In her terrific article published in the March 1991 issue, Susan Demirsky Allan concludes that Slavin did not consider the quality of the programs used for gifted students in his meta-analyses that show little effects from ability grouping ("Synthesis of Research on Grouping in Elementary and Secondary Schools," September 1988).

In his response, ("Are Cooperative Learning and 'Untracking' Harmful to the Gifted?" March 1991), Slavin notes that meta-analyses for the gifted have shown effect sizes for accelerated but not enrichment programs. In addition, there are no major effect sizes for interventions for high-ability students.

Unfortunately, this response again overlooks Susan Allan's criticism. Studies included in a meta-analysis are carefully selected. However, the criteria are usually based solely on the quality of the statistical design with usually no consideration of whether the interventions in the studies were effective. Often, meta-analyses number-crunch the results from good statistical studies of lousy programs.

Does this type of research knowledge extend knowledge? Not really. Overrelying on pessimistic results from poor interventions can inhibit the development of more effective approaches.

To his credit, Dr. Slavin has proposed basic standards for programs to be considered in meta-analyses. These standards, however, are woefully inadequate, and have not been widely accepted.

I share the ideal of getting students from varying backgrounds to work together. I have devoted my professional life to trying to figure out how to get educationally disadvantaged students to succeed in academic coursework. But I do know that simply untracking and relying on cooperative learning will not work. There are huge differences in learning rates between gifted and disadvantaged students. In addition, policies that discourage efforts on behalf of high-ability students will encourage the financially able parents of such children to leave public education. Meta-analysis can not change these realities.

The only real solution is to develop much more powerful interventions for the educationally disadvantaged students so that they can really compete with educationally advantaged students and become eligible for high-quality enrichment and gifted programs. In addition, the only research that matters is that which is done about sustained high-quality interactions between a good teacher and students.