Coursehero >> New Jersey >> Princeton >> ERS 2005

Course Hero has millions of student submitted documents similar to the one below including study guides, homework solutions, papers, exam answer keys and textbook solutions.

Effects The of a Voluntary Summer Reading Intervention on Reading Achievement: Results from a Randomized Field Trial James S. Kim University of California, Irvine 6-30-2006 ABSTRACT The effects of a voluntary summer reading intervention were assessed in a randomized field trial involving 552 students in 10 schools. In this study, fourth-grade children received 8 books to read during summer vacation, and were encouraged by their teachers to practice oral reading at home with a family member and to use comprehension strategies during independent, silent reading. Reading lessons occurred during the last month of school in June, and 8 books were mailed to students on a biweekly basis during July and August. The estimated treatment effects on a standardized test of reading achievement (Iowa Test of Basic Skills) were largest for students who reported owning fewer books at home, less fluent readers, and minority students. These findings suggest that a voluntary summer reading intervention may represent a scaleable and cost-effective policy for improving reading achievement among lower-performing students. 1 Numerous empirical studies indicate that the achievement gap in reading forms and widens during summer vacation rather than during the school year. In a study of summer learning in Atlanta, Heyns (1978) found that the gap between black and white children, and between low- and high-income children widens disproportionately during the months when schools are not in session (p. 187). A synthesis of studies on summer learning loss (Cooper et al., 1996) showed that middle-income students enjoyed reading gains during the summer whereas low-income students lost ground. Longitudinal studies have continued to show that gaps in reading achievement based on children s socioeconomic status grow larger during summer vacation than during the school year (Alexander et al., 2001; Downey et al., 2004). In addition, there is some evidence that summer reading loss is greater for minority students than for White students (Heyns, 1987; Klibanoff & Haggart, 1981; Murnane, 1975; Phillips et al., 1998). Although there are many potential causes of summer reading loss, access to books and voluntary reading are likely to play a critical role in promoting reading achievement outside school (Cunningham & Stanovich, 1998; Entwisle et al., 2000; Heyns, 1978). Some scholars have suggested that policies designed to increase access to books may keep the learning faucet open when schools are closed during summer vacation (Entwisle et al., 2000). Voluntary reading interventions, in which children receive free books and are encouraged to read at home, may represent a cost-effective policy strategy for promoting reading achievement during summer vacation. However, there is little experimental evidence supporting the use of voluntary reading interventions as a large-scale instructional policy. Given the limitations of previous research, the National Reading Panel (2000) underscored a need for rigorous evaluations of the effectiveness of encouraging wide reading on reading achievement (p. 3-27). 2 This experimental study examined the effects of a voluntary summer reading intervention on the reading skills of fourth-grade students in one of the nation s largest public school districts. The intervention attempts to improve reading skills by increasing children s access to books, matching books to children s reading levels and preferences, and encouraging children to read orally with a parent/family member and to practice comprehension strategies learned in school. Finding a cost-effective reading intervention is important for policymakers and practitioners given the current goals of federal education policy. Under the No Child Left Behind Act, public schools must adopt scientifically based reading interventions based on experimental evidence and accelerate the achievement of underperforming subgroups, including students from lowachieving, low-income, and minority backgrounds. Moreover, a voluntary reading intervention that improved reading outcomes for students in Title I schools in particular would complement more intensive reading interventions and compensatory education policies that occur during the regular school year (Borman et al., 2005; D'Agostino & Murphy, 2004). The paper is organized into five sections. Section I summarizes the research literature on correlational studies that examine the effects of access to books and voluntary reading on reading achievement. Section II discusses how findings from the National Reading Panel s review of experimental studies on voluntary reading motivated the design of the present study. Section III describes the methods. Section IV provides estimated treatment effects for all students and subgroups of students, and section V concludes with a discussion of the implications for research and policy. I. Correlational Studies on Access to Books and Voluntary Reading Several studies on summer learning have underscored the potential impact of voluntary reading on children s learning. In pioneering work on summer learning in the Atlanta public 3 schools, Heyns (1978) studied correlates of reading growth on a standardized test of word knowledge among a sample of nearly 3,000 students in grade 6 and 7. Heyns (1978) found that the number of books read during the summer is consistently related to achievement gains; the strength of this relationship often exceeds that of socioeconomic status when prior achievement is controlled (p. 119). Additional results from multivariate analyses indicated that measures of voluntary reading (e.g., number of books read, time spent on daily leisure reading) explained a larger proportion of the variation in reading than other recreational and enrichment activities. Based on these findings, Heyns suggested that [w]hatever the reasons, the unique contribution of reading to summer learning suggests that increasing access to books and encouraging reading may well have a substantial impact on achievement (p. 172). Similarly, Phillips and Chin s (2004) analysis of the Prospects study of Title I from the 1990s revealed that reading with children, encouraging them to read on their own, and providing access to a wide range of new books, improve[d] children s performance on reading comprehension and vocabulary tests (p. 278). In addition to studies focusing on summer learning, numerous studies have revealed strong positive correlations between reading books outside school and improvements in general reading ability (Anderson et al., 1988; Donahue et al., 2001). Analyses of national survey data suggest that policies designed to increase minority children s access to books at home could potentially ameliorate ethnic disparities in reading achievement. For example, Fryer and Levitt (2004) analyzed data from the Early Childhood Longitudinal Survey (ECLS-K), a nationally representative sample of approximately 20,000 children entering Kindergarten in fall 1998, and found that White families (M = 93, SD = 65) reported owning more books than Black families (M = 39, SD = 42,) Latino families (M = 41, SD = 48), and Asian families (M = 49, SD = 56). In multivariate analyses exploring ethnic 4 disparities in test score performance during the first two years of school, Fryer and Levitt found that the inclusion of a composite measure of socioeconomic status and the number of books in children s homes accounted for the entire gap in reading scores between Black students and White students and most of the gap between Latino/a students and White students. Based on these results, Fryer and Levitt argued that the number of books in a child s home was a useful proxy for capturing the conduciveness of the home environment to academic success (p. 452). In addition to the results from the ECLS-K, similar ethnic disparities in book ownership between White and minority students have also been noted on the fourth-grade NAEP reading assessment (Donahue et al., 2001) and the Minority Student Achievement Network s survey of over 40,000 high school students in ten multiethnic suburban school districts (Ferguson, 2002). Ultimately, it is difficult to establish a strong causal link between access to books and reading achievement in multivariate analyses of survey data. Nonetheless, previous research suggest that giving children more books to read during the summer could encourage more reading practice and better reading outcomes for Black and Latino/a students. II. The National Reading Panel s Review of Voluntary Reading Interventions The key question, then, is whether voluntary reading interventions are an effective and scaleable instructional policy for improving reading fluency, vocabulary, and comprehension. To address this question, the National Reading Panel s report, Teaching Children to Read (2000a), reviewed 14 studies that focused on the effects of a widely recommended approach to developing fluent readers encouraging children to read a lot (p. 3-21). Policies designed to encourage voluntary reading have different names (e.g., sustained silent reading), but they share three characteristics: students choose their own books, read silently on their own, and receive little or no feedback on the selection of books or the reading activity from teachers or parents 5 (Filgreen, 2000). As noted by Chall (2000), voluntary reading can be viewed as a studentcentered instructional policy because it assumes that learning is accomplished on one s own, based on one s interests (p. 34). In general, the National Reading Panel s review found little experimental support for the use of voluntary reading as an effective instructional policy (p.328). In particular, the National Reading Panel noted that the inconsistent and inconclusive findings on the effects of voluntary reading could stem from either deficiencies in the instructional procedures themselves or to weaknesses and limitations evident in the study designs (p. 3-27). Four findings from the National Reading Panel s review shaped the design of the current study. First, research design appeared to mediate the treatment effects obtained from the 14 studies. None of the 6 studies using a quasi-experimental design reported positive effects (Carver & Leibert, 1995; Cline & Kretke, 1980; Morrow & Weinstein, 1986; Peak & Dewalt, 1994; Summers & McClelland, 1982; Vollands et al., 1999). In contrast, 5 of the 8 experimental studies showed positive and significant treatment effects (Burley, 1980; Davis, 1988; Holt & O' Tuel, 1989; Langford & Allen, 1983; Manning & Manning, 1984). Thus, all of the positive effects of voluntary reading were restricted to randomized experiments, which yield more valid causal estimates of treatment effects than quasi-experimental studies. Since most of the experimental studies involved a small number of students, a large multi-site randomized field trial would address design limitations in previous research. Second, voluntary reading interventions usually involved students in grade 5 and above and the benefits were strongest among older students. Among the 14 studies reviewed by the National Reading Panel, none involved first-grade students, and the two studies involving second-grade students showed no significant effects (Collins, 1980; Morrow & Weinstein, 1986). 