The Role of Language and Literacy Skills in Science Learning from Kindergarten to 5th Grade: Mitigating Gender, Racial/Ethnic, and Socio-Economic Disparities

Abstract

Despite the acknowledged impact of early science achievement on future success, there is a noticeable gap in research focused on understanding the dynamic longitudinal patterns of children attaining science learning milestones in their early years, as well as few investigations of potential factors that may mitigate gender, racial/ethnic, and socio-economic disparities. This study analyzed nationally representative data from the Early Childhood Longitudinal Study of Kindergarten Cohort 2011 (ECLS-K: 2011), involving 18,174 children from 1328 schools across the United States, selected through a multistage probability sampling process and spanning kindergarten through 5th grade. Using survival analysis with the dependent variable as the time required to attain a specific milestone, the study revealed that boys, non-Black, non-Hispanic, and high-income children reached the science learning threshold earlier than their counterparts—girls, Black, Hispanic, and low-income children. Furthermore, the study underscored the crucial role of language and literacy skills in mitigating these disparities. The study’s implications stress the importance of targeted interventions to address early science education disparities, emphasizing professional development and integrating language and literacy with science learning. The research also enriches the global discourse on educational standards by introducing innovative methodologies to assess both the frequency and duration of science learning milestones.

1. Introduction

Enhancing the scientific proficiency of children in the United States has long been a priority for policymakers and researchers [1,2]. To facilitate science education, states have either adopted the Next Generation Science Standards (NGSS) [3] or incorporated these standards into their educational frameworks [4]. While the national science standards encompass education from kindergarten through 12th grade, existing research in science education tends to concentrate on upper elementary and secondary education settings [5].

A burgeoning body of research underscores the pivotal role of early childhood science experiences in augmenting children’s knowledge, skills, and attitudes essential for future employment, equipping them for an economy that necessitates innovative solutions to intricate problems [6,7,8]. Chesloff [9], for instance, advocates for starting science education in early childhood, contending that fundamental concepts central to science, technology, engineering, and mathematics (STEM)—curiosity, creativity, collaboration, and critical thinking—develop early. The National Research Council (NRC) [10] also emphasizes the critical significance of robust early science education spanning from kindergarten to 3rd grade.

Additionally, educators increasingly acknowledge the importance of understanding the diverse learning needs and strengths of all children in the realm of science education. The NGSS [3] and the Framework for K-12 Science Education [10] aspire to prompt equitable and high-quality science education experiences for every child. Nevertheless, escalating concern revolves around the observed disparities in science achievement, particularly among female, economically disadvantaged, and racial/ethnic minority children, as contrasted with their male, economically privileged, and majority counterparts, which carry significant social, educational, and political implications [11,12,13].

In this context, it is not surprising that the STEM workforce consistently exhibits underrepresentation based on specific demographic characteristics. Investigations into the “STEM pipeline”, reveal a discernible pattern, with female students, racial/ethnic minorities, especially Black and Hispanic students, and those from low-income families, less likely to pursue a STEM major in college. This pattern in turn relates to variations in earlier preparation, academic achievement, and attitudes toward science evident during their school years [11,14,15,16].

Researchers have identified key factors contributing to the underrepresentation of these demographic groups in STEM. First, gender gaps can be attributed to differences in cognitive ability, relative cognitive strengths, and career preferences influenced by biology [17,18]. However, societal beliefs and gender-specific expectations, along with cultural pressures, are deemed more influential than biology alone in shaping career decisions [19]. Second, socioeconomic disparities arise from inherent opportunity gaps, as children from poverty often contend with low-quality educational experiences and are likely to attend underfunded schools with less-experienced teachers and lower expectations [20,21]. Third, racial/ethnic disparities persist in many forms for various reasons. For example, Black and Hispanic students are more likely to depart or change majors to non-STEM disciplines, which is more prevalent for STEM than non-STEM fields [15]. Researchers point to the role of stereotypes, microaggressions, and lack of value and support as key factors contributing to the racial/ethnic disparity in STEM [22,23].

The limited attention given to these disparities has primarily focused on middle and high school, with scant studies documenting the onset of these gaps in the early stages, from kindergarten to primary grades [13]. Therefore, it is imperative to document the early emergence of gender, racial/ethnic, and socioeconomic gaps in science achievement and investigate the array of factors that might mitigate these disparities.

Traditionally, equity in science education has primarily concentrated on providing opportunities and access for underrepresented groups. However, contemporary research highlights effective strategies for diminishing gaps in science achievement through cross-domain learning [4]. For example, engagement in language and literacy activities proves instrumental in supporting children’s science learning given the highly communicative nature of the discipline and its inseparable connection with reading and writing [12,24,25,26]. Language and literacy become integral components of science itself [27], with reading, writing, hearing, and talking science constituting significant aspects of professional scientific endeavors [28]. Notably, some evidence suggests that literacy skills may exert a more substantial influence on science achievement than mathematics skills [29]. Correspondingly, studies investigating the effectiveness of enhancing literacy skills and strategies demonstrate positive outcomes for understanding science concepts [12,25,30,31].

