Investigating if High-Quality Kindergarten Teachers Sustain the Pre-K Boost to Children’s Emergent Literacy Skill Development in North Carolina

Abstract

This study tested the hypothesis that high-quality kindergarten teachers sustain and amplify the skill development of children who participated in North Carolina’s NC Pre-K program during the previous year, compared to matched non-participants (N = 17,330; 42% African American, 40% Non-Hispanic White, 15% Hispanic; 51% male; M_age = 4.5 years at fall of pre-K). Kindergarten teacher quality was measured using a “value-added” approach. NC Pre-K participants outperformed non-participants in the fall of kindergarten (β = 0.22) and 11% of this boost remained evident by the spring of kindergarten. Higher value-added teachers promoted the skill development of all children (β = 0.30 in the spring) but did not differentially benefit the skill development of former NC Pre-K participants compared to non-participants.

Participating in a high-quality pre-kindergarten (pre-K) program can result in a substantial boost to children’s development of early learning skills (Phillips et al., 2017). However, the resulting skill differences between pre-K participants and non-participants sometimes appear to converge during the early elementary grades—with most of the convergence often occurring in kindergarten (Weiland et al., 2021). This convergence may be especially true for emergent literacy skills such as phonemic/phonological awareness and decoding (McCormick et al., 2021), which are a primary focus of instruction during the kindergarten year.

A key question for the field is whether there are certain conditions under which the pre-K boost is less likely to converge (and more likely to sustain) during kindergarten. A growing body of research has investigated factors in the subsequent school environment that may sustain the pre-K program boost. The existing evidence to date documents mixed effects, which precludes broad conclusions (Bailey, Jenkins, et al., 2020). The central aim of this study was to investigate if high-quality kindergarten teachers can sustain the pre-K boost to emergent literacy skill development during kindergarten. We examined children who previously participated in North Carolina’s NC Pre-K program—a targeted pre-K program that is administered in a mixed delivery context and primarily serves children from low-income backgrounds (e.g., income eligibility is defined as at or below 75% of state median income) as well as higher-income peers. The program has robust evidence of providing high-quality services that result in positive outcomes for children (see Watts et al., 2023 for a review). We leveraged statewide formative assessment data collected during kindergarten to measure children’s emergent literacy skill development. We also measured teachers’ contributions to children’s emergent literacy skills as an operationalization of high-quality teaching (i.e., value-added). We then examined if high-quality teachers differentially benefited the skill development of former NC Pre-K participants compared to non-participants.

Background

The Pre-K Boost to Emergent Literacy Skills

Emergent literacy is defined as the knowledge, skills, and attitudes that are precursors to conventional forms of reading (Whitehurst & Lonigan, 1998). Some key components of emergent literacy include the “inside-out” skills that involve “children’s knowledge of the rules for translating the particular writing they are trying to read into sounds” (Whitehurst & Lonigan, 1998, p. 854). Specifically, phonemic/phonological awareness (i.e., the ability to discriminate between units of language such as phonemes, words, and prepositions) and decoding (i.e., the ability to translate units of print into units of sound). Acquiring emergent literacy skills is a major developmental task for young children in the United States, as these skills establish the foundation for later literacy learning. For example, children’s ability to understand how spoken words are represented by sequences of written letters (i.e., the alphabetic principle) is a prerequisite for later reading (Adams, 1990). Research has indeed found that emergent literacy skills at school entry are highly predictive of later reading outcomes (Duncan et al., 2007).

Participating in a high-quality pre-K program appears to result in a large boost to children’s emergent literacy skill development, according to empirical benchmarks for educational research (see Kraft, 2020). In randomized evaluation studies of North Carolina’s NC Pre-K program (Peisner-Feinberg et al., 2019), the federal Head Start program (Puma et al., 2010), and Tennessee’s Voluntary Pre-K program (Lipsey et al., 2018), the end-of-pre-K effect on children’s ability to decode letters and words was 0.22, 0.27, and 0.28 SD, respectively. All of these studies assessed children’s letter and word decoding using the Woodcock-Johnson (WJ) III Letter-Word ID measure. Similarly large magnitude effects were also reported in relation to the WJ III Spelling measure in the Head Start study (0.15) and Tennessee study (0.23) at the end of pre-K (Head Start effect sizes were reported here for children in the 4-year-old cohort of the Head Start Impact Study). Even larger magnitude effects on emergent literacy outcomes have been reported across a range of quasi-experimental studies of public pre-K programs in other states and municipalities (Barnett et al., 2018). Educational stakeholders may feel encouraged by such robust evidence of the pre-K boost in relation to children’s emergent literacy skill development, because pre-K teachers commonly dedicate a majority of their instructional time and emphasis toward content related to emergent literacy skills (Early et al., 2010).

Following evidence of the pre-K boost at the end of the pre-K year, the emergent literacy skill development of pre-K participants and non-participants appears to converge during kindergarten. For example, in the aforementioned randomized evaluation studies of Head Start, NC Pre-K, and Tennessee Pre-K, the pre-K boost to children’s decoding skills was evident at the end of pre-K, but that boost converged completely one year later at the end of kindergarten (Lipsey et al., 2018; Peisner-Feinberg et al., 2020; Puma et al., 2010). Convergence was also observed in a quasi-experimental study of Boston’s Pre-K program. In that study, Weiland et al. (2021) examined children’s phonemic/phonological awareness and decoding skills in kindergarten through second grade using the Dynamic Indicators of Basic Early Literacy Skills (DIBELS) formative assessment suite (Good et al., 2011). Based on these teacher-administered assessments, the Boston Pre-K effect on children’s emergent literacy skills ranged from 0.26 to 0.44 at the beginning of kindergarten and declined to 0.17 by the end of kindergarten. Some stability in effects as well as additional convergence was observed by the end of first grade (0.09–0.17) and second grade (0.08). Statistically significant effects continued to be evident through the end of third grade based on state-standardized assessments of English language arts (0.10). The pattern of convergence found in this quasi-experimental study was not as complete as the pattern found in the three evaluation studies using randomized designs. However, a separate lottery-based study of Boston Pre-K did not find significant effects in third grade on a state-standardized assessment of English language arts (Weiland et al., 2019).

The Pre-K Boost and Subsequent Educational Environments

Two leading hypotheses explain how the pre-K boost may be more or less likely to converge depending on educational environments children experience during elementary school. Perhaps most prominent, the dynamic complementarity hypothesis asserts that the pre-K boost will be sustained (and possibly even enhanced) by higher-quality educational environments during school (Bailey, Duncan, et al., 2020). In the education research literature, complementarity is commonly referred to as the sustaining environments hypothesis (Bailey et al., 2017). Alternatively, dynamic substitutability posits that the pre-K boost is more likely to persist among children who experience fewer opportunities for learning following pre-K (Abenavoli, 2019; Bailey, Duncan, et al., 2020). The substitutability hypothesis produces similar predictions to a theory of compensatory effects, which would also predict that the pre-K boost will buffer children against the adverse effects of lower-quality environments during school (Ramey & Ramey, 1998). In the following section on Teacher Quality, we discuss various mechanisms by which the pre-K boost might be complemented by high-quality teaching in elementary school, or substitutive of low-quality teaching.