6 Indeed, 12 of the 14 studies involved students in grade 5 to high school. In most studies of voluntary reading, researchers usually focused on older students who had sufficiently strong decoding skills to read words independently. Put another way, there is no reason to believe that a voluntary reading intervention would work if students were given free books but were unable to decode the words. Developmental theories of reading assume that decoding ability must be sufficiently strong for students, especially by the beginning of fourth-grade when children must use reading as a tool for learning the new (Chall, 1983). As noted by the National Reading Council (1998), in fourth grade and up, it is taken for granted that they are capable independently and productively of reading to learn (p. 207). Thus, prior research suggests that voluntary reading would benefit students in the upper elementary grades who can decode words but need additional reading practice in order to improve their reading fluency, vocabulary, and comprehension. Third, the failure to match texts to readers was cited as a major reason why voluntary reading interventions were ineffective. For example, among the 14 studies, the National Reading Panel cited a study by Carver and Liebert (1995) as providing one of the clearest tests of the effect of reading by studying students during the summer (p. 3-26). In this study, 43 students in grades 3, 4, and 5 read easy fiction books for six weeks (Monday through Friday), 2 hours each day. The 60 hours of leisure reading time did not translate into higher reading levels, increased reading rate and efficiency, or vocabulary gains. Since students were allowed to self-select texts, most students selected easy reading materials, which were several grade levels below their independent reading level. As a result, Carver and Liebert concluded that their study was seriously flawed in that the apparent difficulty levels used to assign books were not the real levels as measured objectively by a readability formula (p. 43). They added that the results 7 may be explained by a failure to manipulate the difficulty level of the material read (p.44). This study suggests that improving the quality of the match between reader level and text difficulty could enhance the effects of voluntary reading. Fourth, the National Reading Panel s review suggested that guided oral reading and comprehension strategies enhanced the effectiveness of reading practice. Moreover, the National Reading Panel noted that teaching oral guided reading and comprehension strategies did not require a large investment of instructional time and both were easy to implement. The Panel found strong evidence that guided oral reading strategies, in which children receive feedback during oral reading of text, improved fluency, and to a lesser extent, comprehension. This finding is important because fluent readers can read connected text accurately, quickly, and with proper expression, thereby freeing the mind to focus on the meaning of the text (LaBerge & Samuels, 1974; Stahl, 2004). Less fluent readers, however, often lack the skills to understand more challenging texts encountered in fourth-grade reading assessments (Daane et al., 2005). In addition to oral reading of text, the National Reading Panel found that comprehension strategies used by good readers question generation, question answering, summarizing, and re-reading improved understanding of text. Research on the use of multiple strategies also indicated that achievement gains were similar regardless of whether teachers spend 6 or 25 classes teaching these strategies (Rosenshine & Meister, 1994). Moreover, the National Reading Panel found that most studies involved older children, which implies that researchers taught readers who had achieved decoding and other basic reading skills before they were taught strategies (p. 4-51). The Panel s findings indicate that encouraging students to practice oral reading of text and comprehension strategies during silent reading could potentially enhance the effects of reading practice. 8 Overview of Study Design and Research Questions The current study was intended to address both the methodological and instructional limitations of previous research on voluntary reading. With respect to methodology, we conducted an experimental study to examine the causal effects of the intervention on the reading achievement of fourth-graders. In Chall s (1983) stage theory of reading, fourth-grade presents a key transitional point in schooling because it is often assumed that students have mastered the foundational skills needed to read words and are able to read in order to learn the new (Chall, 1983). Therefore, we hypothesized that most fourth-graders could decode words independently but many poor readers need additional opportunities to practice reading books during the extended summer recess.1 Another advantage of targeting fourth-grade students is the availability of data from the grade 4 main NAEP assessment of silent reading and oral reading, which can be used to compare results from this study to a nationally representative sample of grade 4 students (Daane et al., 2005; Perie et al., 2005). To summarize, the experimental design improves the internal validity of the study and the inclusion of fourth-grade students provides a basis for assessing the external validity of the results. With respect to the instructional design, the current intervention addressed multiple factors the characteristics of the students, the books, the activity, and the home context that are likely to shape opportunities to read in the summer and affect reading outcomes (RAND Reading Study Group, 2002). Thus, one goal of the instructional design was to address the variability in access to books, reading levels, and reading preferences between White and minority students. To increase access to books, each student in the treatment group received eight free books to reading during summer vacation. To facilitate a better match between text difficulty and reader ability, a text leveling system was used to provide books that were within 9 each student s independent reading level. One widely used method for matching books to readers is the Lexile Framework (U. S. Department of Education, 2001). This framework bases the readability of individual texts on measures of semantic difficulty (word frequency) and syntactic difficulty (sentence length) as in traditional measures of readability (Chall & Dale, 1995). Like other readability measures, the Lexile Framework has shown that measures of semantic and syntactic difficulty are the best predictors of text difficulty. For this study, the reading level of individual students was based on a standardized test of reading (Iowa Test of Basic Skills), and reading preferences were obtained through a spring survey. Based on information from the reading test and survey, students received books that were within their independent reading level and matched to their reading preferences. Three research questions motivated this study: (1) Did the intervention increase children s access to books at home and literacy related activities during summer vacation? (2) What is the magnitude of the treatment effect for all students and by student ethnicity and income status? (3) Are there differential treatment effects based on summer school attendance, access to books, and reading ability? III. Methods District Context for the Randomized Field Trial The randomized field trial took place in the Lake County Public School District, a large, multi-ethnic school district located in a mid-Atlantic state.2 Earlier studies on summer learning in Lake County indicated that Black and Latino students from both low-income and middleincome families lost ground in reading during summer vacation in the upper elementary grades (3 to 5). In an attempt to address the problem of summer reading loss, we designed the current intervention in order to expand summer learning opportunities and to improve the reading skills 10 of minority students and less skilled readers. Therefore, the practical goal of addressing racial and ethnic disparities in reading achievement motivated the sampling plan. With over 100 elementary schools in Lake County, the district is organized into small sub-districts, each with its own assistant superintendent.3 We selected one sub-district based on two criteria: (1) it included high-poverty schools that administered Title I schoolwide programs; and, (2) it included multiracial schools in which reading scores for Black and Latino students contributed to the federal adequate yearly progress ratings. To select the school sample, we rank ordered the schools by the percentage of Black and Latino students and included enough schools to have approximately 500 students in the study.4 The final sample included four Title I schools and six non-Title I schools with the largest percentage of minority students.5 Our study was designed to have sufficient power to detect small to modest sized effects between .10 and .20 standard deviation units, which was the magnitude of the treatment effects from a pilot study of the current intervention and remedial summer programs (Cooper et al., 2000). We also wanted to gather evidence on the effects of the intervention for less-skilled readers, minority students, and low-income students before scaling up in the lowest performing schools. In short, we planned to use evidence from this study to inform the next phase of research and policy implementation. School principals agreed to implement the experimental design, and students were required to obtain active consent from their parents. To obtain high participation, we used incentives (e.g., ice cream, pizza, extra recess) to encourage students to return consent forms and translated the permission forms into five different languages in order to increase participation among parents whose primary language was not English. Overall, 85% of the students returned their consent forms and participated in the study. 