A statewide evaluation of academic achievement in year-round schools.
by Bradley J. McMillen Time has long been a variable of interest in countless education contexts. A report by the National Education Commission on Time and Learning (1994) discussed at length the importance of time in education, the extent to which time controls what happens in schools, and the need for efficient use of time to promote greater student achievement. One of the major conclusions in that report called for more time devoted to instruction and learning in core subjects. The issue of quantity also was addressed by Carroll's (1963) model of school learning, which designates the amount of instructional time as one of the more malleable factors in determining student learning. Subsequent international comparative investigations, such as the Third International Mathematics and Science Study (Martin, Mullis, Gonzales, Smith, & Kelly, 1999), have shown that countries in which students spend more days in school and more time during the day on mathematics and science instruction demonstrate higher achievement in those subjects. The sheer quantity of instructional time that is available to students is undoubtedly critical in determining student outcomes, and the variety of educational policies concerning the number of credit hours required to earn a diploma or degree and the number of instructional hours in a school year are testament to this relationship. In addition to quantity, researchers have addressed quality of time as it relates to student learning. Research on time on task and academic learning time (Berliner, 1990; Fredrick & Walberg, 1980) focused on the relationship between the amount of time students spend engaged in academic activities and how much they learn. Studies in this area have demonstrated that simply exposing students to classrooms and teachers is not sufficient to affect learning, which implies that the educational quality of the activities and interactions that occur in those settings mediates the relationship between time and learning. Some studies of time and learning focus on how academic performance varies by time of day; however, researchers have not concluded which time of day is best. For example, spelling and memory abilities appear to vary according to the time of day, although some types of errors seem to result more often in the morning, whereas other types of mistakes occur more often in the late afternoon and evening (Dunne, Roche, & Hartley, 1990; Folkard, Monk, Bradbury, & Rosenthall, 1977; Morton & Diubaldo, 1995). Research also has indicated that nonmedicated students with attention deficits have better problem-solving abilities and engage in less off-task behavior in the morning (Zagar & Bowers, 1983). Year-Round Education One of the most common debates about time and learning in recent decades has centered around the adoption of year-round school calendars. According to the National Association for Year-Round Education (NAYRE; 2000), the number of year-round schools operating in the United States has increased from over 400 in the late 1980s to 2,880 during the 1999-2000 school year (National Association for Year-Round Education). Although year-round education exists in many different forms, it involves essentially the reorganization of the traditional school calendar so that the long summer vacation is replaced by several smaller breaks evenly spaced throughout the year. Although year-round is often used to describe extended-year schools (i.e., more than 180 days of instruction per year), I used the term year-round to describe schools that have redistributed their 180 instructional days evenly throughout the calendar year (Worthen & Zsiray, 1994). The definition of year-round was summed up by NAYRE as follows: | |
Year-round education (YRE) reorganizes the school year to
provide more continuous learning by dividing the long summer
vacation into shorter, more frequent breaks.... Students
in a year-round program attend the same classes and
receive the same amount of instruction as students on a nine-month
calendar (usually 180 days).... The year-round calendar
is organized into instructional blocks and vacation
periods that are evenly distributed across 12 months.
| | The proliferation of year-round schools at the national level is evident in North Carolina, where the number of year-round schools has grown from 73 in 1994 to 121 in 2000. Two basic models of year-round education are being used in North Carolina--schoolwide (SW) and school-within-a-school (SWS). In the SW model, all the children in a school attend on a year-round (12 month) calendar. In the SWS model, one group of students attends on a year-round calendar, while the others attend on a traditional 9-month calendar. Teachers in the SWS model also are usually divided into two groups--those who teach in the traditional program and those who teach in the year-round program. The SWS model basically creates one traditional August-May program and one year-round program, each with its own teachers and students, but operating on the same campus with the same administrative staff. There are many types of calendar arrangements in North Carolina's year-round schools; in the most common type, students attend school for 45 days and then break for 15 days. Approximately 94% of year-round schools in North Carolina were on this schedule during the 1998-99 school year (North Carolina Department of Public Instruction (NCDPI; 2000). Although most SW year-round schools in North Carolina operate on a single-track schedule, approximately 13% of them operate as multitrack schools. In those schools, different groups (i.e., tracks) of students begin the school year at different times. Vacation schedules for each track are distributed so that at least one track of students is always out of school. Multitrack year-round schools often are implemented to ease overcrowding. Interest in implementing year-round schools can be attributed to three touted advantages of a year-round calendar: (a) increased student achievement; (b) greater satisfaction