Thomas and Collier (2002) reported that a larger percentage of Spanish-speaking children at a two-way program at the Grant School in Oregon exceeded state norms than did students state and district-wide. This is impressive, but it was only true for one of the two groups tested, consisting of 12 students in one grade (grade 3). In grade 5, Grant students do not do well, scoring lower than district and state norms, but Thomas and Collier note that the Grant group (only 17 children) included late-comers. The Grant school has a very high mobility rate (71% annual). There was no separate analysis of those who have been in the program since kindergarten.

A multiple regression analysis, however, revealed a positive and statistically significant relationship between years in the program and scores on the English reading test, controlling for SES (poverty). Moreover, the effect of years in the program was larger than the effect of SES.

It is difficult to determine whether children at Grant began kindergarten with no English. It is unlikely. At grade 1, Grant students in scored 13.81 on the SOLOM measure of oral English, a test with a range of 5 to 25; a score of 19 is considered “proficient.” By way of comparison, English speakers in the two-way program did not reach this level in Spanish until grade 4.

Group Three: Comparison to English Learners in the Mainstream

We would expect English learners in properly organized two-way programs to outperform comparison children in the mainstream, especially those in “submersion” programs.

In Ajuira (1994), "Hispanic" two-way students in grade 1 did better than comparison students in a mainstream class on a modified version of the Iowa Test of Basic Skills. Ajuria modified the test herself, making it shorter, including only items of “average difficulty” and items rated “fairly easy” (p. 139). The scores ranged from zero to six, but Ajuira does not tell us how this scale was constructed.

3