Furthermore, providing support for spoken and written language fosters equitable participation in content-area learning, particularly benefiting children requiring assistance with language and literacy skills [28,32]. The nature of science learning involves the organization, communication, and sharing of ideas within a community of learners [3,10]. As the expression of ideas can take both oral and written forms, the development of language and literacy is crucial for active participation in science learning [12,27,33]. This importance is heightened for young children who are yet to achieve independent reading and writing skills, as both oral and written discourse serves as a supportive mechanism for engagement in the discipline of science [34].

The significance of language and literacy in science learning is emphasized in national standards documents [4]. The NRC framework [35] advocates for the development of science activities aligned with those undertaken by professional scientists and engineers, including reading science content and engaging in specialized modes of communication through talking and writing. The NGSS [3] similarly underscores the importance of communication through oral and written language, encompassing activities such as reading, writing, using discipline-specific texts, explaining, arguing, discussing, asking, and describing.

While existing studies have explored the connection between language and literacy skills and science learning [12,24,25], they have yet to provide compelling evidence about the specific factors through which language and literacy undergird science learning, especially in endeavors focused on reducing disparities in science achievement [4]. Consequently, this study aimed to investigate the developmental characteristics of children that impact science achievement and the role of language and literacy in addressing disparities across gender, racial/ethnic, and socio-economic backgrounds. Our research sought to compensate for the limitations inherent in previous research by providing a more comprehensive view of the complex interplay between language and literacy skills and science learning.

This study is grounded in developmental and sociolinguistic theories [36,37,38]. First, the Ecological Systems Theory posits that individual development is shaped by context—thus, the uneven distribution of opportunities for early skill development among children is highlighted [38]. Consequently, our hypothesis suggests that the systematic disparities from these contextual differences may contribute significantly to the observed science achievement gaps across diverse groups of children.

Second, this study is informed by the Systemic Linguistics Approach, which underscores the significance of oral and written language for learning across all schooling and disciplines [32,37,38,39,40]. Functional linguists have illustrated how children, through their lexical choices encompassing both general and technical academic language, manifest their proficiency in adopting the language specific to academic fields, thereby positioning themselves as members of distinct discourse communities [37,38,39,40]. Children equipped with knowledge about the available linguistic resources in a given context can make more informed choices when engaging with disciplinary texts [36,38]. In this regard, especially during the transition from instruction grounded in social language to academically oriented teaching, a shift typically experienced in early childhood and elementary school grades, children commonly acquire the essential literacy skills for meaningful participation [41]. As science instruction undergoes a shift towards greater formality and systematization during this transitional period in contrast to earlier years [42], children with inadequate literacy development may confront academic challenges hindering their optimal participation in science learning [4,12,30]. Accordingly, this study hypothesizes that promoting language and literacy skills could be a promising support for children encountering difficulties in science learning.

Implementing the national science standards sparks inquiries into how well children’s performance aligns with these expectations. Methodologically, a prevalent approach entails using mean or standard scores on assessments. Nonetheless, these metrics inadequately address the essential question of whether the observed norms accurately represent children attaining the intended level of performance. Another technique includes categorizing performance levels and determining whether a child is “exceeding the learning threshold” or “falling short of the learning threshold” [43].

Despite the growing interest in studying children’s longitudinal learning dynamics, many studies have not used statistical methods that accurately measure the timing of attainments. In developmental threshold analysis, data are frequently aggregated over time, leading to the formation of summary variables (e.g., averages) and the use of static statistical approaches (e.g., logistic regression). Unfortunately, this methodology conceals the dynamic longitudinal processes at play [43,44]. As a result, a notable gap exists between conceptual inquiries regarding children’s academic achievement aligning with national standards or expectations and the methods employed to investigate these inquiries.

This study used survival analysis to identify subgroups of children at a particular age exhibiting superior achievement and to uncover factors that mitigate disparities. Survival analysis comprises a set of statistical techniques where the dependent variable is the time required to attain a specified threshold [45]. Examining the timing of surpassing the learning threshold offers insights into science achievement as a within-child process rather than merely differences between children [43,44], which is the more typical approach.

In the current study, the primary goals were to examine the significance of language and literacy within the framework of science learning as outlined in the NGSS standards [3] and to address disparities in science achievement across gender, racial/ethnic, and socio-economic backgrounds, using survival analysis. This study focused on two research questions: (1) How did patterns and variations in children’s attainment of the science learning threshold evolve over time? (2) What characteristics of children were associated with exceeding the science learning threshold? The hypotheses were as follows: girls, Black and Hispanic children, and children from low-income backgrounds would attain the science learning threshold later than boys, non-Black and non-Hispanic children, and children from high-income households. Proficiency in language and literacy was expected to enhance science learning, particularly narrowing the learning disparity between boys and girls, Black and non-Black children, Hispanic and non-Hispanic children, and children from low-income and high-income backgrounds.

2. Materials and Methods

2.1. Dataset

This study used data from the Early Childhood Longitudinal Survey-Kindergarten Class of 2010–2011 (ECLS-K: 2011) dataset, managed by the National Center for Educational Statistics (NCES) within the U.S. Department of Education’s Institute of Education Sciences. The ECLS-K: 2011 comprises a nationally representative sample of 18,174 children across 1328 schools, selected through a sophisticated multistage probability sampling process. Parental written informed consent was secured [46]. The data collection process began in 2010, coinciding with the children’s entry into kindergarten, and continued until 2015, marking the conclusion of their 5th-grade year. In each collection period, individual children completed direct assessments of their academic skills.