among parents, teachers, and students; and (c) cost savings. The first two benefits are mentioned often in conjunction with all year-round schools, although cost savings are typically associated only with multitrack year-round schools, which can help postpone the need to build new schools in areas experiencing significant population growth (Inger, 1994). Several reviews of the existing literature on year-round education have been conducted. The general consensus has been that the outcomes of year-round education are at least as positive as (or better than) those achieved under the traditional school calendar. However, the number of quality studies conducted and published in this area is limited (Kneese, 1996). In one of the earliest reviews, Merino (1983) found no differences in achievement between students in year-round and traditional schools. Worthen and Zsiray (1994) later reevaluated the evidence on the effectiveness of year-round education on the basis of several studies that focused on achievement, cost, satisfaction, and other outcomes. They concluded that (a) achievement in year-round schools is equal to or greater than achievement in traditional schools; (b) teachers and students in year-round schools have more positive attitudes; (c) most parents are satisfied with a year-round program if it is well implemented (even though there will always be parents who do not like year-round calendars); and (d) single-track programs cost as much or more than traditional school programs, whereas multitrack programs can result in significant cost savings if the programs are carried out well. Kneese (1996) conducted a similar review of 15 studies that focused specifically on achievement in traditional calendar and year-round schools. She concluded that achievement in year-round schools appears to be slightly higher than in traditional calendar schools. That finding also is echoed in other recent studies (Gandara & Fish, 1994; Shields & Oberg, 1999). Although researchers have not adequately addressed the reasons why achievement may be slightly higher in year-round schools, one possibility is that these schools can use intersessions to provide remediation and enrichment activities, thereby increasing students' exposure to the curriculum (Ballinger, 1995). Another possible explanation comes from research that emphasizes a decline in achievement during the long summer vacation associated with the traditional school calendar (Cooper, Nye, Charlton, Lindsay, & Greathouse, 1996; Entwistle & Alexander, 1992; Parsley & Powell, 1962; Wintre, 1986). Year-round school advocates claim that dividing the long summer vacation period into smaller pieces helps alleviate some of the "forgetting" that occurs over the summer in traditional school programs. However, there is currently no specific scientific evidence to support that contention (R. E. Mitchell & Mitchell, 1999). Some evidence exists that year-round programs may be particularly beneficial, especially in reading, for cognitively and economically disadvantaged students (Cooper et al., 1996; Gandara & Fish, 1994; Handleman & Harris, 1984). Despite several studies on the topic, most existing research on year-round education and achievement suffers from important methodological limitations. The limitations include (a) failure to take student-level factors into account when estimating achievement effects, (b) loss of precision in the dependent variable due to collapsing achievement outcomes into categories such as "at or below grade level" (Shields & Oberg, 1999), (c) failure to report any tests of statistical significance or measures of effect size (Alcorn, 1992), and (d) failure to differentiate between year-round and extended-year schools (Gandara & Fish, 1994). In addition, none of the analyses on year-round education have accounted statistically for the nesting of students within schools, and few have been subjected to the peer review process. The goal of this investigation, therefore, was to provide a more statistically appropriate examination of achievement in year-round and traditional calendar schools. Method According to several archival data sources, 106 public schools in North Carolina operated on a year-round calendar in Grades 3-8 during the 1997-98 school year. I conducted analyses comparing the academic achievement of students who attended the year-round schools and programs during the 1997-98 school year with that of students who attended traditional calendar schools and programs. Data Source I obtained the data for this study from historical databases of the North Carolina Testing Program. Over 95% of public school students in North Carolina participate each year in this statewide testing program (North Carolina Department of Public Testing, 1999a). Testing program data in Grades 3-8 in North Carolina are gathered from end-of-grade (EOG) assessments in reading and mathematics and are reported in the form of normed developmental scale scores. With respect to content, EOG tests are aligned closely with the North Carolina Standard Course of Study, which is the official state curriculum. Test scores are scaled separately for each grade level and subject area, and the normative distribution of scores in each subject area shifts upward slightly from one grade level to the next. Therefore, a student's scale score is expected to increase naturally from one grade level to the next. Because of that feature, scale scores cannot be directly combined or compared across multiple grade levels because the distribution of possible scores is not in the same scale location from one grade level to the next. Because I used data from students across multiple grade levels, I had to convert the scale scores to standard scores before conducting any analyses. Scores were standardized separately for each subject area by grade-level combination. The resulting scores had a mean of 50 and a standard deviation of 10. In addition to achievement test scores, the testing program databases contain demographic information collected from each student. I used that data to create