The Effectiveness of Title I: Synthesis of National-level Evidence from 1966 to 2013

Abstract

This paper reviews prior research to assess the effectiveness of Title I in closing the achievement gaps of disadvantaged students vis-à-vis their non-disadvantaged counterparts. A research synthesis approach is adopted to summarize national assessments of Title I conducted between 1966 and 2011. These analyses are supplemented by the authors' analysis of NAEP data from 1990 to 2013. There is no evidence that early Title I programs significantly reduced achievement gaps nationwide. Studies following NCLB implementation show modest closure of grade 4 gaps of about 0.2 of a standard deviation. Given the modest academic gains attributable to Title I, and considering that the program costs about $15 billion per year, the authors conclude that Title I programs have not been cost effective in closing the achievement gaps

Full text

Universal Journal of Educational Research 4(1): 205-311, 2016 http://www.hrpub.org
DOI: 10.13189/ujer.2016.040126
The Effectiveness of Title I: Synthesis of National-level
Evidence from 1966 to 2013
Sonia Sousa*, David Armor
School of Public Policy, George Mason University, USA
Copyright © 2016 by authors, all rights reserved. Authors agree that this article remains permanently open access under the
terms of the Creative Commons Attribution License 4.0 International License
Abstract This paper reviews prior research to assess the
effectiveness of Title I in closing the achievement gaps of
disadvantaged students vis-à-vis their non-disadvantaged
counterparts. A research synthesis approach is adopted to
summarize national assessments of Title I conducted
between 1966 and 2011. These analyses are supplemented
by the authors' analysis of NAEP data from 1990 to 2013.
There is no evidence that early Title I programs significantly
reduced achievement gaps nationwide. Studies following
NCLB implementation show modest closure of grade 4 gaps
of about 0.2 of a standard deviation. Given the modest
academic gains attributable to Title I, and considering that
the program costs about $15 billion per year, the authors
conclude that Title I programs have not been cost effective in
closing the achievement gaps.
Keywords Title I, NCLB, Education, Achievement
Gaps, Research Synthesis
1. Introduction
Established as part of the Elementary and Secondary
Education Act of 1965 (ESEA), Title I is a U.S.
compensatory education program that provides federal
financial assistance to elementary and secondary schools,
mostly public schools, with a high proportion of children
from low-income families. Since its inception in 1965, Title I
has been the largest single program in the U.S. Department
of Education, accounting for close to 40 percent of the
Department of Education’s total K-12 budget in recent years.
In 2012, its annual funding was about $14.5 billion, and it
reached over 23 million school children.
Title I was established with the original goal of improving
the educational attainment of children in poverty. The No
Child Left Behind Act of 2001 (NCLB) made this goal more
concrete by establishing a requirement to attain 100 percent
proficiency for all students by the 2013-14 school year. One
implication of the NCLB 100-percent-proficiency target is
that Title I aims to raise the academic achievement of
disadvantaged students to match non-disadvantaged students,
thereby closing achievement gaps by the 2013-14 school
year.
This paper seeks to synthetize the national-level evidence
on the effectiveness of the overall Title I compensatory
program in closing the achievement gap between poor and
non-poor students nationwide. The paper also discusses the
costs of that contribution.
1.1. Title I Characteristics
One key characteristic of Title I is that funds are allocated
to eligible schools based on the census estimates of
children’s poverty levels in the school district and the cost of
education in the state. The poverty threshold and the scope of
Title I programs have evolved over time. Currently, Title I
schools with at least 40 percent of children from low-income
families are eligible to use Title I funds, along with other
Federal, State and local funds, to put in place school-wide
assistance programs designed to improve academic
achievement of all students. Title I schools below the
40-percent threshold and those that choose not to operate a
school-wide program can use Title I to fund targeted
assistance programs for students who are failing, or at risk of
failing, to meet the state’s academic achievement standards.
Targeted assistance programs should be designed to meet the
needs of those students and should be developed in
consultation with parents, school staff, and district staff.
