Professional Documents
Culture Documents
Preview
Preview
Preview
W
LONGITUDINAL READING ACHIEVEMENT
IE
EV
by
SHANE M. BASSETT
PR
In the unlikely event that the author did not send a complete manuscript
and there are missing pages, these will be noted. Also, if material had to be removed,
W
a note will indicate the deletion.
IE
EV
UMI 3400051
PR
ProQuest LLC
789 East Eisenhower Parkway
P.O. Box 1346
Ann Arbor, MI 48106-1346
i
Abstract
This research project investigated the effect of kindergarten time schedule on the reading
(n=53), mid-risk (n=59), and low-risk learners (n=44). Students entering kindergarten at
five Reading First schools in September 2004 (n=435) were randomly assigned to full-day
kindergarten (FDK) or half-day kindergarten (HDK). Students were pre-tested using the
Dynamic Indicators of Basic Early Literacy Skills (DIBELS); Letter Naming Fluency and
W
Initial Sounds Fluency subtest scores were added together to form a pretest composite
IE
score. The achievement of the intact sample (n=156) was followed through June 2007. The
students were post-tested at the end of second grade using the Stanford Achievement Test,
EV
Tenth Edition (SAT-10) and DIBELS Oral Reading Fluency (ORF) measure; the raw
scores of these two measures were added together to form a posttest composite
score/dependent variable. Students were administered DIBELS three times each year, and
PR
the SAT-10 at the end of each year. Independent samples t-tests performed on the whole
sample’s (p < .01) pretest scores indicated initial group differences favoring FDK; a t-test
performed on the high risk sub-sample’s (p < .01) pretest scores indicated initial group
differences favoring HDK. From the middle of kindergarten until the end of second grade,
FDK students outperformed their HDK peers on every DIBELS and SAT-10 measure. On
the middle of first grade DIBELS ORF, on the end of second grade DIBELS ORF, and on
the kindergarten, first, and second grade SAT-10 measures, the FDK students in the whole
sample (p < .01), and in the high-risk sub-sample (p < .01 to p < .05) significantly
ii
outperformed the HDK students. ANCOVA results on the whole sample (p < .01) and t-test
results on the high-risk sub-sample (p < .01) favor FDK. Multiple regressions performed on
the whole sample (p < .01) and high-risk sub-sample (p < .05) indicate that the FDK time
condition had a significant effect on the dependent variable after controlling for pretest
scores and initial demographics. The researcher discusses implications of the results for
teachers, principals, school district resource assignment and state policy development in
W
IE
EV
PR
iii
TABLE OF CONTENTS
List of Tables................................................................................................................... v
List of Figures................................................................................................................vi
CHAPTER 1 ...................................................................................................................1
Introduction ..................................................................................................................................... 1
Statement of the problem ............................................................................................................................. 3
Research questions ....................................................................................................................................... 3
Definition of terms ....................................................................................................................................... 4
Limitations and delimitations....................................................................................................................... 4
CHAPTER 2 ...................................................................................................................9
W
Review of the Literature................................................................................................................. 9
Conclusions about alternate-day kindergarten........................................................................................... 10
Early research comparing half-day and full-day kindergarten conditions................................................ 12
Research from the 1990s favoring half-day kindergarten/showing no significant difference ................. 17
IE
Research from the 1990s favoring full-day kindergarten.......................................................................... 18
Summary research favoring full-day kindergarten.................................................................................... 22
Recent research on the relative effectiveness of time schedule on student achievement......................... 25
Specific research about full-day kindergarten for children at high-risk of reading failure ...................... 33
EV
Conclusions from the literature review...................................................................................................... 36
CHAPTER 3 .................................................................................................................38
Methods .......................................................................................................................................... 38
Research questions ..................................................................................................................................... 38
PR
Setting ......................................................................................................................................................... 39
Participants ................................................................................................................................................. 40
Human subjects safeguarding .................................................................................................................... 41
Instrumentation........................................................................................................................................... 42
Research design.......................................................................................................................................... 46
Role of the researcher................................................................................................................................. 47
Procedures .................................................................................................................................................. 47
Data analysis procedures............................................................................................................................ 50
CHAPTER 4 .................................................................................................................61
Results............................................................................................................................................. 61
Initial mean differences in achievement of the intact groups.................................................................... 62
