Skip to main content
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2022 Apr 1.
Published in final edited form as: J Educ Psychol. 2020 Sep 17;113(3):462–476. doi: 10.1037/edu0000635

Differential Patterns of Growth in Reading and Math Skills during Elementary School

Callie W Little 1, Christopher J Lonigan 2,3, Beth M Phillips 3,4
PMCID: PMC8130866  NIHMSID: NIHMS1696524  PMID: 34017147

Abstract

This study investigated developmental trajectories of reading and math using latent-growth-curve analyses across multiple academic skills, measures, and multiple time periods within a single sample. Reading-related growth was marked by significant individual differences during the early elementary-school period and non-significant individual differences during the late elementary-school period. For math-related skills, non-significant individual differences were present for early math growth and significant individual differences were present in late elementary-school. No clear pattern of cumulative, compensatory, or stable development emerged for either reading-related or math skills. These differing growth patterns highlight developmental complexities and suggest domain-specific differences in achievement growth that are potentially associated with contextual factors.

Keywords: reading, math, development, elementary school

Educational Impact And Implications Statement

Our paper addresses the importance of developmental processes in reading and math. We took advantage of a large longitudinal data set to compare the similarity of developmental patterns across several reading and math skills. We found that individual differences in growth rates were greater in the early elementary school years (K-2) for reading than for math, but in later elementary school (3rd-5th grades) individual differences in growth were greater for math than for reading. Our results suggest that domain-specific contextual factors may influence these developmental patterns and these differences could have important implications for the manipulation of associated contextual factors and future intervention development.

Differential Patterns of Growth in Reading and Math Skills during Elementary School

Reading and math skills develop rapidly during the elementary school years (e.g. Bast & Reitsma, 1998; Little, Hart, Quinn, Tucker-Drob, Taylor & Schatschneider, 2016; Bodovsky & Farkas, 2007); however, less is understood about the developmental patterns of change in these skills during this time period (e.g. Parrila, Aunola, Leskinen, Nurmi, & Kirby, 2005; Pfost, Hattie, Dörfler, & Artelt, 2014; Bailey, Watts, Littlefield, & Geary, 2014). In particular, longitudinal studies of reading and math development have yielded mixed evidence for whether inter-individual developmental patterns are stable or vary from early to late elementary school. Additionally, few studies have examined development in multiple academic skills simultaneously, which leaves a gap in our knowledge of whether developmental trajectories are common across academic domains or demonstrate domain-specific patterns. By examining the developmental patterns of several reading and math skills within a single sample in this study, we were able to control for sample-based characteristics while examining whether development displays similar patterns across academic skills and domains, or if there are differences between reading and math development. This information can guide learning support mechanisms that either combine across academic outcomes or are targeted to individual academic domains and skills.

The development of reading and math skills has been found to follow predictable stages that progress from early childhood through elementary school and beyond. The stages of reading development under Chall’s widely accepted theory include two major phases: “learning to read” followed by “reading to learn.” These broader phases consist of several smaller substages. During stages 0 to 2, the pre-reading phase of development, children develop and practice print structure knowledge, basic rules of language, word decoding skills and reading fluency. In stage 3 children transition from the “learning to read” phase to the “reading to learn” phase by developing mastery of fluent reading skills and learning to integrate knowledge and information from what they read. In stages 4 and 5, children increasingly incorporate comprehension strategies to better understand complex ideas (Chall, 1983).

Math development involves building a series of skills such as number knowledge, spatial reasoning, and patterning knowledge. Number knowledge develops in two broad phases, from informal to formal math knowledge (Clements & Sarama, 2009). Early mathematics, or informal math skills, occurs over three stages during which children progress from non-numeric quantity discrimination (e.g. comparing piles of jelly beans), quantity to number-word linkage (e.g. counting the number of jelly beans in a pile), and then to symbolic and non-symbolic quantity relations (e.g. adding or subtracting jelly beans across piles; Krajewski & Schneider, 2009; Siegler & Lortie-Forgues, 2014) beginning prior to school entry and continuing until from around first grade to third grade where more formal math skills begin to develop (e.g., Clements & Sarama, 2009). Formal math knowledge involves the mastery of number competencies, which include number recognition, number comparisons, and understanding number magnitudes (e.g. Mazzocco & Thompson, 2005; Jordan et al., 2009; Merckley & Ansari, 2016). As these comparisons and magnitudes extend to increasingly larger numbers and number arrays, and non-symbolic concepts are mapped onto increasingly more complex symbolic systems, mastery of formal math knowledge progresses (e.g. Lyons & Ansari, 2015; Siegler & Lortie-Forgues, 2014).

Although reading and math skills develop rapidly through elementary school, their phases of development occur along different timelines and appear to depend on both common and unique cognitive and environmental factors (e.g. Daucourt et al., 2020; Landerl et al., 2009). Socio-educational background, genetics, and academic self-concept are likely to influence both reading and math skills in similar ways (e.g. Sirin 2005; Jacobs et al., 2002). However, acquisition of academic skills may be differentially influenced by foundational cognitive skills, such as decoding and phonological processing for reading comprehension (Juel, 1988; Schatschneider et al., 2004), domain-specific number knowledge, and the bidirectional influences of conceptual and procedural knowledge for math skills (e.g. Rittle-Johnson, 2017). Furthermore, contextual factors such as instructional differences (e.g. Bodovski & Farkas, 2007; Jordan et al., 2009) might influence reading and math development. These varying influences may lead to multiple developmental trajectories, whereby some children during early elementary school experience higher rates of growth that slow over time and others have initially slower rates of growth that gradually increase across the elementary years.

From the existing evidence base, three models of development have emerged: (a) a cumulative model in which children with higher initial skills grow more over time than children with lower initial skills (Stanovich, 1986), (b) a compensatory model in which initial low achievers grow at a faster rate than initial high achievers (Parilla et al., 2005; Baumert, Lagy, & Lehmann, 2012; Morgan, et al., 2011), and (c) a stable model of development in which inter-individual rank-order differences and rates of growth remain relatively constant over time (Pfost et al., 2014). The cumulative model of development is characterized by increasing inter-individual variance in a fan-spread pattern resulting from higher performers building on their initial advantage and low performers falling further behind, and this model is represented by a significant positive correlation between intercept and slope parameters in a latent growth curve model. Put another way, students with scores below the mean at the first time point grow more slowly than those who scored above the mean initially, leading to more and more variation in the range of scores over time (see Figure 1a). This type of development suggests that children with initial deficits may experience long-term disadvantage relative to their higher performing peers. Under the compensatory model of development, the correlation between intercept and slope is significantly negative, indicating that the rate of growth increases for children who start with lower scores but remains relatively stable or decreases for initial high performers, allowing the low performers to catch up to their higher performing peers. Figure 1b displays the decreasing variation in scores over time in compensatory development. In this type of developmental model, those initially disadvantaged children appear to overcome skill deficits and catch up to their higher performing peers. A stable model of development theorizes that any gaps in achievement initially present among individuals will remain constant over time (Figure 1c) and is represented by a non-significant correlation between intercept and slope parameters. This model suggests that all children grow at the same rate, regardless of where they started.

Figure 1.

Figure 1.

a) Cumulative growth pattern. b) Compensatory growth pattern. c) Stable growth pattern.

Each model suggests different implications for how reading and math instruction should respond to developmental trajectories. For example, if children who start with lower skills eventually catch up to their peers, do developmentally sensitive periods or instructional changes such as specialized education plans play an important role in the potential “growth spurts” that occur? Alternatively, if children who start lower grow slower or remain in this pattern across elementary school, are current instructional practices ineffective in helping these students gain the necessary reading and math skills? If so, perhaps more intensive interventions should be developed or intervention efforts should be applied before the start of formal schooling rather than later. A better understanding of developmental trajectories may also help illuminate whether developmental patterns are domain-general (e.g. same patterns found for reading and math) or domain-specific (e.g.., different patterns indicated for reading and math). Similar patterns would suggest the same endogenous (e.g., genetic, cognitive) and exogenous (e.g., instructional, experiential) influences on skill development, whereas different developmental patterns might suggest that endogenous or exogenous influences differ across these skills or influence these skills at different developmental time points.

Stanovich (1986) initially proposed a cumulative model of development or “Matthew effect” for reading after synthesizing the existing literature on cognitive processes in reading. Since that time, the cumulative model has been supported in numerous longitudinal studies across a wide range of ages and nationalities (e.g. Bast & Reitsma, 1998; Pfost et al., 2014; Juel, 1988; Campbell et al., 2002; Carreker et al., 2007). In a comprehensive meta-analytic review of longitudinal studies of reading development, Pfost et al. (2014) identified 18 of 78 results that supported a cumulative model of reading development. The time periods included in these longitudinal studies ranged from kindergarten to 9th grade, with most studies following participants through three time points; however, a small portion of these longitudinal studies included five or more time points. Additionally, these longitudinal investigations measured developmental patterns in decoding, reading proficiency, and reading comprehension spanning English, German, Dutch, Finnish, Swedish, & Greek languages, which supplied evidence for the generalizability of a cumulative model of reading development across multiple developmental periods, language orthographies, and measures.

Cumulative growth trajectories for math performance have also been indicated in several studies using U.S., Finnish, and German samples that included students from preschool through 5th grades and used multiple measures of math ability (Aunola, Leskinen, Lerkkanen, & Nurmi, 2004; Bailey, Watts, Littlefield, & Geary, 2014; Bodovski & Farkas, 2007). Results of these studies suggest that learning rates may be unequal among students and that achievement gaps present at the start of formal schooling may increase for both reading and math skills throughout the elementary years. Despite the widespread acceptance of Matthew effects for both reading and math development, however, supportive evidence for alternative models has also been documented.

Other results included in the meta-analytic review by Pfost et al. (2014) provided evidence for compensatory development (33 of 78 studies) or stable development in reading (20 of 78 studies), and, similar to the cumulative evidence, results consistent with compensatory or stable development were representative of a wide range of language orthographies, developmental time periods, and reading measures, including pre-reading skills. Despite the higher number of results indicating a compensatory rather than a cumulative or stable model of development, there were no significant differences in the proportion of developmental patterns. Generally, sample and measurement characteristics overlapped between cumulative, compensatory, and stable results for reading, revealing no clear pattern of moderation.

