Table of Contents | Previous | Next |
CHAPTER 12 - CHILDREN’S ACADEMIC AND SOCIAL OUTCOMES
INTRODUCTION
A major justification for the Head Start-Public School Early Childhood Transition Demonstration Project was that former Head Start children are likely to need additional supports in the first few years in elementary school to ensure that they maintain the benefits of their Head Start participation. This is because low-income children have consistently been shown to be “at risk” for performing below national average on a variety of indicators related to school success, especially academic indicators (S. Ramey & Ramey, 1998; Brooks-Gunn, et al., 1994; Huston, 1992). Remarkably little is known about the actual school performance of a large and nationally diverse sample of former Head Start children, especially using a variety of indicators and informants on an annual basis for the first four years in school. Thus, this study provides much needed information about the early school adjustment of former Head Start children.
An extensive literature indicates that by the end of third grade, children’s academic and social adjustment are strong predictors of their later school success, including their subsequent academic achievement and their probability of completing high school. Both family income status (poverty versus non-poverty) and maternal educational level have been strongly associated with children’s early and later school success. Recent analyses conducted by Currie and Thomas (1997) on the academic achievement levels of former Head Start children nationally (based on children who enrolled in Head Start in the late 1970s and were in 8th grade by 1988) addressed the question of “fade-out” of effects. They discovered that among former Head Start children, there was a large ethnic or racial discrepancy in the quality of public schools they attended. Specifically, they conclude that: “These results show that relative to other black children, black children who attended Head Start subsequently went to schools of low quality in the sense that the black children in these schools have poorer outcomes. The same is not true for white Head Start children. Thus, the results are consistent with the hypothesis that differential fade-out in the effects of Head Start could be caused by subsequent exposure to poor schools among black Head Start children.” (p. 15) This conclusion is in agreement with the results of recent reviews (C. Ramey & Ramey, 1998; S. Ramey & Ramey, 2000) of the long-term effects of early educational interventions for “at risk” children: namely, that for the benefits of early intervention to be sustained, children need continued supports and opportunities for learning, such as participation in reasonably responsive schools that have high expectations for academic achievement.
Before reporting the child outcomes, it is important to underscore two features about this study. First, the sample of children and the schools they attended do not represent the nation as a whole and may not represent accurately the school experiences of all former Head Start children. This is because the local sites were selected based on submitting a competitive application that indicated their interest in and ability to implement a broad program of supports and services to benefit former Head Start children and their families. Thus, the school districts that were selected are much more likely to represent schools that are sensitive to the needs of low income children and are highly motivated to provide these children with a good start in school. Second, this study did not directly measure or estimate possible benefits to the children of their Head Start experience. This was impossible because children were not recruited into the study until they were enrolled in kindergarten. Rather, the assumption that at least some children benefitted is based on other studies that suggest Head Start can be helpful. One of the most recent studies of this, using a large nationally representative sample, concluded that 6-year-olds who went to Head Start performed significantly better on tests of receptive language and reading than did their siblings who had not attended preschool (Currie & Thomas, 1995). Also, a comprehensive literature review commissioned by the Head Start Bureau indicated a variety of benefits (McKey et al, 1985), although this report, as well as other critiques, noted that adequately rigorous studies of Head Start’s actual effectiveness have never been conducted (e.g., ACYF, 1990; GAO 1997; Zigler & Meunchow, 1992).
SOURCES OF INFORMATION
The National Transition Demonstration Research Consortium responsible for the final design and conduct of the National Study relied on five sources of information to describe children’s academic progress. These sources of information were:
(1) Direct assessment of each child in the fall of kindergarten and in the spring of kindergarten and each of the next three years in school, via a one-to-one assessment by a trained child examiner during a relatively brief (20-30 minute) session with the child. This direct child assessment concentrates on three skill areas:
*Receptive vocabulary, because of the importance of verbal comprehension to the everyday school setting and the well-established relationship between children’s vocabulary skills and their general intellectual knowledge. This was estimated by using the Peabody Picture Vocabulary Test -Revised (Dunn & Dunn, 1981). Children’s raw scores were transformed into Rasch-Wright scores that take into account children’s ages and “typical” or average levels of performance from a broad and representative sample of children who participated in the norming for this tool. Although the National Consortium recognized that vocabulary tests have been strongly criticized as showing cultural bias and underestimating the true language competency of many children, it was judged that children’s gains in receptive language skills were important to estimate, and no truly “culture free” test was available as a an alternative.
*Reading skills, because reading remains the single greatest predictor of children’s overall success in school (Lyon, 1997) and because it is a skill that is directly taught to all children, for the first time, in the early elementary school years. Reading skills were measured by the Woodcock-Johnson Tests of Academic Achievement Letter-Word Identification and Passage Comprehension Subtests (Woodcock & Johnson, 1989, 1990). Raw scores were converted to Rasch-Wright age-adjusted scores and can be compared in terms of their distance from “reference scores” that reflect children’s average performance, based on a diverse and representative national sample for norming this test. Scores for the two subtests and an overall Broad Reading score were calculated.
*Math skills, particularly because math also is a central feature of the early elementary school curriculum universally and few children are systematically introduced to mathematics activities beyond counting and number recognition, prior to formal schooling. Rasch-Wright scores for Math Computation and Applied Problems subtests and a Broad Math score were computed according to test manual guidelines.
Collectively, a child’s early receptive vocabulary, reading, and math skills form an important foundation on which increasingly complex academic skills are built. Indeed, language arts and math continue as required subjects throughout the K-12 curriculum in almost all schools nationwide. Further, language comprehension, reading, and math skills are essential for children to make good progress in almost every other subject taught in school.
(2) Teacher ratings of the child’s academic standing, in which the teacher compares the child to others in the classroom as well as to grade-expectations in general. These teacher ratings were completed during the spring of each of the first four years in elementary school for all participating children. The standardized questionnaire for this is the Academic Competence Scale of the Social Skills Rating System (Gresham & Elliott, 1990), comprising 9 questions about children’s reading, math, and overall academic achievement. Teachers’ ratings then are converted to standardized scores with a mean of 100 (representing the national average for the norming sample) and a standard deviation of 15.
(3) School records for each child, which provided information about whether the child was promoted versus retained a grade or placed in special education (and if so, for what types of special needs). The review of children’s school records occurred at the end of each academic year and followed the procedures of the School Archival Records Search (Walker, Block-Pedego, Todis, & Severson, 1991) for extracting reliable information from existing school records.
(4) Parent report of the child’s academic progress, based on an interview of parents using Your Child’s Adjustment to School (Reid & Landesman, 1988). Parents rated their children in the spring of each academic year, using a 10-point scale where 10 represented the highest level of achievement and 0 represented the lowest. Items addressed children’s academic progress and overall school adjustment. Parents also were asked to describe the most positive aspects of their child’s year in school as well as any problems or concerns that arose.
(5) Children’s own appraisal of how they are doing in school, using What I Think of School (Reid & Landesman, 1988), a standardized dialogue in which children report in the spring of each year their perceptions about different aspects of their school experiences, including how much they like school, whether they think they are doing well in their schoolwork, how well they get along with their teachers and peers, how much they try to do their best in school, how important they and their parents think school is, and how much they learn from their teachers.
These five sources of information each provide distinct and useful information about the child’s academic performance during the first four years in elementary school. Taken as a whole, the information provides a broad picture of how these former Head Start children fared during the first four years in public school.