some student-level variables to be employed as covariates in the achievement analyses. Although the focus of the analyses was to examine differences according to year-round and traditional calendar conditions, I included the variables to provide a precise test of differences between calendar conditions and to help rule out competing explanations if any between-group discrepancies were found. The covariates that I used were the same for the SW and SWS schools in both reading and mathematics analyses, and they included each student's previous year EOG test scores, gender, ethnicity, and parents' highest level of education. Sample Selection Procedures Because EOG tests are given only in Grades 3-8 and because I designed this study to determine achievement growth from one year to the next, I used all available data from students who took EOG tests in Grades 4-8 in 1998. The final sample included all North Carolina public school students who took EOG tests in either reading or mathematics in 1998 at a given grade level and who also took the EOG test in that same subject area in 1997 at the previous grade level. Therefore, I excluded those students who did not have available data in a given subject for both years for that subject. Students who were retained during the 1997-98 school year were excluded from all analyses. Sample Characteristics Demographic information for the samples that I used in the reading and mathematics analyses are reported in Table 1. The data are based on student characteristics as reported in North Carolina Testing Program databases for the 1997-98 school year. Because some students had test data available in only one of the two subjects, some small differences exist between the reading analysis sample and the mathematics analysis sample. Demographic information was reported separately for the following four groups of students: (a) students from schools that operated a SW traditional calendar, (b) students from schools that operated a SW year-round calendar, (c) students who attended by the traditional calendar in a school that was operating a SWS model, and (d) students who attended by the year-round calendar in a school that was operating a SWS model. In this study, I included data for over 345,000 students from 1,470 schools serving Grades 3-8. The breakdown of the number of schools by type is given in Table 2. The four groups of students were very similar with respect to gender. However, patterns of differences occurred between the groups regarding grade level, ethnicity, and parental education level (i.e., highest level of education attained). In both the SW and SWS model schools, students who attended according to a year-round calendar were concentrated at the lower grade levels, likely because most year-round programs in North Carolina are in the K-5 range (North Carolina Department of Public Instruction, 2000). In schools operating a SW model, traditional calendar students were more likely to be Caucasian and less likely to be African American than were year-round calendar students (reading, [chi square] = 935.25, p < .05; mathematics, [chi square] = 934.26, p < .05). Also, year-round calendar students reported slightly higher levels of parental education (reading, [chi square] = 549.53, p < .05; mathematics, [chi square] = 544.76, p < .05). In schools operating a SWS model, however, the pattern of ethnicity differences was reversed. In those schools, year-round programs consisted of more Caucasian students and fewer African American students (reading, [chi square] = 43.53, p < .05; mathematics, [chi square] = 46.66, p < .05). As in the SW programs, year-round students reported slightly higher levels of parental education (reading, [chi square] = 14.22, p < .05; mathematics, [chi square] = 15.04, p < .05). Results I used hierarchical linear modeling (HLM) procedures to examine potential differences in achievement between the four groups of students. Traditional regression methods require either aggregating data to the school level prior to analysis, which results in a loss of statistical power and precision, or disaggregating school-level data down to the individual student level, which often results in spuriously significant results that show relationships between variables that may not exist (Hox, 1995). HLM methods avoid both of those problems by properly incorporating both school-level and student-level factors in the same analysis (Bryk & Raudenbush, 1992; Singer, 1998). I conducted the analyses separately for SW schools and for SWS schools using the MIXED procedure from the SAS[R] statistical software package. Each of the two sets of analyses proceeded in three steps. First, I estimated models for both subjects using no independent variables in order to partition the variance into between-school and within-school components. I then used the components to derive intraclass correlations. Second, I calculated a within-schools model using only the student-level covariates. Third, I evaluated the full models including the student-level covariates and the factor of interest (year-round status) to determine whether year-round status demonstrated any relationship to achievement after controlling for the student-level covariates. Schoolwide Data The initial null models for SW schools showed intraclass correlations of .11 for reading and .13 for mathematics, indicating that the majority of variation in achievement in both subjects was found within schools, not between schools. The within-schools models (Table 3) indicated that each of the student-level covariates was related to both reading and mathematics achievement and that these relationships were in the expected direction. Prior achievement had the strongest relationship in both the reading and mathematics analyses, but other student demographic variables also were significant predictors. Female students and Caucasian students demonstrated higher achievement in both subjects. Higher levels of parental education also were associated with higher achievement in reading and mathematics; the magnitude of the effect