Grimmetal Risknotatriskprofiles 2018

See discussions, stats, and author profiles for this publication at: https://www.researchgate.
net/publication/326745321
Early reading skill profiles in typically developing and at-risk first grade
readers to inform targeted early reading instruction
Article in Journal of School Psychology · August 2018

DOI: 10.1016/j.jsp.2018.05.009
CITATIONS READS
0 92
4 authors, including:
Ryan Grimm Nancy S. McIntyre

SRI International University of North Carolina at Chapel Hill
22 PUBLICATIONS 193 CITATIONS 44 PUBLICATIONS 178 CITATIONS
SEE PROFILE SEE PROFILE
Carolyn A Denton
Oregon Research Institute
78 PUBLICATIONS 2,651 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
Longitudinal study of attention and learning in school-aged children with ASD, ADHD, or typical development (Research supported by IES Grant R324A120168; P.
Mundy, P.I.) View project
Reading and language comprehension intervention studies with school-aged children with ASD View project
All content following this page was uploaded by Ryan Grimm on 26 July 2019.
The user has requested enhancement of the downloaded file.

Journal of School Psychology 69 (2018) 111–126
Contents lists available at ScienceDirect
Journal of School Psychology

journal homepage: www.elsevier.com/locate/jschpsyc
Early reading skill profiles in typically developing and at-risk first

T
grade readers to inform targeted early reading instruction☆
⁎
Ryan P. Grimma, , Emily J. Solaria, Nancy S. McIntyrea, Carolyn A. Dentonb
a
School of Education, University of California, Davis, 1 Shields Ave., Davis 95616, CA, United States
b
Children's Learning Institute, University of Texas Health Science Center Houston, 7000 Fannin, UCT 2443, Houston 77030, TX, United States
A R T IC LE I N F O ABS TRA CT
Action Editor: Milena Keller-Margulis This study identified distinct, homogeneous latent profiles of at-risk (n = 141) and not at-risk
Keywords: (n = 149) first grade readers. Separate latent profile analyses were conducted with each sub-
Reading disability group using measures of phonological awareness, decoding, linguistic comprehension, and oral
Reading comprehension reading fluency. This study also examined which measures best differentiated the latent profiles.
Reading intervention Finally, we examined differences on two measures of reading comprehension as a function of
Latent profile analysis profile membership. Results showed two latent profiles of at-risk students and three latent pro-
files of not at-risk students. Latent profiles were generally rank ordered with regard to
achievement across measures. However, the higher performing at-risk profile and the lowest
performing not at-risk profile were nearly identical across measures. Phonological awareness and
decoding measures were best at differentiating latent profiles, but linguistic comprehension was
also important for the lowest performing students. Oral reading fluency was limited to distin-
guishing the highest achieving students from the other profiles, and did not perform well with the
lower achieving profiles. Most of the pairwise comparisons of reading comprehension scores were
consistent across measures, but the nearly identical profiles showed a significant difference on
only one reading comprehension measure. Implications for identifying at-risk first grade readers
and designing targeted early reading interventions for at-risk students are discussed.
In order to comprehend written text, an individual must weave together several reading subcomponent skills; early identification
and remediation of these subcomponent skills is essential to avoid later reading comprehension difficulties. Difficulty acquiring the
necessary sub-skills can lead to struggles with reading comprehension that can persist throughout a student's academic career (Cain &
Oakhill, 2011; Francis, Shaywitz, Stuebing, Shaywitz, & Fletcher, 1996; Torgesen et al., 2001). In a preventative model, early
identification of struggling readers allows educational practitioners to provide intervention that may ameliorate later reading dif-
ficulties or disabilities. The purpose of this study was twofold. First, a two-step process was followed to identify and screen children
who were at-risk for reading difficulties; we utilized a combination of teacher referral and researcher-led screening. Next, once an
appropriate at-risk group was identified, we empirically derived distinct latent profiles based on multiple reading subcomponent
skills to compare the number and nature of latent profiles across an early at-risk group of students and a group not at-risk for reading
difficulties. This procedure was utilized to demonstrate the heterogeneous nature of reading subcomponent skill development within
☆
This work was supported by the Institute of Education Sciences, U.S. Department of Education, grant R324A150091 to the Regents of the University of California.
The content is solely the responsibility of the authors and does not necessarily represent the official views of the Institute of Education Sciences or the U.S. Department
of Education.
⁎
Corresponding author.
E-mail addresses: [email protected] (R.P. Grimm), [email protected] (E.J. Solari), [email protected] (N.S. McIntyre),
[email protected] (C.A. Denton).
https://doi.org/10.1016/j.jsp.2018.05.009
Received 7 February 2017; Received in revised form 2 February 2018; Accepted 14 May 2018
Available online 30 May 2018
0022-4405/ Published by Elsevier Ltd on behalf of Society for the Study of School Psychology.
R.P. Grimm et al. Journal of School Psychology 69 (2018) 111–126
a group considered to be globally at risk of reading difficulties to help inform targeted early intervention efforts. A comparison group
of not at-risk readers was utilized as a reference point for typical reading development, and also to demonstrate the range of reading
profiles at first grade entry.
Multi-tiered systems of support (MTSS) models are comprehensive frameworks providing increasingly intensive academic sup-
port. In preventative MTSS reading models there are typically three tiers of instruction, each providing increasing support for stu-
dents who are most in need (Fletcher & Vaughn, 2009; Fuchs & Fuchs, 2006; Fuchs, Mock, Morgan, & Young, 2003); intervention
need is based on student data. In this framework, all students receive an evidence-based core reading curriculum from the general
education teachers, this is also called Tier 1. When Tier 1 is deemed insufficient, based on student data, Tier 2 provides additional
support, usually in the form of small-group interventions aimed at ensuring that students at-risk for reading difficulties meet grade-
level benchmarks. Students with inadequate response to sufficiently intensive, evidence-based Tier 2 interventions receive more
individualized and intensive Tier 3 interventions. While classroom teachers are often able to intuitively identify struggling students
when their performance is starkly different from higher achieving peers (Begeny, Eckert, Montarello, & Storie, 2008; Gerber, 2005),
the appropriate use of assessments enables a more nuanced picture of students' abilities and targeted intervention needs. This can
inform the intervention design used within MTSS and may provide a more efficient intervention protocol. For instance, administering
assessments in multiple reading domains may reveal targeted intervention needs. In terms of assessing early reading skills devel-
opment, practitioners often focus on precursors of word reading such as phonological processing and letter and sound knowledge, as
well as word level skills (decoding) and oral reading fluency. Linguistic comprehension has been shown to explain variance in reading
comprehension in early elementary (Hoover & Gough, 1990; Kendeou, van den Broek, White, & Lynch, 2009; Storch & Whitehurst,
2002; Tunmer & Chapman, 2012), however it is not commonly assessed in early grades to determine risk status for reading. Given
empirical evidence suggesting reading comprehension is a function of both decoding skills and linguistic comprehension (Hoover &
Gough, 1990; Joshi & Aaron, 2000; Kendeou, van den Broek, et al., 2009; Ouellette & Beers, 2010; Tunmer & Chapman, 2012), it
would seem reasonable to suggest that early readers would be screened for risk in both subcomponent skills.
The Simple View of Reading (Gough & Tunmer, 1986; Hoover & Gough, 1990) has been proposed as a useful way to categorize
reading comprehension subcomponent skills, suggesting that word decoding and listening comprehension are the most significant
predictors of reading comprehension. The Simple View of Reading has been utilized in two ways to help explain the development of
reading comprehension. First, the framework has been used across multiple grade levels to explain the variance in reading com-
prehension performance (Adlof, Catts, & Lee, 2010; Adlof, Catts, & Little, 2006; Kendeou, Savage, & van den Broek, 2009; Kershaw &
Schatschneider, 2012; Nation, Cocksey, Taylor, & Bishop, 2010; Tunmer & Chapman, 2012). It has also been used to describe
subgroups of readers who struggle to develop the necessary sub-skills for successful reading comprehension (Catts, Adlof, & Weismer,
2006; Catts, Hogan, & Fey, 2003; Nation et al., 2010). It is imperative that school-based practitioners are able to identify students
who struggle with early reading so that targeted reading intervention can be implemented.
1. Reading sub-skills
Several studies have used reading sub-skills (e.g., phonological processing, word reading, linguistic comprehension, and oral
reading fluency) to differentiate readers (Catts et al., 2006; Catts, Compton, Tomblin, & Bridges, 2012; Catts, Fey, Zhang, & Tomblin,
1999; Nation et al., 2010). Other researchers have investigated how specific sub-skills predict later word reading and reading
comprehension (Adlof et al., 2006; Hoover & Gough, 1990; Kendeou, Savage, & van den Broek, 2009; Kendeou, van den Broek, et al.,
2009; Kershaw & Schatschneider, 2012; Roth, Speece, & Cooper, 2002; Silverman, Speece, Harring, & Ritchey, 2013; Storch &
Whitehurst, 2002; Wagner & Torgesen, 1987). Since reading sub-skills can predict later reading comprehension, it follows these sub-
skills can also be used to differentiate struggling readers from typically developing readers. Moreover, they may be used to identify
distinct subgroups of struggling and typically developing readers. Many previous studies have treated sub-skills as independent
contributors to reading comprehension, but few have examined how early reading profiles of sub-skills predict reading compre-
hension achievement. Examining profiles of reading sub-skills would enable researchers and practitioners to identify specific relative
strengths and weaknesses that can inform tailored interventions.
1.1. Phonological processing
Prior to word reading, children must develop phonological processing skills. Phonological processing is a multifaceted construct
that involves using phonological information to process oral and written language (Catts et al., 1999; Wagner & Torgesen, 1987). The
contribution of phonological processing to later word reading is well documented (e.g., Roth et al., 2002; Vellutino, Tunmer, Jaccard,
& Chen, 2007; Wagner et al., 1997; Wagner & Torgesen, 1987). Since it is a precursor to word and text reading, phonological
processing can be used to identify potentially struggling readers during the early elementary years. Though it is foundational to
decoding, it is often mastered in early reading development, limiting its ability to distinguish readers as they progress.
1.2. Decoding and reading fluency
As students develop phonological processing skills, they begin to apply these to decoding words. When students decode words,
they use their knowledge of letter-sound correspondence to read individual words. The significant relations between decoding and
reading comprehension have been well-established in the literature (e.g., Chen & Velluntino, 1997; Georgiou, Das, & Hayward, 2009;
Hoover & Gough, 1990; Joshi & Aaron, 2000). Accurate decoding is also a precursor to proficient oral reading fluency, which is an
112
individual's ability to read text quickly and accurately. Proficient reading fluency reduces the cognitive load needed to read in-
dividual words and allows readers to shift attention to comprehending text (LaBerge & Samuels, 1974). Thus, it has been described as
a “bridge” between decoding and reading comprehension (Pikulski & Chard, 2005) and is significantly related to reading compre-
hension (Chard, Vaughn, & Tyler, 2002; Fuchs, Fuchs, Hosp, & Jenkins, 2001; Fuchs, Fuchs, & Maxwell, 1988; Jenkins, Fuchs, van
den Broek, Espin, & Deno, 2003; Kim, Petscher, Schatschneider, & Foorman, 2010; Riedel, 2007; Silverman et al., 2013). As such,
decoding is often targeted in early reading interventions (Foorman, Francis, Fletcher, Schatschneider, & Mehta, 1998; Foorman,
Francis, Shaywitz, Shaywitz, & Fletcher, 1997; Rashotte, MacPhee, & Torgesen, 2001) to improve reading fluency. Yet, interventions
focused on decoding may not be sufficient to enable struggling readers to catch up to their peers. One reason may be that linguistic
comprehension is also key to reading comprehension, even in the early elementary grades.
1.3. Linguistic comprehension
Linguistic comprehension refers to a person's ability to understand spoken language. Extant research has found students who
struggle with reading comprehension also struggle with linguistic comprehension (Catts et al., 1999; Catts et al., 2006; Nation,
Adams, Bowyer-Crane, & Snowling, 1999; Nation et al., 2010). Thus, it makes sense to provide intervention that includes a linguistic
comprehension component to struggling readers; emerging research has begun to demonstrate positive effects of linguistic com-
prehension intervention with respect to reading comprehension outcomes (Bowyer-Crane et al., 2008; Clarke, Snowling, Truelove, &
Hulme, 2010). For example, Catts et al. (2006) found a profile of students who exhibited normal phonological processing skills, but
experienced difficulties with linguistic and reading comprehension. The authors argued intervention for these students should target
linguistic comprehension. Remediating phonological processing and decoding skills may lead to short-term gains in reading com-
prehension in early elementary, but, as texts become more complex, linguistic comprehension difficulties may impede later reading
comprehension if they are not also addressed.
1.4. Differentiating readers according to reading sub-skills
Extant research has identified subgroups of readers using arbitrarily defined cutoff scores on one or more variables (or a com-
posite variable) to differentiate readers based on reading sub skill strengths and weaknesses. Nation et al. (2010) conducted a study in
which they examined the reading development of children between ages 5 and 8 who were accurate word readers, but poor com-
prehenders; poor comprehenders exhibited normal word reading development, but they experienced difficulties with linguistic
comprehension. Both linguistic comprehension and reading comprehension difficulties were persistent across the ages. Since lin-
guistic comprehension difficulties were apparent before the poor comprehenders acquired word reading skills, the authors concluded
that early poor linguistic comprehension can place children at risk for later reading comprehension difficulties. Catts et al. (2003) and
Catts et al. (2006) went a step further by identifying multiple subgroups of poor readers. Catts et al. (2003) used measures of word
reading and linguistic comprehension to subgroup poor readers in early elementary, defining poor readers as those who scored one
standard deviation below the mean on a reading comprehension composite. They also separated the poor readers into four subgroups
based on word reading and linguistic comprehension and found that linguistic comprehension skills in second grade were related to
kindergarten measures of linguistic comprehension. Furthermore, they found the subgroups were moderately stable into fourth
grade. However, the authors also reported that their sample did not form homogenous subgroups. They acknowledged this was
related to their method of imposing a cutoff score and subgroup membership could shift depending on the cutoff score. As described
below, the present study empirically derived subgroups rather than imposing a cutoff score.
Catts et al. (2006) created reading subgroups (i.e., poor comprehenders, poor decoders, and typical readers) from composite
scores in reading comprehension and word recognition using the 25th and 40th percentiles. Specific deficits were defined as scoring
below the 25th percentile in one domain, but above the 40th percentile in the other domain; students were examined at kindergarten,
second, and fourth grade. Results showed the poor comprehenders scored significantly lower on a language comprehension com-
posite compared to poor decoders and typical readers at all three timepoints. In contrast, poor decoders scored significantly lower
than typical readers on the same measure in kindergarten only. However, poor decoders achieved significantly lower scores on the
early measures of phonological awareness. Thus, the analyses suggest subgrouping was a valid methodological approach and that
subgroups were fairly stable across time.
Using latent transition analysis (LTA), Catts et al. (2012) identified four latent classes of readers that were consistent across time,
from kindergarten through tenth grade. The latent classes represented normal readers, readers with word reading disabilities, readers
with comprehension disabilities, and readers with disabilities in word reading and comprehension. Results showed students moved
between latent classes over time, which enabled the researchers to identify students with late-emerging reading comprehension
difficulties; these students had a history of difficulty with linguistic comprehension. The authors posited that their reading com-
prehension difficulties may not have emerged until later grades because most early reading curricula place a heavier focus on
decoding and word reading as compared to comprehension skills. The link between early oral language and reading comprehension
skills has been documented elsewhere (Catts et al., 1999; Kendeou, van den Broek, et al., 2009; Nation et al., 2010; Storch &
Whitehurst, 2002). Therefore, in an MTSS context, identifying early struggling readers and providing targeted intervention based on
phonological and word reading skills may not be sufficient to ameliorate later reading comprehension difficulties for students with
late-emerging reading difficulties. Additionally, identification schemes based solely on phonological and word reading skills will
likely miss students with average word reading skills, but below average linguistic comprehension.
Compton, Fuchs, Fuchs, Elleman, and Gilbert (2008) used LTA to derive latent classes of readers from first through fourth grade.
113
The authors identified two latent classes at first, second, and fourth grades, typically developing readers and students with reading
disabilities. Results showed the latent classes were stable over time and the majority of students remained in their respective latent
class across grades, but 7% of typically developing readers transitioned into the reading disabled class. This was evidence for late-
emerging reading disabilities, which was an important finding as these students are difficult to detect in early elementary by defi-
nition. As these students are less likely to be identified as reading disabled in the early grades, they will not receive the necessary
intervention provided in MTSS frameworks. One of Compton et al.'s (2008) aims was to explore the efficacy of the latent class
indicators in terms of predicting late-emerging RD. However, the indicators were limited to two measures of word reading (word
identification and sight word efficiency) and one measure of passage comprehension. The authors compared models with and without
sight word efficiency and found that its inclusion reduced the number of false negatives; students who would typically not be
identified as struggling readers, but do exhibit RD in later elementary. Importantly, these authors also identified linguistic com-
prehension, phonemic awareness, and phonemic decoding efficiency as promising first-grade predictors of late-emerging RD. While
all three predictors produced false positives, linguistic comprehension produced the least (35) compared to phonemic awareness (79)
and phonemic decoding efficiency (83). While the authors considered the number of false positives identified by linguistic com-
prehension to be unacceptable, it is worth noting linguistic comprehension was measured by a single variable. Using other or multiple
measures of linguistic comprehension may have produced different results. Finally, the authors acknowledged measures other than
early word reading need further exploration as early indicators of later RD. Given these findings and others that have shown students
with late-emerging RD may have typically developing word reading skills (Badian, 1999; Chall, 1983; Leach, Scarborough, &
Rescorla, 2003), it is reasonable to explore additional and multiple measures of reading sub-skills, such as linguistic comprehension,
as indicators of RD.
1.4.1. Use of latent profile analysis in the present study

