Home Articles Reader Opinion Editorial Book Reviews Discussion Writers Guide About TCRecord
transparent 13
Topics
Discussion
Announcements
 

Older Versus Younger: The Role of Entry Age for Students Who Begin Kindergarten With Disabilities


by Michael A. Gottfried & Cameron Sublett - 2019

Background/Context: The age at which children can enter kindergarten continues to be discussed in both educational research and practice, and the debate for whether to increase kindergarten entry age remains active on both sides. A critical oversight has been the lack of attention paid towards entry age for those students who begin school with a disability (“SWDs”). The lack of empirical evidence in this domain is highly concerning given that statewide policies and practices that affect the general schooling population will now also be affecting SWDs who are increasingly being educated in general education classrooms and schools and are hence subject to general educational policies and practices.

Purpose/Objective/Research Question/Focus of the Study: Our study asked the following two research questions: 1. For children who begin kindergarten with a disability, does older versus younger entry age link to differences in short- and long-term achievement outcomes? 2. For children who begin kindergarten with a disability, does older versus younger entry age link to differences in short- and long-term socio-emotional measures?

Population/Participants/Subjects: This study utilizes data from the newly released Early Childhood Longitudinal Study—Kindergarten Class of 1998–1999. Data were collected in several waves: the fall and spring of kindergarten (1998-99), in the spring of 1st grade (1999-00), the spring of 3rd grade (2002), the spring of 5th grade (2004), and the spring of 8th grade (2007). We analyzed data from the elementary school waves.

Research Design: This study examined two sets of outcomes. First was reading and math achievement. The second set included socioemotional scales based on both teacher ratings of child behavior. These outcomes were regressed on a measure for having attended kindergarten at an older entry age as well as a wide span of child and family characteristics. Error terms were clustered at the school level to account for nested data.

Findings: The findings of the current study suggest there is little evidence that older kindergarten entry age links to differences in academic achievement for entrants with disabilities. However, older entrants with disabilities had much fewer instances of problem behaviors compared to those children with disabilities who began school at a younger age. Older entrants with disabilities also had higher social skills compared to entrants with disabilities who entered school at a younger age, though these findings were short-run, with little evidence extending beyond first grade.

Conclusions/Recommendations: Later kindergarten entry may be one way to help SWDs ease the transition into schooling, given that prior research has found that kindergarten entry for SWDs can elevate stress and anxiety for this vulnerable group of students. Hence, SWDs might be especially benefitting greatly from an extra year of maturation before beginning formal schooling. Future research might examine what interventions and developmental services might be best at boosting SWDs’ development during that additional time prior to school.



The age at which children can enter kindergarten continues to be discussed in both educational research and practice, and the debate for whether to increase kindergarten entry age remains active on both sides. Proponents of older kindergarten entry age (“KEA”) support that delaying entry provides children with more time to mature before starting kindergarten (Graue & DiPerna, 2000). This may be particularly relevant given a more challenging kindergarten curriculum that has arisen in recent years (Deming & Dynarski, 2011), which in turn has changed readiness expectations (Bassok, Latham, & Rorem, 2016) as well as teachers’ transition practices (Little, Cohen-Vogel, & Curran, 2016). Proponents of earlier entry age argue that time spent in school can often be more valuable than an additional year spent elsewhere (Stipek, 2002). An earlier start in kindergarten might be especially critical for families who have fewer resources and who could potentially benefit from school networks of supports and resources (Vecchiotti, 2001).


There are many theoretical arguments supporting older versus later age of entry. However, the empirical research generally finds early academic and developmental benefits of older kindergarten entry. Overall, the findings suggest that older entrants have higher test scores compared to younger entrants in early grades (Dagli & Jones, 2012; Datar, 2006; Malone, West, Flanagan, & Park, 2006; National Institute of Child Health and Development, 2007). The test score gaps between older and younger entrants, however, seem to close by the time children reach later elementary grades, thereby suggesting that older KEA may be linked strictly to early benefits (Datar, 2006; Kurdek & Sinclair, 2001; Oshima & Domaleski, 2006; Stipek & Byler, 2001).By middle school, most test score differences have disappeared (Datar & Gottfried, 2015;  Elder & Lubotsky, 2009). Two synthesis reports over the past several decades found the same patterns using an earlier body of research (Shepard & Smith, 1986; Stipek, 2002).


As for the link between entry age and developmental outcomes, the base of literature from which to draw conclusions is not quite as large. Nonetheless, all findings point to either positive or null effects of older entry age. For instance, Spitzer, Cupp, and Parke (1995) found that older kindergarten entrants had higher prosocial ratings and lower frequencies of aggressiveness in kindergarten, though the effects faded by first grade. Stipek and Byler (2001) found that older kindergarten entrants exhibited more positive feelings toward their teachers from kindergarten through 3rd grade; null effects arose on feelings toward their school, academic engagement, and teachers’ ratings of students’ social competence. The National Institute of Child Health and Development (2007) used caregivers’ and teachers’ reports to assess older versus younger frequencies of cooperation, assertion, responsibility, self-control, externalizing, and internalizing behaviors. No significant findings emerged. Finally, Datar & Gottfried (2015) found that older entrants exhibited fewer problem behaviors and greater social skills between kindergarten and 5th grade. Like with achievement, however, gaps between older and younger entrants had closed by middle school.


Most research surrounding the role of older KEA has focused on children in the general population (“GENs”). Little work has focused on any between-group differences, yet the limited work that does exist finds subgroup differences. For example,  Le, Kirby, Barney, Setodji, and Gershwin (2006) found that children from lower-income families had even higher test score gains when delaying kindergarten compared to children from higher-income families. Datar and Gottfried (2015) found children from immigrant backgrounds who enroll in kindergarten at older ages have higher achievement and socio-emotional outcomes in elementary school compared to immigrant children of younger ages. Though little subgroup analysis research exists besides these two studies, the limited body of work highlights that subgroup differences might emerge, and thus having insight into these differences might help to guide both policy and practice.