Although a larger proportion of the literature on math development has indicated cumulative growth patterns, results of a few studies indicate compensatory growth among elementary-aged children (e.g. Jordan, Kaplan, & Hanich, 2002; Salaschek, Zeuch, & Souvignier, 2014). Longitudinal math development is currently under-investigated in comparison to reading development, which may influence the fewer number of results documenting stable or compensatory growth trajectories and, importantly, suggests the need for additional evidence. Taken together, these conflicting results suggest that development in reading and math skills may depend on contextual factors, although it is unclear how much these factors may overlap across reading and math skills. The paucity of studies which examine both reading and math development simultaneously makes it difficult to determine if these contextual factors might be common across skill domains or unique to reading or math.

Many of the previous longitudinal studies have focused on specific developmental periods such as early elementary school (preschool through 2nd grade; Aunola et al., 2004) or later elementary school (4th through 6th grades; Baumert et al., 2012), or have focused on single measures of reading or math (e.g., Baumert et al., 2012; Parrila et al., 2005). Parrila et al. (2005) cautioned that, for any one variable, individual differences may be larger or smaller at different developmental periods, leading to evidence for a cumulative model of development during early elementary years and evidence for a compensatory or stable model later in development. Furthermore, instructional factors have been proposed to influence students’ growth in reading and mathematics (McCoach et al., 2006; Bodovski & Farkas, 2007). There is also some evidence that instructional practices may be different across domains (i.e., more time invested in instruction for reading versus math; Lee & Reeves, 2012), and that these instructional differences also may vary from early to late elementary school (Jordan et al., 2002).

Current Study

Apart from genetic influences, individual differences in developmental trajectories may vary based on the broad learning domain (i.e., reading, math), developmental time period (i.e., early or late elementary school), or domain-specific skill (e.g., decoding, comprehension, fluency, problem solving). For example, the greater attention typically allocated to reading development in early elementary grades might translate into more early opportunities for compensatory growth. Although most previous investigations have explored either math or reading development separately, only a few have examined reading and math within the same sample (e.g. Jordan et al., 2002; LoGerfo, Nichols, & Reardon, 2006). The current study is unique in that it examines the developmental trajectories of both reading and math skill using multiple measures across multiple grades in early and late elementary school within a single sample. This type of investigation allows exploration of these potential domain-, time-, and measure-specific influences on growth in reading and math while controlling for sample characteristics. Evidence for domain-general developmental trajectories might suggest common factors influence acquisition of both reading and math skills. Such a finding might then motivate further research into potential common factors that would be supported by generalized intervention efforts. In contrast, evidence for domain-specific developmental trajectories would suggest factors influencing reading skill differ from those that influence math, spurring a need to investigate catered learning support such as instruction and intervention to each specific skill. Further, if there are differences among the developmental patterns of skills within the broader academic domains, this might suggest even more fine-tuned intercession efforts are warranted.

Method

Participants

Children in this study were recruited when they were in preschool and tested in kindergarten (2007–2010) through 5th grade (2013–2016) as part of a larger study of identifying and reducing risk for reading-related learning disabilities in Florida. Participants were mainly recruited from three school districts in north Florida (one urban, two rural). We were not aware of any specific interventions for reading or math taking place during data collection, although there may have been some interventions occurring at some of the schools. The research was reviewed and approved by the Institutional Review Board (HSC2010.0144). Informed consent was obtained from the parents or guardians of all participants prior to data collection. For this study, data from 921 children who were assessed in kindergarten through 5th grade were used. Attrition rates increased from 1st through 5th grade. Retention rates were 100% in Kindergarten, approximately 80% in 1st grade, 70% in 2nd grade, 40% in grade 3, 35% in grade 4, and 30% in grade 5. The percentages of missing data for all measures are provided in supplementary Tables S1–S4. All assessments were conducted by trained research assistants, individually, in a quiet area of the children’s schools. Research assistants were trained to criterion performance on the measures used in this study through didactic presentations, modeling, a performance assessment of test administration to an adult posing as a child, and live observations of assessment with feedback. Assessments were conducted over several 30- to 45-minute sessions, within a two- to three-week period. At the kindergarten assessment, children ranged in age from 5.08 to 6.92 years (M = 5.98, SD = 3.72 months), and, at the fifth-grade assessment, children ranged in age from 10.42 to 12.00 years (M = 11.04, SD = 3.73 months). Girls represented 44.8% of the kindergarten sample. The sample was diverse in its racial and ethnic make-up; 48% of participants were White; 42% were Black/African American; 3% were Hispanic/Latino; 4% were Multiracial; 2% were Asian, and 2% were classified as other/unknown. Based on parent report, the median income level was between $31,000 and $40,000, which was somewhat below the median income for the local area (i.e., $51,000) at the time of the study and indicates that the sample, on average, represented families in low- to low-middle-SES groups. This is consistent with the larger study, which focused on children at-risk for later difficulties. The majority of children recruited in preschool came from families with lower than average SES.

Measures

The assessments used were chosen partly because they measure skills which are known to develop rapidly during the elementary school years (Chall, 1983; Clements & Sarama, 2014), and to be significantly associated with later school performance and life outcomes (Reyna et al., 2009; Ritchie & Bates, 2013), and also because the particular assessments are all widely accepted, validated, standardized measures of core reading and math skills, and their component skills (e.g. phonological processing, number competency) that have been used in many previous studies of reading and math. Additionally, using multiple skills assessments from the same test along with measuring the same skills across different tests allows for the capture of both broad trends in growth and the examination of specific trends that may be related to an individual skill or measure. Sample-based reliabilities for all measures correspond closely with the reported reliabilities and are provided within the supplementary tables.

Decoding

Test of Word Reading Efficiency (TOWRE).

The TOWRE (Torgesen, Wagner, & Rashotte, 1999) is a timed, norm-referenced, test designed to assess word reading fluency through the Sight Word Efficiency (SWE) subtest and pseudoword reading fluency through the Phonemic Decoding Efficiency (PDE) subtest (Torgesen, et al., 1999). Participants are instructed to read as many real words or pseudowords as quickly as possible in 45 seconds and are scored on the total number of words read correctly within each subtest. Test-retest reliability coefficients range between .89-.90 for the PDE subtest and .91-.93 for the SWE subtest.

Woodcock Reading Mastery-Revised: Word Identification subtest (WID).

The WID subtest (Woodcock, 1998) is an untimed measure of the ability to accurately read aloud single real words of increasing difficulty. There are a total of 106 items and testing is discontinued after 6 consecutive errors. The split-half reliability coefficients for WID on a representative, nation-wide sample of students ranged from .91 to .98.

Woodcock Reading Mastery-Revised: Word Attack Subtext (WA).

The WA subtest (Woodcock, 1998) measures the ability to correctly decode 45 pseudowords or words of low frequency of occurrence in the English language in order of increasing in difficulty. The subtest is discontinued when six consecutive errors are made and scores are calculated from the total number of correct responses. Split-half reliability coefficients ranged from .89-.94.

Phonological Awareness

The Elision (EL) and Blending Words (BW) subtests of the CTOPP (Wagner, Torgesen, & Rashotte 1999) were administered.

Comprehensive Test of Phonological Processing (CTOPP) Elision.

The EL subtest is a phoneme deletion task on which children are required to repeat a word after removing a given syllable or phoneme. According to the test manual, test-retest reliability for the EL subtest ranged from .79 to .88. This subtest is discontinued after children produce three consecutive incorrect responses.

Comprehensive Test of Phonological Processing (CTOPP) Blending Words.

The BW subtest requires children to listen to individually presented sounds and then combine (blend) them together into a real word. The BW subtest is discontinued after children produce three consecutive incorrect responses. Test-retest reliability for the BW subtest ranged from .72 to .88 in the norming sample (ages 5–17 years).

Reading Comprehension

Woodcock Reading Mastery-Revised: Passage Comprehension Subtest (PC).

The PC subtest (Woodcock, 1998) measures an individual’s understanding of written text using a cloze procedure that requires the examinee to supply a missing word, first for sentences and then for paragraphs of increasing complexity. The PC subtest contains 68 items, and testing is discontinued after six consecutive errors. According to the test manual, split-half reliability for passage comprehension ranged from .73 to .94.

Test of Silent Reading Efficiency and Comprehension (TOSREC).

The TOSREC (Wagner, Torgesen, Rashotte, & Pearson, 2010) is a measure of comprehension of connected text. Examinees have to read sentences silently and indicate “yes” or “no” as to the truthfulness of as many of the sentences as possible within a three-minute period. In this study, grade-level test forms of the TOSREC were administered to children in first through fifth grades. According to the test manual, alternate form reliability for the TOSREC ranges from .92 in first grade to .89 in fifth grade.

Math

Test of Early Math Ability (TEMA).

The TEMA (Ginsburg & Baroody, 2003) is a norm-referenced assessment designed to measure early mathematics ability in ages three to eight years. It includes 72 items that include tasks such as producing finger displays for supplied numbers, reading numerals, writing numerals, choosing larger or smaller numbers or sets of numbers, counting, and part-whole conceptualizations. Testing is discontinued after five consecutive incorrect responses. In this study, the TEMA was administered to children in kindergarten through second grade, According to the test manual, test-retest reliability coefficients for the normative sample range from .82 to .93.

Woodcock-Johnson Tests of Achievement III (WJ-III).

The Math Fluency, Applied Problems and Calculation subtests of the WJ-III (Woodcock, McGrew, & Mather, 2001) were administered to children in the third through fifth grades. The Math Fluency subtest assesses the examinee’s ability to answer addition, subtraction, multiplication, and division problems correctly in a limited amount of time (3 min). According to the manual, test-reliability for Math Fluency is .95 for ages 7 to 11 year age range. The Applied Problems subtest assesses the examinee’s ability to analyze increasingly difficult math problems that contain both relevant and extraneous information and then use basic calculations to solve the problems. According to the test manual, median reliability for Applied Problems is .92 for the 5 to 19 year age range. The Calculation subtest requires examinees to perform mathematical computations that are increasingly difficult, ranging from simple (writing single numbers) to advanced (geometric or calculus operations). According to the test manual, median reliability is .85 for the 5 to 19 year age range.

