ERIC Identifier: ED448968
Publication Date: 2000-12-00
Author: Howley, Craig - Strange, Marty - Bickel, Robert
Source: ERIC Clearinghouse on Rural Education and Small Schools
Research about School Size and School Performance in
Impoverished Communities. ERIC Digest.
Many panels and experts have endorsed small schools as educationally
effective, often adding the parenthetical remark that smaller size is especially
beneficial for impoverished students. A recent series of studies, the "Matthew
Project," substantially strengthens the research base on school size and school
performance in impoverished communities, adding evidence to bolster these
claims. This Digest reviews recent thinking about small school size, describes
the aim of the Matthew Project studies, and summarizes findings. Discussion
concludes with a brief section on implications.
REPORTED BENEFITS OF SMALLER SCHOOL SIZE
mid-1990s, there have been efforts to summarize key findings of recent research
on school size. In 1994, Howley focused on influences related to achievement and
attainment (e.g., high school dropout rates), and noted evidence that smaller
size seemed to improve the performance of schools serving impoverished
communities. He also noted that several structural features of schooling had
been reported to bear on the issue of size: grade-span configuration (the number
of grades in a building), educational level (elementary vs. secondary), sector
(private vs. public), location (rural vs. urban), and curricular focus
(comprehensive vs. special purpose) (1).
Both Irmsher's and Raywid's research reviews, by contrast, summarized the
influence on a wider range of outcomes, with each author concluding that a
preponderance of evidence favored smaller size nearly universally. Raywid's
summary pays particular attention to the policy issue of how large "small"
should be. In brief, the upper limit of "small" had (as of 1997) been set at 350
for elementary schools and 900 for high schools.
Interestingly, Howley (1994) pointed out that studies based on "outcomes"
(e.g., achievement, completion rates, attendance) recommend smaller size than
those based on "inputs" (e.g., teacher salaries, instructional materials,
specialized staffing) and Raywid (1999) observes that studies based on the value
of "community" recommend sizes smaller than those based on "outcomes." Thus,
researchers and policy analysts who are most concerned with "community"
(Sergiovanni, 1994) will tend to recommend the smallest schools for nearly
everyone; those concerned with outcomes will advise small schools but only for a
portion of the population; and those most concerned with inputs will recommend
schools that are larger than those recommended by other researchers.
THE MATTHEW PROJECT
Schools grew dramatically larger during
the course of the twentieth century, and a huge professional literature that
addresses school size now exists. But a surprisingly small proportion of this
literature constitutes the "research" base on school size, and comparatively few
studies address the interaction of school size and poverty as a major concern.
Only 22 research reports--during the whole period from 1966 to 2000--define
school size, socioeconomic status (SES), and school-size issues as an important
focus of investigation. Within this literature, however, the studies related to
the Matthew Project are the only ones that pursue the issue systematically
across multiple replications. In fact, the summaries of the literature cited
above rely mostly on the early work in the Matthew Project in making the
judgement that smaller size is "especially" beneficial for at-risk students.
Lee and Smith's 1996 study of high school size is frequently cited in reviews
of the school-size literature, and their findings are consistent with those
reported in the Matthew Project. Typically poverty exerts a strong negative
influence on achievement. In schools with fewer than 301 students, they found
that influence was sharply reduced. They also found that aggregate achievement
(with all else being equal) was highest in high schools enrolling 601-900
students (grades 9-12). The Matthew Project studies take a somewhat different
view of optimal size, concluding instead that optimal size is contingent on
community SES, following the basic principle that the poorer the community, the
smaller the schools should be.
The Matthew Project (2) was inspired by the work of Noah Friedkin and Juan
Necochea (1988). Their study, carried out with California data describing school
performance at four grade levels, concluded that smaller school size benefitted
school performance in impoverished communities, but that larger size benefitted
school performance in affluent communities. Howley (1996) conducted the first
faithful replication of the 1988 study, using West Virginia data, with results
much like those obtained by Friedkin and Necochea. The Matthew Project
subsequently conducted four additional replications (Howley & Bickel, 1999).