The sampled children represented a diverse demographic with roughly equal representation of girls and boys at 49% females and 51% males. The race/ethnic distribution was as follows: 1% American Indian/Alaska Native and Native American/Pacific Islander, 8.5% Asian, 13% Black, 53% Caucasian, 25.3% Hispanic, and 4% two or more races. Regarding home language, 80.6% were English speakers, 18.3% were speakers of languages other than English, and 1.1% reported using multiple languages at home without specifying a primary language. Socioeconomic status (SES) levels were categorized as follows: 21.3% were growing up in households reporting incomes below 100% of the federal poverty level, 23.7% in homes with incomes between 100 and 199% of the federal poverty level, and 54.9% in homes reporting incomes at or above 200% of the federal poverty level.

2.2. Measures

2.2.1. Reading

The reading (language and literacy) assessment evaluated fundamental language and literacy skills, including print familiarity, letter recognition, proficiency in identifying beginning and ending sounds, recognition of common words (sight vocabulary), decoding multisyllabic words, vocabulary knowledge, and reading comprehension. The reading comprehension component included tasks such as identifying explicit information, making nuanced inferences, and critically evaluating text appropriateness and quality. The assessments covered diverse literary genres, and the process began with 12 routing items to determine the difficulty level for subsequent assignments. This measure was constructed for the ECLS-K [46].

2.2.2. Science

The science assessment evaluated knowledge and skills in earth and space science, physical science, life science, and scientific inquiry. Following 15 routing items, children were assigned one of three second-stage forms based on their responses, similar to the reading assessment. To ensure fairness, questions, response options, and any visible text (e.g., graph labels) were read aloud, minimizing the impact of reading skills on science assessment scores. This metric was also created for the ECLS-K [46]. The empirically determined science cut point serves as a benchmark for assessing a child’s success in science [47].

2.2.3. Externalizing Behavior

The study employed teacher ratings to assess children’s externalizing problem behaviors, with teachers indicating the frequency of specific social skills using a four-option scale ranging from “Never” to “Very Often”. Teachers could report the absence of the described behavior for a particular child. The assessment items were adapted from the Social Skills Rating System [48], and higher scores corresponded to a more frequent display of the behavior on the scale [46].

2.3. Analytic Strategy

Research Question 1 aimed to investigate the evolving patterns and variations in children’s attainment of the science learning threshold over time. This was accomplished through the application of the Kaplan–Meier survival analysis method [49,50]. Research Question 2 concentrated on identifying the characteristics of children associated with performance exceeding the established science learning threshold. To address this, a multilevel Cox regression analysis [51,52] was employed. Further information on these analyses is provided below.

Many studies on developmental thresholds explore the association between a variable measured at a specific time and attaining or surpassing a developmental threshold in a subsequent period. However, this approach lacks insights into when attainment of the threshold is most likely and whether early and late achievers differ [43]. Instead of treating threshold attainment as a binary switch from “stayer” to “achiever”, researchers suggest that the time taken to surpass the threshold relates to understanding children’s academic achievement [43,53,54]. The current study aimed to deepen the understanding of children’s developmental threshold attainment by incorporating the timing of reaching a learning milestone into the analysis. Specifically, the empirically determined science cut point [47] is a benchmark for assessing a child’s achievement or success.

Through the application of survival analysis, this study effectively captured the evolving relationship between child developmental characteristics and the attainment of the science learning threshold. It offers a statistical assessment of the changing probabilities associated with science achievement over time. Survival analysis [45,50] estimates the conditional probability of science achievement, treating surpassing the threshold as a time-dependent variable that changes its state depending on the duration taken to achieve the threshold in science learning.

In the study, the primary outcome is the duration until attaining the science learning threshold. If all children reached the threshold, various statistical methods could be applied. However, it is common for some children not to attain the event of interest by the study’s conclusion, leaving their survival times unknown. This phenomenon, known as censoring, may occur if a child has not achieved the threshold by the final semester of data collection in the spring of 2016. Censored survival times can either underestimate or overestimate the actual (yet unknown) time to the event [45,50].

When depicting the survival trajectory of certain children on a timeline (Figure 1), it becomes apparent that their experiences extend beyond the study period. For instance, Child A was monitored from the fall of 2010 until encountering the event (i.e., surpassing the science learning threshold) in the spring of 2012. Consequently, the child’s survival time spans four semesters and is not censored. On the other hand, Child B—also observed since the fall of 2010—was tracked until the conclusion of the study period without reaching the event; essentially, the child fails to attain the science learning threshold. In this case, the child’s survival time amounts to 11 semesters and is censored, specifically right-censored. This unique aspect of longitudinal data underscores the necessity of employing specialized probabilistic techniques, such as survival analysis [53,54].

The nonparametric estimation of survival probability using the Kaplan–Meier (KM) method [49,50] is derived from observed survival times. The $S\left(t\right)$ estimate signifies the probability or proportion of the study population anticipated to persist (i.e., not attaining a developmental threshold) beyond the $t$-th semester. In contrast to conventional methods that exclude censored data from analysis, the Kaplan–Meier method accommodates censored data, allowing for the retention of partial information derived from their inclusion. To address the first research question, which sought to uncover patterns and variations in children’s attainment of the science threshold over time, the Kaplan–Meier method of survival analysis was employed.

