays the average normal curve equivalent (NCE) scores for students in cohort 1 schools. Cohort 1 comprised 63 schools that implemented Reading First from 2004 to 2009 (their last year in the program). The analysis used the average NCE scores for the multiple choice section of the SRF. The NCE is an equal-interval scale with scores dispersed on a normal distribution curve (mean=50; standard deviation = 21.06). Students who score 50 NCEs are considered at grade level. A change of seven NCEs corresponds approximately to one grade level. Scores for the 2004 SRF multiple choice subtest did not include NCE and therefore, the analysis starts with the 2005 results. As seen in the graphic, increases in scores were small. Between 2005 and 2009, mean scores increased from 56 to 60 NCEs for students Virginia Reading First Outcome Evaluation – 21 in kindergarten; from 52 to 54 NCEs for students in first grade; and no increase for students in grade two. Figure 11: Stanford Reading First average NCE Scores for students in cohort 1 schools (between 2004 and 2009) 80 70 NCE 60 50 40 30 20 2005 2006 2007 Kindergarten 2008 Grade 1 2009 Grade 2 A similar analysis was conducted with cohort 2 schools. Cohort 2 includes 19 schools that implemented the program from 2007 to 2010. (Two additional schools entered the program in 2008-2009 and are not included in the analysis depicted in Figure 12, below.) As displayed in Figure 12, between the beginning (2007) and end (2010) of the Reading First program for cohort 2 schools, NCE scores increased five points on average for students in kindergarten, from 55 to 60 NCEs, from 51 to 53 NCEs for students in first grade, and no change for students in second grade. Figure 12: Stanford Reading First average NCE Scores for students in cohort 2 schools (between 2007 and 2010) 80 70 NCE 60 50 40 30 20 2007 2008 Kindergarten 2009 Grade 1 2010 Grade 2 Virginia Reading First Outcome Evaluation – 22 Virginia Standards of Learning Gains across years: Figure 13 shows results for the third-grade Standards of Learning (SOL) English/reading assessment for cohort 1 of Virginia Reading First schools and statewide schools.5 Increases in the percentage of third-grade students who passed this assessment were particularly large between the first and second grant years. In 2004, 58 percent of the students in the cohort 1 schools passed the test, compared to 70 percent in 2005. Between 2004 and 2009, the last grant year for cohort 1 schools, the percentage of third-grade students who passed the SOL reading assessment increased by 43 percent in Reading First schools, compared to 19 percent statewide. Figure 13: Percentage of third-grade students passing the SOL English/Reading assessment Passing+Advanced 100% 80% 60% 77% 78% 72% 71% 70% 80% 76% 84% 81% 86% 83% 2008 2009 58% 40% 20% 0% 2004 2005 State 2006 2007 Cohort 1-Reading First Figure 14 displays the same information for cohort 2 schools. Cohort 2 includes the 19 schools that received Reading First grants in 2007 and remained in the program until 2010. (Two additional schools entered the program in 2008-2009 and are not included in the analysis depicted in Figure 14 below.) Gains for cohort 2 schools were small and comparable to statewide gains. It is important to observe that the cohort 2 schools had no more than four 5 Statewide data retrieved October 1, 2010 from http://www.doe.virginia.gov/statistics_reports/ school_report_card/index.shtml Virginia Reading First Outcome Evaluation – 23 years in the program. The different results for the two cohorts highlight the influence of time in program on student academic performance. Figure 14: Percentage of third-grade students passing the SOL English/Reading assessment - Cohort 2 schools 100% Passing+Advanced 80% 80% 84% 75% 81% 86% 85% 83% 77% 60% 40% 20% 0% 2007 2008 State 2009 2010 VRF Effect size: Although the Virginia Reading First schools fell short of the level of statewide performance, the data suggest strong improvements in the schools that received full program implementation (cohort 1 schools). To assess the magnitude of this improvement over the years, an effect size was calculated. Effect size is a statistical concept that measures the strength of the relationship between two variables during the period of treatment. In this case, the variables are student performance on the Standards of Learning (SOL) English/ reading assessment in the treatment (Virginia Reading First schools) and comparison (statewide) groups. To measure the effect size, a Cohen’s h using transformed population proportions was calculated as follows: ES = 2*arcsin(√𝑝1) - 2*arcsin(√𝑝2 ) where 𝑝1is the proportion of students proficient on the Standards of Learning (SOL) English/ reading assessment