Commentary—Sorting It Out

Description of the Study

Collins and Gan undertook a complex study to examine the impact of grouping practices on student achievement within 135 schools in the Dallas independent school district. Their study addressed three issues: (a) how schools sort students into classes, (b) the effect of these sorting practices on student performance, and (c) differences in effects for different groups of students. They looked at the “tracking effect” on high- and low-achieving students, students who had been identified for special (SPED) and gifted (GT) education, and for students with limited English proficiency (LEP). Using a unique design, they were able to identify each school’s grouping patterns by looking at the range of student scores, on the Texas Assessment of Knowledge and Skills (TAKS) for math and reading, within each grade. Then they created formulas to estimate the average score dispersion across each class. Through this, they were able to identify the level of ability grouping (i.e., grouping based on prior student achievement) used by each school and thus group the schools according to how “tracked” they were.

Although they found wide variability in degree of tracking across the Dallas schools, almost 75% of the schools sorted students using at least one dimension (e.g., ability, GT, SPED, LEP). Nineteen percent of the schools sorted students by math scores, 24% by reading scores, 28% by GT status, 57% by LEP status, and 13% by special education classification. Almost 40% of the schools grouped students using two dimensions, and more than 20% of the schools sorted by at least three dimensions. Using the variation of school’s sorting and the level of dispersion within the classroom, Collins and Gan looked at the impact of tracking on the current achievement of high- and low-scoring students. In addition to looking at the impact of grouping practices based on students’ past achievements, they also looked at the impact for students classified as GT, SPED, and LEP.

The study’s design is robust, and the authors were able to address the possible influence of unobserved variables correlated with sorting, such as parent requests for teachers, through the creation of sorting indices and grade-specific sorting measures for all classes within a given grade. They were also able to work with a large data set (9,325 students from 135 schools) lending confidence in their findings. The authors indicate that data were available for student’s race, economic status (i.e., free and reduced lunch), and gender, and they indicate that these variables did not appear to have been used to sort students. They did not, however, include these variables in their set of observable characteristics to examine the impact of tracking specifically for these groups of students. By omitting this, the authors did not explore possible disparate impacts of tracking based on student’s race, gender, or economic status. Given the early concerns raised within the literature including this would strengthen their design.

Does Sorting Students Improve Scores?

Collins and Gan found a statistically significant positive relationship between their sorting variable (i.e., the degree to which students had been grouped by their previous scores) and student math scores. This suggests that in general, students gain in math from being placed in more homogeneous classrooms when compared with classrooms that include a wider range of test scores. Similarly, although somewhat smaller, benefits were found for students’ reading scores. The authors conclude that “. . . on average, more homogeneous classes are beneficial for students in increasing both math and reading achievement” (p. 18). Next, the authors looked at the possibility that sorting may benefit one group at the detriment of others. The first comparison looked at high- versus low-scoring students, and the authors found that both these groups of students appear to benefit from homogeneous classes. This held true for math and reading achievement. This is of particular interest to educators who worry that ability grouping may benefit students in the top ranges while harming lower scoring students—an argument sometimes used to preclude homogeneous grouping for high-achieving students. The authors indicate that this finding supports the view that teachers are better able to meet student needs when the range of abilities is narrowed.

The second comparison made was students identified as gifted with non-identified students. Findings of this comparison were more complex. Overall, the effects were positive but not statistically significant, with the most positive gains being made by the non-identified students. The authors’ comments on gifted students were limited to two sentences with no additional discussion. Although the authors did not discuss their findings, we might conclude that ability grouping seems to have a positive academic impact for students who have not been identified as gifted while not harming students identified as gifted; however, their gains are not as impressive. Because the authors did not explore their findings regarding gifted students and gave no specific information on their results, it is left to the reader to speculate about why the positive gains for gifted students were not stronger.

The lack of significant positive gains for students identified as gifted makes one wonder what, if anything, was being done differently for these students. This finding may indicate that while gifted students were grouped, the curriculum and instruction remain roughly the same, and so, gains would likely not show for gifted students who were already scoring in the highest ranges. Another possible explanation is that the use of on-grade-level math and reading test scores is not sufficient for determining gains for students within the highest achievement ranges. This may be due to ceiling effects and/or may reflect natural regression toward the mean. A third possible explanation is that the goals for the students identified as gifted may include things not measured by achievement test scores (e.g., creative problem-solving or the deep pursuit of a topic of interest) and that the value-added gains for goals like this are missed when only math and reading achievement scores are used. Finally, one often cited benefit for homogeneous grouping for students who have been identified as gifted is the opportunity for establishing peer relationships, and this was also not reflected in the design of this study. Based on the limited measures for student outcomes used in this study (i.e., scores of math and reading achievement), it would be inaccurate to conclude that there are only slight positive overall benefits for students identified as gifted when ability grouping is used.

The results of the impact of grouping for students identified for special education versus the non-identified students show negative effects for non-special education students and positive, although not significant, effects for special education. The authors do not discuss these findings. Finally, homogeneous classes were shown to be useful for increasing performance of LEP students. Again, the authors did not discuss these results.

Implications for Students Identified as Gifted

The current study fits well with research done on the impacts of ability grouping for gifted students by Karen Rogers. Rogers’s (2007) work on acceleration and subject grouping for gifted, spanning the last several decades, had shown substantive gains in academic growth for students. Through her research, Rogers found that the effects for grouping of gifted students “range from approximately one third of a year’s additional growth (advanced placement or international baccalaureate) to three fifths of a year’s academic gain (mentorship and subject acceleration)” (Rogers, 2007, p. 387). Neihart’s (2007) look at the impact of grouping on socialization for gifted students, however, had mixed conclusions. Neihart summarizes her findings stating that “although the research finds academic and achievement benefits for ability grouping for gifted students, the research does not support the claim of social or emotional benefits for such grouping arrangements” (Neihart, 2007, p. 338) and that “homogeneous grouping arrangements are more strongly associated with positive [social] adjustment outcomes among highly gifted children, although this connection is less clear with moderately gifted students” (Neihart, 2007, p. 338). Neihart calls for additional research looking at the specific social and psychological impacts of group for gifted students with special attention paid to students from minority families and students who are twice exceptional.

The findings of the Collins and Gan study have implications for K-12 gifted programs and for researchers. For K-12 practitioners and parents, these findings provide support for ability grouping, discrediting the idea that although grouping may benefit high-scoring students (e.g., those identified as gifted), it will harm others academically. These findings are in line with Roger’s studies on the impact of grouping for students identified as gifted. The study, however, raises other questions as to why the benefits of grouping for gifted students specifically did not seem to be as strong as previous research indicated it might be. Questions that need to be explored by both practitioners and researchers include the following: What is being done differently for students who have been identified as gifted? What measures best capture the true achievement levels for gifted students in their areas of strength? and How should we document and measure the academic and social value-added benefits for gifted students? We also need to explore the more nuanced impacts of grouping on students who are twice exceptional and gifted students from culturally and linguistically diverse populations.

Caveats and Limitations of the Study

This study used an extant data set to explore important issues related to the impact of grouping practices on a variety of students. The study design was robust, and the authors developed innovative statistical models to examine their question, but we must remember that the study design was correlational and not causal. A limitation of this study is that we do not know what additional “treatments” the students were exposed to. For example, we might logically assume that students identified for special education, LEP programs, and gifted and talented classes are all receiving additional supports and services, and we do not know the nature or impact of these on student achievement. Another limitation is the narrow definition of achievement being defined as math and reading test scores which precludes other important goals for schooling. A final limitation is that the study failed to explore the impact of homogeneous grouping specifically for students by race/ethnicity, economic status, and gender. By not addressing these variables we are still left wondering if grouping might have a disparate impact on these student groups.

The authors make the case that their findings show that ability grouping is beneficial for different groups of students regardless of other interventions and that it is a cost-effective way to increase achievement scores. They argue that by reducing the range of students’ scores within a classroom, we enhance efficiency for teaching methods and resource allocation in the most cost-effective way. Although this may be true, policy decisions must also reflect outcomes for a range of programming options, including those that are not narrowly defined by math and reading scores. It would be unfortunate, for example, if policy makers jumped to the conclusion that we could eliminate special programs, such as those for students identified as gifted, thus saving money, if we implemented a grouping model that established homogeneous classes based on student scores in math and reading.

Conclusion

The major take away from this study is that the authors found positive outcomes for ability grouping across a range of students; thus, undermining the argument that grouping, while good for high-achieving students, is detrimental to others. This is exciting news and, if used in light of the concerns raised above, it will help to make the case for appropriate ability grouping as part of our efforts to create learning environments that promote excellence for all our students.