Francis et al. (2005) argued that the use of cutoff scores on single variables can lead to instability in classification as children who
score near the cutoff boundary may fluctuate between subgroups. Research methodologies that simultaneously examine multiple
measures of reading skills to delineate reading profiles may produce more robust and accurate results. Very few studies have utilized
mixture models to empirically identify latent subgroups of readers (i.e., not rely on predetermined cutoff scores), which is the
approach used in the present study. An advantage of this approach is that individuals can be classified into subgroups according to
patterns of scores across multiple variables. Additionally, classification criteria are data-driven rather than imposed by the re-
searchers. While mixture models also suffer from classification error, this can be measured and accounted for by a given individual's
posterior probability of belonging to a given latent profile. This provides researchers with a measure that can be used to judge the
adequacy of the model, and this classification error can be accounted for when examining covariates, distal outcomes, or other
auxiliary variables specified to be related to latent profile membership. In this study, we utilized multiple measures of each sub-skill
to ensure the robustness of the emergent profiles (Francis et al., 2005). Finally, these analyses are model-based, which allows for
replication. That is, models shown to be preferred in one study can be applied to independent samples and tested for their viability or
rejection.
Latent profile analysis (LPA) is a probabilistic technique that is able to empirically derive categorically distinct profiles of in-
dividuals. This approach is similar to latent class analysis (LCA; which uses binary or ordinal variables), but is conducted with
continuous variables (Gibson, 1959; Lazarsfeld & Henry, 1968). Fig. 1 presents a conceptual diagram of the LPA in the current study.
Separate LPAs were conducted for the subsamples of at-risk and not at-risk students, but we present one diagram, which is the same
model for each subsample. The categorical latent profile variable is represented by a circle. Boxes above the circle represent each of
the observed variables used to measure the latent profile variable, and the arrows indicate the latent variable is measured by the
observed variable. The boxes to the right of the latent variable represent observed reading comprehension measures. Profile-specific
means of these variables were estimated (represented by the arrows), but were not used to measure the latent profile variable. A
separate procedure (described in the Data Analysis section) was used to estimate the profile-specific means of the reading com-
prehension variables.
Elision Blending LWID WA CELF QRI G Rate
Latent WJ PC
profiles
G Comp
Fig. 1. Conceptual diagram of latent profile analyses for both the at-risk and not at-risk groups. LWID = Letter Word Identification; WA = Word
Attack; CELF = Clinical Evaluation of Language Fundamentals – 4 Understanding Paragraphs subtest; QRI = Qualitative Reading Inventory – 5; G
rate = Gray Oral Reading Tests – 5 Rate subscore; WJ PC = Woodcock Johnson IV Passage Comprehension subtest; G comp = Gray Oral Reading
Tests – 5 Comprehension subscore.
114
Using multivariate data, individuals are classified together within a given profile because their pattern of responses across
multiple measures are more similar to each other than individuals assigned to a different profile. Latent profiles differ in their means
across variables, but can also differ in variances and covariances (Masyn, 2013). In this study, variances and covariances were
constrained to equality across classes, which is commonly done for parsimony and is the default specification in Mplus. Deriving
profiles from responses – and simultaneously using the same responses to assign individuals to profiles – is a process that requires
many iterations (i.e., using multiple random start values) as well as optimizing the resulting solutions. Additionally, the enumeration
process used to choose the optimal model is iterative and relies on a number of considerations. Enumeration can be thought of as
analogous to exploratory factor analysis and choosing the number of factors to retain based on multiple fit statistics and substantive
interpretability. The researcher must specify the number of profiles to be estimated, collect and compare multiple fit statistics for all
models, and check each model's interpretability. In this study, 500 random starts with 100 optimizations were used and models with
1–5 profiles were specified because there were model convergence problems with more than five profiles.
1.4.2. Present study