When examining between-group differences, a critical oversight has been the lack of attention paid toward KEA for those students who begin school with a disability (“SWDs”). Our study begins to address this research gap. In recent decades, there has been the push to educate SWDs to the maximum extent possible alongside GEN students in general education schools (McLeskey, Landers, Williamson, & Hoppey, 2012). There is evidence of this trend; most SWDs across the nation now receive a majority of their instruction from within the general educational classroom (U.S. Department of Education, 2012). As the proportion of SWDs continues to grow in the United States (Boyle et al., 2011), so will the number of SWDs who begin and continue through school in the general educational setting (U.S. Department of Education, 2012).


Though the population of children starting kindergarten in general education schools is changing, little attention has been paid to how entry age policies might be linked to outcomes for SWDs. The lack of empirical evidence in this domain is highly concerning given that statewide policies and practices that affect the general schooling population will now also be affecting SWDs who are increasingly being educated in general education classrooms and schools and are hence subject to general educational policies and practices. However, with minimal knowledge of how entry age policies play out differently for different students, it becomes difficult to inform policy decisions that will lead to supportive educational pathways for all. Therefore, our study asked the following two research questions:


1. For children who begin kindergarten with a disability, does older versus younger entry age link to differences in short- and long-term achievement outcomes?

2. For children who begin kindergarten with a disability, does older versus younger entry age link to differences in short- and long-term socio-emotional measures?


ENTRY AGE FOR CHILDREN WITH DISABILITIES


Older Entry


Like the debate of KEA for the general student population, there are proponents on both sides of the KEA-SWD debate. Proponents of an earlier KEA for SWDs rely on the argument that earlier entry to school provides this group of children with earlier access to necessary supports. For instance, kindergarten may be the first opportunity for many parents to gain access to resources for their children (Fenlon, 2005). Earlier entry would imply sooner access to critical instructional supports, special educational services, and disability diagnoses that children may not receive if they had waited an extra year to begin school (Geng, 2012). Earlier entry may also be providing sooner access to peer interaction. It has been established that SWDs often face difficulty integrating with peers (Pearl et al., 1998). Therefore, earlier entry to school would provide SWDs with even more early opportunities to forge relationships and develop socially.


Younger Entry


On the other side, proponents of older entry for SWDs rely on the theoretical argument that earlier entry does not allow for sufficient time for development. Similar to the argument for the GEN population, additional time before starting school may allow SWDs to mature. Specific to the population of SWDs, extra time before starting school enables for more evidence to surface of a child’s specific disability needs. With earlier entry into kindergarten, a child’s disability may still be unknown or not yet detectable, leading to flip-flopping of services through premature judgments on behalf of the school regarding academic abilities and hence leading to inappropriate academic supports (Fenlon, 2005; Geng, 2012). As for social issues, given the added stress that SWDs face with the transition to kindergarten (Janus, Lefort, Cameron, & Kopechanski, 2007; Kazak & Marvin, 1984; Ray, 2003), proponents of later entry support that additional time before starting school may allow for SWDs to mature prior to entry, hence potentially reducing this anxiety and increasing the ability to establish stronger relationships with peers and teachers.


Though there are proponents who support benefits of earlier versus later entry, there is an empirical void in the research to support either side, as mentioned. And given the scarcity of research in this area, it is challenging to do more than speculate. Therefore, our two research questions are highly relevant to moving forward both policy and practice.


We addressed our two questions using a large-scale dataset of students in the United States in which it was possible to track students starting in kindergarten. These data were unique in that they allowed for students’ test scores and social ratings to be linked to school entry cutoff dates and precise age at kindergarten entry.1 Therefore, by isolating the influence of older versus later entry for SWDs across elementary school, our findings can yield insight into what policy levers might be influencing both short- and long-term outcomes of SWDs. This will enable policymakers and practitioners to more efficiently identify key factors to build supportive educational pathways for SWDs.


METHOD


DATASET OVERVIEW


Data for the current study were sourced from the restricted version of the Early Childhood Longitudinal Study—Kindergarten (ECLS-K). Developed by the National Center for Education Statistics at the U.S. Department of Education, ECLS-K contains individual, family, teacher, and school-level characteristics of a nationally representative sample of children entering kindergarten. Because of this, ECLS-K provides rich and comprehensive data on kindergartners from a range of ethnic and socioeconomic backgrounds. Data were collected in several waves: the fall and spring of kindergarten (1998–99), the spring of 1st grade (1999–2000), the spring of 3rd grade (2002), the spring of 5th grade (2004), and the spring of 8th grade (2007) (Tourangeau, Nord, Le, Sorongon, & Najarian, 2009). We analyzed data from the elementary school waves (i.e., K–5) because our socio-emotional scales, as described below, were administered with the same assessment items only in the kindergarten through fifth grade waves.


Given that our research questions focused on the association between older entry age among SWDs, only students in the ECLS-K sample who entered the fall of kindergarten with a disability were included in the analytic sample. Students were identified as having a disability at the start of kindergarten based on a measure found within the dataset. It was constructed from a series of items of in which parents identified their children as being diagnosed with or receiving prior to kindergarten a diagnosis of one of the 13 categories of disabilities recognized under the federal Individuals with Disabilities Education Act (“IDEA”). Approximately 14% of the sample entering kindergarten in the fall of 1998 were considered to have a disability, which is reflective of national estimates. Prior ECLS-K studies on kindergarten entry age have used mean value imputation with missing value indicator dummy variables to address missing values (Datar & Gottfried, 2015; Datar, 2006), and therefore we took the same approach here.2 After imputation, there was a total of approximately n=1,960 SWDs in the analytic sample.


OUTCOMES


We examined the relationship between entry age and academic and socio-emotional student-level outcomes. The academic outcomes were item response theory math and reading scale scores. Assessment in both subject tests occurred in two stages. Stage one was a routing test which determined the difficulty level of the stage two test. Item response measures allowed for the comparison of scores across students, regardless of which second-stage form the students had taken. An additional benefit of using scaled scores is that they enabled for longitudinal assessment of achievement, even when the tests are not identical in each year. As reported in the user’s manuals of ECLS-K, the tests had high alpha reliability coefficients, ranging from 0.92 to 0.99 (Tourangeau, Nord, Le, Sorongon, & Najarian, 2009).


