Beyond the Reading Wars: Exploring the Effect of Child-Instruction Interactions on Growth in Early Reading

Abstract

This study examined the influence of interactions between first graders' fall language-literacy skills (vocabulary and decoding) and classroom instructional practices on their spring decoding scores. Instructional activities were coded as teacher managed or child managed and as explicit or implicit, as well as for change in amount of time spent in the activity over the school year. Findings revealed that specific patterns of instructional activities differentially predicted children's decoding skill growth. Children with low initial decoding scores achieved greater decoding growth in classrooms with more time spent in teacher-managed explicit decoding (TME) instruction. In contrast, for children with initially high decoding scores, amount of TME had no effect. Children with low initial vocabulary scores achieved greater decoding score growth in classrooms with less child-managed implicit (CMI) instruction but with increasing amounts of CMI instruction as the school year progressed. However, children with high initial vocabulary scores achieved greater decoding growth in classrooms with more time spent in CMI activities and in consistent amounts throughout the school year. Children's initial decoding and vocabulary scores also directly and positively affected their decoding score growth. These main effects and interactions were independent and additive, thus children's first-grade decoding skill growth was affected by initial vocabulary and decoding skill as well as type of instruction received-but the effect of type of instruction (TME or CMI amount and change) depended on children's initial vocabulary and decoding scores. Implications for research and educational practices are discussed.

Full text

Beyond the Reading Wars: Exploring the
Effect of Child–Instruction Interactions
on Growth in Early Reading
Carol McDonald Connor, Frederick J. Morrison,
and Leslie E. Katch
University of Michigan
This study examined the influence of interactions between first graders’ fall lan-
guage–literacy skills (vocabulary and decoding) and classroom instructional prac-
tices on their spring decoding scores. Instructional activities were coded as teacher
managed or child managed and as explicit or implicit, as well as for change in amount
of time spent in the activity over the school year. Findings revealed that specific pat-
terns of instructional activities differentially predicted children’s decoding skill
growth. Children with low initial decoding scores achieved greater decoding growth
in classrooms with more time spent in teacher-managed explicit decoding (TME) in-
struction. In contrast, for children with initially high decoding scores, amount of
TME had no effect. Children with low initial vocabulary scores achieved greater de
-
coding score growth in classrooms with less child-managed implicit (CMI) instruc
-
tion but with increasing amounts of CMI instruction as the school year progressed.
However, children with high initial vocabulary scores achieved greater decoding
growth in classrooms with more time spent in CMI activities and in consistent
amounts throughout the school year. Children’s initial decoding and vocabulary
scores also directly and positively affected their decoding score growth. These main
effects and interactions were independent and additive, thus children’s first-grade de
-
coding skill growth was affected by initial vocabulary and decoding skill as well as
type of instruction received—but the effect of type of instruction (TME or CMI
amount and change) depended on children’s initial vocabulary and decoding scores.
Implications for research and educational practices are discussed.
SCIENTIFIC STUDIES OF READING, 8(4), 305–336
Copyright © 2004, Lawrence Erlbaum Associates, Inc.
Requests for reprints should be sent to Carol McDonald Connor, Florida State University and the
Florida Center for Reading Research, City Centre Building, 227 North Bronough Street, Suite 7250,
Tallahassee, FL 32301. E-mail: cconnor@fcrr.org
Do Not Copy

There has been long-standing controversy regarding the best way to teach children
how to read (Ravitch, 2001). The debate has been fueled, in part, because each year
significant numbers of American children fail to reach functional levels of literacy
(National Assessment of Educational Progress, 2000). In essence, the debate has
centered on the efficacy of phonics or code-based instruction versus whole lan
-
guage or meaning-based instruction (Rayner, Foorman, Perfetti, Pesetsky, &
Seidenberg, 2001). Code-based instruction focuses on explicit and systematic
training in decoding including letter recognition, letter–sound correspondence,
phonics, and phonological awareness. Meaning-based instruction views learning
to read as a more natural process (Goodman, 1970) that requires consistent experi
-
ence with meaningful text within a literature-rich environment (Dahl & Freppon,
1995). Unfortunately, as Rayner et al. noted, “the continued dichotomy of reading
philosophies produces fragmented instruction in classrooms rather than the inte
-
grated balance of skills and meaningful applications that research suggests are
needed to produce successful readers” (p. 61). Evidence accumulating systemati
-
cally over the past 20 years has documented that a combination of methods may
better support children’s developing literacy. Most children appear to develop
stronger reading skills when provided explicit decoding instruction in combination
with meaningful reading activities (Guthrie, Schafer, & Huang, 2001; Rayner et
al., 2001; Taylor, Pearson, Clark, & Walpole, 2000). Consequently, there is a grow-
ing trend toward “balanced” instruction in early reading instruction (see also P.
Cunningham & Hall, 1998; Hiebert & Raphael, 1998; Pressley, 1998).
Yet, the promotionofbalanced instruction leavesopen the question of what might
be the best combination of basic skills instruction and meaningful reading activities.
An implicit and largelyuntested assumption in muchliteracy research is thatspecific
instructional practices will be equally effective for all children. This universalistic
view can be found in the literature supporting meaning-based instruction (e.g., Dahl
& Freppon, 1995) as well as that promoting code-based instruction (National Read
-
ing Panel,2000). However, Child Aptitude × Treatmentinteraction research, firstin
-
troduced in the 1970s, revealed that these interactions may be important (Sternberg,
1996). Some recent research has begun to explore the possibility that the efficacy of
instructional practices may vary with the skill level of the student. For example,
Foorman, Francis, Fletcher, Schatschneider, and Mehta (1998) found that children
with weaker phonological awareness at the beginning of the school year demon
-
strated greater growth in decoding skills in the code-based classrooms than did chil
-
dren with stronger phonological awareness. Juel and Minden-Cupp (2000) found an
analogous Reading Group × Classroom Type interaction. In their study, children
who started first grade with weaker reading skills (i.e., low-reading group) made
more progress in classrooms where there was greater emphasis on word recognition
instruction. In contrast, children with stronger reading skills at the beginning of first
grade (i.e., middle or high group) achieved greater reading progress in the classroom
where the teacher emphasized a literature-rich environment with less emphasis on
306
CONNOR, MORRISON, KATCH
Do Not Copy

code-based instruction. A major goal of our study was to further explore the impact
of various instructional practices for children entering first grade with different lan
-
guage and reading skill levels.
CHILD CHARACTERISTICS
Children begin school with widely varying abilities in skills that support early lit
-
eracy development, such as phonological awareness, word decoding, and vocabu
-
lary, and these differences emerge surprisingly early (Stevenson, Parker,
Wilkinson, Hegion, & Fish, 1976). For example Stipek and Ryan (1997) uncov
-
ered large social class differences in both cognitive and early literacy skills among
a sample of preschoolers and kindergartners. A recent observational study docu
-
mented meaningful social class differences in children’s acquisition of expressive
vocabulary as early as 18 months of age (Hart & Risley, 1995).
Further, these individual differences appear to be largely sustained throughout
children’s school careers (Entwisle & Alexander, 1988). Scores on decoding, al-
phabet recognition, and vocabulary tasks at kindergarten entry consistently pre-
dicted academic performance throughout the first 3 years of formal schooling ex-
perience (Dickinson & Tabors, 2001; Hart & Risley, 1995; Morrison & Cooney,
2002; Stevenson et al., 1976). Additional research has revealed stability for
1st-grade reading skill through 11th-grade reading experience and performance,
even after accounting for children’s cognitive abilities (A. Cunningham &
Stanovich, 1997; A. Cunningham, Stanovich, & West, 1994). Altogether, a grow-
ing body of evidence testifies to the very early emergence and stability of variabil
-
ity among children in important literacy and literacy-related skills.
This study attempted to identify and incorporate two important child character
-
istics—decoding and vocabulary—as they interacted with instructional practice.
Decoding skill, specifically alphabet recognition, letter-sound correspondence,
and single word decoding, was selected because it is a foundational skill critical to
the development of proficient reading (Snow, Burns, & Griffin, 1998) and its in
-
struction is both salient and important in first-grade classrooms (Adams, 1990;
Neuman & Dickinson, 2001; Rayner et al., 2001). Children’s vocabulary, an inte
-
gral aspect of language development (Locke, 1997, 1993), varies significantly
among children at school entry, and individual differences appear to be relatively
stable throughout childhood (Hart & Risley, 1995) and beyond (Nippold, 1988).
Further, in two separate studies, there were no significant schooling effects on chil
-
dren’s kindergarten or first-grade vocabulary growth (Christian, Morrison, Frazier,
& Masseti, 2000; Morrison, Smith, & Dow-Ehrensberger, 1995), yet it is an impor
-
tant predictor of later reading success (Anderson & Freebody, 1981; Catts, Fey,
Zhang, & Tomblin, 2001).
BEYOND THE READING WARS 307
Do Not Copy