The second research question, examining the characteristics of children associated with enhancing their attainment of the science learning threshold, was investigated using multilevel Cox regression [51,52]. Using survival analysis, this study could scrutinize the timing of children surpassing the science learning threshold as a within-child process. To investigate the timing of children’s attainment of the developmental threshold, this study adopted a multilevel survival analysis to accommodate the hierarchical structure of children within schools. In multilevel survival analysis, a random effect was employed to capture the variability of the hazard function across different schools. Child characteristics were categorized as time-invariant (e.g., gender) or time-varying (e.g., literacy skills) and tested to ascertain their association with achieving the science learning threshold.

Parameter estimates were converted into the more easily interpreted hazard ratio metric, denoted as $HR=exp\left[\beta \right]$ [50]. In Equation (1), $h_{ij}\left(t\right)$, representing the hazard of surpassing the threshold for child $j$ within each school $i$ at semester $t$, is computed as a function of $h_0\left(t\right)$, the baseline hazard function for each child, $exp\left(v_i\right)$, an exponentiated school-specific random effect, and the exponentiated linear function of the time-varying $X_{ij}\left(t\right)$ and time-invariant predictors $X_j$.

$$\begin{array}{cc}\hfill h_{ij}\left(t\right)=h_0\left(t\right)\times & exp\left(v_i\right)\times \hfill \\ \hfill & exp({\beta }_1{Age}_{ij}\left(t\right)+{\beta }_2{Male}_j\left(t\right)+{\beta }_3{Black}_j\left(t\right)+{\beta }_4{Hispanic}_j\left(t\right)\hfill \\ \hfill & +{\beta }_5{SES}_{ij}\left(t\right)+{\beta }_6{Externalizing Behaviors}_{ij}\left(t\right)\hfill \\ \hfill & +{\beta }_7{Literacy Skills}_{ij}\left(t\right))\hfill \end{array}$$

(1)

3. Results

The following discussion illustrates how survival analysis can be employed to examine patterns and variations in the time it takes for children to attain a specific developmental milestone. Additionally, potential relationships between various child characteristics and the duration required to achieve the desired outcome will be explored.

3.1. Patterns and Variations in Children’s Surpassing the Threshold

The estimate of $S\left(t\right)$, visually represented in Figure 2, delineates the probabilities of children anticipated to remain below the science learning threshold beyond the $t$-th semester. The graph exhibits a consistent decrease, reflecting cumulative probabilities. This implies that the probabilities of staying below the threshold for $t$ semesters depend on having already persisted for $t-1$ semesters. The Kaplan–Meier curve is characterized by stepwise estimates rather than smooth functions, a distinctive feature of non-continuous presentation [49,50]. The vertical distances between horizontals hold significance, illustrating the shifts in the cumulative probability of remaining below the threshold at different points along the curve. Censoring exerts an impact on survival rates, with censored observations typically positioned immediately after the event. Upon adjusting for censoring, it was found that the probability of children remaining below the threshold beyond two years (1st grade), was 99%, beyond three years (2nd grade), was 95%, beyond four years (3rd grade), was 87%, beyond 5 years (4th grade) was 68%, and beyond six years (5th grade) was 42% (

Table 1. Survival Probability.

Semester	Survival Probability	SE	Lower 95% CI	Upper 95% CI
K Spring	1	-	-	-
G1 Fall	1	-	-	-
G1 Spring	0.99	0.00	0.99	1.00
G2 Fall	0.98	0.00	0.97	0.98
G2 Spring	0.95	0.00	0.95	0.96
G3 Spring	0.87	0.00	0.86	0.88
G4 Spring	0.68	0.00	0.67	0.70
G5 Spring	0.42	0.00	0.40	0.43

).

To evaluate whether certain children were more prone to staying below the threshold, Kaplan–Meier models were developed, facilitating a comparison of overall survival rates between two groups. To enhance understanding, the study graphically represented the survival functions and conducted a visual comparison for further analysis. The log-rank test, spanning the entire follow-up period, indicated that non-Black children exhibited a greater likelihood of surpassing the science learning threshold earlier than Black children (${\chi }^2=64.4, df=1, p<0.0001$), and similarly, non-Hispanic children demonstrated a higher probability of exceeding the science learning threshold earlier than Hispanic children (${\chi }^2=266, df=1, p<0.0001$).

Upon examining the disparities in science learning across different racial and ethnic groups, the data underscore substantial gaps that emerge as early as the 2nd grade (Figure 3). The findings indicate a 95% probability for non-Black children and a 97% likelihood for Black children to persist below the threshold beyond the 2nd grade. This discrepancy intensifies in subsequent grades, with an 85% likelihood for non-Black children and a 93% likelihood for Black children to remain below the threshold beyond the 3rd grade. By the 4th grade, the probabilities decrease to 66% for non-Black children and 82% for Black children, highlighting the growing disparities. By the 5th grade, the likelihoods further decrease to 39% for non-Black children and 58% for Black children (

Table 2. Survival probability: Black vs. non-Black.