in the last grant year (2009), and 𝑝2 is the proportion of students in first year of Reading First program (2004)]. Results are displayed in Table 10. Virginia Reading First Outcome Evaluation – 24 Table 10: Effect Size for change in SOL English/reading assessment between 2004 and 2009 Schools Proportion Difference Effect Size Effect Virginia Reading First – Cohort 1 43% 0.561 Medium State 19% 0.349 Small The effect size for change on the third grade SOL reading assessment is higher for the Virginia Reading First schools (ES = 0.561) than schools statewide (ES = 0.349). The effect size of change on the SOL in the Virginia Reading First schools is interpreted as medium, while change at the state level is interpreted as small. It is of note that effect sizes in situations where an effective treatment is not administered in a controlled setting tend to be small. Therefore, the schools that participated for six years in the Virginia Reading First program demonstrated considerable improvement in SOL results. Time in program (students): Students who took the Standards of Learning (SOL) English/reading assessment as third graders in 2010 were tracked back to their first year in Virginia Reading First schools. Each third-grade student who took the 2010 test fit in one of four categories: (1) Students who were in the school for four years and took the SRF in spring 2007 as kindergarteners (591 students); (2) Students who were in the school for three years and took the SRF for the first time as first graders in spring 2008 (143 students); (3) Students who were in the school for two years and took the SRF for the first time as second graders in spring 2009 (198 students); and (4) Students who entered the school in grade three in 2010 (400 students). Students who were retained in a grade or those who exited the school and might have returned in a later year were not included in the analysis. Figure 15 displays the average scaled score for the 2010 SOL English/reading assessment for each of the four groups. As seen in the graphic, the mean scaled score increased as the number of years in a Virginia Reading First school increased. The average SOL scaled score for students who attended a Virginia Reading First school for four years was significantly higher than students who had attended for three or fewer years. Virginia Reading First Outcome Evaluation – 25 Figure 15: Average scaled scores for the 2010 SOL English/reading assessment third grade by years of attendance in Virginia Reading First Schools 474.6 Scaled scores 480 470 452.0 460 450 446.8 438.9 440 430 420 410 400 1 year 2 years 3 years 4 years* *Tukey’s HSD post-hoc analysis; significance at α = 0.05 The analysis was then repeated using proficiency levels. Figure 16 displays the percentage of students who scored at the proficient level and those who scored at the advanced level. As suggested by the graphic, the longer the students remained in a Virginia Reading First school, the greater their likelihood of passing the Standards of Learning (SOL) English/reading assessment. Specifically, these students were more likely to score at the advanced level than students who had been in Virginia Reading First schools two or fewer years. The percentage of four-year students scoring advanced on the SOL is statistically significantly higher than students who attended Virginia Reading First schools for two years or less.6 Statistical significance was calculated at α = 0.05. The alpha was adjusted for multiple comparisons using four groups and calculated to be α* = 0.017. Logistic regression analysis controlled for differences of race/ethnicity, SIED status and SWD status. 6 Virginia Reading First Outcome Evaluation – 26 Figure 16: Proficiency levels on the SOL English/reading assessment for grade three by years in schools 4 years 42.8% 3 years 38.6%* 39.2% 2 years 30.8% 49.0% 1 year 23.2% 43.5% 0% 10% 20% 23.5% 30% Pass 40% 50% 60% 70% 80% 90% Advanced NOTE: (*) indicates that the difference between this and other proportions is statistically significant at α = 0.05 Virginia Reading First Outcome Evaluation – 27 CONCLUSIONS In 2010, the last year of the Virginia Reading First program, 21 schools in 17 school divisions served 5,339 students from kindergarten to grade three. Of these, 60 percent were classified as economically disadvantaged, 47 percent were minorities, seven percent were students with limited English proficiency, and nine percent were students with disabilities. Results on the Stanford Reading First (SRF) showed that students scoring at grade level comprised 73 percent of kindergarteners, 66 percent of first-grade students, and 65 percent of students in second grade. For the three grades, the two weakest components