Annotated Bibliography Project

212

eadlines in the Atlanta Journal-Constitution (“Kindergarten,” 2003) read, “Kindergarten: Is older wiser?” “Some say ‘redshirting’ improves

readiness.” The article featured success stories for children with summer birthdays who delayed entrance to kinder- garten. Parents of preschoolers, after reading newspaper articles or talking with other parents, or both, are keenly aware that some parents purposefully delay their child’s entrance to kindergarten. A number of parents begin the process even earlier by having their child repeat a 3-year- old preschool program. That phenomenon, especially com- mon in affluent communities and among boys, causes par- ents with a 5-year-old summer-birthday child to wonder whether they should delay enrollment even when their child seems ready for kindergarten.

According to the newspaper article mentioned in the preceding paragraph, “The National Center of Education Statistics reports that 6% to 9% of kindergarten-aged chil- dren in the United States start a year late.” However, those figures may be even higher. Gifted Child Today (“Delaying Entrance Into Kindergarten,” 2004) reported that “Accord- ing to the Census Bureau, 22% of first graders were 7 or older in Oct. 2002, up from 13% in 1970.” “Redshirting” is more common among White parents than it is among par- ents from other races (Diamond, Reagan, & Bandyk, 2000).

Abundant early childhood research concerns kinder- garten readiness (e.g., Andrews & Slate, 2002; Kurdek & Sinclair, 2001), as well as contradicting reports on the ben- efits of redshirting. Although most researchers seem to agree with the short-term academic and behavioral benefits of redshirting, there is a wide range of conflicting reports on its long-term benefits. Rusch (1998), for example, argued that being an older student in a class can backfire, especially in the upper grades. Conversely, Deutsch (2003) suggested that the youngest children in a classroom suffer more psychiatric disorders than do other students. The benefit (or harm) of redshirting is just a small part of the problem for an individual. At the classroom level, redshirt- ing widens the age range for students and makes teaching practices more difficult. The presence of substantially older children (i.e., more than a year older) can affect other chil- dren academically and socially. Therefore, the increasingly popular trend of delaying kindergarten entry must be eval- uated carefully.

To investigate redshirting, researchers should first deter- mine whether there is an academic performance difference between children with summer birthdays and children with fall birthdays. We expected a large difference at the kinder- garten level because educators and parents agree that 1 year can make a substantial difference in learning at a young age. Several studies confirm the effect of age on academic achievement at the kindergarten level (e.g., Kurdek & Sin- clair, 2001; Meisels, 1992). It is not clear, however, when or if the difference in academic performance diminishes as students grow older. Gifted Child Today (“Delaying Entrance Into Kindergarten,” 2004) reported that the aca- demic advantages of redshirting diminish by the third grade, whereas Crosser (1991) stated that the difference is still distinct for fifth or sixth graders, especially for boys. Kurdek and Sinclair (2001) found no effects of age for either reading or mathematics achievement at fourth grade.

Address correspondence to T. C. Oshima, Department of Educa- tional Policy Studies, College of Education, P. O. Box 3977, Atlanta, GA 30302-3977. (E-mail to [email protected])

Copyright © 2006 Heldref Publications

Academic Performance Gap Between Summer-Birthday and Fall-Birthday

Children in Grades K–8

T. C. OSHIMA CHRISTOPHER S. DOMALESKI Georgia State University

ABSTRACT Much interest exists among parents and researchers regarding the benefits and drawbacks of delaying kindergarten entrance to acquire academic advantage (“red- shirting”). How evident is this assumed advantage at the kindergarten level and beyond? The authors evaluated large- scale test data from Grades K–8 to investigate the difference in performance between younger children (summer birthday) and older children (fall birthday). The performance gap evi- dent in kindergarten decreased rapidly in Grades 1–3 but persisted up to Grade 5, until leveling off at middle school. The performance gap in the early grades that resulted from birth date was much larger than was the gap caused by gen- der difference.

Key words: academic performance gap, fall- and summer- birthday students, Grades K–8

H

March/April 2006 [Vol. 99(No. 4)] 213

Our purposes in this study were as follows:

1. To first delineate the difference at the kindergarten level; and

2. To subsequently investigate the difference through the elementary and middle school grades by using a statewide criterion reference test.

Method

Data

We used data for kindergarten students from a large dataset, “The Early Childhood Longitudinal Study of the Kindergarten Class of 1998–1999” (National Center for Education Statistics, 2001). From 21,260 kindergarten stu- dents, we identified 3,862 students as “younger children with summer birthdays (June, July, and August)” and 2,693 students as “older children with fall birthdays (September, October, and November).” To qualify for the younger sum- mer-birthday group, the child had to be a first-time kinder- garten student who was born in June, July, or August and be less than 67 months of age at the time of fall testing. To be in the older fall-birthday group, the child had to be a first-time kindergarten student who was born in September, October, or November and be 67 months of age or older at the time of fall testing. Test data came from two test admin- istration periods (fall and spring). We collected data for elementary and middle grades from a statewide criterion- referenced test administered to approximately 115,000 stu- dents per grade in spring 2002. In this southeastern state, children had to have been 5 years old as of September 1 to enter kindergarten. Therefore, children born in the sum- mer were the youngest children in the kindergarten class unless they were held back for a year. For each grade, about 10,000 children qualified as “younger children with sum- mer birthdays”; another 10,000 qualified as “older children with fall birthdays.” From each pool of approximately 10,000 students, we randomly selected 3,000 students for the present study. The state reported test scores in Grades 1–8 in scaled scores; a score of 300–349 indicated “meeting expectations,” and a score of 350 and over indicated “exceeding expectations.”

Analysis

We examined students’ reading and mathematics perfor- mance in this cross-sectional study. For kindergarten stu- dents, we added affective variables (approaches to learning, self-control, and social interaction), as well as physical variables (height and weight). The study also includes demographic variables—race and gender. Race included eight levels for kindergarten students: (a) White, (b) African American, (c) Hispanic—Race Specified, (d) His- panic—Race Not Specified, (e) Asian, (f) Native Hawai- ian, (g) Other—Pacific Islander, (h) American Indian or Alaska Native, and (i) Multiracial, and six levels for Grade 1–8 students: (a) Asian, (b) African American, (c) His-

panic, (d) Native American/Alaskan, (e) White, and (f) Multiracial.

In the first part of the analysis, we investigated the mean difference between the two groups of interest for the age variable (summer- vs. fall-birthday children). To make the comparison meaningful across all kinds of measures and tests, we used effect size (ES) as an indicator for the differ- ence throughout this study, as well as the independent t test. We calculated ES as the mean difference of two groups over pooled standard deviation. We plotted ES across grades to observe the general trend over time. We repeated the analysis within each gender group.

In the second part of the analysis, we used multiple regression to compare the relationships between three independent variables (race, gender, and age) and the dependent variables (reading or mathematics) to identify how age affected the test scores in relation to other demo- graphic variables. Race, gender, and age were categorical variables that we transformed into k dummy variables where k was the number of categories, minus 1. For race variables for kindergarten students, we created seven dummy variables. For Grades 1–8, we transformed the race variable into five dummy variables (see Pedhazur, 1997 for procedures regarding dummy coding).

Results

Figure 1 shows ES for the difference between two groups (older fall-birthday children and younger summer-birthday children) for kindergarten students. Positive ES indicates that the mean for the older children was higher than that of the younger children. Stevens (1999) posited that ES = .20 is small, ES = .50 is medium, and ES > .80 is large. Fur- thermore, Stevens explained that medium ES is apparent

FIGURE 1. Effect sizes for younger versus older kindergarten students by academic, affective, and physical variables.

0.60

0.50

0.30 0.40

0.20

0.10

0.00 –0.10

Re ad

in g

M ath

em ati

cs

Ge ne

ral K

no wl

ed ge

Ap pr

oa ch

es to

L ea

rn in

g

Se lf-

Co nt

ro l

So cia

l I nt

era cti

on

H eig

ht

W eig

ht

Variable

Fall Spring

E ff

ec t

Si ze

214 The Journal of Educational Research

to a researcher. Stevens provided the IQ difference between semiskilled workers and professionals or managers as an example of medium ES (.50). On the basis of Stevens’s criteria, the differences of those two groups are considered to be average for academic variables, as well as for physical variables.