Analyses

A series of univariate latent-growth-curve models was conducted to examine the development of reading and math skills and to investigate individual differences in developmental patterns for these skills (McArdle & Bell, 2000). Latent growth models allow for the estimation of initial status or intercept in a trait of interest as well as the slope or change trajectory for that trait over time (Duncan & Duncan, 2004). The mean intercept and slope parameters represent the average initial status and average pattern of growth within a sample, whereas the intercept and slope variance estimates represent the presence of individual differences in initial status and development. To explore developmental trajectories from kindergarten through the elementary grades, the first time point was set as the intercept in all models. Models were estimated in Mplus 7.4 (Muthén & Muthén, 1998–2012), with missing data handled using full-information maximum likelihood (FIML) estimation1. Full-information maximum likelihood was chosen as the estimator for these analyses because, like multiple imputation, this method provides “…important information about the underlying marginal distributions of the incomplete variables and hence may reduce the bias that would result from the listwise deletion of cases” (Enders & Bandalos, 2001, p. 435). Observed raw scores were used in models for decoding and phonological awareness skills. Due to non-normal distributions among the kindergarten raw scores for Woodcock Reading Mastery-Revised and WJ-III subtests, developmental (“W”) scores were used in models for these skills. Model fit was evaluated using multiple criteria, including the χ2 statistic, the root mean square of approximation (RMSEA), standardized root mean square residual (SRMR), Bentler’s Comparative Fit Index (CFI; Hu & Bentler, 1999), and the Tucker-Lewis Index (TLI). Better fitting models are indicated by χ2 values lowest and closest to the degrees of freedom, with values that are non-significant preferred (Kline, 2011). Values of CFI and TLI above .95 indicate close model fit, and for RMSEA, values less than .08 indicate adequate model fit (Browne, Cudeck & Bollen, 1993). The SRMR statistic represents good model fit when values are below .05 and acceptable model fit when values are below .08 (Hooper, Coughlan, & Mullen, 2008).

Results

Descriptive statistics are presented first, followed by the latent-growth-curve analyses for reading and math at all available grades, then the latent-growth-curve analyses sectioned into the earlier time points (grades K-2) and the later elementary years (grades 3–5). The mean age of participants at each grade and descriptive information (means, standard deviations, minimum and maximum values, skewness and kurtosis) calculated on raw or W scores for decoding, phonological awareness, reading comprehension, and math at all available grade levels are provided in the supplementary tables S5–S8.

Latent-Growth-Curve Analyses

Kurtosis values were above 2 for the TOWRE-decoding in kindergarten, WRM-reading comprehension at grade 3, and math calculation at grades 3 and 4, indicating outliers were present within the data. To investigate the influence of these outliers, models were fit with extreme data points present and again with extreme data points removed. The results revealed equivalent model fit indices and closely similar parameter estimates (within a tenth of a decimal point) between the two sets of analyses. Due to the high similarity between results under these two conditions, results of the analyses in which all data points were included in model estimation are reported.

Reading-related skills.

Model fit indices, intercept, and slope parameters for latent-growth-curve models across all available time points for decoding, phonological awareness, and reading comprehension variables are shown in Table 1. Absolute fit indices indicated adequate fit for all models. For the TOSREC-reading comprehension, a model that included just linear growth provided the best fit to the data, and, for WID-decoding, a model that included just linear and quadratic growth represented the best fit. All other reading-related variables were best represented by models that included linear, quadratic, and cubic slopes. Hence, some element of curvilinear growth was present for all reading-related variables except for TOSREC-reading comprehension.

Table 1.

Fit indices and parameter values from the latent growth curve models across all available grade levels.

Variable χ2 (df) CFI TLI SRMR RMSEA Intercept Linear Quadratic Cubic

M Var M Var M Var M Var
TOWRE SWE 65.09(7) 0.98 0.96 0.12 0.10 12.40 160.40 29.72 286.00 −5.39 44.26 0.36 0.52
TOWRE PDE 11.67(7) 1.00 1.00 0.03 0.03 5.33 40.04 11.88 96.80 −1.30 12.02 0.04 0.11
WID 6.84(6) 1.00 0.99 0.09 0.01 504.21 420.50 −8.30 7.74 −0.49 0.14 --- ---
WA 15.84(7) 1.00 0.99 0.24 0.04 458.83 242.36 30.03 222.36 −6.92 32.72 0.59 0.38
CTOPP BW 27.22(7) 0.98 0.96 0.04 0.06 7.64 7.06 4.44 2.72 −1.41 0.48 0.15 0.01
CTOPP EL 4.00(7) 1.00 1.00 0.01 0.00 4.75 7.95 4.29 12.26 −0.62 2.57 0.03 0.04
PC 40.20(7) 0.99 0.97 0.45 0.07 435.25 250.28 34.52 226.73 −6.84 35.32 0.52 0.41
TOSREC 138.23(10) 0.92 0.92 0.20 0.12 15.91 87.70 7.55 9.46 --- --- --- ---

Notes: TOWRE SWE = Test of Word Reading Efficiency Sight Word Efficiency, TOWRE PDE = Test of Word Reading Efficiency Phonemic Decoding Efficiency, WID = Woodcock Reading Mastery Test Word Identification, Word Attack = Woodcock Reading Mastery Test Word Attack, CTOPP BW = Comprehensive Test of Phonological Processes Blending Words, CTOPP Elision = Comprehensive Test of Phonological Processes Elision, Passage Comp = Woodcock Reading Mastery Tests Passage Comprehension, TOSREC = Test of Silent Reading Efficiency and Comprehension, M = mean estimate, Var = variance estimate. Bolded estimates indicate significance at p < .05.

Mean intercepts and the intercept variances were significant for all decoding, phonological awareness, and reading comprehension variables, indicating the presence of inter-individual differences in these skills at kindergarten. With the exceptions of WID-decoding and CTOPP BW–phonological awareness, all reading-related variables had significant mean slope and slope variances, indicated inter-individual differences in the development of these skills between kindergarten and fifth grade. The correlation between intercept and slope for WID- decoding was significant and positive (r = .42, p < .001), indicating a cumulative growth pattern in which decoding at kindergarten was positively associated with how developmental changes in decoding performance occurred through fifth grade. For WA-decoding and TOSREC-reading comprehension, the correlations between intercept and slope were significantly moderate and negative (r = −.21, p = .034 and −.48, p < .001, respectively), indicating a compensatory growth pattern or a negative influence from initial status to change in these skills over time. Correlations between intercept and slope were non-significant for TOWRE SWE-decoding (r = −.11, p = .11), TOWRE PDE-pseudo word decoding (r = .02, p = .86), CTOPP BW-phonological awareness (r = −.26, p = .46), CTOPP EL-phonological awareness (r = .17, p = .48), and WRM-reading comprehension (r = −.15, p = .13), indicating stable growth pattern for these measures.

As seen in Figures 2a through 2c there were large increments in growth in reading-related skills between kindergarten and third grade; however, between third and fifth grades these increments become smaller, as the rates of change decreased and leveled off, suggesting that differential growth patterns were occurring across these two developmental periods. To investigate the potentially different developmental patterns occurring between earlier and later elementary grades, two sets of latent-growth-curve analyses were conducted for all skills: One set of analyses for grades kindergarten through grade 2 and one set of analyses for grade 3 through grade 5. Mean scores for TOSREC-reading comprehension indicated relatively little change across grades 1 and 2, and a decrease in scores from grade 4 to grade 5. Latent-growth-curve models of TOSREC-reading comprehension grades 1 through 3, and from grades 3 through 5 resulted in poor model fit, negative slope variances, and non-positive definite theta or psi matrices, suggesting weak or unstable growth patterns were present; therefore, these analyses were not included. Results of the two sets of linear models for all other reading variables are shown in Tables 2 and 3. Model fit indices revealed good-to-excellent fit for all outcomes. Similar to the models that included all time points, mean and variance estimates for intercepts were significant for kindergarten to grade 2 models and for grade 3 through grade 5 models for all outcomes.

Figure 2.

Figure 2.

Growth patterns of reading-related skills across grades.

Table 2.

Fit indices and parameter values from the latent growth curve models across K-2nd grades for decoding, phonological awareness, and reading comprehension.

Reading χ2 (df) CFI TLI SRMR RMSEA Intercept Linear

M Var M Var
TOWRE SWE (K-2nd) 0.23 (1) 1.00 1.00 0.00 0.00 12.37 198.53 20.47 82.15
TOWRE PDE (K-2nd) 0.04 (1) 1.00 1.00 0.00 0.00 5.31 43.90 9.54 30.68
WID (K-2nd) 0.66 (1) 1.00 1.00 0.01 0.00 387.13 989.90 26.49 90.82
WA (K-2nd) 0.11 (1) 1.00 1.00 0.00 0.00 458.78 261.19 18.50 56.99
CTOPP BW (K-2nd) 1.56 (1) 1.00 1.00 0.01 0.03 7.60 7.94 2.24 0.99
CTOPP EL (K-2nd) 0.80 (1) 1.00 1.00 0.01 0.00 4.75 9.42 3.16 3.20
PC (K-2nd) 5.66 (1) 1.00 0.99 0.01 0.07 435.28 325.93 22.76 79.85

Note: TOWRE SWE = Test of Word Reading Efficiency Sight Word Efficiency, TOWRE PDE = Test of Word Reading Efficiency Phonemic Decoding Efficiency, WID = Woodcock Reading Mastery Test Word Identification, Word Attack = Woodcock Reading Mastery Test Word Attack, CTOPP BW = Comprehensive Test of Phonological Processes Blending Words, CTOPP Elision = Comprehensive Test of Phonological Processes Elision, Passage Comp= Woodcock Reading Mastery Tests Passage Comprehension, TOSREC = Test of Silent Reading Efficiency and Comprehension. M = mean estimate, Var = variance estimate. Bolded estimates indicate significance at p < .05.