11 Table 1 provides descriptive statistics for the demographic and achievement variables. Each minority subgroup includes at least 10% of the sample, facilitating analyses of separate treatment effects for White and minority students.6 In addition, the sample means for the reading measures are similar to national norms. The spring mean ITBS score corresponded with a national percentile rank (NPR) of 52. The mean reading fluency, which is a combination of reading accuracy and rate, was 120 words correctly read per minute. This figure is similar to the national norm of 123 reported in a 2005 norming study of nearly 16,000 students in grade 4 (Behavioral Research & Teaching, 2005) and the 119 average from the 2002 NAEP study of oral reading in grade 4 (Daane et al., 2005). Finally, the reading attitude score (M = 58.45) translates into a NPR of 58 based on norms from a national study of elementary school children s attitudes toward reading (McKenna & Kear, 1990). [INSERT TABLE 1] Measures Five sets of measures were collected for this study, and descriptions of the measures are provided below. 1. Student Demographic Measures. Student demographic variables were obtained from the Lake County Public Schools testing and evaluation office. The demographic variables were dummy-coded (1 = yes, 0 = no) and included data on gender, free lunch status, English language proficiency, and whether or not the student was White, Black, Latino, Asian, or other ethnicity. The other ethnicity category included 2 Native American students and 21 multiethnic students. These demographic variables are used to determine the performance of subgroups as required by the No Child Left Behind Act s accountability rules.7 12 2. Reading Outcomes. All students were administered a pre and posttest measure of silent reading and oral reading. The Iowa Test of Basic Skills (ITBS), Form A, was the primary outcome used to measure reading achievement. The ITBS was nationally normed in 2000 and is a widely used standardized test of general reading ability. The complete battery of the ITBS, Form A (Total Reading), Level 10, was administered to grade 4 students in June 2005 and Form B was administered to grade 5 students in September 2005. The KR-20 reliability coefficients for the ITBS reading test are above .93 in the spring and fall (Hoover et al., 2003). Spring ITBS scores were also used to obtain a reading level for each student that was converted into Lexiles. The ITBS also provides a Lexile range for each student. A 100-point Lexile range on the ITBS corresponds to the independent reading level used to match books to students (U. S. Department of Education, 2001). The ITBS provides a 100-point Lexile range (+50 Lexiles above student s observed score and -50 Lexiles below), which represents each student s independent reading level. Each student s Lexile range was used to select books within the student s independent reading level. To assess oral reading fluency, retired elementary school teachers administered an oral fluency assessment from the Dynamic Indicators of Basic Early Literacy Skills (DIBELS). Reliability coefficients (alternate forms) on the DIBELS oral fluency assessment have ranged from .89 to .94 (Good & Kaminski, 2003; Tindal et al., 1983), and the test-retest reliability from this study was .89. The oral fluency assessment (Good & Kaminski, 2003) is individually administered to students and provides a measure of the number of words correctly read per minute (WCPM), which is a widely used indicator of oral reading fluency (Fuchs et al., 2001). Each student read a grade level passage in the spring and fall for one minute. To compute the 13 WCPM, each tester subtracted the number of decoding errors from the total number of words read.8 3. Spring Reading Survey. Students were administered a spring reading survey, which included a 20-item Elementary Reading Attitude Survey (ERAS) and a 25-item reading preferences survey. The Elementary Reading Attitude Survey (ERAS) is a 20-item scale used to measure attitudes toward academic reading (10 items) and recreational reading (10 items). The ERAS was normed in 1989 in a national sample of 18,138 students in grades 1 to 6. For grade 4, Cronbach alpha reliabilities were .89 for the full scale, .83 for the recreational subscale, and .83 for the academic subscale (McKenna & Kear, 1990). The reading preferences survey asked students how much they enjoyed reading books from one of 25 categories of children s books. The prompt asked students how they would feel reading books from a particular genre/category of children s literature. Each response option included a picture (smiley faces) representing 1 of 4 options: (a) I don t like it, (b) It s okay, (c) I like it, and (d) I really like it! The reading categories were developed using an article on reading preferences by the American Library Association (2003), and recent editions of the National Council of Teachers of English s (NCTE) Adventuring with Books series (McClure & Kristo, 2002). The reading categories were validated using other published surveys on children s reading preferences (Galda et al., 2000; Ivey & Broaddus, 2001; Monson & Sebesta, 1991; Summers & Lukasevich, 1983).9 In addition, four elementary school English language arts teachers checked the face validity of the categories and helped to revise and simplify the readings categories used in the spring reading preference survey. There were 25 reading categories.10 A total of 240 books were used in the study and were placed into the reading 14 categories.11 Each book was available through two major children s books sellers (Scholastic, Borders), and contained information on text difficulty (Lexile level). Survey data was used to match 8 books to children s preferences. In particular, matched books were selected for each child by a computer algorithm that merged data from two files. One file contained a Lexile level and preference categories for each of 240 available book titles. The second file contained each student s Lexile range from the spring ITBS and reading preferences from the spring survey. The algorithm generated a list of the eight best matches for each child.12 4. Fall Reading Survey. The fall survey included measures of reading activity during the summer and access to books at home. The literacy habits survey (Paris et al., 2004) has been used in previous studies of out of school reading habits involving beginning readers (grade 1 to 3). The measure was adapted for this study, which involves students in the upper elementary grades and focuses specifically on reading during summer vacation. Each of the items begins with the statement, During summer vacation and is followed by 1 of 5 activities: (1) how often did you read at home for fun? (2) how often did you read books or stories at bedtime? (3) how often did you read books? (4) how often did your parents help you read at home? (5) how often did you read out loud to someone at home? The four response options were (a) never or hardly ever, (b) once or twice a month, (c) once or twice a week, and (d) almost every day. The internal reliability coefficient for the 5-item literacy habits measure was .71. Two measures of book ownership were also included in the fall survey. The first pertained to children s books and asked students Some homes have 0 books for kids while others have more than 50 books for kids. About how many books for kids do you have in your home? Response options included (a) 0-10 books, (b) 11 to 20 books for kids, (c) 21 go 30 15 books for kids, (d) 31 to 50 books for kids, and (e) more than 50 books for kids. The second measure was taken from the National Assessment of Educational Progress (NAEP) grade 4 reading assessment. It asked students about how many books are in your home? which is followed by 1 of 4 options: (a) few (0-10), (b) enough to fill one shelf (11-25), (c) enough to fill one bookcase, (d) enough to fill several bookcases (more than 100). To facilitate interpretation of analyses, we converted the scale to two new measures: (1) the percentage of students who reported owning more than 100 books, and (2) the percentage of students who reported owning more than 50 books. 6. Summer Reading Postcards. Students in the treatment group received a postcard with each book. The postcard encouraged students practice the oral reading and silent reading skills taught by their teachers. The first set of questions asked students to write down the title of the book, and whether they read the book (yes, no). The next question asked students: what did you do to better understand this book? (check all that apply). The options included, (a) I re-read parts of this book, (b) I asked questions about this book, (c) I made predictions about this book, (d) I summarized parts of this book, and (e) I made connections (text to text, text to self). The final question prompted students to read orally with a parent or family member, and to ask whether they read the passage more smoothly, with more expression, and knew more words after a second reading. A line was provided for an adult signature indicating that the student read aloud from a 100-word passage. Procedures for Teacher Training, Reading Lessons, and Delivery of Treatment In June 2005, all participating teachers in fourth-grade attended a 2 hour training session after school to learn how to administer the reading tests and surveys and how to teach the reading lessons. A veteran English language arts teacher who had participated in a 2004 pilot study led 16 the training session. In particular, she reviewed two strategies for encouraging both silent reading and oral reading. First, teachers were asked to review five comprehension strategies that help children understand text during silent reading as noted by the National Reading Panel (2000). During the training, the lead teacher used the children s book, The Wreck of the Zephyr, to model five comprehension strategies: re-reading, asking questions, making predictions, summarizing, and making connections to self and to other text. Second, teachers were asked to instruct their students about paired reading (Topping, 1987), a widely used oral fluency strategy in which a student chooses a favorite part of a book (100 words) to read out loud to a parent or family member. During the second reading, the student asks if he/she read with more prosody that is, did they read smoothly, with expressiveness, and knowledge of more words. These lessons were designed to help students practice oral reading with their parents or another family member. A letter (English and Spanish) accompanied each postcard, encouraging a family member to read aloud with students.13 During the last two weeks of school, teachers modeled the comprehension strategies and paired reading in their classrooms and assigned homework assignments asking students to practice filling out the postcards at home.14 The postcard asked students to check comprehension strategies used while reading and to obtain a signature from a parent or family member after reading aloud from the text. Since both treatment and control students were in the classrooms during the reading lessons, this study