The next set of outcomes included five scales that related to students’ socio-emotional behaviors. These include two scales of problem behaviors and three scales of classroom social skills. These scales were based on the more general Social Skills Rating System (SSRS; Gresham & Elliot, 1990) and were modified by the National Center for Education Statistics for the purposes of data collection.


All five socio-emotional scales were teacher-rated. The two scales we refer to as student problem behaviors measured internalizing and externalizing behaviors. Internalizing behaviors were characterized by frequencies of loneliness, anxiety, sadness, and low self-esteem. In contrast to internalizing behaviors, externalizing behaviors were characterized by frequencies of anger, impulsivity, disturbing others, and fighting. The scale we used to measure students’ internalizing behaviors consisted of four items; the scale we used to measure externalizing behaviors consisted of five items. For both scales higher scores indicated increased internalizing or externalizing behaviors. Therefore, higher scores on these scales would represent unfavorable outcomes.


The three scales we refer to as student social skills measured self-control, approaches to learning, and the interpersonal skills among the students in the ECLS-K sample. Self-control measured a student’s ability to exert control over her or his temper, handle stress, and respect others’ property and ideas. This was a 4-item scale with high scores representing increased self-control, a favorable outcome. Approaches to learning measured a student’s ability to be organized for learning, adapt to change, focus, persist and follow classroom rules. This was a 7-item scale with high scores representing desirable approaches to learning, a favorable outcome. Last, interpersonal skills, measured a student’s ability to socialize with his or her peers, maintain friendships, empathize, listen, and express ideas. This was a 5-item scale with high scores representing higher interpersonal skills, a favorable outcome. As reported in the ECLS-K user’s manual, all five scales had high internal consistency with alpha coefficients ranging from 0.79 to 0.89 (Tourangeau, Nord, Le, Sorongon, & Najarian, 2009).


Table 1. Control Measures

 

 

 

 

 

 

 

 

Control measures

 

Mean

 

sd

 

 

 

 

Kindergarten entrance age

5.46

 

0.39

 

 

 

 

Demographic data

 

 

 

Male

0.68

 

 

White

0.68

 

 

Black

0.12

 

 

Hispanic

0.12

 

 

Asian

0.02

 

 

Other

0.05

 

 

English is home language

0.95

 

 

 

 

 

 

Pre-Kindergarten care

 

 

 

Center-based care

0.81

 

 

Care by relatives

0.23

 

 

Care by nonrelatives (non-center)

0.25

 

 

Head Start

0.20

 

 

 

 

 

 

Family data

 

 

 

Below poverty line

0.21

 

 

Number of adults in household

1.99

 

0.66

Number of siblings in household

1.49

 

1.09

Mother has less than high school education

0.03

 

 

 

 

 

 

School Data

 

 

 

School size (shown here:0–149 students)

0.06

 

 

Less than 10% minority in school

0.38

 

 

Private school

0.19

 

 

Northeast

0.22

 

 

Midwest

0.27

 

 

South

0.36

 

 

West

0.15

 

 

 

 

 

 

n

1,960

 

 

 

 

 

 

Note. In this table, sd is presented for continuous variables.

 

 

 


ENTRY AGE


Table 1 presents all independent variables taken from the first wave of data collection. Because ECLS-K contains information on every student’s birthdate as well as the start date of kindergarten, we were able to compute all students’ entry age with precision.  In more detail, we took the age at which students in the ECLS-K sample began kindergarten (measured in months) and divided that number by 12. Table 1 presents mean student age at the start of kindergarten for students in the ECLS-K sample who entered kindergarten with a disability.


CONTROL VARIABLES


We took advantage of the data available in ECLS-K and included a set of control variables. We were careful in constructing these sets of controls so that the measures were sourced from the same wave as a given outcome variable. For instance, ECLS-K contains information on time-varying student characteristics such as socioeconomic status (SES). Because it is entirely possible that SES can change from kindergarten to, for example, 5th grade, a single measure of SES taken during the kindergarten wave of data collection may not have sufficiently controlled for the influence of SES in that particular year. As a result, we ensured that we first controlled for any time-invariant characteristics, such as race. When these characteristics were time varying, we were sure to include the wave-specific measure for this characteristic.3


Time-invariant controls included gender, race/ethnicity, and an indicator of whether English was a student’s first language. In addition, we included controls for students’ pre-kindergarten care including whether students attended center-based care or Head Start in pre-kindergarten or remained in the care of a relative in this year. Time-varying controls included family-level variables. These measures were: an indicator of being below the poverty threshold, household composition (i.e., the number of adults and siblings in the household), and mother’s education level.4 Finally, the last set of control measures represented school-level characteristics. These measures included the following: school size (which were a series of categorical measures for total enrollment, such as 1–149 students), the percentage of minority students, an indicator of private school status, and an indicator for geographic region.


Prior research (Datar & Gottfried, 2015; Datar, 2006) found few, if any, differences in the observable characteristics of nondisabled students entering kindergarten at older versus younger ages. A question that remained at the start of this study, however, was whether the same pattern held true for the sample of SWDs. Table 2 lists observable characteristics once again; however, in this table, each measure was broken out by KEA, which we have divided into four age brackets, as guided by prior literature (Dagli & Jones, 2013; Malone et al., 2006): less than 5 years, 5 to 5.5 years, 5.5 to 6 years, and greater than 6 years of age. Roughly 6% of the SWDs in the sample entered kindergarten at less than 5 years of age; 45% began kindergarten between 5 and 5.5 years; nearly 40% entered between 5.5 and 6; and just 5% entered kindergarten over 6 years of age. Table 2 is consistent with what prior research had found for the general student population (Datar & Gottfried, 2015; Datar, 2006). On the whole, there did not appear to be systematic differences between SWDs entering kindergarten at different ages. Like in Datar and Gottfried (2015), there did appear to be geographic differences, hence further motivating the need to include these as controls in our analyses to follow.