increased steadily across the six measured levels. The full between-schools models indicated no statistically significant difference in reading or mathematics achievement between students in year-round and traditional schools after controlling for the student-level covariates. With respect to the overall models, prior achievement and the other student-level covariates accounted for a large majority of the explainable between-school and within-school variation in achievement (Table 3). Those indicators of explained variance are not analogous to a squared multiple correlation in linear regression, however, and should not be interpreted as such. They are merely relative indicators of the proportion of school-level variation and student-level variation that is explained by the variables in the model (Snijders & Bosker, 1994). The addition of the year-round variable to the models resulted in no improvement in the measures for either reading or mathematics. School-Within-a-School Data The analyses for the SWS programs also consisted of an initial null model that did not include independent variables. In those analyses, students were not only nested within schools--they were also nested within year-round conditions within schools, implying a three-level HLM design. Some initial calculations, however, indicated that the between-schools variance component was not significantly different from zero. That component was therefore dropped from the final models, yielding a two-level model with students nested within year-round conditions within schools. The intraclass correlations for the models were .08 for reading and. 12 for mathematics, indicating that (as in the SW analyses), the majority of the variation in achievement was between students. HLM results for SWS programs were essentially similar to the previous SW analyses, with one exception in mathematics. Similar to the SW results, prior achievement and parental education level were associated positively with achievement in both reading and mathematics. Also, Caucasian students demonstrated higher achievement than did non-Caucasian students. Unlike the SW analyses, however, female students did not demonstrate higher achievement in mathematics than did males in the SWS analyses (see Table 4). All other relationships in both the within-track and between-track models mirrored those of the SW results. As in the SW analyses, the full between-track models indicated no statistically significant difference in reading or mathematics achievement between students in year-round and traditional tracks. Again, prior achievement and the other student-level covariates accounted for the majority of the explainable variation in achievement (see Table 4); the year-round status variable contributed nothing to the fit of the model. Differential Effects Because previous research has raised the possibility that more disadvantaged students might benefit from year-round education (Cooper et al., 1996; Gandara & Fish, 1994; Handleman & Harris, 1994), I generated additional regression models for both SW schools and SWS programs. Those models examined potential interactions between year-round status and prior achievement, parent education level, and ethnicity. They were identical to the previously reported full models (see Tables 3 & 4), except for the addition of one interaction term. The models for SW schools revealed statistically significant interactions between year-round status and prior achievement for reading (t = 5.19, p< .05) and mathematics (t = 2.34, p < .05); lower achieving students in year-round schools demonstrated slightly higher achievement than their traditional school counterparts. That relationship was stronger in reading than in mathematics, but in both cases, the differences were not large (approximately .05 standard deviations). Models examining the interactions between year-round status and ethnicity showed a similarly small relationship, but only in mathematics (reading, t =.67, p >.50; mathematics, t = 2.30, p< .05). Compared with their counterparts in traditional schools, Caucasian students scored approximately .04 standard deviations higher in mathematics in year-round schools, whereas non-Caucasian students scored at similar levels in both types of schools. I found no interactions with respect to parent education level (reading, t = -1.93, p >.05; mathematics, t = -.04, p > .96). The models for SWS programs yielded no statistically significant interactions between either year-round status and prior achievement (reading, t = .37, p > .71; mathematics, t = -.54, p > .58) or year-round status and ethnicity (reading, t = -.39, p > .69; mathematics, t = 1.11, p > .27). With respect to parent education level, however, an interaction was found for reading (t = -2.18, p< .05), but not for mathematics (t = .20, p >.84). Although students whose parents had some post-secondary education demonstrated higher reading achievement in both types of programs, this difference was slightly larger among students in traditional tracks. Similar to the SW analyses, however, the difference was small in magnitude, approximately .03 standard deviations. Discussion Initial analyses indicated no statistically significant differences in either reading or mathematics achievement between students attending school by a year-round calendar and those attending by a traditional August-May calendar during the 1997-98 school year after controlling for prior achievement, student gender, student ethnicity, and parent education level. The results were consistent for students in SW schools as well as those in SWS. I found some interactions implying that lower achieving and Caucasian students may benefit slightly from being on a year-round calendar. One result suggested that students whose parents have high levels of education may do better under a traditional school calendar. The effects were all very small in magnitude, however, and were not found for all subjects and program types. The results of this study are somewhat consistent with previous studies indicating that achievement of