To date, a number of studies have examined factors during elementary school in relation to convergence of the pre-K boost. This includes classroom-level observations of teacher-child interaction quality (Burchinal et al., 2022; Carr et al., 2019; McCormick et al., 2022); observations of teachers’ instruction, planning, and environment based on the Charlotte-Danielson rubric (Pearman et al., 2020; Swain et al., 2015); measures of structural classroom quality, including teacher degree (Bailey, Jenkins, et al., 2020; Bassok et al., 2019; Jenkins et al., 2018; Magnuson et al., 2007); teacher-reported measures of teacher-child relationships (Ansari et al., 2020); measures of instructional content (Burchinal et al., 2022; Jenkins et al., 2018; McCormick et al., 2022); and a measure of the proportion of former pre-K participating peers in the classroom (Burchinal et al., 2023). Previous studies have also examined school- and school district-level characteristics (Burchinal et al., 2022; Jenkins et al., 2018; Watts et al., 2023) including measures of school- or school district-level achievement growth (Carr, Peisner-Feinberg, et al., 2021; Carr, Watts, et al., 2021; Pearman et al., 2020; Swain et al., 2015). The predominant pattern of findings across these studies reveals limited evidence of complementarity or substitutability between the pre-K boost and subsequent educational environments during kindergarten. Moreover, largely null results were found in a meta-analysis of published studies investigating the interaction between early childhood educational intervention and later educational quality (Bailey, Jenkins, et al., 2020). Taken together, these findings could suggest that the pre-K boost is neither complemented by higher-quality school environments nor substitutive of lower-quality school environments. Alternatively, it is possible that the most critical ingredients of the elementary school environment have not yet been investigated in relation to convergence of the pre-K boost.

Among the aforementioned studies, there are several studies that have investigated NC Pre-K and subsequent school environments (Burchinal et al., 2022; Carr, Peisner-Feinberg, et al., 2021; Carr, Watts, et al., 2021; Watts et al., 2023). Similar to the current study, two previous studies have examined child outcomes in kindergarten while using matching methods to compare the outcomes of former NC Pre-K participants and matched non-participants (Burchinal et al., 2022; Carr, Peisner-Feinberg, et al., 2021). With regard to children’s emergent literacy skills, both studies found no evidence of complementarity or substitutability between NC Pre-K participation and subsequent educational environments as measured by kindergarten teacher-child interaction quality, instructional content, school-wide academic proficiency, or school-wide academic achievement growth. Evidence of complementarity was found for child language skills and working memory skills in one study (Carr, Peisner-Feinberg, et al., 2021). However, neither of these studies investigated a measure of teacher quality that is directly related to children’s emergent literacy skills development—which is the primary aim of the current study.

Two separate studies have also tested the dynamic complementarity and substitutability hypotheses in relation to longer-term outcomes, while using a research design that examined variability in NC Pre-K program funding (Carr, Watts, et al., 2021; Watts et al., 2023). These studies examined the rollout of NC Pre-K program funding across counties and years to assess the effects of increases in NC Pre-K funding. These studies also examined the effect of funding on the academic achievement outcomes of all children in the public-school population, including former NC Pre-K participants as well as non-participants—referred to as “spillover” effects. Finally, these studies considered longer-term effects on student achievement outcomes in fifth grade (Watts et al., 2023) and eighth grade (Carr, Watts, et al., 2021). Both of these studies documented evidence of dynamic substitutability for the effect of NC Pre-K funding on children’s reading and math achievement outcomes in lower-quality subsequent school environments (e.g., larger effects on eighth grade reading and math outcomes in school districts with lower levels of academic achievement growth; Carr, Watts, et al., 2021). Taken together, these NC Pre-K funding studies provide some evidence of substitutability based on broad measures of subsequent school environments using a research design that focused on NC Pre-K program funding and longer-term outcomes. Alternatively, research based on matched comparison designs have not found evidence of complementarity or substitutability in relation to emergent literacy skills in kindergarten.

Teacher Quality

There are many ways to measure the quality of the subsequent educational environments children experience after they graduate from pre-K and transition to elementary school. Both theory and research suggest that the quality of children’s interactions with their teachers is a primary driver of learning and development within the early childhood classroom environment (Burchinal, 2018; Hamre & Pianta, 2010). In the current study, we examined teacher “value-added” to emergent literacy skills as a measure of teacher quality during kindergarten. Value-added is commonly measured as the average learning gain among all students in a teacher’s classroom and is used to index a teacher’s overall contribution to children’s learning (Hanushek & Rivkin, 2010). By directly measuring the extent to which teachers promote learning gains in an outcome of interest (e.g., emergent literacy), the value-added approach can circumvent the need to examine a variety of individual characteristics that may be related to a teacher’s ability to promote learning gains (e.g., education, experience, and pedagogy). In other words, the value-added approach directly measures a key construct of interest when evaluating the dynamic complementarity and substitutability hypotheses: teachers’ ability to promote learning gains in the same measure used to assess the pre-K boost. Prior research has found measures of teacher value-added to be robust and unbiased predictors of teacher impacts on student learning in the later elementary and middle grades (e.g., Chetty et al., 2014), including first grade (Palardy & Rumberger, 2008). To our knowledge, the current study is the first to investigate teacher value-added in relation to children’s learning gains during kindergarten, and to investigate teacher value-added in relation to complementarity or substitutability of the pre-K program boost.

Scholars have speculated about the ways in which the pre-K boost might be dynamically complemented by high-quality teaching in elementary school or substitutive of low-quality teaching. With regard to dynamic complementarity, a consensus statement by leading researchers described how “pre-K can be viewed as powering up early learning, for which the elementary grades need to provide essential charging stations that sustain and amplify the learning gains made by children in pre-K. Absent re-charging, progress will likely be stalled, and the benefits from any boost provided by pre-K education may be lost” (Phillips et al., 2017, p. 25). Separately, scholars have also describe how a pattern of “catch up” in the early grades can lead to pre-K effect convergence, as the children who did not participate in pre-K learn at a more rapid pace than the pre-K participants (e.g., Abenavoli, 2019). Catch-up could also signify that kindergarten teachers have focused their instruction on those students who were initially low-performing at the beginning of the year so that their performance by the end of the year converges with that of initially higher-performing peers. If former pre-K participants enter kindergarten at comparatively higher skill levels, they may receive less instructional attention than non-pre-K attenders. This idea has inspired some observers to argue that kindergarten teachers should differentiate and individualize their instruction to further support the advanced learning of former pre-K participants following pre-K, which would presumably lead to sustaining the pre-K boost (Phillips et al., 2017). Because the value-added measure directly assesses the extent to which teachers promote larger average gains in learning, the measure should be well-aligned to capture the classroom environments that provide “essential charging stations” for continued learning for children who participated in pre-K. Thus, if predictions of dynamic complementarity are accurate, then high value-added teachers should facilitate a process of “skills begets skills” (Cunha et al., 2006, p. 703) as children progress beyond the foundational skills they learned during pre-K to learn more advanced skills at a more rapid pace during kindergarten. In the context of complementarity, convergence is more likely to occur in elementary classrooms with lower value-added teachers as the skill development of former pre-K participants slows—allowing non-participants to acquire those foundational skills and catch-up to their pre-K peers.