Title I program provides some guidelines but school districts
and schools have great flexibility to decide where and how to
focus the funds.
NCLB introduced some significant changes to the Title I
characteristics which went into effect in the 2002-03 school
year. Four of these changes are worth mentioning here given
their implications for the program’s goals, costs and
participation rates: (a) 100-percent state proficiency as the
academic achievement target, (b) school-specific
interventions, (c) teacher quality, and (d) parental choice.
First, NCLB established a clear achievement target: 100
percent state proficiency in core subjects for all students by
the 2013-14 school year. As noted above, this means that, in

206 The Effectiveness of Title I: Synthesis of National-level Evidence from 1966 to 2013
practice, Title I’s goal now goes beyond raising academic
achievement of poor students – it aims to close the
achievement gaps between poor and non-poor students by
2013-14. Second, schools and districts failing to make
adequate yearly progress towards the state proficiency
targets are identified as needing improvement and are subject
to specific interventions designed to improve their
performance and provide additional options to students.
Third, NCLB requires that all teachers (including Title I
teachers) of core academic subjects are highly qualified.
Fourth, parents of students attending continuously failing
schools must be given the option of obtaining supplemental
educational services from an approved public or private
provider chosen by the parents and funded by Title I.
1.2. Costs and Participation
Figure 1 shows the trends in Title I appropriations
(adjusted for inflation) as well as the number of students
served by Title I compensatory programs. The funds shown
cover only the Title I compensatory program and do not
include funds for implementing components and programs
of NCLB other than those funded by Title I and related
programs.
Between 1966 and 1990 funding fluctuated between $6
and $8.3 billion (in 2012 dollars), swinging to the lower end
during the 1980s. Funding increased in the early 1990s
reaching $10 billion by 1992 and steadied around this figure
until 2000. Funding grew again in the early 2000s exceeding
$14.5 billion by 2008. Allocations in connection with the
American Recovery and Reinvestment Act (ARRA) pushed
spending to $15.5 billion in 2009 and $15.3 in 2010,
returning to $14.5 billion in 2012.
Participation rates were not recorded during the early
years of Title I. Until 1980 there were only broad guidelines
on how the funds were to be spent, and individual student
participants were not identified. Starting with the 1981
reauthorization act, Title I recipients (either individuals or
whole schools) were counted as participants. Many students
were identified for special pull-out sessions that offered
more intense instruction or tutoring than was available in
regular classrooms. Student participation rates remained
fairly stable at about 5 million students from 1980 to 1995. It
rose steadily in the late 1990s and 2000s, reaching
approximately 20 million students by 2005 and 24 million in
2012. The main reason for the increase in the rate of
participation since the late 1990s was the reduction of the
threshold for school-wide programs. Before 1995, if a school
had 75 percent of its students below the poverty line, the
entire school enrollment would be counted as program
participants. That threshold was reduced to 60 percent in
1995-96, then to 50 percent, and it is currently at 40 percent.
Naturally, reducing the threshold has led to very large
increases in Title I enrollment, which currently stands at
nearly one-half of the national K-12 student population.
Per capita expenditures have been fairly low, particularly
after 1997. Prior to 1997, per capita funding (in 2012 dollars)
was about $1500 per student, but starting in 1997 per capita
funding dropped to $900, and it continued to drop to just over
$600 in 2012. This drop is due to lowering the threshold for
school-wide programs as the change counts more students as
participants.