INSTRUCTIONAL FACTORS
There is also appreciable variability in amount and type of language arts instruction
provided to children across classrooms (Juel & Minden-Cupp, 2000; Taylor et al.,
2000; Wharton-McDonald, Pressley, & Hampston, 1998). For instance, Juel and
Minden-Cupp (2000) described four classrooms that varied substantially in amounts
ofmeaning andcode-based instructionprovidedoverthecourse ofthe schoolyear.
Teachers have been observed using elements of both code-based and mean
-
ing-based instruction while varying the amount of each type during the school year
(Juel & Minden-Cupp, 1998; Rayner et al., 2001; Taylor et al., 2000). Dichoto
-
mous or categorical comparisons of classroom instructional practices may be inad
-
equate to describe fully teachers’ literacy practices. Grouping teachers by their pri
-
mary focus (e.g., code- vs. meaning-based) may oversimplify what is actually
happening in classrooms and undermine our ability to examine the complex effects
of both the amount and types of reading instruction teachers provide. Although in
-
struction has been defined using a number of methods, we wanted to describe in-
struction in a way that could be used with a range of child outcomes including de-
coding, comprehension, fluency, and so forth across a variety of classroom grades
and settings. Further, we wanted to capture the complexity of instruction while
representing it in measurable variables. To this end, and relying on recent studies
of children’s reading skill development, we selected three specific dimensions of
instruction: (a) explicit versus implicit instruction, (b) teacher-managed versus
child-managed instruction, and (c) change in type and amount of instruction over
time. Each of these dimensions is discussed more fully in the rest of this article. A
fourth dimension, word level versus higher order, was not included in this study
because it overlapped with the explicit versus implicit dimension. This would not
have been the case if our outcome had been, for example, reading comprehension,
which might be differentially impacted by implicit word level (e.g., alphabet activ
-
ities vs. higher order activities—sustained silent reading).
Explicit Versus Implicit Instruction
The first dimension focuses on whether instruction is explicit or implicit in promot
-
ing growth of a particular skill, such as word decoding. For example, Foorman and
her colleagues (see Foorman et al., 1998) grouped classrooms along the explicit–im
-
plicit continuum focusing on decoding skill instruction (i.e., Direct code, Embedded
code, Implicit code). In our coding scheme, if, as in this study, word decoding is the
targeted skill, then instructional activities such as blending onsets and rhymes or
teaching letter-sound correspondence would be considered explicit because the
children’s attention is primarily directed to components of worddecoding strategies.
In contrast, activities like teacher-led discussions, in which the child’s attention is
more explicitly focused on comprehension (i.e., extracting meaning from text),
308
CONNOR, MORRISON, KATCH
Do Not Copy

could still influence word-decoding skills in an implicit or incidental fashion. At an
intermediate point would be implicit phonics (e.g., Torgesen et al., 2001) where
teachers expose children to word lists that contain similar spelling–sound corre
-
spondences but do not explicitly teach spelling–sound correspondences. Note that if
reading comprehension had been the outcome of interest, different sets of instruc
-
tional activities would have been defined as explicit or implicit.
As defined, explicit decoding instruction encompasses much of what has been
described as code-based instruction including teaching of phoneme–grapheme
correspondences, phonological awareness (e.g., onset-rime segmentation, blend
-
ing phoneme, segmenting phonemes), and letter names and sounds. Implicit de
-
coding instruction includes meaning-based activities like teachers reading to stu
-
dents, discussions about books, teachers and students reading together, and
students reading and writing independently.
Teacher-Managed Versus Child-Managed Instruction
The second dimension refers to the degree that instructional activity and the child’s
attention are primarily under the direction of the teacher (e.g., when the teacher is
explicitly instructing the children in word decoding strategies) or primarily con-
trolled by the child (e.g., in sustained silent reading). This dimension is quite simi-
lar to methods of instruction such as prescriptive (teacher-managed) and
responsive (teacher-managed moving to child-managed) (see Rayner et al., 2001)
as well as “child-centered” and “teacher-directed” (Bredekemp & Copple, 1997).
However, there are some important differences because the activity is coded ac-
cording to whether the teacher or the child is responsible for directing attention to
or “managing” the learning. Thus, activities that might be considered child-cen
-
tered activities, such as discussions about books, would be considered
teacher-managed because the teacher is managing the learning. However,
teacher-directed activities, such as completing worksheets, would be considered
child-managed because the child is responsible for his or her own learning.
Change in Amount of Instructional Activities
Over the School Year
One provocative finding in the Juel and Minden-Cupp (2000) study was that some
teachers changed their instructional emphasis over the course of the school year.
For example, one teacher began the year with a strong focus on explicit,
teacher-managed decoding instruction that tapered off as the year progressed and
as children mastered basic skills. In this class, children with weaker fall reading
skills (i.e., children in the low reading group) achieved stronger spring decoding
scores than did children in the low reading group in other classrooms. Thus, how
much time the class spends in particular activities at certain times of the school
BEYOND THE READING WARS 309
Do Not Copy

year may be important to consider. In this study, we observed first-grade class
-
rooms three times during the school year (i.e., fall, winter, and spring). Hence, we
were able to examine the impact of changes in amount of instructional activities
over the course of the school year.
HYPOTHESIS
This study examined first-grade students’ decoding skill growth and how it was af
-
fected by classroom instructional practices, the skills with which the children be
-
gan school, and the interactions between instruction and child characteristics. We
predicted that children with stronger fall vocabulary or decoding skills would
achieve stronger decoding skill growth in classrooms that provided more
child-managed implicit decoding instruction and less teacher- managed explicit
decoding instruction. In contrast, we expected children with weaker fall vocabu
-
lary or decoding scores to achieve stronger decoding skill growth in classrooms
that provided more teacher-managed explicit decoding instruction and less
child-managed implicit decoding instruction. Finally, we anticipated that changes
in amount of different types of decoding instruction over the course of the school
year would affect children’s decoding growth.
METHODS
Participants
One hundred eight first-grade children taught by 42 teachers participated in this
study as part of a larger longitudinal study of schooling conducted in a large mid
-
western city. Children who were English proficient and had no identified disability
were recruited from schools in the participating school district. Children were re
-
cruitedover3consecutive years, with most entering the study during Years1 and2.
Descriptive information is provided in Table 1. Of the participants in this study,
44% of the children were girls, 65% were White, 28% were African American, 3%
were Hispanic, 2% were Asian, and 2% belonged to other ethnic groups or were
not identified. Children’s race–ethnicity did not contribute significantly to our out
-
come in preliminary hierarchical linear models, t(41) = .27, p = .32, and so was not
included in our final model. Children’s cognitive abilities were assessed at the be
-
ginning of the study using the Stanford-Binet Intelligence Scale–4th Edition
(Thorndike, Hagen, & Sattler, 1986), which provides a full-scale intelligence quo
-
tient (IQ). On average, this sample demonstrated IQs within normal limits. Further,
IQ did not significantly contribute to our outcome in preliminary hierarchical lin
-
ear models, t(90) = –.44, p = .66, and so was not included in the final model.
310
CONNOR, MORRISON, KATCH
Do Not Copy

Measures
Individual child assessment.
Children were assessed using a battery of tests
and tasks in the fall and spring of their first-grade year. These tests included the Pea-
body Picture Vocabulary Test–Revised (Dunn & Dunn, 1987), which assesses re-
ceptive vocabulary (Vocabulary), and the Reading Recognition subtest of the Pea-
body Individual Achievement Tests–Revised (Markwardt, 1989), which assesses
letter identification, letter-soundcorrespondence, and single wordrecognition skills
(Decoding). Raw scores were used for statistical analyses. When reported as grade
equivalents, raw or fitted scores were converted to grade equivalent scores using the
tables provided in the examiners’ manual.
Parent questionnaire: Mother’s educational level and home literacy.
Par
-
ents completed questionnaires during their 1st year in the study, which provided de
-
scriptive information about the family, including mothers’ educational levels, ex
-
pressed in years. Home literacy environment scores (Home Literacy) were also
based on the results of parents’responses on a questionnaire. Home Literacy, a com
-
posite measure that ranged from 3 (low)to18(high), was derived from parents’ re
-
sponses regardinghow frequently theyused a library card, number of adult and child
magazine subscriptions, number of newspaper subscriptions, how often the family
read together, number of children’s books, hours of television the child watched per
week, and how frequently parents read to themselves (Griffin & Morrison, 1997).
Reliability of the measure was adequate (Cronbach’s α = .72). These parent and
home variables were included in the models because of their well-documented asso
-
ciation with children’s academic and reading success (Snow et al., 1998).
School district, teachers, and classroom observation—Instructional
variables.
As noted previously, all children and teachers were recruited from the
same school district in a large midwestern city. The school district in this study re
-
BEYOND THE READING WARS 311
TABLE 1
Descriptive Statistics for Student Participants
White/Asian African American Total
M SD M SD M SD
Decoding fall 26.0 13.5 18.8 11.42 23.27 13.14
Decoding spring 43.7 15.0 33.0 13.3 39.63 15.22
Vocabulary fall 90.6 14.0 73.0 12.2 83.85 15.79
Vocabulary spring 99.1 12.0 80.2 13.8 91.91 15.68
Home literacy score 14.5 2.1 10.1 3.1 12.93 3.29
Mother’s educational level (years) 17.4 2.2 13.2 2.7 15.98 3.10
IQ (full-scale standard score) 107.5 13.5 90.9 11.0 101.21 14.95
Note. All reported values are raw scores unless otherwise indicated.
Do Not Copy