Alternatively, the dynamic substitutability hypothesis would suggest that the pre-K boost is more likely to persist among children who go on to experience fewer learning opportunities in kindergarten. If children enter kindergarten environments in which opportunities to learn emergent literacy skills are limited, then the pre-K boost may be more likely to persist because the children who did not participate in pre-K would be less likely to rapidly learn the foundational emergent literacy skills that were promoted in pre-K. As mentioned above, this hypothesis is highly related to predictions of buffering or compensatory effects, in which pre-K “may build skills in children that make them resilient to adverse conditions, such as poor-quality experiences at school” (e.g., Abenavoli, 2019, p. 1113). In the context of substitutability, the foundational literacy skills children gained in pre-K might enable them to maintain a skill advantage in a kindergarten context where the teacher is less effective in promoting further skill development (i.e., lower value-added). Substitutability would also predict that convergence is more likely to occur in elementary classrooms where higher value-added teachers enable non-pre-K participants to progress at an advanced rate of skill development and catch-up to their more advanced pre-K peers. This dynamic substitutability hypothesis is consistent with other developmental theories and research indicating that children who experience adverse environmental risk factors such as poverty are most likely to benefit from early childhood intervention (e.g., McCartney et al., 2007). Using a statistical test, support for the dynamic substitutability hypothesis would be evidenced by a positive main effect of pre-K participation and kindergarten teacher quality as well as a negative interaction between these two inputs; while support for the dynamic complementarity hypothesis would be evidenced by a positive interaction (Bailey, Jenkins, et al., 2020). In the current study, we extend previous research by examining the extent to which NC Pre-K participation and kindergarten teacher value-added can dynamically benefit children’s emergent literacy skill development during kindergarten—a period of time when pre-K convergence is most likely to occur.

Methods

The aim of the current study was to draw comparisons between the emergent literacy skill development of children who did or did not participate in North Carolina’s NC Pre-K program. Importantly, we examined if a measure of kindergarten teacher quality (i.e., value-added) moderated differences between the skill development of former NC Pre-K program participants and non-participants. Consistent with prior research by Weiland et al. (2021), we examine children’s emergent literacy skills based on the DIBELS measure, which was assessed by kindergarten teachers at the fall, winter, and spring of the kindergarten year.

Our study took place in three phases (see Table S1). In the first phase, we measured teacher value-added to emergent literacy skill outcomes during kindergarten. In the second phase, we used multi-variable matching methods to construct a matched sample of children who participated in the NC Pre-K program and children who did not participate, with exact matching for a measure of economic disadvantage given that NC Pre-K is a targeted pre-K program that primarily serves children from low-income backgrounds. In the third phase of our study, we examined NC Pre-K and comparison group differences in emergent literacy skill development during kindergarten and the moderation of group differences by teacher value-added.

Consistent with experimental evidence from a randomized control trial of NC Pre-K (Peisner-Feinberg et al., 2020; Peisner-Feinberg et al., 2019), we expected NC Pre-K participants to outperform non-participants in emergent literacy skills in the fall of kindergarten. We also expected this skill difference to begin to converge, on average, during kindergarten. If evidence of moderation were observed, then support for dynamic complementarity would be indicated by former NC Pre-K participants demonstrating higher (or comparable) rates of skill development than non-participants when enrolled in kindergarten classrooms with higher value-added teachers, and lower rates of skill development than non-participants when enrolled in classrooms with lower value-added teachers. Conversely, support for the dynamic substitutability hypothesis would be indicated by former NC Pre-K participants demonstrating higher (or comparable) rates of skill development than non-participants when enrolled in kindergarten classrooms with lower value-added teachers, and lower rates of skill development than non-participants when enrolled in classrooms with higher value-added teachers. Our study is largely exploratory (rather than confirmatory) because we rely on novel data that has not previously been used to estimate the NC Pre-K boost and we test competing hypotheses regarding dynamic complementarity and substitutability using a novel measure of kindergarten teacher quality.

Sample

Our sample was drawn from children enrolled in North Carolina public schools. Across Phases I, II, and III of our study, we used school records data on the population of children enrolled in North Carolina public schools (obtained from the NC Education Research Data Center [NCERDC] at Duke University). In Phases II and III of the study, we also use birth records data on the subpopulation of public-school children who were born in North Carolina (obtained from NC State Center for Health Statistics [NCSCHS]). The identified birth records and school records of individual children were matched by the NCERDC. Among all kindergarten students who were age-eligible to participate in NC Pre-K during the 2013–14 program year and then recorded complete literacy scores in kindergarten during the 2014–15 school year, a total of 79% of all school records were matched to a birth record and all matches were based on exact first and/or last name (or close/partial/nickname match) and exact date of birth (or 2/3 of birth date components in < 1% of matches).

The first phase of our study included all children with complete emergent literacy assessment scores and complete school records who were enrolled in kindergarten during the 2013–14, 2014–15, or 2015–16 school years. The second phase of the study included children who were (1) age-eligible to participate in NC Pre-K during the 2013–14 program year; (2) enrolled in kindergarten during the 2014–15 school year; (3) had matched school records and birth records; and (4) had no missing information related to emergent literacy scores, school records, and birth records. The third phase of the study included only those children who were matched from the Phase II analyses. Children who switched classrooms/teachers between the fall and spring of kindergarten constituted 5% of our Phase III analysis sample. In our Phase I analyses, the emergent literacy scores of children who switched classrooms between the fall and spring of kindergarten were duplicated and assigned to each respective classroom to allow these scores to be reflected in the value-added scores of each classroom/teacher children were enrolled with during the school year. In our Phase II and III analyses, children were exclusively assigned to their spring of kindergarten teacher. The overall pattern of findings for the Phase II and III analyses were robust to the assignment of children to their fall kindergarten classroom.

Regarding the sociocultural composition of our sample, we focused on children who were enrolled in North Carolina public schools during kindergarten across all three phases of our study, because these children and their families might have also sought to enroll in the publicly funded NC Pre-K program during the previous year—if given the opportunity to do so. Additionally, we largely focused on children enrolled in kindergarten during the 2014–15 school year given the missing data limitations presented by more recently collected data (e.g., data on maternal education and emergent literacy skills assessments were missing for more recent cohorts of students). In Phase II of our analyses, we constructed a “matched” sample of NC Pre-K participants and non-participants using multi-variable matching methods that sought to identify non-participants with similar sociodemographic characteristics as the NC Pre-K participants, including race/ethnicity, gender, and economic disadvantage. These efforts were limited by our inability to collect additional data beyond the administrative data sources that we have relied on (i.e., school records, birth records, NC Pre-K records), as well as our inability to consider children who were missing data on key analytic variables drawn from these data. Beyond the requirements listed above, we place no additional restrictions on the social, cultural, or demographic composition of our sample. The sociodemographic characteristics of our analysis sample is further detailed below in the section titled Phase II Results: Matching.

Data

NC Pre-K Participation

We obtained administrative records on children who enrolled in NC Pre-K from the North Carolina Division of Child Development & Early Education (DCDEE). We identified children who enrolled in NC Pre-K during the 2013–14 program year. To be conservative in our measure of NC Pre-K participation, we defined NC Pre-K participants as only those children who attended one or more days of NC Pre-K. Among the children in our Phase III analysis sample, more than 90% of the NC Pre-K participants attended NC Pre-K for more than 157 days during the program year, with a median of 277 days attended (M = 254, SD = 64).