Figure 1. Trends in Title I Appropriations and Participation [1,2]
$1,000,000,000
$2,000,000,000
$3,000,000,000
$4,000,000,000
$5,000,000,000
$6,000,000,000
$7,000,000,000
$8,000,000,000
$9,000,000,000
$10,000,000,000
$11,000,000,000
$12,000,000,000
$13,000,000,000
$14,000,000,000
$15,000,000,000
$16,000,000,000
0
5,000,000
10,000,000
15,000,000
20,000,000
25,000,000
30,000,000
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1975*
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
Appropriations, 2012 dollars
Enrollment
Participants
Appropriation, 2012 dollars

Universal Journal of Educational Research 4(1): 205-311, 2016 207
2. Materials and Methods
We conduct a research synthesis supplemented by an
analysis of more recent longitudinal national-level NAEP
data (1990–2013) to summarize the evidence on the outcome
measure of interest to this paper: national-level achievement
gap between poor and non-poor students. Research synthesis
is a process through which we integrate the quantitative
evidence provided by two or more research studies
concerning a particular question, but not necessarily using an
identical outcome measure across studies. A research
synthesis lies between a literature review, which describes
the authors’ findings without presenting the quantitative
results in a systematic manner, and a meta-analysis, which
employs statistical methods to synthetize quantitative
evidence in the form of an overall estimate of the
effectiveness of a program/intervention. A research synthesis
is particularly well suited to synthetize quantitative evidence
based on a small number of studies and/or studies that use
incomparable outcome measures [3]. We adopt a research
synthesis approach, as opposed to a meta-analysis, because
the disparate methodologies used to assess the effectiveness
of Title I nationwide over the years makes it very difficult to
produce an overall quantitative estimate of the effects of
Title I on the achievement of below-poverty line students.
A rigorous assessment of Title I effectiveness would have
to meet What Works Clearinghouse standards as set by the
Institute for Education Sciences of the U.S. Department of
Education. This would require selecting only studies
employing randomized controlled trials (RCT) or true
experimental designs. We did not apply this criterion
because such rigorous evaluation studies have never been
conducted for the overall Title I compensatory program
nationwide, despite the fact that the Congress has
specifically mandated three National Assessments of Title I
since the late 1980s (more on this below). The most likely
explanation for the absence of randomized designs is the
near universal implementation of Title I, which makes it
virtually impossible to find a randomly assigned control
group that does not receive Title I services. A RCT study
would require to randomly assign classrooms, schools, or
school districts to Title I or non-Title I conditions. Given that
Title I is a long standing nearly universal program, most
would find ethical, political or legal objections to depriving
eligible poor children from receiving Title I services [4,5].
2.1. Search Procedure and Criteria
We identified prior evaluations of Title I effectiveness
using Google, Google scholar and ProQuest search engines,
covering the period from 1970 to 2015. Depending on the
engine search capability we employed the following
keywords alone or in combination: Title I, no child left
behind, impact, academic achievement. We scanned the title
and abstract/executive summary to select only studies within
the scope of this research synthesis, which is defined by the
following four criteria: (a) evaluations of Title I/NCLB as a
whole, as opposed to evaluations of specific components; (b)
evaluations of Title I/NCLB that are nationally
representative, as opposed to assessments of state- or
local-level effects of Title I; (c) evaluations that assess the
effect of Title I/NCLB on students’ academic achievement,
as opposed to, for example, effects of Title I/NCLB on
school spending, and (d) evaluations that look at students
from disadvantaged poverty backgrounds as compared to
their non-disadvantaged poverty counterparts. We then
scanned the reference sections of the studies selected to
explore other potential studies that would fit within the scope
of this synthesis.
Studies assessing specific components of Title I, such as
the impacts of remedial reading programs in [6] do not meet
the requirements of this synthesis. Studies that assess the
impact of specific NCLB provisions, as for example, [7-10]
that access school choice, supplemental educational services
options, teacher quality and other accountability provisions
are also outside the scope of this study. State- or local-level
evaluations of Title I, such as [11-13] are also out of the
scope of this synthesis. The study by Cascio and
colleagues[14] is outside the scope because it addresses only
the South and the outcomes studied are school spending and
dropout rates. Studies that look at comparisons between
groups defined by features other than poverty backgrounds
are also outside of the scope of this synthesis. This is the case,
for example of [15], which compares public vs non-public
schools and high-standards vs non-high standards states.