Longitudinal mean differences in whole sample and sub-sample achievement....................................... 65
Pretest-posttest analysis with ANCOVA and t-tests ................................................................................. 72
Pretest-posttest analyses on whole sample ........................................................................................... 72
Pretest-posttest analyses on high-risk sample....................................................................................... 73
Pretest-posttest analyses on mid-risk sample........................................................................................ 74
Pretest-posttest analyses on low-risk sample........................................................................................ 74
Multiple regression analysis....................................................................................................................... 78
Multiple regression results for whole sample....................................................................................... 81
Multiple regression results for high-risk sample .................................................................................. 82
iv
Multiple regression results for mid-risk sample ................................................................................... 83
Multiple regression results for low-risk sample ................................................................................... 84
Summary of results..................................................................................................................................... 85
CHAPTER 5 .................................................................................................................87
Discussion ....................................................................................................................................... 87
Comparison of results to past research ...................................................................................................... 88
Length of study...................................................................................................................................... 90
Stratification of sample into risk categories.......................................................................................... 91
Methodology.......................................................................................................................................... 91
Limitations.................................................................................................................................................. 94
Implications ................................................................................................................................................ 95
Teacher and school level resource assignment..................................................................................... 95
School district resource assignment...................................................................................................... 97
Regional and statewide resource assignment and policy ..................................................................... 99
Future research ......................................................................................................................................... 101
Conclusions .............................................................................................................................................. 103
W
References...................................................................................................................104
APPENDICES............................................................................................................116
Appendix A .................................................................................................................116
IE
Appendix A.1................................................................................................................................ 116
Appendix A.2................................................................................................................................ 117
EV
Appendix A.3................................................................................................................................ 118
Appendix A.4................................................................................................................................ 119
Appendix A.5................................................................................................................................ 120
Appendix A.6................................................................................................................................ 121
PR
LIST OF TABLES
Table 2 Measures of Beginning, Middle and End of Kindergarten, First, and Second
Grade……………………………………………………………………...55
Table 3 Initial Group Differences by Time Condition and Risk Level ……………63
W
Table 4 Mean Stanine Rankings of Students on the Standard Achievement Test,
Table 5
IE
Mean Scores on DIBELS ORF1 and ORF2 by Time Condition and Initial
Risk Level………………………………………………………………...69
EV
Table 7 SPSS Coding for Each Variable Included in the Multiple Regression
Analysis…………………………………………………………………...79
vi
LIST OF FIGURES
Figure 3 Initial Group Differences by Time Condition and Initial Risk Level……..64
W
Figure 5 SAT-10 Reading Achievement by Time Condition and Initial Risk Level
…………………………………………………………………………….70
Figure 6
IE
Longitudinal DIBELS Oral Reading Fluency by Time Condition and Initial
Risk Level………………………………………………………………...71
EV
Figure 7 DIBELS Oral Reading Fluency Achievement by Time Condition and Initial
Risk Level………………………………………………………………...71
PR
CHAPTER 1
Introduction
to Robinson (1987), 39 states offered half-day kindergarten (HDK) in 1986 and eight states
provided full-day kindergarten (FDK). By 1990, 40% of the kindergarten population was
W
attending full-day kindergarten (Kauerz, 2005). By 2005, 63% of kindergarten students
kindergarten will help their children better adjust and perform in school, and meet the need
EV
for high-quality, affordable child-care for working, and single parents. Parrett (2005) and
Tirado (2001) have reported that full-day kindergarten is one intervention among many that
PR
has been absolutely essential to improving the achievement realities of the severely high-
One of the major issues affecting any school district as they begin to implement a
kindergarten to all children, how would a school district determine which students to allow
In one medium sized school district, where the current study took place, the School
Board implemented one full-day kindergarten class at each district school. Teachers,
2
parents, and administrators had differing ideas about how to assign students to the full-day
kindergarten program. There was no consistent opinion between any of these groups. The
senior administration preferred a lottery system of assignment. The rationale for this point
of view rested in their belief that parents would perceive a system based on need as unfair.