in the SRF were vocabulary development and reading comprehension. Alternatively, phonemic awareness was the strongest component with 80 to 96 percent of the students scoring at grade level. Scores on the phonics component should be taken with caution as this component of the SRF is not aligned with the Virginia Standards of Learning, particularly for grade one. When scores are disaggregated by subgroups, scores for boys and girls were similar at kindergarten, but differences start to appear as they move to upper grades. In grade two, girls outperformed boys in almost all components of the SRF. Although the academic gap between white students and minority students (African Americans and Hispanics) still remained, a decrease in the gap was noted. Vocabulary and reading comprehension were the two components that require the most attention for SIED, Hispanics, African Americans, and LEP students. Reading fluency is another area where students with disabilities need further help. Seventy-three percent of the students in grade three in the Virginia Reading First schools passed the Standards of Learning (SOL) English/reading assessment. Of these, 44 percent scored at the proficient level and 31 percent attained the advanced level. The majority (60 percent or more) of the students passed the test in almost all subgroups, except for students with disabilities and students with limited English proficiency. For these two groups, the passing rate was 50 percent and 56 percent, respectively. Addressing the Research Questions 1. What are the demographic characteristics of students in Virginia Reading First schools? Virginia Reading First Outcome Evaluation – 28 Virginia Reading First schools served a large number of students that are traditionally less likely to obtain academic success. In the last year of the grant (2009-2010 academic year), the Virginia Reading First program involved 21 schools with 5,339 students from kindergarten to grade three. Of these, 60 percent were classified as economically disadvantaged, and 47 percent were students from minority backgrounds. Across the grant period, from September 2003 through June 2010, Virginia Reading First schools served a total of 132,038 students; 56 percent were classified as economically disadvantaged, 9.8 percent were students identified as having disabilities, and four percent were identified as having limited English proficiency. Minority students comprised about half of the total student population. 2. What are the student characteristics in Virginia Reading First schools compared to Virginia’s overall population of students in kindergarten to grade three? Compared to Virginia’s overall student population for the same age group, Virginia Reading First schools were more likely to serve minority students and students classified as economically disadvantaged. In 2010, 53 percent of the students from kindergarten to grade three in Virginia Reading First schools were white, while 47 percent were students from minority background, particularly African Americans (35 percent), and Hispanics (six percent). Statewide, for the same age group, the percentages were: 55 percent white, 24 percent African American, and 11 percent Hispanics. Regarding SIED, in 2004, 57 percent of third graders in Virginia Reading First schools were classified as economically disadvantaged, compared to 62 percent in 2010. 3. What percentage of students in kindergarten to grade two in schools receiving Reading First subgrants achieved proficiency on the grade-level outcome measure in each year? Between 2004 and 2010, the percentage of students at grade level in the kindergarten SRF assessment increased by 13 percent in phonemic awareness, 22 percent in phonics, 24 percent in vocabulary development, 19 percent in reading fluency, and 29 percent in reading comprehension. By spring 2010, 88 percent of kindergarteners were scoring at grade level in phonemic awareness, 76 percent were at grade level in phonics, 67 percent in reading fluency, while about half were at grade level in vocabulary development and reading comprehension. For first graders, the percentage of students at grade level during the same period increased by 25 percent in phonics, six percent in reading fluency, four percent in reading Virginia Reading First Outcome Evaluation – 29 comprehension, three percent in phonemic awareness, and two percent in vocabulary development. By spring 2010, 82 percent of first-grade students were scoring at grade level in phonemic awareness, 73 percent were at grade level in reading comprehension, 61 percent were at grade level in reading fluency, and 61 percent were at grade level in vocabulary development. Results in phonics (37 percent at grade level) are not