ported that they supported a whole-language approach to literacy instruction and
encouraged teachers to provide significant amounts of student-initiated reading
and writing activities. Teachers were recruited for the study if a target child at
-
tended their classroom. Classroom instruction was observed over the course of the
school year during three all-day sessions in the fall, winter, and spring (Connor,
Morrison, & Griffin, 2002). Trained observers recorded a narrative description of
the school day and the amount of time spent on specific instructional activities. Re
-
searchers then coded the activities described in the narratives for type of instruc
-
tion and length of time the type of activity lasted (in minutes). Descriptions of the
activities are provided in the Appendix. For approximately 10% of the observa
-
tions, two observers observed the same classroom and independently recorded a
narrative of the classroom activities. Agreement was calculated by time and de
-
scription of activity with the number of minutes the observers were in agreement
divided by minutes in agreement plus minutes where the observers disagreed.
Interobserver agreement for the narrative was 95%. Interrater reliability for the
coded activities was calculated in much the same way. The number of minutes the
two independent raters coded a description using the same activity code divided by
agreements plus disagreements. Interrater reliability was 86%. For this study, only
language arts activities were included.
Dimensions of instruction.
Instructional activities were coded using the first
two dimensions of instruction described previously: explicit versus implicit and
teacher-managed versus child-managed (see Table 2). The third dimension, change
in amount of instruction over time, is described more fully in the next section.
312
CONNOR, MORRISON, KATCH
TABLE 2
Instructional Activities Comprising Each Dimension of Instruction: Teacher-
Versus Child-Managed and Explicit Versus Implicit
Teacher Managed Child Managed
Explicit Alphabet activity Spelling
Letter sight–sound
Initial consonant stripping
Word segmentation
Implicit Vocabulary Student read aloud, individual
Teacher read aloud Sustained silent reading
Student read aloud, choral Reading comprehension activity
Teacher-managed group writing Student independent writing
Writing instruction Student group writing
Discussion
Conventions of print
Listening comprehension
Note. Descriptions of activities are provided in the Appendix.
Do Not Copy

Instructional activities were coded as explicit or implicit as they pertained to word
decoding. Hence instructional activities that taught letters, letter-sound association,
phonological awareness, spelling, or decoding words were defined as explicit decod
-
ing instructional activities. Other language arts activities were defined as implicitly
teaching decoding. For example, the opportunity to read independently while explic
-
itly focusing on fluency implicitly supported children’s decoding skills.
Teacher-managed instruction included those activities in which the teacher was
the primary director of the children’s attention—for example, teacher and student
discussions surrounding a particular book or teacher scaffolding–coaching. Other
teacher-managed activities included teachers reading aloud and direct instruction
in letter-sound relations. Child-managed instruction included those instructional
activities where the student was primarily controlling his or her focus of atten
-
tion—for example, reading independently (e.g., sustained silent reading) and com
-
pleting worksheets independently.
Computing classroom variables and change over time.
Because we
observed every classroom three times over the course of the school year (there
were no missing classroom observations), variables representing types of instruc-
tion were expressed as both the number of minutes for amount of instruction
(amount centered at winter observation) as well as change in the amount of instruc-
tion over the school year (i.e., slope, which is the third dimension of instruction).
Variables were computed using the following method. For each observation, the
amount of time, in minutes, spent in each type of instructional activity was com-
puted. Using Hierarchical Linear Modeling (HLM; Raudenbush & Bryk, 2002),
instructional growth curves for each classroom were modeled at Level 1 using
month of the observation centered at the winter evaluation, with the individual
classroom modeled at Level 2. Empirical Bayes residuals were calculated for each
classroom using the HLM software (Version 5.0). This provided an instruction
amount score and an instruction slope score for each classroom teacher based on
the observations in his or her classroom.
Each classroom amount score represents the amount of time in minutes that a
particular teacher spent in one type of activity, which was above or below the fitted
mean for all teachers, centered at the winter observation. As an illustration, sup
-
pose Teacher A’s instruction amount score in teacher-managed explicit (TME) in
-
struction was 3.4. This means that, on average, she spent 3.4 min more than the av
-
erage number of minutes teachers in the study spent in this type of instruction,
which was 7.4 min. In other words, she spent a total of 10.8 min (7.4 + 3.4) per day
in TME instruction. In contrast, hypothetical Teacher B’s instruction amount score
was –2.5. This means that she spent 2.5 min less than the average time spent on this
type of instruction compared to the mean of all teachers, or 5.1 min (7.4–2.5) per
day. Instruction slope score was calculated in the same way, but using the slope
rather than the intercept. Thus, there was a fitted mean slope for all of the class
-
BEYOND THE READING WARS 313
Do Not Copy

rooms (fitted slope TME instruction = –.58) and an individual teacher slope score,
which represented his or her difference from the mean slope. This procedure was
followed for each dimension of instruction variable (see Table 2). These variables
were then used in the final model at Level 2, as described next. To recap, these
amount and slope variables included TME, child-managed explicit (CME),
teacher-managed implicit (TMI) and child-managed implicit (CMI).
Analytic Strategy
HLM (Raudenbush & Bryk, 2002) was used to control for the nested nature of the
data: Children nested in classrooms. On average, there were more than two chil
-
dren per classroom. However three classrooms had as many as six children,
whereas a few had only one. Had we used regression instead of HLM, we might
have overestimated the effect of instruction on children’s outcomes because we
would not have accounted for the shared classroom variance (Raudenbush & Bryk,
2002). Child-level variables were entered at Level 1; classroom variables were en-
tered at Level 2 (see Equation 1).
Level 1
Y
ij
= β
0j
+ β
1j
(Mother’s Education)
ij
+ β
2j
(Home Literacy)
ij
+
β
3j
(fall Vocabulary)
ij
+ β
4j
(fall Decoding)
ij
+ r
ij
Level 2
β
0j
= γ
00
+ γ
10
(Classroom instruction variables)
j
+ u
0j
β
1j
= γ
10
+ u
1j
β
2j
= γ
20
β
3j
= γ
30
+ γ
31
(Classroom instruction variables)
j
β
4j
= γ
40
+ γ
41
(Classroom instruction variables)
j
Y
ij
, which is the Decoding score for Child i in Class j, is a function of the respec
-
tive coefficients (β) at Level 1 as they pertain to Mothers’ Educational Levels,
Home Literacy, fall Vocabulary, and fall Decoding, as well as a residual (r
ij
). β
0j
is a
function of the fitted mean for the group of students (γ
00
) plus the effect of the
classroom instruction variables for Classroom j, plus error (u). γ
10
represents the
effect of mother’s education on spring Decoding. γ
20
represents the effect of Home
Literacy environment on spring Decoding score. γ
30
represents the effect of fall Vo
-
cabulary, and γ
40
represents the effect of fall Decoding on spring Decoding. γ
31
rep
-
resents the interaction between instruction variables and fall Decoding. γ
41
is the
interaction between fall Vocabulary and instruction variables. The error at the level
314
CONNOR, MORRISON, KATCH
Do Not Copy

of the classroom is represented by u
j
. For all models presented, residuals were as
-
sumed to be normally distributed with means of zero.
RESULTS
Variation in Classroom Instruction
Results revealedsignificantdifferencesin amount and change (i.e., slope) of instruc
-
tion type provided in classrooms over the course of the school year. As depicted in
Table 3 and Figure 1, overall, teachers provided significant amounts of each type of
instruction—TME, CME, TMI, and CMI. They also significantly changed the
amount of TME, CME, and CMI (but not TMI) instruction they provided over the
course of the school year. In general, teachers provided more TME and CME at the
beginning of the school year and significantly less as the year progressed (see Figure
1), as indicated by significant and negative slope coefficients (see Table 3). In con-
trast, teachers increased the overall amount of CMI instruction over the course of the
school year, as demonstrated by the significant positive slope coefficient.
Teachers differed substantially in the total amount of language arts instruction
they provided (i.e., sum of TME, CME, TMI, and CMI) ranging from 5 to 134 min of
language arts instruction per day on the actual days observed (M = 63.5, SD = 26.5).
Table 3 provides means (centered at the winter observation) and standard deviations
for each type of instruction calculated using HLM. The length of the school day did
not differ across classrooms because all of the teachers taught within the same school
district.
BEYOND THE READING WARS 315
TABLE 3
Descriptive Information for Instructional Variables
MSDSE
Level 2
Variance (u)
Level 1
Variance (r)
Total Language Arts amount 63.50 10.27 NA NA NA
TME amount 7.43*** 3.53 1.04 24.39*** 69.77
TME slope –.82* .58 .32 .83
CME amount 4.27*** 2.52 .62 9.61*** 23.49
CME slope –.58** .62 .19 .58
TMI amount 23.26*** 3.52 1.52 32.29* 211.63
TMI slope –.72 1.56 .62 5.40*
CMI amount 28.54*** 9.45 2.11 130.12*** 200.20
CMI slope 2.26** 1.99 .67 8.61**
Note. Fitted means (in minutes) and slopes (in minutes per month) computed using Hierarchical
Linear Modeling. NA = data not applicable; TME = teacher-managed explicit; CME = child-managed
explicit; TMI = teacher-managed implicit; CMI = child-managed implicit.
*p < .05. **p < .01. ***p < .001.
Do Not Copy

