Association for Public Policy Analysis & Management

In 2003, Chicago introduced Double-Dose Algebra, requiring additional support coursework for ninth-grade students with incoming math skills below the national median. Importantly, the policy also affected how schools assigned students to algebra classes, resulting in skill-based segregation on the basis of the cut-score. Consequently, students below the cutoff tended to take double-dose algebra with lower-skill peers, while classroom peer skills improved post-policy for those above the cutoff.  

Using a regression discontinuity design, prior studies showed overall positive impacts on both short- and long-run outcomes, including algebra scores, graduation, and college enrollment.  Another RD study examined the independent effects on algebra scores of the policy’s two mediators—doubled algebra instruction and classroom peer skills. This study revealed that Double Dose was most effective in schools with low level of skill-based segregation.  However, the results of these RD-based studies are not generalizable to the population away from the cutoff. Also, they are unable to identify the two mediators’ effects away from the cutoff. These are important limitations not only because policies such as Double Dose are typically aimed at a broad range of low-skill students.  Also, the policy is likely to affect overall achievement distribution through its unintended effects on high-achieving students through peer ability change.

To address these limitations, we use an interrupted time-serious design and estimate: 1) the overall policy impacts on low- and high-achieving students (those away from the cutoff) and 2) the independent impact of double-dose algebra and classroom peer ability on low-achieving students and that of classroom peer ability on high-achieving students.  Specifically using two pre- and two post-policy cohorts, we model the outcome for each site (i.e., school) as a function of time, capturing the outcome trend for school j, and estimate the school-specific policy effect defined as the discontinuity/interruption in the outcome at the time of policy implementation. We exploit between-school variability in policy-induced school-specific discontinuity in double-dose enrollment and classroom peer skills to identify the average independent impacts of these two mediators. 

We found that for low-skill students, double-dose enrollment increased by 80 percentage points and their classroom peer skills declined by -.14 SD post-policy (p<.001), on average. The average ITT effect on their 10^th-grade algebra scores was 0.30 SD. The effect of double-dose enrollment and that of peer skills on algebra scores were, respectively, 0.21 SD (p<.001) and 0.04 (n.s.). However, these effects differed by special education status. The ITT and double-dose effects on algebra scores were smaller for students with disabilities than regular-education students.  The positive effect of classroom peer ability was found for regular-education students, but not for students with disabilities.  For high-skill students, their classroom peer ability increased by 0.13 SD(p<.001). The average ITT effect on algebra scores was 0.23 SD, and the effect of classroom peer skills was .28 SD(p<.001).

This study shows that double-dose algebra benefited both low- and high-skill students away from the cutoff through different mechanisms.  Benefits of doubled instruction were diminished by declines in classroom peer ability, particularly for regular education students.