Table 3.

Fit indices and parameter values from the latent growth curve models across 3rd through 5th grades for decoding, phonological awareness, and reading comprehension.

Reading χ2 (df) CFI TLI SRMR RMSEA Intercept Linear

M Var M Var
TOWRE SWE (3rd-5th) 0.44 (1) 1.00 1.00 0.01 0.00 59.98 185.88 4.05 4.00
TOWRE PDE (3rd-5th) 3.44 (1) 1.00 1.00 0.01 0.06 19.50 138.41 4.94 6.86
WID (3rd-5th) 0.91 (1) 1.00 1.00 0.03 0.00 477.74 539.69 9.31 8.66
WA (3rd-5th) 1.65 (1) 1.00 1.00 0.05 0.04 498.87 255.16 4.01 4.99
CTOPP BW (3rd-5th) 0.05 (1) 1.00 1.00 0.00 0.00 12.20 10.01 0.20 1.13
CTOPP EL (3rd-5th) 1.98 (1) 1.00 0.99 0.02 0.04 12.31 21.12 0.91 0.47
PC (3rd-5th) 4.91 (3) 1.00 1.00 0.46 0.04 487.40 141.39 5.23 --

Note: TOWRE SWE = Test of Word Reading Efficiency Sight Word Efficiency, TOWRE PDE = Test of Word Reading Efficiency Phonemic Decoding Efficiency, WID = Woodcock Reading Mastery Test Word Identification, Word Attack = Woodcock Reading Mastery Test Word Attack, CTOPP BW = Comprehensive Test of Phonological Processes Blending Words, CTOPP Elision = Comprehensive Test of Phonological Processes Elision, Passage Comp = Woodcock Reading Mastery Tests Passage Comprehension, TOSREC = Test of Silent Reading Efficiency and Comprehension. M = mean estimate, Var = variance estimate. Bolded estimates indicate significance at p < .05.

With the exception of CTOPP BW-phonological awareness and WA-decoding slope, mean and variance estimates demonstrated different patterns of significance for the kindergarten through grade 2 models than for grade 3 through grade 5 models. Specifically, the variances of the slope parameters were statistically significant for outcomes between kindergarten and grade 2, but the variances of the slope parameters were not statistically significant for outcomes between grades 3 and 5. For k-2 decoding, correlations between intercept and slope were moderately significant and negative for TOWRE SWE-decoding (r = −.38, p < .001), WID (r = −.48, p < .001), and WA (r = −.37, p < .001), but non-significant for TOWRE PDE-pseudo word decoding (r = −.08, p < .07). The correlation between intercept and slope for CTOPP BW-phonological awareness was moderately negative and significant (r = −.43, p < .04) but was non-significant for CTOPP EL-phonological awareness (r = −.12, p = .23). For WRM-reading comprehension, the correlation was large, negative, and significant (r = −.52, p < .001).

The slope-intercept correlations for decoding in grade 3 to grade 5 was large, negative, and significant for TOWRE SWE-decoding (r = −.62, p < .001) and WA-decoding (r = −.69, p < .001), and moderate, negative, and significant for TOWRE PDE-pseudo word decoding (r = −.34, p < .03) and WID-decoding (r = −.48, p < .001). Correlations between intercept and slope for phonological awareness in the later grades were moderately negative and significant for CTOPP BW-phonological awareness (r = −.45, p < .05), and large, negative, and significant for CTOPP EL-phonological awareness (r = −.57, p < .001). For WRM-reading comprehension the slope parameter was constrained to zero due to a negative estimated slope variance; therefore, no slope-intercept correlation was estimated.

Math-related skills.

Because of the change in the math measures used across grades, it was not possible to conduct latent-growth-curve analyses from kindergarten through grade 5. Instead, latent-growth-curve models were estimated separately from kindergarten to grade 2 and from grade 3 to grade 5. Model fit indices, intercept, and slope parameters for the models for the math outcomes are shown in Table 4. All models provided excellent fit to the data. The pattern of growth from kindergarten to grade 2 in early math ability is shown in Figure 3a, and the patterns of growth in math fluency, applied problems, and calculation from grade 3 to grade 5 are shown in Figure 3b.

Table 4.

Fit indices and parameter values from the latent growth curve models for early math ability, math fluency, applied problems, and calculation.

Math χ2 (df) CFI TLI SRMR RMSEA Intercept Linear

M Var M Var
TEMA (K-2nd) 0.95 (1) 1.00 1.00 0.01 0.00 28.23 59.04 12.99 3.41
WJ III MF (3rd-5th) 0.94 (1) 1.00 1.00 0.38 0.00 496.30 38.82 3.67 6.08
WJ III AP (3rd-5th) 0.03 (1) 1.00 1.00 0.01 0.00 495.99 321.61 8.41 −0.81
WJ III Calc (3rd-5th) 1.26 (1) 1.00 1.00 0.05 0.02 499.08 128.89 10.08 16.41

Notes: TEMA = Test of Early Mathematics Ability, WJ III MF = Woodcock Johnson III Tests of Achievement Math Fluency, WJ III AP = Woodcock Johnson III Tests of Achievement Applied Problems, WJ III Calculation = Woodcock Johnson III Tests of Achievement Calculation. M = mean estimate, Var = variance estimate. Bolded estimates indicate significance at p < .05.

Figure 3.

Figure 3.

Growth patterns of math skills across grades.

For all math outcomes, the mean and variance estimates for the intercept were significant, indicating differences between children in initial skill levels, and the mean estimate for the slope was significant, indicating an increase in scores across time. Values for slope variance were not significant for the TEMA-early math ability in kindergarten through grade 2 or for Applied Problems in grade 3 to grade 5, which indicated that although scores increased on the measures over time, there were no inter-individual differences in the rate of growth on these measures. In contrast, the slope variance estimates for math fluency and calculation were significant in grade 3 through grade 5, indicating that there were individual differences in the rate of growth on these measures. For math fluency, the correlation between intercept and slope was small, positive, and significant (r = .16, p < .03), indicating a cumulative growth pattern in which higher math fluency in third grade was positively associated with development of math fluency through fifth grade. In contrast, correlations between intercept and slope for early math ability (r = .75, p = .18), applied problems (r = −.16, p = .23), and calculation (r = .05, p = .68) were non-significant, indicating stable growth patterns.

Multivariate growth models

Following the univariate analyses, a post-hoc exploration of the apparent differences in growth patterns within and across broad skill domains was conducted using a series of multivariate, latent-growth-curve models. These multivariate-latent-growth models extended univariate models to examine the associations between latent characteristics of change in two or more skills, simultaneously (e.g. Curran et al., 1997). Models were fit between reading skills in grades K-2nd and in grades 3–5, between math skills in 3rd-5th grades, and between reading and math skills in K-2nd and 3rd-5th grades. These grade-level splits were chosen to align with the two developmental periods that were identified within the univariate analyses. Fixed quadratic, random intercept and slope models were fit because only three time points were used within these analyses. These models estimate curvilinear mean growth over time and the variance of growth at the intercept but hold the variance in the quadratic term fixed to zero to allow for model identification. Two sets of models were conducted, one set that allowed the slope variances of each skill to be freely estimated and one set that constrained slope variances to equality. If growth in one skill is substantially different from growth in the other, then a model with variances constrained will provide a significantly worse fit to the data than a model with variances freely estimated. Performing a series of chi-square difference tests across these two sets of models provides a formal comparison of growth patterns within reading and math skills, and importantly, across reading and math skills to supply additional evidence for either domain-general or domain-specific developmental trajectories. Due to the large number of model comparisons, results for the within-reading and within-math models are presented in supplementary online tables S9–S11.

Results for reading and reading-related skills in grades K-2nd (see Table S9) indicated unconstrained models fit the data better across all sets of skills with the exceptions of WA-decoding with WRM-reading comprehension, and WA-decoding with TOWRE PDE-pseudo word decoding. Furthermore the between-slope correlations were significant for all pairs of skills with the exceptions of TOWRE SWE-decoding and TOWRE PDE-pseudo word decoding with BW-phonological awareness. In grades 3–5, unsurprisingly, opposite patterns were evident (see Table S10). Models with slope variances constrained to equality were not significantly different from models with unconstrained slope variances, suggesting similar growth trajectories between reading skills during this time period. The slope-slope correlations were, accordingly, non-significant.

Only one math measure was available in K-2nd grades and this measure was different than those available in 3rd-5th grades; therefore, multivariate growth within math skills was only modelled in grades 3–5 (see Table S11). Results varied across skills. For Math Fluency and Applied Problems, the slope-slope correlation was non-significant, and the model with variances constrained did not produce significantly worse fit to the data than the unconstrained model. However, multivariate models between Math Fluency and Applied Problems and between Applied Problems and Calculation indicated the constrained models provided a significantly worse fit to the data with significant slope-slope correlations present for Math Fluency with Calculation but not for Applied Problems with Calculation.

Cross-domain multivariate models are presented in Table 5. Reading and math developmental trajectories were first examined in K-2nd grades. Models with unconstrained slope variances fit the data better for the relations between TEMA-early math ability and most reading skills, with significant slope-slope correlations present between TEMA-early math ability and WRM-reading comprehension, and between TEMA-early math ability and TOWRE PDE-pseudo word decoding. The modeled associations between TEMA-early math ability and both phonological awareness measures, however, indicated the unconstrained models presented a better fit to the data. A significant slope-slope correlation was indicated for the relation between TEMA-early math ability and CTOPP EL-phonological awareness but not for the relation between TEMA-early math ability and CTOPP BW-phonological awareness. The results of chi-square difference tests between reading and math skills in grades 3–5 indicated that, generally, constrained models did not fit the data significantly worse than unconstrained models, although there were several exceptions: TOWRE SWE-decoding with Calculation, WRM-reading comprehension with Calculation, TOWRE PDE-pseudo word decoding with Math Fluency, TOWRE PDE-pseudo word decoding with Calculation, WA-decoding with Calculation, CTOPP BW-phonological awareness with Math Fluency, and CTOPP EL-phonological awareness with Math Fluency. The slope-slope correlations were significant between TOSREC-comprehension and Math Fluency, WRM-passage comprehension and Applied Problems, and between WA-decoding and Calculation but not between any other sets of skills.