Teachers also differed in the emphasis they placed on the type of instruction, as
indicated by the significant variance at Level 2 for these variables (see Table 3).
Some teachers spent relatively more time on TME, others on CMI, others spent lit
-
tle time on either TME or CMI, but there was not a simple trade-off as evidenced
by the low and nonsignificant correlation between the amounts for each type of in
-
struction (see Table 4). On average, teachers provided significantly more CMI than
TME—7.4 min per day of TME compared to 28.5 min per day of CMI (t = 14.26).
Of the change variables, only CMI and TMI slopes varied significantly among
teachers. Some teachers started the school year providing less time in CMI or TMI
and increased the amount as the year progressed, whereas other teachers provided
the same amount all year long. Still others decreased the amount of TMI provided.
TME amount and slope were highly and negatively correlated (t r = –.88) so that
teachers who provided higher amounts of TME instruction in the fall tended to de
-
crease the amount taught more sharply over the school year (i.e., steeper declining
316
CONNOR, MORRISON, KATCH
FIGURE 1 Graph of the fitted growth curves of the amount (in minutes/day) and slope (in
minutes per month) in the four classroom instruction variables, teacher-managed explicit
(TME), child-managed explicit (CME), teacher-managed implicit (TMI), and child-managed
implicit (CMI).
Do Not Copy

slopes), whereas teachers who provided smaller amounts of TME in the fall tended
to provide smaller amounts all year long (i.e., flat slope). In contrast, there was no
appreciable correlation between CMI instruction amount and slope (r = .27); there
was no prevailing systematic pattern of CMI instruction among teachers over the
school year. Some teachers provided higher or lower amounts of CMI all year long,
some provided less as the year progressed, and others provided more.
Classroom Instruction Effects
Overall, classroom instruction had a complex effect on students’ decoding skill
growth. The results of hierarchical linear modeling are presented in Table 5. It
should be noted that causal claims are beyond the scope of this study because we
did not experimentally manipulate the type of instruction children received.
All instructional amount variables were included in the model at Level 2 for the
intercept (β
0
) to control for the total amount of language arts instruction provided.
As noted previously, the length of the school day was the same for all classrooms,
thus our model controlled for the total amount of language arts instruction (vs. other
classroom activities). In other words, the effect of a particular type of instruction is
controlling for the amount of all other types of instruction provided. Because all con
-
tinuous variables were centered at the grand mean, the effect of a particular variable
holds all other variables constant at their grand mean for the sample.
CMI slope was included in the model because it varied significantly by class
-
room. The fall Decoding × TME Amount, Decoding × CMI Amount, and De
-
coding × CMI Slope interactions were entered into the model at Level 2. The fall
BEYOND THE READING WARS 317
TABLE 4
Correlation Between Teacher Variable Amounts and Slopes
TMEa TMEsl TMIa TMIsl CMEa CMEsl CMIa CMIsl
TMEa —
TMEsl –.88** —
TMIa .062 –.068 —
TMIsl –.039 .045 –.99** —
CMEa –.047 .069 –.053 .051 —
CMEsl .047 –.070 .053 –.051 –.99** —
CMIa –.148 .137 –.026 .002 –.131 .131 —
CMIsl –.024 .020 .039 –.049 –.217 .217 .270 —
totLA .235 –.234 .365* –.378* .084 –.084 .824* .448*
Note. Hierarchical Linear Model Tau r provided for each variable’s intercept/slope correlation.
All others are Pearson correlations. a = teacher variable amounts; sl = slopes; TME = teacher-managed
explicit; TMI = teacher-managed implicit; CME = child-managed explicit; CMI = child-managed im
-
plicit; totLA = total amount of Language Arts instruction.
*p < .05. **p < .01.
Do Not Copy

Vocabulary × TME Amount, Vocabulary × CMI Amount, and Vocabulary × CMI
Slope interactions were also entered into the model at Level 2.
Results of Hierarchical Linear Modeling
Type of instruction had a significant but complex effect on children’s achievement,
which was dependent on children’s fall Vocabulary and Decoding scores (see Ta
-
ble 5). Overall, this model accounted for approximately 70% of the variance in
children’s spring reading decoding scores (Level 1 r variance in unconditional
model = 231.60). There were significant main effects for children’s fall vocabulary
and fall decoding scores. Overall, children who began the school year with higher
vocabulary and decoding scores tended to achieve higher spring decoding scores.
318
CONNOR, MORRISON, KATCH
TABLE 5
Hierarchical Linear Model Results for the Instructional Variables’ Effect on
Spring Decoding Scores, Controlling for Mother’s Educational Level and
Home Literacy Environment Centered at the Grand Mean of the Sample
Fixed Effects Coefficient SE t-ratio Approximate df
For Intercept β
0
Spring decoding, γ00 39.71 .74 53.40 36
TME amount, γ01 .06 .21 .30
CME amount, γ02 .60 .42 1.42
TMI amount, γ03 .20 .26 .77
CMI amount, γ04 .03 .11 .25
CMI slope, γ05 .52 .44 1.20
MomEd, β
1
, γ10 .50 .37 1.34 41
Litscore, β
2
, γ20 –.01 .35 –.01 90
Fall decoding, β
3
, γ30 .88 .05 16.86*** 90
TME Amount × Decoding, γ31 –.02 .01 –2.08*
CMI Amount × Decoding, γ32 –.002 .01 –.17
CMI Slope × Decoding, γ33 .06 .05 1.27
Fall vocabulary, β
4
, γ40 .14 .07 1.98* 90
TME Amount × Vocabulary, γ41 –.002 .01 –.19
CMI Amount × Vocabulary, γ42 .02 .01 3.23**
CMI Slope × Vocabulary, γ43 –.11 .04 –2.64**
Random Effects Variance df χ
2
Intercept, U
0
1.0 17 25.51****
MomEd, U
1
.57 22 33.95*
Level 1, r 69.40
Note. Deviance = 758.75. Amount is in minutes/day and change over the course of the school
year. Slope is in minutes change per month, centered at the mean. TME = teacher-managed explicit;
CME = child-managed explicit; TMI = teacher-=managed implicit; CMI = child-managed implicit;
MomEd = mother’s educational level; Litscore = home literacy environment,
*p < .05. **p < .01. ***p < .001. ****p < .10.
Do Not Copy

There were no significant main effects for TME, TMI, CME, or CMI amounts or
CMI slope on spring decoding scores (i.e., nonsignificant coefficients γ
01
, γ
02
, γ
03
,
γ
04
, γ
05
). However, there were significant interactions between type of instruction
and fall vocabulary and decoding scores. TME interacted negatively and signifi
-
cantly with children’s fall decoding score (i.e., coefficient γ
31
), controlling for the
other variables. CMI amount interacted positively and CMI slope interacted nega
-
tively with children’s fall vocabulary scores (i.e., coefficients γ
42
and γ
43
, respec
-
tively). These interactions are presented in Figure 2. Mother’s educational level
and home literacy score did not significantly affect spring decoding scores (i.e., co
-
efficients γ
10
and γ
20
, respectively).
BEYOND THE READING WARS 319
FIGURE 2 Child Skill × Instruction Type interaction effect for spring decoding raw scores.
Top: Child fall decoding raw score (25th, 50th, and 75th percentiles of the sample) by TME
amount holding child fall vocabulary raw score constant at the mean for the sample. Middle:
Child fall vocabulary raw score (25th, 50th, and 75th percentiles of the sample) by CMI amount.
Bottom: CMI slope holding child fall decoding raw score constant at the mean for the sample.
Do Not Copy

The results of our model generally supported our hypotheses. As noted, we
found significant interactions between children’s fall decoding score and TME
amount but not for CMI. The reverse was the case for children’s fall Vocabulary;
interactions were significant for CMI amount and CMI slope but not for TME. We
had expected parallel patterns because of the documented association between
children’s vocabulary and decoding skills (Anderson & Freebody, 1981;
Scarborough, 1990). To investigate further, we computed the correlation between
fall decoding and vocabulary and found it surprisingly low (r = .26, p = .008).
Our model indicated that instructional effects varied with both fall vocabulary and
decodinglevelssimultaneouslyand independently;plus therewas a main effectfor fall
vocabulary and fall decoding on spring decoding scores. Thus the association between
amount of TME and CMI provided and children’s decoding growth, according to our
model, was independently affected by both their fall decoding and vocabulary skills.
Therefore, we decided to investigate these complex effects further by computing the
patterns of instruction associated with more versus less decoding skill growth for chil
-
dren with different patterns of fall vocabulary and decoding skill levels.
Because all of the variables in our model were continuous, there was an almost
infinite number of potential child skill patterns and corresponding effects of pat-
terns of instruction. To illustrate these interactions, we used our model to compute
the effect of these interactions for four distinct child patterns of skills: (a) children
with low fall vocabulary and decoding scores, which fell at the 25th percentile ac-
cording to published norms (the raw scores representing the respective percentiles
were selected using tables provided in the examiners’ manuals); (b) children with
high fall vocabulary and decoding scores, which fell at the 90th percentile; (c) chil-
dren with low fall vocabulary (25th percentile) but high fall decoding (90th percen-
tile); and (d) children with high fall vocabulary (90th percentile) but low fall de
-
coding (25th percentile). The 25th and 90th percentiles were selected because they
fell within the range of the data. Very few children in our sample had scores falling
at the 10th percentile for either decoding or vocabulary. To do this, we used the raw
score that, according to the norms, corresponded with the 25th and 90th percentile
for first-grade students and entered it in to the model to compute the fitted spring
Decoding outcome scores based on the varying patterns of instruction. These re
-
sults are presented in Figures 3 through 6 and described next.
Low vocabulary/low decoding skills.
For children whose fall vocabulary
and decoding scores fell at the 25th percentile for published test norms (n =12
within 1 standard error of measurement [SEM]), the model predicted significant
main effects and interaction effects as displayed in Figure 3. For these children, fit
-
ted results demonstrated that they achieved greater growth in decoding scores in
classrooms with more TME, whereas they achieved less growth in decoding scores
with less TME (Figure 3). In addition, the more CMI instruction these children re
-
320
CONNOR, MORRISON, KATCH
Do Not Copy