Together the seven states in which related studies have been conducted (AK,(3)
CA, GA, OH, MT, TX, WV) reflect the range of schooling conditions in the United
States: ethnicity, locale, poverty, region, and school district organization.
This series of related studies is important because replication is one
time-honored method of validating research findings. Most studies in education,
in fact, have dubious validity because they are seldom replicated. For this
reason, these studies constitute an almost unique and particularly useful series
in the research literature on school size. It is the only series of studies that
focuses on the hypothesis that size mitigates the influence of poverty.
MATTHEW PROJECT FINDINGS
The Matthew Project investigated
the possible academic "excellence" and "equity" effects of school size in
Montana, Georgia, Ohio, and Texas at different levels of school socioeconomic
status (Howley & Bickel, 1999). Achievement test scores analyzed for each
state were those required by the state education agency (SEA). The studies
included virtually every school in each state studied.
Excellence was measured in terms of "effect size" at varying levels of school
socioeconomic status. Effect size is simply the ratio of an existing difference
in terms of standard deviation units. An effect size of +0.5 is a substantial
positive difference, meaning that larger school size would be predicted to
"increase" achievement. An effect size of -0.5 is a substantial negative
difference, meaning that larger school size would likely "decrease" achievement.
Equity was measured by dividing the schools in a given state into two
groups--smaller schools vs. larger schools divided at the median of size. Then,
the researchers computed the correlations between achievement and socioeconomic
status within the groups. In one study (Bickel & Howley, 2000), there were,
however, four groups: smaller schools within smaller districts, smaller schools
within larger districts, larger schools within smaller districts, and larger
schools within larger districts.
Excellence effects of size varied substantially by state, but like Friedkin
and Necochea (1988) and Howley (1996), the Matthew Project studies found that
the influence of size varied by SES level, with size exerting a negative
influence on achievement in impoverished schools, but a positive influence on
achievement in affluent schools. That is, all else equal, larger school size
benefits achievement in affluent communities, but it is detrimental in
impoverished communities. The practical import of the excellence findings can be
translated as follows (Howley & Bickel, 1999):
Between 41 and 90 percent of schools (depending on grade level tested) would
likely produce "lower" average scores if the schools were larger, or (in these
schools) higher scores if they were smaller. At the ninth grade level, 90
percent of schools are too big to maximize achievement. These schools serve 89
percent of Ohio's ninth graders.
Between 26 and 57 percent of schools (depending on grade level tested) would
likely produce lower average student scores if the schools were larger, or
higher scores if smaller. At the 10th grade level, 57 percent of the schools are
too big to maximize achievement. These schools serve almost half (46 percent) of
Texas' 10th graders.
Between 36 and 68 percent of the schools (depending on grade level tested)
would likely produce lower scores if the schools were larger, or higher scores
if the schools were smaller. At the eighth grade level, 52 percent of the
schools serving 48 percent of the students are too big to maximize achievement.
The percentage of schools "at risk" in this way (i.e., too large to maximize
achievement) is even greater at the elementary level.
In Montana there is only weak evidence that the effect of school size on the
average academic achievement of students depends on the level of poverty among
the students in the school. The effect was statistically significant only in
grade 4. Montana's unique results are probably due to the fact that its schools
are more uniformly small and income is more evenly distributed than in any of
the other three states.
While the excellence effects are substantial but varied in strength, the
"equity effects" are remarkably strong and consistent from state to state. They
show that in smaller schools, regardless of state, the relationship between
achievement and SES is substantially weaker in the smaller schools than in the
larger schools. The dividing line between "smaller" and "larger" in these
studies varies from state to state, as median size varies by grade level tested
and by state.