Table 5.

Correlations between slopes of reading outcomes and math outcomes in kindergarten-second grade and third grade - fifth grade

K-Grade 2 Grade 3 - Grade 5


TEMA WJ-III Math Fluency WJ-III Applied Problems WJ-III Calculation
WRM Passage Comprehension .29u .26c .69c NPDu
WRM Word Identification .13u .26c ---b, c .24c
WRM Word Attack .19u .08c NPDc .90c
TOWRE SWE .21u .26c .76c .08u
TOWRE PDE .38u NPDu .36c ---b, u
CTOPP Blending Words .97c −.49u .14c −.15c
CTOPP Elision .95c .20u ---b, c ---b, c
TOSREC ---a .39c .45c .19c

Notes.

a

Not evaluated;

b

model yielded inappropriate solution (i.e., correlation > 1.0);

u

unconstrained slope model;

c

constrained slope model. NPD = non positive definite solution.

WRM = Woodcock Reading Mastery; WJ-III = Woodcock-Johnson , 3rd edition; CTOPP = Comprehensive Test of Phonological Processing; TOWRE Test of Word Reading Efficiency; SWE = Sight Word Efficiency; PDE = Phonemic Decoding Efficiency; TOSREC = Test of Silent Reading Efficiency and Comprehension; TEMA = Test of Early Mathematics Ability. Correlations shown in bold are significant at p < .05.

Discussion

The goal of this study was to examine the developmental trajectories of reading and math across multiple measures and multiple time periods within a single sample and to investigate whether development showed common or unique patterns across academic domains or across subskills within domains. Growth-curve analyses revealed two salient trends in development: First, for reading-related skills, a consistent pattern of two-stage growth emerged in which there were significant individual differences in growth during the early elementary-school period (i.e., kindergarten to second grade) and non-significant individual differences in growth during the late elementary-school period (i.e., third to fifth grade). Second, for math-related skills, the opposite pattern of two-stage growth was present in which there were no significant individual differences in growth for early math skills and significant individual differences in growth for two of three measures of math ability in the late elementary-school period. These trends support domain-specific developmental processes associated with time-based factors (early versus late elementary school). The presence of individual differences in early but not late elementary school highlights complexities in developmental trajectories that were not detected by simple examination of the slope-intercept correlation across all time points and supports theories that patterns of reading and math development may vary across time (Jordan et al., 2002). There were several correlations that were medium to large in magnitude but non-significant, further suggesting that a reliance on significance tests alone could lead to over-interpretation of the results. We conclude that there was not substantial evidence in support of any clear patterns of cumulative, compensatory, or stable development for either reading-related or math skills, indicating that within domains there also may be contextual factors which differentially influenced developmental trends.

Reading-Related Skills

In general, results of reading-related skills from kindergarten through fifth grade indicated that curvilinear growth patterns were present with greater increases in the early grades that began to decelerate around grade 3. Follow-up analyses of growth patterns in early versus later grades revealed significant mean growth and substantial individual differences in growth rates for reading-related outcomes in kindergarten through second grade. Mean growth rates remained mostly significant but decreased in third through fifth grades, and the extent of individual differences in growth rates for reading-related outcomes was much lower for the later grades. Overall, these findings indicate that there are substantial individual differences in growth rates for reading-related outcomes during early elementary school, but the extent of individual differences in growth rates for these skills is inconsequential in later elementary school. Similar to the results of this study, Pianta et al. (2008) reported that the majority of the change in reading skills occurred by the third grade in their longitudinal study of 791 children in first through fifth grades. Children in kindergarten through second grade fall within the “learning to read” phase of Chall’s (1983) developmental model of reading, during which reading comprehension and reading component skills are building rapidly (e.g. Kim, Petscher, Schatschneider, & Foorman, 2010).

Math Skills

In contrast to reading-related skills, mean growth rates for math skills were significant in both early and late elementary school and the overall pattern for math growth indicated individual differences were present in late elementary school but not in early elementary school. Persistent mean growth across all time points may indicate that math knowledge accrues and builds on itself over a longer period of time than reading. The transition from informal to formal math knowledge begins to occur as early as kindergarten, but the shift is more pronounced by third grade, which also aligns with the increases in individual differences found for most of the math skills measured within this study (Clements & Sarama, 2009). Math skills are hypothesized to increase as children gain knowledge of number sequence and are then able to generalize this knowledge to larger and larger arrays of numbers (e.g. Siegler & Lortie-Forgues, 2014). This trajectory indicates a possible compounding of skill development which would further correspond with both the findings of significant mean growth over time and increased individual differences in later grades.

Patterns of Growth

To focus on developmental patterns across elementary school and include all available time points, patterns of growth for reading were examined using the k-5 models. Examination of slope-intercept correlations revealed no consistent growth pattern within or across reading-related skills. That is, the level of reading-related skills children had in kindergarten was not consistently associated with more or less growth through 5th grade. A cumulative growth pattern was observed for WID-decoding (i.e., children with initially higher skills gained more skills than children with initially lower skills); a compensatory growth pattern was observed for WA-decoding and TOSREC-reading comprehension (i.e., children with initially lower skills gained more skills than children with initially higher skills); and a stable growth pattern was indicated for TOWRE SWE-decoding, TOWRE PDE-pseudo word decoding, CTOPP BW-phonological awareness, CTOPP EL-phonological awareness, and WRM-reading comprehension (i.e., growth in skills was not differentially related to initial skill level). Although stable growth seemed to be present for the majority of measures, the type of reading-related skill measured did not appear to be linked to any one pattern of growth, suggesting that variability in growth patterns were not driven by subskills of reading but potentially by contextual factors that influenced those subskills.

In the two-stage models, individual differences in growth appeared to be domain-specific for reading and math skills. For reading, individual differences were present in early but not late elementary school, but for math individual differences were apparent in late elementary school that were not indicated in the early elementary years. Post-hoc multivariate model fit comparisons also supported the conclusions that growth trajectories were more varied in earlier elementary grades, but begin to stabilize across reading skills in the later elementary grades. Additionally, these analyses provided evidence that growth patterns were associated among reading skills in the earlier grades, but these associations were no longer significant by late elementary school. The early associations may reflect the importance of co-development between reading skills during Chall’s initial stages of ‘learning to read’, that have less influence in the later stages of ‘reading to learn’ (Chall, 1983).

A cumulative developmental pattern was indicated for math fluency skills in third-fifth grades, and stable growth patterns were suggested for early math skills, applied problem solving, and calculation. Results of the multivariate math models also indicated growth patterns generally consistent with increased individual differences in the later grades, though the developmental patterns were mixed across sets of skills. Using several measures of math skills allowed for the detection of more than one developmental pattern among math abilities, suggesting how children develop mathematical knowledge may be more complex than indicated when assigning only one pattern of growth (e.g. cumulative or compensatory).

Cross-domain, multivariate analyses revealed three significant associations in growth between reading and math skills in both early and later elementary grades, though for most skills, domain-specific growth patterns were indicated. In each of the cases where significant associations were present, the measures for reading and math were both designed to capture similar skills across the domains. TOSREC and Math Fluency are both measures of fluency skills, WRM-comprehension and Applied Problems both assess linguistic comprehension ability, and WA-decoding and Calculation assessments both potentially access facility with symbol recognition. The majority of the multivariate analyses revealed independent associations between the developmental trajectories of reading and math skills, however, suggesting domain-specific factors may have a greater influence on development than domain-general influences. These differences in developmental patterns for reading and math may be influenced by both endogenous and exogenous factors (Hart et al., 2009; LoGergo et al., 2006). In the next sections we discuss several endogenous and exogenous factors that could influence developmental trends for reading and math, although many of these potential factors were not measured within the current study (e.g. instructional type and time), or were measured, but were associated with attrition levels (e.g. inattention and SES).

Endogenous Factors Influencing Growth

Evidence from the quantitative genetic literature has indicated that genes influence approximately half of the variation in reading and math ability (e.g. Little et al., 2017; Kovas et al., 2007). Other endogenous factors such as socio-educational background, and cognitive ability (Jacobs et al., 2002) also have been associated with these skills, although a large body of evidence from behavior genetic research suggests that these factors are also somewhat influenced by genetics (Hart et al., 2019). Domain-specific cognitive skills may have been associated with developmental differences, although there were no consistent patterns of growth among the reading-related or math-related skills within the present sample to suggest this. Many of these endogenous factors; however, are not believed to operate specifically on a single academic domain, but rather across academic domains in a similar fashion (e.g. Sirin, 2005; Jacobs et al., 2002), suggesting that exogenous factors such as instructional contexts may be the primary influence on differences in developmental trends across reading and math.

Exogenous Factors Influencing Growth

Exogenous factors such as classroom instruction or classroom environment may play a role in influencing developmental patterns in reading across the elementary years (Pfost et al., 2014; Pianta et al., 2008). Exposure to literacy instruction decreases across first through fifth grades (Pianta et al., 2008; Sonnenschein, Stapleton, & Benson, 2010), and instructional practices typically shift from teacher-directed, code-focused (e.g. phonics instruction) to child-directed meaning-focused (i.e. self-paced reading; Sonnenshchein et al., 2010), corresponding with decreases in the rate and variance of growth for reading skills across these time periods. There may be additional environmental factors that influence how children develop reading skills in early elementary school that are no longer influential by the transition to late elementary school. Thoroughly understanding and identifying which factors influence reading development is an important extension of the current findings. Early identification of children at risk for reading difficulty can serve to inform preventative intervention, reduce achievement gaps, and improve developmental trajectories at a critical period in development (Berkman, Sheridan, Donahue, Halpern, & Crotty, 2011; Fantuzzo et al., 2005; Lee & Burkham, 2002; Vogler, DeFries, & Decker, 1985).

