Table 7
Attainment by Grade 5

With respect to question (2), whether two-way students do better than English learners in the mainstream, two studies suffer from very small samples, short durations, and lack of measures for initial competence. In another, Thomas and Collier’s Houston data, two-way students do better than those in submersion but not as well as those in ESL-only. This data, however, is uncontrolled for a number of important factors.

With respect to question (3), two-way children outperform children in transitional bilingual education in two studies, but do worse in another. None of the studies provides us assurance that the children were at similar levels when starting school, one study, the Amigos project, ends at grade 2 and has a small sample size, and in the Houston study scores for all children were very high at grade 1, and are lower in subsequent years. In addition, children in two-way in Houston were superior and had high scores after only one year of school, suggesting a selection bias and considerable previous English competence. The only direct comparison of two-way with a developmental program showed evidence that two-way children did better, but sample sizes were modest and the developmental children appeared to be closing the gap between the two programs (note that Thomas and Collier, 2002. contains an indirect comparison of two-way with developmental bilingual education, with two-way doing better.)

Two-way children do as well or better than native speakers of English in five studies. In three of the five, the sample sizes were small or modest, one of the “two-way” programs was really immersion with some oral first language support, and durations were short in two studies.

11