Emergent Literacy Skills

Data on children’s emergent literacy skills were obtained from the NCERDC. Kindergarten teachers administered direct assessments of children’s emergent literacy skills using various subtests from the DIBELS Next^® formative assessment suite (Good et al., 2011). Assessments were administered in the fall, winter, and spring of the school year. While one subtest was administered across all three timepoints, other subtests were added or removed at different timepoints to measure the increasing complexity of children’s skill development over time. In the fall of kindergarten, the subtests included First Sound Fluency (FSF) and Letter Naming Fluency (LNF). In the winter, the subtests included FSF, LNF, Phoneme Segmentation Fluency (PSF), and Nonsense Word Fluency (NWF). In the spring of kindergarten, the subtests included LNF, PSF, and NWF (removing FSF). A composite score was calculated at each timepoint based on scores from the administered subtests. The Supplemental Materials contain additional information about the DIBELS Next^® subtests, benchmark ranges, and transformation of the composite score into the Lexile^® Framework for Reading scale score (MetaMetrics^®, 2014).

Child and Family Background Characteristics

We considered a range of child and family background characteristics as covariates in our analyses. The following covariates were drawn from the NCERDC school records and used across all phases of analyses: Sex (Male=1), economic disadvantage (measured as qualification for free- and/or reduced-price lunch; yes=1), English language learner (yes=1), race (measured using all available categories from the NCERDC school records, including Non-Hispanic White, African American, Hispanic, Asian American/Pacific Islander, Native American, or Mixed/Other Race), age (in years; calculated at kindergarten entry in the Phase I analyses and at pre-K entry in the Phase II and III analyses), and kindergarten teacher/classroom membership. In our Phase I analyses, we also controlled for designated disability status (yes=1). We also considered absences during kindergarten as a covariate in the Phase I analyses and in the Phase III sensitivity analyses. The following covariates were drawn from the NCSCHS birth records and used in the Phase II and II analyses: Birth weight (pounds), mother education at child’s birth (years), teenage mother (yes=1), mother age at child’s birth (years), mother married (yes=1), no father information on birth record (yes=1), mother immigrant (yes=1), and child was mother’s first born (yes=1). Finally, we also report on a measure of school-wide economic disadvantage, which is indexed by the percent of economically disadvantaged students in the school. This measure was drawn from the NCERDC school records and used only for descriptive reporting.

Analyses

Phase I Analyses: Measuring Teacher Value-Added

Following prior research by Chetty et al. (2014), we used multi-level regression models to estimate kindergarten teacher value-added to children’s emergent literacy skill outcomes in the spring of kindergarten (see Supplemental Materials for more detail). Based on this model, we derived our measure of teacher value-added—representing the mean skill level among children in the teacher’s classroom in the spring of kindergarten adjusted for children’s fall-of-kindergarten skill level (i.e., residualized change) and other covariates (see Chetty et al., 2014).

Phase II Analyses: Matching

To construct a matched sample of NC Pre-K participants and non-participants, we used the PSMATCH procedure in SAS^® to conduct multi-variable matching based on the child and family background covariates. Importantly, we used exact matching for kindergarten classroom/teacher membership in our Phase II analyses to ensure that each NC Pre-K participant was matched to a non-participating peer who subsequently enrolled within the same kindergarten classroom (see Supplemental Materials for more detail).

Phase III Analyses: Moderation of the NC Pre-K Boost

A series of multi-level growth curve models were used to address the main research aims. Separate models considered linear and quadratic growth using either teacher random effects or teacher fixed effects. In these models, we examined the NC Pre-K boost in relation to children’s emergent skills in the fall of kindergarten and in relation to growth during kindergarten. We also examined teacher value-added in relation to growth during kindergarten. Importantly, we examined if teacher value-added moderates the NC Pre-K boost in relation to growth during kindergarten, which could provide support for the dynamic complementarity or dynamic substitutability hypotheses (see Supplemental Materials for more detail).

Results

Phase I Results: Measuring Teacher Value-Added

Results of the multi-level regression analyses to measure teacher value-added are presented in table S2, with three separate models for each of the three school years. All school record covariates were found to be reliable predictors of children’s spring-of-kindergarten emergent literacy scores (p < .05; the Hispanic and/or Multi-Race categories were not reliably different from the reference group [Non-Hispanic White] during some school years, but, nonetheless, race differences were found across all school years). Based on the results of these models we calculated the jackknife measure of teacher value-added. The jackknife measure was highly correlated with the 2014–15 measure (r = 0.63), which suggests that the jackknife scores provided an adequate proxy for observed teacher value-added scores during the 2014–15 school year. This correlation is based on a sample of 5,021 teachers with value-added scores during the 2013–14 and/or 2015–16 school years, excluding 595 teachers who had recorded a value-added score during the 2014–15 school year only. Descriptive statistics for the jackknifed measure show a Lexile^® score value of –168.28 (SD = 108.07) as the adjusted (i.e., residualized change) classroom-level average in the spring of kindergarten. This measure was found to be normally distributed (see Figure S1), with a skewness value of 0.22 indicative of a symmetrical distribution (i.e., between −0.50 and 0.50) and a kurtosis value of 0.57 indicative of a broad platykurtic distribution (i.e., < 3). Additionally, we find that 19% of teachers’ value-added scores would be indicative of students reaching “beginning reader” status in the spring of kindergarten (i.e., a Lexile^® score value of −75L is indicative of a “beginning reader” according to the Lexile Framework for Reading).

Phase II Results: Matching

Based on the input sample of children in the 2014–15 cohort (n = 76,477), multi-variable matching resulted in a matched sample of 17,330 children, with equal numbers of children in the NC Pre-K (n = 8,665) and comparison groups (i.e., non-NC Pre-K participants; n = 8,665). After matching, we found adequate balance in the covariate distributions between the matched NC Pre-K and comparison groups, with standardized mean differences ≤ |0.09| for all school and birth record covariates (see column “c-d” in Table 1; a criterion of ≤ |0.10| for adequate balance was defined by Austin, [2014]). In many cases, this represented a substantial reduction in the group differences observed prior to matching (e.g., 0.57 SD for economic disadvantage prior to matching and 0.00 SD after matching; see columns “a-b” and “c-d” in Table 1).

Table 1.

