Gene V. Glass and Mary Lee Smith claim in their report, "Meta-Analysis of Research on the Relationship of Class-Size and Achievement" (ED 168 129), that their integration of data from 80 previous studies through complex regression analysis techniques revealed a "clear and strong relationship" between decreases in class size and increases in academic achievement. Due to numerous errors in methodology, these findings are highly suspect. These errors include the incorrect assumption that the size of their sample assured randomization, the compounding of statistical errors already existing in the original studies, a lack of adequate concern for statistical significance, and a failure to investigate or account for all the important variables involved in the problem. An additional difficulty is that Glass and Smith failed to take into account the realistic constraints of educational finance, thereby reducing the possible value of their findings. (Author/PGD)