2.2. Data
Our search yielded five peer reviewed studies and reports
by the U.S. Department of Education that meet our data
selection criteria stated above. These studies are: (a) the
Borman and D’Agostino meta-analysis [16], which
synthetizes evaluation studies conducted from 1966 to 1993;
(b) the Prospects Study [17], which uses a large national
sample of students and covers the 1991 – 1994 period; (c) the
1999 National Assessment of Title I and its follow-up studies
[18-19], using national data from late 1980s to late 1990s; (d)
the 2007 National Assessment of Title I and its follow-up
studies [6,7,20,21], which cover the period from 1992 to
2007, and (e) the recent study by Dee and Jacob [22], which
also analyses the period from 1992 to 2007. We supplement
these sources with our own trend analyses of longitudinal
national-level NAEP data from 2007 to 2013 to bring the
academic achievement data up to date. Taken together, this
research synthesis covers the period from 1966 to 2013.
3. Results
3.1. The Borman and D’Agostino Meta-analysis

208 The Effectiveness of Title I: Synthesis of National-level Evidence from 1966 to 2013
Figure 2. Mean Standardized Effect Sizes of Title I Participation from the Borman and D’Agostino Meta-Analysis (period covered: 1966 – 1993)
The Borman and D’Agostino meta-analysis [16]
synthetizes 17 federal Title I evaluations conducted from
1966 to 1993. The outcome measure is the achievement gap
between Title I participants and a control group. Figure 2
presents the mean standardized effect size, broken down by
subject (math and reading) and by grade. Only studies
employing a two-wave, pretest/posttest design [23] are
included in the meta-analysis and reflected in the effect size
of Figure 2. The standardized effect sizes represent the
achievement gaps between Title I participants and
non-participants as a fraction of one standard deviation (sd).
The results of this meta-analysis suggest three major
findings. First, the overall effect of Title I over the 1966 to
1993 period is positive but modest in magnitude at only 0.11
sd. Second, for grades 1 to 6, Title I seems to have a stronger
effect in math than in reading over this period. The effect
sizes of Title I on students’ achievement in math range from
0.21 to 0.26 sd in grades 1 to 6, whereas the effect sizes for
reading range from -0.01 to 0.11 sd. Third, the effect sizes of
Title I in math decline significantly after grade 6 and
resembles the effects in reading. Specifically, from grade 7
through grade 12, the effects of Title I in math decline to
values ranging from 0.08 to 0.14 sd, while the effects in
reading remain relatively similar to those in elementary
grades, ranging from 0.08 to 0.11 sd. In short, the Borman
and D’Agostino meta-analysis suggests that the effect of
Title I is stronger in math programs and during the student’s
elementary grades, although even here the effect is modest at
about 0.2 sd.
3.2. The Prospects Study
The Prospects study [17] relies on a nationally
representative sample of about 40,000 students from 364
schools, from grades 1, 3 and 7, from 1991 to 1994. Using
the Comprehensive Test of Basic Skills (CTBS) to assess
students’ academic achievement, the study estimates the
effect of Title I by means of multivariate statistical
techniques (namely, hierarchical linear models), to control
for differences between Title I participants and
nonparticipants on a set of student, family, and school
characteristics.
Table1 summarizes the effects broken down by number of
years exposed to Title I, cohort, and subject area – math and
reading. For each cohort and duration of exposure to Title I,
the Prospects study estimates two related outcome measures:
(a) achievement gaps, or the achievement score differences
between Title I participants and nonparticipants and (b)
changes in the achievement gap between participants and
nonparticipants over the 1991 – 1994 period. We report the
significant results of both outcome measures as standardized
effect sizes. We obtained the standardized effects by
dividing the scale-score points by the standard deviation of
45 scale-score points, the mid-point of the standard deviation
range reported in the study. To illustrate for the 1st grade
cohort in math, -0.64 sd means that students in 1st grade
exposed to Title I for one year score, on average,
approximately two-thirds of a standard deviation in math
below those who are not exposed to Title I. The entry for
over-time gap change means that the 1st grade math score
gap of -0.64 sd does not significantly change over the 1991 –
1994 period.