Most teachers, on the other hand, suggested a system based on need, where the most high-
risk students would receive full-day kindergarten. More affluent parents preferred the full-
day schedule because it could ease their child-care expenses, and they knew it would be a
way for their students to get a head start in academic subjects. Many working-class parents
W
simply wished to lower their child care expenses. Parents from lower-SES families
preferred a needs-based system they knew would benefit their children. In short, there was
IE
a great deal of divisiveness around how to assign students to full-day kindergarten. These
positions were based on assumptions, opinions, and emotions, with little reference to any
EV
When the school district decided that they could afford to fund one full-day
PR
kindergarten classroom at each school, they also decided to use this opportunity to compare
the achievement of these students to those in the traditional half-day program. Thus, the
school district randomly assigned registered students from each school into full-day and
half-day kindergarten until the full-day kindergarten at each school was 50% full. Then,
they filled the remaining 50% of slots in each full-day kindergarten class with randomly
assigned students who were registered by the day before school started in September 2004.
Late registrants were placed into half-day kindergarten classes. The school district asked
the researcher to track the achievement of this class of kindergarten students at least
through the end of second grade. The current researcher presented the preliminary findings
3
to the School Board at Spring 2004, Winter 2005, Spring 2006, and Fall 2006 meetings.
Based on the Spring 2004 findings, which showed that full-day kindergarten benefited
children from all No Child Left Behind (NCLB) sub-groups, the school district
implemented full-day kindergarten in all elementary schools in the district. This research
study constitutes the final quantitative evaluation of the original intervention, and will
inform the continued implementation of full-day kindergarten in the school district and
others like it. It is hoped that this study will shed light on the effects of the full-day
W
Statement of the problem
The general purpose of this research study was to provide quantitative data in
IE
regard to full-day kindergarten implementation, rigorous data analysis, and well-conceived
the purpose of this study was to compare the longitudinal effect of full-day kindergarten
PR
and half-day kindergarten time schedules on the reading achievement of high-risk, mid-
risk, and low-risk kindergarten students over a three year period, while controlling for
Research questions
Thus, there were three research questions that extended from the overall purpose of
this research:
Definition of terms
1. An alternate-day schedule is defined as when students attend three full days one week
and two full days the next week with the remaining days off, or attend two full days each
week with a half-day schedule on Wednesday, for a total of five full days of instruction
W
2. A full-day schedule occurs when students attend all day, every day.
3. A half-day schedule is when students attend kindergarten for a half day five times per
IE
week, either in the morning or afternoon.
that have been prescribed and monitored at the schools participating in the study. The
students in this study attended schools that have reading programs that are prescribed and
PR
5. The operational definition of reading achievement means the scores on the Dynamic
Indicators of Basic Early Literacy Skills’ (DIBELS) measures and the total reading raw
score or stanine ranking on the Stanford Achievement Test, Tenth Version (SAT-10).
6. The operational definitions of high-risk, mid-risk, and low-risk are arbitrarily defined by
There are several key limitations to this study. Limitations are defined here as
threats to internal validity. First, during the duration of the study, the researcher was in a
5
direct leadership role in the school district in question, and was deeply involved in
implementing and measuring the effects of a wide array of programs connected to this
study. The chance of researcher bias affecting the outcome of this study is high,
Second, the random assignment process was inherently flawed. There were two
registration periods during which participants enrolled in kindergarten. Families who were
more informed at navigating the school system may have been the first to register in the
first registration period. The more informed parents usually come from higher SES groups.
