Comment

Jon Lorence, A. Gary Dworkin, Laurence A. Toenjes, and Antwanette N. Hill begin with a critique of the existing research literature on grade retention. They argue that their study, based on a sample of 736 retained students and 28,351 socially promoted controls, overcomes the weaknesses of previous studies because it is larger, uses a matched comparison group, and follows achievement over a six-year period. In contrast to prior research syntheses showing no academic benefit from retention, Lorence and his colleagues find a positive effect that is striking in the first year following retention (one-half standard deviation achievement gain of retained students over controls). This benefit diminishes over time but still shows a slight positive effect when retained students reach seventh grade. The authors attribute the positive effects of retention as well as the achievement gains found for both retained and socially promoted students to the strictures of the Texas accountability system, which sanctions districts and schools if they fail to raise scores for low- performing students.

Summary of Existing Research

Lorence and his colleagues acknowledge that major research reviews of the literature have found grade retention to be ineffective but cast aspersions on negative findings by referring to the views of educational researchers and their entrenched perceptions. They attempt to explain away negative evidence, calling it biased or flawed. However, their arguments are misleading in several respects. First, the authors quote one or two positive studies and then hold them up against meta-analyses or research reviews that included many dozens of studies.⁴⁰ For example, the 1987 S. E. Peterson, J. S. DeGracie, and C. R. Ayabe study was one of many included in the 1989 C. T. Holmes meta- analysis, and the 1994 K. L. Alexander, D. R. Entwisle, and S. L. Dauber study was included in the 1999 J. P. Heubert and R. M. Hauser review.⁴¹ Evidence from these positive studies was considered but outweighed by findings from many more negative studies. Second, the authors imply that new, rigorous studies find positive effects for retention while only old, flawed studies find negative effects. Many of the studies cited in research summaries used statistical controls and matching strategies as good as those employed by Lorence and his colleagues. The Holmes meta-analysis used degree of study control as a variable in examining variation in study findings; and Heubert and Hauser explicitly considered methodological adequacy in weighing findings from Alexander, Entwisle, and Dauber as well as from other studies. Thus, the authors. seem to be going out of their way to build a case for retention instead of trying to evaluate--on balance--where the weight of evidence lies.

Lorence, Dworkin, Toenjes, and Hill have correctly identified me and my colleagues as critics of retention policies based on the evidence. But in their concluding section, they seriously misrepresent our work. They say that "proponents of social promotion would argue that the emotional hardships experienced by retained children and their parents are unnecessary because such children will mature sufficiently in a few years and will then be able to learn the required curriculum." This sounds like our summary of research on kindergarten readiness, kindergarten retention, and within-grade age effects, whereby we showed that kindergarten retention for reasons of social immaturity was unwarranted because within-grade age effects disappear. However, we explicitly distinguished kindergarten retention for reasons of immaturity from retention used as a remedy for poor academic performance. To my knowledge, no one who criticizes the effectiveness of retention as a means to improve achievement would claim that poor achievement can be overcome by ignoring it. Instead, the argument would be to use instructional interventions that are more effective than repeating a grade in school.

Methodological Strengths and Weaknesses

quasi-experimental studies suffer from some methodological shortcomings. Lorence and his colleagues need not claim that their study is impeccable and all other studies seriously flawed. Study...