Race/Ethnicity	Semester	Survival Probability	SE	Lower 95% CI	Upper 95% CI
Black	K Spring	1	-	-	-
	G1 Fall	1	-	-	-
	G1 Spring	1	-	-	-
	G2 Fall	0.98	0.00	0.97	0.99
	G2 Spring	0.97	0.00	0.95	0.98
	G3 Spring	0.93	0.01	0.91	0.96
	G4 Spring	0.82	0.01	0.78	0.85
	G5 Spring	0.58	0.02	0.54	0.63
Non-Black	K Spring	1	-	-	-
	G1 Fall	1	-	-	-
	G1 Spring	0.99	0.00	0.99	1.00
	G2 Fall	0.98	0.00	0.97	0.98
	G2 Spring	0.95	0.00	0.94	0.96
	G3 Spring	0.85	0.00	0.84	0.87
	G4 Spring	0.66	0.00	0.64	0.67
	G5 Spring	0.39	0.00	0.37	0.41

).

Similarly, upon scrutinizing the disparities in science learning between Hispanic and non-Hispanic children, the data show substantial gaps emerging as early as the 2nd grade (Figure 4). The findings reveal a 93% probability for non-Hispanic children and a 98% likelihood for Hispanic children to persist below the threshold beyond the 2nd grade. This gap intensifies in subsequent grades, with an 81% likelihood for non-Hispanic children and a 95% likelihood for Hispanic children to remain below the threshold beyond the 3rd grade. By the 4th grade, the probabilities decrease to 59% for non-Hispanic children and 83% for Hispanic children, underscoring the widening disparities. By the 5th grade, these likelihoods further decline to 32% for non-Hispanic children and 56% for Hispanic children (

Table 3. Survival probability: Hispanic vs. non-Hispanic.

Race/Ethnicity	Semester	Survival Probability	SE	Lower 95% CI	Upper 95% CI
Hispanic	K Spring	1	-	-	-
	G1 Fall	1	-	-	-
	G1 Spring	1	-	-	-
	G2 Fall	0.99	0.00	0.99	0.99
	G2 Spring	0.98	0.00	0.98	0.99
	G3 Spring	0.95	0.00	0.94	0.96
	G4 Spring	0.83	0.00	0.81	0.85
	G5 Spring	0.56	0.01	0.53	0.59
Non-Hispanic	K Spring	1	-	-	-
	G1 Fall	1	-	-	-
	G1 Spring	0.99	0.00	0.99	1.00
	G2 Fall	0.97	0.00	0.96	0.98
	G2 Spring	0.93	0.00	0.92	0.94
	G3 Spring	0.81	0.00	0.80	0.83
	G4 Spring	0.59	0.01	0.57	0.61
	G5 Spring	0.32	0.00	0.30	0.34

).

3.2. Factors Mitigating the Disparities in Science Achievement

Multilevel Cox regression was employed to investigate factors influencing children’s success in surpassing the science learning threshold, with a specific emphasis on addressing disparities in science achievement. The results of the multilevel survival analysis are presented in

Table 4. Multilevel Cox regression.

	Model 1			Model 2
	Estimate	Standard Error	Hazard Ratio (HR)	Estimate	Standard Error	Hazard Ratio (HR)
Main Effects
Age	0.013 *	0.005	1.013	0.013 *	0.005	1.013
Male	0.155 ***	0.046	1.167	2.114 ***	0.564	8.282
Black	−0.288 **	0.096	0.749	−2.588 *	1.185	0.075
Hispanic	−0.162 *	0.066	0.850	−1.931 *	0.788	0.144
SES	0.164 ***	0.042	1.178	1.421 **	0.494	4.141
Externalizing Behavior	0.027	0.044	1.027	0.022	0.044	1.022
Literacy Skills	0.076 ***	0.002	1.080	0.105 ***	0.010	1.111
Moderations
Male × Literacy Skills				−0.013 ***	0.003	0.986
Black × Literacy Skills				0.016 *	0.008	1.016
Hispanic × Literacy Skills				0.012 *	0.005	1.012
SES × Literacy Skills				−0.009 **	0.003	0.990

* p < 0.05, ** p < 0.01, *** p < 0.001.

. A hazard ratio (HR) of 1 suggests no association between the predictor and the survival outcome variable. An HR greater than 1 indicates an increased hazard of event occurrence (in this context, surpassing the threshold) for higher values of the predictor, while an HR less than 1 suggests a reduced hazard of event occurrence for higher values of the predictor [50].

The likelihood of surpassing the science learning threshold significantly increased when children were older ($\beta =0.013, HR=1.013, se=0.005)$, male ($\beta =0.155, HR=1.167, se=0.046)$, non-Black ($\beta =-0.288, HR=0.749, se=0.096)$, non-Hispanic ($\beta =-0.162, HR=0.850, se=0.066)$, and from high-income families ($\beta =0.164, HR=1.178, se=0.042)$. These findings underscore disparities within gender, racial/ethnicity and socioeconomic status groups. Furthermore, children with advanced language and literacy skills exhibited a significantly heightened likelihood of surpassing the threshold of science learning ($\beta =0.076, HR=1.080, se=0.002)$, highlighting the crucial role that language and literacy proficiency play in fostering scientific learning. Children’s externalizing behaviors were not a significant predictor for the threshold.