W
As a result, membership in full-day kindergarten could have been populated at 50% of
capacity by this first group of higher SES registrants. The second registration period
IE
probably brought in more diverse students, which is why the school district populated the
remaining half of the full-day kindergarten classes from students in the second registration
EV
period. It is widely believed that the most high-risk students usually register after the school
year has begun. All of these extra registrants, who registered after the second registration
PR
period ended, were automatically placed in half-day kindergarten. This could possibly have
kindergartens, which may skew the comparison of achievement between full-day and half-
day kindergarten groups. The data analysis has accounted for this limitation by examining
initial group differences by risk level, and by controlling pretest scores, but there was no
Third, there are potential history and mortality issues with the research sample.
Although there is an intact sample, the school district redrew its’ boundaries between the
2004-2005 school year and the 2005-2006 school year. Practically, this means that there
6
may have been several students who were in a non-Reading First school during
greatly affects the control over curriculum that was secured in kindergarten by the Reading
First grant assurances. Children entering Reading First schools from non-Reading First
controlled for this factor by only including students who entered into the five Reading First
schools in fall 2004. Mortality will only be partially controlled by looking at the intact
sample, which will have scores across all years of the study. There was no way to
W
determine, however, the effect of an unequal number of half-day and full-day students
when they were first pre-tested, and were much older, developmentally, in second grade.
EV
There may have been many interventions that occurred between the beginning of
kindergarten and the end of second grade that may have matured students’ reading abilities,
PR
and also muddied the clear effect of time schedule on student reading achievement.
There are also several delimitations that have occurred in this study. Delimitations
First, researcher bias is a risk because there was only one researcher collecting the
Second, the achievement scores in this research study were taken from schools that
were under the Oregon Reading First Grant Assurances. The Reading First Grant is a
federal grant program under the No Child Left Behind Act that provides money to schools
comprehensive and specific. Although the grant assurances provided a set of controlled
such as paid grade level team meetings once a month for the entire K-3 teaching staff - this
same high level of control has led to limited generalization. Results may only be
generalizable to the school district in question, or to similar school districts under similar
grant assurances.
Third, the researcher compiled DIBELS data that had been collected throughout the
W
school district in the fall, winter, and spring across a three-year time period. There could
have been an interaction effect of pre-testing and retesting students three to twelve times
IE
each year using these measures. Because students were tested so many times with DIBELS,
they may have scored higher each time simply due to the fact that they learned how to take
EV
DIBELS better. The results based on DIBELS may not be able to be generalized to a
population that has not been pre-tested with DIBELS or experienced being repeatedly
PR
tested with DIBELS measures over a long period. The same threat exists for SAT-10
testing for the same reasons, although SAT-10 testing took place only once per year, so the
Fourth, the researcher, some teachers, all principals, and all reading coaches knew
that there would be a study comparing the achievement of the experimental and control
groups. The reactive effect of the experimental arrangement on these parties in how they
approached full-day kindergarten implementation may have affected whether the results
can be generalized to a population that is perhaps less favorable or less opposed to the
selection bias beyond those mentioned in the limitations section that would deem the
W
IE
EV
PR
9
CHAPTER 2
types of kindergarten scheduling programs that are available to students. There are three
types of schedules that have been studied in the existing research literature. They are half-
W
day, full-day and alternate-day kindergarten schedules. In a half-day schedule, students
attend kindergarten for half of a day five times per week. An alternate-day schedule is
IE
when students attend three full days one week and two full days the next week with the
remaining days off, or attend two full days each week with a half-day schedule on
EV
Wednesday, for a total of five full days in school over two weeks. A full-day schedule
This review of literature is organized into six subsections. First, research that
compares the effect of half-day and alternate-day schedules on the reading achievement of
kindergarten students is reported. Second, early research (1970-1989) comparing the effect
completed in the early to mid-1990s that compares the effect of half-day and full-day
programs. Next, the summative and meta-analytical studies through 1997 that make
published since 1997 is examined. After that, the research on the observed effects of half-
day and full-day schedules on the reading achievement scores of high-risk children is
The results of the research that compare the effects of half-day and alternate-day
schedules on reading achievement are not clear when examined superficially. Two major
studies, for example, used the same instrument to measure achievement and reported
W
different results. On one hand, Cleminshaw and Guidubaldi (1979) found that there are
significant differences in the achievement scores of the two groups, as measured by the
IE
Metropolitan Readiness Test. The alternate-day group’s mean (74.88) was significantly
higher than the half-day group’s mean (66.63). These differences favor a full-day alternate
EV
approach. On the other hand, Clements and Gullo (1984) found that there was no
significant difference between reading achievement scores as measured by the same test.