students in year-round schools is equal to that of students in traditional schools and that year-round calendars may be particularly beneficial for lower achieving students. Differences between the results obtained in this study and those that reported higher overall achievement in year-round schools may be due to methodological distinctions in data analysis and various definitions of year-round schools. For example, Gandara and Fish (1994) reported achievement benefits for students in year-round schools. However, the year-round schools in that study had more than 220 instructional days in a calendar year, leaving open the question of whether those gains might have been attributable to the added number of instructional days as opposed to the use of a year-round calendar. Similarly, Shields and Oberg (1999) found that achievement is slightly higher in multitrack year-round schools. However, all their analyses were conducted at the school level and did not include any student-level factors that may have been related to year-round status and may also have affected observed achievement. It is also possible that the distribution of variance between the levels of the model diminished the likelihood of finding any meaningful relationships between the outcomes and school- or program-level predictor variables. Tables 3 and 4 show that once the student-level covariates entered into the model, there was little explainable variation remaining. That evidence, coupled with the relatively low intraclass correlations for each model, imply that any school- or program-level predictor variables were unlikely to contribute much additional clarity to any of the models. However, even models calculated without the student-level covariates (i.e., with year-round status as the sole predictor variable) showed no statistically significant differences between year-round and traditional calendar students in either subject area for both SW and SWS models, providing additional confirmation of the lack of evidence of any year-round effect. Given that many year-round programs are magnet programs and therefore may draw students from outside the school's normal attendance zone, the consideration of student-level covariates in studies of year-round schools is essential. Although the inclusion of covariates associated with achievement is not a cure for nonrandom assignment, it does provide a strict test of the potential effects of the year-round calendar on achievement. Not only do studies of year-round education often suffer the limitations inherent in retrospective nonexperimental research, but also local decisions about where year-round calendars are implemented and the potential effects of school choice may produce systematic differences between students in year-round and traditional calendar schools and programs. For example, all SWS year-round programs and 69% of SW year-round schools in North Carolina are schools of choice (North Carolina Department of Public Instruction, 1999b), which may have some bearing on the demographic differences between the traditional and year-round students in this study. Other studies that focus exclusively on multitrack programs also document substantial differences in student populations and their corresponding relationship to achievement outcomes. Specifically, R. E. Mitchell and Mitchell (1999) noted that socioeconomic homogeneity within tracks occurs in situations where families are allowed to choose the track in which their child will enroll and that differences in academic achievement between tracks are accounted for largely by these demographic differences. Unfortunately, this study does not speak to the potential differences between single-track and multitrack SW year-round schools. No information noting whether the 67 year-round schools were single track or multitrack was available in the extant database. Therefore, the schoolwide year-round schools in this study likely include both types of schools. However, a recent survey of year-round schools in North Carolina indicated that 87% of them are single track (North Carolina Department of Public Instruction, 2000), which would imply that the SW schools in the present study were largely single track. Future studies of year-round schools should address the possible differences between single and multitrack schools and whether either of these models has unique benefits for students, teachers, or families. Researchers also need to differentiate between the effects of a year-round calendar and additional instructional time on student achievement. In North Carolina, almost all year-round programs offer some form of remediation and/or enrichment during intersessions; 57% of the programs have mandatory intersession remediation for students who are behind academically (North Carolina Department of Public Instruction, 1999b). That particular factor may be at least partially responsible for the slight benefits demonstrated in some cases for lower achieving students in this investigation. The question of whether the total amount of instructional time or the distribution of that time across the calendar year might be responsible for any achievement advantages for year-round schools has yet to be addressed in the research in this area. Further investigations that (a) consider the length of time that a school has been operating year round and (b) measure possible differences in pedagogical techniques between traditional and year-round schools also are needed. Such studies could help document factors that might mediate the effects of year-round education on student outcomes and could determine whether a year-round schedule results in changes in day-to-day instructional activities. Although increased achievement is often touted as a benefit of year-round education, the results of my investigation suggest that the merits of year-round education might best be judged on factors other than achievement. I found some statistically significant interactions indicating that some students may benefit more from a year-round calendar, but these effects are probably too small to be educationally significant by most standards. The consideration of other circumstances such as potential cost savings and stakeholder preferences, which vary from location to location, may provide a more reasonable basis for decisions about whether to keep or to adopt year-round calendars. | |