In the present study, we use an independent sample to address multiple objectives cited in the individual studies mentioned
above. The research questions for the present study are: 1) How many reader profiles would emerge for each risk status group and
what would be the qualitative differences between the profiles? 2) Which measures best differentiate the reading profiles? and 3)
How do the latent profiles differ in terms of reading comprehension achievement?
2. Methods
2.1. Participants
Data for this study come from a larger study investigating the efficacy of a reading intervention. All data in this study were
collected prior to the beginning of the intervention. First-grade students were drawn from 30 classrooms and 15 schools from a
suburban and rural region of California and an urban region of Texas. The sample consisted of N = 290 students, with 47.1% female
and 77.7% qualifying for free lunch. The ethnic breakdown of the sample was 44.8% Latino(a), 23.4% African-American, 19.2%
Caucasian, 3.5% Asian, and 8.7% “Other” or “Mixed.” Initial screenings (described below) identified n = 141 students as at-risk and
n = 149 students as not at-risk. Nine or ten students were recruited from each classroom with approximately half of the recruited
students considered at-risk. All students were enrolled in general education classrooms full time and none of the students were
receiving special education services.
2.2. Determining at-risk status
At-risk status was determined using a two-step process. Risk status was used to identify students in need of the reading inter-
vention provided by the larger project. First, teachers ranked all of their students according to their judgment of students' reading
skills, but they were not asked to officially designate risk status. Students ranked in the lower performing 50% of each class were then
screened to identify those most at-risk of experiencing reading difficulties. Thus, while teacher judgment was used to identify po-
tential students for screening, their judgment was not used to define risk status, nor was it a variable in this study. We used the Texas
Primary Reading Inventory (TPRI; Children's Learning Institute & Texas Institute for Measurement, Evaluation, and Statistics, 2010),
which is designed to identify students in grades K-3 at-risk for reading difficulties and as a diagnostic tool. First, students were
administered three screens assessing letter sounds, word reading, and blending phonemes. If students passed all three screens, they
were considered not at-risk and were discontinued from the screening process. If students failed at least one of the three screens, they
were administered a brief word-reading screen that consisted of simple words that students receive exposure to prior to the beginning
of first grade. Prior research (Denton et al., 2010; Mathes et al., 2005) has demonstrated that this measure can successfully identify at-
risk first graders. Finally, students were administered a listening comprehension screen from the TPRI. Examiners orally read two
passages to students. Following each passage, examiners asked them to answer six literal and inferential questions, yielding a total of
12 questions. Students were considered at-risk if they scored five or below (out of 15) on the word-reading screen and had an average
score of three or below (out of six) on the listening comprehension screen. These cut scores were chosen because they were consistent
with prior studies (Denton et al., 2010; Mathes et al., 2005) that have used the TPRI. Students could only qualify as at-risk if they
qualified in both word reading and listening comprehension constructs; students were disqualified if they exhibited difficulties in
only one construct. We did not include classrooms with less than four students who qualified as at-risk. If a classroom contained more
than five students who qualified, we selected the five lowest-performing students for purposes of the reading intervention. The not at-
risk group was comprised of five students from each classroom who were randomly selected from the 50% of students who were
teacher-nominated as not at-risk. Fig. 2 presents a flow chart summarizing the selection of the at-risk and not at-risk subsamples.
2.3. Procedures
Examiners received group training on the assessment battery and were required to meet reliability criteria for each assessment
with the trainers. Trainers administered the battery to the individual examiners in a role-playing situation. Examiners were required
to administer each assessment with a minimum of 90% accuracy per assessment. Students were administered the assessment battery
in quiet areas outside of their classrooms. To ensure the examiners continued to accurately administer the assessment battery, trainers
115
Teachers rank students in their classrooms (n = 767)
Lower-performing 50% of students Not at-risk students randomly selected

from each classroom screened from the higher-performing 50% of
for risk of reading difficules each classroom
(n = 384) (n = 149)
Pass first 3 Fail any of first 3

screening measures screening measures
(n = 212) (n = 172)
Disconnue screening. Administer word

Child deemed not at-risk, reading and listening
but not in the higher- comprehension
performing 50% screening measures
Risk defined by both

word reading and
listening comprehension
(n = 141)
Fig. 2. Flow chart summarizing the process used to determine at-risk status and choosing the not at-risk subsample.
checked all assessments at the end of each student's assessment session. Any errors were immediately corrected before the student
returned to class. Data were collected within a 4-week period during September and October. Finally, all data were double-entered
and, if errors occurred, were corrected by checking the original testing forms.
2.4. Measures
We converted scaled and standard scores to z-scores using national norms for all measures except QRI-5, which is non-normed.
Scores on this measure were converted to z-scores using within-sample means and standard deviations.
2.4.1. At-risk screen

The TPRI was utilized to screen students to determine if they were at-risk for reading difficulties. The TPRI is designed to identify
students in grades K-3 at-risk for reading difficulties and as a diagnostic instrument to assist in planning instruction. To screen first
graders for risk, the following measures from the TPRI were utilized: Letter Sound Identification, Word Reading, Blending Phonemes,
and Listening Comprehension. In the Letter Sound Identification subtest, students are presented with a short list of letters and asked
to identify the most common sound for each item; the subtest has 10 items, reliability is reported at 0.91. The Word Reading measure
contains 5 items and requires students to read the words out loud, reliability is 0.96. The Blending Phonemes subtest has 5 items,
reliability is reported to be 0.92. For this subtest, the assessor presents individual phonemes to the student and they are required to
use the phonemes presented to form a real word. The Listening Comprehension subtest requires students to listen to two passages and
answer a series of questions orally after listening, with a total possible score on each passage of 6 points; reliability is reported at 0.72.
2.4.2. Phonological awareness

The Elision and Blending Words subtests of the Comprehensive Test of Phonological Processing (CTOPP; Torgesen, Wagner, &
Rashotte, 1999) were used to measure phonological awareness. For the Elision subtest, the examiner asks an individual to repeat a
word presented orally while omitting one of the word's sounds (e.g., “say bold. Now say bold without /b/”). For the Blending Words
subtest, an examiner plays a recording that asks the individual to combine separate sounds into whole words. For example, the first
practice item presents the individual with “What word do these sounds make? Can-dy” with the correct response being candy. The
examiner's manual reports a test-retest reliability of 0.88 for both subtests with individuals aged five to seven years. For the current
sample, this study calculated Cronbach's alpha of 0.90 for Elision and 0.88 for Blending Words. In terms of validity, the CTOPP
manual reports appropriate content-description, criterion-prediction, and construct-identification validity for these subtests. Ad-
ditionally, both subtests have high factor loadings on a phonological awareness latent factor.
116
2.4.3. Decoding
Letter-Word Identification (LWID) and Word Attack (WA) subtests of the Woodcock-Johnson IV (WJ-IV; Schrank, Mather, &
McGrew, 2014) were used to assess decoding skills. The LWID subtest presents individual letters/letter combinations and words in a
list format. The WA subtest presents lists of pseudowords to assess phonological decoding. The technical manual reports split-half
reliability coefficients for the ages in this study between 0.96 and 0.98 for LWID and 0.94 and 0.96 for WA. Cronbach's alpha for
LWID for the current sample was 0.96 and 0.89 for WA.
2.4.4. Linguistic comprehension

Two separate measures of linguistic comprehension were utilized. First, the Understanding Spoken Paragraphs subtest from the
Clinical Evaluation of Language Fundamentals, Fourth Edition (CELF-4; Semel, Wiig, & Secord, 2006) was used. An assessor read
three short passages to students and asked five comprehension (literal and inferential) questions following each passage, yielding a
total of 15 questions. Possible scaled scores range from 1 to 19. The test manual reports a Cronbach's alpha of 0.69 and 0.65 and split-
half reliability of 0.74 and 0.73 for six-year olds. The test manual also reports multiple forms of validity evidence based on response
process, structural equation models, and correlations with the CELF, Third Edition. Additionally, the manual reports validity with
regard to varying demographic characteristics as well as varying disorders related to language development. Cronbach's alpha for the
current sample was calculated as 0.81.
The second measure was an adapted version of the Qualitative Reading Inventory-5 (QRI-5; Leslie & Caldwell, 2011). Students
were read a passage then asked six questions regarding explicit and implicit details of the passage. We analyzed the raw score (i.e.,
total number of correct responses) as this assessment is non-normed. We calculated a Cronbach's alpha of 0.66 for the study sample.
The technical manual reports an alternate-form reliability, meaning instructional levels associated with two alternative stories
matched 91% of the time for first graders. It also reports standard errors of measurement ranging from 0.14–0.16 for the stories used
in this study. Cronbach's alpha for this sample was 0.65. The manual provides evidence of content, criterion-related, and construct
validity.
2.4.5. Oral reading fluency

The fluency measures used was the Rate subscale of the Gray Oral Reading Test-5 (GORT-5; Wiederholt & Bryant, 2012). Students
were asked to read aloud at least two passages. Students receive a score for each passage based on the number of seconds it took them
to read the passage. We analyzed a final normed score that accounts for the total number of passages read. The examiner's manual
reports Cronbach's alpha of 0.86 for six-year olds using Form A. Cronbach's alpha for this sample was 0.84. The technical manual also
reports appropriate content-description, criterion-prediction, and construct-identification validity.
2.4.6. Reading comprehension