The significance of advanced language and literacy skills in increasing the likelihood of surpassing the science learning threshold was underscored, especially when examining specific demographic groups. For female children ($\beta =-0.013, HR=0.986, se=0.003)$, Black children ($\beta =0.016, HR=1.016, se=0.008)$, Hispanic children ($\beta =0.012, HR=1.012, se=0.005)$, and those from low-income families ($\beta =-0.009, HR=0.990, se=0.003)$, higher literacy skills played a pronounced role. These findings indicate that language and literacy skills can serve as a reinforcing factor, helping to mitigate disparities between boys and girls, Black and non-Black children, Hispanic and non-Hispanic children, and children from high-income and low-income families.

4. Discussion

Using nationally representative data from the Early Childhood Longitudinal Study of Kindergarten Cohort 2011 (ECLS-K: 2011), this research investigated the emergence of disparities in science achievement among underrepresented groups and their counterparts from kindergarten through 5th grade. Additionally, this study examined the potential mitigating influence of language and literacy skills on these disparities. The primary objective was twofold: first, to explore the dynamic patterns and variations in children’s attainment of the science learning threshold over time, employing the Kaplan–Meier survival analysis method; second, to investigate the characteristics of children linked to their performance that surpassed the threshold of science learning using the multilevel Cox regression technique. This investigation was conducted using multilevel Cox regression analyses with a specific emphasis on the role of language and literacy skills in alleviating disparities in science achievement.

In the inferential analyses, the following findings emerged: (1) After accounting for censoring, the probability of children remaining below the threshold beyond 1st grade was 99%, beyond 2nd grade was 95%, beyond 3rd grade was 87%, beyond 4th grade was 68%, and beyond 5th grade was 42%. (2) Boys tended to surpass the science learning threshold earlier than girls. Non-Black and non-Hispanic children tended to surpass this threshold earlier than their Black and Hispanic counterparts. Children from high-income families exhibited a higher probability of exceeding the science learning threshold earlier than their counterparts from low-income families. (3) Children with advanced language and literacy skills demonstrated a significantly increased probability of surpassing the science learning threshold. (4) Elevated language and literacy skills played a mitigating role in reducing the disparities between boys and girls, Black and non-Black children, Hispanic and non-Hispanic children, and children from high-income and low-income families.

Our findings indicated that overall, there was an increase in the likelihood of children surpassing the science learning threshold, even in the early primary grades. While many children demonstrated their competence in science learning at a high level, concerning disparities emerged from the early years. Thus, it is a critical time to intervene and boost interest and competence starting from early childhood through the elementary years. Recent strides in science education reform are evident in the United States through the development of the Framework for K-12 Science Education [10] and the NGSS [3]. Notably, these national science standards underscore the importance of incorporating kindergarten to 3rd grade in advancing science education [4,34]. The National Science Teachers Association (NSTA) also underscores the importance of integrating science activities into early education, emphasizing that this approach not only nurtures children’s curiosity and a sense of wonder in learning but also establishes a foundational framework for a sustained and evolving journey of scientific exploration throughout their lives [55]. Nevertheless, research has predominantly focused on science achievement in middle and high school, with comparatively less attention directed towards the early stages of development [5,13].

Additionally, the national science standards strongly advocate for delivering equitable and high-quality science education experiences to all children from their early years [3,10]. This commitment is driven in part by the observed disparities in science achievement between underrepresented groups, including girls, Black and Hispanic children, and children from low-income families and their counterparts in the field of science, creating a significant political concern [11,12,13]. Remarkably, there has been scant emphasis on unraveling the disparities in science achievement, a surprising oversight given the extensive literature dedicated to addressing gaps in reading and math skills [56]. Addressing this literature gap, the present study explores the patterns and variations of children surpassing the threshold of science learning during their early years, focusing on the potential mitigating influence of language and literacy skills on these disparities. The following discussion outlines how the findings of this research align with previous studies and articulates its contribution to the existing literature.

The current study uncovered that male students were more likely to exceed the science learning threshold at an earlier stage compared to their female counterparts. This finding aligns with prior research indicating the early onset of gender gaps in science achievement [13]. NGSS Appendix D, titled “All Standards, All Students”, recognizes girls as a marginalized group in science education, attributing this to women’s enduring underrepresentation in science careers [3]. Recent research indicates that while children today more frequently associate women with science compared to earlier decades, the persistent stereotype of scientists as males endures and strengthens as children advance through school [57]. As a result, even among individuals with significant scientific aptitude, women are more prone than men to choose non-STEM careers [58]. This study contributes to the existing literature by offering a nuanced understanding of the patterns associated with gender disparities in children’s science achievement, starting in their early years. This issue warrants attention and proactive measures to ensure greater gender inclusivity in STEM fields.

The present study highlighted socioeconomic disparities in science achievement between children from socioeconomically disadvantaged households and their more advantaged counterparts [11]. Various mechanisms contribute to the impact of socioeconomic status on science learning support, which, in turn, is associated with overall science achievement. Socioeconomic status plays a pivotal role in shaping children’s science achievement, influencing both the quantity and quality of educational materials (e.g., science books, kits, online learning tools, laboratory equipment) and the range of experiences available to them (e.g., extracurricular activities, field trips to science museums, participation in science camps) [59]. Furthermore, schools with a student body from low socioeconomic backgrounds often experience shortages of fully qualified science teachers [60]. This study enhances our understanding of socioeconomic disparities in science achievement by providing nuanced insights into the complex patterns associated with such disparities in children’s science performance, beginning in their early years.