ceived in first grade, the less growth in Decoding they demonstrated. Yet the less
CMI they received, the more growth in decoding they exhibited (Figure 3).
There was also a significant interaction with CMI slope for children with low
fall vocabulary and decoding scores (Figure 3). Fitted results demonstrated that
children achieved stronger growth in decoding in classrooms with less CMI in the
fall but with increasing amounts through the spring of the school year (i.e., steep
positive slope). In contrast, children achieved less growth in spring decoding
scores in classrooms with steady amounts of CMI all year long (i.e., flat slope).
Note that CMI amount and CMI slope were independent effects and so these ef
-
fects were evident for classrooms with either high or low amounts of CMI. Never
-
theless, fitted results indicated that low amounts of CMI in combination with a
steep CMI slope (increasing amounts) was the pattern of instruction associated
with greater decoding skill growth for these children.
High vocabulary/high decoding.
For children whose fall vocabulary and
decoding scores fell at the 90th percentile using published norms (n = 8 within 1
SEM plus 8 whose scores fell above the 90th percentiles for vocabulary and decod
-
ing), a contrasting pattern emerged, as displayed in Figure 4. For these children, fit
-
BEYOND THE READING WARS 321
FIGURE 3 Effect of instruction (teacher-managed explicit [TME] and child-managed im-
plicit [CMI] amount and slope [CMI change]) on children’s spring decoding raw scores for chil-
dren who began the school year with scores falling at the 25th percentile of standardized norms
for fall vocabulary and decoding.
Do Not Copy

ted results revealed virtually no effect for amount of TME on their fall decoding
scores (see Figure 4). However, when in classrooms with higher amounts of CMI,
these children achieved greater growth in decoding scores by spring, whereas with
lower amounts of CMI, they achieved less growth in decoding scores (see Figure 4).
Here too, the effect of CMI slope had to be considered. For these children, a steady
dose of more CMI all year long (i.e., flat slope) yielded more growth in decoding,
whereas a steeper positive slope (i.e., less CMI in the fall increasing over the course
of the school year) yielded less growth in decoding scores (see Figure 4).
Low vocabulary/high decoding.
Only one child in our sample demon
-
strated scores that fit this profile—fall vocabulary scores at the 25th percentile but
decoding scores at the 90th percentile (n = 1 within 1 SEM). Another child demon
-
strated observed scores that fell above the 90th percentile for decoding and below
the 25th percentile for vocabulary. Thus, these results should be interpreted cau
-
tiously. Fitted results indicated that amount of TME had little effect on spring de
-
coding scores (see Figure 5). Fitted results also demonstrated that when children
received smaller amounts of CMI, they achieved stronger growth in spring decod
-
ing scores, whereas when they received higher amounts of CMI, they achieved less
growth in spring decoding scores (see Figure 5). Further, children receiving in
-
322
CONNOR, MORRISON, KATCH
FIGURE 4 Effect of instruction (teacher-managed explicit [TME] and child-managed im-
plicit [CMI] amount and slope [CMI change]) on children’s spring decoding raw scores for chil-
dren who began the school year with scores falling at the 90th percentile for fall vocabulary and
decoding using standardized norms.
Do Not Copy

creasing amounts of CMI over the school year (i.e., steep positive slope) achieved
stronger growth in decoding scores by the spring. However, when they received
steady amounts of CMI over the school year (i.e., flat slope) they achieved less
growth in decoding scores (see Figure 5).
High vocabulary/low decoding.
For children with fall vocabulary scores
falling at the 90th percentile for published norms but decoding at the 25th percen
-
tile (n = 10 within 1 SEM), fitted results indicated that children in classrooms with
higher amounts of TME achieved more growth in decoding scores. When they
were in classrooms with lower amounts of TME, children achieved less growth in
decoding scores (see Figure 6). Further, with higher amounts of CMI, children
achieved stronger growth in decoding scores and with lower amounts of CMI, they
achieved less growth (see Figure 6). For these children, a steady amount of CMI all
year long (i.e., flat slope) yielded greater decoding score growth whereas lower
amounts in the fall increasing over the school year yielded less growth in decoding
scores (see Figure 6).
BEYOND THE READING WARS 323
FIGURE 5 Effect of instruction (teacher-managed explicit [TME] and child-managed im
-
plicit [CMI] amount and slope [CMI change]) on children’s spring decoding raw scores for chil
-
dren who began the school year with fall decoding scores falling at the 90th percentile but fall
vocabulary falling at the 25th percentile using standardized norms.
Do Not Copy

Cumulative Impact of Instructional Variables:
Instruction Associated With Stronger Decoding Growth
The model can be used to elucidate the complex associations between child char
-
acteristics (i.e., fall vocabulary and decoding) and dimensions of instruction (i.e.,
TME amount, CMI amount, and CMI slope) and their combined effect on chil
-
dren’s spring decoding scores. To this end, we estimated the patterns of instruction
associated with stronger and weaker decoding growth for children in each of the
four patterns of skills (i.e., low vocabulary/low decoding, etc.). Again, because the
instructional variables were continuous, we chose values within the actual range of
our data for amounts of TME and CMI as well as CMI slope—the 10th percentile
and 90th percentiles of the sample (see Figures 7–10). Thus, meaningful values for
both child characteristics and types of instruction (amount and slope) were entered
into our model and the fitted spring decoding scores were computed. Note that al
-
though we use the term more effective instruction to describe patterns of instruc
-
tion associated with stronger decoding score growth, we do not intend to imply
causality. Moreover, the “causal effect” may go both ways—students respond to
instruction and teachers modify instruction to accommodate the skills of their stu
-
dents. Addressing these issues is beyond the scope of this study.
324
CONNOR, MORRISON, KATCH
FIGURE 6 Effect of instruction (teacher-managed explicit [TME] and child-managed im-
plicit [CMI] amount and slope [CMI change]) on children’s spring Decoding raw scores for
children who began the school year with fall decoding raw scores falling at the 25th percentile
and fall vocabulary raw scores falling at the 90th percentile of standardized norms.
Do Not Copy

For children with low fall vocabulary and decoding scores, the model indicated
that a more effective pattern of instruction included higher amounts of TME (90th
percentile for our sample) and smaller amounts of CMI (10th percentile for our sam
-
ple) that increased over the school year (90th percentile for our sample). A less effec
-
tive pattern of instruction included smaller amounts of TME (10th percentile for our
sample) and greater amounts of CMI (90th percentile for our sample) all year long
(slope at 10th percentile for our sample). As depicted in Figure 7, two children who
started first grade with identical decoding scores (½ year below grade level, n =12
within one SEM) showeddramatically different rates of decoding growthdepending
BEYOND THE READING WARS 325
FIGURE 7 Effects of more and less effective patterns of instruction on spring decoding grade
equivalent score for children who began the school year with low fall vocabulary raw scores
(25th percentile standardized norms) and low fall decoding raw scores (25th percentile stan
-
dardized norms).
FIGURE 8 Effects of more and less effective patterns of instruction on spring decoding grade
equivalent score for children who began the school year with high fall vocabulary (90th percen
-
tile standardized norms) and high fall decoding scores (90th percentile standardized norms).
Do Not Copy

on whether they received a more or less effective instructional pattern. As can be
seen, one child exposed to a more effective instructional regime achieved a decoding
score increase of almost two fitted grade equivalents. In contrast, the other child, re
-
ceiving a less effective instructional pattern, demonstrated only limited growth in
decoding scores or less than half a grade-equivalent. The children’s fitted spring de
-
coding scores differed by more than two grade-equivalents.
A contrasting pattern of effective instruction was revealed for children who had
high fall vocabulary and decoding scores (see Figure 8, n = 8 within 1 SEM). For
these children, instruction had a smaller effect overall; the difference between
more and less effective patterns of instruction on children’s spring decoding scores
326
CONNOR, MORRISON, KATCH
FIGURE 9 Effects of more and less effective patterns of instruction on spring decoding grade
equivalent score for children who began the school year with low fall vocabulary (25th percen
-
tile standardized norms) and high fall decoding scores (90th percentile standardized norms).
FIGURE 10 Effects of more and less effective patterns of instruction on spring decoding
grade equivalent score for children who began the school year with high fall vocabulary (90th
percentile standardized norms) and low fall decoding scores (25th percentile standardized
norms).
Do Not Copy

