This research is sort of old, though. If I have time later, I may be able to find some newer meta-analyses.
Here is a more recent review.
http://www.ascd.org/ASCD/pdf/journals/ed_lead/el199904_loveless.pdfApril 1999 | Volume 56 | Number 7
Understanding Race, Class and Culture Pages 28-32
Will Tracking Reform Promote
Social Equity?
Although supporters of detracking believe that it leads to
greater social equity, we must carefully examine the
research, which suggests that tracking reform has potential
dangers.
Tom Loveless
...
One way to narrow the gap between high and low achievers is to boost low-ability students'
learning while either holding steady or lowering everyone else's. A study by Argys, Rees, and
Brewer suggests that detracking works in precisely this manner. The analysis focused on 10th
graders in the National Educational Longitudinal Study of 1988 (NELS) sample. When assigned
to heterogeneous math classes rather than to low tracks, low-ability 10th graders gained about
5 percentage points on achievement tests. Detracking helped them. Average students, in
contrast, lost 2 percentage points from detracking, and high-ability students lost even more,
about 5 points (Argys, Rees, & Brewer, 1996). The achievement gap was indeed narrowed, but
apparently at the expense of students in regular and high tracks, representing about 70
percent of 10th graders in the United States. Overall, achievement was approximately 2
percentage points lower in detracked schools.
...
Algebra
Research is unclear on whether tracking's effects vary by subject area. Several studies note
that math teachers are resistant to tracking reform (Loveless, 1994; Gamoran & Weinstein,
1998). The middle and high school math curriculum is usually organized hierarchically, with
progress through courses predicated on successful completion of prerequisites. Even within a
particular math course, topics build upon one another so that students acquire knowledge
sequentially. As a consequence, math teachers are intuitively skeptical of assigning students
who can't do basic arithmetic to the same classes as students ready to solve complex algebra
problems.
This intuition may be on the mark. Researchers at Johns Hopkins University analyzed NELS
data to find out what happened to the math achievement of 8th graders who were grouped in
different ways (Epstein & MacIver, 1992). Students in heterogeneously grouped algebra classes
didn't learn as much as students in tracked algebra classes. This held true for all ability levels—
high, average, and low. In contrast, when survey courses in math were heterogeneously
grouped, low-ability students benefitted. Tracking apparently doesn't affect all math courses
identically. This finding assumes added importance as the idea of all students taking an algebra
course at an earlier age gains popularity. If the finding is valid, then tracking reform may
seriously diminish the prospect that universal 8th grade algebra will boost U.S. math
achievement.