examines the value-added of giving children free books and encouraging reading practice at home during summer vacation. Children in the treatment group received their books and postcards by airmail during July and August. Thus, the only difference between the two groups is that the treatment group received additional resources 17 books, postcards, and lessons during summer vacation whereas control students received these materials after fall posttests. IV. Results Analytic Strategy A total of 552 students received consent to participate in the study and took pretests in June 2005. Students were randomly assigned to treatment and control groups within their English language arts classroom (i.e., the blocking variable). Randomization of students within each of the 34 classroom blocks was intended to create probabilistically equivalent treatment and control groups with respect to teacher effectiveness in reading and language arts classroom. All analyses include fixed effects for the classroom in which students were randomly assigned to experimental conditions. As would be expected from the random assignment of students to the treatment and control group, there were no statistically significant differences between the two groups at the beginning of the experiment on numerous demographic and achievement characteristics. Results from Table 2 suggest that the percentage of minority students, low-income students, students with limited English proficiency, and mean scores on the reading measures were statistically equivalent in both groups. When these analyses were conducted for each of the four major ethnic groups, there were no statistically significant differences between treatment and control groups on most observed characteristics. However, among Asian students, the control group (M = 45%) had a significantly larger percentage of females than the treatment group (M = 28%), suggesting baseline differences in the gender composition of the two groups. [INSERT TABLE 2] 18 A total of 552 students completed baseline ITBS reading tests, including 282 treatment group students and 270 control group students. Due to attrition during the summer, the final analytic sample included 486 students, including 252 students in the treatment group and 234 students in the control group. A total of 66 students moved during the summer. Since these students had missing posttest scores, they were not part of the final analytic sample. Although missing data represents a potential threat to internal validity, there was no systematic relationship between missing ITBS test scores and experimental condition, 2 (1, N = 552) = .95, p =.33. Among students who moved during the summer, there was also no statistically significant difference in mean ITBS scores for the 30 students in the treatment group (M = 199.43, SD = 26.75) and 36 students in the control group (M = 194.44, SD = 22.35), t (64) = .83, p = .41. This suggests that the initial reading skills of treatment and control group students excluded from the analyses were statistically equivalent. Using data for the analytic sample of 486 students, we conducted analyses to address the three main study questions. First, we conducted an analysis of variance (ANOVA) to examine whether the intervention increased access to books and reading activities during the summer. Second, we fit a series of ordinary least squares (OLS) regression models to estimate treatment effects for all students and each of the four major ethnic groups, and interactions between the treatment and income status. Third, we estimated treatment effects for subgroups of students based on summer school attendance, access to books at home, and reading achievement. Research Question #1: Did the intervention increase children s access to books at home and literacy related activities during summer vacation? An analysis of variance on each of the two measures of books at home revealed two relevant findings. First, the ANOVA comparing the mean percentage of students who reported owning more than 100 books revealed a significant main effect of ethnicity, F (3, 467) = 16.492, 19 p < .001, but no significant difference by experimental condition nor the interaction between ethnicity and condition. Second, the ANOVA comparing the mean percentage of students who reported owning 50 kids books revealed a significant main effect of ethnicity, F (3, 467) = 13.113, p < .001. However, there were no significant differences by condition or the interaction between ethnicity and condition. Table 3 shows the actual percentage of students in each of the four ethnic groups and their self-reported data on book ownership. Bonferroni adjustments were made to compare pairwise differences between White and minority students. As can be seen in the table, the mean for White students was significantly higher than the mean for each of the minority subgroups on the two measures of book ownership. [INSERT TABLE 3] The analysis of variance revealed no significant main effect on the literacy habits scores, F (1, 462) = .49, p = .48, nor an interaction between the treatment and ethnicity, F (3, 462) = .04, p = .99. However, one pattern is apparent in figure 1, which plots the literacy habit scores for each of the four ethnic groups. With the exception of Asian students, average scores on the 4point literacy habits measure are higher, on average, for the treatment group than the control group.15 While there were no significant differences between the treatment and control group, the magnitude of the differences are suggestive of the idea that children in the experimental condition engaged in more literacy-related activities than control students. For example, the mean score on the literacy habits survey is at or above 2.3 for treatment students in each of the four subgroups of students. Since a score of 2.3 corresponds to a self-reported behavior that occurred between once or twice a month (mean score = 2.0) and once or twice a week (mean score = 3.0), this figure implies that students in the treatment group engaged in literacy related 20 activities approximately once a week (i.e., read for fun, read books, read at bedtime, read with parents, and read aloud with a family member). Data from postcards returned by 147 students in the treatment group allow us to examine whether students followed through on the directions for reading their books during the summer. Slightly over half of all students in each of the four ethnic groups returned at least one postcard, including 56% of White students, 51% of Black students, 58% of Latino students, and 51% of Asian students. Thus, postcards were available for more than half of all students in each of the four ethnic groups. On average, students from each of the four ethnic groups returned at least 3 postcards.16 However, reliable inferences about summer reading cannot be drawn from postcards due to missing data from nearly half the children in the treatment group. [INSERT-FIGURE 2] Research Question #2: What is the magnitude of the treatment effect for all students and by student ethnicity and income status? Since the purpose of this study was to examine the effects of a voluntary summer reading on student reading outcomes, individual students were assigned to treatment and control groups and analyses involved the estimation of treatment effects at the student-level. In other words, this study did not involve random assignment of schools to experimental conditions and there was no attempt to use hierarchical linear models (Raudenbush & Bryk, 2002) as is common in studies of schoolwide interventions such as Success for All (Borman et al., 2005). Ordinary least squares (OLS) regression is the most appropriate method for analyzing the achievement of individual students during summer vacation (Burkham et al., 2004).17 Formally, the following OLS model was used to estimate the treatment effects: (1) Yi = 0 + 1Ti + 2Xi + 3RBi + i 21 where Yi represents the fall reading scores for the ith student, Ti is a binary variable representing the experimental condition (i.e., treatment or control group), Xi is the baseline covariate (i.e., spring ITBS score), RBi is the randomization block (i.e., fourth-grade English language arts classroom) in which a student was randomly assigned to treatment or control groups, and i is the residual for the ith student. Using equation (1), the goal is to obtain a parameter estimate for 1, which represents the treatment effect. This intention-to-treat estimate provides an unbiased estimate of the intervention s impact on posttest reading scores and the baseline covariate is designed to improve the precision of the estimated treatment effect. To facilitate interpretation of the results, standardized scores (i.e., z-scores) were used in the OLS models. Thus, the coefficient for the treatment effect represents the effect size, or standardized mean difference, between the treatment and control group. The regression results for the ITBS reading score for all students and by ethnicity are displayed in Table 4. Among all students, the estimated treatment effect (B = .08, SE = .04, t = 1.90, p = .059) is marginally significant. The treatment effect for all students, however, is smaller than the treatment effect for Black and Latino students. For example, the estimated treatment effect is significant for Black students (B = .22, SE = .09, t = 2.59, p = .011), marginally significant for Latino students (B = .14, SE = .08, t = 1.76, p = .081), and nonsignificant for White students (B = .11, SE = .09, t = 1.23, p = .22). Although the treatment effect for each of these subgroups is positive, there were negative and non-significant effects for Asian students (B = -.17, SE = .11, t = -1.55, p = .125). This idiosyncratic finding may stem from baseline differences in the gender composition of the treatment and control group. Among Asian students, the control group included a significantly larger percentage of females than males. In general, females are more likely to read more on their own and do better on reading 22 tests than males (Blackburn, 2003; Donahue et al., 2001). In our sample of Asian students, the z-score for females was .08 compared to -.19 for males on the spring ITBS. Given baseline differences in the gender composition of Asian students at the beginning of the experiment, we estimated a treatment effect for all students excluding data for Asian students. In the re-analysis, the estimated treatment effect for all students was positive and significant (B = .14, SE = .05, t = 2.89, p = .004). To examine differential treatment effects based on free lunch status, we also tested for interactions between the treatment effect and income status for all students and for each of the four ethnic groups. None of these interaction terms was statistically significant.18 [INSERT TABLE 4] Table 5 displays the regression results for oral fluency reading outcomes, which was defined