reliable, as the phonics component of SRF does not align with the Virginia Standards of Learning for grade one. At second grade, between 2004 and 2010, the percentage of students at grade level increased by 24 percent for phonics, nine percent in reading comprehension, six percent in vocabulary development, three percent in phonemic awareness, but no change in average for reading fluency. By spring 2010, 96 percent of the students in grade two were scoring at grade level in phonemic awareness, 55 percent in reading fluency, 54 percent in vocabulary development, 46 percent in reading comprehension and 43 percent in phonics. Despite the low score in phonics, this was the SRF component with the greatest improvement across the years. 4. How did student achievement in grade three compare between schools receiving Reading First subgrants and the state average? Third-grade students made strong improvements on the Standards of Learning (SOL) English/reading assessment. In 2004, 58 percent of third graders in Virginia Reading First schools passed this assessment, compared to 72 percent statewide. In 2010, 75 percent of the third graders in Reading First schools passed the Standards of Learning (SOL) English/reading assessment, compared to 83 percent statewide, an eight percentage point difference. The increase in percentage of students passing the assessment between 2004 and 2010 was 29 percent for Virginia Reading First Schools, compared to a 15 percent increase statewide. A calculation of effect size on the SOL results across the time of the program showed that improvements in reading for students in Virginia Reading First schools were stronger than those for students statewide. 5. How did reading achievement on the SRF assessment compare to national norms provided by Harcourt? In 2010, students in kindergarten scored, on average, at or above 73 percent of kindergarten students nationwide in the multiple choice component of the SRF; first-grade students scored at or above 58 percent of first graders taking the test nationwide, and second Virginia Reading First Outcome Evaluation – 30 graders scored at or above 53 percent of the national sample of second graders. These results were very similar to results for the 2009 SRF assessment. Across seven years of the Virginia Reading First program, kindergarten students scored above the national norm, while students in first and second grades scored at the norm provided by Harcourt. 6. How did student performance in Reading First schools vary by subgroup? Regarding gender, kindergarten girls and boys performed similarly on the SRF, except for the reading comprehension component, where girls outperformed boys.7 In grade one, girls outperformed boys in all but two components of the test (vocabulary development and phonics). At grade two, girls outperformed boys in all components but one (vocabulary development). At grade three, girls outperformed boys in the Standards of Learning (SOL) English/reading assessment; mean scaled score for girls was 465 and for boys the mean scaled score was 449. Results for students identified as economically disadvantaged (SIED) were compared to the test results for students who were not identified as economically disadvantaged (Not SIED). Not SIED outperformed SIED in all grade levels for both the SRF and SOL assessments. The SIED mean scale in the grade 3 SOL was 442 and the Not SIED was 482. An analysis of data from Cohort 1 schools (the schools that were in the program for six years) showed a decrease in the achievement gap between SIED and Not SIED students. On the 2006 SOL English/reading assessment,8 68.9 percent of Not SIED passed the assessment, compared to 51.7 percent for SIED. On the 2009 SOL, the last year for Cohort 1 schools, 89 percent of Not SIED and 76.4 percent of SIED passed the SOL English/reading assessment. Therefore, not only did both subgroups improve their level of performance, but the achievement gap decreased by 30.2 percent. Students with disabilities (SWD) had their assessment results compared to students who did not receive special education services (Not SWD). Not SWD outperformed their SWD peers in all grades and assessments; their mean scaled score was 463 for the SOL, compared to a 404 mean scaled score for the SWD. The word “outperformed” is used here to indicate statistical significant differences in test results for the subgroups, with a 95 percent confidence level. 8 SOL data from years prior to 2006 were not disaggregated for SIED and Not SIED students. 