In addition, information about children’s social adjustment was obtained from teachers, parents, and children themselves. Teachers and parents each rated children using the Social Skills Questionnaire within the Social Skills Rating System (Gresham & Elliott, 1990), a scale that concerns the child’s broad social and emotional adjustment. The items for parents and teachers are similar, reflecting many aspects of children’s positive social skills, ability to get along with others, use of socially appropriate behavior, independence in social areas, cooperativeness, and lack of serious behavior problems. These ratings then are converted into a standardized score, using a scale where 100 represents the national average for the norming sample with a standard deviation of 15.
FINDINGS ABOUT CHILDREN’S OUTCOMES
Academic performance based on direct assessment of children
The direct assessment of children provides a standardized method for measuring children’s academic skills each year. Eighty-seven percent of the children who were assessed in kindergarten were also assessed in the second or third grade. In this longitudinal sample, seventy-six percent had direct assessment measures available at every year.
These analyses focused on three central questions for the National Study:
(1) How do these former Head Start children do academically in the early
elementary school years? That is, do the children show a pattern of
consistent growth in their academic competency or do they stay relatively
level or decline (that is, “lose the gains” they may have shown
from Head Start)? What is their level of performance each successive year
on standardized tests compared to national means or expectations for children
of comparable ages?
(2) Does participation in the Demonstration Transition Program result in children showing a significant boost in their academic performance compared to those in the comparison condition? To the extent that children in the Transition Demonstration schools might have received a more responsive, individualized, and developmentally appropriate education, then their academic progress might be expected to be better than those in the comparison condition. More specifically, it was hypothesized that children who received all 4 years of schooling in a Transition Demonstration school would show the greatest benefits, based on the belief that continuity of positive experiences in children’s early elementary school education would maximize their academic gains, by increasing children’s comfort with and knowledge about school practices and expectations, as well as by encouraging early successes that would generalize from one year to the next. In fact, only 60% of the original group received the full 4 years of Transition Demonstration intervention (that is, the full “dose”) as planned, and 25% received only 1 or 2 years of planned intervention.
(3) To what extent does the academic performance of former Head Start children show an association with their family characteristics or with child characteristics? The tremendous diversity in the former Head Start families, much greater than expected, led to an early recognition that family characteristics were likely to affect participation in the Transition Demonstration Program, and possibly to affect the degree to which children benefitted from (or needed) the extra supports. Child characteristics, such as gender, age at school entry, disability status, or their initial academic competence when they first entered school, were considered as well, given an impressive diversity in the former Head Start sample that was apparent even at the first round of child assessments. If strong influences are detected between academic progress and family or child characteristics, then it is possible that these factors might also influence the extent to which children benefit (more or less) from their participation in the Transition Demonstration Program.
Performance on Standardized Assessments
Figure 12.1 (page 146) presents a composite picture of the performance of former Head Start children (combining the demonstration and comparison groups) during the first four years in school, based on the children’s mean scores on the four standardized Woodcock-Johnson measures of reading and math skills.1 Separate lines are displayed for children in the two treatment groups. A shaded area indicates national means or expected “average” scores for children of comparable ages (with confidence intervals drawn to reflect standard error of measurement). This composite picture reveals three strong and robust findings from this study:
The first finding is that these former Head Start children entered school somewhat below other children nationally in terms of scopes. This finding affirms what many other studies have consistently shown -- that children from very low income families start school at a disadvantage in terms of their average entry level academic competency. Further, the children’s performance on the Peabody Picture Vocabulary Test, indicating receptive vocabulary skills, places these children at an even more marked disadvantage relative to a national sample, placing them just a little more than one standard deviation below national norms during their first year in public school.
The second finding -- perhaps one of the strongest and most important of the National Study -- is that these children show significant improvement on all 5 measures during these first four years in school. In fact, for the two measures of reading skills and the two of math skills, former Head Start children quickly rise to perform very close to or essentially right at the national averages -- with average standardized test scores for both demonstration and comparison groups of 97 and 98 respectively for Broad Reading Scores and 100 and 101 for Broad Math in the spring of children’s fourth year in school. These average scores are especially impressive, since children who were in Special Education are included, as well as those who repeated a grade. The findings indicate that the sharpest rise in children’s scores occurs between the kindergarten year and the end of the second year in school (first grade, for most of the children) with a somewhat slower incline thereafter. A very slight decline in this rate of improvement appears between the third and fourth years in school only in children’s Letter-Word Identification, and is so small as to be not educationally relevant. The area in which children’s relative gains were the least was on the Peabody Picture Vocabulary Test. However, the children did show significant gains that brought their performance inside the normal range of 100 plus or minus one standard deviation. The effect size for this gain from the kindergarten year through the end of the fourth year in school was .29, a modest significant gain. In contrast, the magnitude of gains for reading and math, measured in effect sizes, were on the order of .36 and 1.01, respectively.
The third finding is that when looking at a composite of the children’s performance from these sites, the year by year performance of the children in the Transition Demonstration and the comparison groups is remarkably similar each year. In the combined sample, comparison children had slightly higher baseline scores than did demonstration children although this very small difference varied from site to site in terms of magnitude and direction. However, it is important to note this national portrayal might be obscuring some group differences that could appear at some local sites, but not others. Given the tremendous diversity across sites in the types of programs implemented, this would not be an unexpected outcome. Similarly, if some groups of children, but not others, benefitted from the Transition Demonstration Program, then this simple comparison of groups by treatment condition may not convey all relevant findings. Thus, a pre-planned series of analyses explored many possible factors or explanatory models that could account for changes in individual children’s growth curves on these 5 academic indicators. The results of these are summarized below.
|
Growth Curve Analyses
Growth curve analyses are used to provide a more sensitive and accurate portrayal of the changes in individual children’s performance over time. The hypothesis being tested is whether children in the Transition Demonstration group show more rapid gains and whether they maintain or accelerate their skill levels in subsequent years at a higher level than comparison children do. These involved calculating individual growth curves for each child on all 5 measures related to academic competence. Specifically, a technique known as B-spline allows modeling the exact shape of the curve over time, and then permits analysis and display of these curves in ways that take into account differences in children’s ages and their entry level skills. A more detailed description of this statistical methodology is provided in Technical Report 5. The results of the growth curve analysis provide a much more accurate depiction of the rate and nature of children’s changes in test performance over time.2
These growth curve analyses were studied for individual sites and reviewed nationally in terms of regularities in patterns of these growth curves. Overall, only five sites had statistically significant, usually small magnitude, differences in children’s growth curves as a function of treatment. Further, the differences were not consistent across test measures and the demonstration and comparison groups did not differ significantly in any consistent way.
Examples of differences in the patterns across the 28 sites included in these analyses are worthy of some scrutiny. In Figure 12.2, an example is provided of the most common pattern, observed in 20 sites, in which children in both the demonstration and comparison group show a rapid early rise over the first two or three years in school in scores on the math and reading subtests and somewhat less of an increase in their receptive language skills. These sites essentially mirror what the national composite showed above. Across the sites, the slope of this early gain, as well as the exact entry levels and final achievement of the children varied. Further, for many sites, a slight drop-off in scores between the third and fourth years in school was noted. In general, however, the patterns toward clear gains from kindergarten to the end of the fourth year in school was remarkable. A second pattern, observed in five sites, was that of improvement to a much lesser degree. Finally, there were only 2 sites that did not show the general national pattern of improvement in children’s scores.
|
Hierarchical Linear Models Applied to Outcomes.
