6
Table 3 presents scores on the California Achievement Test for two cohorts. The results are impressive: two-way children outperform comparisons in all grade levels for both cohorts, and achieve above the expected (grade level) score in grade 3 in one cohort. There are, however, reasons to be somewhat cautious:
| (1) | The sample sizes are very small. |
| (2) | The study only goes up to grade 3 |
| (3) | There is unusual variability in the grade 3 1989-90 score (standard deviation = 2.6). (The sd of 7.3 for comparisons in grade 1 appears to be an error; more likely, it was .73). |
| (4) | The 1990-91 cohort shows a decline from grades 2 to 3, making less than one year progress in one year. |
It is noteworthy that “Spanish amigos” children were behind comparisons on a measure of “communication skills” (done in an interview) in grade 1 but performed better than comparisons in grades 2 and 3, increasing their lead in grade 3.
Table 3
"Spanish Amigos" English Competence (grade level equivalents)
|
|||||||||||||||||||||||||||||||||||||||||||||
From: Cazabon, Lambert and Hall (1993)
In their Houston report, Thomas and Collier go a long way toward controlling for pre-existing differences among the groups. In their report, data is provided for two-way bilingual education as well as two alternative programs: transitional bilingual education (TBE) and “developmental bilingual education.” All three provide equivalent amounts of Spanish and English up to the end of grade 3. In grade 4, the transitional program contains more English, while the other two maintain 50-50 Spanish-English.
Table 4
Comparison of Bilingual Programs/Cross-Sectional (NCE/Percentiles)
|
from: Thomas and Collier (2002)
6