was less than half a grade-equivalent and children’s scores improved over the
school year even with the less effective pattern of instruction.
The model predicts substantial differences in achievement for more and less ef
-
fective patterns of instruction for the other two groups of children (see Figures 9 and
10). More effective instruction for children who had low vocabularyand high decod
-
ing scores (n = 2) included lower amounts of TME and higher amounts of CMI all
year long (i.e., flat slope). Contrasting more and less effective patterns of instruction
revealed more than one grade equivalent difference in spring decoding (see Figure
9). For children who had high vocabularyand lowdecoding scores (n = 10 within one
SEM), more effective instruction included higher amounts of TME and lower
amounts of CMI that increased over the school year (i.e., steep positive slope). Com
-
paring more and less effective patterns of instruction yielded a fitted grade equiva
-
lent difference of about one grade (see Figure 10).
DISCUSSION
The results of this study demonstrated that the effects of specific first-grade instruc-
tional practices on children’s decoding skills depended in large part on children’s en-
tering vocabulary and decoding skills. Children came to first grade with varyinglan-
guage and literacy skills. Classroom instruction also variedin amount and included a
mixture of code-based and meaning-based activities in differing proportions across
classrooms, which were encompassed in three dimensions of instruction (explicit
vs. implicit, teacher- vs. child-managed, and change).
Not surprisingly, children with weaker fall decoding skills achieved greater decod-
ing skill growth in classrooms with more teacher-managed explicit decoding instruc
-
tion (i.e., TME), whereas children with stronger decoding skills attained less decoding
skill growth in the same classrooms. Children with stronger vocabularies achieved
strongerdecoding skill growthin classroomswith manyopportunitiesfor independent
reading and writing activities (i.e., CMI) throughout the school year, whereas children
with weaker vocabulary skills achieved stronger decoding skills when opportunities
for independent reading and writing were kept to a minimum in the fall but were in
-
creased as the school year progressed. It is important to note that these effects occurred
simultaneously. Because the correlation between fall vocabulary and decoding was
weak, there were appreciable numbers of children with low decoding and high vocab
-
ulary skills, as well as a few children with high decoding and low vocabulary, in addi
-
tion to those with high vocabulary and high decoding, or low vocabulary and low de
-
coding. Fitted results indicated that depending on children’s decoding and vocabulary
skills, different but predictable patterns of instruction were associated with stronger
decoding skill growth.Forexample, children with low fall decoding skills and high fall
vocabulary skills achievedgreater decodingskill growthin classroomswith more time
spent in TME and high amounts of CMI all year long.
BEYOND THE READING WARS 327
Do Not Copy

Dimensions of Instruction
Overall, the coding scheme used in this study proved fruitful in identifying dimen
-
sions of instruction (explicit vs. implicit, teacher- vs. child-managed, and change in
amount over time) that varied among classrooms and that predicted growth in decod
-
ing. The variables appeared to capture important elements of literacy instruction
within both code-based and meaning-based philosophies and provided a way to
quantify and compare them. In general, the pattern of findings supported aspects of
both sides in the reading wars, but instructional effects differed for children with dif
-
fering skills.
It is notable that systematic changes in focus of instruction over the course of
the school year were apparent for this group of teachers. Overall, teachers focused
more on directly instructing basic decoding skills (i.e., TME instruction) at the be
-
ginning of the year but decreased this emphasis as the year progressed and, pre
-
sumably, as children’s decoding skills improved. This pattern of decreasing
amounts of TME over time was consistent among teachers. In contrast, overall,
CMI instruction increased in amount over the school year but varied significantly
among teachers. This suggests that teachers may have been responding to chil-
dren’s increasing independence as they learned classroom routines and better man-
aged their own learning as the school year progressed.
Two types of instruction, teacher-managed implicit decoding instruction (TMI) and
child-managed explicit decoding instruction (CME), had no significant effect on chil-
dren’s decoding skill growth. Overall, in the observed classrooms, there was little lan-
guage arts time devoted to CME activities, such as phonics workbooks or alphabet
worksheets. Thus this study cannot speak to the effect of such activities on children’s
decoding skill growth. TMI activities, such as discussions about books and teachers
reading to the students, did not appear to have a systematic effect on children’s decod
-
ing skill growth. However, such activities may be important for children’s reading
comprehension growth and emerging attitudes toward reading; more study is needed.
Conceptualizing instructional practices using dimensions of instruction pro
-
vided flexibility coupled with a fine-grained yet quantifiable view of classroom ac
-
tivities.This allowedus to modelthe complexinstructional patterns observedas well
as their interaction with child characteristics as they affected child outcomes. This
would not have been possible had we defined instruction one-dimensionally. Fur
-
ther, dimensions of instruction were specific to the outcome of interest. Had reading
comprehension rather than decoding been the outcome of interest, the dimension of
explicit versus implicit would have changed substantially. Given the high degree of
specificity observed for schooling effects (Christian et al., 2000) it is reasonable to
infer that explicit decoding instruction would have had an indirect impact on chil
-
dren’s reading comprehension growth. Rather, instructional activities explicitly fo
-
cusing on reading comprehension (Palincsar & Brown, 1984; Wixson, 1983) might
be expected to have a greater effect.
328
CONNOR, MORRISON, KATCH
Do Not Copy

Child Factors
The skills with which children began first grade had a direct and important effect
on their decoding skill growth. Overall, the effect of these child factors was greater
than the effect of classroom instruction practices and directly influenced the
strength of the instruction effects. Classroom instruction had a much greater influ
-
ence, either positively or negatively, for children who began first grade with
weaker vocabulary and decoding skills than it did for children with stronger skills.
As can be seen by comparing Figures 7 and 8, the potential difference in decoding
growth between more and less effective patterns of instruction was about half a
grade equivalent for children with high fall vocabulary and decoding skills but
more than two full grade equivalents for children with low fall vocabulary and de
-
coding skills. Children who began the year with stronger vocabulary and decoding
skills achieved some decoding skill growth almost regardless of classroom instruc
-
tional practices. Strong decoding and vocabulary skills may have acted to offset
the negative effects of less effective instructional practices and provided a path of
resilience (Werner, 2000) not available for children with weaker skills. Teacher’s
instructional practices had a greater effect—both positive and negative—for the
children who lacked these apparent protective factors. Children whose decoding
skills were below grade level remained below grade level in the face of poorly fit-
ting instructional patterns.
The interaction between children’s fall vocabulary and CMI amount and slope
underscore the association between oral language and reading skills. The relation
between vocabulary and reading skills has been well documented (Anderson &
Freebody, 1981; Scarborough, 1990). However, the interaction between vocabu-
lary and CMI instruction is intriguing and deserves further study. We speculate that
stronger vocabulary skills may support children’s implicit decoding learning be
-
cause they have a greater repertoire of words to rely on when they encounter un
-
known words. Further, there is a documented association between vocabulary and
phonological awareness (Foy & Mann, 2001; Gathercole & Baddeley, 1989;
Metsala, 1991). Thus children with strong vocabularies may have greater access to
more effective word-attack strategies (Stanovich, 1980).
Although this study focuses on vocabulary and decoding, there are clearly other
child characteristics that may influence children’s achievement and interact with in
-
structional activities, including phonological awareness skills, other oral language
skills, and self-regulation. In addition, the child factors of interest may vary when the
literacy outcome is fluency, comprehension, or writing skills. All merit further study.
Research Implications
From the foregoing discussion, it is evident that at the level of the individual child,
instructional activities and patterns that are considered high quality for one child
BEYOND THE READING WARS 329
Do Not Copy

may be considered poor quality for another. Substantial amounts of CMI (or mean
-
ing-based instruction) for a child with strong vocabulary skills may be associated
with stronger decoding gains, whereas the same amount of CMI for a child with
weaker vocabulary skills may be associated with substantially less progress. On a
broader level, our findings suggest that the effort to define and search for “quality”
in instructional environments may be somewhat misdirected. By focusing our ef
-
forts toward identifying those instructional variables that emerge as causally rele
-
vant in the classroom environment (amount, type, and change) and how they inter
-
act with child characteristics, we can more accurately identify “high-quality” or,
rather, effective instruction for individual children.
This study demonstrated the usefulness of classroom observation and the cod
-
ing system utilized for quantifying the nature and specificity of instructional ef
-
fects. Transcription and coding of the classroom observations yielded indices of
the absolute and proportional amounts of instructional time devoted to various lan
-
guage arts activities (e.g., word decoding, initial consonant stripping, sustained si
-
lent reading for language arts) as well as changes in amount provided over the
school year. These continuous multidimensional variables provided a more infor-
mative view of classroom instructional activities than would have been available
using categorical or one-dimensional variables. However, there are clearly other
dimensions of instruction that may be important to consider. For example,
teacher-managed instruction may be whole-class, small group, or individual, and
our dimensions did not capture this difference. Further, there may be differences in
outcomes for child-managed instruction that encompasses working with peers
compared to children working individually. In addition, there is a documented as-
sociation between teacher warmth and responsiveness to their students and student
achievement (deKruif, McWilliam, & Ridley, 2000; Mahoney & Wheeden, 1999).
Further, teachers’ ability to manage their classrooms is associated with children’s
learning (Brophy & Good, 1986; Taylor et al., 2000). Teacher warmth-responsivity
and teacher control-discipline are dimensions that should be studied further.
The interaction between child characteristics and patterns of instruction sug
-
gests that classroom instruction may be more effectively studied at multiple levels
considering both classroom practices and the characteristics of the children in the
classroom. Further, more distal sources of influence, such as school district policy
and community expectations, may well affect classroom instruction efficacy. Tay
-
lor et al. (2000) observed that school policies, atmosphere, beliefs, and reform ef
-
forts affected what happened in the classroom. This effect was apparent in this
study as well. Overall, as noted previously the school district in this study sup
-
ported a whole-language approach to teaching reading and encouraged teachers to
provide significant amounts of sustained silent reading time for their students. The
effect was evident when we compared the mean amount of TME provided (7.4 min
per day) with the mean amount of CMI provided (28.5 min per day). The ramifica
-
tion for children who began school with low decoding and vocabulary skills was
330
CONNOR, MORRISON, KATCH
Do Not Copy