-0.10
-0.05
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0.40
1 2 3 4 5 6 7 8 9 10 11 12
Mean Standardized Effect Size (sd)
Grades
Math
Reading
Overall effect
size: 0.11sd
Overall effect
size: 0.2sd

Universal Journal of Educational Research 4(1): 205-311, 2016 209
Tab le 1. Summary of Effects of Title I Participation Detected by the Prospects Study (Period Covered: 1991 – 1994)
Academic Subject and Years of
exposure to Title I
1st Grade 3rd Grade
Achievement gaps Over-time gap change Achievement gaps Over-time gap change
Math (CTBS standard test)
1 year -0.64 not significant -0.49 not significant
2 years -0.62 not significant -0.82 not significant
3 years -0.60 not significant -0.64 not significant
4 years not applicable not applicable -1.02 0.11
Reading (CTBS standard test)
1 year -0.62 not significant -0.38 not significant
2 years -1.13 not significant -0.53 -0.10
3 years -1.38 not significant -0.44 -0.08
4 years not applicable not applicable -0.76 not significant
The results of the nationally representative Prospects
study suggest three major findings. First, students exposed to
Title I between 1991 and 1994 score, on average,
considerably lower than students who did not receive Title I
assistance. In both academic subjects and cohorts, participant
students show moderate to large1 negative achievement gaps
(vis-à-vis nonparticipant students), on average, after being
exposed to Title I for a minimum of one year and up to four
years. For example, after three years of Title I exposure, the
1st-grade cohort scores, on average, 0.6 sd lower in math and
1.4 sd lower in reading compared to scores of students never
exposed to Title I. For the 3rd-grade cohort, after 3 years of
Title I exposure, on average, participant students score below
nonparticipant students by 0.64 sd in math and by 0.44 sd in
reading. Second, the negative achievement gaps are wider
among students who have more years of Title I exposure than
among students who have less years of Title I exposure.
Third-graders with four years of Title I assistance show very
large achievement gaps (vis-à-vis nonparticipants) in both
math (-1 sd) and reading (-0.8 sd); whereas the negative
achievement gaps after one year of Title I exposure are
moderate to large in both math (-0.5 sd) and reading (-0.4 sd).
Third, the negative achievement gap between participants
and nonparticipants remain relatively unchanged over time
between 1991 and 1994. The over-time gap changes
(between participants and nonparticipants) are either
insignificant or where there are significant changes they tend
to increase the disadvantage of participants (negative effect
size). In only one case the over-time gap change is
significant and positive, albeit small (0.1 sd), and this is the
effect of Title I on math for 3rd-graders after four years of
Title I exposure. This positive over-time gap change of 0.1 sd
indicates that by the time 3rd grade participants completed 6th
grade, they had gained only about one-tenth of a standard
deviation more in math achievement than nonparticipant
students.
1 Following [17], we classify the magnitude of the effect as: (a) small, if it is
below 0.22 sd or 10 scale-score points; (b) moderate, if it is between 0.22 sd
and 0.56 sd (or between 10 and 25 scale-score points); and (c) large, if it is
over 0.56 sd or 25 scale-score points.
3.3. The 1999 National Assessment and Follow-up
Studies
The 1999 National Assessment of Title I was mandated by
the Congress as part of the 1994 reauthorization of ESEA
[18]. A few follow-up studies conducted between 1999 and
2001 are summarized in [19]. Unlike the Prospects study, the
1999 National Assessment and its follow-up studies do not
compare the academic achievement of Title I participants
versus nonparticipants. Rather, they conduct a trend analysis
of national-level NAEP achievement test scores in reading
and math from 1986 to 1999 for subgroups of students from
disadvantaged backgrounds and thereby likely to benefit
from Title I, as compared to those of non-disadvantaged
counterparts. These groups are: (a) 9-year-old students in
high-poverty schools (schools were 75 percent or more
students receive free- or reduced-price lunches), from 1986
to 1999 and (b) low-achieving (below the 10th percentile)
4th-grade students, from 1992 to 1999. These subgroups are
likely to benefit from Title I because Title I is designed to
support schools with high concentration of poverty,
particularly students in greatest risk of failing, and because
most Title I funds serve elementary schools. NAEP scores
provide a uniform basis for comparing achievement progress
nationwide.
