The age-grouped classroom is quite new to education and is largely a product of the industrial revolution and especially of the scientific management movement, though there were somewhat earlier threads of similar character. Three main forces have driven the age grouping practice. First, the realization that children’s cognitive capabilities differ from those of adults, stimulated by Freud and Piaget, led to the belief that children of different ages need different cognitive resources, need different learning methods, and have different cognitive dispositions for acquiring various forms of knowledge. The extreme form of this view is that children need age mates to support the particular kinds of thinking and emotion that their level of maturation involves. So, for example, the very idealistic kibbutz movement in Israel arranged living facilities so that children lived with age mates rather than with their families (of course, community was valued over family in this culture and there were practical considerations as well, but still, a major investment in social change was driven by beliefs in the importance of age-peer interaction).

The second major force toward age grouping was the drive for universal education. This, and the third force, efficiency demands, forced a move from a tutorial approach to a didactic approach in age-grouped classes. When a wealthy person had three or four children to educate and no communal resources to provide that education, hiring a single tutor to teach all the children made great sense. However, when a community had all of its children to educate, specialization of labor became appropriate, and the most obvious and socially compatible specialization seemed to be by age. Had society set its sights higher at this time, the specialization could just as well have been first by subject matter, but the initial goals of universal literacy were low enough that it was assumed that any teacher already had mastered the content. Rather, teacher preparation focused on teaching methods, especially adaptation to children of different ages.

Through all of this time, minimal attention was paid to the variability of learning rates among children. We do not have complete data on how disparate educational achievement would be if we did not limit the speed at which children could progress, but there is every reason to believe that it would be dramatic (cf. Gettinger, 1984, which cites other findings of a 5:1 ratio among students in time to learn). Even within a highly constrained system, in which faster learners are not afforded sufficient learning opportunities, there is ample evidence of major variability. We present here a few modest examples of the evidence that supports this view.

Consider the standardized educational achievement tests we give to college bound students. In the case of the College Board’s Scholastic Achievement Test, for example, scores of college bound seniors have a standard deviation of about 100, by design. What is less-often discussed is the average rate of growth in SAT scores over the course of schooling. Now that we have a substantial number of children taking the SAT as early as 11 or 12 years of age, there is plenty of evidence to support the view that SAT scores grow an average of about 35 points per year (Bond, personal communication, ca. 1985).<4> This means that the standard deviation of the test scores at the end of secondary education is about three years worth of normative learning. While the construction of the test limits the absolute precision of the following claim, it still seems worthwhile to infer that only about 2/3 of high school seniors have attained a level of intellectual capability within three years of the average. To include all but perhaps 5% of students, we would need to have a range of perhaps 12 years of normative education, i.e., the mean plus or minus two standard deviations. In general, this reasoning, and examination of a variety of data sources, suggests that in age-grouped education, the range of achievement is roughly equal to the average - if we look at students after n years of schooling, they will range over 0 to 2×n years of normative achievement.

A very different view is available from the entry end of the public schooling world, though the conclusion is roughly the same. Lesgold, Resnick, and Hammond (1985) conducted a longitudinal study of learning to read. Over the first three years of schooling, children were tested up to eight times in an effort to understand the differential effects of two different approaches to reading instruction. For our purposes, though, what is important is the variability in outcomes of learning over the three grades. At the end of the third grade, when students are assumed to have learned the basics of reading, oral reading speeds in the top 20-25% of the students averaged about 100 words/min on passages similar in difficulty to those in their readers but not prepracticed, roughly the speed of conversation and clearly fast enough to support higher-level thinking processes. The bottom 20-25% of students, though, averaged only 50 words/min for one curriculum and 80 words/min for the other, slow enough to be a substantial barrier to understanding. Even with multiple reading groups, major differences in achievement remain, with many students reading so slowly after three years that they will have trouble doing much meaning processing of what they read and others clearly done with the basics of learning to read after a year or so.

While primary reading instruction is partly individualized, with each classroom usually having several different "reading groups," this individualization tends to end after the third grade. It is interesting to note that Lesgold et al.(1985) observed substantial differences in achievement that did not decrease with years of schooling over the primary grades, yet the capability of the instructional system to support continued individualization past Grade 3 is generally quite limited. While dispersion of achievement did not narrow over grades, Lesgold et al.(1985) found that dispersion of "reading level" was maximal at second grade. That is, students begin primary education in a system that allows reading to be acquired at differential rates, but between second and third grade, even though achievement continues to vary widely over students, classification variance decreases—our society expects almost all children to reach the top reading level by the end of third grade, so they do (on paper), independent of their real competence, as measured by reading of passages not previously drilled in class.