Table 2. Descriptive Statistics for SWD Sample by KEA

[39_22577.htm_g/00002.jpg]

Note. In this table, sd is presented for continuous variables.



METHOD


We began by employing the following ordinary least squares (OLS) baseline model:


[39_22577.htm_g/00004.jpg]


where Y represents the academic achievement or socio-emotional outcome for student i in survey wave t. The elements on the right side of the model are as follows: KEA (central to this analysis) represents a student’s age at kindergarten entry; I represents the set of individual student characteristics (both time invariant and time varying); F is a set of family background characteristics (both time invariant and time varying); and S is a set a of school-level factors in year t. Because student observations are nested within schools and hence their experiences are not fully independent, we have chosen to cluster the error term at the school level. Note that we have estimated each outcome separately by survey wave—a necessity given that teacher-rated socio-emotional outcomes were scored by different teachers in each survey wave.


A constant challenge for researchers using secondary or observational data is the issue of omitted variable bias. It is quite possible that a number of underlying mechanisms may have, in addition to KEA, simultaneously affected students’ academic achievement or socio-emotional behaviors. For instance, students in highly involved families might have entered kindergarten at a later age for reasons unobserved to the researcher. It would be highly likely that this high level of involvement might have also influenced our set of outcomes. Therefore, in the OLS model above, it was not possible to identify whether the association between KEA and our set of outcomes was due to entry age or parental involvement.


While large-scale data sets like ECLS-K offer an array of control measures to account for some of the influence of these underlying mechanisms, it is impossible to account for all potential influences. In these situations, and in situations when randomized experimentation is not feasible, researchers can employ a number of research designs that can significantly mitigate the bias resulting from omitted variables. One such technique is instrumental variable estimation (IV), which relies on the use of a unique variable within a given data set that satisfies the dual conditions of predicting the explanatory variable (e.g., KEA) and being entirely unrelated with any omitted variables. In satisfying these conditions, the instrument allows researchers to essentially identify an exogenous portion of an endogenous predictor variable to estimate the impact on a given outcome (Murnane & Willett, 2011).


IV proceeds in a two-stage process. In the first stage, our treatment variable (i.e., KEA) was regressed on the exogenous predictor of entry age in addition to any control measures to be used in the second stage. For the current study, this first stage model is represented with the following model,


[39_22577.htm_g/00006.jpg]


where DAYS represents this exogenous predictor of entry age: the number of days between the school kindergarten entry cutoff in a given state and student’s birthday, and thus, in this case functions as the instrument; Z represents all the control measures included in the original OLS model described above. It is important to note that our use of the number of days between the kindergarten entrance cutoff and a student’s birthday as an exogenous predictor of KEA has been corroborated in prior research (Datar and Gottfried, 2015; . This instrument was selected in this study, as well as in prior literature, thanks to the fact that the number of days between a student’s birthday and the entry cutoff date was plausibly exogenous because it is highly improbable that students’ birthdays were planned around a given state’s cutoff date for school entry. Consequently, we argue that this introduces exogenous variation sufficient to model our outcomes of interest while also reducing the impact of omitted variable bias.


Interestingly, kindergartners whose birthdays were close to (but before) the cutoff will be younger than kindergartners whose birthdays were close to (but after) the cutoff. As a most extreme example, students whose birthdays were only a few days before the cutoff would be almost an entire year older than those students whose birthdays were a few days after the cutoff. Given this example, note that we also controlled for birth month in every model, even in our baseline models, due to the fact that children born in the months closer to the cutoff will have naturally higher/lower ages compared to children born in the winter or spring (Datar, 2006).


In the second stage of using an IV analysis, the dependent variable was student academic achievement (measured by math and reading IRT test scores) or socio-emotional behaviors. The outcome was then regressed onto the fitted-values of the treatment variable, KEA, generated by the first stage model, in which an arguably random source of variation predicted this entry age. Control measures were also included along with birth month fixed effects and a school clustered-adjusted error term.


Importantly, the power of IV estimation to produce improved estimates over OLS rests on the appropriateness of the strategy. The instrument must (1) be correlated with endogenous predictors and (2) be unrelated or orthogonal to the errors (Buam, Shaffer, & Stillman, 2003). Beyond these requirements, however, instruments must provide sufficient explanatory power. One way of testing the association between an instrument and the endogenous predictors as well as the testing for satisfactory explanatory power is to examine F statistics associated with the first stage regression. We observed a range of F statistics, depending on the outcome. This range was from F = 13.21 to 108.92, which was beyond the suggested F statistic of 10 as reported by Staiger and Stock (1997).


RESULTS


ENTRY AGE AND ACHIEVEMENT


Table 3 presents the OLS and IV estimates of the relationship between KEA and both reading and math achievement, as measured by achievement scores. All models included the set of control measures we previously described but were not presented for the sake of clarity. The outcomes have been standardized and, therefore, the reported coefficients can be interpreted as increases or decreases in standard deviation units (i.e., effect sizes).


Table 3. Reading and Math Achievement

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

IRT Achievement Scores

 

 

 

 

 

 

 

 

 

Reading Achievement

 

Math Achievement

 

OLS

 

IV

 

OLS

 

IV

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Fall Kindergarten

0.11

 

0.14

 

0.32***

 

0.35*

 

(0.08)

 

(0.16)

 

(0.09)

 

(0.15)

Spring Kindergarten

0.10

 

0.17

 

0.27**

 

0.41*

 

(0.08)

 

(0.15)

 

(0.10)

 

(0.17)

Spring Grade 1

0.05

 

0.21

 

0.12

 

0.29

 

(0.09)

 

(0.16)

 

(0.10)

 

(0.18)

Spring Grade 3

0.10

 

0.12

 

0.08

 

0.08

 

(0.10)

 

(0.18)

 

(0.10)

 

(0.19)

Spring Grade 5

0.09

 

0.17

 

0.00

 

0.12

 

(0.09)

 

(0.17)

 

(0.09)

 

(0.17)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Note: Robust standard error in parentheses
*** p < 0.001, ** p < 0.01, * p < 0.05.  