By third grade, many of the factors influencing reading growth may no longer be sensitive to intervention. Alternatively, the reduction of individual differences by third grade may be a reflection of the reduction in specific reading instruction that begins to occur at this time, indicating that more rather than less reading instruction and intervention is needed during this period to reduce gaps between higher and lower performing children. Results of the current study suggest these potential developmental changes in response to intervention could be vital directions for future research and provide new information about the patterns of growth in reading skills spanning all of elementary school and using multiple measures of reading.

The pattern of decreasing individual differences from early to late elementary school was consistent across the majority of the reading measures, although there were two exceptions: TOWRE PDE-pseudo word decoding) and CTOPP BW-phonological awareness. For TOWRE PDE-pseudo word decoding, significant variance in growth was present in both early and late elementary school, indicating individual differences in growth of this skill persisted across the elementary school years, rather than dissipating in third through fifth grades. Decoding in this study was measured using both timed and untimed real-word and pseudo-word formats. The exception to the developmental patterns for decoding was for timed pseudo-word reading (TOWRE PDE), suggesting that mastery of the underlying alphabetic principle may influence development of these skills. Similar to the proposed triangle model of reading, which posits two routes to successful decoding: phonological (TOWRE PDE-pseudo word decoding) and orthographic (TOWRE SWE-decoding; Harm & Seidenberg, 2004), the persistent individual differences for TOWRE PDE-pseudo word decoding may indicate that some children are still increasing their mastery of this pathway to decoding beyond the second grade. Furthermore, adding a timed constraint to pseudo-word decoding may tap into additional factors, such as processing speed, and contribute to lasting individual differences on this task through late elementary school (Shanahan et al., 2006). Replication studies should be conducted to determine whether these different patterns of individual differences in decoding growth persist across multiple samples and measures.

For CTOPP BW-phonological awareness, no significant variance in growth was detected throughout the elementary years, indicating no individual differences were present for growth on this task, even in early elementary school. There was no evidence of floor or ceiling effects within the current sample although the standard deviations for CTOPP BW-phonological awareness were slightly lower than those for CTOPP EL-phonological awareness, suggesting some restriction of range for the CTOPP BW-phonological awareness task (see supplementary online materials). Previous studies have found high convergence between phonological awareness tasks and high factor loadings for elision and phonemic blending tasks on a latent variable of Phonological Awareness (Wagner et al., 1997; Anthony & Lonigan, 2004), indicating suitable construct validity--although developmental change in these skills may begin to diverge over time. The reason for significant variance in growth for CTOPP EL-phonological awareness in early elementary school but not for CTOPP BW-phonological awareness in the present study remains unclear; however, further investigation of the different developmental patterns among measures of phonological awareness may be able to determine sources of these differences.

In contrast with reading, the quantity of mathematics instruction may increase across the elementary school years (e.g. Bodovski & Farkas, 2007; Jordan et al., 2009) and increased exposure to formal math concepts may trigger inter-individual differences in how children respond to math instruction (Pianta et al., 2008), such as those found within this study. Underlying differences in cognitive ability may be present at early elementary school; however, the exposure to math instruction may increase beyond a critical threshold at around third grade, stabilizing environmental influences and allowing individual differences in math ability to flourish (e.g., gene-by-environment interaction; Taylor, et al., 2010). Evidence suggesting that formal math knowledge begins to develop in kindergarten indicates the opportunity for classroom instruction to enhance these skills earlier than current practice dictates (e.g. Engel, Claessens, & Finch, 2013). However, these implications should be considered with caution until future research can explore them by comparing developmental trajectories for students across different levels of exposure to instruction, and by closely examining additional factors (e.g., advantaged versus disadvantaged status) which may influence response to instruction.

Limitations

Despite the strengths of the current study, which include the use of multiple academic domains measured simultaneously, multiple measures of each domain, an adequate sample size to address the questions, a relatively diverse sample, and a longitudinal design stretching over the entire elementary-school period, there were some limitations. First, there was attrition across time points, reaching approximately 64% by fifth grade; however, use of latent-growth models accounts for measurement error, increasing reliability and power, and reduces sample size requirements (Duncan & Duncan, 2004). Second, the goal of this study was to examine unbiased learning trajectories for reading and math; therefore, covariates which may be associated with missingness of the data were not included. Future investigations should further examine these developmental patterns while including covariates such as SES and inattention. Third, test forms for TOSREC-reading comprehension differed across the time points, which may have introduced additional measurement error into the growth model, leading to the erratic growth pattern observed for this measure. Fourth, although reading ability from kindergarten through fifth grades was measured with the same assessments at all time points, measures of early math ability and later math ability were not the same. Fifth, there was negligible variance on the growth parameters for many of the measures in which the within- or across-domain correlations were significant. This suggests that these significant correlations may reflect an increase in statistical power, rather than general influences of factors such as maturation, exposure, or instruction. Finally, data were not available for classroom exposure to reading or math instruction for children in the study. Future longitudinal studies might benefit from including these measures to directly investigate the influence that quantity and quality of reading and math instruction may have on individual differences in growth.

Summary and Conclusions

This study is one of the first to examine both within and between construct development in reading and math skills using a comprehensive series of univariate and multivariate analyses. Results suggested evidence for influences of broad learning domain and developmental time period on the growth of academic skills throughout the elementary-school period, although determining which specific contextual factors might be at play was beyond the scope of the current study. Although there was statistically significant and positive growth across the entire elementary-school period for most reading-related and math skills (i.e., all children were gaining skills), the presence of individual differences in growth was not consistent across time and skill domain (i.e., there were not always differences in rates of growth between children). Furthermore, multivariate growth models indicated that growth patterns across reading and math skills were generally uncoupled and dissimilar to each other. Evidence for domain-specific developmental trends suggested that the contextual factors that influence reading-related skills may differ from those that influence math. Future longitudinal studies of achievement should strive to include multiple academic domains, including reading and math, to examine whether or not these domain- and time-based differences can be further substantiated. Further evidence in support of these differences could have important implications for the manipulation of associated contextual factors and future intervention development. Understanding factors associated with growth may guide educators in ways to leverage these factors to help close achievement gaps for children at risk of reading or math difficulties.

Supplementary Material

Supplementary materials

Acknowledgements

This research was supported by grants from the Eunice Kennedy Shriver National Institute of Child Health and Human Development (HD052120) and the Institute of Education Sciences, United States Department of Education (R305F100027). The views expressed herein are those of the authors and have neither been reviewed nor approved by the granting agency.

Footnotes

1

Sensitivity analyses were conducted using only participants who had data at each time point and the results indicated that the overall trends observed using the full data set and FIML were similar to those indicated using a truncated data set.