that they were less likely to be in classrooms with patterns of instruction that were
effective for them. However, children with high fall decoding and vocabulary skills
were more likely to be in classrooms that were effective for them. Not surprisingly,
the school district had a documented achievement gap that increased as children
progressed through school (i.e., a Mathew Effect; Stanovich, 1986).
These results highlight the value of bringing an ecological approach to the study
of children’s development and the multiple sources of influence—proximal, distal,
and chronological—that directly and indirectly impact children’s literacy learning
(Bronfenbrenner, 1986). Observing and modeling classrooms as complex systems
subject to proximal (children, teachers, and instructional practices) and distal
sources of influence (home, school, and community) and changes in these influ
-
ences over time, may better inform our understanding of instruction and how best
to serve all students.
Implications for Classroom Practice
A central implication of these findings is that appropriate amounts of individual-
ized instruction can lead to significant decoding skill growth. Indeed, in other stud-
ies of instruction, researchers observed that the most effective teachers provided
additional explicit instruction in basic decoding skills to those children who
needed it (Wharton-McDonald et al., 1998). The dilemma, of course, is that indi-
vidualizing instruction inserts a level of complexity into designing and implement-
ing effective classroom practices, especially if teachers’ classrooms include many
students with very different skill levels. Yet, by understanding which instructional
activities are most effective in promoting children’s skill growth and in what
amounts over the school year, teachers can become more mindful of the effect of
their instructional practices. For example, early in the school year, teachers may
ask their students with strong vocabulary and decoding skills to read or write inde
-
pendently (CMI) while they provide explicit decoding instruction (TME) to chil
-
dren with weaker decoding and vocabulary skills.
Although a thorough discussion of specific classroom strategies is beyond the
scope of this article, use of flexible small student groups based on such assessments
of student abilities may approximate the kind of individualized instruction that will
optimize each child’s learning. Further, there is good evidence that how teachers in
-
teract with their students affects student achievement (e.g., scaffolding–coaching
vs. telling; Taylor et al., 2000). Clearly, to provide individualized instruction, teach
-
ers need to have some idea of the initial skill level of their students. Here another im
-
portant educational implication of our findings is that systematic assessment of chil
-
dren’s ability levels in important language and literacy skills should be a routine part
of each year’s classroom practices. Initial assessments during early fall should be
complemented by ongoing testing during the school year to monitor progress and
BEYOND THE READING WARS 331
Do Not Copy

adjust instruction accordingly (Juel & Minden-Cupp, 2000; Rayner et al., 2001;
Taylor et al., 2000; Westat, 2001; Wharton-McDonald et al., 1998).
In summary, the results of this study highlight the importance of understanding
interactions between the skills that children bring to school and the instructional
strategies they encounter in the classroom. These findings indicate that appropri
-
ately targeted instructional strategies can have a dramatic impact on growth of chil
-
dren’s early reading skills and their prospects for academic success.
ACKNOWLEDGMENTS
Carol McDonald Connor is now with the College of Education at Florida State
University and the Florida Center for Reading Research. Funding for this study
was provided by the National Institute of Child Health and Human Development
Grant R01 HD27176.
We thank Steve Raudenbush for his consultation regarding the analyses used in
thisstudy. We also thank Seung-Hee Son, Elizabeth Griffin,andmembers of the Uni-
versity of Michigan and Loyola projects as well as the parents, children, teachers,
andschool districtpersonnel withoutwhom thisstudy wouldnot havebeenpossible.
REFERENCES
Adams,M.J.(1990).Beginningtoread: Thinking andlearningaboutprint.Cambridge, MA: MIT Press.
Anderson, R. C., & Freebody, P. (1981). Vocabulary knowledge. In J. T. Guthrie (Ed.), Comprehension
and teaching: Research reviews (pp. 77–117). Newark, DE: International Reading Association.
Bredekemp, S., & Copple, C. (Eds.). (1997). Developmentally appropriate practice in early childhood
programs. Washington, DC: National Association for the Education of Young Children.
Bronfenbrenner, U. (1986). Ecology of the family as a context for human development: Research per
-
spectives. Developmental Psychology, 22, 723–742.
Brophy, J. E., & Good, T. L. (1986). Teacher behavior and student achievement. In M. C. Wittrock
(Ed.), Handbook of research on teaching (3rd ed., pp. 328–375). New York: Macmillan.
Catts, H. W., Fey, M. E., Zhang, X., & Tomblin, J. B. (2001). Estimating the risk of future reading diffi
-
culties in kindergarten children: A research-based model and its clinical implementation. Language,
Speech, and Hearing Services in Schools, 32, 38–50.
Christian, K., Morrison, F. J., Frazier, J., & Masseti, G. (2000). Specificity in the nature and timing of
cognitive growth in kindergarten and first grade. Journal of Cognition and Development, 1, 429–449.
Connor, C. M., Morrison, F. J., & Griffin, E. (2002). Measuring classroom instruction. Unpublished
technical report, University of Michigan Pathways to Literacy Project, Ann Arbor, MI. (Available
from pathwaystoliteracy@umich.edu)
Cunningham, A., & Stanovich, K. (1997). Early reading acquisition and its relation to reading experi
-
ence and ability 10 years later. Developmental Psychology, 33, 934–945.
Cunningham, A., Stanovich, K., & West, R. (1994). Literacy environment and the development of chil
-
dren’s cognitive skills. In E. Assink (Ed.), Literacy acquisition and social context. (pp. 70–90). Lon
-
don: Wheatsheaf/Prentice Hall.
332 CONNOR, MORRISON, KATCH
Do Not Copy

Cunningham, P., & Hall, D. (1998). The four blocks: A balanced framework for literacy in primary
classrooms. In K. R. Harris, S. Graham & D. Deshler (Eds.), Teaching every child every day:
Learning in diverse schools and classrooms (pp. 32–76). Cambridge, MA: Brookline.
Dahl, K. L., & Freppon, P. A. (1995). A comparison of innercity children’s interpretations of reading
and writing instruction in the early grades in skills-based and whole language classrooms. Reading
Research Quarterly, 30(1), 50–74.
deKruif, R. E. L., McWilliam, R. A., & Ridley, S. M. (2000). Classification of teachers’ interaction be
-
haviors in early childhood classrooms. Early Childhood Research Quarterly, 15, 247–268.
Dickinson, D. K., & Tabors, P. O. (2001). Beginning literacy with language. Baltimore: Brookes.
Dunn, L., & Dunn, L. (1987). Peabody Picture Vocabulary Test–Revised. Circle Pines, MN: American
Guidance Service.
Entwisle, D. R., & Alexander, K. L. (1988). Factors affecting achievement test scores and marks of
black and white first graders. The Elementary School Journal, 88, 449–472.
Foorman, B. R., Francis, D. J., Fletcher, J. M., Schatschneider, C., & Mehta, P. (1998). The role of in
-
struction in learning to read: Preventing reading failure in at risk children. Journal of Educational
Psychology, 90, 37–55.
Foy, J. G., & Mann, V. (2001). Does strength of phonological representations predict phonological
awareness in preschool children? Applied Psycholinguistics, 22, 301–325.
Gathercole, S. E., & Baddeley, A. D. (1989). Evaluation of the role of phonological STM in the develop
-
mentofvocabularyinchildren:Alongitudinalstudy. JournalofMemoryandLanguage, 28, 200–213.
Goodman, K. (1970). Reading: A psycholinguistic guessing game. In H. Singer & R. B. Ruddell (Eds.), Theo-
retical models and processes of reading (pp. 259–272). Newark,DE:International Reading Association.
Griffin, E., & Morrison, F. J. (1997). The unique contribution of home literacy environment to differ-
ences in early literacy skills. Early Child Development and Care, 127–128, 233–243.
Guthrie, J. T., Schafer, W. D., & Huang, C.-w. (2001). Benefits of opportunity to read and balanced in-
struction on the NAEP. Journal of Educational Research, 94(3), 145–162.
Hart, B., & Risley, T. R. (1995). Meaningful differences in the everyday experience of young American
children. Baltimore: Brookes.
Hiebert, E. H., & Raphael, T. E. (1998). Early reading instruction. New York: Harcourt Brace.
Juel, C., & Minden-Cupp, C. (1998). Learning to read words: Linguistic units and strategies. Ann Ar
-
bor, MI: Center for the Improvement of Early Reading Achievement.
Juel, C., & Minden-Cupp, C. (2000). Learning to read words: Linguistic units and instructional strate
-
gies. Reading Research Quarterly, 35, 498–492.
Locke, J. L. (1993). The child’s path to spoken language. Cambridge, MA: Harvard University Press.
Locke, J. (1997). A theory of neurolinguistic development. Brain and Language, 58, 265–326.
Mahoney, G., & Wheeden, A. C. (1999). The effect of teacher style on interactive engagement of pre
-
school-agedchildrenwithspeciallearning needs. Early Childhood Research Quarterly, 14(1), 51–68.
Markwardt, F. C. (1989). Peabody Individual Achievement Test–Revised. Circle Pines, MN: American
Guidance Service.
Metsala, J. (1991). Young children’s phonological awareness and nonword repetition as a function of
vocabulary development. Journal of Educational Psychology, 91(1), 3–19.
Morrison, F., & Cooney, R. (2002). Parenting and academic achievement: Multiple paths to early literacy.
In M. Bristol-Power (Ed.), Parenting and the child’s world: Influences on academic, intellectual, and
social-emotional development (pp. 141–160). Mahwah, NJ: Lawrence Erlbaum Associates, Inc.
Morrison, F. J., Smith, L., & Dow-Ehrensberger, M. (1995). Education and cognitive development: A
natural experiment. Developmental Psychology, 31, 789–799.
National Assessment of Educational Progress. (2000). U.S. Department of Education. Washington,
DC: National Center for Educational Statistics.
Neuman, S. B., & Dickinson, D. K. (2001). Handbook of early literacy research. New York: Guilford.
Nippold, M. (Ed.). (1988). Later Language Development. Boston: College Hill.
BEYOND THE READING WARS
333
Do Not Copy