 

 

 

 

 

 

 


Looking first at the OLS estimates, there appears to be no relationship between entry age and reading scores. On the other hand, the results from the OLS math models indicated that SWDs who entered school at an older age had higher scores on math tests in kindergarten, though this association was no longer statistically significant past kindergarten. There was a similar overall pattern when looking at the IV estimates for SWDs shown in Table 3. For SWDs, there was still no benefit of older entry age on reading outcomes. As for math, the interpretation remained the same as in the OLS models—SWDs who entered kindergarten at an older age had higher math scores in kindergarten. The size of this association was larger in the IV models than in the OLS models, suggesting a previous underestimation in the size of [39_22577.htm_g/00008.jpg]. As before, however, the link between older entry and math outcomes faded by 1st grade.


ENTRY AGE AND SOCIO-EMOTIONAL OUTCOMES


Problem Behaviors


Table 4 provides the OLS and IV estimates of the associations between older KEA and internalizing and externalizing behaviors. As before, the outcomes have been standardized such that coefficients can be interpreted in standard deviation units. Lower values in both internalizing and externalizing behaviors were associated with fewer frequencies of these behaviors. Therefore, negative coefficients of KEA on problem behaviors indicated a favorable finding.


Relying on the OLS model alone, the estimates of older entry age and internalizing and externalizing behaviors among students who entered kindergarten with a disability would have suggested that, by and large, older KEA was not linked to differences in frequencies of problem behaviors. The one exception appears to be a statistically significant increase of internalizing behaviors in grade 5. However, this finding appeared to be an anomaly given all other OLS findings.


On the other hand, the IV estimates in the table do suggest that older entry was associated with lower values in both internalizing and externalizing behaviors. Hence, once considering the IV strategy, the association between older entry and these outcomes among SWDs appeared to have been underestimated: the OLS models would have suggested no association, whereas the IV strategy suggested otherwise. The overall interpretation from Table 3 is that older kindergarten entrants tended to display fewer problem behaviors across both measures, though the associations dissipated by the time these students reached later elementary grades.


Table 4. Problem Behaviors

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Problem Behaviors

 

 

 

 

 

 

 

 

 

Internalizing

 

Externalizing

 

OLS

 

IV

 

OLS

 

IV

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Fall Kindergarten

-0.17

 

-0.46*

 

-0.04

 

-0.43*

 

(0.11)

 

(0.22)

 

(0.11)

 

(0.21)

Spring Kindergarten

-0.15

 

-0.13

 

0.01

 

0.60**

 

(0.11)

 

(0.20)

 

(0.11)

 

(0.19)

Spring Grade 1

0.11

 

0.00

 

-0.01

 

-0.35

 

(0.11)

 

(0.21)

 

(0.11)

 

(0.21)

Spring Grade 3

-0.01

 

0.03

 

-0.16

 

-0.45*

 

(0.10)

 

(0.21)

 

(0.10)

 

(0.22)

Spring Grade 5

0.28**

 

-0.04

 

0.04

 

-0.27

 

(0.09)

 

(0.20)

 

(0.09)

 

(0.20)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Note: Robust standard error in parentheses
*** p < 0.001, ** p < 0.01, * p < 0.05.  

 

 

 

 

 

 

 


Social Skills


Table 5 presents the OLS and IV estimates of the relationship between later kindergarten entry and students’ social skills. These skills include “approaches to learning,” “self-control,” and “interpersonal skills.” Each of these outcomes has been standardized. Higher values on these outcomes represented favorable outcome



Table 5. Social Skills

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Social Skills

 

 

 

 

 

 

 

 

 

 

 

 

 

Approaches to learning

 

Self control

 

Interpersonal skills

 

OLS

 

IV

 

OLS

 

IV

 

OLS

 

IV

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Fall Kindergarten

0.26*

 

0.83***

 

0.03

 

0.15

 

0.16

 

0.43*

 

(0.10)

 

(0.19)

 

(0.10)

 

(0.19)

 

(0.10)

 

(0.19)

Spring Kindergarten

0.17

 

0.60**

 

0.01

 

0.11

 

0.08

 

0.20

 

(0.10)

 

(0.19)

 

(0.10)

 

(0.19)

 

(0.10)

 

(0.20)

Spring Grade 1

0.05

 

0.27

 

0.05

 

0.34

 

0.04

 

0.33

 

(0.09)

 

(0.19)

 

(0.10)

 

(0.19)

 

(0.09)

 

(0.19)

Spring Grade 3

0.16

 

0.41*

 

0.08

 

0.30

 

0.15

 

0.30

 

(0.09)

 

(0.19)

 

(0.10)

 

(0.20)

 

(0.10)

 

(0.21)

Spring Grade 5

-0.14

 

0.11

 

-0.07

 

0.10

 

-0.03

 

0.02

 

(0.09)

 

(0.18)

 

(0.08)

 

(0.17)

 

(0.08)

 

(0.18)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Note: Robust standard error in parentheses
*** p < 0.001, ** p < 0.01, * p < 0.05.

 

 

 

 

 

 

 

 

 

 

 


The OLS estimates in Table 5 suggested that older entry was linked to higher approaches to learning behaviors in the fall of kindergarten, after which this association diminished and became no longer significant. Also, in the OLS models, there appeared to be no relationship between self-control and interpersonal skills or older entry for SWDs.


There was a slightly different interpretation when looking at the IV estimates, however. Like the OLS estimates, the IV estimates indicated that KEA was positively related to approaches to learning, though the size and longevity of the effects were much greater in the IV models compared to the OLS models. In contrast to the OLS results where the findings were null, the IV estimates suggested that older entry was in fact associated higher interpersonal skills in the fall of kindergarten. The OLS models would suggested only null findings.