This study used the Ecological Systems Theory to conduct a comprehensive analysis of disparities in early science learning. Within this theoretical framework, gender gaps were examined across various ecological layers, taking into account cognitive factors, societal beliefs, gender-specific expectations, and cultural pressures influencing career decisions [17,18,19,61]. The application of this theory extended to socioeconomic disparities, emphasizing the existence of inherent opportunity gaps. Children from economically disadvantaged backgrounds often encountered low-quality educational experiences and were more likely to attend underfunded schools staffed by less-experienced teachers with lower expectations [20,21,59,60,62]. Racial and ethnic disparities were explored within the same ecological framework, investigating factors such as departures from STEM majors, stereotypes, and the absence of support [15,22,23,63,64]. Overall, the Ecological Systems Theory provided a robust foundation for understanding the intricate interplay of individual and contextual factors contributing to early science learning disparities across gender, socioeconomic status, and racial/ethnic dimensions.

Traditionally, research in science education has concentrated on increasing accessibility for underrepresented groups. However, this study distinguishes itself by adopting a novel approach, delving into cross-domain learning strategies to tackle and narrow science achievement gaps [4]. This study, informed by the Systemic Linguistics Approach highlighting the importance of both oral and written language in disciplinary learning, builds on existing research [32,37,38,39,40]. Functional linguists have shown that children, by making lexical choices that encompass both general and technical academic language, demonstrate proficiency in adopting language specific to academic fields, aligning themselves with distinct discourse communities [37,38,39,40]. By extending existing research, this study contributes to our understanding of the role of language and literacy in science learning, particularly during critical developmental transitions, and offers empirical evidence supporting the theoretical framework.

The fundamental activities integral to scientific inquiry encompass reading, writing, listening, and arguing [12,25,26]. An academic discipline serves as a framework for understanding, and the knowledge it encapsulates is intricately interwoven with symbols, predominantly words [36,37]. Language and literacy possess a unique capacity to represent the physical world in a clear and unequivocal manner [28,40]. Notably, language and literacy are pivotal tools employed by both teachers and children to communicate and structure the process of science learning [12,24,25]. Furthermore, they are used in a manner akin to scientists themselves, facilitating activities such as designing, recording, analyzing, comparing, explaining, evaluating, and theorizing about the observable natural world [4,26]. Despite this understanding, recent studies have yet to supply compelling evidence of effective strategies for enhancing science education through cross-domain learning [24]. In employing cutting-edge longitudinal data analysis techniques, the present study affirms previous findings that underscore the significance of addressing language and literacy skills for successful learning and performance in science.

The current study found that disparities in science learning become evident by the 2nd grade, intensify through the 3rd grade, and continue to widen through the 4th and 5th grades. Elevated literacy skills played a pivotal role in mitigating these disparities between boys and girls, Black and non-Black children, Hispanic and non-Hispanic children, and children from high-income and low-income families. This finding holds significance as the transition from socially oriented language instruction to academically focused instruction typically occurs between the 2nd and 4th grades [65]. During this period, children acquire the necessary language and literacy skills for effective participation in academic tasks [4,28]. As science education transitions into a more structured and systematic phase [42], children possessing fundamental language and literacy skills demonstrate the ability to overcome challenges in science learning [4,12,30]. Through survival analysis, this study extends previous research by offering a statistical examination of dynamic probabilities associated with science achievement. It investigates the evolving protective impact of language and literacy skills, narrowing the gap between boys and girls, Black and non-Black children, Hispanic and non-Hispanic children, and children from high-income and low-income families.

The implications of the current study apply to the practical, policy, and research realms. First, the gender, racial/ethnic, and socio-economic disparities in science achievement carry substantial implications for access to careers in STEM as well as in health professions [3,13,57,58]. The findings underscore that these disparities manifest early and are firmly entrenched by 3rd grade. Consequently, initiatives and policies bridging these disparities may benefit from targeting children during their formative years, possibly extending interventions to the early elementary and preschool stages. Importantly, studies reveal that many early childhood teachers (PreK-3) face challenges in effectively teaching science, often stemming from a deficiency in foundational knowledge in the subject [66]. Indeed, many early childhood teachers express a sense of unpreparedness to provide science education [67,68]. Professional development emerges as a pragmatic approach to address this gap, offering a viable solution for creating inclusive science classrooms in early childhood.

Second, the findings underscore the importance of integrating targeted strategies into professional development and teacher education programs to address inequities and challenge stereotypes effectively. Teachers, especially in early childhood, need awareness of how language use, curricular materials, and pedagogical choices impact students’ feelings of belonging and competence in science learning. Furthermore, promoting successful role models from underrepresented groups in science and implementing supportive structures, such as classroom-based interventions and extracurricular activities, are crucial to encourage engagement among girls, Black and Hispanic children, and children from low-income families in science learning [3,28].

Third, the findings suggest implications for curriculum development, emphasizing the need to enhance language and literacy development alongside science learning. Particularly crucial for students juggling the demands of learning science while developing English language and literacy proficiency, further research is required to understand the nuanced influence of language and literacy across diverse science branches and learning environments. Providing educators with proper training can better prepare them to anticipate and address these challenges effectively.

