The dependent variable of the study was students' performance on language assessment tests geared towards ELL students, and the independent variable was that of performance on general academic assessment test. However, there was an underlying assumption that the first component of the assessment, namely that of the measured language proficiency was 'correct.' Other forms of proficiency, such as general life skills, classroom performance, and teacher perception of student improvement were not tabulated nor assessed through quantitative or qualitative methodology.

There are other problems with the study. First of all, the definition of poverty by assessing student participation in school lunch programs is problematic, given this assumes such programs measure poverty in absolute terms, while the accuracy of their measurements of student poverty has been called into question in recent years. Secondly, the study admits that it has not arrived at a solution of one of standardized testing's dilemmas regarding ELL students with little or no proficiency in English: "Waiting for students to reach a certain level of English proficiency before standardized academic testing is not pragmatic in the era of accountability," it admits but ELL learners constitute a highly diverse group (Mahon 2006:495). The study makes broad conclusions based upon a Hispanic population of Spanish speaking ELL students, but the different educational needs, types of proficiency, and overall success...

Consider this contradictory scenario to the study's finding that English language skills and academic achievement will always positively correlate: a Hispanic child speaking better English in an impoverished home might have a higher language ability than a recent immigrant child of two professors from France, but the academic level of the latter child within several years might be vastly superior to that of the former, simply because of the resources he or she had access to subsequently.
From an educator's standpoint, the greatest weakness of the study is that it did not compare the different types of assessment programs within the context of the study, to see if, for example, 'pull out' programs yielded superior results to dual language or transitional programs. But because the economic status of the schools was so different, no substantiated conclusions could be reached, anyway. In short, the study was too homogenous in terms of only studying Hispanic-speaking students to be useful, and also not sufficiently segmented in terms of the possible variables that could affect or inform the results.

Works Cited

Mahon, Elizabeth a. (2006, Summer). "High-stakes testing and English Language

Learners: Questions of validity." Bilingual Research Journal. 30.2: 479-496. Retrieved 15 Apr 2008 at http://brj.asu.edu/vol30_no2/art11.pdf