References

  1. Anthony JL, & Lonigan CJ (2004). The nature of phonological awareness: Converging evidence from four studies of preschool and early grade school children. Journal of Educational Psychology, 96, 43–55. doi: 10.1037/0022-0663.96.1.43 [DOI] [Google Scholar]
  2. Aunola K, Leskinen E, Lerkkanen MK, & Nurmi JE (2004). Developmental dynamics of math performance from preschool to grade 2. Journal of Educational Psychology, 96, 699–713. doi: 10.1037/0022-0663.96.4.699 [DOI] [Google Scholar]
  3. Bailey DH, Watts TW, Littlefield AK, & Geary DC (2014). State and trait effects on individual differences in children’s mathematical development. Psychological Science, 25, 2017–2026. doi: 10.1177/0956797614547539 [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Bast J, & Reitsma P. (1998). Analyzing the development of individual differences in terms of Matthew effects in reading: results from a Dutch Longitudinal study. Developmental Psychology, 34, 1373–1399. doi: 10.1037/0012-1649.34.6.1373 [DOI] [PubMed] [Google Scholar]
  5. Baumert J, Nagy G, & Lehmann R. (2012). Cumulative advantages and the emergence of social and ethnic inequality: Matthew effects in reading and mathematics development within elementary schools?. Child Development, 83, 1347–1367. doi: 10.1111/j.1467-8624.2012.01779.x [DOI] [PubMed] [Google Scholar]
  6. Berkman ND, Sheridan SL, Donahue KE, Halpern DJ, & Crotty K. (2011). Low health literacy and health outcomes: an updated systematic review. Annals of Internal Medicine, 155, 97–107. doi: 10.7326/0003-4819-155-2-201107190-00005 [DOI] [PubMed] [Google Scholar]
  7. Bodovski K, & Farkas G. (2007). Mathematics growth in early elementary school: The roles of beginning knowledge, student engagement, and instruction. The Elementary School Journal, 108, 115–130. doi: 10.1086/525550 [DOI] [Google Scholar]
  8. Bradley RH, & Caldwell BM (1982). The consistency of the home environment and its relation to child development. International Journal of Behavioral Development, 5, 445–465. doi: 10.1177/016502548200500404 [DOI] [Google Scholar]
  9. Browne MW, Cudeck R, & Bollen KA (1993). Alternative ways of assessing model fit. In Bollen KA & Longs JS (Eds.) Testing structural equation models. (pp. 136–162). Newbury Park, CA: Sage. doi: 10.1177/0049124192021002005 [DOI] [Google Scholar]
  10. Byrne B, Olson R, & Samuelsson S. (in press). Behavior-genetic studies of literacy development: A commentary for professionals in psychology and education. In Kilpatrick D, Joshi M, & Wagner R. (Eds.), Reading Difficulties and Disabilities at School - Evidence-Based Assessment and Intervention Practice. (pp. 1–38). Springer, Cham. [Google Scholar]
  11. Campbell FA, Ramey CT, Pungello E, Sparling J. and Miller-Johnson S. 2002. Early childhood education: Young adult outcomes from the Abecedarian Project. Applied Developmental Science, 6(1): 42–57. [Google Scholar]
  12. Carreker SH, Neuhaus GF, Swank PR, Johnson P, Monfils MJ, & Montemayor ML (2007). Teachers with linguistically informed knowledge of reading subskills are associated with a Matthew effect in reading comprehension for monolingual and bilingual students. Reading Psychology, 28(2), 187–212. [Google Scholar]
  13. Chall JS (1983). Stages of reading development. New York, NW: McGraw-Hill. [Google Scholar]
  14. Clements DH, & Sarama J. (2014). Learning and teaching early math: The learning trajectories approach. Routledge. [Google Scholar]
  15. Daucourt MC, Erbeli F, Little CW, Haughbrook R, & Hart SA (2020). A meta-analytical review of the genetic and environmental correlations between reading and attention-deficit/hyperactivity disorder symptoms and reading and math. Scientific Studies of Reading, 24(1), 23–56. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Duncan TE, & Duncan SC (2004). An introduction to latent growth curve modeling. Behavior Therapy, 35, 333–363. doi: 10.1016/S0005-7894(04)80042-X [DOI] [Google Scholar]
  17. Engel M, Claessens A, & Finch MA (2013). Teaching students what they already know? The (mis) alignment between mathematics instructional content and student knowledge in kindergarten. Educational Evaluation and Policy Analysis, 35, 157–178. doi: 10.3102/0162373712461850 [DOI] [Google Scholar]
  18. Fantuzzo JW, Rouse HL, McDermott PA, & Sekino Y. (2005). Early childhood experiences and kindergarten success: A population-based study of a large urban setting. School Psychology Review, 34, 571–588. [Google Scholar]
  19. Friend A, DeFries JC, & Olson RK (2008). Parental education moderates genetic influences on reading disability. Psychological Science, 19, 1124–1130. doi: 10.1111/j.1467-9280.2008.02213.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Friend A, DeFries JC, Olson RK, Pennington B, Harlaar N, Byrne B, ... & Keenan JM (2009). Heritability of high reading ability and its interaction with parental education. Behavior Genetics, 39, 427–436. doi: 10.1007/s10519-009-9263-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Guarino CM, Hamilton LS, Lockwood JR, & Rathbun AH (2006). Teacher Qualifications, Instructional Practices, and Reading and Mathematics Gains of Kindergartners. Research and Development Report. NCES 2006–031. National Center for Education Statistics. [Google Scholar]
  22. Ginsburg H, & Baroody AJ (2003). TEMA-3: Test of Early Mathematics Ability. (3rd ed.). Austin, TX: Pro-ed. [Google Scholar]
  23. Gough PB, & Tunmer WE (1986). Decoding, reading, and reading disability. Remedial and Special Education, 7, 6–10. doi: 10.1177/074193258600700104 [DOI] [Google Scholar]
  24. Griffin EA, & Morrison FJ (1997). The unique contribution of home literacy environment to differences in early literacy skills 1. Early Child Development and Care, 127, 233–243. doi: 10.1080/0300443971270119 [DOI] [Google Scholar]
  25. Harm MW, & Seidenberg MS (2004). Computing the meanings of words in reading: cooperative division of labor between visual and phonological processes. Psychological Review, 111, 662–720. doi: 10.1037/0033-295X.111.3.662 [DOI] [PubMed] [Google Scholar]
  26. Hart B, & Risley TR (2003). The early catastrophe: The 30 million word gap by age 3. American Educator, 27, 4–9. [Google Scholar]
  27. Hart SA, Ganley CM, & Purpura DJ (2016). Understanding the Home Math Environment and Its Role in Predicting Parent Report of Children’s Math Skills. PloS ONE, 11: e0168227. doi: 10.1371/journal.pone.0168227 [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Hart SA, Soden B, Johnson W, Schatschneider C, & Taylor J. (2013). Expanding the environment: gene× school-level SES interaction on reading comprehension. Journal of Child Psychology and Psychiatry, 54, 1047–1055. doi: 10.1111/jcpp.12083 [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Hart SA, Petrill SA, Thompson LA, & Plomin R. (2009). The ABCs of math: A genetic analysis of mathematics and its links with reading ability and general cognitive ability. Journal of Educational Psychology, 101, 388–402. doi: 10.1037/a0015115 [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Hart S, Little C, & van Bergen E. (2019). Nurture might be nature: Cautionary tales and proposed solutions. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Haworth C, Kovas Y, Harlaar N, Hayiou-Thomas ME, Petrill SA, Dale PS, & Plomin R. (2009). Generalist genes and learning disabilities: a multivariate genetic analysis of low performance in reading, mathematics, language and general cognitive ability in a sample of 8000 12-year-old twins. Journal of Child Psychology and Psychiatry, 50, 1318–1325. doi: 10.1111/j.1469-7610.2009.02114.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Hooper D, Coughlan J, & Mullen M. (2008). Structural equation modelling: Guidelines for determining model fit. Journal of Business Research Methods, 6, 53–60. doi: 10.21427/D7CF7R [DOI] [Google Scholar]
  33. Hu LT, & Bentler PM (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling: A Multidisciplinary Journal, 6, 1–55. doi: 10.1080/10705519909540118 [DOI] [Google Scholar]
  34. Jacobs JE, Lanza S, Osgood DW, Eccles JS, & Wigfield A. (2002). Changes in children’s self-competence and values: Gender and domain differences across grades one through twelve. Child development, 73(2), 509–527. [DOI] [PubMed] [Google Scholar]
  35. Johnson AD, Martin A, Brooks-Gunn J, Petrill SA (2008). Order in the house! Associations among household chaos, the home literacy environment, maternal reading ability, and children’s early reading. Merrill-Palmer Quarterly. 54,445–472. doi: 10.1353/mpq.0.0009 [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Jordan NC, Kaplan D, & Hanich LB (2002). Achievement growth in children with learning difficulties in mathematics: Findings of a two-year longitudinal study. Journal of Educational Psychology, 94, 586–597. doi: 10.1037//0022-0663.94.3.586 [DOI] [Google Scholar]
  37. Kline RB (2011). Convergence of structural equation modeling and multilevel modeling. In Williams M. (Ed.), Handbook of Methodological Innovation. (pp. 562–589). Thousand Oaks, CA: Sage. doi: 10.4135/9781446268261.n31 [DOI] [Google Scholar]
  38. Kim YS, Petscher Y, Schatschneider C, & Foorman B. (2010). Does growth rate in oral reading fluency matter in predicting reading comprehension achievement?. Journal of Educational Psychology, 102, 652–667. doi: 10.1037/a0019643 [DOI] [Google Scholar]
  39. Krajewski K, & Schneider W. (2009). Early development of quantity to number-word linkage as a precursor of mathematical school achievement and mathematical difficulties: Findings from a four-year longitudinal study. Learning and Instruction, 19, 513–526. doi: 10.1016/j.learninstruc.2008.10.002 [DOI] [Google Scholar]
  40. Landerl K, Fussenegger B, Moll K, & Willburger E. (2009). Dyslexia and dyscalculia: Two learning disorders with different cognitive profiles. Journal of experimental child psychology, 103(3), 309–324. [DOI] [PubMed] [Google Scholar]
  41. Lee VE, & Burkham DT (2002). Inequality at the starting gate: Social background differences in achievement as children begin kindergarten. Washington, DC: Economic Policy Institute. [Google Scholar]
  42. Lee J, & Reeves T. (2012). Revisiting the impact of NCLB high-stakes school accountability, capacity, and resources: State NAEP 1990–2009 reading and math achievement gaps and trends. Educational Evaluation and Policy Analysis, 34, 209–231. doi: 10.3102/0162373711431604 [DOI] [Google Scholar]
  43. Lenchner O, Gerber MM, & Routh DK (1990). Phonological awareness tasks as predictors of decoding ability: Beyond segmentation. Journal of Learning Disabilities, 23, 240–247. doi: 10.1177/002221949002300407 [DOI] [PubMed] [Google Scholar]
  44. Lesaux NK (2012). Reading and reading instruction for children from low-income and non-English-speaking households. The Future of Children, 22, 73–88. doi: 10.1353/foc.2012.0010 [DOI] [PubMed] [Google Scholar]
