[Sorry for the formatting on this, and I’m too tired to fix it right now. Please click link to see original research. -CAR]
However, new results from a study of a large,
comprehensive dataset on US student achievement
seriously challenge assumptions
of private school superiority
overall, and find substantial
differences between different t
ypes of private schools. Ba
sed on the 2003 National Assessment
of Educational Progress (NAEP) mathematics ex
am, this analysis compares achievement in
public, charter, and different type
s of private schools. When co
mpared with other subjects (like
reading, for instance), math is more heavily in
fluenced by school than home experiences, so
studying math achievement provides
clearer insights into the rela
tive performance of different
types of schools. The 2003 NAEP samples are ove
r ten times larger than
in any previous NAEP
administration, providing achievement and student,
teacher, and administrator survey data on
over 190,000 4
th
graders (up from 13,855 in 2000) in
7485 schools, and more than 153,000 8
th
graders (up from 15,930 in 2000) in 6092 schools.
Earlier studies of charter schools based on
these data were disputed because researchers ha
d to rely on a web tool that did not allow for
simultaneous analysis of multiple student- and school
-level variables. This new analysis of the
complete raw data employs advanced statistica
l techniques (hierarchical
linear modeling) to
study the relationship between school type and
mathematics achievement while controlling for
demographic differences in the pop
ulations served by the schools.
Major Findings
Without controlling for student background differe
nces, private schools scored higher than non-
charter public schools, as would be expected.
However, this study examines these patterns
further, determining whether they are due simp
ly to the fact that
higher proportions of
disadvantaged students are enrolled
in public schools, and the exte
nt to which the gaps persist
3
after controlling for potential student- and
school-level confounding variables, including
measures of socioeconomic status, race/ethnicity,
gender, disability, limited English proficiency,
and school location.
Overall, the study demonstrates th
at demographic differences between
students in public and private
schools more than account for the relatively high raw scores of
private schools. Indeed, after controlling for these differences, the presumably advantageous
“private school effect” disappears, and even reverses in most cases.