Two measures were used to assess reading comprehension. The first was the Passage Comprehension (PC) subtest of the WJ-IV,
which requires individuals to read short sentences and passages with a missing word. They are asked to supply the word. The
examiner's manual reports split-half reliability coefficients for this age range between 0.93 and 0.98. Cronbach's alpha for the current
sample was 0.90. The second measure was the Comprehension score from the GORT-5. Individuals read increasingly complex pas-
sages and were asked five open-ended comprehension questions after each passage. This test is discontinued when students are
unable to fluently read two consecutive passages according to the test's stopping criterion. The technical manual reports Cronbach's
alpha of 0.92 for this age range and was 0.72 for this sample.
2.5. Data analysis plan
First, we compared the teacher-nominated at-risk and not at-risk groups across all variables to ensure the groupings were sup-
ported empirically. We conducted a series of t-tests with all of the variables. To account for multiple comparisons, we employed a
Bonferroni correction and used an adjusted p-value of 0.006 as the cutoff for statistical significance.
The latent profile analyses (LPA) were conducted in multiple steps with each subgroup separately. The same steps were performed
for each subgroup. A conceptual diagram is presented in Fig. 1. To account for the nested nature of the data, we clustered students at
the teacher level. We began by fitting a model with one profile then increased the number of profiles by one with each subsequent
iteration. Model fit statistics were recorded and compared to determine the optimal number of profiles within each subgroup.
Additionally, we used substantive reasoning to ensure the chosen model made conceptual sense (Muthén, 2003). All models were run
using Mplus version 7.4 (Muthén & Muthén, 1998–2015) using Full Information Maximum Likelihood estimation. This estimator
allowed students to be included as long as they had data on at least one observed variable unless they were missing data on the distal
outcomes (i.e., reading comprehension variables). As these models are known to converge on local solutions, we used a large number
of random start values (McLachlan & Peel, 2004).
We used commonly employed fit statistics, specifically the Bayesian Information Criterion (BIC), Adjusted BIC (ABIC) where
lower values indicated better fit. We also used two likelihood-based indices, the Lo-Mendell-Rubin Likelihood Ratio Test (LMR) and
the Bootstrap Likelihood Ratio Test (BLRT). These models provide a p-value to compare a k profile model to a k - 1 profile model. A
non-significant p-value indicates the additional profile did not significantly improve the model. Finally, we also calculated values for
the Approximate Weight of Evidence (AWE) criterion, with lower values indicating better fit. See Nylund, Asparouhov, and Muthén
(2007) and Masyn (2013) for additional information on the fit statistics.
Once the preferred unconditional model (i.e., without distal outcomes) was chosen, the distal outcomes (i.e., reading
117
Table 1
Descriptive statistics disaggregated by risk status and t-tests.
At-risk Not At-risk
Measure n M SD Min Max n M SD Min Max t
Elision 141 6.82 2.66 1.00 16.00 147 10.35 2.68 3.00 19.00 11.20
Blending words 141 9.34 2.07 4.00 13.00 147 11.53 2.47 4.00 18.00 8.15
Letter word 141 77.19 12.17 40.00 112.00 148 100.34 14.41 70.00 135.00 14.78
Word attack 141 83.82 16.88 42.00 124.00 148 107.40 11.60 73.00 132.00 13.90
CELF 141 6.58 3.74 1.00 14.00 147 9.12 3.43 1.00 15.00 6.00
QRI 141 2.44 1.59 0.00 6.00 149 3.57 1.67 0.00 6.00 5.90
GORT rate 140 5.39 1.21 2.00 11.00 148 9.01 2.60 4.00 15.00 15.30
WJ PC 140 81.76 10.96 40.00 113.00 148 100.14 11.17 78.00 128.00 14.08
GORT comp 141 4.34 1.02 2.00 11.00 148 8.31 2.53 4.00 14.00 17.32
Note. Scores are norm-referenced except for QRI. CELF = Clinical Evaluation of Language Fundamentals - 4; QRI = Qualitative Reading Inventory -
5; GORT rate = Gray Oral Reading Tests - 5 Rate subscore; WJ PC = Woodcock Johnson IV Passage Comprehension subtest; GORT comp = Gray
Oral Reading Tests - 5 Comprehension subscore; All t-tests significant at p < 0.001.
comprehension variables) were included using the BCH approach (Asparouhov & Muthén, 2014; Bakk & Vermunt, 2016; Bolck,
Croon, & Hagenaars, 2004; Vermunt, 2010). This approach avoids shifts in the latent profiles that can occur when auxiliary variables
(i.e., distal outcomes) are included in the model. After choosing the preferred unconditional model, weights are applied to each
individual based on posterior probabilities of membership in each latent profile. This treats latent profile membership as known.
Finally, the distal outcome variables are included and a multiple group analysis is performed treating the latent profiles as observed
subgroups. This approach avoids shifts in the latent profiles when the distal outcomes are included, which can occur with other
methods of including distal outcomes in mixture models (Asparouhov & Muthén, 2014). Mean estimates of the two reading com-
prehension variables were then estimated for each latent profile and compared for significant differences. Within each risk group,
significant differences were examined using a series of Wald tests, which were provided by the BCH approach and available in the
Mplus output. To test for significant differences across risk groups, we calculated pooled variances and conducted t-tests. This yielded
six t-tests for each reading comprehension measure. To control for familywise error rate, we applied a Bonferroni correction and used
a p-value of 0.008 to indicate significant differences.
3. Results
3.1. Descriptive statistics
Table 1 presents descriptive statistics disaggregated by risk status. The mean scores for the not at-risk group were significantly
higher compared to the at-risk group across all variables, as indicated by t values in the far right column of Table 1. All t-tests were
significant at p < 0.001. Additionally, the maximum score achieved by the not at-risk group was greater than the maximum score
achieved by the at-risk group for every variable except QRI-5. For most variables, this was also true of the lowest score achieved by
each subgroup.
Table 2
Fit statistics of the latent profile analyses by risk status.
# of classes LL # of parameters BIC ABIC AWE LMR p-value
At-risk
1 −1204.97 14 2479.23 2434.93 2590.51 –
2 −1103.10 22 2315.07 2245.46 2489.95 0.06
3 −1053.88 30 2256.22 2161.30 2494.69 0.21
4 −1009.64 38 2207.33 2087.10 2509.39 0.26
5 Non-positive definite
Not at-risk
1 −1367.71 14 2805.47 2761.17 2917.53 –
2 −1223.55 22 2557.18 2487.56 2733.27 0
3 −1172.44 30 2495.00 2400.06 2735.12 0.01
4 −1141.56 38 2473.27 2353.01 2777.42 0.22
5 −1122.98 46 2476.15 2330.57 2844.32 0.43
Note. Bold values indicate the preferred model for a given fit statistic. LL = Log-likelihood; BIC = Bayesian Information Criterion; ABIC = Adjusted
BIC; AWE = Approximate Weight of Evidence; LMR = Lo-Mendell-Rubin Likelihood Ratio Test.
118
Reading Profiles
Phono Awareness Decoding Linguistic Comp Fluency
2
1.5
1
0.5
Z-scores
0
-0.5
-1
-1.5
-2
-2.5
Elision Blending Letter Word Word Attack CELF QRI GORT rate
ID
At-Risk Global (43.6%/21.0%) At-Risk Low Fluency (56.4%/27.6%)

Not At-Risk Low Fluency (33.3%/17.2%) Average (41.1%/21.4%)
Above Average (25.6%/12.8%)
Fig. 3. Item-profile plot of at-risk and not at-risk students. Separate LPAs were conducted for each subsample, but are presented on one plot for ease
of interpretation. Solid lines denote latent profiles of at-risk students. Dashed lines denote latent profiles of not at-risk students. Percentages before
the slash represent the percent within each risk status group. Percentages after the slash represent the percent of the overall sample.
3.2. Latent profiles for at-risk readers
We examined models consisting of one to four profiles as models with five and six profiles did not converge. Fit statistics can be
seen in Table 2. Neither the BIC nor ABIC reached a minimum value. However, the AWE reached a minimum value at two profiles.
Similarly, the LMR value approached significance at two profiles, but not at three or four profiles. The BLRT never became non-
significant and was considered uninformative. Next, we examined the item-profile plots to ensure the emergent profiles were sub-
stantively meaningful (Muthén, 2003). The two-profile plot is discussed in further detail below. The three- and four-profile models
contained profiles that were redundant with the two-profile model. Specifically, the three-profile model contained two profiles with
overlapping means (within 0.3 z-score points) on four of the seven indicators. The four-profile model contained two profiles with
overlapping means on six of the seven indicators and another two profiles with overlapping means on four of the seven indicators.
Moreover, one of the redundant profiles in the three- and four-profile models consisted of a small proportion of the sample (5.7%),
which was further evidence they were not theoretically viable. Thus, we retained the two-profile model.
Fig. 3 presents the item-profile plot of all profiles. Though we conducted LPAs separately with at-risk and not at-risk readers, we
present all profiles on a single plot to foster interpretability and comparisons. Both the separate and combined item-profile plots were
used to interpret and the label the emergent plots. The two profiles of at-risk readers can be seen at the bottom Fig. 3 and are denoted
by solid lines. The profile at the very bottom of the plot scored lowest on all items and is demarcated by circular markers. They scored
nearly two standard deviations below average on five of the seven measures. Thus, we labeled this profile At-Risk Global and they
consisted of 21.0% of the overall sample. The average posterior probability for this profile was 0.94. The profile demarcated by a
solid line with triangle markers performed below average on most measures, but was marked by poor performance on the GORT rate
measures. This profile was labeled At-Risk Fluency and consisted of 27.6% of the overall sample and the average posterior probability
was 0.95.
3.3. Latent profiles for not at-risk readers
We examined models consisting of one to five profiles as the model with six profiles suffered from convergence problems. The
two-profile model was supported by the AWE. The three-profile model was supported by the LMR p-value. However, findings with the
BIC were more nuanced. A minimum BIC value supported the four-profile model, but it has also been shown that the point at which
the BIC displays an “elbow” (i.e., subsequent BIC values indicate diminished returns) may also be indicative of the preferred model
(Nylund et al., 2007). This elbow occurred with the three-profile model.
Having statistical support for the two-, three-, and four-profile models, we proceeded by examining these item-profile plots for
substantive consideration (Muthén, 2003). The two-profile plot consisted of one profile that scored approximately one standard
deviation above average on phonological awareness and decoding variables, but between average and one-half standard deviation
above average on linguistic comprehension and fluency variables. The second profile scored between average and one standard
119
deviation on all variables. The three-profile model was better differentiated in that one profile consistently scored above average, one
profile consistently scored near average, and one profile consistently scored near or below average. The four-profile model included a
profile that was redundant with at least one other profile on six of the seven measures. Thus, the four-profile model was not par-
simonious and was considered substantively untenable. We retained the three-profile model as it had both statistical and substantive
support.
The three profiles are presented in Fig. 3 and are denoted by dashed lines. The profile with square markers scored remarkably
similar across all measures to the At-Risk Low Fluency profile. Though their scores nearly matched the At-Risk Low Fluency profile, we
termed them Not At-Risk Low Fluency to reflect teacher nomination of risk status. However, we emphasize the term “Not At-Risk” is
based on teacher judgment and we elaborate on this further in the Discussion section. This profile was composed of 17.2% of the
overall sample and the average posterior probability was 0.94. The profile demarcated by diamond markers hovered between one-
half of a standard deviation below and above average on all measures. We labeled this profile Average and it consisted of 21.4% of the
students and the average posterior probability was 0.92. Finally, the profile at the top of the plot denoted by X markers scored highest
on all measures. This profile scored near or above one standard deviation on the phonological awareness and decoding measures as
well as above average on linguistic comprehension and fluency measures. We termed this profile Above Average and it consisted of
12.8% of the overall sample with an average posterior probability of 0.97.
3.4. Differentiating profiles by measure
Comparing measures in terms of their ability to differentiate reading profiles can be done by visually examining the distance
between profiles on each measure (see Fig. 3). The phonological awareness and decoding measures exhibited the greatest variation
across profiles. However, the Not At-Risk Low Fluency and At-Risk Low Fluency profiles exhibited nearly identical mean scores across
these sets of measures. The remaining three profiles were well-differentiated by these measures.
The linguistic comprehension measures performed in parallel across profiles. That is, all profiles scored higher, on average, on the
QRI-5 measure than the CELF measure. The At-Risk Global profile scored markedly lower on these two measures than the other four
profiles. The Above Average profile was also clearly delineated from the other profiles by these two linguistic comprehension mea-
sures. However, the Average, Not At-Risk Low Fluency, and At-Risk Low Fluency profiles were not well-differentiated by CELF and QRI-
5.
GORT rate delineated the three not at-risk profiles, but there was virtually no differentiation among the at-risk students.
3.5. Reading comprehension prediction
We included two measures of reading comprehension as distal outcomes of the latent profiles, which allowed us to estimate
profile-specific means. Figs. 4 and 5 summarize the results for WJ Passage Comprehension and GORT-5 Comprehension, respectively.
Results were nearly identical for both reading comprehension outcomes, lending validity to the emergent profiles. In terms of WJ
Passage Comprehension, At-risk Global had the lowest mean (M = 75.37), which was more than one and a half standard deviations
below the national average. Both At-risk Low Fluency (M = 86.61) and Not At-risk Low Fluency (M = 88.63) were within one standard
deviation below the national norm. The Average profile (M = 100.32) scored at the national average and the Above Average profile
(M = 114.80) scored approximately one standard deviation above the national average. The means for the At-Risk Low Fluency and
Not At-Risk Low Fluency profiles were not significantly different (p = 0.25). All other pairwise mean comparisons were significant at
p < 0.001.
A slightly different picture emerged when examining the means of GORT-5 comprehension, though the rank ordering of the
means remained the same. Both the At-risk Global (M = 3.96) and At-risk Low Fluency (M = 4.63) profiles scored approximately two
WJ PC Means
120
114.80
100
100.32
80 86.61 88.63
75.37
60
40
20
0
At-Risk At-Risk Not At-Risk Average Above
Global Fluency Low Fluency Average
Fig. 4. Profile-specific means of WJ-IV Passage Comprehension. All pairwise comparisons were significantly different except for the means of the At-
Risk Fluency and Not At-Risk Fluency profiles (p = 0.25).
120
GORT Comp Means