  45. Little CW, Hart SA, Quinn JM, Tucker-Drob EM, Taylor J, & Schatschneider C. (2016). Exploring the Co-Development of Reading Fluency and Reading Comprehension: A Twin Study. Child Development, 88, 934–945. doi: 10.1111/cdev.12670 [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. LoGerfo L, Nichols A, & Reardon SF (2006). Achievement gains in elementary and high school. Washington D.C.: Urban Institute. Retrieved from: http://www.urban.org/ [Google Scholar]
  47. Mazzocco MM, & Thompson RE (2005). Kindergarten predictors of math learning disability. Learning Disabilities Research & Practice, 20(3), 142–155. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. McArdle JJ, Bell RQ. (2000). An introduction to latent growth curve models for developmental data analysis. In: Little TD, Schnabel KU, Baumert J, (Eds.) Modeling longitudinal and multiple-group data: Practical issues, applied approaches, and scientific examples. (pp. 69–107). Mahwah, NJ: Erlbaum. doi: 10.1016/S0005-7894(04)80042-X [DOI] [Google Scholar]
  49. McCoach DB, O’connell AA, Reis SM, & Levitt HA (2006). Growing readers: A hierarchical linear model of children’s reading growth during the first 2 years of school. Journal of Educational Psychology, 98, 14–28. doi: 10.1037/0022-0663.98.1.14 [DOI] [Google Scholar]
  50. Moos RH (1990). Conceptual and empirical approaches to developing family-based assessment procedures: Resolving the case of the Family Environment Scale. Family Process, 29, 199–208. doi: 10.1111/j.1545-5300.1990.00199.x [DOI] [PubMed] [Google Scholar]
  51. Moos RH, & Moos BS (1994). Family environment scale manual. Palo Alto, CA: Consulting Psychologists Press. [Google Scholar]
  52. Morgan PL, Farkas G, & Wu Q. (2011). Kindergarten children’s growth trajectories in reading and mathematics: Who falls increasingly behind?. Journal of Learning Disabilities, 44, 472–488. doi: 10.1177/0022219411414010 [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Muthén LK, & Muthén BO (2012). Mplus Version 7 user’s guide. Los Angeles, CA: Muthén & Muthén. [Google Scholar]
  54. Parrila R, Aunola K, Leskinen E, Nurmi JE, & Kirby JR (2005). Development of individual differences in reading: Results from longitudinal studies in English and Finnish. Journal of Educational Psychology, 97, 299–319. doi: 10.1037/0022-0663.97.3.299 [DOI] [Google Scholar]
  55. Pennington BF, McGrath LM, Rosenberg J, Barnard H, Smith SD, Willcutt EG, ... & Olson RK (2009). Gene× environment interactions in reading disability and attention deficit/hyperactivity disorder. Developmental Psychology, 45, 77–99. doi: 10.1037/a0014549 [DOI] [PMC free article] [PubMed] [Google Scholar]
  56. Petrill SA, Pike A, Price T, & Plomin R. (2004). Chaos in the home and socioeconomic status are associated with cognitive development in early childhood: Environmental mediators identified in a genetic design. Intelligence, 32, 445–460. doi: 10.1016/j.intell.2004.06.010 [DOI] [Google Scholar]
  57. Petrill SA, Hart SA, Harlaar N, Logan J, Justice LM, Schatschneider C, ... & Cutting L. (2010). Genetic and environmental influences on the growth of early reading skills. Journal of Child Psychology and Psychiatry, 51, 660–667. doi: 10.1111/j.1469-7610.2009.02204.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. Pfost M, Hattie J, Dörfler T, & Artelt C. (2014). Individual differences in reading development: A review of 25 years of empirical research on Matthew effects in reading. Review of Educational Research, 84, 203–244. doi: 10.3102/0034654313509492 [DOI] [Google Scholar]
  59. Pianta RC, Belsky J, Vandergrift N, Houts R, & Morrison FJ (2008). Classroom effects on children’s achievement trajectories in elementary school. American Educational Research Journal, 45, 365–397. doi: 10.3102/0002831207308230 [DOI] [Google Scholar]
  60. Plomin R, & Bergeman CS (1991). The nature of nurture: Genetic influence on “environmental” measures. Behavioral and Brain Sciences, 14, 373–386. doi: 10.1017/S0140525X00070278 [DOI] [Google Scholar]
  61. Purpura DJ, Schmitt SA, & Ganley CM (2017). Foundations of mathematics and literacy: The role of executive functioning components. Journal of Experimental Child Psychology, 153, 15–34. doi: 10.1016/j.jecp.2016.08.010 [DOI] [PubMed] [Google Scholar]
  62. Purpura DJ, Baroody AJ, & Lonigan CJ (2013). The transition from informal to formal mathematical knowledge: Mediation by numeral knowledge. Journal of Educational Psychology, 105, 453–464. doi: 10.1037/a0031753 [DOI] [Google Scholar]
  63. Reyna VF, Nelson WL, Han PK, & Dieckmann NF (2009). How numeracy influences risk comprehension and medical decision making. Psychological Bulletin, 135, 943–973 doi: 10.1037/a0017327 [DOI] [PMC free article] [PubMed] [Google Scholar]
  64. Ritchie SJ, & Bates TC (2013). Enduring links from childhood mathematics and reading achievement to adult socioeconomic status. Psychological Science, 24, 1301–1308. doi: 10.1177/0956797612466268 [DOI] [PubMed] [Google Scholar]
  65. Rivera SM, Reiss AL, Eckert MA, & Menon V. (2005). Developmental changes in mental arithmetic: Evidence for increased functional specialization in the left inferior parietal cortex. Cerebral Cortex, 75, 1779–1790. doi: 10.1093/cercor/bhi055 [DOI] [PubMed] [Google Scholar]
  66. Salaschek M, Zeuch N, & Souvignier E. (2014). Mathematics growth trajectories in first grade: Cumulative vs. compensatory patterns and the role of number sense. Learning and Individual Differences, 35, 103–112. doi: 10.1016/j.lindif.2014.06.009 [DOI] [Google Scholar]
  67. Samuelsson S, Byrne B, Olson RK, Hulslander J, Wadsworth S, Corley R, ... & DeFries JC (2008). Response to early literacy instruction in the United States, Australia, and Scandinavia: A behavioral-genetic analysis. Learning and Individual Differences, 18, 289–295. doi: 10.1016/j.lindif.2008.03.004 [DOI] [PMC free article] [PubMed] [Google Scholar]
  68. Sarama J, & Clements DH (2009). Early childhood mathematics education research: Learning trajectories for young children. New York, NY: Routledge. doi: 10.4324/9780203883785 [DOI] [Google Scholar]
  69. Institute SAS (2014). Base SAS 9.4 Procedures Guide. SAS Institute. [Google Scholar]
  70. Scarr S, & McCartney K. (1983). How people make their own environments: A theory of genotype→ environment effects. Child Development, 424–435. [DOI] [PubMed] [Google Scholar]
  71. Shanahan MA, Pennington BF, Yerys BE, Scott A, Boada R, Willcutt EG, ... & DeFries JC (2006). Processing speed deficits in attention deficit/hyperactivity disorder and reading disability. Journal of Abnormal Child Psychology, 34, 585–602. doi: 10.1007/s10802-006-9037-8 [DOI] [PubMed] [Google Scholar]
  72. Shaw P, Greenstein D, Lerch J, Clasen L, Lenroot R, Gogtay NEEA, … & Giedd J. (2006). Intellectual ability and cortical development in children and adolescents. Nature, 440, 676–679. doi: 10.1038/nature04513 [DOI] [PubMed] [Google Scholar]
  73. Siegler RS, & Lortie-Forgues H. (2014). An integrative theory of numerical development. Child Development Perspectives, 8(3), 144–150. [Google Scholar]
  74. Soden B, Christopher ME, Hulslander J, Olson RK, Cutting L, Keenan JM, … & Petrill SA (2015). Longitudinal stability in reading comprehension is largely heritable from grades 1 to 6. PloS ONE, 10, e0113807. doi: 10.1371/journal.pone.0113807 [DOI] [PMC free article] [PubMed] [Google Scholar]
  75. Sowell ER, Thompson PM, Leonard CM, Welcome SE, Kan E, & Toga AW (2004). Longitudinal mapping of cortical thickness and brain growth in normal children. Journal of Neuroscience, 24, 8223–8231. doi: 10.1523/JNEUROSCI.1798-04.2004 [DOI] [PMC free article] [PubMed] [Google Scholar]
  76. Sirin SR (2005). Socioeconomic status and academic achievement: A meta-analytic review of research. Review of Educational Research, 75, 417–453. doi: 10.3102/00346543075003417 [DOI] [Google Scholar]
  77. Sonnenschein S, Stapleton LM, & Benson A. (2010). The relation between the type and amount of instruction and growth in children’s reading competencies. American Educational Research Journal, 47, 358–389. doi: 10.3102/0002831209349215 [DOI] [Google Scholar]
  78. Sowell ER, Thompson PM, Leonard CM, Welcome SE, Kan E, & Toga AW (2004). Longitudinal mapping of cortical thickness and brain growth in normal children. Journal of Neuroscience, 24, 8223–8231. doi: 10.1523/JNEUROSCI.1798-04.2004 [DOI] [PMC free article] [PubMed] [Google Scholar]
  79. Stahl SA, & Heubach KM (2005). Fluency-oriented reading instruction. Journal of Literacy Research, 37, 25–60. doi: 10.1207/s15548430jlr3701_2 [DOI] [Google Scholar]
  80. Stanovich KE (1986). Matthew effects in reading: Some consequences of individual differences in the acquisition of literacy. Reading Research Quarterly, 21, 360–407. doi: 10.1177/0022057409189001-204 [DOI] [Google Scholar]
  81. Taylor J, & Hart SA (2014). A chaotic home environment accounts for the association between respect for rules disposition and reading comprehension: A twin study. Learning and Individual Differences, 35, 70–77. doi: 10.1016/j.lindif.2014.07.020 [DOI] [PMC free article] [PubMed] [Google Scholar]
  82. Taylor J, Roehrig AD, Hensler BS, Connor CM, & Schatschneider C. (2010). Teacher quality moderates the genetic effects on early reading. Science, 328, 512–514. doi: 10.1126/science.1186149 [DOI] [PMC free article] [PubMed] [Google Scholar]
  83. Torgesen JK, Wagner RK, Rashotte CA, Rose E, Lindamood P, Conway T, & Garvan C. (1999). Preventing reading failure in young children with phonological processing disabilities: Group and individual responses to instruction. Journal of Educational Psychology, 91, 579–593. doi: 10.1037/0022-0663.91.4.579 [DOI] [Google Scholar]
  84. Tucker-Drob EM (2012). Preschools reduce early academic achievement gaps: A longitudinal twin approach. Psychological Science, 23, 310–319. doi: 10.1177/0956797611426728 [DOI] [PMC free article] [PubMed] [Google Scholar]
  85. Vogler GP, DeFries JC, & Decker SN (1985). Family history as an indicator of risk for reading disability. Journal of Learning Disabilities, 18, 419–421. doi: 10.1177/002221948501800711 [DOI] [PubMed] [Google Scholar]
  86. Wagner RK, Torgesen JK, Rashotte CA, & Pearson NA (2010). Test of Silent Reading Efficiency and Comprehension. Austin, TX: Pro-Ed. [Google Scholar]
  87. Wagner RK, Torgesen JK, & Rashotte CA (1999). CTOPP: Comprehensive Test of Phonological Processing. Austin, TX: Pro-Ed. [Google Scholar]
  88. Wagner RK, Torgesen JK, Rashotte CA, Hecht SA, Barker TA, Burgess SR, … & 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. doi: 10.1037/0012-1649.33.3.468 [DOI] [PubMed] [Google Scholar]
  89. Woodcock RW (1998). Woodcock Reading Mastery Tests-Revised, examiner’s manual. Circle Pines, MN: American Guidance Service. [Google Scholar]
  90. Woodcock RW, McGrew KS, & Mather N. (2001). Woodcock-Johnson Tests of Achievement. Itasca, IL: Riverside Publishing. [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary materials

RESOURCES