as the number of words read correctly in one minute. The estimated treatment effects were not statistically significant in the OLS models for all students and for each of the four ethnic groups. In short, the treatment students did no better than control students on a grade level measure of oral reading fluency in the fall. [INSERT TABLE 5] Question #3: Are there differential treatment effects based on summer school attendance, access to books, and reading ability? We estimated treatment effects for different subgroups of students based on their access to books at home and their reading skills. The first row of Table 6 suggests that the effect size was similar for students who did and did not attend summer school. The effect sizes, however, were largest for children who reported owning fewer than 100 books (ES = .10) or fewer than 50 kids books (ES = .13). When we examine subgroup impacts based on spring ITBS and reading fluency scores, several findings are noteworthy. There were slightly larger effects for students 23 whose spring ITBS scores were below the median (ES = .10) and for students in the low fluency group (ES = .17) [INSERT TABLE 6] Table 7 displays cross tabulations between column variables denoting student ethnicity and row variables denoting summer school attendance, access to books, and reading ability. The large chi-square statistics reveal significant associations between student ethnicity and each of the subgroup categories. More precisely, a larger percentage of minority students reported owning fewer than 100 books or 50 kids books than White students. In addition, a larger percentage of minority students than White students scored below the median on the ITBS and the measure of reading fluency. [INSERT TABLE 7] Discussion The purpose of the field trial was to determine for whom and under what conditions a voluntary summer reading intervention would be most effective. Given the current demands of federal education policy and the emphasis on narrowing academic achievement disparities, a cost-effective and scaleable voluntary summer reading intervention would support the work of policymakers and practitioners. We conducted a randomized field trial to assess the effects of a voluntary summer reading intervention on the reading achievement of Grade 4 students in a multiracial school district. In this intervention, resources were construed as the combination of additional books and reading lessons. The explicit skill instruction in oral reading of connected text and use of comprehension strategy during silent reading were designed to support reading practice among less skilled readers. 24 The main study findings are tentative but promising. The magnitude of the average treatment effect for all students was .08 standard deviations, which is small by social science standards (Cohen, 1988; Lipsey & Wilson, 1993). In addition, there were no significant effects for children who reported owning more than 100 books in general or 50 kids books in particular, and non-significant effects were also observed for fluent readers. This first finding suggests that wide scale adoption of the current intervention in all schools would not be an effective policy for improving the reading achievement of less skilled readers. In other words, the treatment effects would be diminished if all children received the intervention. The more promising results indicate that the positive effects were concentrated among students who reported owning fewer books at home and students with weaker reading skills. In particular, the effect sizes representing standardized mean differences between the treatment and control group on the posttest were largest for students who reported owning fewer than 100 books (ES = .10) or fewer than 50 kids books (ES = .13). Furthermore, the intervention had a positive impact for students whose reading fluency was below national norms (ES = .17). The range of effect sizes (.10 to .17) suggest that the average student in the treatment group would be expected to outscore the average student in the control group by approximately 4 to 5 percentile ranks. To provide a more substantive interpretation of these results, it is useful to compare the treatment effects from this study to remedial summer school programs. This comparison is relevant and appropriate because the benefits of the current intervention were restricted largely to children with fewer books and weaker skills, and such students are likely candidates for a remedial summer school program. In a meta-analysis of summer programs, Cooper et al. (2000) found that random assignment studies of remedial programs yielded an average effect size of .14, which is similar to the magnitude of the positive effects observed in the current study. 25 What explains the relatively larger effects for less-skilled readers and students who own fewer books? Our explanation fastens on both correlational and experimental evidence. First, correlational research indicates that (1) reading practice is positively associated with reading growth, and (2) this relationship is not linear. With respect to the first point, the National Reading Panel (2000) noted that there are literally hundreds of studies that find that the best readers read the most and that poor readers read the least (p. 3-21). With respect to the second point, Anderson, Wilson, and Fielding s (1988) research revealed that the relationship between time spent reading books and reading achievement among fifth-graders was curvilinear. In other words, there are diminishing returns to achievement among skilled readers and children who already engage in independent reading of books in print-rich home environments. As a result, Anderson and his colleagues found that reading comprehension rises sharply between 0 and above 10 minutes a day of book reading and then levels off (p. 297). Anderson et al. s research suggest that less fluent readers and children with fewer books at home would benefit the most from a voluntary intervention that provided children with free books and encouraged reading practice at home during summer vacation. Second, experimental evidence from this study suggests that minority students reaped the largest reading gains because they were concentrated in subgroups defined by low ownership of books and low reading fluency. The treatment effects for Black students (ES = .22) and Latino students (ES = .14) are similar in magnitude to the effects for less skilled readers and students with fewer books at home. Given the disparities in ownership of books, an intervention that increased children s access to books may potentially have a larger incremental impact on reading achievement for minority children than White children. In our sample, nearly twice as many White students than minority students reported owning more than 100 books overall or more 26 than 50 kids books. Prior analyses (Fryer & Levitt, 2004) of data from ECLS-K suggest that the number of books in a child s home was correlated with higher reading scores, underscoring the importance of addressing resource gaps between White and minority students. The gap in book ownership in Lake County mirrors those observed in other multiracial school districts such as the 10 public school districts comprising the Minority Student Achievement Network (MSAN). In surveying over 40,000 secondary school students in the MSAN, Ferguson (2002) found that 79% of White students (n = 19,216) reported owning 100 or more books compared to 29% of Black students (n = 2,715), 47% of Latino/a students (n = 2,617), and 65% of Asian students (n = 4,972). In light of these disparities in book ownership, the initial goal of the intervention was to increase book ownership and encourage reading practice at home. To the extent that Black students have fewer books in their homes and weaker reading skills than White students, it is plausible that providing more books and encouraging additional reading practice would have a relatively larger impact on the reading skills of minority children. Implications for Policy and Practice The main empirical findings have several implications for policy and practice. The current intervention should be viewed as a complementary approach for improving reading achievement among less-skilled readers in the summer and for sustaining reading gains among children who can decode words. First, voluntary summer reading interventions represent one cost-effective policy alongside many existing summer programs for at-risk readers in urban school districts. There is also a need for more effective out-of-school learning policies, especially for minority families.19 Although minority students had weaker reading skills than White students, the majority of students in Lake County did not attend summer school programs. Addressing summer reading loss through a cost-effective intervention is also important because 27 some research suggests that Black students (Klibanoff & Haggart, 1981; Murnane, 1975; O'Brien, 1999) lose ground in reading during the summer relative to White students even after controlling for socioeconomic status (Phillips & Chin, 2004). The current intervention cost approximately $57.00 per student, including the price for books, postage, and labor. Ballpark estimates from a preliminary cost-effectiveness analysis suggest that the current intervention would cost substantially less than giving a student a voucher to attend summer school for free.20 A study evaluating the effects of a mandatory summer program versus voluntary summer reading intervention would provide better estimates of costs and program impacts. Second, a logical implementation strategy might include efforts to support the reading skills of children in schools receiving federal Reading First grants. Since the goal of Reading First is to prevent reading difficulties from kindergarten to third-grade, it represents a major investment in helping children master basic reading skills. The current intervention could build on the accomplishments of the Reading First legislation. Indeed, there is evidence that both White and minority students have acquired the phonemic awareness and decoding ability needed to read words well by the end of third-grade. Beyond the ability to decode words, children need access to books and guidance on effective forms of reading practice that to improve their reading skills. Having teachers instruct children how to practice simple reading activities and providing books may represent a fun activity that capitalizes on children s acquired skills (e.g., decoding ability) in fourth-grade and their responsiveness to teacher encouragement. Limitations and Future Research Despite the promising effects for lower-performing readers, additional research is needed to address limitations of the study design. The results showed no difference in the estimated treatment effects based on student income status. This