In light of a recent revelation that Gersten (1985) included erroneous information on one of two programs for English Language Learners (ELLs), the authors re-calculate results of their earlier meta-analysis of program effectiveness studies for ELLs in which Gersten's studies had behaved as outliers (Rolstad, Mahoney & Glass, 2005). The correction resulted in a change in mean effect size from 0.08 to 0.19 for all outcome measures, from -0.06 to 0.14 for (English) reading, from 0.08 to 0.17 for (English) math, and from -0.01 to 0.10 for all Transitional Bilingual Education (TBE) studies. The revelation of Gersten's coding error, and the inconsistency of Gersten's studies with other studies reviewed, increases confidence in the conclusion that an "investigator effect" suppresses results favoring TBE in these studies. Removing Gersten's studies from the meta-analysis renders an effect size of 0.17 for TBE, nearly as high as for Developmental Bilingual Education (DBE). The authors argue that the most informative result is the effect size reported for studies involving ELLs in both treatment and control groups, with an average effect size for TBE of 0.23. The new analysis therefore strengthens the conclusions previously reached in the authors' original research supporting TBE over English-only approaches, and DBE over TBE. (Contains 1 table and 2 endnotes.)