Our data show that older kindergarten students are 1.95 in taller than are their younger peers in the fall (ES = .50). In other words, older kindergarten students average one half of a standard deviation taller than younger kinder- garten students. That physical difference is probably appar- ent to parents and educators. For academic variables, ES also is about .50 (.38 for reading, .55 for mathematics, and .50 for general knowledge). In other words, differences existed between the two groups in academic areas as much as in the physical areas. The differences observed for affec- tive variables (approaches to learning, self-control, and

social interaction) were small (see Figure 1). Also shown in Figure 1 is the slight decrease in ES from fall to spring, except ES for height, possibly indicating a slight narrowing of the gap during the kindergarten year.

Table 1 reports the means and standard deviations for reading and mathematics scores for the two groups across kindergarten through Grade 8. For kindergarten, the scores are expressed in a T score. The T score is a standardized score with a mean of 50 and a standard deviation of 10. Table 1 also lists p values for the test of equal means (inde- pendent t tests) and equal variances (Levine’s test). Because of a large number of significance tests combined with large sample sizes, we used a conservative level of alpha (.001) as an indicator of significance. The differences of means were significant from kindergarten to Grade 5 for reading and mathematics. The differences were not signifi- cant for Grades 6–8. Also noteworthy was a consistent

TABLE 1. Means and Standard Deviations for Reading and Mathematics Scores

Reading Mathematics Grade Younger Older p Younger Older p

Kindergartena Fall

M 44.82 50.75 <.001 47.09 53.48 < .001 SD 16.37 14.61 <.001 11.77 11.52 .901

Spring M 46.77 51.62 <.001 48.13 53.85 < .001 SD 14.57 13.05 <.001 10.67 10.32 .430

1b M 338.31 349.67 <.001 327.85 339.46 < .001 SD 36.30 38.37 .072 29.68 31.98 < .001

2 M 339.80 348.38 <.001 328.89 335.51 < .001 SD 41.28 43.24 .141 28.87 31.84 < .001

3 M 338.26 343.61 <.001 328.81 332.47 < .001 SD 34.93 35.75 .293 29.04 29.56 .241

4 M 343.09 348.37 <.001 317.45 320.64 < .001 SD 43.91 46.84 .002 30.60 31.97 .041

5 M 334.24 338.06 <.001 324.50 327.94 < .001 SD 33.07 33.62 .239 28.82 29.52 .210

6 M 343.24 344.68 .205 324.02 324.83 .361 SD 42.78 45.01 .013 33.74 35.01 .018

7 M 339.96 341.26 .145 321.38 322.61 .072 SD 33.63 35.17 .034 25.73 27.36 .001

8 M 350.20 351.36 .339 320.49 320.22 .759 SD 46.57 47.51 .402 32.75 33.76 .132

Note. Younger = children with summer (June, July, August) birthdays; Older = children with fall (Sep- tember, October, November) birthdays. a For kindergarten, n (younger students) = 3,862 and n (older students) = 2,693 in the fall; n (younger students) = 3,718 and n (older students) = 2,586 in the spring. b For Grades 1–8, n (younger students) = 3,000 and n (older students) = 3,000.

March/April 2006 [Vol. 99(No. 4)] 215

trend of standard deviations. For Grades 1–8, although sel- dom reaching significance, the younger group’s variation was less than was that of the older group’s variation. That trend reversed for kindergarten students.

To track mean differences over time, we plotted ES against Grades 1–8 (see Figure 2; ES likely declined rapid- ly during Grades 1–3). However, a small ES (.10) remained during Grades 3–5. ES reached near zero for Grades 6–8. Figure 2 also illustrates graphs for girls and boys. For girls, a fairly large ES exhibited for mathematics in Grade 1 quick- ly declined by Grade 2. For boys in Grades 2–8, reading tended to exhibit larger ES values than did mathematics, indicating that being a younger aged student may have more effect on reading than on mathematics.

Table 2 shows the results from regression analyses. The R2 values are reported under race, gender, and age. For example, the R2 value of .182 under race for kindergarten (fall) for reading indicates that 18.2% of variation in read-

ing can be explained by determining race alone. It is simi- lar that 0.03% and 3.3%, respectively, of variation in read- ing can be explained by gender and age alone. The R2change refers to the difference in R2 between the full model (race, gender, and age as independent variables) and the reduced model (race and gender as independent variables), indicat- ing the unique contribution of age in addition to race and gender. In other words, a significant R2change indicates that adding age in the model would significantly improve the prediction of the test score when the model already con- tains race and gender.

One can make several observations from Table 2: First, age (whether a child was younger or older in the class) was a significant predictor for reading and mathematics through Grade 5. Significant R2change for Grades K–5 also indicated that age improved the prediction after control- ling for race and gender. Second, at any given grade, race always had the highest R2 with the test scores. The effect of race was fairly stable across Grades K–8. Third, gender was a significant predictor for reading but not for mathe- matics. Fourth, the order of strength for reading in terms of relationship with test scores was race, age, then gender through second grade. For Grades 3–5, the order was race, gender, then age; for Grades 6–8, race then gender. In other words, age was a better predictor of reading than was gender through Grade 2; gender became a better predictor than age for Grades 3–5. Fifth, the order of strength for mathematics was race, then age through Grade 5; race was the only significant predictor for Grades 6–8. For example, at the beginning of the kindergarten year, 10.6% of varia- tion in mathematics scores could be explained by race, and 6.8% could be explained by age. Gender explained little (< 0.1%) variation.

Discussion

For kindergarten students, the difference observed for academic areas (reading, mathematics, and general knowl- edge) was as large as was the difference for the height of children from the two groups. By spring, the academic dif- ference narrowed somewhat. For elementary school chil- dren, ES of the difference of the two groups across five grades declined. There was a rapid decrease up to the third grade and a gradual or no decrease between the third and fifth grades; the difference still existed at fifth grade. Con- versely, the difference was near zero in the middle school years. Although redshirting is more common for boys (Bent, May, & Kundert, 1996), the gap between the two groups was not always more pronounced for boys.

How large was the impact of age, whether the child was younger or older in the class, on test scores when compared with other demographic variables, such as race and gender? Whereas race explained 10–15% of variation in test scores in general, age explained up to about 7% of variation, depending on the grade level. That percentage was much larger than the percentage that we obtained from gender

FIGURE 2. Effect sizes for younger versus older children over time (Grades 1–8) for combined sample, female sam- ple, and male sample.

0.5

0.3

0.4

0.2

0.1

0

–0.1

Grade

Reading Mathematics

1 2 3 4 5 6 7 8

A

E ff

ec t

Si ze

0.5

0.3

0.4

0.2

0.1

0

–0.1

Grade 1 2 3 4 5 6 7 8

B

E ff

ec t

Si ze

0.5

0.3

0.4

0.2

0.1

0 –0.1

Grade 1 2 3 4 5 6 7 8

C

E ff

ec t

Si ze

216 The Journal of Educational Research

(1% for reading and <.01% for mathematics). In other words, age difference was as important, or more important, than was gender difference in the early grades.

A plethora of research exists on gender differences. For example, many researchers have shown that reading scores for boys are far behind those of girls (e.g., Gambell & Hunter, 1999; Phillips, Norris, Osmond, & Maynard, 2002; Sommers, 2000), a phenomenon that we confirmed. All through Grades K–8, girls outperformed boys. Researchers also suggest that boys outperform girls in mathematics, per- haps more so during adolescence (e.g., Hedges & Nowell, 1995; Royer & Wing, 2002). In this study, however, we did not observe gender difference for mathematics. Researchers should acknowledge gender difference in academic perfor- mance, as well as identify any difference that may exist by birth date. We showed that differences of birth date can be a much larger factor regarding academic performance than gender differences in the early grades and that the differ- ences continue throughout the elementary school years.

We provide data on the issues related to the controver- sial practice of academic redshirting. It was not our inten- tion to argue for or against the practice. We did not exam- ine the population that delayed kindergarten entrance. Redshirting is based partly on the assumption that older children have an advantage over younger children in a classroom. We delineated how much difference there actu- ally is in a given grade.

Numerous issues need to be considered before a decision is made regarding whether a parent should hold a summer- birthday child back for a year: (a) Does the child tend to learn better with older peers? (b) Does the child enjoy interacting with older children? (c) Is she or he easily hurt by not being able to perform as well academically (and physically) as her or his peers? and (d) Is the child a boy, and does he have a reading problem? Although any deci-

sion concerning academic redshirting at the kindergarten level (or at a younger age) should be individually based, research in these areas (such as type of children who thrive in an environment in which they are the youngest) should help parents make an intelligent decision concerning red- shirting, which is a decision that cannot be reversed once it is made.

We showed the magnitude and duration of the academic gap exhibited by age differences of 7 to 11 months (summer- birthday children vs. fall-birthday children). With those objective data, researchers can investigate the practice of redshirting, which has a profound impact on individuals, as well as on teachers and schools.

REFERENCES

Andrews, S. P., & Slate, J. R. (2002). Public and private prekindergarten programs: A comparison of student readiness. Educational Research Quarterly, 25, 59–73.

Bent, D., May, D. C., & Kundert, D. K. (1996). The incidence of delayed school entry: A twelve-year review. Early Education and Development, 7, 121–135.

Crosser, S. L. (1991). Summer birth date children: Kindergarten entrance age and academic achievement. The Journal of Educational Research, 84, 140–146.

Delaying Entrance Into Kindergarten. (2004, Summer). Gifted Child Today, 27, 6.

Deutsch, N. (2003). Youngest kids in class suffer more psychiatric disor- ders. Medical Post, 39, 47.

Diamond, K. E., Reagan, A. J., & Bandyk, J. E. (2000). Parents’ concep- tion of kindergarten readiness: Relationships with race, ethnicity, and development. The Journal of Educational Research, 94, 93–100.

Gambell, T. J., & Hunter, D. M. (1999). Rethinking gender differences in literacy. Canadian Journal of Education, 24, 1–16.

Hedges, L., & Nowell, A. (1995). Sex differences in mental test scores, variability and numbers of high-scoring individuals. Science, 269, 41–45.

Kindergarten: Is older wiser? (2003, September 20). The Atlanta Journal- Constitution, p. G1.

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

TABLE 2. R2 Values for Race, Gender, and Age, With Reading or Mathematics as the Dependent Variable

Reading Mathematics Grade Race Gender Agea R 2 Race Gender Agea R 2

Kindergarten Fall .182* .003* .033* .022* .106* < .001 .068* .053* Spring .128* .005* .028* .020* .095* < .001 .067* .053*

1 .069* .016* .022* .020* .084* < .001 .034* .031* 2 .084* .006* .010* .011* .118* .002 .012* .011* 3 .104* .014* .006* .006* .131* .001 .004* .004* 4 .120* .008* .003* .003* .118* < .001 .003* .002* 5 .102* .004* .003* .003* .159* .001 .003* .002* 6 .128* .007* < .001 < .001 .112* < .001 < .001 < .001 7 .140* .009* < .001 < .001 .159* < .001 < .001 < .001 8 .106* .007* < .001 < .001 .128* < .001 < .001 < .001

a Refers to either younger or older students. b R 2change refers to the difference in R 2 between the full model (race, gender, and age) and the reduced

model (race and gender), indicating the unique contribution of age above and beyond race and gender. *p = .001.

March/April 2006 [Vol. 99(No. 4)] 217

Meisels, S. (1992). Doing harm by doing good: Iatrogenic effects of early childhood enrollment and promotion policies. Early Childhood Research Quarterly, 7, 155–174.

National Center for Education Statistics. (2002). ECLS-K base year pub- lic-use data files and eletronic codebook. Washington, DC: Author.

Pedhazur, E. J. (1997). Multiple regression in behavioral research: Explanation and prediction (3rd ed.). New York: Holt, Rinehart and Winston.

Phillips, L. M., Norris, S. P., Osmond, W. C., & Maynard, A. M. (2002). Relative reading achievement: A longitudinal study of 187 children from

first through sixth grades. Journal of Educational Psychology, 94, 3–13. Royer, J. M., & Wing, R. E. (2002). Making sense of sex differences in

reading and math assessment: The practice and engagement hypothe- sis. Issues in Education, 8, 77–85.

Rusch, L. (1998). Delaying kindergarten. Parents, 73, 129. Sommers, C. H. (2000). The war against boys: How misguided feminism is

harming our young men. New York: Simon & Schuster. Stevens, J. (1999). Intermediate statistics: A modern approach (2nd ed.).

Hillsdale, NJ: Erlbaum.