Another data analytic approach, based on growth curves as well, involved applying a complex hierarchical linear model (HLM) to these children’s test scores to evaluate whether multiple influences could be contributing to patterns of change, and the extent to which local site factors may be contributing to outcomes. The same hypothesis is being tested, but the statistical methodology is considered particularly well suited to the study’s design. Specifically, hierarchical linear modeling (HLM) is well suited to handling multiple data irregularities in a complex, longitudinal data set. The results are both complex and informative. First, the findings about treatment effects are neither large nor highly consistent across all outcomes or present in models. Second, some of the HLM analyses do reveal some very small differences in the growth curve patterns of children in the treatment versus comparison groups. When detected, these group differences favored children in the Transition Demonstration group. Third, post hoc analyses to explore the likely basis for theses differences associated with participating in the Transition Demonstration program have not provided clear support for the major hypothesis that sites with the strongest programs would yield the largest benefits, or that sites with the least “competition” from comparison schools would show greater group differences than those with strong competition (that is, Head Start-like supports provided to families and children in both treatment and comparison groups).
Our guiding principle in interpreting the results has been that there must be consistent findings and a reasonably plausible explanation of these findings to substantiate a major conclusion. At the same time, we think that even small or inconsistent findings should at least be mentioned in this final report -- in order to indicate the complexity of issues and to stimulate continued analyses of this unique longitudinal study.
Dr. Margaret Burchinal lead the analyses and authored the following section on the HLM analyses which describes patterns of development from entry to kindergarten through the child’s third or fourth year of school. Child and family, school, and community characteristics that have theoretical links to transition to school were identified. The following variables were included:
| Child/family: | |
| |
gender (child) |
| |
whether English is the primary language in household |
| |
whether the child ever had a IEP for a learning-related disabilities |
| |
maternal education |
| |
whether parent has partner in household |
| |
whether parent is depressed |
| |
family involvement |
| |
parental attitudes |
| School: | |
| |
treatment & treatment implementation |
| |
teacher ratings of school climate |
| |
transition practices |
| |
percent of children in poverty for LEA |
| Community | |
| |
Neighborhood scales (Barriers) |
| |
location of LEA - from inter-city to rural |
Hierarchical linear models were conducted on the five major outcomes (Woodcock Johnson-Reading, Woodcock Johnson-Math, parent rating of children’s social skills, and teacher ratings of children’s social skills) to describe patterns of development of reading, math, and social skills as reported by the parent and teacher. Individual quadratic growth curves were estimated and individual growth parameters were estimated from the selected child/family, school, and community blocks of variables.
Results are shown on the following pages. The top row of the tables list the number of assessments included in the analyses. The next three rows describe the individual growth curve parameters, listing the mean and random-effects variance for the individual growth curve parameters.
The test of the mean coefficient asks whether the parameter estimate, averaged over the sample, differed from zero. This is interesting for the linear and quadratic slopes because it indicates the extent to which children’s scores are changing over time. The test of the mean of the intercept is less interesting because it tests whether the mean score was zero, a value that is not possible on any of the outcomes. The test of variances provides information about whether there are systematic individual differences in terms of overall level, linear rate of change, and quadratic rate of change. As shown in Table 31, the children showed substantial differences in overall level, linear change, and quadratic change for math, reading, and language scores over time.
The rest of the table lists the results of tests of the joint contribution of the child/family, school, and community blocks and individual regression coefficients for predicting the individual growth curve parameters. The block test indicates whether the set of variables is reliably related to that individual growth curve parameter (e.g., the block test for the school variables in predicting the individual growth linear slope parameter tests whether aspects of the school were related to the rate of acquisition of that skill). The individual coefficients indicate the magnitude of the association for a specific predictor, given all of the other variables in the model.
Math (see Table 31). The analyses indicated that children were clearly acquiring math skills during the four years of this study. The individual growth curves suggest that children were scoring at about 425 on average when they were 6 years of age. The rate of acquisition was steeper during the first years of school than during second and third grade, resulting in significant linear and quadratic individual growth curve parameter estimates. These individual growth curve estimates were significantly related to the selected child/family, school, and community predictors.
| |
The child and family block of variables significantly predicted children's overall level, but not rate of linear or quadratic change over time. Higher math scores were associated with being a girl and having a mother with more education, more responsive and nonrestrictive child rearing attitudes, and less depression. |
| |
The school block of variables significantly predicted children's overall level of math skills,linear rate of change, and quadratic rate of change. Each of these variables is discussed separately. |
|
|
|
|
|
|
|
|
| |
The community block was related only the estimated individual intercept. Children in communities with more barriers tended to have slightly lower math scores over time. |
Reading (Table 31). The analyses indicated that children were clearly acquiring reading skills during the four years of this study. The individual growth curves suggest that children were scoring at about 400 on average when they were 6 years of age. The rate of acquisition was steeper during the first years of school than during second and third grade, resulting in significant linear and quadratic individual growth curve parameter estimates. These individual growth curve estimates were significantly related to the selected child/family, school, and community predictors.
| |
The child and family block of variables significantly predicted children's overall level, linear rate of change, and quadratic rate change over time. Higher reading scores at age 6 were associated with being a girl, and having a mother with more education, more responsive and nonrestrictive child rearing attitudes, and less depression. Girls tended to acquire reading skills slightly more quickly, but also showed more deceleration in second and third grade. Similarly, children with mothers with more education, with a partner in the household, with depressive symptoms, and more progressive childrearing attitudes showedfaster acquisition of reading skills, especially during the first years of school. | |
| |
The school block of variables significantly predicted children's overall level of math skills, linearrate of change, and quadratic rate of change. Each of these variables is discussed separately. | |
|
||
|
||
|
||
|
||
| |
The community block was related to all three individual growth curve parameters. Children in communities with more barriers tended to have slightly lower reading scores over time and to gain reading skills slightly more slowly. Children in more urban settings tended to score slightly higher, show gains across time, and less deceleration during second and third grade. |
|
Language (Table 31). The analyses indicated that children were clearly acquiring language skills during the four years of this study. The individual growth curves suggest that children were scoring at about 85 on average when they were 6 years of age. The rate of acquisition was steeper during the first years of school than during second and third grade, resulting in significant linear and quadratic individual growth curve parameter estimates. These individual growth curve estimates were significantly related to the selected child/family, school, and community predictors.