7 Virginia Reading First Outcome Evaluation – 31 Regarding race/ethnicity, the analysis focused on the three largest groups within this subgroup: African Americans, Hispanics, and whites. White students tended to outperform African American students in most components of the SRF, except phonemic awareness and reading fluency at kindergarten, and speaking vocabulary at first grade. In these three cases, mean scores for African American and white students were similar. White students also outperformed Hispanic students in nearly all components, except in the phonemic awareness test for first and second grades. In these two tests, Hispanic and white students had similar scores. An analysis of data from Cohort 1 schools found a decrease in the achievement gap between white students and their African American and Hispanic peers for the grade 3 SOL English/reading. In the 2004 SOL, 66.8 percent of the white students passed the test, compared to 51.4 percent of African Americans and 52.4 percent of Hispanics. In the 2009 SOL, 87.5 percent of the white students passed the test, compared to 75.5 percent of African Americans and 77.8 percent of Hispanics. Therefore, the achievement gap on the SOL between African American and white students decreased by 22.1 percent, while the gap between Hispanic and white students decreased by 32.6 percent. 7. Has student achievement in all subgroups shown consistent improvement over time? Focusing solely on the statewide assessment, the percentage of students with disabilities in Virginia Reading First schools who passed the SOL English/reading assessment increased by 24 percent between 2004 and 2010. In 2010, 73 percent of the students with disabilities in Virginia Reading First schools had passed the Standards of Learning (SOL) English/reading assessment, compared to 53 percent in 2005. A decrease in the academic gap was also observed for students from minority backgrounds, as described above. Time spent in a Virginia Reading First school had a positive impact on student outcome, as students who had been in the schools for four years showed stronger improvements on the test compared to students who had been in the schools for two or fewer years. The longer the students stayed in the program, the greater their likelihood to score at the advanced level. Mobility is a reality for these schools, though. For instance, in 2010, one third of the third-grade classes in Virginia Reading First schools were comprised of students who had enrolled in the school for the first time that year. Virginia Reading First Outcome Evaluation – 32 Next Steps The analyses of reading scores for Virginia Reading First schools provided evidence of the impact of the program on the reading skills of all participant students, but particularly those students who are most at risk for academic failure. Evidence of this success is the improvement in scores, particularly for the Virginia Standards of Learning (SOL) English/ reading assessment. The VDOE, school divisions, and schools should celebrate these results, while ensuring that the instructional strategies learned during the Reading First program are not forgotten as the grant funds expire. While celebrating success is important, it is equally important to revisit the results of the evaluation reports to understand where further efforts are required or where changing of strategies may be needed. One area that deserves attention is the discrepancy between results on the phonemic awareness component of the SRF and the vocabulary development and reading comprehension components. While it is elating to see that 96 percent of students in second grade are at grade level in phonemic awareness, it is of concern that fewer than half of the same students are at grade level in reading comprehension. It is true that phonemic awareness is a foundational skill for reading, but without a strong vocabulary and comprehension of reading passages, students will have difficulty reading in the upper grades, where more complex skills are required. Likewise, it is invigorating to see that academic gaps between the different student subgroups are being decreased. However, students that are economically disadvantaged still lag behind their peers from upper socioeconomic strata, at least regarding their achievement in standardized assessments. At this time, the Virginia Department of Education, like many other state agencies across the nation, is actively involved in the development of a comprehensive literacy plan birth to grade twelve. This is an important time to revisit the findings from the Reading First program, and re-examine these findings from the perspective of new advances in reading research. This merging of evidence derived from practice and research findings can help create strong foundations for a plan that will ensure that all students graduate from Virginia schools with strong literacy skills. Virginia Reading First Outcome Evaluation – 33