One limitation of this trend analysis approach is that it is
hard to claim that achievement progress, even for those
students likely to receive Title I, are in fact attributable to
Title I. Although Title I is the largest single federal
educational program, Title I accounts for only about 3
percent of total resources invested in elementary and
secondary education by federal, state, and local authorities
combined. However, the expansion of school-wide programs
funded by Title I, which started in 1996, blurred the
distinction between program participants and
nonparticipants making it hard to find approaches that
contrast these two groups [18].

300 The Effectiveness of Title I: Synthesis of National-level Evidence from 1966 to 2013
Reading Mathematics
A) Average scale scores on the Trend NAEP of 9-year-old public school students, by poverty level
Reading Mathematics
B) Average scale scores on the Main NAEP of 4th-graders public school students, by performance percentile
Figure 3. Performance on NAEP for Students Most Likely to Benefit from Title I Services, 1990 to 1999
The results of the 1999 National Assessment are
summarized in Figure 3. Our main outcome measure of
interest is the achievement gap between disadvantaged
groups, measured in this study as students in low-poverty
schools (Figure 3A) and students in the 10th achievement
percentile (Figure 3B), and their non-disadvantaged
counterparts.
The data required for conducting significance tests and
calculating standardized effect sizes is neither reported nor
available at the NAEP data portal [24]. Still, achievement
score-point trends indicate that during the period from 1988
to 1999, students from disadvantaged backgrounds fail to
reduce their achievement gaps vis-à-vis their
non-disadvantaged counterparts. Two major findings
support this overall assessment. First, over the decade from
late 1980s to late 1990s the achievement gaps widen for both
groups of disadvantaged students, as compared to their
non-disadvantaged counterparts and it does so in both
reading and math. The gap between highest- and
lowest-poverty schools for 9-year-old students, as measured
by average NAEP scores, widens from a 27-point gap in
1988 to a 40-point gap in 1999 in reading; and from a
20-point gap in 1986 to a 29-point gap in 1999 in math. The
performance gap for the lowest-performing 4th -graders as
compared to the overall average widens from 47 to 50 points
between 1992 and 1996 in reading and from 42 to 44 points
from 1990 to 1996 in math. Second, achievement gaps in the
late 1990s are substantial, equivalent to several grade levels.
190 189
180 184 188 186
217 221 220 220 224 226
211 209 209 209 210 210
120
140
160
180
200
220
240
260
1988 1990 1992 1994 1996 1999
NAEP mean scale scores
Highest-poverty school
Low-poverty schools
All schools
208 213 208
215 217 212
228
237 236 239 238 241
219
230 228 229 230 231
120
140
160
180
200
220
240
260
1986 1990 1992 1994 1996 1999
NAEP mean scale scores
Highest-poverty school
Low-poverty schools
All schools
168
156
165
192 187 192
215 212 215
120
140
160
180
200
220
240
260
1992 1994 1996
NAEP mean scale scores
10th percentile 25th percentile
Average
171 176 180
193 197 201
213
220 224
120
140
160
180
200
220
240
260
1990 1992 1996
NAEP mean scale scores
10th percentile 25th percentile
Average

Universal Journal of Educational Research 4(1): 205-311, 2016 301
A 10-point difference in NAEP scale scores is considered
roughly equivalent to one grade level [19]. In the late 1990s
9-year-old students in the highest-poverty schools have an
average gap vis-à-vis the lowest-poverty schools equivalent
to four grade levels in reading (40-point gap) and nearly
three grade levels in math (29-point gap). Similarly, the
lowest-performing 4th -graders have a gap vis-à-vis the
overall average equivalent to five grade levels in reading
(50-point gap) and four grade levels in math (44-point gap).