result conflicts with the faucet theory of 28 learning (Cooper et al., 1996; Entwisle et al., 2000) which suggests that providing additional books and instruction would benefit poor children more than middle-class children. Thus, a major limitation of the present study was the use of a coarse binary measure of socioeconomic status (SES) that is, whether or not a student received free and reduced-price lunch. Usually, studies have employed granular measures socioeconomic status (SES) such as family income (Heyns, 1978) or a composite based on parental income, occupational status, and education (Downey et al., 2004; Entwisle et al., 1997). A more sensitive measure of SES may have revealed differential treatment effects for low-income children. In addition to using more precise measures of SES, future work should examine whether the treatment effects are generalizable across different school districts and are cumulative across multiple summers. A larger experiment involving multiple schools district would allow us to assess the external validity of the results by exploring treatment effects in sites with diverse social and economic conditions, local education policies, and samples of teachers and students. The present study was undertaken in a school district with strong support from the assistant superintendent and school principals. It is unclear whether the intervention could be effectively implemented in multiple districts where administrative support may vary across sites. Moreover, since this study examined changes in reading achievement over a single summer, a longitudinal study is needed to examine long-term impacts. Finally, randomized experiments have been criticized for providing little information on the causal mechanisms that produce outcome differences between treatment and control groups (Maxwell, 2004). Indeed, the survey items used to assess access to books and literacy activities were unable to capture significant differences between treatment and control groups. Data from this study were too limited to provide detailed information on the processes that ultimately led to 29 better reading outcomes among less-skilled readers, children with fewer books at home, and children from minority backgrounds. Additional ethnographic research could reveal the nature of the social interaction between students and their families around books. Ethnographies of children s home environment (Chin & Phillips, 2004; Lareau, 1989) have shown that minority and low-income parents often lack the knowledge to help their children select appropriately challenging and interesting books. Such insights, gained through ethnographic research, could shed light on the etiology of summer reading loss and how it varies across families from different social class and ethnic backgrounds. Ideally, future research should include an experimental design and ethnographic data on the summer reading experiences of students from various ethnic and social backgrounds. Conclusion The main findings warrant cautious optimism about current efforts to address academic achievement disparities. Caution is needed until the current findings are replicated in a study that includes a more diverse sample of districts and schools and employs a richer set of reading measures. Optimism, however, may also be justified given the promising findings for lowerperforming students. Encouraging voluntary reading during summer vacation may be one costeffective strategy for helping struggling readers acquire the skills needed to succeed in school. 30 Tables and Figures TABLE 1 Characteristics of Students at the Beginning of the Study (N = 552) Variable % Min Female 0.47 White 0.33 Black 0.19 Latino 0.26 Asian 0.17 Other 0.05 Free-Reduced Lunch 0.39 Limited English Proficiency 0.38 Title I School 0.26 Age (Months) 108 Iowa Test of Basic Skills (DSS) 142 Iowa Test of Basic Skills (NPR) 1 Oral Reading Fluency (WCPM) 6 Elementary Reading Attitude Survey (Total) 23.00 Note: DSS = Developmental Standard Score, NPR = National Percentile Rank, WCPM = Words Correctly Read Per Minute. Max M SD 140 123.45 4.74 263 202.78 24.08 99 51.97 28.08 242 120.27 37.83 80 58.45 11.12 31 TABLE 2 Comparison of Baseline Achievement and Demographic Characteristics by Condition Treatment Control Variable M SD M SD t df White 0.35 0.48 0.32 0.47 0.72 550 Black Latino Asian Other Title I School Age (Months) Female Limited English Proficiency Free-Reduced Lunch Iowa Test of Basic Skills (DSS) Oral Reading Fluency (WCPM) Elementary Reading Attitude Survey 0.19 0.26 0.17 0.04 0.26 123.57 0.45 0.56 0.37 0.39 0.44 0.37 0.20 0.44 0.19 0.26 0.18 0.05 0.26 0.40 -0.14 0.44 -0.01 0.39 -0.46 0.22 -0.51 0.44 4.58 0.09 0.62 550 550 550 550 550 550 550 550 550 550 527 542 p 0.471 0.890 0.995 0.647 0.607 0.929 0.538 0.367 0.840 0.145 0.647 0.990 0.974 4.89 123.32 0.50 0.50 0.48 0.49 0.55 0.43 0.50 -0.90 0.50 0.20 0.50 -1.46 0.46 0.01 203.24 24.10 202.30 24.09 120.29 39.89 120.25 35.60 58.43 11.39 58.46 10.86 -0.03 Note: Scores on the fluency assessment and total attitude score do not add up to 552 due to missing data. 32 TABLE 3 Book Ownership by Student Ethnicity White Black Latino Asian Percentage who reported owning more than 100 books 0.58 0.24* 0.24* 0.32* Percentage who reported owning more than 50 kids books 0.61 0.30* 0.33* 0.35* *Pairwise comparison revealed significant difference at the .05 level with Bonferroni adjustments 33 TABLE 4 Ordinary Least Squares Models Predicting Treatment Effect on ITBS (Total Reading Scores) All White Black Latino Asian Variables B (SE) B (SE) B (SE) B (SE) B (SE) Treatment .08~ .11 .22* .14~ -0.17 (.04) (.09) (.09) (.08) (.11) Spring ITBS .87** (.02) -0.07 (.05) 0.76 486 .84** (.04) -0.03 (.10) 0.71 160 .83** (.05) -.17~ (.09) 0.76 93 .77** (.05) -0.12 (.09) 0.69 125 .88** (.07) 0.07 -0.13 0.71 85 (Constant) R2 N Note: All models include fixed effects for the randomization block. Standard errors in parentheses. The model for "other ethnic students" (21 multiethnic, 2 Native American) revealed non-significant treatment effects. ~p<.10, *p<.05, **p<.01 TABLE 5 Ordinary Least Squares Models Predicting Treatment Effect on Oral Reading Fluency (WCPM) All White Black Latino Asian Variables B (SE) B (SE) B (SE) B (SE) B (SE) Treatment -2.09 -2.83 -1.79 -2.21 -.41 (1.50) (2.73) (3.31) (2.81) (3.95) Spring-WCPM .83*** (0.02) 11.08*** -3.07 0.80 450 0.86*** (.04) 7.24 -5.51 0.80 150 0.83*** (.05) 12.21~ -6.72 0.81 85 0.77*** (.05) 17.00** -6.37 0.73 116 0.77*** (0.05) 20.14* (8.54) 0.75 80 (Constant) R2 N Note: All models include fixed effects for the randomization block. Standard errors in parentheses. Sample sizes for OLS models predicting fluency are not equal to the ITBS analysis due to missing data on the fall fluency assessment. ~p<.10, *p<.05, **p<.01 34 TABLE 6 Treatment Effects Based on Summer School Participation, Book Ownership, Reading Achievement Characteristic of Individual Students B SE Lower Bound Upper Bound t p 1. Did you go to summer school in a Lake County Public School? Yes (n = 83) 0.091 0.095 -0.098 0.280 0.962 0.339 No (n = 398) 0.077 0.050 -0.020 0.175 1.554 0.121 2. About how many books are in your home? (more than 100) Yes (n = 186) 0.057 0.078 -0.097 No (n = 294) 0.097 0.051 -0.004 0.206 0.728 0.467 0.198 1.887 0.060 3. About how many books for kids do you have in your home? (more than 50 books for kids) Yes (n = 206) 0.024 0.072 -0.117 0.166 0.339 0.735 No (n = 272) 0.131 0.056 0.021 0.242 2.348 0.020 4. Spring Iowa Test of Basic Skills (total reading) scores above median? Yes (n= 246) 0.064 0.066 -0.065 No (n = 240) 0.104 0.058 -0.010 0.193 0.982 0.327 0.218 1.805 0.072 5. Spring reading fluency (Words Read Correctly Per Minute) above median? Yes (n = 236) 0.023 0.063 -1.000 0.146 0.367 0.714 No (n = 232) 0.167 0.063 0.044 0.290 2.669 0.008 Note: Samples are unequal due to missing data from the fall survey. 35 TABLE 7 Relationship Between Student Ethnicity and Summer School, Book Ownership, Reading Achievement Characteristic of Individual Students White Black Latino Asian 1. Did you go to summer school in a Lake County Public School? Yes 7 19 30 19 No 93 81 70 81 Chi Square = 27.682 (df = 3), p < .001 2. About how many books are in your home? (more than 100) Yes 58 No 42 Chi Square = 44.761 (df = 3), p < .001 25 75 25 75 32 66 3. About how many books for kids do you have in your home? (more than 50 books for kids) Yes 61 30 33 No 39 70 67 Chi Square = 36.458 (df = 3), p < .001 4. Spring Iowa Test of Basic Skills (total reading) scores above median? Yes 73 38 No 27 62 Chi Square = 73.182 (df = 3), p < .001 5. Spring reading fluency (Words Read Correctly Per Minute) above median? Yes 65 39 No 35 61 Chi Square = 36.856 (df = 3), p < .001 35 65 29 71 42 58 33 67 46 54 36 Figure 1: Literacy Habits by Ethnicity 2.6 2.5 Estimated Marginal Means 2.4 2.3 0=control 1=treatment Black Latino Asian 2.2 White 37 References Alexander, K. L., Entwisle, D. R., & Olson, L. S. (2001). Schools, achievement, and inequality: A seasonal perspective. Educational Evaluation and Policy Analysis, 23, 171-191. Anderson, R. C., Wilson, P. T., & Fielding, L. G. (1988). Growth in reading and how children spend their time outside of school. Reading Research Quarterly, 23(3), 285-303. Behavioral Research & Teaching. (2005). Oral reading fluency: 90 years of measurement. Eugene, OR: Author. Blackburn, M. (2003). Boys and literacies: What difference does gender make? Reading Research Quarterly, 38, 276-287. Borman, G. D., Slavin, R. E., Cheung, A., Chamberlain, A. M., Madden, N. A., & Chambers, B. (2005). Success for all: First-year results from the national randomized field trial. Educational Evaluation and Policy Analysis, 27, 1-22. Burkham, D. T., Ready, D. D., Lee, V. E., & LoGerfo, L. F. (2004). Social-class differences in summer learning between kindergarten and first grade: Model specification and estimation. Sociology of Education, 77, 1-31. Burley, J. E. (1980). Short-term, high intensity reading practice methods for upward bound students: An appraisal. Negro Educational Review, 31, 156-161. Carver, R. P., & Leibert, R. E. (1995). The effect of reading library books at different levels of difficulty upon gain in reading ability. Reading Research Quarterly, 30, 26-48. Chall, J. S. (1983). Stages of reading development. New York: McGraw-Hill. Chall, J. S. (2000). The academic achievement challenge: What really works in the classroom? New York: Guilford Press. Chin, T., & Phillips, M. (2004). Social reproductive and child-rearing practices: Social class, children's agency, and the summer activity gap. Sociology of Education, 77, 185-210. Cline, R. K. J., & Kretke, G. L. (1980). An evaluation of a long-term SSR in the junior high school. Journal of Reading, 23, 503-506. Cohen, J. (1988). Statistical power analysis for the behavioral sciences. Hillsdale, NJ: Erlbaum. Collins, C. (1980). Sustained silent reading periods: Effects on teachers' behaviors and students' achievement. Elementary School Journal, 81, 108-114. Cooper, H., Charlton, K., Valentine, J. C., & Muhlenbruck, L. (2000). Making the most of summer school: A meta-analytic and narrative review. Washington, DC: Society for Research in Child Development. Cooper, H., Nye, B., Charlton, K., Linday, J., & Greathouse, S. (1996). The effects of summer vacation on achievement test scores: A narrative and meta-analytic review. Review of Educational Research, 66, 227-268. Cunningham, A. E., & Stanovich, K. E. (1998). What reading does for the mind. American Educator, Spring/Summer, 8-15. D'Agostino, J. V., & Murphy, J. A. (2004). A meta-analysis of reading recovery in the united states. Educational Evaluation and Policy Analysis, 26, 23-39. Daane, M. C., Campbell, J. R., Grigg, W. S., Goodman, M. J., & Oranje, A. (2005). Fourthgrade students reading aloud: NAEP 2002 special study of oral reading (NCES 2006469). Washington, DC: U. S. Department of Education. Institute of Education Sciences, National Center for Education Statistics. Davis, Z. T. (1988). A comparison of the effectiveness of sustained silent reading and directed reading activity on students' reading achievement. The High School Journal, 72, 46-48. 