Introduction
RE: Jensen, B. T. (2007). The relationship between Spanish use in the classroom and the mathematics achievement of Spanish-speaking kindergartners. Journal of Latinos and Education, 6, 267–280.
After publication of Bryant Jensen's article The Relationship Between Spanish Use in the Classroom and the Mathematics Achievement of Spanish-Speaking Kindergartners in the third issue of the sixth volume of this journal, an error in the interpretation of the analyses was found by the author. Although the analytical methods and results shown in the tables and figures were correct, an interpretation of the results was inaccurate.
A primary purpose of this paper was to explore teacher characteristics associated with higher math performance of Spanish-speaking children during kindergarten. As the paper indicated, three teacher characteristics, identified in the literature as related to academic performance of young Hispanic English learners, were identified—Spanish use in classroom instruction, perspectives of English knowledge for school readiness, and years of experience teaching kindergarten. Multiple regression analyses indicated that the latter two characteristics did not account for a substantial amount of variance over and above Spanish use in the classroom. It was concluded, therefore, that of the teacher characteristics assessed, Spanish use in the classroom was the only one to uniquely predict variations in math achievement of Spanish-speaking kindergartners in the nationally representative sample—albeit with a small effect size (i.e., r 2 = .04). Moreover, it was found that the significance and size of this effect varied across levels of socioeconomic status, maternal education, and student English proficiency as shown in Figure 2.
Until this point, there were no errors. The reported effect sizes from the overall sample and subsamples across levels of the aforementioned sociodemographic characteristics were correct. The error found was a misinterpretation of the direction of the relationship between Spanish use and math scores. That is, it was reported that Spanish use was associated with higher math scores when, in actuality, Spanish use was associated with lower scores. While perusing the ECLS-K data, the author discovered that the positive correlations between Spanish use and math scores were interpreted incorrectly, assuming the teacher variable was coded as 1 = no and 2 = yes (indicating whether Spanish was used for classroom instruction). However, the author discovered that, as indicated in the ECLS-K electronic codebook, “yes” was coded 1, and “no” coded 2. This means for the overall sample and subsamples alike, where interpretable effects were found, Spanish use was associated with lower math scores, as shown in the attached correlation table.