12
10 11.07
8 8.50
6
5.96
4 4.63
3.96
2
0
At-Risk Global At-Risk Low Not At-Risk Average Above
Fluency Low Fluency Average
Fig. 5. Profile-specific means of GORT Comprehension. All pairwise comparisons were significantly different.
standard deviations below the national norm. The Not At-risk Low Fluency profile (M = 5.96) scored more than one standard deviation
below the national norm. This was in contrast to scores on the WJ Passage Comprehension measure in which both the At-risk Low
Fluency and Not At-risk Low Fluency profiles scored within one standard deviation of the national average. The Average profile
(M = 8.50) scored below the national average, but was within one standard deviation. The Above Average profile (M = 11.07) scored
within one standard deviation above the national norm. All pairwise mean comparisons were significant at p < 0.001, including the
comparison of the At-Risk Low Fluency and Not At-Risk Low Fluency means, which contrasted with the WJ Passage Comprehension
findings.
4. Discussion
4.1. Heterogeneous reading profiles
The goal of this study was to examine whether meaningful subgroups of at-risk and not at-risk readers could be identified in a
sample of first grade readers. Characterizing subgroups of readers using multiple measures may better identify students who require
intervention and subsequently inform intervention design that accounts for students' nuanced skill profiles and better targets in-
dividual needs. This approach may also help educational practitioners identify students who hover near cutoffs for meeting at-risk
criteria. An important feature of this study was identifying these subtypes empirically rather than using somewhat arbitrary cutoff
scores to classify students, as has been used in previous research (Catts et al., 2003; Catts et al., 2006; Nation et al., 2010). Second,
this study compared measures of multiple subcomponent skills of reading comprehension to examine which measures best differ-
entiated the emergent subgroups. Finally, we explored whether the emergent subgroups were consequential in terms of reading
comprehension achievement. Establishing this relation has important implications as it would provide evidence that students at-risk
for reading comprehension difficulties require varied approaches to intervention. In order to answer the study's research questions,
two latent profile analyses were conducted to identify subgroups of at-risk readers and subgroups of not at-risk readers based on
subcomponent skills of reading comprehension. Two subgroups of at-risk readers and three subgroups of typically developing readers
were identified. Next, two measures of reading comprehension as a function of subgroup membership were examined. Categorically
distinct subgroups were identified; however, these subgroups were generally rank-ordered, especially with respect to the phonolo-
gical awareness and decoding variables. While this finding can be interpreted as suggestive of a continuum of skill development, we
argue that pinpointing an individual student's prereading skills along a continuum is less beneficial to applied educators than a
heuristic that can group students with similar proficiencies. This is because, it is possible, that the latter approaches provide a more
useful foundation to design intervention for multiple students simultaneously. We discuss these findings as well as implication for
practices and limitations of the current study below.
A critical component of the MTSS framework is the identification of students who are at-risk for developing later reading diffi-
culties. Early identification is important, as extant data demonstrate the need for intensive, early reading intervention for optimal
outcomes (Francis et al., 1996; Lyon & Fletcher, 2001; Torgesen et al., 2001). This study demonstrates that early identification of at-
risk readers is a difficult task. Teachers were asked to rank their students in terms of reading skills, and only students in the bottom
half of the rankings were administered the screener to determine at-risk status. Yet, a surprising finding emerged, in which the At-Risk
Low Fluency and Not At-Risk Low Fluency profiles bore a striking similarity in terms of achievement across the measures (see Fig. 3).
This suggests that the 17.2% of students who comprised the Not At-Risk Low Fluency profile would likely benefit from reading
intervention, but they were not identified as such by their teachers in the beginning of first grade. That is, even though these students
were not ranked by their teachers as performing in the lower 50% of their class, there are some students who hover near cutoffs for
risk and, while difficult to identify, these students may also be candidates for some amount of intervention. This finding is consistent
with prior research investigating teachers' judgments regarding students' oral reading fluency may not be accurate (Hamilton &
Shinn, 2003; Meisinger, Bradley, Schwanenflugel, Kuhn, & Morris, 2009). These findings support previous researchers who suggest
that universal screening is the best way to identify at-risk readers in the early grades (Glover & Albers, 2007; Jenkins, Hudson, &
Johnson, 2007).
121
In contrast to the two Low Fluency profiles, the subgroup that would likely be most easily identified as at-risk for reading diffi-
culties in first grade was the At-Risk Global profile. Notably, this profile contained one-fifth of our study sample. While it may be
expected that students in this profile would score markedly lower on measures of phonological awareness and word reading, this
profile also exhibited substantial difficulties with linguistic comprehension. Thus, screening procedures might be improved by in-
cluding measures of linguistic comprehension in the early grades. There is an emerging body of evidence that supports the relation
between linguistic comprehension and later reading comprehension (Catts et al., 1999; Kendeou, van den Broek, et al., 2009;
Ouellette & Beers, 2010; Storch & Whitehurst, 2002). Early reading intervention often focuses on word reading skills (Foorman et al.,
1997; Foorman et al., 1998; Rashotte et al., 2001); our results suggest targeting linguistic comprehension skills for the most at-risk
students may also be necessary. Remediating word reading skills may result in short-term gains in reading fluency and reading
comprehension, but students may continue to struggle (Denton et al., 2013) or students might develop reading difficulties later in
school (Catts et al., 2012; Compton et al., 2008) as text becomes more complex. Linguistic comprehension skills have been shown to
be a stronger predictor of reading comprehension in later grades (Hoover & Gough, 1990; Kershaw & Schatschneider, 2012; Storch &
Whitehurst, 2002; Vellutino et al., 2007), so early interventions that include both word reading and linguistic comprehension may be
more beneficial than interventions focused solely on word reading skills with respect to long-term reading comprehension outcomes.
4.2. Most efficient measures to differentiate reading profiles
Given our results concerning the importance of linguistic comprehension, is reading fluency a sufficient measure to be used
unilaterally to predict reading comprehension performance? The GORT rate measure differentiated the three profiles of not at-risk
students, but did not perform well in terms of delineating the students most at-risk. The At-Risk Global, At-Risk Low Fluency, and Not
At-Risk Low Fluency subgroups performed similarly with respect to reading fluency. However, the At-Risk Global profile was most
differentiated from the other two profiles by the other subcomponents of reading comprehension: phonological awareness, decoding,
and linguistic comprehension. Therefore, in first grade, these three skill domains appear better able to distinguish students who
present the greatest risk for reading comprehension difficulties. This is not to suggest that measuring reading fluency in first grade is
not beneficial. The reading fluency measure did reliably distinguish the two highest performing profiles (Average and Above Average)
from the remaining profiles, but identifying such coarse differences may be of limited utility when the focus is on identifying students
in need of intervention. Had reading fluency been used as a unilateral measure to differentiate students, it is unlikely that it would
have distinguished between the three lower performing profiles. With respect to predicting reading comprehension difficulties,
reading fluency may be useful, but other domains, including linguistic comprehension should also be considered.
The measures with the greatest variation and, thus, the greatest potential to differentiate first grade readers were the phonological
awareness and decoding measures. This is consistent with prior empirical research that has demonstrated the link between these skills
and reading comprehension in early elementary (Chen & Velluntino, 1997; Georgiou et al., 2009; Hoover & Gough, 1990; Joshi &
Aaron, 2000). However, as noted above, there was a surprising similarity between the At-Risk Low Fluency and Not At-Risk Low
Fluency profiles on all phonological awareness and decoding variables. Even though the phonological awareness and decoding
measures demonstrated the most potential to differentiate students, these measures essentially were unable to delineate these pro-
files. Together, these profiles accounted for approximately 45% of the sample. This is a sizeable number of students who appear to be
performing slightly below average on most of the phonological awareness and decoding measures. As such, these students may be
thought to be on the cusp for experiencing future reading difficulties. This suggests that students who perform below average, but
perhaps not overtly so, on phonological awareness and decoding measures should be referred for screening if universal screening is
not available. It is possible that both profiles of students would benefit from Tier 2 support that would alleviate future reading
difficulties.
The CELF and QRI-5 were used to measure linguistic comprehension. Overall, there was less variation in these measures across the
profiles compared to both the phonological awareness and decoding measures, but this is consistent with the Simple View with
respect to early elementary students (e.g., Hoover & Gough, 1990; Kendeou, van den Broek, et al., 2009; Vellutino et al., 2007). Thus,
it was not surprising these two measures did not differentiate the three middle profiles in Fig. 3. However, both measures clearly
delineated the At-Risk Global profile from the remaining profiles. This suggests these linguistic comprehension measures may be used
in conjunction with the phonological awareness and decoding measures when the aim is to identify students with the greatest need
for intervention.
The CELF and QRI-5 performed in parallel across the profiles. That is, every profile scored higher on the QRI-5 than the CELF. This
was somewhat surprising as the CELF is a normed measure whereas the QRI-5 is not. Since these results were consistent across
profiles, these measures may be considered to be providing similar information regarding students' linguistic comprehension skills.
This is a promising finding as the QRI-5 is a commonly used assessment by teachers and school psychologists. A drawback, however,
is that administering a single early elementary QRI-5 story – as was done in this study – yields six comprehension questions compared
to administering three passages from the CELF Understanding Spoken Paragraphs subtest, which contains 15 comprehension ques-
tions. Fewer questions may make it more difficult for teachers and school psychologists to interpret QRI-5 results as there is an
increased potential for at-risk students' scores to be artificially inflated by capitalizing on chance. This may explain why both the at-
risk and not at-risk subgroups had the same range of scores (see Table 1) on the QRI-5. While the range of scores on the CELF was
similar for both subgroups, there was greater variation in the means of each subgroup compared to the QRI-5 means. Moreover, since
the CELF is normed, results are more readily interpreted and may be more easily communicated to parents and other education
professionals.
122
4.3. Predicting reading comprehension achievement
Profile membership was used to predict two measures of reading comprehension, the passage comprehension subtest of the WJ-IV
and the comprehension subscale of the GORT. Generally, results were as might be expected intuitively. The Above Average profile,
which performed the highest on all measures, also achieved the highest means on both reading comprehension measures, while the
At-Risk Global profile, which performed lowest on all measures, achieved the lowest reading comprehension means. The reading
comprehension means of the remaining profiles were also aligned with the general rank ordering of achievement patterns across the
measures as depicted in Fig. 3. These intuitive findings lend validity to the emergent profiles and support the notion of categorically
distinct latent profiles as a useful heuristic for school practitioners seeking to identify readers in need of intervention.
The reading comprehension results showed significant differences for the vast majority of the pairwise mean comparisons be-
tween profiles. All of the profile-specific means of the GORT comprehension subscale were significantly different from each other.
However, the WJ-IV passage comprehension means of the At-Risk Low Fluency and Not At-Risk Low Fluency profiles were not sig-
nificantly different from each other. Additionally, the WJ-IV passage comprehension means of these two profiles were both within
one standard deviation of the national norm, while the GORT comprehension means of these two profiles were both more than two
standard deviations below the national norm. This may reflect differences between the formats of the two measures. The WJ-IV
passage comprehension subtest presents a rebus and cloze format while the GORT comprehension subtest utilizes open-ended
questions read by the examiner. Additionally, the stopping criterion for the GORT is met when a student is unable to read the passage
fluently, whereas the stopping criterion for the WJ-IV passage comprehension is met when a student misses six consecutive items.
Differences in format may indicate these two reading comprehension tasks rely on reading sub-skills in different ways. Keenan,
Betjemann, and Olson (2008) examined prior versions of these subtests and found only a modest correlation. Moreover, their study
demonstrated the WJ passage comprehension subtest was more strongly related to decoding than the GORT, and suggested this may
be because the WJ consists of shorter passages. This is also true of the newer versions utilized in this study, so their rationale may also
explain the differences found here. Finally, and perhaps, more importantly, Keenan et al. (2008) found the relation between decoding
and reading comprehension for the WJ passage comprehension subtest was dependent on the level of reading development with the
effect being stronger for younger and less skilled readers. Therefore, the non-significant difference in the WJ-IV passage compre-
hension means between the At-Risk Fluency and Not At-Risk Low Fluency profiles may reflect the fact that their decoding skills were
strikingly similar. Keenan et al. (2008) found the GORT reading comprehension was dependent on both decoding and linguistic
comprehension. In the present study, the significant differences between the At-Risk Fluency and Not At-Risk Low Fluency profiles on
GORT reading comprehension may simply stem from the fact that multiple measures across multiple reading sub-skills were utilized,
which could have enhanced the profile-specific differences on the GORT.
These results may also be related to the difficulty in identifying at-risk students who hover near the middle range of reading
achievement. As discussed earlier, 17.2% of our sample (the Not At-Risk Low Fluency profile) was deemed not at-risk by teachers,
though our analyses demonstrate their reading skill profile was almost identical to a subgroup of 27.6% of our sample the teachers
deemed at-risk (the At-Risk Low Fluency profile). This speaks to the importance of universal screening to identify students who appear
to be on the cusp of requiring reading intervention. Further, Tier 2 intervention might provide a greater benefit to students on this
cusp than general curricula since at-risk status is difficult to discern. If educators wish to be cautious, then it would behoove them to
refer all of the middle-performing students for further screening.
4.4. Implications for practice
The results of this study suggest screening procedures used in first grade should begin with phonological awareness and decoding
measures as these exhibited the greatest variation. This is consistent with literature exploring the utility of gated screening with this
age group (Clemens, Shapiro, & Thoemmes, 2011; Compton et al., 2010). Gated screening involves first administering an efficient,
brief assessment to all students followed by a more comprehensive battery of assessments to those who may be at-risk. Results from
this study suggest phonological awareness and decoding assessments are the first screenings that should be conducted. Since the At-
Risk Low Fluency profile demonstrated concomitant weaknesses in linguistic comprehension, these assessments should be included in
a secondary battery. Since the At-Risk Low Fluency profile exhibited weaknesses in both areas, it would seem phonological awareness