Fourth, this study advances our understanding of science learning achievement by examining the occurrence and duration of attainment of learning milestones. The establishment of the national science standards prompts inquiries into how children align with these expectations. Categorizing performance levels, such as “exceeding the learning threshold” and “falling short of the learning threshold” [43], becomes essential in this context. Despite calls for investigating children’s attainment of the learning threshold, most studies lack statistical approaches that quantify the timing of attainment. These temporal considerations carry implications for both theoretical frameworks and methodologies. Conceptually, incorporating the time children attain or surpass learning thresholds into developmental models allows research questions to explore when this achievement is most likely and what factors influence early attainment. Methodologically, survival analysis provides a robust set of statistical tools for addressing questions about the dynamic probabilities associated with achievement in cross-domain learning.

The current study is subject to several noteworthy limitations. First, the analyses were correlational, preventing the establishment of causal links between language and literacy skills and science learning due to the non-manipulable and non-randomly assigned nature of the existing dataset. Second, the absence of a discipline-specific literacy measure, due to the use of a secondary data analysis approach, restricts our understanding of literacy to encompass skills beyond reading and writing, including reasoning and communication in diverse content areas [69]. Third, the study lacks an intersectional analysis, neglecting the interplay of demographic factors, such as girls who are Hispanic from low-income families. Fourth, while the findings offer insights into the roles of language and literacy in science learning across diverse demographic groups, generalizability is limited by the study’s exclusive focus on outcomes in the US. This geographical constraint poses potential limitations when extrapolating findings to other countries. Fifth, the study’s reliance on data from 2010–2016 may not accurately reflect current conditions for young children’s learning, considering the dynamic changes in demographics, health crises including COVID-19, and broader societal shifts. While replicating the study with a more recent sample is necessary, the choice of using ECLS-K data is based on its size and meticulous longitudinal design, which captures crucial developmental nuances over an extended period. This decision is further influenced by the scarcity of projects engaging in comprehensive longitudinal data collection in early childhood and elementary education. Finally, the ECLS-K data underscore the onset of the achievement gap in kindergarten; however, existing research indicates that these disparities may manifest even earlier, potentially from birth [70,71]. Regrettably, our dataset’s limited scope and secondary analysis hindered an exploration of this early stage. Finally, this study aimed to assess performance levels by categorizing children as “exceeding the learning threshold” or “falling short of the learning threshold”. The transformation of a continuous science achievement variable into a categorical one involved grouping values into discrete categories, potentially resulting in information loss. Nevertheless, the conventional method of utilizing mean or standard scores in assessments inadequately addresses the fundamental query of whether the observed norms authentically reflect children attaining the desired level of performance. This research introduces novel methodologies to explore inquiries concerning the alignment of children’s performance with the expectations set by science standards.

In light of the study’s limitations, future research must address these constraints. First, future research should use experimental designs to establish causal links between language and literacy skills and science learning, thereby overcoming the correlational nature of the present study. Second, there is a critical need for developing and implementing discipline-specific literacy measures that extend beyond conventional reading and writing skills, encompassing reasoning and communication within the realm of science [28,32,69]. Third, researchers should integrate intersectional analyses, considering demographic factors such as gender, race/ethnicity, and socioeconomic status, to ensure a more comprehensive understanding of how these variables interact and impact science learning outcomes. Fourth, undertaking international comparative studies and leveraging more recent, extensive datasets, including longitudinal data, are indispensable for enhancing findings’ generalizability and temporal relevance. Finally, delving into disparities from birth through early childhood will yield insights into the precise initiation of academic gaps, thereby guiding the development of targeted interventions at an early stage.

5. Conclusions

While it is widely recognized that early science achievement significantly influences children’s knowledge, skills, and dispositions critical for future success, few studies have examined the dynamic longitudinal patterns of children surpassing the science learning threshold during their early years. There is also limited exploration of potential factors mitigating gender, racial/ethnic, and socio-economic disparities. This study utilized nationally representative data from the Early Childhood Longitudinal Study of Kindergarten Cohort 2011 (ECLS-K: 2011), including 18,174 children from 1328 schools across the United States, selected through a multistage probability sampling process and spanning kindergarten through 5th grade. This study employed the Ecological Systems Theory to analyze gender, socioeconomic, and racial/ethnic disparities in early science learning, exploring the multifaceted influences across different ecological layers. Additionally, the Systemic Linguistics Approach informed the emphasis on language and literacy skills, recognizing their pivotal role in mitigating science learning disparities, particularly during the transition from social to academically oriented teaching in early childhood and elementary school grades.

The findings highlighted the early onset and development of science achievement gaps in early childhood. Moreover, they emphasized the mitigating impact of language and literacy skills on these disparities, carrying significant implications for practice, policy, and future research. In particular, establishing national standards naturally prompts investigations into how children align with these expectations. Categorizing performance levels, such as “exceeding the learning threshold” and “falling short of the learning threshold”, becomes essential. Employing survival analysis techniques, this study shows that boys, non-Black, non-Hispanic, and high-income children tended to surpass the science learning threshold earlier than their counterparts: girls, Black, Hispanic, and low-income children. Furthermore, children with advanced language and literacy skills exhibited a significantly higher probability of surpassing the science learning threshold, playing a mitigating role in reducing disparities across gender, racial/ethnic, and socio-economic groups.