In all four states, smaller schools cut the variance in achievement
associated with SES by between 20 and 70 percent, and usually by 30-50 percent,
depending on grade level. In Georgia, Ohio, and Texas, smaller schools reduce
the negative effect of poverty on average student achievement in every grade
tested. In Montana, smaller schools significantly cut poverty's effect in two of
the three grades tested.
A great deal of effort has been
directed to breaking the bond between poverty and school achievement since the
late 1960s, but with depressingly little benefit. Yet, findings from the Matthew
Project clearly imply small schools help to thwart threats that poverty imposes
on school performance. Many state accountability systems, of course, monitor
school performance, though no SEAs have yet acknowledged that large size further
handicaps improvement efforts in impoverished communities. However, the Florida
legislature recently imposed upper limits on school size (to take effect two
years hence), and in California, the Gates Family Foundation has sponsored a
multimillion dollar initiative to create new small schools. According to the
Matthew Project findings, it would be important for such efforts to focus on
creating new small schools in impoverished communities
People who assert that "all schools be small schools" have other values in
mind than merely improving achievement; and providing smaller schools in very
affluent communities could well prove to be counterproductive in terms of
achievement. Even in affluent communities, however, schools serving 1,500 or
more students might have diseconomies of scale and bureaucratic operating modes
that are not educationally hospitable. Indeed, a wide consensus seems to have
emerged (cf. Fulton, 1996) that schools larger than 1,000 are unwise choices for
any community. The consensus clearly suggests that schools in impoverished
communities should be much, much smaller.
(1) For instance, schools of differing grade-span configuration but the same
enrollment are not really the same size in terms of their impacts on students;
the one with fewer grades is larger. Elementary schools are generally smaller
than high schools (with grade-span controlled), and private schools are
generally substantially smaller than public schools. Comprehensive high schools
are generally larger than special-purpose schools.
(2) Sponsored by the Rural School and Community Trust. For details about
findings and methodology consult http://www.ruraledu.org/matthew.html.%20
(3) This includes a study by Huang and Howley (1993).
Bickel, R., & Howley, C. (2000). The
influence of scale on school performance: A multi-level extension of the Matthew
principle. Education Policy Analysis Archives, 8(22).
Friedkin, N., & Necochea, J. (1988). School system size and performance:
A contingency perspective. Educational Evaluation and Policy Analysis, 10(3),
Fulton, M. (1996). The ABC's of investing in student performance. Denver, CO:
Education Commission of the States.
Howley, C. (1996). Compounding disadvantage: The effects of school and
district size on student achievement in West Virginia. Journal of Research in
Rural Education, 12(1), 25-32.
Howley, C. (1994). The academic effectiveness of small-scale schooling (an
update) (ERIC Digest). Charleston, WV: ERIC Clearinghouse on Rural Education and
Small Schools. (ERIC Document Reproduction Service No. ED 372 897)
Howley, C. B., & Bickel, R. (1999). The Matthew project: National report.
Randolph, VT: Rural Challenge Policy Program. (ERIC Document Reproduction
Service No. ED 433 174)
Huang, G., & Howley, C. (1993). Mitigating disadvantage: Effects of
small-scale schooling on student achievement in Alaska. Journal of Research in
Rural Education, 9(3), 137-49.
Irmsher, K. (1997). School Size (ERIC Digest). Eugene, OR: ERIC Clearinghouse
on Educational Management. (ERIC Document Reproduction Service No. ED 414 615)
Lee, V., & Smith, J. (1996). High school size: Which works best and for
whom? Educational Evaluation and Policy Analysis, 19(3), 205-27.
Raywid, M. A. (1999). Current Literature on Small Schools (ERIC Digest).
Charleston, WV: ERIC Clearinghouse on Rural Education and Small Schools, 1999.
(ERIC Document Reproduction Service No. ED 425 049)
Sergiovanni, T. (1994). Organizations or communities? Changing the metaphor
changes the theory. Educational Administration Quarterly, 30(2),214-26.