If there is one intervention in advanced education that has been studied extensively, it is acceleration—that is, allowing students who have mastered the material to move ahead a full grade or take a given subject (such as math or science) at a higher grade level, rather than keeping them in classes with age-based peers. In general, this research finds that academic acceleration (both grade skipping and subject acceleration) produces strong positive achievement effects for advanced students with no negative effects on non-accelerated peers. A new NBER working paper from David Card and Laura Giuliano—yes, the same duo who conducted groundbreaking research on universal screening—examines whether being placed in an advanced math track in middle school impacts students’ college and/or career trajectories.
They follow approximately 23,000 students attending an unnamed large urban district in Florida for more than a decade. The intervention is known as GEM (Great Explorations in Mathematics) and it compresses the standard sixth-grade math curriculum to make room for pre-algebra (typically a full-year course on its own), then moves students into Algebra I in seventh grade and Geometry in eighth. In other words, students are essentially moved two years ahead of their peers. Students who enter the track must maintain at least a B to continue in it. Card and Giuliano use district information linked to data from the National Student Clearinghouse, which provides them with postsecondary outcome data up to six years after high school graduation for sampled students, all of whom started fifth grade in the district between 2003–04 and 2006–07.
The researchers use a regression discontinuity design, taking advantage of the fact that eligibility for GEM is determined by whether a student scores at or above the 80th percentile on the math portion of the fifth-grade Florida Comprehensive Assessment Test (FCAT). In other words, students just above the cut were more likely to enter GEM in sixth grade than their peers just below the cut. However, while they can estimate causal effects by comparing outcomes within a narrow band around the cutoff, they can’t tell us how GEM would work for students who scored well below or well above it.
Interestingly, Card and Giuliano also seek to examine students’ non-academic traits (such as work ethic, conscientiousness, and persistence) that could enhance the benefits from GEM. They use students’ average GPA in non-math high school subjects (English, history, etc.) as a proxy, controlling for fifth-grade test scores in those subjects, and call it the “non-cognitive skill” index.
Lo and behold, it turns out that the index—rather than math ability—is the strongest predictor of who actually completes the GEM math sequence through eighth grade. Among students at the lowest end of the index’s distribution, almost nobody finishes; among those at the top, completion rates are about 60 percent. Since the program is a sustained commitment over multiple years where students have to maintain a B to stay in the track, success is about more than just mathematical ability.
In the longer term, GEM raised advanced math course-taking in high school for the overall sample, increased the probability of completing calculus (by 10–13 percentage points), and nudged PSAT math scores up slightly. Somewhat more surprising is that GEM had no overall effect on whether students enroll in college or finish a degree. The authors suggest this is likely because GEM admission criteria—and thus the sample—is limited to relatively high-achieving kids who were probably headed to college, regardless of whether they took Algebra I in seventh or ninth grade. Still, GEM did increase the likelihood of earning a STEM or business degree (weirdly, a combined category in this study) by about 6 percentage points, a roughly 25 percent increase over non-GEM students.
The effects vary a lot by gender and non-cognitive skill index score (a.k.a. GPA). For example, boys scoring high on the non-cognitive index see the biggest calculus enrollment gains (33 percentage points by eleventh grade) and the biggest bump in majoring in STEM (12 percentage points). Puzzlingly, girls with low non-cognitive scores—that is, strong in math but weaker in work habits or persistence—see the largest college gains of any group, with selective college entry, bachelor’s completion, and STEM/business degrees all running around 13 to 14 percentage points above comparable non-GEM girls. By contrast, boys who are low on the non-cognitive index show very few gains from GEM versus non-GEM boys. One possible explanation suggested by the authors: Being placed in the advanced math track may have changed what was expected of females and what girls expected of themselves.
In the end, it’s tricky to use GPA in non-math courses to represent non-cognitive ability. Clearly, it is not an isolated signal of persistence or work habits because it intersects with academic ability, not to mention other things, like how rigorously a teacher grades. Also, because GEM requires a B to stay in the program, GPA is also tied mechanically to those who remain in the track. So take the index with a grain of salt.
Yet that doesn’t take away the fact that targeted math acceleration has the potential to increase participation in advanced math classes in high school and the share of students who major in more math-intensive fields in college. Given the lack of acceleration taking place in K–12 education across the land, we need more programs and strategies like GEM to incentivize it.
SOURCE: David Card and Laura Giuliano, “Targeted Acceleration in Middle School Math: Impacts on College Entry, Degree Completion, and STEM,” NBER Working Papers (October 2026).