Table 1.--Student Demographics, by Calendar Condition
Reading
T-SW YR-SW
Variable n % n %
Gender
Female 172,345 50 6,806 50
Male 172,630 50 6,772 50
Ethnicity
Asian American 4,195 1 180 1
African American 92,222 27 5,184 38
Hispanic 5,639 2 258 2
American Indian 5,421 1 279 2
Multiracial 1,916 1 91 1
Caucasian 235,506 68 7,582 56
Other/unknown 76 -- 4 --
Parental education level
Less than high school 35,416 10 1,082 8
High school 151,473 44 5,124 38
Trade/business school 16,108 5 670 5
Community college/
technical school 50,724 15 1,820 13
Four-year college 70,824 21 3,671 27
Graduate school 20,430 6 1,211 9
Grade level (in 1997-98)
4 69,753 20 3,756 28
5 68,488 20 3,697 27
6 69,230 20 2,135 16
7 69,428 20 2,057 15
8 68,076 20 1,933 14
Total 344,975 13,578
Reading
T-SWS YR-SWS
Variable n % n %
Gender
Female 3,896 51 1,702 49
Male 3,773 49 1,671 51
Ethnicity
Asian American 79 1 19 1
African American 2,801 36 1,027 30
Hispanic 78 1 22 1
American Indian 67 1 20 1
Multiracial 31 1 18 1
Caucasian 4,609 60 2,267 67
Other/unknown 4 -- 0 --
Parental education level
Less than high school 800 10 290 9
High school 3,411 44 1,407 42
Trade/business school 467 6 200 6
Community college/
technical school 1,120 15 621 18
Four-year college 1,552 20 737 22
Graduate school 319 4 118 3
Grade level (in 1997-98)
4 1,867 24 1,018 30
5 1,848 24 970 29
6 1,344 18 570 17
7 1,612 21 413 12
8 998 13 402 12
Total 7,669 3,373
Mathematics
T-SW YR-SW
Variable n % n %
Gender
Female 172,507 50 6,802 50
Male 173,285 50 6,801 50
Ethnicity
Asian American 4,219 1 181 1
African American 92,518 27 5,195 38
Hispanic 5,677 2 256 2
American Indian 5,436 2 280 2
Multiracial 1,924 1 93 1
Caucasian 235,942 68 7,594 56
Other/unknown 76 -- 4 --
Parental education level
Less than high school 35,654 10 1,087 8
High school 151,826 44 5,143 38
Trade/business school 16,126 5 671 5
Community college/
technical school 50,784 15 1,824 13
Four-year college 70,940 21 3,666 27
Graduate school 20,462 6 1,212 9
Grade level (in 1997-98)
4 70,025 20 3,767 28
5 68,699 20 3,707 27
6 69,328 20 2,140 16
7 69,497 20 2,057 15
8 68,243 20 1,932 14
Total 345,792 13,603
Mathematics
T-SWS YR-SWS
Variable n % n %
Gender
Female 3,899 51 1,707 50
Male 3,802 49 1,673 50
Ethnicity
Asian American 79 1 19 1
African American 2,817 37 1,023 30
Hispanic 78 1 23 1
American Indian 70 1 22 1
Multiracial 31 1 18 1
Caucasian 4,622 60 2,275 67
Other/unknown 4 -- 0 --
Parental education level
Less than high school 810 11 295 9
High school 3,431 45 1,406 42
Trade/business school 468 6 200 6
Community college/
technical school 1,121 15 624 18
Four-year college 1,552 20 737 22
Graduate school 319 4 118 3
Grade level (in 1997-98)
4 1,885 24 1,022 30
5 1,859 24 975 29
6 1,346 18 570 17
7 1,612 21 412 12
8 999 13 401 12
Total 7,701 3,380
Notes. Some percentages may not add to 100 due to rounding.
T-SW = traditional school (entire school on traditional calendar);
T-SWS = traditional program, school-within-a-school model;
YR-SW = year-round school (entire school on year-round calendar);
and YR-SWS = year-round program, school-within-a-school model.
Table 2.--Number of Schools, by Type
Type of school n %
Schoolwide traditional 1,364 93
Schoolwide year-round 67 4
School-within-a-school 39 3
Table 3.--Achievement in Schoolwide Year-Round and Schoolwide
Traditional Calendar Schools
Reading
Within
Variable B (se) t
Previous end-of-grade .80 (.001) 763.94 *
test score (1997)
Gender (a) .44 (.02) 25.84 *
Ethnicity (b) -1.00 (.02) -45.93 *
Parental education--high
school graduate 1.03 (.03) 33.39 *
Parental education--trade/business
school 1.60 (.05) 32.55 *
Parental education--community
college/technical school 1.88 (.04) 51.08 *
Parental education--4-year college 2.52 (.04) 68.74 *
Parental education--graduate school 3.03 (.05) 61.75 *
Calendar (c)
% between-school variation 99.5
explained
% within-school variation 70.2
explained
Reading
Between
Variable B (se) t
Previous end-of-grade .80 (.001) 763.94 *
test score (1997)
Gender (a) .44 (.02) 25.84 *
Ethnicity (b) -1.01 (.02) -45.94 *
Parental education--high
school graduate 1.03 (.03) 33.39 *
Parental education--trade/business
school 1.60 (.05) 32.54 *
Parental education--community
college/technical school 1.88 (.04) 51.07 *
Parental education--4-year college 2.51 (.04) 68.72 *
Parental education--graduate school 3.03 (.05) 61.74 *
Calendar (c) -.10 (.11) -0.95
% between-school variation 99.5
explained
% within-school variation 70.2
explained
Mathematics
Within
Variable B (se) t
Previous end-of-grade .81 (.001) 785.00 *
test score (1997)
Gender (a) .11 (.02) 6.65 *
Ethnicity (b) -.82 (.02) -37.13 *
Parental education--high
school graduate .87 (.03) 28.82 *
Parental education--trade/business
school 1.35 (.05) 27.86 *
Parental education--community
college/technical school 1.64 (.04) 45.58 *
Parental education--4-year college 2.35 (.04) 65.38 *
Parental education--graduate school 2.98 (.05) 61.55 *
Calendar (c)
% between-school variation 85.3
explained
% within-school variation 71.0
explained
Mathematics
Between
Variable B (se) t
Previous end-of-grade .81 (.001) 785.00 *
test score (1997)
Gender (a) .11 (.02) 6.65 *
Ethnicity (b) -.82 (.02) -37.15 *
Parental education--high
school graduate .87 (.03) 28.81 *
Parental education--trade/business
school 1.35 (.05) 27.85 *
Parental education--community
college/technical school 1.64 (.04) 45.57 *
Parental education--4-year college 2.35 (.04) 65.36 *
Parental education--graduate school 2.98 (.05) 61.54 *
Calendar (c) -.23 (.19) -1.25
% between-school variation 85.3
explained
% within-school variation 71.0
explained
Note. The reference group for dummy-coded parent education level
factors was less than high school. Percentages of variance explained
were calculated in reference to the null model.
