Disordered Data and Murky Models
Research Study
Few education research studies have garnered greater attention in education-policy circles than the analyses of programs for language minority students conducted by Virginia Collier and her George Mason University colleague Wayne Thomas. In fact, much of the public attention received by their most recent paper occurred before the report was self-published in 2002. Since then, the paper has been cited and its findings discussed in public documents and proceedings around the country, particularly to support teaching English learners in their non-English native language.
This critique, by Boston University professor of political science Christine H. Rossell, offers the first systematic review of the 2002 “National Study of School Effectiveness for Language Minority Students’ Long-Term Academic Achievement,” and its findings. Among its conclusions:
• Numerous assertions of educational benefits remain unsubstantiated, or worse, influenced by significant factors ignored by the authors.
• While Thomas and Collier repeatedly refer to various assertions that “the research to date has found,” in fact, there is no research other than theirs that has found this and their research often does not withstand close scrutiny using standards of social science research.
• The research design contains fundamental flaws, such as the absence of a control group — similar students in alternative program(s) — and the absence of statistical control (if there is no random assignment) for other variables that explain achievement, render many significant conclusions highly questionable.
Find Archived Articles: