Kindergarten practices and first-grade achievement for Latino Spanish-speaking, Latino...

19
Pergamon Journal of School Psychology, Vol. 33, No. 2, pp. 123-141, 1995 Copyright © 1995 Society for the Study of School Psychology Primed in the USA 0022-4405/95 $9.50 + .00 0022-4405(95)00002-X Kindergarten Practices and First-Grade Achievement for Latino Spanish-Speaking,Latino English-Speaking, and Anglo Students Merith Cosden, JulesZimmer, Carla Reyes, and Maria del Rosario Gutierrez University of California, Santa Barbara A disproportionate number of Latino students, both immigrant and native-born, are having serious problems in the school system. Little is known about the early schooling experiences of these children, or the impact of these experiences on later school out- comes. This concern is raised in the context of escalating academic demands within kindergarten programs. This escalation has had the result that some parents hold children out of school until they are older and more ready to succeed, while schools have increasingly relied on retention in kindergarten to prevent children who are not ready from going on to first grade. This study assessed the impact of gender, ethnicity/ home language, holding out, retention, and social promotion on first-grade achieve- ment. Among the significant findings: Latino students were more likely to be "ad- vanced" (socially promoted to first grade despite concerns about their performance) than were Anglo students, and none of the interventions employed by parents (holding out) or schools (retention and social promotion) had their intended positive effects. In order for early identification of students with academic needs to be a useful endeavor, interventions designed to promote the academic growth of these students need to follow. Keywords: Kindergarten; Latino; Retention; Social promotion. Over the past 10 years there has been a significant increase in the number of Latino children, both native-born and immigrant, entering the school system (McDonnell & Hill, 1993; U.S. General Accounting Office, 1993). Many schools have problems retaining Latino students through high school, and a disproportionate number of Latino students who remain in school experience academic problems (Barona & Garcia, 1990). Although these school problems often appear to have an early genesis, there has been little systematic research on the early schooling experience of Latino students, or the impact of these early experiences on later school outcomes (McDonnell & Hill, 1993). There have also been significant changes in early education during this time. In particular, there has been an escalation of the academic demands of Received May 3, 1994; final revision received September 16, 1994. Address correspondence and reprint requests to Merith Cosden, PhD, Counseling/Clinical/ School Psychology, Department of Education, University of California, Santa Barbara, Santa Barbara, CA 93106. 123

Transcript of Kindergarten practices and first-grade achievement for Latino Spanish-speaking, Latino...

Pergamon

Journal of School Psychology, Vol. 33, No. 2, pp. 123-141, 1995 Copyright © 1995 Society for the Study of School Psychology

Primed in the USA 0022-4405/95 $9.50 + .00

0022-4405(95)00002-X

Kindergarten Practices and First-Grade Achievement for Latino Spanish-Speaking, Latino

English-Speaking, and Anglo Students

Merith Cosden, Jules Zimmer, Carla Reyes, and Maria del Rosario Gutierrez

University of California, Santa Barbara

A disproportionate number of Latino students, both immigrant and native-born, are having serious problems in the school system. Little is known about the early schooling experiences of these children, or the impact of these experiences on later school out- comes. This concern is raised in the context of escalating academic demands within kindergarten programs. This escalation has had the result that some parents hold children out of school until they are older and more ready to succeed, while schools have increasingly relied on retention in kindergarten to prevent children who are not ready from going on to first grade. This study assessed the impact of gender, ethnicity/ home language, holding out, retention, and social promotion on first-grade achieve- ment. Among the significant findings: Latino students were more likely to be "ad- vanced" (socially promoted to first grade despite concerns about their performance) than were Anglo students, and none of the interventions employed by parents (holding out) or schools (retention and social promotion) had their intended positive effects. In order for early identification of students with academic needs to be a useful endeavor, interventions designed to promote the academic growth of these students need to follow.

Keywords: Kindergarten; Latino; Retention; Social promotion.

Over the past 10 years there has been a significant increase in the number of Latino children, both native-born and immigrant, entering the school system (McDonnell & Hill, 1993; U.S. General Accounting Office, 1993). Many schools have problems retaining Latino students through high school, and a disproportionate number of Latino students who remain in school experience academic problems (Barona & Garcia, 1990). Although these school problems often appear to have an early genesis, there has been little systematic research on the early schooling experience of Latino students, or the impact of these early experiences on later school outcomes (McDonnell & Hill, 1993).

There have also been significant changes in early education during this time. In particular, there has been an escalation of the academic demands of

Received May 3, 1994; final revision received September 16, 1994. Address correspondence and reprint requests to Merith Cosden, PhD, Counseling/Clinical/

School Psychology, Department of Education, University of California, Santa Barbara, Santa Barbara, CA 93106.

123

124 Iournal of School Psychology

many kindergarten programs, which have been associated with a mandate to prepare children for first-grade work (Freeman & Hatch, 1989; Shepard & Smith, 1988). This change appears to be relatively pervasive. In a recent survey conducted by the National Center for Educational Statistics, for exam- ple, 70 % of the teachers surveyed stated that they would hesitate to promote a child if they felt the child was not ready for the demands of first grade. In addition, only 27 % of the teachers also felt that by the end of kindergarten all of their children would be ready for first-grade work (National Center for Educational Statistics, 1993).

In response to these curricular changes, some parents have decided to hold their child out of school an extra year to increase their "success" in kindergar- ten. This is most likely to occur if the child is a boy and would be among the youngest in his cohort if he entered school on schedule (Mergendoller, Bellis- imo, & Horan, 1990; Shepard & Smith, 1988). Parents argue that by holding their child out of school they are giving that child extra time to develop the skills needed for kindergarten. This parental decision is often supported by teachers' attitudes. In the survey by the National Center for Educational Statistics (1993), for example, over half of responding kindergarten teachers agreed that if a child appears unready for kindergarten, he or she should wait a year before enrolling.

Many districts have developed their own prekindergarten screening pro- grams (Cannella & Reiff, 1989). In a study of districts engaged in prekinder- garten screening practices, Walsh, Ellwein, Eads, and Miller (1991) found that sex, socioeconomic status (SES), and age were significant predictors of whether students would be excluded from kindergarten. Young, male students who were participating in the free-lunch program were the category of stu- dents most likely to be asked to wait a year before entering school, whereas older, female students who did not participate in the free-lunch program were the least likely to be excluded.

Although holding children out of school is designed to increase their readi- ness for school, this practice may contribute to the acceleration of academic demands in kindergarten by priming classes with children who have higher levels of academic skill (Shepard & Smith, 1986). Little is known about the impact of holding out, either on the child who is held out or on the child's classmates (Mergandollar et al., 1990).

Furthermore, in conjunction with this escalation in academic demands, kindergarten teachers have been called upon to identify children with aca- demic problems before their entry into first grade. Once children with aca- demic problems are identified, schools must decide what to do for them. Typically schools have vacillated between the use of in-grade retention as a method for allowing children to catch up to their peers, or promotion for social reasons despite concerns about the child's ability to perform well aca- demically at the next grade level. The negative impact of social promotion, "advancing" children to the next grade level whether or not they have mastered

Cosden et al. 125

the skills taught at their current grade level, has been noted by employers, educators, and parents and by the children themselves, who find themselves inadequately prepared for future educational or vocational opportunities (Carstens, 1985; Shepard & Smith, 1989). Thus, it has become common for schools to utilize a policy of in-grade retention, asking students to repeat kindergarten if they do not have the skills viewed as necessary for success in first grade (Byrnes & Yamamoto, 1986; Smith & Shepard, 1988).

However, reviews of retention research find that students gain little aca- demically as a function of being retained, even when they are compared with children with similar initial skills who were not retained (Shepard & Smith, 1987 ). The only instances in which retention has resulted in positive academic outcomes are those in which special remediation, along with class repetition, has been offered during the student's second year (Holmes, 1989). Even when combined with additional intervention, however, retention may have negative social and emotional consequences if students perceive retention as a personal failure (Holmes & Matthews, 1984; Niklason, 1987; Smith & Shepard, 1988). For example, Byrnes (1989) followed teacher and parent surveys on retention with interviews of children who had been retained, or who had classmates who had been retained. Many of the children in her study reported negative effects from retention, despite their teachers' reports that the retention was successful. In addition, studies indicate that students who are retained are at greater risk for dropping out of school, even when compared with children who have similar prior achievement levels but who have not been retained (Grissom & Shepard, 1989).

Nevertheless, many educators continue to believe that retention is effective, particularly when used in the lower grades. It has been suggested that this belief persists because opportunities for continued observation of the students are rarely readily available to teachers. That is, while teachers may correctly perceive that retained students have higher levels of achievement in their second year in kindergarten than they did in their first, this enhanced perfor- mance rarely continues into the next grade level, and this is less visible to the retaining teacher (Smith, 1989; Tomchin & Impara, 1992). Shepard (1989), in a review of kindergarten retention research, found negative effects associ- ated with retention in kindergarten; students retained in kindergarten not only failed to catch up with their peers academically but often experienced negative social stigma in association with their retention. Likewise, Mantzico- poulos and Morrison (1992) found that children who had been retained in kindergarten, although they had evidenced some academic gains during their second year in kindergarten, had lost these benefits by the end of first grade.

It appears that current kindergarten practices may not be meeting the needs of many students. In a recent study of three school districts, by the end of kindergarten 40 % of the children were "off track," that is, they had been held out of school an extra year, retained in kindergarten, or advanced to first grade with concerns about their skills (Cosden, Zimmer, & Tuss, 1993).

126 Journal of School Psychology

Some groups of students appear to be more vulnerable than others to these practices. Factors such as age, gender, and ethnicity appear to play a signifi- cant role in these intervention decisions. For example, it appears that younger boys are more likely to be retained than other students (Fowler & Cross, 1986; Langer, Kalk, & Searls, 1984). While some studies suggest that younger boys have the greatest need for in-grade retention, there are indications that this practice is more a function of parents' and teachers' perceptions that it is acceptable to retain younger students, particularly boys, than it is a function of the students' behavior (Shepard & Smith, 1986). Children who evidence similar behavior problems, but are older, are less likely to be retained.

A few studies have also documented that students from ethnic minority groups are more likely to be retained or identified with early school problems. Cosden et al. (1993) found that a significantly greater proportion of Latino than Anglo students were identified as having school problems in two of three districts studied. In one district Latino students were more likely than Anglo students to be retained. In the other district Latino students were more like- ly than Anglo students to be advanced (socially promoted); this social ad- vancement occurred when teachers felt that the students were not ready for first grade work but, for a variety of other reasons, did not want to retain them.

This study, however, did not assess differences among students as a func- tion of home language. This is an important factor, given the large number of Latino students now entering the school system whose primary or sole lan- guage is Spanish (McDonnell & Hill, 1993). Little is known about the early schooling experiences of these students. Bourdieu's (1991) conceptualization of literacy as cultural capital explains the problems that Spanish-speaking immigrants have in addressing their needs with community agencies. That is, limited speaking and reading skills in English prevent these families from accessing the cultural knowledge needed to successfully use community re- sources. Similarly, if parents lack access to information about the school cul- ture, they cannot anticipate school readiness issues in the same way as would families with greater access to knowledge about the school system.

Furthermore, the one common characteristic of most present-day immi- grant families is low income status (McDonnell and Hill, 1993, p. 49). In many instances immigrant children live in communities where the level of poverty is generally high and school achievement is generally low. Thus, there are many factors that may contribute to poor school performance for these students. A better understanding of these variables can assist schools in devel- oping more effective early intervention strategies.

The present study has several objectives. First, it was the purpose of this study to describe patterns of holding out, retention, and advancement in a school district serving both Latino and Anglo students. The degree to which students in this district were held out, and thereby started a year older than required by state law, and the extent to which students who entered kindergar-

Cosden et al. 127

ten on schedule were retained in kindergarten or socially advanced to first grade, were assessed.

A second, related objective was to determine the relationship of certain student characteristics to holding out, retention, and advancement. The litera- ture suggests that age, gender, and ethnicity may all be related to these early schooling decisions. From prior research, a relationship between holding-out practices, gender, and age was anticipated: It was predicted that a greater number of younger students, and males, would be held out an extra year before starting school. It was also predicted, from findings reported in the literature, that retention would be associated with gender, age, and ethnicity, and that younger children, boys, and Latinos would be more likely to be retained. Finally, it was hypothesized that advancement would be related to ethnicity, Latino students being more likely than Anglo students to be ad- vanced in lieu of promotion.

In addition, this study was designed to explore the relationship of both ethnicity and home language to holding out, retention, and advancement patterns. It was hypothesized that Latino, Spanish-speaking students would be retained or advanced more than other groups of students, including Latino students for whom English was a home language. This hypothesis was based on the linguistic problems faced by these students and their families in the community.

Finally, this study was designed to assess the relationship of holding out, retention, and advancement on first-grade achievement. The few controlled studies that have previously addressed these issues have not found positive effects for either holding out or retention. This study analyzed the relationship of ethnicity and home language to outcomes for students after they had been held out, retained, or advanced. Whether these practices had similar out- comes for different groups of students was assessed.

METHODS

Subjects

This study was conducted in a school district in southern central California. The district had an average daily attendance of 9,248. Approximately 26 % of the students were Anglo and 68 % were Latino. Data were collected on chil- dren entering kindergarten in 1990 across all 11 elementary schools in the district. Although a total of 987 students entered kindergarten, data were available on only 616 students at the end of first grade; this is largely a function of the high degree of mobility in this rural geographic area.

The district reported 17 % of its students to be in families receiving Aid for Dependent Children (AFDC). The socioeconomic status (SES) of the district as a whole was relatively low, as determined by the California Assessment Program (CAP), which assigns numbers to families to indicate level of parent

128 Journal of School Psychology

occupation. According to CAP designations, a value of 1 is assigned to un- skilled employees, 2 to skilled and semiskilled employees, and 3 to professional or semiprofessional employees. This school district received an average SES score of 1.6.

Of the 616 students on whom data were available, 373 took the Primary 1, Stanford Achievement Test (SAT), 8th Edition (Psychological Corporation, 1992) at the end of first grade, while 243 students took the Primary 1, Aprenda: La Prueba de Logros en Espanol (Psychological Corporation, 1989). Table 1 presents the demographic characteristics of the students taking each test.

Data Source

Demographic data on students (age, gender, birthdate) were obtained from computerized school records. School records were also used to identify chil- dren who had been held out of school (i.e., were overage when they entered school for the first time), and to indicate who had been promoted, retained in kindergarten, or advanced from kindergarten to first grade. Very few students in the district had been retained in first grade (fewer than 10 according to school records). In order to answer the questions posed by this study without confound by these students, they were not included in the database.

The school district had written policies for retention and social promotion, which they labeled advancement. Regular promotion to the next grade occurred

Table 1 Demographic Characteristics of First Graders Taking the Stanford Achievement Test

(SAT) or the Aprenda: La Prueba de Logros en Espafiol

SAT Aprenda

n % n %

Gender Males 187 50.13 125 51.44 Females 186 49.87 118 48.56

Ethnicity/home language Anglo 131 35.12 -- Latino/English 162 43.43 -- Latino/Spanish 80 21.45 243 100

Kindergarten status Normal age of entry 324 87.57 195 80.91 Retained 17 4.59 19 7.88 Held out 29 7.84 27 11.20

End of kindergarten status Normal promotion 335 90.54 197 81.74 Advanced to first grade 35 9.46 44 18.26

Cosden et al. 129

for students who scored at the fourth stanine or higher on district- administered standardized tests of reading, language, and math. Students were advanced rather than promoted when they failed to perform at the fourth stanine but appeared to be working at capacity and making reasonable progress. Retention was the final alternative for students who did not meet the criteria for either promotion or advancement. Students who attended school less than a specified number of days could also be retained. A Pupil Placement Committee convened to decide cases in which retention, advance- ment, or promotion was in question.

While these policies were designed for all grade levels, they could not be applied as easily to kindergarten students, since achievement tests were not administered until the end of first grade. No other formal criteria had been established for evaluating students' performance; thus more latitude was given to teachers, or the Pupil Placement Committee, in making kindergarten reten- tion and advancement decisions.

Test Scores

Students in the district took an achievement test at the end of first grade, either the SAT (Psychological Corporation, 1992) or the Aprenda (Psycho- logical Corporation, 1989). The linguistic skills of the student when he or she first entered school, as assessed by the Bilingual Syntax Measure (BSM) (Psychological Corporation, 1975) determined which test was taken. For stu- dents whose initial scores indicated Spanish as a primary language, the BSM was readministered each year. Thus, some students whose home language was listed as Spanish were administered the SAT on the basis of newer BSM scores. Both the SAT and the Aprenda were administered in small groups at the school sites in accordance with procedures defined by the test developer.

Although the SAT and the Aprenda were considered comparable tests by the district, a content analysis of the two indicated that there were differences in specific test items, the number of items within subtests, and the format of items within subtests (Reyes & Cosden, in preparation). While tables are available for converting Aprenda scores to scores that are appropriate for comparison to the SAT (Psychological Corporation, 1991), conversions for all subtests were extrapolated from scores on only a few subtests. Further- more, the student sample on which the conversions were based was not repre- sentative of the students taking the Aprenda in this school district. For these reasons, students' outcomes on the SAT and the Aprenda were analyzed separately.

RESULTS

The students who had been held out, retained, or socially advanced were identified. The relationship of age, ethnicity/home language, and gender to these practices was assessed by chi-square analysis techniques.

130 Journal of School Psychology

Holding-Out Patterns

Fifty-six students (9.15 % of the kindergarten sample) entered school a year later than they could have under state law. These students were identified as held-out. Chi-square analyses were conducted on the entire sample to assess differences in holding-out patterns as a function of age, gender, and ethnicity/ home language.

The impact of age on holding-out decisions was determined by coding birth months into quartiles that reflected the age the children would have been in relation to their peers had they entered school on schedule. Children in Quart- ile 1 were born between September and November, making them the youn- gest children in a cohort with December 1 as the cutoff for school entry. Children in Quartile 2 were born between June and August, children in Quartile 3 between March and May, and children in Quartile 4 between December and February. Coded in this manner, birth month was related to holding-out patterns, X2(3, N = 611) = 14.79, p < .05. Children in the youngest part of the cohort (Quartile 1 ) were more likely to be held out ( 16% ) than students in Quartile 2 (9 .58%), Quartile 3 (4 .72%), or Quartile 4 (5 .19%) .

No difference in holding-out patterns was found as a function of gender, X (1, N = 611) = 1.76, p > .05. Approximately 11% of males and 7.59% of females entered school a year late, but this difference was not statistically significant.

Furthermore, no difference in holding-out was found as a function of eth- nicity/home language, X 2 (2, N = 611) = .66, p > .05. Holding-out was evenly distributed across ethnic groups, with 9.02 % of Anglo students, 7.74 % of English-speaking Latino students, and 9.97% of Spanish-speaking Latino students entering school a year late.

Retention Patterns

A total of 36 students (5.88 % of the sample) had been retained. None of the factors under study were significantly associated with the likelihood of being retained. Age of entry was not significantly related to retention, X 2 (1, N = 617) = 2.65, p > .05. Retention rates were 7.27% for students in Quartile 1, 7.69% for students in Quartile 2, 5.43% for students in Quartile 3, and 2.60 % for students in Quartile 4. None of the retained children had been held out.

Gender did not have a significant impact on retention, X 2 ( i, N = 617) = 2.65, 'p > .05. Boys were more likely to be retained (7.35% of the male sample) than girls (4.28% of the female sample), but this difference was not statistically significant. Retention did not vary, either, as a factor of ethnicity/ home language, X 2 (2, N = 617) = 1.44, p > .05. In this cohort, 3.76% of Anglo students, 5.92% of English-speaking Latino students, and 7.14% of Spanish-speaking Latino students had been retained.

Cosden et al. 131

Social Advancement Patterns

Seventy-nine students, 12.76% of the kindergarten sample, were advanced rather than being promoted or retained in kindergarten. Advancement did not vary as a function of birth month, X 2 (3, N = 612) = 4.96, p > .05, although a greater proportion of younger students than older students were advanced ( 17.68 % of students in Quartile 1, 11.97 % of Quartile 2, 11.81% of Quartile 3, and 10.79% of Quartile 4).

Advancement did vary as a function of gender; a greater proportion of boys (15.86%) than girls (9.90%) were advanced, X 2 (1, N = 612) = 4.83, p < .05.

The likelihood of being advanced also varied as a function of ethnicity/ home language, X 2 (2, N = 612) = 8.21, p < .05. Only 6.77% of Anglo students had been advanced, compared with 11.31% of English-home- language Latino students and 16.39% of Spanish-home-language Latino stu- dents.

Holding Out and Achievement

Multivariate analysis of variance ( M A N O V A ) was used to assess the relation- ship of holding out, retention, and social advancement to achievement. Sepa- rate analyses were conducted on scores obtained from the SAT and from the Aprenda. Each student who had been held out, retained, or advanced was matched with a student within his or her school who had the same gender and ethnicity/home language, and a birth month in the same quartile. The stu- dents were matched by these criteria to assure that any differences found in achievement could be attributed to the interventions and not just to differ- ences in the demographic composition of the samples.

Matches were created in the following manner. For each school, a list of all first-grade students, and their demographic characteristics, was obtained. The students were then divided into two groups: those who had taken the SAT and those who had taken the Aprenda. In order to create a matched sample for students who had taken the SAT, the students who had been held out, re- tained, or advanced were identified, and their birth month, gender, and ethnicity/home language were noted. At each school a second pool of first- grade students was created, consisting of those students who had taken the SAT but who had not been held out, retained, or advanced. The students who had been held out, retained, or advanced were individually matched to stu- dents in the second pool. This was accomplished by identifying all students in the second pool who had the same configuration of birth month, gender, and ethnicity/home language as the target child's. In most instances three to five students fit these criteria. When that occurred, one student was randomly selected to be part of the matched sample. In the few instances in which only one child met the criteria, he or she was automatically used in the matched sample.

132 Journal of School Psychology

To create a matched sample for students taking the Aprenda a similar set of procedures was followed. Within each school, first-grade students who had been held out, retained, or advanced were identified, and a pool of remaining first-grade students was created. Gender and birth month were used as the matching criteria, as ethnicity and home language were the same for all students. For each target child, a list of other first-grade children at her or his school with the same birth month and gender was created. In instances in which more than one child met the matching criteria, one was randomly selected. If, within a school, only one child had the same birth month and gender configuration as that of a target child, that child automatically became part of the matched sample.

The SAT had been administered to 29 students who had been held out. These students were matched by gender, ethnicity/home language, and birth month with a similar number of students who had not been held out, following the procedures described above. The students in the two groups had similar birth months but were a year apart in age.

Holding out did not have a significant impact on achievement scores on the SAT, multivariate F(5, 37) = 1.58, p > .05. The interaction between holding out and ethnicity/home language was also nonsignificant, multivari- ate F(10, 74) = 1.09, p > .05. The children who had been held out did not score significantly higher on the SAT than the children who had not been held out, even though the held-out children were a full year older than the students in the second group.

Twenty-seven children who had been held out had been assessed by the Aprenda; they were matched, by gender and birth month, with 27 children who had not been held out, by the matching procedures described above. Again, no significant differences in achievement were found between groups, F(4, 43) = 1.63, p > .05, even though the students who had been held out were a year older than their matched peers. Holding out was not significantly related to first-grade achievement.

Retention and Achievement

Seventeen students who had taken the SAT had been retained in kindergar- ten. To assess the impact of retention on achievement, a contrast group was selected that consisted of a subsample of nonretained, normally promoted students who were matched by birth month, gender, and ethnicity/home lan- guage to the retained students. The procedures for selecting these students are described above. By virtue of the matching of the students by birth month, the retained students were a full year older than the promoted students with whom they were being compared, although both groups were entering first grade.

Significant differences were obtained between retained and nonretained students, multivariate F(5, 24) = 3.38, p < .05. The students who had

Cosden et al. 133

been retained in kindergarten scored lower than students who had not been retained, even though the retained children were a year older than their peers. Further analysis indicated that these differences were obtained across all subtests (see Table 2 ).

The interaction between retention and ethnicity/home language was non- significant, multivariate F(10, 40) = 1.48, p > .05. That is, the retained children performed more poorly than their nonretained peers regardless of their ethnicity/home language.

Nineteen retained children had been assessed on the Aprenda. They were compared with a contrast group of 19 students who had entered school on schedule and who were matched by birth month and gender to the retained students in the manner described above. No significant differences between retained and nonretained groups were obtained, multivariate F(4, 31) = 1.66, p > .05. Thus, for students who had taken the Aprenda, retention was not a significant factor in predicting achievement.

Advancement and Achievement

Thirty-five students who had taken the SAT had been advanced. A sample of 35 promoted students, matched by gender, age, and ethnicity/home language, was selected for comparison. As with the other MANOVAs, all promoted students who matched the characteristics of the advanced sample were pooled, and 35 students were randomly selected to compose the contrast group. Thus, the students in the two groups were the same age and gender, and had similar ethnic characteristics.

Advancement had a significant impact on achievement scores, multivariate F(5, 54) = 2.59, p < .05. Univariate analyses found differences across all areas of achievement, the advanced students scoring significantly lower than the students who had been promoted (see Table 3). The interaction of ethnic- ity/home language and advancement was nonsignificant, F( 10, 108) = .891, p > .05. That is, advancement had similar effects on students who had taken the SAT regardless of ethnicity/home language.

Forty-four students assessed by the Aprenda also had been advanced in school. These students were compared with a sample of 44 students who had been promoted and who were matched to the advanced children by age and gender. Advancement did not have a significant impact on achievement for students who had taken the Aprenda, multivariate F(4, 7 6 ) = 1.41, p > .05.

Retention Versus Advancement and Achievement

MANOVAs were also conducted to compare the achievement levels of the students who had been retained with those who had been advanced. The results of the M A N O V A comparing the advanced and the retained students

Tab

le 2

A

chie

vem

ent

Scor

es f

or R

etai

ned

and

Pro

mot

ed S

tude

nts

SA

T

Pro

mo

ted

(n

=

17)

Ret

ain

ed (

n =

17

)

Ap

ren

da

Pro

mo

ted

(n

=

19)

Ret

aine

d (n

=

19)

M

SD

M

S

D

F

M

SD

M

S

D

F

Tot

al R

ead

ing

42

.78

14.3

1 27

.72

11.8

7 9.

83*

30.3

4 18

.98

26.7

1 20

.24

0.53

T

otal

Mat

h

54.9

5 19

.12

36.4

7 17

.96

7.45

* 38

.21

14.5

4 39

.07

19.1

9 0.

02

Tot

al L

ang

uag

e 52

.77

21.5

6 24

.63

16.0

8 16

.42"

44

.36

14.5

0 44

.87

22.3

1 0.

06

Bas

ic B

atte

ry

44.0

8 15

.51

26.8

6 13

.76

10.3

4"

34.1

1 14

.53

31.5

9 20

.67

0.47

T

otal

Bat

tery

44

.44

14.8

2 27

.51

13.9

5 10

.37"

Not

e. T

he

mul

tiva

riat

e an

alys

is f

or th

e S

AT

was

sig

nifi

cant

, m

ulti

vari

ate

F(5

, 24

) =

3.

38,

p <

.0

5. T

he

mul

tiva

riat

e an

alys

is f

or th

e A

pre

nd

a w

aS n

ot

sign

ific

ant,

mul

tiva

riat

e F

(4,

31 )

=

1.66

, p

>

.05.

All

ret

aine

d st

uden

ts w

ere

incl

uded

in

the

anal

yses

. A

sam

ple

of p

rom

ote

d s

tude

nts

was

sel

ecte

d on

the

bas

is o

f th

eir

mat

ch t

o th

e re

tain

ed s

tude

nts

by

tes

t ta

ken,

bir

th m

on

th,

gend

er,

and

eth

nic

ity

/ho

me

lang

uage

. R

etai

ned

stud

ents

wer

e a

year

ol

der

than

pro

mo

ted

stu

dent

s.

*p <

.0

5.

Tab

le 3

A

chie

vem

ent

Scor

es f

or A

dva

nce

d a

nd

Pro

mot

ed S

tud

ents

SA

T

Pro

mo

ted

(n

=

35

) A

dvan

ced

(n =

35

)

Ap

ren

da

Pro

mo

ted

(n

=

44)

Adv

ance

d (n

=

44)

M

SD

M

S

D

F

M

SD

M

S

D

F

¢J1

Tot

al R

eadi

ng

34.5

0 20

.65

22.5

2 10

.06

9.59

* 29

.62

16.7

5 21

.33

11.4

1 5.

61"

Tot

al M

ath

43.3

5 19

.43

27.9

6 17

.39

11.2

0"

37.7

5 17

.18

33.7

1 13

.20

1.61

T

otal

Lan

guag

e 33

.39

21.3

0 26

.15

18.2

0 5.

29*

44.9

3 15

.52

40.4

2 14

.89

1.0

Bas

ic B

atte

ry

33.1

7 19

.39

20.3

7 11

.80

12.7

8"

33.8

0 16

.66

27.4

2 12

.33

3.29

T

otal

Bat

tery

33

.87

18.3

3 21

.58

11.7

3 12

.09"

Not

e. T

he

mul

tiva

riat

e an

alys

is f

or t

he S

AT

was

sig

nifi

cant

, m

ulti

vari

ate

F(5

, 45

) =

2.

59,

p <

.0

5.

Th

e m

ulti

vari

ate

test

for

the

Ap

ren

da

was

not

si

gnif

ican

t, m

ulti

vari

ate

F(4

, 76

) =

1.

41,

p >

.0

5. A

ll ad

vanc

ed s

tude

nts

wer

e in

clud

ed i

n th

e an

alys

es.

A s

ampl

e o

f p

rom

ote

d s

tude

nts

was

sel

ecte

d on

the

bas

is o

f th

eir

mat

ch t

o th

e ad

vanc

ed s

tude

nts

by

tes

t ta

ken,

age

, g

end

er a

nd

eth

nici

ty/h

ome

lang

uage

. *p

<

.05.

136 Journal of School Psychology

who had taken the SAT was nonsignificant, F(5, 41) = 1.16, p > .05, as was the M A N O V A comparing students who had taken the Aprenda, F(4, 54) = .93, p > .05. Thus, there were no differences in achievement at the end of first grade between the students who had been advanced to first grade after a year of kindergarten and those who had repeated kindergarten. The students in both groups, however, performed more poorly than did the stu- dents in their contrast group.

Regression

Two simultaneous, multiple regression analyses were conducted to assess the relative impact of the variables under study on overall achievement. Separate analyses were run for the SAT and the Aprenda. For the SAT gender, home language, ethnicity, age, retention, holding out, and advancement were en- tered, with the SAT's Basic Battery score as the criterion measure. The regres- sion analysis was significant, F(7, 347) = 12.71, p < .001, R = .45, R 2 = .20. The variables that had statistically significant relationships to the Basic Battery score were ethnicity and advancement. A multiple regression analysis was also conducted on the Aprenda Basic Battery score. As all the students taking this test were Latino and Spanish-speaking, the variables entered into the equation were gender, age, retention, holding out, and advancement. This test was also significant, F(4, 210) - 4.56, p < .05, R = .28, R 2 = .08. Both gender and advancement were significant predictors of achieve- ment. These variables had not been significant in the MANOVAs. However, there were a greater number of subjects in this analysis; furthermore, these variables accounted for only a small proportion (8 % ) of the total variance.

DISCUSSION

This study has several implications for school entry, early retention, and promotion decisions. First, it is apparent that a large number of students are having difficulty successfully entering the school system. In this rural, lower-income district, approximately 9% of students entered school overage, 6 % were retained at the end of kindergarten, and almost 13 % were advanced to first grade despite concerns about their scholastic performance. Thus, al- most a third of the students in this district had fallen outside of established standards for school entry and promotion by the end of kindergarten.

Second, it appears that teachers can, and do, identify students whom they perceive as having academic problems by the end of kindergarten. Decisions to advance or retain students in kindergarten in this district were based on teachers' judgments, as standardized tests were not administered until the end of first grade. Given the low test scores of retained and advanced students at the end of first grade, these data offer some support for the utility of teachers' perceptions in making student assessments (Hayes, 1990). However, the ex- tent to which first-grade test scores reflected the problems observed by teach-

Cosden et al. 137

ers in kindergarten, and the extent to which test scores were affected by advancement and retention decisions themselves, could not be determined in this study. For example, a self-fulfilling prophesy may take place, with stu- dents placed in retained and advanced categories treated differently across subsequent school years.

Neither holding out, nor retention, nor advancement had their intended effects on the students' achievement. With regard to holding out, children who entered school a year later than necessary did not have higher achievement scores at the end of first grade than children who had similar birth months but who entered on schedule. Thus, there was no evidence, based on these aca- demic measures, that younger children who entered on schedule were at a disadvantage relative to their peers who waited a year to enter school. This is important, given the pervasive belief that younger children will be more suc- cessful if they enter school a year older. This belief was operative in this study, too, as younger children were more likely to have been held out. There are several caveats to these findings, however. Only academic outcomes were assessed; thus, it is possible that holding out could have had an impact on other aspects of behavior and adjustment. Further, the analysis of group effects does not negate the possibility that individual students may have re- sponded positively to the intervention. Finally, the impact of holding-out practices on classroom climate and the need for teachers to deal with a greater diversity of skills in children across a wider age range also need to be assessed.

Children who were retained or advanced performed more poorly than their peers at the end of first grade. This suggests, at the very least, that neither intervention had a positive impact on the students' achievement. These find- ings confirm other evaluations of retention in kindergarten (e.g., Mantzico- poulos & Morrison, 1992; Shepard, 1989) and also speak against social ad- vancement, even in kindergarten. In interpreting these data, one needs to consider the instructional implications of being retained or advanced. Re- tained students in this district often had a second year in the same kindergar- ten program, or in a similar class within the same school. Advanced students were placed in regular first-grade classes, but without special remedial assis- tance. Reviews of similar practices suggests that students demonstrate aca- demic improvements only when they receive additional academic assistance the year after they have been identified (Holmes, 1989).

The students were matched by birth month, gender, ethnicity, and home language for analysis of holding-out, retention, and advancement effects. Nevertheless, there were many uncontrolled factors that may have influenced these outcomes, including parents' levels of education and the quality of their relationships with the schools. It is possible, for example, that achievement differences between advanced and promoted students could also be explained, in part, by differences in the parents' ability to access school resources. Fur- ther, in-depth studies are needed to determine other factors that play a role in early decision making by the schools and in early achievement.

Latino students were more likely than Anglo students to be advanced,

138 Journal of School Psychology

Spanish-speaking Latino students being the most vulnerable to advancement. Advancement was associated with similar student outcomes regardless of a student's ethnicity. However, given that a disproportionate number of Latino students were identified for advancement at the end of kindergarten, it is not surprising that a disproportionate number of Latino students scored lower on the SAT. It must be noted that both ethnicity and advancement were associ- ated with lower SAT scores, suggesting that some Latino students who were not advanced also did poorly on the achievement test.

It must be noted that the mean achievement score for all students in this district was relatively low, being below the 50th percentile relative to other schools in the state of California. This may create a "floor effect" in the data, which could explain the low proportion of test score variance accounted for by the regression analysis. These low scores may be associated with the low socioeconomic conditions in the district, or the high level of mobility that is associated with the economy in the area. Nevertheless, certain groups of students were more likely to have lower achievement scores than others.

CONCLUSIONS AN D RECOMMEN DATIONS

Early identification of students having academic problems is useful only if followed by interventions designed to assist them in catching up to their peers. A greater proportion of Latino than Anglo students in this study were identified with academic problems early in their schooling. The interventions utilized in this d is t r ic t -holding out, retention, and advancemen t -d id not have their intended positive effects on any of the groups of children. Waiting a year before entering school did not have a positive impact on students' academic performance, and children identified with problems in kindergarten continued to have problems in first grade whether they were retained or advanced.

Clearly, different types of interventions are needed to assure the smooth entry of these students into the school system. Holmes (1989) found that retention resulted in academic gains only when it was paired with additional, specialized help the second year. The National Association for the Education of Young Children (NAEYC) advocates flexible grading and promotion stan- dards for the first several years of schooling; this position is based on the premise, however, that instruction needs to meet children "where they are" rather than requiring that children come to school prepared to meet the de- mands of the curriculum (Bredekamp, 1987). While providing greater indi- vidualization of instruction in the early grades, or in any grade, may appear to be a costly proposal, the alternatives, as evidenced in this study, are more costly in many ways. Methods for implementing ungraded curricula in the early grades are currently being tested in the state of Kentucky, with some promising preliminary findings (Hovda, Kyle, & McIntyre, 1994).

The level of mobility in this district must also be considered in interpreting

Cosden et al. 139

these data. From the end of kindergarten to the end of first grade, 38 % of the students had moved and were not available for follow-up. To the extent that mobility impairs the continuity of academic experiences for students, not having these students in the database may result in a conservative estimate of achievement for this district. The impact of mobility on students' achieve- ment, and the relationship between ethnicity/home language and mobility patterns, require further study.

This study indicates that the needs of Latino students must be addressed early in their schooling. McDonnell and Hill (1993) noted that there are no national educational policies for immigrant children. Several investigators (e.g. , Lynch & Stein, 1987; Sue & Padilla, 1986) have suggested that school- ing problems for Latino children are at least in part due to limited access to information about the school system. This lack of information is a function of both cultural and linguistic barriers. Different conceptualizations of what it means to prepare children for school can have a significant impact on chil- dren's early experiences as well as on teachers' perceptions of children's school- related skills. Greater efforts are needed to make the school culture accessible to all families, and to assure that school practices are designed so that all children can be successful.

REFERENCES

Barona, A., & Garcia, E. E. (Eds.). (1990). Children at risk. Washington, DC: Na- tional Association of School Psychologists.

Bourdieu, P. (1991). Language and symbolic power. Cambridge: Harvard University Press.

Bredekamp, S. (Ed.). ( 1987 ). Developmentally appropriate practice in early childhood pro- grams serving children from birth through age 8. Washington, DC: National Association for the Education of Young Children.

Byrnes, D. A. (1989). Attitudes of students, parents and educators toward repeating a grade. In L. Shepard & M. Smith (Eds.), Flunking grades: Research and policies on retention (pp. 108-131 ). Bristol, PA: Falmer.

Byrnes, D. A., & Yamamoto, K. (1986). Views on grade retention. Journal of Research and Development in Education, 20, 14-20.

Cannella, G. S., & Reiff, J. C. (1989). Mandating early childhood entrance/retention assessment: Practices in the United States. Child Study Journal, 19, 83-99.

Carstens, A. A. (1985). Retention and social promotion for the exceptional child. School Psychology Review, 14, 48-63.

Cosden, M., Zimmer, J., & Tuss, P. (1993). The impact of age, gender, and ethnicity on kindergarten entry and retention decisions. Educational Evaluation and Policy Anal- ysis, 15, 209-222.

Fowler, M., & Cross, A. (1986). Preschool risk factors are predictors of early school performance. Journal of Developmental and Behavioral Pediatrics, 7, 327-341.

Freeman, E. B., & Hatch, J. A. (1989). What schools expect young children to know and do: An analysis of kindergarten report cards. Elementary School Journal, 89, 595- 605.

Grissom, J. B., & Shepard, L. A. (1989). Repeating and dropping out of school. In L. Shepard & M. Smith (Eds.), Flunking grades: Research and policies on retention (pp. 34-63). Bristol, PA: Falmer.

140 Journal of School Psychology

Hayes, A. (1990). The context and future of judgment-based assessment. Topics in Early Childhood Special Education, 10, 1-12.

Holmes, C. T. (1989). Grade level retention effects: A meta-analysis of research studies. In L. A. Shepard & M. L. Smith (Eds.), Flunking grades: Research and policies on retention (pp. 16-33). Bristol, PA: Falmer.

Holmes, C. T., & Matthews, K. M. (1984). The effects of nonpromotion on elemen- tary and junior high school pupils: A meta-analysis. Review of Educational Research, 54, 225-236.

Hovda, R., Kyle, D., & McIntyre, E. (1994). Nongraded primary programs within a context of educational reform. Paper presented at the American Educational Research Association, New Orleans, LA.

Langer, P., Kalk, J. M., & Searls, D. T. (1984). Age of admission and trends in achievement: A comparison of Blacks and Caucasians. American Educational Research Journal, 21, 61-78.

Lynch, E., & Stein, R. (1987). Parent participation by ethnicity: A comparison of Hispanic, Black, and Anglo families. Exceptional Children, 53, 105-111.

Mantzicopoulos, P., & Morrison, D. (1992). Kindergarten retention: Academic and behavioral outcomes through the end of second grade. American Educational Research Journal, 29, 182-198.

McDonnell, L. M., & Hill, P. T. (1993). Newcomers in American schools: Meeting the educational needs of immigrant youth. Santa Monica, CA: Rand.

Mergendoller, J., Bellisimo, Y., & Horan, C. (1990). Kindergarten holding out: The role of school characteristics, family background, and parental perceptions. Paper presented at the annual meeting of the American Educational Research Association, Boston, MA.

National Center for Educational Statistics. (1993). Public school kindergarten teachers' views on children's readiness for school. Washington, DC: United States Department of Education, Office of Educational Research and Improvement.

Niklason, L. B. (1987). Do certain groups of children profit from a grade retention? Psychology in the Schools, 24,339-345.

Psychological Corporation. ( 1975 ). Bilingual Syntax Measure. San Antonio, TX: Har- court Brace Jovanovich.

Psychological Corporation. (1989). Primary 1 Aprenda: La Prueba de Logros en Espanol. San Antonio, TX: Harcourt Brace Jovanovich.

Psychological Corporation. ( 1991 ). Aprenda: La Prueba de Logros en Espanol national norms booklet. San Antonio, TX: Harcourt Brace Jovanovich.

Psychological Corporation. (1992). Prima~y 1 Stanford Achievement Test. (8th ed. ). San Antonio, TX: Harcourt Brace Jovanovich.

Shepard, L. A. (1989). A review of research on kindergarten retention. In L. A. Shepard & M. L. Smith (Eds.), Flunking grades: Research and policies on retention (pp. 64-78). Bristol, PA: Falmer.

Shepard, L. A., & Smith, M. L. (1986). School readiness and kindergarten retention: A policy analysis. Educational Leadership, 44, 78-86.

Shepard, L. A., & Smith, M. L. (1987). Effects of kindergarten retention at the end of first grade. Psychology in the Schools, 24, 346-357.

Shepard, L. A., & Smith, M. L. (1988). Escalating academic demand in kindergar- ten: Counterproductive policies. Elementary School Journal, 89, 135-145.

Shepard, L. A., & Smith, M. L. (1989). Introduction and overview. In L. A. Shepard & M. L. Smith ( Eds. ), Flunking grades: Research and policies on retention (pp. 1-15 ). Bristol, PA: Falmer.

Smith, M. L. (1989). Teacher's beliefs about retention. In L. A. Shepard & M. L. Smith (Eds.), Flunking grades: Research and policies on retention. Bristol, PA: Falmer.

Smith, M. L., & Shepard, L. A. (1988). Kindergarten readiness and retention: A qualitative study of teachers' beliefs and practices. American Educational Research Journal, 25, 307-333.

Cosden et al. 141

Sue, S., & Padilla, A. (1986). Ethnic minority issues in the United States: Challenges for the educational system. In California State Department of Education, Bilingual Education Office and Evaluation, Dissemination and Assessment Center, Califor- nia State University, Los Angeles, Beyond language: Social and cultural factors in school- ing language minority children. Los Angeles: Author.

Tomchin, E. M., & Impara, J. C. (1992). Unraveling teachers' beliefs about grade retention. American Educational Research Journal, 29, 199-223.

U.S. General Accounting Office. ( 1993 ). School age demographics: Recent trends pose new educational challenges. Briefing report to congressional requesters: GAO/HRD-93- 105BR. Washington, DC: GAO.

Walsh, D., Ellwein, M. C., Eads, G., & Miller, A. ( 1991 ). Knocking on kindergar- ten's door: Who gets in? Who's kept out? Early Childhood Research Quarterly, 6, 89- 100.