(a) 0 = boy; 1 = girl. (b) 0 = Caucasian; 1 = non-Caucasian,
(c) 0 = year-round school; 1 = traditional school.
* p <.05.
Table 4.--Achievement in School-Within-a-School Year-Round and
School-Within-a-School Traditional Tracks
Reading
Within
Variable B (se) t
Previous end-of-grade
test score (1997) .79 (.01) 130.41 *
Gender (a) .40 (.10) 4.01 *
Ethnicity (b) -.91 (.12) -7.52 *
Parental education--high school
graduate 1.40 (.18) 7.86 *
Parental education--trade/business
school 1.96 (.26) 7.46 *
Parental education--community
college/technical school 2.11 (.21) 10.03 *
Parental education--4-year college 2.79 (.21) 13.23 *
Parental education--graduate school 3.37 (.31) 10.88 *
Calendar (c)
% between-school variation
explained 94.6
% within-school variation
explained 69.8
Reading
Between
Variable B (se) t
Previous end-of-grade
test score (1997) .79 (.01) 130.40 *
Gender (a) .40 (.10) 4.01 *
Ethnicity (b) -.92 (.12) -7.53 *
Parental education--high school
graduate 1.40 (.18) 7.86 *
Parental education--trade/business
school 1.96 (.26) 7.47 *
Parental education--community
college/technical school 2.12 (.21) 10.04 *
Parental education--4-year college 2.79 (.21) 13.23 *
Parental education--graduate school 3.37 (.31) 10.88 *
Calendar (c) .11 (.19) .55
% between-school variation
explained 94.5
% within-school variation
explained 69.8
Mathematics
Within
Variable B (se) t
Previous end-of-grade
test score (1997) .80 (.01) 130.80 *
Gender (a) -.13 (.10) -1.34
Ethnicity (b) -.98 (.12) -7.95 *
Parental education--high school
graduate 1.37 (.17) 7.85 *
Parental education--trade/business
school 1.82 (.26) 7.06 *
Parental education--community
college/technical school 1.97 (.21) 9.51 *
Parental education--4-year college 2.95 (.21) 14.25 *
Parental education--graduate school 3.55 (.31) 11.60 *
Calendar (c)
% between-school variation
explained 84.1
% within-school variation
explained 70.3
Mathematics
Between
Variable B (se) t
Previous end-of-grade
test score (1997) .80 (.01) 130.80 *
Gender (a) -.13 (.10) -1.33
Ethnicity (b) -.98 (.12) -7.95 *
Parental education--high school
graduate 1.37 (.17) 7.86 *
Parental education--trade/business
school 1.82 (.26) 7.07 *
Parental education--community
college/technical school 1.97 (.21) 9.52 *
Parental education--4-year college 2.95 (.21) 14.26 *
Parental education--graduate school 3.55 (.31) 11.61 *
Calendar (c) .22 (.33) .67
% between-school variation
explained 83.9
% within-school variation
explained 70.3
Note. The reference group for dummy-coded parent education level
factors was less than high school. Percentages of variance explained
were calculated in reference to the null model.
(a) 0 = boy; 1 = girl. (b) 0 = Caucasian; 1 = non-Caucasian,
(c) 0 = year-round school; 1 = traditional school.