and decoding assessments would be sufficient to identify them. However, extant literature has identified profiles of readers who
demonstrated adequate decoding skills, but experienced difficulty with linguistic comprehension (e.g., Catts et al., 2006; Nation
et al., 2010). Though this study was unable to identify a comparable group of students, prior findings would suggest best practice
would include a screening measure of linguistic comprehension in addition to decoding. For instance, Compton et al. (2010) included
a measure of oral vocabulary in their gated screening procedures and found it was informative. Restricting interventions to word
reading skills leaves linguistic comprehension skills – a critical subcomponent of reading comprehension in later grades – un-
addressed, which may lead to the re-emergence of reading comprehension difficulties as students progress through school.
Teachers often assess oral reading fluency as an indicator of reading progress. This study supports this practice, but only in terms
of examining coarse differences among students. That is, the reading fluency measure used in this study delineated the two highest
performing profiles from the three lower performing profiles. It did a much poorer job of distinguishing among students in the three
lower performing profiles, especially when compared to the other measures used in this study. It is likely that these students' word
reading skills were not developed enough to allow for much variation in their oral reading fluency skills. Thus, teachers and other
practitioners may be measuring oral reading fluency before it is able to provide useful information. For instance, Silverman et al.
(2013) found oral reading fluency mediated the relation between decoding and reading comprehension, but in a sample of fourth
123
grade students. Moreover, prior studies of first grade reading screening have shown fluency in identifying individual words is a more
effective measure than fluency in reading connected text when identifying first grade struggling readers (Clemens et al., 2011;
Compton et al., 2010). This does not suggest oral reading fluency should not be a target of early intervention; indeed, we advocate for
its inclusion as a component of reading intervention. However, basing instructional and intervention decisions on oral reading
fluency may be fruitless, or worse, misguided with respect to at-risk first graders. A better strategy would be to tailor intervention
based on word reading and linguistic comprehension skills until word reading enables oral reading fluency to reach a proficiency
level that is informative.
4.5. Limitations
LPA is generally regarded as a large-sample technique and the size of each subsample in this study borders on the lower end of
what is usually considered acceptable. However, given that there were only a small number of profiles identified within each sub-
sample, and that each profile had a high average posterior probability (all > 0.92), there is evidence of a strong signal in the data
supporting these findings. Nevertheless, future studies should examine whether these findings can be replicated in other samples of
early first grade students, accounting for demographic characteristics. Another limitation attributable to sample size is this study did
not identify a profile of readers who exhibited adequate decoding skills, but poor linguistic comprehension, which has been found in
prior studies (Catts et al., 2006; Nation et al., 2010). Larger numbers of each subsample may have allowed this type of profile to be
identified.
Second, accurately measuring reading comprehension in first grade is a difficult task, as evidenced by the inconsistent results
concerning the WJ and GORT comprehension measures. While the rank ordering of profile-specific reading comprehension scores
aligned was consistent, statistically significant differences among the profiles were not. Thus, these results may be more informative
for the design of interventions targeting reading sub-skills than reading comprehension directly.
Third, it is unclear how the results might have changed if a different sampling procedure was used. For example, the not at-risk
sample did not include students who were ranked in the lower 50% of each classroom, but passed the screening. If these students
were re-included in the not at-risk sample, the nature and proportions of the not at-risk profiles might have been different. Perhaps
the Not At-Risk Low Fluency might have contained a larger proportion of the not at-risk students. Additionally, screening all students
might have yielded a different at-risk sample compared to screening only students ranked by teachers in the lower 50% of their
classrooms in terms of reading skills. Furthermore, students were only included in the at-risk sample if they failed all of the three
screening measures. It is possible that a more liberal benchmark for defining risk – such as including all students who did not pass any
one of the first three screening measures – could have resulted in a different at-risk sample, which could have yielded different
profiles. While it is impossible to know exactly how these decisions might have impacted the results, it is likely that the changes listed
above would have mostly affected the At-Risk Low Fluency and Not At-Risk Low Fluency profiles as those profiles exhibited the greatest
overlap.
5. Conclusions
Prior research has often defined risk status by utilizing relatively arbitrary cutoff scores on individual measures or combinations of
them. Recent advances in statistical modeling now enable researchers to empirically classify students based on multivariate response
patterns. This study identified distinct profiles of at-risk and not at-risk students in the fall of first grade. However, this study was
cross-sectional and it is not clear if these profiles would emerge in other grades or even at later points in first grade. Furthermore,
while relations between the emergent profiles and reading comprehension were apparent, we do not know if these relations would be
stable over time, especially if the latent profiles themselves are not stable. Further research in these areas is warranted.
Identifying at-risk readers in early elementary is a difficult task, and this study suggests teacher judgment is not sufficient. This
supports universal screening, but the measures used in these procedures should be carefully examined for their utility. We found oral
reading fluency to be least useful with this sample, but this may not apply to other samples, ages, or grades. Linguistic comprehension
was identified as a critical risk factor for reading difficulties and this should be a target of intervention in early elementary. Little is
known about the long-term effects of early linguistic comprehension on later reading comprehension, but this is a promising area of
study that deserves further attention and may enable educational practitioners to develop better reading interventions.
References
Adlof, S. M., Catts, H. W., & Lee, J. (2010). Kindergarten predictors of second versus eighth grade reading comprehension impairments. Journal of Learning Disabilities,
43, 332–345. http://dx.doi.org/10.1177/0022219410369067.
Adlof, S. M., Catts, H. W., & Little, T. D. (2006). Should the simple view of reading include a fluency component? Reading and Writing, 19, 933–958. http://dx.doi.org/
10.1007/s11145-006.
Asparouhov, T., & Muthén, B. (2014). Auxiliary variables in mixture modeling: Using the BCH method in Mplus to estimate a distal outcome model and an arbitrary
secondary model (Mplus Web Note: No. 21). Retrieved from: http://www.statmodel.com/download/asparouhov_muthen_2014.pdf.
Badian, N. A. (1999). Reading disability defined as a discrepancy between listening and reading comprehension. Journal of Learning Disabilities, 32, 138–148. http://dx.
doi.org/10.1177/002221949903200204.
Bakk, Z., & Vermunt, J. K. (2016). Robustness of stepwise latent class modeling with continuous distal outcomes. Structural Equation Modeling: An Interdisciplinary
Journal, 23(1), 20–31. http://dx.doi.org/10.1080/10705511.2014.955104.
Begeny, J. C., Eckert, T. L., Montarello, S. A., & Storie, M. S. (2008). Teachers' perceptions of students' reading abilities: An examination of the relationship between
teachers' judgments and students' performance across a continuum of rating methods. School Psychology Quarterly, 23, 43–55. http://dx.doi.org/10.1037/1045-
124
3830.23.1.43.
Bolck, A., Croon, M., & Hagenaars, J. (2004). Estimating latent structure models with categorical variables: One-step versus three-step estimators. Political Analysis, 12,
3–27. http://dx.doi.org/10.1093/pan/mph001.
Bowyer-Crane, C., Snowling, M. J., Duff, F. J., Fieldsend, E., Carroll, J. M., Miles, J., ... Hulme, C. (2008). Improving early language and literacy skills: Differential
effects of an oral language versus a phonology with reading intervention. Journal of Child Psychology and Psychiatry, 49, 422–432. http://dx.doi.org/10.1111/j.
1469-7610.2007.01849.x.
Cain, K., & Oakhill, J. (2011). Matthew effects in young readers: Reading comprehension and reading experience aid vocabulary development. Journal of Learning
Disabilities, 44, 431–443. http://dx.doi.org/10.1177/002221941140042.
Catts, H. W., Adlof, S. M., & Weismer, S. E. (2006). Language deficits in poor comprehenders: A case for the simple view of reading. Journal of Speech, Language, and
Hearing Research, 49, 278–293. http://dx.doi.org/10.1044/1092-4388(2006/023).
Catts, H. W., Compton, D., Tomblin, J. B., & Bridges, M. S. (2012). Prevalence and nature of late-emerging poor readers. Journal of Educational Psychology, 104,
166–181. http://dx.doi.org/10.1037/a0025323.
Catts, H. W., Fey, M. E., Zhang, X., & Tomblin, B. (1999). Language basis of reading and reading disabilities: Evidence from a longitudinal investigation. Scientific
Studies of Reading, 3, 331–361. http://dx.doi.org/10.1207/s1532799xssr0304_2.
Catts, H. W., Hogan, T. P., & Fey, M. E. (2003). Subgrouping poor readers on the basis of individual differences in reading-related abilities. Journal of Learning
Chall, J. S. (1983). Learning to read: The great debate. New York: McGraw-Hill.
Chard, D. J., Vaughn, S., & Tyler, B. J. (2002). A synthesis of research on effective interventions for building reading fluency with elementary students with learning
disabilities. Journal of Learning Disabilities, 35(5), 386–406. http://dx.doi.org/10.1177/00222194020350050101.
Chen, R. S., & Velluntino, F. R. (1997). Prediction of reading ability: A cross-validation study of the simple view of reading. Journal of Literacy Research, 29, 1–24.
http://dx.doi.org/10.1080/10862969709547947.
Clarke, P. J., Snowling, M. J., Truelove, E., & Hulme, C. (2010). Ameliorating children's reading-comprehension difficulties: A randomized controlled trial.
Psychological Science, 21, 1106–1116. http://dx.doi.org/10.1177/0956797610375449.
Clemens, N. H., Shapiro, E. S., & Thoemmes, F. (2011). Improving the efficacy of first grade reading screening: An investigation of word identification fluency with
other early literacy indicators. School Psychology Quarterly, 26, 231–244. http://dx.doi.org/10.1037/a0025173.
Compton, D. L., Fuchs, D., Fuchs, L. S., Bouton, B., Gilbert, J. K., Barquero, L. A., ... Crouch, R. C. (2010). Selecting at-risk first-grade readers for early intervention:
Eliminating false positives and exploring the promise of a two-stage gated screening process. Journal of Educational Psychology, 102, 327–340. http://dx.doi.org/
10.1037/a0018448.
Compton, D. L., Fuchs, D., Fuchs, L. S., Elleman, A. M., & Gilbert, J. K. (2008). Tracking children who fly below the radar: Latent transition modeling of students with
late-emerging reading disability. Learning and Individual Differences, 18, 329–337. http://dx.doi.org/10.1016/j.lindif.2008.04.003.
Denton, C. A., Nimon, K., Mathes, P. G., Swanson, E. A., Kethley, C., Kurz, T. B., & Shih, M. (2010). Effectiveness of a supplemental early reading intervention scaled up
in multiple schools. Exceptional Children, 76(4), 394–416. http://dx.doi.org/10.1177/001440291007600402.
Denton, C. A., Tolar, T. D., Fletcher, J. M., Barth, A. E., Vaughn, S., & Francis, D. J. (2013). Effects of tier 3 intervention for students with persistent reading difficulties
and characteristics of inadequate responders. Journal of Educational Psychology, 105, 633–648. http://dx.doi.org/10.1037/a0032581.
Fletcher, J. M., & Vaughn, S. (2009). Response to intervention: Preventing and remediating academic difficulties. Child Development Perspectives, 3, 30–37. http://dx.
doi.org/10.1111/j.1750-8606.2008.00072.x.
Foorman, B. R., Francis, D. J., Fletcher, J. M., Schatschneider, C., & Mehta, P. (1998). The role of instruction in learning to read: Preventing reading failure in at-risk
children. Journal of Educational Psychology, 90, 37–55. http://dx.doi.org/10.1037//0022-0663.90.1.37.
Foorman, B. R., Francis, D. J., Shaywitz, S. E., Shaywitz, B. A., & Fletcher, J. M. (1997). The case for early reading intervention. In B. Blachman (Ed.). Foundations of
reading acquisition and dyslexia: Implications for early intervention (pp. 243–264). Mahwah, NJ, US: Lawrence Erlbaum Associates Publishers.
Francis, D. J., Fletcher, J. M., Stuebing, K. K., Lyon, G. R., Shaywitz, B. A., & Shaywitz, S. E. (2005). Psychometric approaches to the identification of LD: IQ and
achievement scores are not sufficient. Journal of Learning Disabilities, 38, 98–108. http://dx.doi.org/10.1177/00222194050380020101.
Francis, D. J., Shaywitz, S. E., Stuebing, K. K., Shaywitz, B. A., & Fletcher, J. M. (1996). Developmental lag versus deficit models of reading disability: A longitudinal,
individual growth curves analysis. Journal of Educational Psychology, 88, 3–17. http://dx.doi.org/10.1037//0022-0663.88.1.3.
Fuchs, D., & Fuchs, L. S. (2006). Introduction to response to intervention: What, why, and how valid is it? Reading Research Quarterly, 41, 93–99. http://dx.doi.org/10.
1598/rrq.41.1.4.
Fuchs, D., Mock, D., Morgan, P. L., & Young, C. L. (2003). Responsiveness-to-intervention: Definitions, evidence, and implications for the learning disabilities con-
struct. Learning Disabilities Research and Practice, 18, 157–171. http://dx.doi.org/10.1111/1540-5826.00072.
Fuchs, L. S., Fuchs, D., Hosp, M. K., & Jenkins, J. R. (2001). Oral reading fluency as an indicator of reading competence: A theoretical, empirical, and historical
analysis. Scientific Studies of Reading, 5, 239–256. http://dx.doi.org/10.1207/s1532799xssr0503_3.
Fuchs, L. S., Fuchs, D., & Maxwell, L. (1988). The validity of informal measures of reading comprehension. Remedial and Special Education, 9, 20–28. http://dx.doi.org/
10.1177/074193258800900206.
Georgiou, G. K., Das, J. P., & Hayward, D. (2009). Revisiting the “Simple View of Reading” in a group of children with poor reading comprehension. Journal of Learning
Gerber, M. M. (2005). Teachers are still the test: Limitations of response to instruction strategies for identifying children with learning disabilities. Journal of Learning
Gibson, W. A. (1959). Three multivariate models: Factor analysis, latent structure analysis, and latent profile analysis. Psychometrika, 24, 229–252.
Glover, T. A., & Albers, C. A. (2007). Considerations for evaluating universal screening assessments. Journal of School Psychology, 45, 117–135. http://dx.doi.org/10.
1016/j.jsp.2006.05.005.
Gough, P. B., & Tunmer, W. E. (1986). Decoding, reading, and reading disability. Remedial and Special Education, 7, 6–10. http://dx.doi.org/10.1177/
074193258600700104.
Hamilton, C., & Shinn, M. (2003). Characteristics of word callers: An investigation of the accuracy of teachers' judgments of reading comprehension and oral reading
skills. School Psychology Review, 32, 228–240.
Hoover, W. A., & Gough, P. B. (1990). The simple view of reading. Reading and Writing: An Interdisciplinary Journal, 2, 127–160. http://dx.doi.org/10.1007/
bf00401799.
Jenkins, J. R., Fuchs, L. S., van den Broek, P., Espin, C., & Deno, S. L. (2003). Sources of individual differences in reading comprehension and reading fluency. Journal
of Educational Psychology, 95, 719–729. http://dx.doi.org/10.1037/0022-0663.95.4.719.
Jenkins, J. R., Hudson, R. F., & Johnson, E. S. (2007). Screening for at-risk readers in a response to intervention framework. School Psychology Review, 36, 582–600.
Joshi, R. M., & Aaron, P. G. (2000). The component model of reading: Simple view of reading made a little more complex. Reading Psychology, 21, 85–97. http://dx.doi.
org/10.1080/02702710050084428.
Keenan, J. M., Betjemann, R. S., & Olson, R. K. (2008). Reading comprehension tests vary in the skills they assess: Differential dependence on decoding and oral
comprehension. Scientific Studies of Reading, 12, 281–300. http://dx.doi.org/10.1080/10888430802132279.
Kendeou, P., Savage, R., & van den Broek, P. (2009). Revisiting the simple view of reading. British Journal of Educational Psychology, 79, 353–370. http://dx.doi.org/10.
1348/978185408X369020.
Kendeou, P., van den Broek, P., White, M. J., & Lynch, J. S. (2009). Predicting reading comprehension in early elementary school: The independent contributions of
oral language and decoding skills. Journal of Educational Psychology, 101, 765–778. http://dx.doi.org/10.1037/a0015956.
Kershaw, S., & Schatschneider, C. (2012). A latent variable approach to the simple view of reading. Reading and Writing, 25, 433–464. http://dx.doi.org/10.1007/
s11145-010-9278-3.
Kim, Y. S., Petscher, Y., Schatschneider, C., & Foorman, B. (2010). Does growth rate in oral reading fluency matter in predicting reading comprehension achievement?
125
Journal of Educational Psychology, 102, 652–667. http://dx.doi.org/10.1037/a0019643.