3.4. The 2007 National Assessment and Follow-up
Studies
The 2007 National Assessment of Title I was mandated by
Congress as part of the 2001 NCLB Act and commissioned
by the Institute of Education Sciences in the Department of
Education [6,7,20]. A set of follow-up evaluation studies
conducted since 2007 are summarized in a report released in
2009 and conducted by the Policy and Program Studies
Service in the Department of Education [21]. The 2007
National Assessment analyses national-level NAEP trends
for groups of disadvantaged students considered as the most
likely beneficiaries of Title I under the 2001 NCLB mandate.
This methodological approach is similar to that followed by
the 1999 National Assessment. The two national assessments
vary slightly in the groups of disadvantaged students they
consider. Like its predecessor study, the 2007 National
Assessment also evaluates students in highest-poverty
schools, but unlike its predecessor it includes racial and
ethnic minorities, specifically black and Hispanic students,
as opposed to low-achieving students.
As with the 1999 study, the major limitation of the 2007
National Assessment is that the achievement trend data do
not isolate the impact of Title I. Rather it measures the effect
of Title I and the entire educational system on students’
achievement. Simple trend analyses such as the ones done in
this study cannot separate the effects of Title I from the
effects of other state and local improvement programs,
demographic changes, and other factors that may affect
student achievement trends.
Figures 4 to 6 summarize the national-level trends in
NAEP test scores for 4th- and 8th-graders in reading and math,
by race/ethnicity and by school poverty level, from 1990 to
2007. We do not include the NAEP scores in science in our
analysis because there are only three data points: 1996, 2000
and 2005. We draw on these NAEP trends to assess the
effects of Title I on students’ achievement according to our
outcome measure of interest: the achievement gaps between
disadvantaged groups, measured in this case by students in
low-poverty schools and students from racial/ethnical
minorities, and their non-disadvantaged counterparts. We
report gaps in scale-score points, obtained directly from the
figures, and also as effect sizes, using the corresponding
standard deviations obtained from the NAEP data portal
[24].
The results by race/ethnicity in Figure 4 show that in
reading the black-white gap in 4th grade reduces by 5 score
points or 0.13 sd, from 32 points (0.93 sd) in 1992 to 27
points (0.8 sd). Most of the improvement occurs in a single
interval, between 2000 and 2002. The reduction in the
black-white gap in 8th grade reading is smaller, from 29
points (0.89 sd) to 26 points (0.8 sd); a closure of 3 points or
0.09 sd. The change in the Hispanic-white gaps in reading is
similar in magnitude to the black-white gap in both grades.
In math, there is nearly continuous improvement between
1990 and 2007 for all groups in both grades, although the
gains have been greater in 4th grade. The 4th grade math
black-white gap narrows by 6 points or 0.12 sd, from 32
points (1.1 sd) to 26 points (0.98 sd), while the 8th grade
math gap narrows slightly (2 points; less than 0.01 sd) from
33 points (0.965 sd) to 31 points (0.964 sd). The
Hispanic-white math gap widens slightly for 4th graders (by
only 1 score point; 0.07 sd) and also for 8th graders (by just 2
score points; 0.08 sd).
Figure 5 shows an unfavorable pattern when the
highest-poverty schools are compared to the lowest-poverty
schools. Between 1992 and 2007, the poverty gap in reading
widens slightly for both 4th graders (by 2 score points; 0.04
sd) and 8th graders (by 3 score points; 0.08 sd). In math, the
poverty gap widens 7 score points (0.24 sd) for 4th graders
(from 1990 to 2007) and 20 points (0.59 sd) for 8th graders
(from 1990 to 2005). Despite the improvement in math
scores for both groups and both grades over the 1990 – 2007
period, the gains are greater for low-poverty schools than for
high-poverty schools, which results in the widening of the
poverty gaps at the school level.
When we consider the percentage of students at or above
the NAEP proficiency level between 1992 and 2007, shown
in Figure 6, we find that the improvement of white students is
greater than both black and Hispanic students in both
subjects and in both grades. In reading, between 1992 and
2007, black-white gap widens by 3 percentage points in 4th
grade and by one percentage point for 8th graders. The
widening of the Hispanic-white reading gaps is similar in
magnitude. In math, there is a continuous improvement in
the percentage of students performing at or above the NAEP
proficiency level between 1990 and 2007, for all groups and
in both grades. However, whites improved at a faster rate and
as a result the gaps widens substantially. The black-white
math gap widens 22 percentage points for 4th graders and 17
percentage points for 8th graders. The Hispanic-white gap
widens by 18 percentage points in 4th -grade and by 15
percentage points in 8th grade.