| |
The child and family block of variables significantly predicted children’s overall level, but not rate of linear or quadratic change over time. Higher language scores were associated with having English spoken at home, and having a mother with more education, more responsive and nonrestrictive child rearing attitudes, and more involvement in educational activities. Boys showed slightly more gains over time in language than did girls. Additionally, children whose families did not speak English at home, and children whose mothers were depressed showed slightly more gains in language skills over time, although the gains for children whose mothers were depressed were largely during the first years of school. | |
| |
The school block of variables significantly predicted children’s linear and quadratic rate of change in language skills. Treatment was not related to language scores, but children at schools that used more transition practices tended to show slightly greater gains and less “leveling off” in second and third grades (Figure 12.11). Children in LEAs with more poverty showed slightly greater gains in language initially, but began to lag behind other children by second to third grade. |
|
| |
The community block was related only the estimated individual intercept. Children in communities with more barriers tended to have lower language scores over time. |
|
| Math | Reading | Language | |
|---|---|---|---|
| Number of assessments | 11016 | 11022 | 10556 |
| Individual Growth Curves | B (var) | B (var) | B (var) |
| Intercept | 421.05 (109.2***) | 395.89 (204.5***) | 83.17 (31.25***) |
| Linear Slope | 25.38***(18.96***) | 35.86 ***(154.94***) | 8.74***(2.80***) |
| Quadratic Slope | -1.49***(.92***) | -2.49 ***(10.35***) | -1.04*** (.39***) |
| Population Growth Curve-with child, family, school, and community predictors (includes random age-squared term) | |||
| Site | *** | *** | *** |
Predictors of Intercept |
B (se) |
B (se) |
B (se) |
| Child, family | (block test, p<.0001) | (block test, p<.0001) | (block test, p<.0001) |
| Child gender | 1.02* (.45) | 2.37** (.62) | -.15(.24) |
| English as home lang | -.59 (1.00) | .71 (1.35) | 8.78***(.552) |
| M. education | 1.82*** (.29) | 2.62*** (.39) | 1.41***(.15) |
| Partner in HH | .41 (.48) | .83 (.65) | .35 (.25) |
| M. depression | -0.5217 | -0.8128 | -.37 (.25) |
| Family Involvemt | -.14 (.33) | -.01 (.45) | -.53** (.17) |
| Parent Attitudes (PDI) | 1.68*** (.31) | .84* (.42) | 1.64***(.16) |
| School | (block test, p<.0001) | (block test, p<.0001) | (block test, p=.17) |
| Demonstration trtmt | .61 (.48) | .99 (.66) | .50 (.26) p=.051 |
| School Climate | -.39 (.34) | -.39 (.48) | -.10 (.18) |
| Transition Practices | 1.23*** (.18) | 2.84***(.25) | .08 (.10) |
| % children in poverty | -.16 (.11) | -.10 (.14) | -.06 (.06) |
| Community | (block test, p=.023) | (block test, p=.007) | (block test, p=.0005) |
| Barriers | -.27** (.10) | -0.0462 | -.21*** (.05) |
| locale (urban to rural) | -.15 (.24) | -0.2048 | .02 (.13) |
| Predictors of Linear Rate of Change: Age | |||
| Child, family | (block test, p=.06) | (block test, p<.0001) | (block test, p<.0001) |
| Child gender | -.42 (.40) | 1.78** (.69) | -0.104 |
| English as home lang | -1.41 (.85) | -1.27 (1.42) | -2.38*** (.43) |
| M. education | .42 (.25) | 1.34** (.43) | .25 (.13) |
| Partner in HH | .52 (.42) | 2.02** (.72) | -.26 (.21) |
| M. depression | .15 (.42) | 1.51* (.72) | .50* (.21) |
| Family Involvemt | .57* (.29) | -.87 (.49) | .03 (.15) |
| Parent Attitudes (PDI) | .07 (.26) | 2.18*** (.46) | -.12 (.14) |
| School | (block test, p<.0001) | (block test, p<.0001) | (block test, p<.0001) |
| Demonstration trtmt | .90* (.41) | 1.57* (.71) | -.19 (.21) |
| School Climate | -2.14*** (.43) | -2.92*** (.63) | .04 (.23) |
| Transition Practices | 1.97*** (.23) | 5.54*** (.33) | -0.0533 |
| % children in poverty | .12*** (.02) | .01 (.04) | .04***(.01) |
| Community | (block test, p=.27) | (block test, p<.0001) | (block test, p=.99) |
| Barriers | -.13 (.09) | -.51** (.15) | .00 (.04) |
| locale (urban to rural) | .07 (.10) | -.58** (.18) | .00 (.05) |
| Population Growth Curve, cont.- with child, family, school, and community predictors (includes random age-squared term) | |||
| Predictors of Quadratic Rate of Change: Age-squared | |||
| Child, family | (block test, p=.50) | (block test, p=.0001) | (block test, p=.04) |
| Child gender | -.03 (.13) | -0.1008 | .12 (.07) |
| English as home lang | .07 (.27) | .01 (.44) | .26 (.15) |
| M. education | -.13 (.08) | -.41** (.13) | -.08 (.05) |
| Partner in HH | -.09 (.13) | -0.121 | .11 (.07) |
| M. depression | .04 (.13) | -.37 (.22) | -0.0105 |
| Family Involvemt | -.16 (.09) | -.20 (.15) | .04 (.05) |
| Parent Attitudes (PDI) | .02 (.09) | -.42** (.14) | .04 (.05) |
| School | (block test, p<.0001) | (block test, p<.0001) | (block test, p<.0001) |
| Demonstration treatment | -.22 (.13) | -0.1056 | .03 (.07) |
| School Climate | .84*** (.14) | 1.08*** (.21) | .00 (.08) |
| Transition Practices | -.83*** (.08) | -2.22*** (.11) | .17***(.04) |
| % children in poverty | -.03** (.01) | -.01 (.01) | -.01 (.004) |
| Community | (block test, p=.85) | (block test, p=.0002) | (block test, p=.69) |
| Barriers | .00 (.03) | .08 (.05) | -.01 (.02) |
| locale (urban to rural) | -.02 (.03) | .20*** (.05) | .00 (.02) |
Teacher ratings on SSRS (Table 32). The analyses indicated that teachers tended to rate the children close to the expected mean (100), but that second and third grade teachers were rating the children less positively than kindergarten and first grade teachers. Children showed systematic individual differences in their overall level and linear rate of change over time. In contrast, the quadratic, not linear, rate of change averaged over children was significantly different from zero. These individual growth curve estimates were significantly related to the selected child/family, school, and community predictors.
| |
The child and family block of variables significantly predicted children’s overall level, but not rate of linear or quadratic change over time. Higher teacher ratings were associated with having English not spoken at home, and having a mother with a partner in the home, less depression, and more responsive and nonrestrictive child rearing attitudes. | |
| |
The school block of variables significantly predicted children’s intercept and quadratic rate of change in language skills. Treatment was not related to language scores, but children at schools with higher scores on school climate were rated higher on average. Children in LEAs with more poverty tended to lag farther behind other children by second to third grade (see figure 12.12). |
|
| |
The community block was related only the estimated individual intercept. Children in communities with more barriers tended to have lower teacher ratings over time. | |
Parents ratings on SSRS (Table 32). The analyses indicated that parents tended to rate the children below the expected mean (100), but their ratings increased over time. Children showed systematic individual differences in their overall level and linear rate of change over time. These individual growth curve estimates were significantly related to the selected child/family and community predictors, but not school predictors.