PR
The alternate day group’s mean (72.00) was not significantly different from the half-day
group’s mean (71.93). Two completely different results from the same instrument merit a
After a thorough reading, one realizes that there are stark differences in the research
designs employed in these studies. Clements and Gullo (1984) used a causal-comparative
design without random selection of participants, while Cleminshaw and Guidubaldi (1979)
used a randomly selected sample. Also, Clements and Gullo (1984) used a pretest-posttest
format to control for initial differences in achievement while Cleminshaw and Guidubaldi
(1979) only administered a posttest. In addition, Clements and Gullo (1984) made little
11
attempt to control for other variables, while Cleminshaw and Guidubali (1979) went to
great lengths to control extraneous factors. Finally, Cleminshaw and Guidubaldi (1979)
took their participants from four similar schools with different teachers for each of four
treatment groups. There were only fifteen participants in each treatment group. Clements
and Gullo (1984) selected participants from one school and both treatment groups were
taught by the same teachers, but in different years. There were over 90 participants studied
each year. The fact that these studies are so different in their design may explain the
W
Several years later, Hildebrand (1997) set out to draw some stronger conclusions
about the effect of half-day and alternate-day schedules. The research design that
IE
Hildebrand employed was superior to both of the studies mentioned above for the
experimental conditions. Second, her sample size was larger (n=147) and participants were
equally and randomly assigned to the three scheduling conditions: half-day, alternate-day,
PR
and full-day kindergarten. Third, all three middle-class to lower-class groups of students
teachers, under Chapter 1 funding and assistance. Hildebrand used a pretest-posttest design
and conducted an analysis of covariance (ANCOVA) on the data to control for the initial
group differences, which were present. Hildebrand used an instrument called the Test of
Early Reading. It was found that there was no significant difference between the reading
scores of students in the alternate-day and half-day schedules, but that the full-day students
exhibited superior achievement. Only a study that was fully experimental could refute the
the effectiveness of alternate-day kindergarten. Thus, this literature review now shifts to an
area of study with a larger research base that is more relevant to the purpose of this study,
academic achievement emerged in the literature in the early 1970s and continued through
1988. Because these studies are over 20 years old or older, their findings are reported, but
W
the discussion of their specific methodological strengths or flaws is limited. It is widely
agreed that the research pre-1988 was rife with methodological flaws (Puleo, 1988; Elicker,
IE
2000; Clark, 2001). Some of these flaws come from small sample size, lack of random
assignment, or lack of pre-testing to control for initial group differences (Baskett, Bryant,
EV
children each were randomly assigned to full-day and half-day kindergarten scheduling
students. Reading levels were measured at the end of first grade and compared. Johnson
(1974) concluded that no significant differences were found between the two scheduling
conditions as measured by the Walker Readiness Test and the Stanford Early School
Achievement Test. Nieman and Gastright (1975), Alper and Wright (1979), and Oelerich
(1979), on the other hand, all reported in their studies [n=190, n=198, n=25 respectively],
13
that there was a statistically significant difference in favor of full-day kindergarten as
measured by the Metropolitan Readiness Test. Hatcher, Schmidt, and Cook (1979) reported
no significant difference between the full-day and half-day condition. In short, the research
from the 1970s offers no clear conclusion about the comparative effectiveness of either
schedule. Unfortunately, research from the early 1980s only confounds the researcher in
On one hand, there were many studies in the 1980s that showed significant
W
Ziomek and Harris (1980) found that there was a statistically significant difference in favor
of full-day kindergarten based on the scores on the Metropolitan Readiness Test. Adcock
IE
(1980) indicated in a report of a study on 189 kindergarten students, that full-day
half-day kindergarten students on the Metropolitan Achievement Test. In the same year,
Humphrey (1980) revealed, in a medium-sized school district study (n=179), that full-day
PR
kindergarten students performed significantly higher than the half-day students on the
California Achievement Test, Boehm Test of Basic Concepts, and the Gates-MacGintie
reported that full-day kindergarten students tended to have higher academic report card
marks, a lower rate of grade retention, and higher standardized achievement test scores.