LaBerge, D., & Samuels, S. (1974). Toward a theory of automatic information processing in reading. Cognitive Psychology, 6, 293–323. http://dx.doi.org/10.1016/
0010-0285(74)90015-2.
Lazarsfeld, P. F., & Henry, N. W. (1968). Latent structure analysis. Boston, MA: Houghton Mifflin.
Leach, J. M., Scarborough, H. S., & Rescorla, L. (2003). Late-emerging reading disabilities. Journal of Educational Psychology, 95, 211–224. http://dx.doi.org/10.1037/
0022-0663.95.2.211.
Leslie, L., & Caldwell, J. (2011). Qualitative reading inventory-V (QRI-V). Boston, MA: Pearson.
Lyon, G. R., & Fletcher, J. M. (2001). Early identification, prevention and early intervention for children at-risk for reading failure. Publication of the Council for Basic
Education.
Masyn, K. (2013). Latent class analysis and finite mixture modeling. In T. Little (Vol. Ed.), The Oxford handbook of quantitative methods in psychology. Vol. 2. The Oxford
handbook of quantitative methods in psychology (pp. 375–393). Oxford, UK: Oxford University Press.
Mathes, P. G., Denton, C. A., Fletcher, J. M., Anthony, J. L., Francis, D. J., & Schatschneider, C. (2005). The effects of theoretically different instruction and student
characteristics on the skills of struggling readers. Reading Research Quarterly, 40(2), 148–182. http://dx.doi.org/10.1598/rrq.40.2.2.
McLachlan, G., & Peel, D. (2004). Finite mixture models. John Wiley & Sons.
Meisinger, E. B., Bradley, B. A., Schwanenflugel, P. J., Kuhn, M. R., & Morris, R. D. (2009). Myth and reality of the word caller: The relation between teacher
nominations and prevalence among elementary school children. School Psychology Quarterly, 24, 147.
Muthén, B. (2003). Statistical and substantive checking in growth mixture modeling: Comment on Bauer and Curran (2003). Psychological Methods, 8, 369–377. http://
dx.doi.org/10.1037/1082-989X.8.3.369.
Nation, K., Adams, J. W., Bowyer-Crane, C. A., & Snowling, M. J. (1999). Working memory deficits in poor comprehenders reflect underlying language impairments.
Journal of Experimental Child Psychology, 73, 139–158. http://dx.doi.org/10.1006/jecp.1999.2498.
Nation, K., Cocksey, J., Taylor, J. S. H., & Bishop, D. V. M. (2010). A longitudinal investigation of early reading and language skills in children with poor reading
comprehension. The Journal of Child Psychology and Psychiatry, 51, 1031–1039. http://dx.doi.org/10.1111/j.1469-7610.2010.0225.x.
Nylund, K. L., Asparouhov, T., & Muthén, B. O. (2007). Deciding on the number of classes in latent class analysis and growth mixture modeling: A Monte Carlo
simulation study. Structural Equation Modeling, 14, 535–569. http://dx.doi.org/10.1080/10705510701575396.
Ouellette, G., & Beers, A. (2010). A not-so-simple view of reading: How oral vocabulary and visual-word recognition complicate the story. Reading and Writing, 23,
189–208. http://dx.doi.org/10.1007/s11145-008-9159-1.
Pikulski, J. J., & Chard, D. J. (2005). Fluency: Bridge between decoding and reading comprehension. The Reading Teacher, 58(6), 510–519. http://dx.doi.org/10.1598/
rt.58.6.2.
Rashotte, C. A., MacPhee, K., & Torgesen, J. K. (2001). The effectiveness of a group reading instruction program with poor readers in multiple grades. Learning
Disability Quarterly, 24, 119–134. http://dx.doi.org/10.2307/1511068.
Riedel, B. W. (2007). The relation between DIBELS, reading comprehension, and vocabulary in urban first-grade students. Reading Research Quarterly, 42, 546–567.
http://dx.doi.org/10.1598/RRQ.42.4.5.
Roth, F. P., Speece, D. L., & Cooper, D. H. (2002). A longitudinal analysis of the connection between oral language and early reading. The Journal of Educational
Research, 95, 259–272. http://dx.doi.org/10.1080/00220670209596600.
Schrank, F. A., Mather, N., & McGrew, K. S. (2014). Woodcock-Johnson IV tests of achievement. Rolling Meadows, IL: Riverside.
Semel, E. M., Wiig, E. H., & Secord, W. (2006). CELF 4: Clinical evaluation of language fundamentals. Pearson: Psychological Corporation.
Silverman, R. D., Speece, D. L., Harring, J. R., & Ritchey, K. D. (2013). Fluency has a role in the simple view of reading. Scientific Studies of Reading, 17, 108–133.
http://dx.doi.org/10.1080/10888438.2011.618153.
Storch, S. A., & Whitehurst, G. J. (2002). Oral language and code-related precursors to reading: Evidence from a longitudinal structural model. Developmental
Psychology, 38, 934–947. http://dx.doi.org/10.1037//0012-1649.38.6.934.
Torgesen, J. K., Alexander, A. W., Wagner, R. K., Rashotte, C. A., Voeller, 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. http://dx.doi.org/10.1097/
00004703-200108000-00033.
Torgesen, J. K., Wagner, R., & Rashotte, C. A. (1999). CTOPP comprehensive test of phonological processing. Austin, TX: Pro-Ed.
Tunmer, W. E., & Chapman, J. W. (2012). The simple view of reading redux: Vocabulary knowledge and the independent components hypothesis. Journal of Learning
Vellutino, F. R., Tunmer, W. E., Jaccard, J. J., & Chen, R. (2007). Components of reading ability: Multivariate evidence for a convergent skills model of reading
development. Scientific Studies of Reading, 11, 3–32. http://dx.doi.org/10.1080/10888430709336632.
Vermunt, J. K. (2010). Latent class modeling with covariates: Two improved three-step approaches. Political Analysis, 18, 450–469. http://dx.doi.org/10.1093/pan/
mpq025.
Wagner, R. K., & Torgesen, J. K. (1987). The nature of phonological processing and its causal role in the acquisition of reading skills. Psychological Bulletin, 101,
192–212.
Wagner, R. K., Torgesen, J. K., Rashotte, C. A., Hecht, S. A., Barker, T. A., Burgess, S. R., ... Garon, T. (1997). Changing relations between phonological processing
abilities and word-level reading as children develop from beginning to skilled readers: A 5 year longitudinal study. Developmental Psychology, 33, 468–479. http://
dx.doi.org/10.1037//0012-1649.33.3.468.
Wiederholt, J. L., & Bryant, B. R. (2012). Gray Oral Reading Test–Fifth Edition (GORT-5): Examiner's manual. Austin, TX: Pro-Ed.
126
View publication stats