| |
The child and family block of variables significantly predicted children’s overall level and rate of linear change over time. Higher parent ratings were associated with being a girl, having English not spoken at home, and having a mother with more education, less depression, less education activities, and more responsive and nonrestrictive child rearing attitudes. Children in which English was not spoken in the household showed larger changes over time in parent ratings | |
| |
The school block of variables did not significantly predict children’s intercept or linear and quadratic rate of change. However, treatment was related to parents’ ratings at age 6 (see Figure 12.13). Treatment parents rated their children higher on the SSRS than comparison parents. |
|
| |
The community block was related to the estimated individual intercept and quadratic rate of change. Children in communities with more barriers tended to have lower parent ratings over time, with slightly less deceleration in the rate of change. | |
| Teacher | Parent | |
|---|---|---|
| Number of assessments | 8251 | 10810 |
| Individual Growth Curves | B (var) |
B (var) |
| Intercept | 98.19 (101.2***) | 90.87 (117.07***) |
| Linear Slope | .54 (5.92***) | 2.27***(9.81**) |
| Quadratic Slope | -.40*** (1.69) | -.12 (.42) |
| Population Growth Curve- with
child, family, school, and community predictors (includes random age-squared term) |
||
| Site | *** | *** |
Predictors of Intercept |
B (se) |
B (se) |
| Child, family | (block test, p<.0001) | (block test, p<.0001) |
| Child gender | .92 (.63) | -2.89** (.48) |
| English as home language | -5.90*** (1.34) | -3.39** (1.03) |
| M. education | .65 (.40) | 1.03*** (.30) |
| Partner in HH | 1.79** (.67) | -.85 (.51) |
| M. depression | -1.90** (.66) | -2.77*** (.50) |
| Family Involvement | -.55 (.46) | -3.47*** (.35) |
| Parent Attitudes (PDI) | 1.87*** (.43) | 1.53*** (.33) |
| School | (block test, p<.0001) | (block test, p=.26) |
| Demonstration treatment | -.35 (.67) | 1.14* (.51) |
| School Climate | 3.62*** (.60) | .09 (.39) |
| Transition Practices | -.54 (.33) | -.03 (.20) |
| % children in poverty | .03 (.12) | .06 (.11) |
| Community | (block test, p=.023) | (block test, p<.0001) |
| Barriers | -.36*** (.10) | -.74*** (.11) |
| locale (urban to rural) | -.00 (.29) | .48 (.25) |
| Predictors of Linear | Rate of Change: Age | |
| Child, family | (block test, p=.76) | (block test, p=.004) |
| Child gender | .09 (.76) | .30 (.43) |
| English as home language | -.63 (1.57) | -3.27*** (.89) |
| M. education | .06 (.47) | .23 (.27) |
| Partner in HH | .32 (.79) | .36 (.45) |
| M. depression | .28 (.79) | .60 (.45) |
| Family Involvement | .45 (.55) | -.52 (.31) |
| Parent Attitudes (PDI) | -.79 (.50) | .30 (.29) |
| School | (block test, p=.12) | (block test, p=.58) |
| Demonstration treatment | .94 (.78) | .43 (.45) |
| School Climate | 1.03 (.84) | .38 (.47) |
| Transition Practices | .63 (.47) | -.27 (.26) |
| % children in poverty | .13** (.05) | -.01 (.03) |
| Community | (block test, p=.88) | (block test, p=.051) |
| Barriers | -.08 (.16) | -0.0198 |
| locale (urban to rural) | -.02 (.20) | -.04 (.11) |
| Population Growth Curve, cont.-with child, family, school, andcommunity predictors (includes random age-squared term) | ||
| SSRS Teacher | SSRS Parent | |
Predictors Quadratic Rate of Change: Age-squared |
||
| Child, family | (block test, p=.89) | (block test, p=.74) |
| Child gender | .23 (.23) | 0 |
| English as home language | .22 (.47) | .56 (.31) |
| M. education | .01 (.14) | .00 (.09) |
| Partner in HH | -.07 (.24) | -.04 (.15) |
| M. depression | .03 (.24) | -.11 (.15) |
| Family Involvement | -.09 (.17) | .04 (.11) |
| Parent Attitudes (PDI) | .15 (.15) | -.02 (.10) |
| School | (block test, p=.02) | (block test, p=.87) |
| Demonstration treatment | -.29 (.23) | -.15 (.15) |
| School Climate | -.25 (.26) | -.08 (.16) |
| Transition Practices | -.07 (.14) | .02 (.09) |
| % children in poverty | -.04** (.01) | -.00 (.01) |
| Community | (block test, p=.96) | (block test, p=.02) |
| Barriers | .01 (.05) | .09** (.03) |
| locale (urban to rural) | .00 (.06) | .01 (.04) |
|
|
|
|
|
|
|
|
|
|
|
Summary of HLM findings. Undoubtedly, many types of influences contribute to variations in children’s academic and language test scores. The central focus of the study was on the importance of having special types of supports and services provided to maximize positive transition-to-school experiences for former Head Start children. The HLM analysis that considered the greatest number of factors that potentially could influence children’s achievement scores did not find consistent advantages -- of being in the Demonstration, attending schools with a more positive school climate, and the availability of more “transition practices” (such as gathering information about a child prior to entering kindergarten and ensuring curriculum continuity in the early elementary school years).
Because many sites did not implement the Transition Demonstration program in a strong or uniform way (see Chapter 5), we decided to explore in a post hoc descriptive way the extent to which these findings from the National Study may have been diluted in magnitude because of weak program implementation. Further, as discussed in Part 2 of this report, a majority of sites experienced local competition, where the comparison schools frequently provided similar Head Start-like supports to the children and families, sometimes with other special grant funds or local initiatives. In these post hoc analyses, we compared sites that did not have strong competition and also looked at those that implemented stronger versus weaker educational programs and parent involvement programs. Contrary to what we hypothesized, even the stronger programs with the least competition did not show evidence that participating children were more likely to have individual growth curves that were more favorable than those in less positive school settings. Thus, at this point, we conclude that the advantages of the Transition Demonstration Program are not confirmed strongly in any of the analyses. We recognize that this conclusion may be weakened, however, if almost all of those schools had been enacting effective programs.
Teachers’ Judgements of Children’s Academic Achievement
Teachers’ ratings of children’s academic achievement correlated significantly with direct assessment of children’s reading, math, and receptive vocabulary skills. The correlations were much higher between both the reading and math subtest scores and teachers’ ratings of school achievement than they were between receptive vocabulary and teachers’ overall appraisal of children’s academic performance. Specifically, the correlations between teacher ratings in second and third grade and children’s scores on standardized achievement tests of Letter-Word Recognition (r=.54, p<0.0001) and Passage Comprehension (r=.50, p<0.0001) were somewhat higher than for Math Computation (r=.45, p<0.001) and Applied Problems (r=.50, p<0.001), which in turn were much higher than the correlation with the Peabody Picture Vocabulary Test (r=.31, p<0.001). Not unexpectedly, all of these correlations were somewhat lower during the kindergarten year, except for the Peabody Picture Vocabulary Test, which was essentially the same. During the kindergarten year, teachers may have fewer opportunities to directly assess children’s emerging literacy and math skills, since these are not taught systematically until the first grade.