National Reading Panel. (2000). Teaching children to read: An evidence-based assessment of the scientific lit
-
erature on reading and its implications for reading instruction (Summary). Washington, DC: Author.
Palincsar, A. S., & Brown, A. L. (1984). Reciprocal teaching of comprehension-fostering and monitor
-
ing activities. Cognition and Instruction, 1, 117–175.
Pressley, M. (1998). Readinginstructionthatworks:The case for balanced teaching.New York: Guilford.
Raudenbush, S. W., & Bryk, A. S. (2002). Hierarchical linear models: Applications and data analysis
methods (2nd ed.). Thousand Oaks, CA: Sage.
Ravitch, D. (2001). It is time to stop the war. In T. Loveless (Ed.), The great curriculum debate: How
should we teach reading and math (pp. 210–228). Washington, DC: Brookings Institutional Press.
Rayner, K., Foorman, B. R., Perfetti, C. A., Pesetsky, D., & Seidenberg, M. S. (2001). How psychological
science informs the teaching of reading. Psychological Science in the Public Interest, 2(2), 31–74.
Scarborough, H. S. (1990). Very early language deficits in dyslexic children. Child Development, 61,
1728–1743.
Snow, C. E., Burns, M. S., & Griffin, P. (Eds.). (1998). Preventing reading difficulties in young children.
Washington, DC: National Academy Press.
Stanovich, K. E. (1980). Towards an interactive-compensatory model of individual differences in the
development of reading fluency. Reading Research Quarterly, XVI, 32–71.
Stanovich, K. E. (1986). Matthew effects in reading: some consequences of individual differences in
the acquisition of literacy. Reading Research Quarterly, 21, 360–407.
Sternberg, R. J. (1996). Matching abilities, instruction, and assessment: Reawakening the sleeping gi-
ant of ATI. In I. Dennis (Ed.), Human abilities: Their nature and measurement (pp. 167–181).
Hillsdale, NJ: Lawrence Erlbaum Associates, Inc.
Stevenson, H., Parker, T., Wilkinson, A., Hegion, A., & Fish, E. (1976). Longitudinal study of individ-
ual differences in cognitive development and scholastic achievement. Journal of Educational Psy-
chology, 68, 377–400.
Stipek, D. J., & Ryan, R. H. (1997). Economically disadvantaged preschoolers: Ready to learn but fur-
ther to go. Developmental Psychology, 33, 711–723.
Taylor, B. M., Pearson, D. P., Clark, K., & Walpole, S. (2000). Effective schools and accomplished
teachers: lessons about primary-grade reading instruction in low-income schools. The Elementary
School Journal, 101(2), 121–165.
Thorndike, R. L., Hagen, E. P., & Sattler, J. M. (1986). Stanford-Binet Intelligence Scale (4th ed.). Chi
-
cago: Riverside.
Torgesen, J. K., Alexander, A. W., Wagner, R. K., Rashotte, C. A., Voelier, K. K. S., & Conway, T.
(2001). Intensive remedial instruction for children with severe reading disabilities: Immediate and
long-term outcomes from two instructional approaches. Journal of Learning Disabilities, 34, 33–58.
Werner, E. E. (2000). Protective factors and individual resilience. In J. P. Shonkoff & S. J. Meisels (Eds.),
Handbook of early childhood intervention: 2nd edition. New York: Cambridge University Press.
Westat. (2001). The longitudinal evaluation of school change and performance in Title 1 schools: Final
report (No. 2001–20). Washington, DC: U.S. Department of Education.
Wharton-McDonald, R., Pressley, M., & Hampston, J. M. (1998). Literacy instruction in nine
first-grade classrooms: Teacher characteristics and student achievement. The Elementary School
Journal, 99(2), 101–128.
Wixson, K. K. (1983). Postreading question-answer interactions and children’s learning from text.
Journal of Educational Psychology, 30, 413–423.
Manuscript received November 7, 2002
Accepted August 22, 2003
334
CONNOR, MORRISON, KATCH
Do Not Copy

APPENDIX
Coding Scheme for Classroom Observations
Language Arts: Time spent (number of minutes) engaged in activities that require
reading, writing, or reading/writing related things butthat are not focused on gaining
informationabout anothercontent area(science, socialstudies, math,drama, etc.).
a. Teacher read aloud: The teacher reads from a picture book, a chapter book,
or magazine, etc.; provides a book-on-tape for the children to listen to; or
shows a video wherein a story is presented.
b. Student read aloud, individual: A single child reads aloud, in a small group
or whole class, from a picture book, chapter book, magazine, or own writing.
c. Student read aloud, choral: More than one child reads aloud from picture
book, chapter book, magazine, poster, etc.
d. SSR (silent sustained reading): Children sit quietly and read to themselves
e. Teacher-managed group writing: The teacher is at blackboard–easel, work-
ing with children on a group writing activity. Children offer the content of
the written piece but the teacher puts the ideas into complete sentences,
with appropriate punctuation, etc.
f. Writing instruction: The teacher tells the children how to do things that
will help them to become independent writers, such as how to engage in ad-
vanced organizing (e.g., webbing, outlining), how to move from outline to
written product, how to proofread, and edit. This also includes instruction
in the different forms of writing (expository versus demonstration, etc.).
g. Teacher model writing: The teacher, without input from the children,
stands at the blackboard–easel and produces some sort of written product
(depending on the level of the students, it could be as small as a sentence).
The intent of the writing must be to model the act of writing and an appro
-
priate product.
h. Student group writing: The children are working in pairs or small groups to
produce a written product (such as a story). Not all of the children will ac
-
tually be doing the writing, but should be engaged in discussions about
what will be written.
i. Student independent writing: Children are quietly writing a story, poem, or
journal entry by themselves.
j. Spelling: Children are taking a spelling test or copying or practicing spell
-
ing words.
k. Discussion: Children are reviewing a storyline from a book the teacher has
been reading aloud (just prior to the teacher continuing with her reading),
or previewing a book the teacher is about to read. Children are responding
to questions that go beyond simple comprehension.
BEYOND THE READING WARS 335
Do Not Copy

l. Reading comprehension activity: Children are completing worksheets re
-
lated to material they have read, or are writing in response to something
they have read.
m. Listening comprehension: Children respond to questions about material
teacher is reading to them presently, such as who is the main character, or
what just happened.
n. Alphabet activity: Children are engaged in work that focuses their attention
on a particular letter of the alphabet. For example, they might have to make
a letter out of clay, color a paper that shows a particular letter and items that
begin with that letter, or put their body in the shape of a letter.
o. Letter sight–sound: Children are engaged in activities that focus their at
-
tention on the relationship between written form of individual letters and
the sound those letters represent. Included here are activities such as “signs
for sounds” wherein the teacher orally produces a single letter sound, and
the children circle the letter (from an array of letters on a prepared paper)
that represents that sound. This subactivity must combine the written form
and oral sounds that represent the written form. If no written form is used,
then the activity is more appropriately coded as initial consonant stripping
or word segmentation.
p. Initial consonant stripping: Children are identifying the beginning (initial)
consonant sound of words, aurally and not visually. If the activity is visual
and aural, then code the activity as letter sight–sound. Additionally, if the
initial sound is a vowel and only orally presented, then code the activity as
word segmentation.
q. Word segmentation: Children are engaged in activities wherein they break
words into subcomponents (syllables, sub-syllables or phonemes), orally; or
they are charged with constructing whole words from orally presented word
segments. Included here are activities such as learning word families (children
are presented with a rime, and must find onsets that make real words; this is of
-
ten an oral–written activity, but the initial response is oral). For an activity to
be coded as word segmentation rather than letter sight–sound, the intent of the
activity should be at the word level and not the letter level.
r. Vocabulary: The teacher and/or children are discussing the meaning of a
word or a phrase. This is a rarely used code, primarily because word-defi
-
nition discussions rarely last the minimum 1-min interval.
s. Conventions of print: The children are engaged in activities focused on
grammar or punctuation. Included here are activities such as “Daily Oral
Language” wherein children have to correct the errors in a sentence that the
teacher has written on the blackboard.
336
CONNOR, MORRISON, KATCH
Do Not Copy