* p <.05.
| | NOTE I thank Carolyn Cobb, Kris Kaase, and Gongshu Zhang for their feedback on design and analysis issues in this study. REFERENCES Alcorn, R. D. (1992). Test scores: Can year-round schools raise them? Thrust for Educational Leadership, 21(6), 12-15. Ballinger, C. (1995). Prisoners no more. Educational Leadership, 53(3), 28-31. Berliner, D. C. (1990). What's all the fuss about instructional time? In M. Ben-Peretz & R. Bromme (Eds.), The nature of time in schools: Theoretical concepts, practitioner perceptions (pp. 3-35). New York: Teachers College Press. Bryk, A., & Raudenbush, S. (1992). Hierarchical linear models: Applications and data analysis methods. Newbury Park, CA: Sage. Carroll, J. B. (1963). A model of school learning. Teachers College Record, 64, 723-733. Cooper, H., Nye, B., Charlton, K., Lindsay, 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. Dunne, M. P., Roche, F., & Hartley, L. R. (1990). Effects of time of day on immediate recall and sustained retrieval from semantic memory. Journal of General Psychology, 117, 403-410. Entwistle, D. R., & Alexander, K. L. (1992). Summer setback: Race, poverty, school composition, and mathematics achievement in the first two years of school. American Sociological Review, 57, 72-84. Folkard, S., Monk, T. H., Bradbury, R., & Rosenthall, J. (1977). Time of day effects in schoolchildren's immediate and delayed recall of meaningful material. British Journal of Psychology, 68, 45-50. Fredrick, W. C., & Walberg, H. J. (1980). Learning as a function of time. The Journal of Educational Research, 73(4), 183-194. Gandara, P., & Fish, J. (1994). Year-round schooling as an avenue to major structural reform. Educational Evaluation and Policy Analysis, 16, 67-85. Handleman, J., & Harris, S. (1984). Can summer vacation be detrimental to learning? An empirical look. Exceptional Children, 31, 151-157. Hox, J. J. (1995). Applied multilevel analysis. Amsterdam: T. T. Publikates. Inger, M. (1994). Year-round education: A strategy for overcrowded schools. (ERIC Document Reproduction Service No. ED 378 267) Kneese, C. C. (1996). Review of research on student learning in year-round education. Journal of Research and Development in Education, 29(2), 61-72. Martin, M. O., Mullis, I. V. S., Gonzalez, E. J., Smith, T. A., & Kelly, D. L. (1999). School contexts for learning and instruction: IEA's Third International Mathematics and Science Study (TIMSS). Chestnut Hill, MA: Boston College. Merino, B. J. (1983). The impact of year-round schooling: A review. Urban Education, 18(3), 298-316. Mitchell, R. E., & Mitchell, D. E. (1999). Student segregation and achievement tracking in year-round schools. Manuscript submitted for publication. Morton, L. L., & Diubaldo, D. (1995). Circadian differences in hemisphere-linked spelling proficiencies. International Journal of Neuroscience, 81, 101-110. National Association for Year-Round Education. (2000). Available: http://www.nayre.org National Education Commission on Time and Learning. (1994). Prisoners of time. Washington, DC: U S. Government Printing Office. North Carolina Department of Public Instruction. (1999a). The 1997-98 North Carolina state testing results. Raleigh, NC: Author. North Carolina Department of Public Instruction. (1999b). [A survey of year-round schools]. Unpublished raw data. North Carolina Department of Public Instruction. (2000). Year-round schools and achievement in North Carolina (Evaluation Brief Vol. 2[2]). Raleigh, NC: Author. Parsley, K. M., & Powell, M. (1962). Achievement gains or losses during the academic year and over the summer vacation period: A study of trends in achievement by sex and grade level among students of average intelligence. Genetic Psychology Monographs, 66(2), 285-342. Shields, C. M., & Oberg, S. L. (1999). What can we learn from the data? Toward a better understanding of the effects of multi-track year-round schooling. Urban Education, 34(2), 125-154. Singer, J. (1998). Using SAS PROC MIXED to fit multilevel models, hierarchical models, and individual growth models. Journal of Educational and Behavioral Statistics, 24, 323-355. Snijders, T. A., & Bosker, R. J. (1994). Modeled variance in two-level models. Sociological Methods and Research, 22, 342-363. Wintre, M. G. (1986). Challenging the assumption of generalized academic losses over summer. The Journal of Educational Research, 79(5), 308-312. Worthen, B., & Zsiray, S. (1994). What twenty years of educational studies reveal about year-round education. Chapel Hill: North Carolina Educational Policy Research Center. Zagar, R., & Bowers, N. D. (1983). The effect of time of day on problem solving and classroom behavior. Psychology in the Schools. 20, 337-345. Address correspondence to Bradley J. McMillen, North Carolina Department of Public Instruction, Division of Accountability Services, 301 North Wilmington Street, Raleigh, NC 27601. (E-mail: bmcmille@dpi.state.nc.us) BRADLEY J. McMILLEN is senior evaluation consultant, North Carolina Department of Pubic Instruction. He is interested in educational evaluation and policy research. |
|