| Brief Description | Resourceful | Single Parent Welfare | Foreign Language | Highly Mobile | Mother Absent | Chronic Health Problem | Recently Homeless | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| D | C | D | C | D | C | D | C | D | C | D | C | D | C | |
| WJ-Reading Standard Score | ||||||||||||||
| K | n=800 | n=761 | n=574 | n=484 | n=110 | n=85 | n=93 | n=84 | n=88 | n=76 | n=53 | n=52 | n=61 | n=45 |
| 90.61 | 90.11 | 87.7 | 87.02 | 86.17 | 89.01 | 86.75 | 83.61 | 87.81 | 87.55 | 84.7 | 90.92 | 86.11 | 85 | |
| (13.36) | (12.96) | (12.7) | (12.25) | (12.85) | (14.43) | (13.05) | (12.2) | (13.31) | (12.51) | (10.13) | (11.84) | (10.72) | (11.81) | |
| 3 | n=816 | n=778 | n=593 | n=500 | n=122 | n=93 | n=103 | n=85 | n=91 | n=77 | n=57 | n=58 | n=63 | n=47 |
| 100.22 | 101.41 | 94.93 | 95.67 | 95 | 99.35 | 96.39 | 94.58 | 96.3 | 97.65 | 96.75 | 98.52 | 95.51 | 92.26 | |
| (16.32) | (15.66) | (17.08) | (16.9) | (14.86) | (15.28) | (17.67) | (15.06) | (15.76) | (15.91) | (17.01) | (18.64) | (14.39) | (15.12) | |
| WJ-Math Standard Score | ||||||||||||||
| K | n=800 | n=761 | n=574 | n=485 | n=110 | n=85 | n=93 | n=84 | n=88 | n=76 | n=53 | n=52 | n=61 | n=45 |
| 86.11 | 86.79 | 82.77 | 81.74 | 81.88 | 82.65 | 83.68 | 78.69 | 81.89 | 82.93 | 80.98 | 86.31 | 81.3 | 81.36 | |
| (15.93) | (16.21) | (16.14) | (15.55) | (17.4) | (16.92) | (16.95) | (17.47) | (15.46) | (15.01) | (15.87) | (15.49) | (15.2) | (16.93) | |
| 3 | n=816 | n=779 | n=592 | n=500 | n=122 | n=93 | n=103 | n=85 | n=91 | n=77 | n=57 | n=58 | n=63 | n=47 |
| 104.08 | 104.28 | 96.8 | 98.29 | 100.54 | 103.2 | 97.55 | 95.14 | 97.44 | 99.05 | 97.68 | 99.67 | 96.13 | 100.21 | |
| (17.66) | (16.81) | (18.85) | (19.07) | (14.99) | (16.97) | (18.72) | (17.06) | (17.52) | (16.66) | (20.00) | (21.33) | (16.09) | (19.18) | |
| Social Skills Rating System-Teacher (Academic ) | ||||||||||||||
| K | n=796 | n=770 | n=587 | n=488 | n=117 | n=91 | n=97 | n=83 | n=90 | n=77 | n=54 | n=55 | n=60 | n=46 |
| 92.86 | 93.12 | 87.72 | 87.63 | 92.64 | 95.75 | 87.44 | 86.04 | 87.57 | 87.78 | 88.76 | 90.76 | 86.4 | 88.61 | |
| (15.14) | (14.84) | (15.41) | (14.92) | (13.75) | (14.14) | (15.74) | (15.44) | (16.48) | (12.99) | (15.69) | (15.43) | (16.37) | (14.93) | |
| 3 | n=817 | n=779 | n=592 | n=499 | n=122 | n=92 | n=103 | n=85 | n=91 | n=77 | n=57 | n=58 | n=63 | n=47 |
| 91.4 | 92.06 | 87.41 | 86.15 | 92.84 | 93.79 | 87.95 | 86.99 | 85.04 | 87.03 | 88.25 | 90.02 | 85.75 | 87.34 | |
| (14.68) | (14.39) | (14.66) | (14.88) | (13.28) | (14.32) | (15.57) | (14.29) | (14.65) | (14.9) | (17.39) | (16.5) | (17.28) | (14.34) | |
* Longitudinal sample includes scores for family units in which there is a family interview, child assessment, and teacher interview in kindergarten (fall and/or spring) and end point (second and/or third grade). Data from 29 sites included in analysis.
Overall, two interesting findings emerged about teachers’ ratings of children’s academic competence. One was that teachers tended to rate children somewhat lower compared to national norms than their actual test performance indicated. The reasons for this are not readily apparent. One possibility is that teachers may be making comparisons within the classroom between former Head Start children and non-Head Start children, and may see these children as still somewhat lower in their overall academic competency -- even if objective indicators place them close to or above national averages. Another possibility is that teachers’ ratings may take into account much more than just the children’s math and reading skills that are assessed on the Woodcock-Johnson, and thus may reflect some true differences in children’s overall fund of knowledge or other classroom skills. In fact, teachers’ average ratings of the children place them toward the low end of the normal range.
A second finding is that teachers’ ratings do not suggest any pattern of gains or growth in children’s relative academic standing, unlike the direct assessments of their skills. Again, the reasons for this are not readily apparent. Each year, different teachers rate the children. Do these teacher ratings indicate possible bias or negative appraisal of children that follow them from year to year and obscure true gains they are realizing? Or do the ratings by independent teachers prevent teachers from observing the relative gains former Head Start children have made since their kindergarten entry? In the absence of information about the trends for classmates who were not previously in Head Start and those who are not from low income homes, it is difficult to reach any conclusion.
Parent Ratings of Children’s Academic Adjustment
The parents’ ratings of children provide an exceptionally positive picture. Even though a scale of 0-10 points was provided, the parents as a group used only the upper ranges. Table 34 summarizes parent ratings, providing means and standard deviations where 10 is the highest rating. No significant group differences were found between Transition Demonstration and comparison groups at either the start or the end of the study period. Also, parent ratings did not show much change between the spring of kindergarten and the end of the third or fourth year in school, although their overall estimation of how well their children were doing was slightly lower at the end for how well children were doing academically, how well they got along with their teachers, and their overall school adjustment.
Although parent appraisals of their children’s school performance are undoubtedly important for families and for children, the limited range in these scores indicates that parents’ ratings do not show an orderly relationship to either teacher appraisals or to children’s objective test performance.
| First year | Final year | |||
|---|---|---|---|---|
| Demonstration Mean (s.d.) | Comparison Mean (s.d.) | Demonstration Mean (s.d.) | Comparison Mean (s.d.) | |
| How much child likes school | 8.7 (1.76) | 8.7 (1.73) | 8.0 (1.97) | 8.0 (1.98) |
| How much effort child puts into trying to do well in school | 8.3 (1.76) | 8.33 (1.72) | 7.82 (1.95) | 7.8 (1.96) |
| How well child actually does in school | 7.7 (1.79) | 7.7 (1.81) | 7.4 (1.93) | 7.4 (2.00) |
| How well child gets along with teacher | 8.6 (1.76) | 8.6 (1.89) | 8.2 (2.14) | 8.1 (2.10) |
| How well child get along with other children at school | 7.8 (1.85) | 8.0 (1.82) | 7.7 (1.91) | 7.7 (1.95) |
| Rating of overall adjustment to school | 8.4 (1.62) | 8.4 (1.72) | 8.1 (1.87) | 8.0 (1.86) |
Children’s Impressions of Their Own School Adjustment
|
|
Like parents, children presented a positive picture of their school adjustment, although there was somewhat greater diversity in children’s own ratings of how well they were doing. This is particularly noteworthy since children used only a 3-point scale indicating very positive, moderate, or somewhat low or negative impressions for each item. Figure 12.14 summarizes children’s positive appraisals of schools. Age-trends are apparent for 7 of the 8 items analyzed. Specifically, older children provide less positive ratings for how much they like school; how well they think they are doing in school -- the area with the most marked decline over the first four years in school; and how well they get along with peers. Over these same years, they also indicate that they think school is increasingly important, that they really want to do well in school, and that they try hard to do their best -- with over 90% of the former Head Start children rating school as very important and indicating they try at high levels to do well. They also confirm that the vast majority of their parents think doing well in school is very important -- with 95 percent rating their parents as valuing school at the highest level by the end of the fourth year in school. These impressions clearly defy a negative stereotype that these children and their parents do not believe that school is important or that the children are not motivated to do their best. Interestingly, even in the spring of kindergarten, the clear majority of children (81%) rate their teachers as being very good at helping them learn new things, and these positive appraisals of teachers increase to even higher levels (88%) by the end of their fourth year in school.