Grimmetal Risknotatriskprofiles 2018

Uploaded by

Copyright:

Available Formats

Grimmetal Risknotatriskprofiles 2018

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Grimmetal Risknotatriskprofiles 2018

Uploaded by

Copyright:

Available Formats

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

Article in Journal of School Psychology · August 2018

Ryan Grimm Nancy S. McIntyre

SEE PROFILE SEE PROFILE

The user has requested enhancement of the downloaded file.

Contents lists available at ScienceDirect

Journal of School Psychology

Early reading skill proﬁles in typically developing and at-risk ﬁrst

1.1. Phonological processing

1.2. Decoding and reading ﬂuency

1.3. Linguistic comprehension

1.4. Diﬀerentiating readers according to reading sub-skills

1.4.1. Use of latent proﬁle analysis in the present study

Elision Blending LWID WA CELF QRI G Rate

1.4.2. Present study

2.2. Determining at-risk status

Teachers rank students in their classrooms (n = 767)

Lower-performing 50% of students Not at-risk students randomly selected

Pass ﬁrst 3 Fail any of ﬁrst 3

Disconnue screening. Administer word

Risk deﬁned by both

2.4.1. At-risk screen

2.4.2. Phonological awareness

2.4.4. Linguistic comprehension

2.4.5. Oral reading ﬂuency

2.4.6. Reading comprehension

2.5. Data analysis plan

Measure n M SD Min Max n M SD Min Max t

3.1. Descriptive statistics

At-Risk Global (43.6%/21.0%) At-Risk Low Fluency (56.4%/27.6%)

3.2. Latent proﬁles for at-risk readers

3.3. Latent proﬁles for not at-risk readers

3.4. Diﬀerentiating proﬁles by measure

3.5. Reading comprehension prediction

GORT Comp Means

4.1. Heterogeneous reading proﬁles

4.2. Most efficient measures to differentiate reading profiles

4.3. Predicting reading comprehension achievement

4.4. Implications for practice

Journal of Educational Psychology, 102, 652–667. http://dx.doi.org/10.1037/a0019643.

View publication stats

You might also like