DIFFERENCES BY DISABILITY


In our final set of analyses, we examined whether differences arose for SWDs with different types of disabilities. Given the sampling design of the dataset, there is not sufficient sample size to examine students in each of 13 federal categories of disabilities. Therefore, SWDs were grouped based on Gottfried (2014): emotional or learning disabilities, speech impairments, mental retardation or developmental delays, and physical impairments. Based on these groupings, the IV models were rerun separately each specific group. Table 6 presents the KEA coefficients for the model specified by group row and outcome column. Only grades K and 1 are presented, as no significant findings emerged whatsoever in later grades.



Table 6. By Disability Category

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Reading Achievement

Math Achievement

Internalizing

Externalizing

Approaches to Learning

Self Control

Interpersonal Skills

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Learning or Emotional Disabilities

 

 

 

 

 

 

 

Fall Kindergarten

0.35

0.42

-0.12

-0.10

0.23

-0.35

-0.23

Spring Kindergarten

0.55*

0.41

0.04

-0.05

-0.03

-0.14

-0.11

Spring Grade 1

0.29

-0.01

0.27

-0.24

0.07

-0.07

-0.46

Speech Impairments

 

 

 

 

 

 

 

Fall Kindergarten

-0.68

-0.25

-0.05

-0.34

0.21

-0.22

-0.08

Spring Kindergarten

-0.22

-0.26

-0.35

-0.22

0.55

-0.17

-0.25

Spring Grade 1

-0.14

-0.07

0.03

0.62

-0.02

-0.30

-0.12

Mental Retardation or Developmental Delays

 

 

 

 

 

 

 

Fall Kindergarten

-2.12

0.78

-0.57

-1.19

0.05

0.03

0.13

Spring Kindergarten

-3.24

0.81

0.18

-0.40

-1.39

-1.28

-0.49

Spring Grade 1

-0.63

-0.75

0.25

-0.66

-0.32

-0.02

-0.02

Physical Impairments

 

 

 

 

 

 

 

Fall Kindergarten

0.24

0.54**

-0.14

-0.05

0.62***

0.18

0.36*

Spring Kindergarten

0.19

0.52**

0.01

-0.01

0.24

0.04

-0.00

Spring Grade 1

0.32

0.36

0.07

-0.19

0.47**

0.31

0.15

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

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

 

 

 

 

 

 

 


The results in the table do not suggest a great deal of heterogeneity by disability, as evidenced by the general lack of statistical significance throughout the table. With that said, there was some evidence that children with physical disabilities may have higher math scores, higher frequencies of approaches to learning, and higher interpersonal skills when entering kindergarten at an older age. Still, it does not appear that the age at which SWDs enter kindergarten matters differently for different disability groups. Although not shown, the results for OLS models also looked similar to these. We do urge some caution with these findings; while there was a general lack of statistical significance, many of the estimated coefficients were quite large. Perhaps with greater statistical power a different pattern would have emerged.


DISCUSSION


There are more SWDs than ever before entering general kindergarten. Hence, an increasing number of SWDs are held accountable to general schooling policies and practices, such as kindergarten entry cutoffs. There are a multitude of aforementioned studies on how children from the general population have fared in regard to older versus younger kindergarten entry; it may certainly be the case that the generally positive results for the general population in regard to kindergarten entry have provided an impetus for state and district policy to move forward with increasing school cutoffs. However, SWDs have been overlooked. Not understanding the experience of SWDs in this regard has been a significant oversight. Prior to this study, then, policy has taken hold without a full exploration of the heterogeneity in the student population. Much of the debate, in other words, has relied on speculation and assumption.


In addition to exploring this overlooked area of research, the current study was strengthened by the use of a plausibly exogenous instrumental variable. Because parents are, arguably, unable to influence the number of days between school entry cutoff and a student’s birthday, the current study was able to exploit seemingly random variation to estimate the association between KEA coefficients and academic and socio-emotional outcomes. This represents an improvement over basic OLS modeling which, as our results suggested, biased estimates downward. The current study was also strengthened through the use of longitudinal data. Our results indicated that, in many cases, the association of older entry to our set of outcomes tapered over time. Most often the association with older entry into kindergarten among SWDs was less persistent and diminished more rapidly compared to GENs. This was an important finding with significant policy implications that would have been undetected without the availability of longitudinal data.


LIMITATIONS


The current study is not without some limitations. First, our study was based on secondary data analysis and thus lacked a true experimental design. Hence, while we have attempted to reduce the influence of omitted variable bias, it is not for certain that the estimates are causal. Therefore, we caution against using language that would insinuate any unequivocal causal effects. Further research might explore extraneous variables that could be missing in the analysis and hence prohibiting from a true causal estimate from being derived. These might include measures pertaining to family involvement at home and school with regard to having children with disabilities, teacher preparation and experience and attitudes toward working with students with disabilities, and school resources and support systems. Likely, however, attaining these data exist on a smaller scale than using nationally representative data.


Second, our study must rely on outcomes that were constructed in the dataset for the general student population. Outcomes that are tailored more specifically to SWDs might yield different findings that are potentially more relevant to this group of students. Finally, while longitudinal, the outcomes of interest did not go past grade 5. Therefore, we can assess the effects of KEA through the end of elementary school, though the differences between older and younger school entrants may emerge throughout adolescence. Smaller scale longitudinal studies that follow children from birth into adulthood may be useful in this regard. Nonetheless, though there are some limitations, this study was significant in bringing to surface an understudied yet highly critical issue.


CONCLUSIONS


Overall, the results provided a number of important conclusions. With regard to academic achievement, there was no relationship between older entry age and reading test scores for SWDs in the sample, and while older entry age was related to increased math scores, this relationship diminished completely by the end of kindergarten. This suggests that, as a policy lever, delaying kindergarten entry may have limited utility for improving academic achievement for SWDs.


When looking at teacher-rated problem behavior outcomes, our model estimates suggested that older entrance into kindergarten was associated with lower frequencies of internalizing and externalizing behaviors among SWDs. One caveat, however, was that the associations between KEA coefficients and problem behaviors among SWDs were strongest at the start of kindergarten and for internalizing only lasted through the fall of kindergarten; for externalizing, the statistical significance carried over into 1st grade.