Descriptive Statistics Before and After Matching

	Before Matching				After Matching				Standardized Mean Difference
	(a)		(b)		(c)		(d)		a-b	c-d	c-a
	NC Pre-K		Comparison		NC Pre-K		Comparison
	n = 19,494		n = 56,983		n = 8,665		n = 8,665
	M	(SD)	M	(SD)	M	(SD)	M	(SD)
DIBELS Next Composite Score
Fall	30	(22)	32	(25)	30	(22)	25	(22)	−0.08	0.24	0.01
Winter	137	(50)	145	(53)	137	(49)	134	(53)	−0.16	0.07	0.01
Spring	143	(44)	151	(47)	143	(43)	142	(46)	−0.16	0.04	0.00
Lexile Scale Score
Fall	−1083	(266)	−1059	(299)	−1079	(265)	−1145	(269)	−0.08	0.24	0.01
Winter	−516	(295)	−465	(315)	−512	(291)	−533	(313)	−0.16	0.07	0.01
Spring	−210	(302)	−157	(324)	−210	(298)	−221	(314)	−0.16	0.04	0.00
School Record Covariates
Male (Yes=1)	0.51	(0.50)	0.50	(0.50)	0.51	(0.50)	0.51	(0.50)	0.02	0.00	0.00
Economic Disadvantage (Yes=1)	0.78	(0.41)	0.50	(0.50)	0.79	(0.40)	0.79	(0.40)	0.57	0.00	0.03
English Language Learner (Y es=1)	0.14	(0.35)	0.10	(0.30)	0.10	(0.29)	0.10	(0.29)	0.13	0.00	−0.13
Race: Non-Hispanic White (Yes=1)	0.31	(0.46)	0.52	(0.50)	0.40	(0.49)	0.40	(0.49)	−0.42	0.00	0.19
Race: African American (Yes=1)	0.38	(0.48)	0.24	(0.43)	0.42	(0.49)	0.42	(0.49)	0.30	0.00	0.10
Race: Hispanic (Yes=1)	0.23	(0.42)	0.17	(0.37)	0.15	(0.35)	0.15	(0.35)	0.17	0.00	−0.22
Race: Asian American/Pacific Islander (Y es=1)	0.01	(0.12)	0.03	(0.16)	0.00	(0.05)	0.00	(0.05)	−0.08	0.00	−0.12
Race: Native American (Yes=1)	0.03	(0.16)	0.01	(0.11)	0.02	(0.15)	0.02	(0.15)	0.12	0.00	−0.03
Race: Multi-Race (Yes=1)	0.04	(0.19)	0.03	(0.17)	0.01	(0.09)	0.01	(0.09)	0.04	0.00	−0.18
Age at Pre-K Entry (Years)	4.51	(0.29)	4.50	(0.29)	4.52	(0.29)	4.50	(0.29)	0.03	0.09	0.04
Days Absent	7.02	(9.09)	6.29	(8.61)	7.04	(8.92)	7.65	(10.70)	0.08	−0.06	0.00
Birth Record Covariates
Child Birth Weight (Pounds)	7.06	(1.30)	7.18	(1.29)	7.01	(1.34)	7.05	(1.28)	−0.10	−0.03	−0.04
Mother Education at Child's Birth (Years)	11.66	(2.57)	12.92	(2.95)	11.83	(2.23)	11.79	(2.36)	−0.43	0.02	0.07
Teenage Mother (Yes=1)	0.17	(0.38)	0.12	(0.32)	0.18	(0.39)	0.19	(0.39)	0.17	−0.02	0.03
Mother Age at Child's Birth (Years)	25.08	(5.74)	27.18	(6.15)	24.55	(5.48)	24.79	(5.70)	−0.34	−0.04	−0.09
Mother Married (Yes=1)	0.37	(0.48)	0.57	(0.50)	0.35	(0.48)	0.37	(0.48)	−0.38	−0.04	−0.05
No Father Information on Birth Record (Yes=1)	0.24	(0.43)	0.16	(0.37)	0.24	(0.43)	0.27	(0.44)	0.19	−0.06	0.00
Mother Immigrant (Yes=1)	0.24	(0.43)	0.19	(0.39)	0.16	(0.36)	0.14	(0.34)	0.13	0.05	−0.20
Child was Mother's First Born (Yes=1)	0.37	(0.48)	0.41	(0.49)	0.38	(0.49)	0.39	(0.49)	−0.08	0.00	0.02
Kindergarten Environment
Teacher Value-Added (jackknife)	−161	(106)	−163	(106)	−162	(105)	−162	(105)	0.02	0.00	−0.01
School-Wide Economic Disadvantage (%)	0.78	(0.23)	0.63	(0.28)	0.77	(0.22)	0.77	(0.22)	0.54	0.00	−0.03

Note. Standardized mean differences between columns (e.g., a-b represents the standardized mean difference between NC Pre-K and comparison groups before matching). Days absent was used as a covariate in the Phase I analyses and Phase III sensitivity analyses.

The matched sample was racially and ethnically diverse with 42% of children reporting as African American, 40% as Non-Hispanic White, 15% as Hispanic, 2% as Native American, 1% as multi-race, and < 1% Asian American/Pacific Islander. The matched sample was nearly evenly distributed by sex, with 51% male. A portion of the sample was administratively identified as English language learners in kindergarten (10%). Finally, a majority of children in the sample were administratively identified as economically disadvantaged in kindergarten (79%).

We also examined the standardized mean differences in the DIBELS and Lexile scores between the NC Pre-K and comparison groups before and after matching. Before matching, we found negative standardized mean differences indicative of lower emergent literacy skills for the NC Pre-K group at the fall, winter, and spring of kindergarten (–0.08, –0.16, and –0.16 SD, respectively). However, after matching, we found positive standardized mean differences indicative of higher emergent literacy skills for the NC Pre-K group at the fall, winter, and spring of kindergarten (0.24, 0.07, and 0.04 SD, respectively).

The matched sample of NC Pre-K participants (n = 8,665) was found to differ from the input sample of NC Pre-K participants (n = 19,494) along several sociodemographic characteristics (see column “c-a” in Table 1). In particular, we found standardized mean differences > |0.10| indicating that the matched NC Pre-K sample—compared to the input sample—had a lower proportion of English language learners (–0.13 SD), a higher proportion of Non-Hispanic Whites (0.19 SD), a higher proportion of African Americans (0.10 SD), a lower proportion of Hispanics (–0.22 SD), a lower proportion of Asian American/Pacific Islanders (–0.12 SD), a lower proportion of multi-racial children (–0.18 SD), and a lower proportion of children with immigrant mothers (–0.20 SD) (see Table 1 for percentage point differences). The matched and input samples of NC Pre-K participants were very similar with regard to emergent literacy skills in the fall, winter, and spring of kindergarten, with standardized mean differences all ≤ |0.01|.

Phase III Results: Moderation of the NC Pre-K Boost

Multi-level growth curve models compared skill differences between the matched NC Pre-K group (i.e., program participants) and the comparison group (i.e., non-participants) as well as moderation of group differences by kindergarten teacher value-added (see Table 2). Comparable findings were evident across all model specifications. Therefore, in the text below, we summarize findings and report coefficients derived from the teacher random effects model with linear and quadratic growth. We also compare our findings to the statewide average of children’s emergent literacy skills, which was calculated using Lexile scores from all available students (N = 107,151) assessed in the fall, winter, and/or spring of kindergarten during the 2014–15 school year.

Table 2.

Multi-Level Growth Curve Model Results: Longitudinal Analyses of Growth in Emergent Literacy Skills during Kindergarten (N = 17,330)

	Lexile Scale Score
	Model 1		Model 2		Model 3		Model 4
Parameters	b	(SE)	b	(SE)	b	(SE)	b	(SE)
Intercept	−1078.82***	(3.88)			1130.69***	(3.82)
Linear	453.60***	(2.11)			768.22***	(5.32)
Quadratic					−157.50***	(2.46)
NC Pre-K	54.26***	(3.83)	54.34***	(3.84)	60.06***	(3.69)	59.10***	(3.79)
NC Pre-K × linear	−26.87***	(1.97)	−26.94***	(1.97)	−61.65***	(5.63)	−55.54***	(6.12)
NC Pre-K × quadratic					17.39***	(2.66)	14.30***	(2.91)
Value-added	7.12**	(2.20)			0.01	(2.20)
Value-added × linear	45.40***	(1.23)			87.72***	(2.91)
Value-added × quadratic					−21.14***	(1.33)
NC Pre-K × value-added	6.12	(3.81)	6.12	(3.81)	6.92	(3.67)	6.89	(3.77)
NC Pre-K × value-added × linear	−2.01	(1.96)	−2.00	(1.96)	−6.87	(5.60)	−6.65	(6.09)
NC Pre-K × value-added × quadratic					2.43	(2.65)	2.32	(2.89)
Linear Growth	✓		✓		✓		✓
Quadratic Growth					✓		✓
Teacher Random Effects	✓				✓
Teacher Fixed Effects			✓				✓

Note.