An earlier report indicated that only a small percentage of former Head Start children (Ramey, Lanzi, Phillips, & Ramey, 1998) provide extremely negative self-appraisals of their progress in school and how much they like school -- about 7 percent. These children, in fact, are ones who show significantly poorer academic progress as indicated by both their test scores on the Woodcock-Johnson Tests of Achievement (see Figure 12.15) and their teachers’ ratings of their academic and social development (Figure 12.16). Note that the children’s receptive language scores (PPVT) do not differ significantly at school entry or thereafter.
|
|
These findings indicate that children’s self-appraisals, even as early as kindergarten, when they are quite young and cognitively not very sophisticated in making comparative judgments, reflect that children’s own reports about school adjustment provide meaningful information. Further analyses of children’s self-reports indicate that including their self-appraisal in a statistical model predicting school success significantly increases the fit of this model. That is, children’s impressions of how much they like school and how well they are doing provides useful information over and above their entry level skills, teacher ratings, and parent ratings (S. Ramey, Lanzi, Phillips, & Ramey, 1999). Further, as Figure 12.17 shows, former Head Start children who report in kindergarten that things are not going well are significantly more likely to be placed in special education by the end of their fourth year in school. This is particularly noteworthy because their rates of special education placement were comparable in kindergarten. Yet, the children who reported less favorably on their school experiences had more than a twofold increase in special education assignment -- from 1 percent in kindergarten to 24 percent four years later.Indeed, those with low academic self-ratings also have higher than average rates of self-reported difficulty in getting along with peers and teachers, and do not think their parents place as high a value on doing well in school.
Social Adjustment
In kindergarten, teachers’ ratings of children’s social development indicate that children in both the demonstration and comparison groups are essentially at the national average (national mean = 100 with 3 points standard error of measurement). Teachers continued to rate children’s social adjustment as quite close to national average.
| Teacher | Parent | |||
|---|---|---|---|---|
| Demonstration Mean (s.d.) | Comparison Mean (s.d.) | Demonstration Mean (s.d.) | Comparison Mean (s.d.) | |
| First Year | 99.5 (16.3) | 99.2 (16.1) | 90.4 (15.2) | 91.6 (15.0) |
| Second Year | 98.0 (16.7) | 98.1 (16.4) | 93.4 (15.6) | 95.2 (15.8) |
| Third Year | 98.3 (16.6) | 97.7 (16.6) | 95.3 (16.2) | 97.6 (16.0) |
| Fourth Year | 97.3 (16.5) | 96.8 (17.0) | 96.8 (15.9) | 97.7 (15.9) |
In contrast, parents initially rate their children almost 10 points lower than did teachers, placing them at the low end of the normal range. Over the four years in school, however, parents’ ratings show a clear increase, such that by the end of the third or fourth year in school, the national mean for parents’ ratings is just slightly below (by 2 to 3 points) the national average of 100.
Teachers’ and parents’ ratings of individual children’s social adjustment showed only very low levels of correlation, from r = 0.15 (p < 0.001) in the spring of kindergarten to a high of r = 0.20 (p < 0.0001) at the end of the fourth year in school. This is commonly the case with ratings of children’s behavior and probably reflects differences in how children behave in home versus school, as well as the fact that parents and teachers may have different standards. This low correlation indicates several likely influences: parents and teachers may be attending to somewhat different aspects of a child’s behavior; children may display different levels or types of behavior in school than they do at home; and parents and teachers may have different standards in mind when rating their children’s social behavior. The finding that parents initially gave much lower ratings of their children than did teachers was not expected. Further, even though overall averages for teacher and parent ratings were similar four years later, their ratings for an individual child still were only modestly in agreement. This indicates that some children who are rated very highly by their teachers are not necessarily the same children rated the most highly by their parents (and vice versa). The overall increase over time in parents’ ratings may reflect their appreciation of children’s adjustment to school, increased opportunities to observe children’s social competencies in new areas, and/or that children are truly displaying increases in social development at home that are in line with parent expectations.
As in the area of academic competence, the results regarding significant differences between demonstration and comparison groups are complex. No significant differences in teacher ratings were found, although HLM analyses detected a small effect when considering parent ratings, when taking into account baseline group differences. Further, site by site analyses did not reveal any consistent pattern of significant treatment effects. Because the children showed good levels of social adjustment (close to the national average of 100) at all time periods, based on teacher ratings, there was no reason to expect that the Transition Demonstration intervention would result in above average performance.
Multi-perspective definition of “successful transitions”
The National Transition Research Consortium endorsed a definition of “successful transitions” for children that emphasizes multiple, interactive components, integrating good academic and cognitive development with social adjustment, positive feelings about school and learning, and mutually supportive relationships among children, families, and schools. Following on this conceptualization of transition as a multi-faceted construct, a multi-perspective definition of transition-to-school was developed. Six indicators of successful transition were identified and scored in the final year of study participation (the child’s third or fourth year of school):3
| |
The parent indicated that the child’s academic and overall adjustment to school were good; | |
| |
The child indicated that his or her own adjustment to school was good, by indicating that they liked school and felt that they did well in school; | |
| |
The teacher indicated satisfactory social adjustment; | |
| |
The teacher indicated satisfactory academic adjustment, comparing the child to others in the classroom and to grade level expectations; | |
| |
Objective, standardized assessments of reading skills placed the child’s achievement in reading at or above average, based on national norms; and | |
| |
Objective, standardized assessments of math skills placed the child’s achievement in math at or above average, based on national norms. |
|
Dichotomous indicators (yes = criterion was met; no = criterion was not met) were created for each of the six indicators and then summed to obtain an overall transition score (range 0 to 6). The transition scores were then divided into three categories: (1) Highly successful transition, (transition score of 5 or 6); (2) Moderately successful transition (transition score of 2, 3, 4); and (3) Poor transition (transition score of 0 or 1). Thus, to be characterized as having achieved a “highly successful transition,” the child had to be broadly judged as having made a good adjustment to school and have exhibited average to above average achievement in key academic areas.
Just under a third of these former Head Start children (30%) met these stringent criteria and were considered to have made a highly successful transition to school. Only 14 percent made poor transitions, and the remainder were moderately successful.
Analyses indicated that children of caregivers not born in the United States were nearly twice as likely to experience a highly successful transition as children of U.S.-born caregivers. This association did not appear to be related to the language spoken in the home, and thus may reflect a different attitude toward school and learning on the part of these families. Not unexpectedly, a greater level of education on the part of the caregiver was also associated with a greater likelihood of a highly successful transition for the child and less likelihood of a poor transition. In addition, analyses indicated that children of families experiencing greater and more chronic poverty were less likely to experience highly successful transitions and more likely to experience poor transitions. Similarly, children whose families moved more often during the early school years were less likely to experience highly successful transitions and more likely to experience poor transitions.
Special education placement and grade retention
|
Based on reviews of individual school records, children placed in special education and those retained in grade (that is, were not promoted to the next grade) at least once during the first four years in school, were identified. Figure 12.18 shows the percentage of children who were in special education for at least one year in the first four years in school: 24 percent of those in the Transition demonstration group and 20 percent of those in the comparison group. This small but statistically significant difference might be attributable to increased vigilance for the children in the Transition Demonstration group, although this cannot be affirmed. Because the children in the two treatment conditions did not differ significantly on other tests of academic and social development, it is unlikely that these differences in special education placement reflect greater learning disability or social maladjustment. Indeed, whether special education placement benefits children or not continues to be vigorously debated (e.g., Detterman & Thompson, 1997; S. Ramey & Ramey, 1998) and is likely to vary depending on the quality of special education supports in a school district. A noteworthy finding in this study is the tremendous variation in rates of special education across sites. These rates ranged from less than 5 percent (in 10 sites) to more than 20 percent (in 2 sites).