The estimates of the association between KEA and problem behaviors provide an interpretation of mixed sentiments. On one hand, it appears that older kindergarten entry was linked to lower frequencies of problem behaviors for SWDs. Hence, later kindergarten entry may be one way to help SWDs ease the transition into schooling, given that prior research has found that kindergarten entry for SWDs can elevate stress and anxiety for this vulnerable group of students (Janus et al., 2007). Hence, SWDs might be especially benefitting greatly from an extra year of maturation before beginning formal schooling. While not possible with the data used in this study, future research might examine what interventions and developmental services might be best at boosting SWDs’ development during that additional time prior to school. On the other hand, however, the IV estimates suggest that the benefits of older entry were short-lived: the positive associations do not persist beyond the first grade. As the decision to delay school entry can be potentially very costly for parents, we must ask if the benefits of delayed enrollment on reducing short-term problem behaviors outweigh the costs. We therefore must interpret these findings with policy objectives in mind, depending on whether the goal is to promote short- versus long-term developmental outcomes, or both. The policy objective itself would yield different interpretations to these findings.


With regard to social skills, the interpretation between older entry and these outcomes was slightly more mixed. On one side, older entry age was linked to higher approaches to learning outcomes for SWDs, and the size of this association was quite large at just over eight tenths of a standard deviation. However, the statistical significance of this association did not persist past the 1st grade. Entry age was not linked to higher levels of self-control, and was associated only with increased interpersonal skill through the fall of kindergarten.


In sum, our results provide reasons to be both skeptical and supportive of policies geared at increasing KEA for SWDs. That older entry age was not strongly associated with increased math achievement or reading achievement among SWDs is reason to doubt the efficacy of delaying of school entry age in order to boost academic performance. Furthermore, in instances where older entry age was associated with positive achievement outcomes, these relationships never persisted through 1st grade. For these reasons, we are skeptical that any potential expansion of delayed enrollment policies will yield substantial and sustained academic benefits.


With this said, it may be the case that any positive association between entry age and academic or socio-emotional outcomes is beneficial, regardless of persistence. This is for two reasons. First, even though a SWD may experience a shorter-lived benefit from older entry age, it is plausible that this association could boost performance in other areas of child development. In other words, it may be the case that a subtle boost in, for example, interpersonal skills among SWDs in kindergarten could lead to other intangible, and hitherto this point, undiscovered benefits. As a consequence, the impact of older entry age policies may be more meaningful than our results suggest.


Second, despite the fact that any positive associations with older entry age and our outcomes did not persist very long into a child’s schooling trajectory, again it may be the case that older KEA influences other mechanisms in a child’s trajectory that relate to long-term or more sustained academic or socio-emotional improvement. For instance, the association between older entry age and internalizing and externalizing behaviors was strongest among the SWD sample in our study and, while a critical assessment of these results may highlight how quickly they diminished, it may be the case that strong associations among these students, however short-lived they may be, actually lead to lasting impacts in other, less quantifiable dimensions, such as quality of life, child happiness, and family well-being.


Notes


1. While a more recent version of these data, Early Childhood Longitudinal Study of 2010–11, is available, it is not possible to calculate each student’s precise age at kindergarten entry because exact state cutoff dates for school entry were not provided. Additionally, the 2010–2011 dataset at present does not extend through elementary school as the dataset in our present study did.


2. While previous research has upheld the relative benefits of multiple imputation to mean value imputation with dummy indicators (e.g., (McCleary, 2002; Royston, 2004; Schafer, 1999), Stata does not currently provide support for multiple imputation and instrumental variables estimation, our analytic approach as described below.


3. Another concern was the potential for a student’s disability status to change over the course of the ECLS-K study. To ensure that late entry was not predictive of changes in disability status, we ran our models to follow with a “change in disability” status as the outcome and kindergarten entry age as the key predictor. The estimates were not statistically significant: students who entered kindergarten at an older age were not more likely to experience a “change in disability” status over the course of the waves of ECLS-K.


4. Note that we ran our models to follow with and without the time-varying control variables. The results did not change.


References


Bassok, D., Latham, S., & Rorem, A. (2016). Is kindergarten the new first grade? AERA Open, 2, 2332858415616358.

Bedard, K., & Dhuey, E. (2006). The persistence of early childhood maturity: International evidence of long-run age effects. The Quarterly Journal of Economics, 121(4), 1437–1472.

Boyle, C. A., Boulet, S., Schieve, L. A., Cohen, R. A., Blumberg, S. J., Yeargin-Allsopp, M., … Kogan, M. D. (2011). Trends in the prevalence of developmental disabilities in US children, 1997–2008. Pediatrics, 127(6), 1034–42.

Datar, A., & Gottfried, M. A. (2015). School Entry Age and Children’s Social-Behavioral Skills: Evidence From a National Longitudinal Study of U.S. Kindergartners. Educational Evaluation and Policy Analysis, 37(3), 333–353. 

Dagli, U. Y., & Jones, I. (2012). The effects of on-time, delayed and early kindergarten enrollment on children’s mathematicsematics achievement: Differences by gender, race, and family social-economic status. Educatoinal Sciences: Theory and Practice, 12(4), 3061–3074.

Datar, A. (2006). Does delaying kindergarten entrance give children a head start? Economics of Education Review, 25(1), 43–62.

Deming, D., & Dynarski, S. (2011). The lengthening of childhood. The Journal of Economic Perspectives, 22(3), 71–92.

Elder, T. E., & Lubotsky, D. H. (2009). Kindergarten entrance age and children’s achievement: Impacts of state policies, family background, and peers. Journal of Human Resources, 44(3), 641–683.

Fenlon, A. (2005). Paving the way to kindergarten for young children with disabilities. Arlington, VA.

Geng, L. (2012). Reason to delay kindergarten.

Gottfried, M. A. (2004). Classmates with disabilities and students' noncognitive outcomes. Educational Evaluation and Policy Analysis, 36(1), 20-43.