38 Donahue, P. L., Finnegan, R. J., Lutkus, A. D., Allen, N. L., & Campbell, J. R. (2001). The nation's report card: Fourth-grade reading 2000, NCES 2001-499. Washington, D.C.: U.S. Department of Education, Office of Educational Research and Improvement, National Center for Education Statistics. Downey, D. B., von Hippel, P. T., & Broh, B. A. (2004). Are schools the great equalizer? Cognitive inequality during the summer months and the school year. American Sociological Review, 69, 613-635. Duffett, A., Johnson, J., Farkas, S., Kung, S., & Ott, A. (2004). All work and no play? New York: Public Agenda. Entwisle, D. R., Alexander, K. L., & Olson, L. S. (1997). Children, schools, and inequality. Boulder: Westview Press. Entwisle, D. R., Alexander, K. L., & Olson, L. S. (2000). Summer learning and home environment. In R. D. Kahlenberg (Ed.), A notion at risk (pp. 9-30). New York: The Century Foundation Press. Ferguson, R. F. (2002). What doesn't meet the eye: Understanding and addressing racial disparities in high-achieving suburban schools. Naperville, IL: North Central Regional Educational Laboratory. Filgreen, J. (2000). The SSR handbook: How to organize and maintain a sustained silent reading program. Portsmouth, NJ: Heinemann. Fuchs, L. S., Fuchs, D., Hosp, M. K., & Jenkins, J. R. (2001). Oral reading fluency as an indicator of reading competence: A theoretical, empirical, and historical analysis. Scientific Studies of Reading, 5, 241-258. Galda, L., Ash, G. E., & Cullinan, B. E. (2000). Children's literature. In M. L. Kamil, P. B. Mosenthal, P. D. Pearson & R. Barr (Eds.), Handbook of reading research, vol. 3 (pp. 361-380). Mahwah, NJ: Lawrence Erlbaum Associates. Good, R. H., & Kaminski, R. A. (2003). Dynamic indicators of basic early literacy skills, 6th edition. Eugene, OR: Institute for the Development of Educational Achievement, University of Oregon. Heyns, B. (1978). Summer learning and the effects of schooling. New York: Academic Press, Inc. Heyns, B. (1987). Schooling and cognitive development: Is there a season for learning? Child Development, 58, 1151-1160. Holt, S. B., & O' Tuel, F. S. (1989). The effect of sustained silent reading and writing on achievement and attitudes of seventh and eighth grade students reading two years below grade level. Reading Improvement, 26, 290-297. Hoover, H. D., Dunbar, S. B., Frisbie, D. A., Oberley, K. R., Ordman, V. L., Naylor, R. J., et al. (2003). The Iowa tests, guide to research and development. Itasca, IL: Riverside Publishing. Ivey, G., & Broaddus, K. (2001). "just plain reading": A survey of what makes students want to read in middle school classrooms. Reading Research Quarterly, 36(4), 350-377. Klibanoff, L. S., & Haggart, S. A. (1981). Summer growth and the effectiveness of summer school (rep. No. 8). Mountain View, CA: RMC Research Corporation. LaBerge, D., & Samuels, S. J. (1974). Toward a theory of automatic information processing in reading. Cognitive Psychology, 6, 293-323. Langford, J. C., & Allen, E. G. (1983). The effects of u.S.S.R. On students' attitudes and achievement. Reading Horizons, 23, 194-200. 39 Lareau, A. (1989). Home advantage: Social class and parental intervention in elementary education. New York: The Falmer Press. Lipsey, M. W., & Wilson, D. B. (1993). The efficacy of psychological, educational, and behavioral treatment: Confirmation from meta-analysis. American Psychologist, 48, 1181-1209. Manning, G. L., & Manning, M. (1984). What models of recreational reading make a difference. Reading World, 23, 375-380. Maxwell, J. A. (2004). Causal explanation, qualitative research, and scientific inquiry in education. Educational Researcher, 33, 3-11. McClure, A. A., & Kristo, J. V. (2002). Adventuring with books, a booklist for pre-k-grade 6. Urbana, IL: National Council for Teachers of English. McKenna, M. C., & Kear, D. J. (1990). Measuring attitude toward reading: A new tool for educators. The Reading Teacher, May, 626-639. Monson, D. L., & Sebesta, S. (1991). Reading preferences. In J. Flood (Ed.), Handbook of research on teaching the English language arts (pp. 664-673). New York: Macmillan. Morrow, L. M., & Weinstein, C. S. (1986). Encouraging voluntary reading: The impact of a literature program on children's use of library centers. Reading Research Quarterly, 21, 330-346. Murnane, R. J. (1975). The impact of school resources on the learning of inner-city children. Boston: Ballinger Press. National Reading Panel. (2000a). 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. Washington, DC: National Institute of Child Health and Human Development. National Research Council (Ed.). (1998). Preventing reading difficulties in young children. Washington, D.C.: National Academy Press. O'Brien, D. M. (1999). Three essays on early academic achievement of minority and disadvantaged students. Unpublished Dissertation for the Degree of Doctor of Philosophy in Political Economy, The University of Texas at Dallas, Dallas. Paris, S. G., Pearson, P. D., Cervetti, G., Carpenter, R., Paris, A. H., DeGroot, J., et al. (2004). Assessing the effectiveness of summer reading programs. In G. Borman & M. Boulay (Eds.), Summer learning: Research, policies, and programs (pp. 122-161). Mahwah, NJ: Lawrence Erlbaum Associates. Peak, J., & Dewalt, M. W. (1994). Reading achievement: Effects of computerized reading management and enrichment. ERS Spectrum, 12, 31-34. Perie, M., Moran, R., & Lutkus, A. D. (2005). NAEP 2004 trends in academic progress: Three decades of student performance in reading and mathematics (NCES 2005-464). Washington, DC: U.S. Department of Education, Institute of Education Sciences, National Center for Education Statistics. Phillips, M., & Chin, T. (2004). How families, children, and teachers contribute to summer learning and loss. In G. Borman & M. Boulay (Eds.), Summer learning: Research, policies, and programs (pp. 255-278). Mahwah, NJ: Lawrence Erlbaum Associates. Phillips, M., Crouse, J., & Ralph, J. (1998). Does the black-white test score gap widen after children enter school? In C. Jencks & M. Phillips (Eds.), The black-white test score gap (pp. 229-272). Washington, D.C.: Brookings Institution Press. RAND Reading Study Group. (2002). Reading for understanding. Santa Monica, CA: RAND. 40 Raudenbush, S. W., & Bryk, A. S. (2002). Hierarchical linear models. Thousand Oaks, CA. Rosenshine, B., & Meister, C. (1994). Reciprocal teaching: A review of the research. Review of Educational Research, 64, 181-221. Stahl, S. A. (2004). What do we know about fluency? Findings of the national reading panel. In P. McCardle & V. Chhabra (Eds.), The voice of evidence in reading research (pp. 187212). Baltimore: Paul H. Brookes Publishing Co. Sturm, B. W. (2003). The information and reading preferences of north carolina children. School Library Media Research Retrieved April 20, 2005, from http://www.ala.org/ala/aasl/aaslpubsandjournals/slmrb/slmrcontents/volume62003/readin gpreferences.htm Summers, E. G., & Lukasevich, A. (1983). Reading preferences of intermediate-grade children in relation to sex, community, and maturation (grade level): A canadian perspective. Reading Research Quarterly, 18, 347-360. Summers, E. G., & McClelland, J. V. (1982). A field-based evaluation of sustained silent reading (SSR) in intermediate grades. Alberta Journal of Educational Research, 28, 100-112. Tindal, G., Marston, D., & Deno, S. L. (1983). The reliability of direct and repeated measurement (research report 109). Minneapolis, MN: Univeristy of Minnesota Institute for Research on Learning Disabilities. Topping, K. J. (1987). Paired reading: A powerful technique for parent use. The Reading Teacher, 40, 608-614. U. S. Department of Education. (1998). Summer activities of students enrolled in grades 1-12. Washington, DC: U. S. Department of Education, National Center for Education Statistics. U. S. Department of Education. (2001). Assessing the lexile framework: Results of a panel meeting, NCES 2001-08. Washington, DC: U. S. Department of Education, National Center for Education Statistics. U. S. Department of Education. (2005). The condition of education 2005 (NCES 2005-94). Washington, DC: National Center for Education Statistics, U. S. Government Printing Office. Vollands, S. R., Topping, K. J., & Evans, R. M. (1999). Computerized self-assessment of reading comprehension with the accelerated reader: Action research. Reading and Writing Quarterly, 15, 197-211. Notes 1 According to analyses of ECLS-K, virtually all children do well on basic reading tasks, including the ability to recognize letters, beginning and end sounds, and sight words. However, there are significant racial/ethnic disparities related to more difficult reading comprehension (e.g., literal inference, extrapolation, and evaluation). See Fryer and Levitt (2005), Table 7. 2 Pseudonyms are used to maintain the anonymity of the school district. 3 There are over 100 elementary schools Lake County and schools are organized into smaller administrative and geographic sub-units, each with its own assistant superintendent. The assistant superintendent of one sub-unit agreed to be the official research sponsor and provided the administrative support to carry out the experiment. This sub-unit had both Title I and nonTitle schools. The non-Title I schools were ethnically diverse and had sufficiently large Black 41 and Latino enrollments whose achievement counted for AYP calculations. We decided not to target only Title I schools because they were located in 7 different sub-units and managed by 7 different assistant superintendents. In short, it was administratively easier to conduct the study in 1 sub-unit. Thus, the decision to target the one sub-unit with both Title I schools and racially and ethnically diverse non-Title I schools facilitated the administration of the experiment and the analyses of subgroup analyses. 