Anderson (1985) also found that the “most mature” full day students perform significantly
better than the “most mature” half-day students on the Stanford Early School Achievement
14
Test (n=60). In a ten-year longitudinal study on a small sample (n=44), Oelerich (1984)
concluded that full-day, every-day kindergarten was the superior condition, given the
results of the Metropolitan Achievement Test. Two large-scale school district studies were
completed by the Madison Metropolitan School District (1985) using voluntary assignment
(n=383), and the New York City Board of Education (1985) using random assignment
On the other hand, other researchers from the early 1980s claim that there are no
W
significant differences between the achievement of students attending half-day and full-day
kindergarten. McClinton and Topping (1981) used unknown sampling techniques, and
IE
Herman (1984) used random assignment (n=66). However, both found that there were no
on the California Achievement Test. Robertson (1984) made a comparison of full-day and
half-day student achievement that revealed that neither schedule significantly improved the
PR
achievement of students over the other group. Savitz and Drucker (1984) found that full-
day kindergarten did not lead to higher Test for Analysis and Placement scores (n=19).
District, for instance, Uguroglu and Nieminen (1986) concluded that there were small
differences in the academic readiness for first grade that the half-day and full-day programs
produced. Similarly, the Pasco School District (1987) found that the full-day kindergarten
program was far more effective than half-time models for students who live in poverty, or
who are of average or above average achievement levels. In a follow-up to the 1983 study
15
cited above, Evansville-Vanderburg School District (1988) concluded that full-day students
1988) in Ohio County School District showed that the additional instructional time in full-
With the exception of Johnson (1974), the studies mentioned above are older
studies that are not cited often in the current research literature. There are two studies,
however, from the 1980s that are consistently cited in the current literature. The results of
these two studies are reported below with considerably more analysis than those reported
W
above.
differences between full-day and half-day students across three cohorts of kindergarten
EV
students who were in first, second, and third grade at the time of the study. The sample
background. They were not randomly selected, but all available students were part of the
sample in this ex post facto design (Airasian & Gay, 2003). The researchers used the
this study. This measure was administered at the end of first, second, and third grade to
students who had been in both kindergarten schedules. There was no significant difference
in the achievement of children from both time schedules. Even though the sample size of
this study is adequate, the conclusion of this causal-comparative study is weak because the
study did not employ random assignment and exercised little control over the independent
schedules and reading achievement, Sergesketter and Gilman (1988) reported no difference
researchers used the Gates-McGintie Reading Test to measure the differences in reading
achievement between four full-day kindergarten groups and four half-day kindergarten
groups that were matched by socioeconomic level. These researchers used random
sampling to select and assign 187 half-day kindergarten students and 223 full-day
W
the full-day kindergarten scores were higher on both the comprehension and vocabulary
sections of the selected test, data analysis by the authors revealed the results not statistically
IE
significant. Because there was no pretest, the authors could not have ruled out initial
Puleo (1988) evaluated most of the studies mentioned above in his review and
critique of full-day kindergarten research up to 1988. Puleo came to the conclusion that
PR
most of the studies lacked the methodological rigor needed to draw more clear conclusions
and to conceive stronger implications. At that point in the history of the kindergarten time
the effectiveness of full-day and half-day kindergarten on student reading achievement, but
with similarly mixed results. More time is spent reviewing these research studies since they
are newer, and have more bearing on this research study. In addition, the researchers often