^*p < .05,

^**p < .01,

^***p < .001.

All models include school & birth record covariates.

In the fall of kindergarten, we found that former NC Pre-K participants outperformed the matched comparison group in emergent literacy skill attainment on average (β = 0.22, b = 60.06, p = .001) (see Figure 1, Panel B). By the winter of kindergarten, 26% of the initial skill difference between the two groups remained (post-hoc estimate: β = 0.05, b = 15.80, p = .001) and 11% remained by the spring of kindergarten (post-hoc estimate: β = 0.02, b = 6.31, p = .13). The NC Pre-K and comparison groups both demonstrated substantial rates of growth in emergent literacy skills during kindergarten—following a non-linear pattern such that the highest rates of growth occurred between the fall and winter of kindergarten (linear: β = 0.13, b = 768.22, p = .001; quadratic: β = –0.06, b = –157.50, p = .001). The comparison group demonstrated higher rates of growth in emergent literacy skills than the NC Pre-K group during kindergarten (linear: β = –0.24, b = –61.65, p = .001; quadratic: β = 0.07, b = 17.39, p = .001). In particular, the comparison group gained an average of 633 Lexile score points between the fall and winter of kindergarten while the NC Pre-K group gained an average of 589 points (post-hoc estimate of difference: β = –0.17, b = –44.26, p = .001). Between the winter and spring of kindergarten, the comparison group gained an average of 300 points while the NC Pre-K group gained an average of 291 points (post-hoc estimate of difference: β = –0.03, b = –9.49, p = .003). Furthermore, former NC Pre-K participants demonstrated an average rate of growth in emergent literacy skills that was higher than the statewide average between the fall and winter of kindergarten (i.e., 598 compared to 571 points) and similar to the statewide average between the winter and spring of kindergarten (i.e., 291 compared to 303 points). Therefore, the skills attained by NC Pre-K children were not “fading out” during kindergarten while the comparison group caught up to former NC Pre-K participants.

Figure 1.

Growth in Emergent Literacy Skills during Kindergarten

Kindergarten teacher value-added was positively associated with growth in children’s emergent literacy skills during kindergarten (linear: β = 0.29, b = 87.72, p = .001; quadratic: β = –0.07, b = –21.14, p = .001) as the association between teacher value-added and skill level emerged in the winter (post-hoc estimate: β = 0.22, b = 65.99, p = .001) and further increased in the spring (post-hoc estimate: β = 0.30, b = 90.89, p = .001), but, as expected, was not evident in the fall (β = 0.00, b = 0.01, p = .99). Importantly, kindergarten teacher value-added was not found to moderate the NC Pre-K boost to growth in children’s emergent literacy skills during kindergarten (linear: β = –0.03, b = –6.87, p = .22; quadratic: β = 0.01, b = 2.43, p = .36), indicating that the higher rate of growth observed among the comparison group was evident when kindergarten teacher value added was both higher and lower (see Figure 1, Panels A & C).

Sensitivity Analyses

A series of sensitivity analyses were undertaken to assess the robustness of our results to alternative analytic specifications (see the Supplemental Materials section for more detail). Surprisingly, we still found virtually no evidence of moderation of the NC Pre-K boost by teacher value-added across these alternative specifications (see Table S3). This includes null moderation results for a model with a non-linear term for teacher value-added, a separate model controlling for child absences, and a separate model excluding children in classrooms where a jackknifed measure of teacher value-added was missing. A statistically significant moderation effect emerged in a model that relaxed the requirement for within-kindergarten classroom matching and allowed NC Pre-K participants to be matched to any other non-participating child across North Carolina, regardless of kindergarten context—resulting in an increased sample size (n = 38,910) compared to our primary analyses (n = 17,330). This significant moderation effect was suggestive of dynamic complementarity and was similar in magnitude to the null moderation effect found in our primary analyses.

Discussion

There is a wealth of evidence demonstrating that a high-quality pre-K can boost children’s development of early learning skills (Phillips et al., 2017). Today, education stakeholders are eager to identify factors that sustain the pre-K boost into elementary school. This is the key issue we sought to address in the current study.

Results Summary

We found that former NC Pre-K participants outperformed matched non-participants in emergent literacy skills at the beginning of kindergarten. However, the difference between NC Pre-K participants and matched non-participants converged over the course of the school year as matched non-participants demonstrated higher rates of skill development to catch up to their NC Pre-K peers. By the end of kindergarten, the difference between these two groups was approximately 11% the magnitude of the initial group difference that was found at the beginning of kindergarten. Nonetheless, both former NC Pre-K participants and matched non-participants demonstrated substantial rates of growth in emergent literacy skills during kindergarten. Thus, the emergent literacy skill development of former NC Pre-K participants did not “fade out” during kindergarten.

We also found that kindergarten teacher value-added (a measure of teacher quality) was positively associated with growth in emergent literacy skills for all children during kindergarten—including former NC Pre-K participants and matched non-participants. However, teacher value-added did not differentiate the skill development of NC Pre-K participants compared to non-participants during kindergarten, but, instead, the benefit of teacher value-added was similar in magnitude for both groups. Thus, students who did not participate in NC Pre-K partially caught up to the emergent literacy skill level of students who previously participated in NC Pre-K, regardless of whether they attended a kindergarten classroom with a higher or lower value-added teacher. This pattern of findings was consistent with the overarching goals of formative assessment systems like the DIBELS, which enable teachers to identify low-performing students and target them for accelerated growth.

We can compare the magnitude of our findings to established benchmarks in educational research. For example, among children in the 2011 Early Childhood Longitudinal Study–Kindergarten Cohort (ECLS-K), the average difference between students’ fall and spring literacy scores during kindergarten was 1.21 in effect size units (Hedberg, 2016). Using this benchmark, we found that the NC Pre-K boost to children’s emergent literacy skill development in the fall of kindergarten was 18% of the ECLS-K benchmark (i.e., 0.22/1.21). We also found that the association between teacher value-added and children’s emergent literacy skills was 25% of the ECLS-K benchmark in the spring of kindergarten (i.e., 0.30/1.21). Additionally, all of the aforementioned effect size units can be described as “large” according to empirical benchmarks established for educational research (i.e., > 0.20) (see Kraft, 2020). These findings indicate that NC Pre-K participation and teacher value-added can boost children’s skill development in educationally meaningful ways.

We have also illustrated our findings relative to the benchmarks for emergent literacy skill development defined by the DIBELS assessment system (see the Supplemental Materials section for more detail). At the beginning of the kindergarten year, we found that the average NC Pre-K participant scored within the “at benchmark” range, while the average non-participant scored within the “below benchmark” range—illustrating the NC Pre-K boost to emergent literacy skills at the beginning of kindergarten (see Figure 1). Non-participants grew toward the “at benchmark” level of former NC Pre-K participants by the winter of kindergarten and continued to close the gap through the spring of kindergarten. Moreover, our findings demonstrate a pattern of “catch-up” in which non-participants grew at a faster rate during kindergarten, with no sign that the former NC Pre-K participants lost ground in their emergent literacy skill development. Instead, former NC Pre-K participants begin kindergarten “at benchmark” and continued to make progress in this range throughout the year, while non-participants progressed from scoring “below benchmark” at the beginning of kindergarten to scoring “at benchmark” by the end of kindergarten.

Children in our sample still had more room left to grow in emergent literacy skills by the end of kindergarten, which was evidenced by the average scores of former NC Pre-K participants and matched non-participants remaining below the statewide average across the school year (see Figure 1, Panel B). Therefore, it seems entirely possible that pre-K boost to emergent literacy skills could have sustained for the children in our sample if former NC Pre-K participants had demonstrated even higher rates of growth during kindergarten—narrowing the gap for these children whose families largely qualified as economically disadvantaged. Scholars have previously referred to emergent literacy as a “constrained” skill and speculated that the pre-K boost may be more likely to convergence for “constrained” skills compared to “unconstrained” skills such as language (Bailey et al., 2017; Johnson et al., 2023; McCormick et al., 2021). A key tenet of the constrained skill theory suggests that “constrained skills have a ceiling; the learner can achieve perfect performance” (Snow & Matthews, 2016, p. 58). Contrary to this hypothesis, the measure of emergent literacy skills used in our study assessed a set of competencies that increased in complexity over time and allowed children in our sample to demonstrate substantial rates of growth during kindergarten—with even more room left to grow by the end of kindergarten. Therefore, it seems likely that other factors may be leading to the pattern of convergence found in our study (i.e., catch up).

Non-pre-K participants may be more likely to catch up to their former pre-K peers in the context of public-school settings where there is widespread use of the DIBELS formative assessment system, and the benchmark ranges defined by the DIBELS. Knowing where children score in these benchmark ranges may incentivize teachers and school administrators to focus on providing extra supports to those children who score “below benchmark” at the beginning of the school year—supporting them to reach “at benchmark” scores by the end of the year. Indeed, a survey or kindergarten through sixth grade teachers found that teachers’ primary use of the DIBELS assessment suite relates to raising the skill levels of low-performing students, including identifying students who are at-risk for later reading difficulties as well as designing interventions and conducting progress monitoring (Hoffman et al., 2009). However, it was surprising that we still observed this pattern of catch-up in our study even in the kindergarten classrooms where teachers were most effective at promoting emergent literacy skill development (see Figure 1, Panel A).

As we consider the pattern of “catch-up” that was found in our study across classrooms with low, average, and high levels of teacher value-added, it is possible that another type of “substitution” may be at play in our data. In particular, it is possible that the experience of participating in kindergarten will act as a substitute for not participating in pre-K, regardless of whether teaching quality is high or low in kindergarten. This idea differs from the predictions of dynamic substitutability that we have articulated above, in which the pre-K boost was hypothesized to substitute or buffer children against a low-quality kindergarten experience. Instead, the idea that attending kindergarten could act as a substitute for attending pre-K is consistent with the idea that public education is primarily compensatory in nature. For example, Barnett (2011) has described the following scenario: “if early educational interventions have only small initial impacts, the compensatory efforts of public schools for children who enter school behind may succeed in offsetting the entire advantage” (p. 975). In this way, a kindergarten education might be most effective at promoting the skill development of children who enter school with the lowest skill levels—therefore substituting for the pre-K boost. However, we cannot test the effect of having no exposure to kindergarten in our data.

Building on Previous Research

Our findings support and extend the existing evidence base. Our study documented evidence of an initial NC Pre-K boost to emergent literacy skills as well as convergence—in part—during kindergarten. Although our study focused on short-term outcomes related to children’s emergent literacy skills, a larger body of research has documented longer-term benefits resulting from the rollout of funding for the NC Pre-K program (Bai et al., 2020; Dodge et al., 2016; Ladd et al., 2013). The benefits include positive effects for reading outcomes through eighth grade (Bai et al., 2020). Collectively, this body of research suggests that NC Pre-K programming can produce both short-term and long-term benefits, even in the context of short-term convergence for emergent literacy skills reported here. Therefore, the evidence base clearly warrants further support and expansion of NC Pre-K programming in North Carolina.

In the current study, we also found that teacher value-added was positively associated with children’s emergent literacy skill development. This finding was consistent with other studies investigating the impacts of teacher value-added on student learning in later grades, including first grade (Palardy & Rumberger, 2008). However, our study was potentially the first to link teacher value-added to children’s skill development during kindergarten. This finding indicates that measuring teacher value-added in kindergarten may offer practical insights into the effectiveness of teachers’ practice, if used in combination with professional development supports.

Another novel contribution of the current study is that we tested the hypothesis that kindergarten teacher value-added would lead to differential rates of growth among former NC Pre-K participants compared to non-participants through dynamic complementarity or substitutability. Given the null interaction reported between NC Pre-K participation and kindergarten value-added, the findings of our study did not support the notion that the NC Pre-K boost to emergent literacy was complemented by high-value-added teachers in kindergarten nor substitutive of low-value-added teachers. Despite a strong theoretical basis for such hypotheses (Bailey, Duncan, et al., 2020), the null interaction observed in our study is largely consistent with the extant literature, which has documented few reliable findings of complementarity or substitutability (Bailey, Jenkins, et al., 2020). This includes two prior studies of NC Pre-K, which found no evidence of complementarity or substitutability between NC Pre-K participation and subsequent educational environments in relation to children’s emergent literacy skills during kindergarten (Burchinal et al., 2022; Carr, Peisner-Feinberg, et al., 2021). Those studies examined broader measures of the subsequent educational environment and did not examine teacher value-added. Therefore, our null results contribute to narrowing the field of inquiry with regard to these hypotheses. Perhaps the results of our study simply indicate that high-quality teaching in kindergarten is important for the emergent literacy skill development of all children, regardless of whether they participated in NC Pre-K or not.

It should be noted that our study only observed a single measure of emergent literacy skills, and these findings do not preclude the possibility that dynamic effects could exist for other measures of child skills and behaviors. Indeed, prior studies have documented evidence of dynamic interactions for the effect of NC Pre-K in relation to language and executive functioning skills in kindergarten (Carr, Peisner-Feinberg, et al., 2021) and in relation to longer-term reading and math outcomes in fifth and eighth grade (Carr, Watts, et al., 2021; Watts et al., 2023). Collectively, these findings demonstrate that NC Pre-K can produce dynamic and broad ranging benefits for children’s skill development in kindergarten and later grades, even in the context of short-term convergence for emergent literacy skills reported here.

Opportunities for Future Research

Future research may continue to investigate factors that sustain the pre-K boost to children’s emergent literacy skills into elementary school. We discuss three promising avenues for future research. First, it is possible that greater specificity is needed when measuring the quality of children’s educational experiences in elementary school. Teacher value-added has been described as an approach to identifying a teacher’s “overall” contribution to student learning (Hanushek & Rivkin, 2010, p. 267). Perhaps measuring overall teacher quality was not specific enough to identify sustaining environments for individual children. Instead, researchers may need to assess the ways in which teachers differentiate and individualize their instruction to match the competencies of the individual children in their classrooms. Moreover, individualized and differentiated instruction may be necessary to further support the advanced learning of former pre-K participants and lead to sustaining the pre-K boost (Phillips et al., 2017). Indeed, a study by McCormick et al. (2022) examined the extent to which individual children experienced content-rich and cognitively demanding instruction in kindergarten as a sustaining environment—finding that these factors were associated with sustaining the pre-K boost to emergent literacy skills to some degree. Measuring children’s individual classroom experiences seems to be a fruitful path forward for future research investigating pre-K convergence.