4 Due to administrative and financial costs, we decided not to assign schools at random to experimental and treatment conditions. Instead, we randomly assigned students to conditions. Based on Cooper et al. s (2000) meta-analysis of summer school programs and a previous pilot experiment, we anticipated a small effect size between .10 and .20 standard deviations. Using this information, we conducted a power analysis by assuming an effect size of between .10 and .20, and a pretest that was highly correlated with the posttest (r = .80). Under these circumstances, approximately 500 students would be needed to have an 80% chance of finding a significant differences between treatment and controls on a two-tailed test with alpha = .05. 5 All 4 Title I schools administered a schoolwide program, which, under the Elementary and Secondary Education Act, permits these schools to use their federal dollars to support programs serving all students regardless of their income status. 6 Relative to the public school enrollment (K-12) in the United States in fall 2003, the study sample has a slightly larger percentage of minority students. Figures for the nation are as follows: White students = 58.3%, Black students = 16.1%, Latino students = 18.6%, and other = 7.0%. See table 4.1 in the Condition of Education 2005 (U. S. Department of Education, 2005). 7 Information on summer school attendance was obtained through the fall reading survey, which asked whether they attended a summer school program in a public school. About 17% of the sample reported attending summer school, which is similar to figures on summer school attendance in nationally representative surveys of elementary and secondary school students (Burkham et al., 2004; U. S. Department of Education, 1998). 8 As noted in the manual for administration for the DIBELS oral fluency assessment (Good & Kaminski, 2003), all words omitted, substituted, and hesitations of more than 3 seconds were scored as errors while self-corrections within 3 seconds were scored as correct. Each examiner adhered to scripted directions, stating: Please read this (point to passage) out loud. If you get stuck, I will tell you the word so you can keep reading. When I say stop, I may ask you to tell me about what you read, so do your best reading. Start here (point to the first word of the passage). Begin. After the passage was read, students were asked: Please tell me all about what you just read. Try to tell me everything you can. Begin. 9 For example, Summers and Lukasevich (1983) conducted a literature review of preference surveys and found 5-30 common themes in surveys measuring the reading interests and preferences of intermediate grade students. The 14 categories from their survey are similar to those in the NCTE list and include: adventure, history/geography, animals, children/family, sports, humor, science, poetry, fantasy, travel, romance, nature study, and mystery. 10 For the reading preference, the following categories were used: (1) science nonfiction (earth science, space, technology, (2) prehistoric life, dinosaurs, (3) animal nonfiction (zoology, mammals, marine life, reptiles, et.), (4) historical nonfiction and biography, (5) historical fiction, (6) African and African American stories (real and realistic fiction), (7) Asian and Asian American stories (real and realistic fiction), (8) Latino and Latino Americans (Spanish speaking) and stories (real and realistic fiction), (9) Native American stories (real and realistic fiction), (10) 42 family and everyday life stories, (11) school life stories, (12) struggle and survival stories (realistic adventure, heroes, heroines, etc.), (13) poetry, (14) sports biographies, (15) sports fiction, (16) animal fantasy, (17) science fiction fantasy, (18) time travel fantasy, (19) heroic and adventure fantasy, (20) supernatural tales and fantasy (ghosts, magic, and monsters), (21) realistic animal stories, (22) mystery stories, (23) stories of other girls my age (real and realistic fiction), (24) stories of other boys my age (real and realistic fiction), (25) traditional literature (fables, myths and legends, tall tales, etc..). 11 We decided not to use a book fair to match books to readers for two reasons. First, principals and teachers wanted to implement an administratively easy and simple procedure for matching books to reader s preferences. A book fair was deemed too burdensome to set up at the end of the school year given existing administrative tasks associated with the end of the school year. Second, we wanted to employ a uniform procedure for ensuring that books were within each student s independent reading level. Prior research indicated that allowing children to self-select texts would produce poor matches between texts and readers. 12 The computer program (Python) was used to create two input: (1) student information on reading preferences and reading levels (Lexile level), and (2) book information on reading categories and reading levels (Lexile level). For each student, the student file contains the Survey ID, lexile range determined by the ITBS test, and the results of the reading preference survey. The book file contains a list of the 240 titles available to match to each student, along with the lexile level, reading level, and the list of categories. The output is a list of students matched to a specific book and a score for the match between the student and the book. A high score indicates a good match between child and book, and books with the 8 highest scores were matched to each child. 13 The family letter was drafted by teachers and principals: Dear Parent (Family Member), Please encourage your child to read this book and complete the postcard. It will help your child if he or she reads out loud to you, or to an older brother or sister. After you listen to your child reading out loud a second time, tell him or her how they improved. There is also a place for your signature. Please sign the postcard indicating that you listened to your child read a part of the book. The postcard does not require a stamp; all you need to do is put it in the mail. It is important to return the postcard even if your child has not finished the book. After the postcard has been returned, you may certainly encourage your child to finish the book, read it again, or reread favorite parts of it. The information on the postcard will help us understand the results and improve the program next year. Thank you for your time and effort towards making your child s summer reading a successful experience. 14 All students were told that they would receive books and postcards either in the summer or fall. Teachers followed a uniform script in which they explained that, regardless of when they received the books, children should read their books and then follow the direction for the completing the postcards and reading activities. 15 Among Asian students, the idiosyncratic finding on the literacy habits scores may stem from initial baseline differences in the gender composition of the two groups. The control group had a larger percentage of females than the treatment. Since females read more on their and do better on reading tests, on average, then males, the results for the control group may have been biased upward creating the near 0 difference on the literacy habits measure. 43 16 The mean number of returned postcards by ethnicity were as follows: Asian students (M = 5.33, SD = 2.10), White students (M = 3.02, SD = 1.86), Black students (M = 4.00, SD = 2.11). Latino students (M = 3.69, SD = 2.29), Asian students (M = 5.33, SD = 2.10) 17 As noted by Burkham et al. (2004) in their analysis of summer learning using ECLS-K data, this study did not investigate children s progress in school, but instead investigated academic learning when children were out of school during the summer months. Thus, the nested nature of research questions that is typical of school-effects studies is not applicable here. Instead of the multilevel methods that are typically required for school-effects studies (i.e., hierarchical linear models, or HLM), we used the major analytic method appropriate for multivariate analyses of individual children: ordinary least-squares (OLS) regression (p. 9). 18 All p-values for the treatment by free lunch coefficient were larger than .10. However, for Asians students the interaction term was negative and significant (B = -.448, SE = .156, t = -2.05, p = .044). 19 According to a recent Public Agenda survey on out-of-school time (Duffett et al., 2004), Black and Latino/a parents were more likely than White families to express concern about the quality of summer learning experiences available to their children. 20 As noted in the presentation of the results, the set of positive and significant effects ranged between .13 and .22 standard deviation units, which can be compared to effect size estimates in Cooper et al. s (2000) comprehensive meta-analysis of summer school programs. According to the most relevant findings from this meta-analysis (Cooper et al., 2000), random assignment studies of remedial summer school yielded an effect size of .14, which is line with the treatment effects observed in the current study. The effect size of .13 to .22 standard deviations imply that the average student in the treatment group would be expected to score between 5 and 9 percentile ranks higher than the average student in the control group. The estimated program effect from remedial summer school and the current intervention can be used to generate some ballpark costeffectiveness ratios. Assuming that program effects are .14 standard deviations, we would expect the two interventions to improve achievement by approximately 5 percentile ranks. In other words, the average child who exceeded 50% of her peers would be expected to exceed approximately 55% of her classmates after either a remedial summer school program of a voluntary summer reading intervention. In Lake County, remedial summer school costs $360 per pupil, and this figure would represent the cost of a voucher that paid for the tuition. The present intervention cost approximately $56.70 per student, including the price of books, postage, and labor. Although the cost-effectiveness ratio would favor the voluntary summer reading intervention, these are merely preliminary and rudimentary estimates and they provide no estimate of costs and effects over multiple school districts and summers. 44

Find millions of documents here - Study Guides, Homework Solutions, Papers, Exam Answer Keys and more. Course Hero has millions of course related materials that will enable you to learn better, faster and get an A in all your courses.
Below is a small sample set of documents: