Does Sorting Students by Ability Improve Learning or Increase Disparity?

The Issue:

Whether or not to separate students into classrooms according to their ability is a contested issue. Proponents argue that it is a low-cost tool to improve learning since instruction and resources can be more precisely tailored, making education more effective. Opponents counter that this type of stratification exacerbates initial differences in opportunities, setting students placed in the lower tracks on a path to lower educational attainment and decreased earnings, with the potential of aggravating economic inequality and perpetuating intergenerational disadvantage. Existing research has not come to a consensus on this debate. What is more, relatively little is known about the extent to which schools in the United States engage in this behavior in elementary and secondary schools. New research looking at 4th to 8th grade public school students in Texas provides insight about the prevalence of the practice and how it impacts student achievement.

Relatively little is known about the scope and nature of tracking — the sorting of students into classes based on ability — in the U.S.

The Facts:

• In contrast to systems common in other countries that stream students to different schools or programs of study, the ways by which students are assigned to classrooms according to ability in the United States are often informal. Because of this, relatively little is known about the scope and nature of tracking — the sorting of students across classes within a school based on ability — in the United States. Nonetheless, a nationally representative survey of public school teachers and principals suggests that about one quarter of public elementary schools and about half of public middle schools in the United States tracked students by ability in 2017-2018 (see here). The assignment of students to classrooms based on ability could be inadvertent on the part of the school, such as if parents successfully push for specific teacher assignments, or purposeful, such as if administrators use achievement as a factor in class and course assignments. To facilitate tracking, schools or districts could adjust class sizes or offer more advanced or remedial course offerings. There are also relevant state policies regarding special student populations, such as gifted and talented students, English learners and students with disabilities, that result in placing students along different tracks.
• One way to estimate the extent to which schools track students is by measuring how much a student’s prior test scores can be predicted by those of the student’s classmates. If students are randomly assigned to classes within a given school and grade, class indicators will provide very little information about students’ prior test scores. Using administrative data from the universe of public-school students in Texas between 2011 to 2019, we calculate two data-driven measures of tracking for grades 4 to 8 across math classes according to prior math scores. While the study is limited to one state, Texas accounts for 10% of the school-aged population of the United States and our data cover more than 1,200 school districts and 8,800 schools. 
• The extent to which public schools sort students by ability in Texas varies from school to school and by grade. Consistent with national survey findings, we find there is a greater degree of tracking in middle school than in elementary school grades. This is to be expected since sorting by ability will rise as students begin to take courses that are differentiated not only by level of difficulty and pace but also by subject content. The average middle school student in our sample is in a school that realizes about 35% of its potential to track students, but this ranges from no tracking at all to realizing almost 75% of the potential to track. In addition, we find that the factor that seems the most important determinant for the extent of tracking is the range of ability present among the students: School-grade cohorts of students that have a wider dispersion of test scores have more tracking than school-grade cohorts that are more homogeneous. Schools appear to operationalize tracking through the classification of students into categories such as gifted and disabled. (We observe that cohorts that are more tracked have higher shares classified as both gifted and disabled.) Greater math curricular differentiation is also associated with greater tracking.
• Contrary to popular perception, we find that there is a greater extent of sorting of students by ability within schools than across public schools in Texas. A popular perception in the United States is that, because school assignment is based primarily on residential location, much of the sorting takes place across districts and schools. However, in the Texas elementary and middle schools we analyzed, tracking by ability within schools overwhelms any sorting by ability that takes place across schools. In fact, only 10% and 17% of the variation in prior scores (within grade-years) is explained by districts and schools, while 44% is explained by classes. In addition, within the schools themselves, we find evidence of greater sorting of students based on prior test scores than of sorting based on race/ethnicity and socioeconomic status (as measured by eligibility for reduced-price or free school meals).
• What are the implications of tracking for students? To explore how tracking impacts test score growth for students, we mapped the students’ positions in the statewide test score distribution in 3rd grade to their positions in the test score distribution five years later. We find that for students with the lowest test scores, exposure to tracking is unrelated to future test score growth. For those initially at the top, however, exposure to more tracking is beneficial. For example, our results suggest that a one standard deviation increase in exposure to middle-school tracking would lead to a 1.3 percentile increase in predicted test scores 5 years after 3rd grade for students initially at the 75th percentile. We find that for students with the lowest test scores attending schools with a greater degree of tracking is associated with smaller average class sizes (which is not the case for those with higher initial test scores). These findings are consistent with tracking increasing inequality in educational outcomes, but primarily by benefiting those already at the top.