Future research might also examine if the benefits of NC Pre-K can “spillover” to children who did not experience NC Pre-K but go on to enroll in the same kindergarten classrooms as former NC Pre-K participants. For example, it could be the case that having a higher proportion of NC Pre-K children in a kindergarten classroom would enable teachers to cover more advanced content and facilitate greater opportunities for peer learning. In other words, the presence of more pre‐K peers in a kindergarten classroom could lead to complementarity or substitutability. Although evidence to support this hypothesis was not found in recent research in Virginia (Burchinal et al., 2023), the issue has not been examined in relation to NC Pre-K. In the current study, our primary analyses considered kindergarten classroom environments in which non-NC Pre-K participants were exposed to a former pre-K participant—excluding classroom environments where non-NC Pre-K children lacked exposure to a NC Pre-K peer. The findings of the current study may therefore be consistent with the spillover hypothesis, because the skills of non-NC Pre-K participants grew closer to the skills of former NC Pre-K participants who were enrolled in the same kindergarten classrooms.

Unfortunately, our study was not able to examine instructional alignment across pre-K and kindergarten. Despite widespread recognition of the importance of alignment, prior research in North Carolina has found relatively weak vertical alignment between the standards, curricula, and assessments used across pre-K and kindergarten systems (Cohen-Vogel et al., 2020). Educators also report facing many obstacles to achieving alignment within and across systems (Cohen-Vogel et al., 2020). It is likely that more systematic efforts are needed to align classroom curriculum and instruction across pre-K and kindergarten, which may contribute to sustaining the pre-K boost during kindergarten. For example, the National P-3 Center has developed a framework “to improve the quality and coherence of children’s learning opportunities” across preschool through third grade (Kauerz & Coffman, 2013, p. 1). Policymakers and education stakeholders across North Carolina also recognize the importance of educational alignment, and some promising efforts are currently underway. For example, statewide rollout of the LETRS^® program began in 2021 to provide both NC Pre-K and elementary school teachers with training related to the science of reading (Language Essentials for Teachers of Reading and Spelling). Such efforts may help to sustain the NC Pre-K boost to emergent literacy skills in particular, and future research should be undertaken to examine the possible benefits of this program.

Strengths and Limitations

There are both strengths and limitations to the extent to which the results of this study may be broadly generalizable. First, we focused on pre-K programming and child development within the context of North Carolina and, therefore, these results were not intended to generalize to other state contexts. A consensus statement from leading pre-K researchers noted that “pre-K programs are not all equally effective” (Phillips et al., 2017, p. 23). Moreover, NC Pre-K is an established program with a robust body of evidence demonstrating positive effects (see Watts et al., 2023 for a review). The generalizability of our findings may be constrained in that light.

We can also consider how our findings may generalize within the context of North Carolina. Although we relied on statewide administrative data, we utilized matching procedures to enhance the potential for causal inferences to be drawn from our study. In particular, our primary analyses limited the matched sample of NC Pre-K participants to only those children with a non-participating peer enrolled in the same kindergarten classroom. This resulted in a sample of matched NC Pre-K participants that was substantially smaller than the input sample of NC Pre-K participants and presented some differences in sociodemographic characteristics (see column “c-a” in Table 1). We conducted sensitivity analyses that relaxed the requirement for within-classroom matching and retained > 99% of children in the NC Pre-K input sample (see Supplemental Materials). Those findings revealed similar estimates as those shown in our primary analyses. Therefore, the results of our study may offer some degree of generalizability to the broader state context. Further evidence to support the generalizability of our findings is demonstrated by the similarity between the estimates found in our study and those of a recent randomized control trial of NC Pre-K (Peisner-Feinberg et al., 2019). In the current study, we documented evidence of a 0.22 SD boost to children’s emergent literacy skills in the fall of kindergarten based on a teacher-administered assessment of children’s emergent literacy skills (i.e., DIBELS). Similarly, the randomized control trial also documented evidence of a 0.22 SD boost to children’s emergent literacy skills in the spring of pre-K, based on a researcher-administered measure (i.e., WJ III Letter-Word ID). The similarities in findings from these studies suggests some degree of generalizability across different methodologies, assessment tools, and school years in which the studies were conducted.

Our study was not able to measure children’s skill development across a broad range of early learning domains (e.g., language, mathematics). Therefore, our findings related to children’s emergent literacy may not generalize to other developmental domains. Moreover, we have noted that prior research suggests that the pre-K boost may be more likely to sustain into kindergarten in relation to children’s development of other “unconstrained” skills (Johnson et al., 2023; McCormick et al., 2021). Nonetheless, we have argued for the importance of focusing on emergent literacy skills in our study because the acquisition of these skills constitutes a major developmental task for young children and establishes the foundation for later literacy learning. Moreover, these skills constitute a primary focus of instruction in pre-K and kindergarten.

Some strengths of this study should be noted. First, in order to draw equitable comparisons between the outcomes of NC Pre-K participants and non-participants, we implemented “exact” matching on important child and family socioeconomic characteristics, and for kindergarten classroom membership. In addition to promoting equitable comparisons based on observed sociodemographic characteristics, implementing “exact” matching for kindergarten classroom membership may have accounted for other unobserved factors related to self-selection and school-level decision-making that would lead children to enroll in the same kindergarten classrooms. Second, we utilized a jackknife (leave-year-out) measure of teacher value-added that approximated teacher quality during the focal school year (i.e., 2014–15) by using scores from the previous and/or subsequent years. This method aimed to remove potential bias that could result by assessing child skills and teacher value-added using DIBELS during the same school year. Finally, we showed that our findings were robust to alternative analytic specifications.

There are also some limitations to this study that should be noted. First, we relied on quasi-experimental methods to estimate the NC Pre-K boost and moderation by teacher value-added—methods that may be prone to issues associated with selection bias. However, there was a high degree of consistency between the findings in our study and findings from previous randomized studies. Second, we were unable to examine the preschool experiences of non-NC Pre-K participants, such as Head Start. Indeed, prior research has shown that pre-K effects are smaller or null when compared to children who participated in Head Start, which offers comparable services to NC Pre-K (e.g., Zhai et al., 2011). Finally, we did not measure the experiences of individual children in kindergarten nor the alignment of curriculum across pre-K and kindergarten but relied on a classroom-wide measure of teacher quality instead.

Conclusion

The current study contributes new evidence to a growing body of research investigating factors that may sustain the pre-K boost into elementary school. This study focused on the NC Pre-K program—a high-quality, targeted pre-K program. This study then explored the potential contribution of kindergarten teacher quality (teacher value-added) to sustaining the NC Pre-K boost to children’s emergent literacy skills. The findings did not support the idea that kindergarten teacher quality differentiated the skill development of former NC Pre-K participants compared to matched non-participants. Instead, the current study documented a large, positive association between teacher value-added and growth in children’s emergent literacy skills, regardless of NC Pre-K participation. We advocate for pre-K programming and its benefits to be viewed as part of a continuum of high-quality educational experiences to be built on sequentially across the transition to elementary school. We suggest that future research focus on differentiated instruction, peer effects, and instructional alignment across pre-K and kindergarten.