Of those children who were placed in special education at some time during their first four years in school, approximately 40 percent of these were placed prior to or during kindergarten, 23 percent during the second year of school, 20 percent during the third year, and 18 percent during the fourth year (see Figure 12.19). The timing of the children’s placement into special education did not differ significantly by treatment condition. That is, Transition Demonstration schools did not systematically identify and place children in special education programs either earlier or later than those in comparison schools.
|
The frequency and duration of out-of-class placements for special education students were expected to relate to participation in the Transition Demonstration Project, since mainstreaming or inclusion of special education students was a stated goal for the Transition Demonstration program. Although the percentage of children being fully included (that is, receiving all instruction in a regular education classroom) did not differ significantly over the four years as a function of the treatment group, there was a consistent pattern of partial-day inclusion favoring the demonstration schools (see Figure 12.20). For the years after kindergarten, the percentage of special education children in demonstration schools maintained in the regular classroom for at least part of the day was significantly greater than in comparison schools. Similarly, the percentage of special education children in demonstration schools who were assigned to a full-day program out of the regular classroom was lower than the percentage of special education children in the comparison schools. The difference between the two groups increased each year until third grade, when it diminished. About one-fifth of the special education children were also grade-retained. Specifically, 22 percent of those in the Transition Demonstration and 20 percent in the comparison group were held back at least one grade.
|
Concerning overall rates of grade retention, there were no significant differences between treatment groups: 11.8 percent of those in the Transition Demonstration group and 10.7 percent in the comparison group were held back at least once in the first four years in school. Figure 12.21 shows overall percentages of children who were: grade-retained only, placed in special education only, both grade-retained and placed in special education, and promoted annually and never placed in special education. The majority of former Head Start children were never grade-retained or placed in special education -- nearly seven of every 10 children. Nonetheless, rates of placement in special education are significantly higher than the national average, which is about 10 percent overall (U.S. Department of Education, 1997). An estimate for Head Start children nationally was 14 percent (O’Brien et al., 1997). Reliable national data about rates of grade retention were not available (to our knowledge) for these years by grade level.
|
Figure 12.22 shows that children who were either placed in special education and/or grade retained performed at significantly lower levels on all measures than did other children. Especially impressive is the finding that former Head Start children who are either grade-retained or placed in special education are virtually indistinguishable from each other as groups. That is, it is not clear from these data why decisions were made for special education versus grade retention. Further, both groups of children continue to show both academic and social progress over the years, although not as marked in their standardized test scores on reading and math as were children who were grade-promoted and not placed in special education.
|
SUMMARY FINDINGS
In conclusion, former Head Start children in this study showed quite favorable early school adjustment by all indicators. Despite entering kindergarten with scores somewhat below national norms, the children significantly improved their performance in both reading and math, such that they were essentially at national averages by the end of second and third grades. In other words, although Head Start children entered school with academic skills that were somewhat weaker than average, this does not appear to have impeded their ability to make rapid progress once formal schooling began. Vocabulary skills, as measured by the Peabody Picture Vocabulary Test - Revised, indicated somewhat lower levels of progress, although scores still were within the normal range. Further, teachers’ ratings of the social skills of these former Head Start children place them at national norms, from kindergarten on through the first four years in public school.
Both parents and children confirm this portrayal of good school adjustment for the overwhelming majority of former Head Start children. Their impressions of children’s overall school adjustment, academic performance, social development, and motivation to do well in school -- a high value for this group of children and parents -- are consistently positive as a group. A small subgroup of children, however, reported early impressions of school that were less favorable. For these children (7 percent), their school progress was much less favorable both academically and socially. Indeed, the children with early non-optimal school impressions were at very high risk for special education placement.
Family characteristics, as expected, related to aspects of children’s academic and social adjustment. Consistent with extensive reports elsewhere (cf. S. Ramey & Ramey, 1998), children from families with higher levels of resources (including parent education, parent employment, income, two parents active in the child’s life, stable residence) entered school with generally higher skills and continued to perform at higher levels. Children from families where English was not the primary language spoken at home showed excellent progress in both reading and math, and very high levels of social skills, despite their lower scores on receptive vocabulary and early reading comprehension. Even children experiencing considerable life challenges such as homelessness, frequent moves, or a parent with a chronic health problem, showed good early gains in academic skills.
1The sample used in these analyses included children who had individualized assessments of academic skills (Woodcock Johnson Tests of Achievement) in kindergarten and in either the third or fourth year of school. The total sample, as noted above, included 5,914 former Head Start children. (back)
2Extensive analyses showed that the conclusions are unchanged when analyses are conducted on those children who have 3, 4 or 5 sets of test scores (5 representing a full set of scores from fall and spring of kindergarten and spring of each of the subsequent three years in school). Therefore, the results presented in this report are based on the children who had complete test scores at all ages (65% of the total in the longitudinal analysis data set). The sample included a total of 3,474 subjects. This sample did not include those children who had been retained in grade, but did include children placed in special education programs. (back)
3Successful transitions analyses were completed using a total of 5,914 former Head Start children and families who were present at baseline and at the end of the study (either third or fourth year data available) and had child assessment data available. A total of 28 sites were represented within this data set. The multi-perspective successful transitions outcome variable was defined using six indicators scored in the final year of study participation (using the later of the third or fourth year data). There informants (family, child, teacher) provided responses that were scored as follows: a) The parent indicated that the child’s adjustment to school was good. That is, the parent, completing Your Child’s Adjustment to School, rated both the child’s adjustment and overall school adjustment as 5 or greater (on a 10-point scale). Higher ratings on both questions were required to meet criterion.
b) The child indicated that his or her own adjustment to school was good by indicating (via What I Think of School) that they liked school and felt that they did well in school (ratings of 2 or 3 on a 3-point scale). Higher ratings on both questions were required to meet criterion.
c) The teacher indicated satisfactory social adjustment [Social Skills Rating System, Social Skills Questionnaire, standard score of 95 or better (mean=100, standard deviation=15)].
d) The teacher indicated satisfactory academic adjustment [Social Skills Rating System, Academic Competence Scale, standard score of 90 or better (mean=100, standard deviation=15)].
e) The child achieves at an average level (for age) in reading (Woodcock Johnson Tests of Achievement, Broad Reading, standard score of 95 or better (mean=100, standard deviation=15).
f) The child achieves at an average level (for age) in math (Woodcock Johnson Tests of Achievement, Broad Math, standard score of 95 or better (mean=100, standard deviation=15).Binary indicators (0 = did not meet criterion; 1 = met criterion) were created for each of the six indicators. The six binary scores were then summed to create an overall adjustment score (range 0 to 6), and a categorical “success” variable was created as follows:
Poor success = overall adjustment score of 0 or 1
Moderate success = overall adjustment score of 2, 3, or 4
High success = overall adjustment score of 5 or 6.The SUCCESS score was created for a total of 4,558 children. The remaining 1,356 children were missing key information on one or more of the six individual indicators (typically the teacher ratings of social and academic adjustment). Key covariates considered for inclusion in the predictive models included: gender, age at entry into kindergarten, child has a health condition that interferes with school attendance, poverty status of family (in poverty at both beginning and ending assessments, at one assessment, or at neither assessment), language spoken in the home (English versus other), caregiver employed full-time, caregiver educational status, caregiver has chronic health condition that interferes with ability to care for child, family moves with frequency, family has been homeless at some time in the early school years, initial treatment condition, years in a demonstration school, and level of implementation of Transition Demonstration Program. (back)
| Table of Contents | Previous | Next |