Graue, M. E., & DiPerna, J. (2000). Redshirting and early retention: Who gets the “gift of time” and what are its outcomes? American Educational Research Journal, 37(2), 509–534.

Gresham, F. M., & Elliot, S. M. (1990). Social skills rating system. Circle Pines, MN: American Guidance Service. Retrieved from http://www.researchconnections.org/childcare/resources/305

Janus, M., Lefort, J., Cameron, R., & Kopechanski, L. (2007). Starting kindergarten: Transition issues for children with special needs. Canadian Journal of Education, 30, 628–648.

Kazak, A. E., & Marvin, R. (1984). Differences, difficulties, and adaptation: Stress and social networks in families with a handicapped child. Family Relations: Journal of Applied Family & Child Studies, 33, 67–77.

Kurdek, L. A., & Sinclair, R. J. (2001). Predicting reading and mathematics achievement in fourth-grade children from kindergarten readiness scores. Journal of Educational Psychology, 93(3), 451–455.

Le, V.-N., Kirby, S. N., Barney, H., Setodji, C. M., & Gershwin, D. (2006). School Readiness, Full-Day Kindergarten, and Student Achievement: An Empirical Investigation. Santa Monica, CA: RAND.

Little, M. H., Cohen-Vogel, L., & Curran, F.C. (2016). Facilitating the transition to kindergarten: What ECLS-K data tell us about school practices then and now. AERA Open, 2, 1–18.

Malone, L. M., West, J., Flanagan, K. D., & Park, J. (2006). The early reading and mathematicsematics achievement of children who repeated kindergarten or who began school a year late (No. NCES 2006-064). Washington, DC.

McCleary, L. (2002). Using multiple imputation for analysis of incomplete data in clinical research. Nursing Research, 51(5), 339–343.

McLeskey, J., Landers, E., Williamson, P., & Hoppey, D. (2012). Are we moving toward educating students with disabilities in less restrictive settings? Journal of Special Education, 46, 131–140.

Murnane, R. J., & Willett, J. B. (2011). Methods matter: Improving causal inference in educational and social science research. New York, NY: Oxford University Press.

National Institute of Child Health and Development (NICHD). (2007). Age of entry to kindergarten and children’s academic achievement and socioemotional development. Early Education and Development, 18(2), 337–368.

Oshima, T. C., & Domaleski, C. S. (2006). Academic performance gap between summer-birthday and fall-birthday children in grades k-8. The Journal of Educational Research, 99(4), 212–217.

Pearl, R., Farmer, T. W., Van Acker, R., Rodkin, P. C., Bost, K. K., Coe, M., & Henley, W. (1998). The social integration of students with mild disabilities in general education classrooms: Peer group membership and peer-assessed social behavior. Elementary School Journal, 99, 167–185.

Ray, L. D. (2003). The social and political conditions that shape special-needs parenting. Journal of Family Nursing, 9, 281–304.

Royston, P. (2004). Multiple imputation of missing values. The Stata Journal, 4(3), 227–241.

Schafer, J. L. (1999). Multiple imputation: a primer. Statistical Methods in Medical Research, 8(1), 3–15.

Shepard, L. A., & Smith, M. L. (1986). Synthesis of research on school readiness and kindergarten retention. Educational Leadership, 44(3), 78–86.

Smith, J. (2009). Can regression discontinuity help answer an age-old question in education? The effect of age on elementary and secondary school achievement. The B.E. Journal of Economic Analysis & Policy, 9(1), 1–30.

Spitzer, S., Cupp, R., & Parke, R. D. (1995). School entrance age, social acceptance, and self-perceptinos in kindergarten and first grade. Early Childhood Research Quarterly, 10(4), 433–450.

Staiger, D., & Stock, J. H. (1997). Instrumental variables reression with weak instruments. Econometrica, 65(3), 557–586.

Stipek, D. (2002). At what age should children enter kindergarten? A question for policy makers and parents. Social Policy Report, 16(2). Retrieved from http://cepa.stanford.edu/content/what-age-should-children-enter-kindergarten-question-policy-makers-and-parents

Stipek, D., & Byler, P. (2001). Academic achievement and social behaviors associated with age of entry into kindergarten. Journal of Applied Developmental Psychology, 22(2), 175–189.

Tourangeau, K., Nord, C., Lê, T., Sorongon, A. G., & Najarian, and M. (2009). Combined user’s manual for the ECLS K-eighth-grade and K–8 full sample data files and electronic codebooks (NCES 2009–004), 99, 1–217.

U.S. Department of Education. (2012). Transforming American education: Learning powered by technology. Washington, DC. Retrieved from https://www.ed.gov/sites/default/files/NETP-2010-final-report.pdf

Vecchiotti, S. (2001). Kindergarten: The overlooked school year. New York, NY: Foundation for Child Development. Retrieved from http://fcd-us.org/resources/kindergarten-overlooked-school-year?destination=resources/search?page=34




Cite This Article as: Teachers College Record Volume 121 Number 3, 2019, p. 1-24
https://www.tcrecord.org ID Number: 22577, Date Accessed: 10/22/2021 3:08:25 PM

Purchase Reprint Rights for this article or review
 
Article Tools
Related Articles

Related Discussion
 
Post a Comment | Read All

About the Author
  • Michael Gottfried
    Gevirtz Graduate School of Education; University of California, Santa Barbara
    E-mail Author
    MICHAEL A. GOTTFRIED, PhD, is an Associate Professor in the Gevirtz Graduate School of Education. His research focuses on the economics of education and education policy. Recent publications include "Linking Getting to School With Going to School" (Educational Evaluation and Policy Analysis).
  • Cameron Sublett
    Pepperdine University
    E-mail Author
    CAMERON SUBLETT, PhD, is an Associate Professor of Education in the Graduate School of Education and Psychology at Pepperdine University. His research examines policy, leadership, and research methods in K–16 education contexts.
 
Member Center
In Print
This Month's Issue

Submit
EMAIL

Twitter

RSS