302 The Effectiveness of Title I: Synthesis of National-level Evidence from 1966 to 2013
* p < .05
Figure 4. Performance on main NAEP for Public School Students by Race and Ethnicity, 1990 to 2007

Universal Journal of Educational Research 4(1): 205-311, 2016 303
* p < .05
Figure 5. Performance on main NAEP for Public School Students by School Poverty Level, 1990 to 2007

304 The Effectiveness of Title I: Synthesis of National-level Evidence from 1966 to 2013
* p < .05
Figure 6. Students at or above the NAEP Proficiency Level by Race and Ethnicity, 1990 to 2007
In short, the 2007 National Assessment of Title I suggests
that Title I was largely ineffective in closing the achievement
gap of students from disadvantaged backgrounds over the
period from the early 1990s to the late 2000s. As measured
by overall scale scores, there is a modest closure of the
black-white and Hispanic-white gaps in the 4th grade over
this period, but very little change in the 8th grade gaps. There
is no closure in the achievement gaps between high-poverty
schools and more affluent schools at either grade over the
same period. The black-white gap closure for 4th graders
represents a standardized effect of only about 0.13 sd in
reading and 0.12 sd in math, and these are achieved mostly in
a single interval – between 2000 and 2002. As measured by
proficiency rates, the black-white and Hispanic-white gaps
show little change in reading, but the math gaps worsen
substantially. This suggests that the improvements in
minority math scores gap in the 4th grade might be mainly at
the lowest end of the performance continuum, raising some
of the lower scores to some degree but not sufficient to cross
the proficiency threshold. Part of these improvements for the
lowest-performing students might have been due to special
accommodations in testing administrations, which were
adopted in the late 1990s and whose use increased markedly
between 2000 and 2003.
3.5. The Dee and Jacob Study
The study by Dee and Jacob [22] analyzes state-level

Universal Journal of Educational Research 4(1): 205-311, 2016 305
NAEP panel data from 1992 to 2007 to determine whether
NCLB has influenced student achievement nationally. They
use a comparative interrupted time series (CITS) approach
(also known as an interrupted time series with a
non-equivalent control group) to contrast states with and
without school accountability policies in place prior to
NCLB. Their outcome measure is over-time achievement
gaps between states without (treatment states) and with
(control states) pre-NCLB accountability policies.
Specifically, the study compares the deviation from prior
trend for the states that were arguably affected by NCLB
provisions (those without pre-NCLB accountability policies;
the treatment group) with the analogous deviation for states
that were less affected by NCLB (those with pre-NCLB
accountability policies; the control group). The rationale is
that the deviations from prior achievement trends within
states with pre-NCLB accountability policies (control states)
provide a good counterfactual for what would have happened
in states without pre-NCLB policies (treatment states) if
NCLB had not been implemented.
One advantage of this study is that the panel-based
research design allows distinguishing the effect of NCLB
from the effects of other state and local educational changes,
as well as other social and economic changes that took place
during the period of analysis. This is a limitation of both the
1999 and 2007 National Assessment studies. One limitation
of the study by Dee and Jacob is that it has data for only a
subset of states ranging from 19 to 39 depending on the
subgroup (see Table 2). This raises issues about the
representativeness of the results nationally. Some of the
states missing from this analysis have very large student (and
minority) populations, including Florida, Illinois, New
Jersey, and Pennsylvania. The number of states considered in
the analysis is particularly small for Hispanics (ranging from
16 to 22 states depending on the grade-subject) and for
blacks (ranging from 27 to 32 states), which raises concerns
about the representativeness of the results for these
subgroups with respect to the nation as a whole. Table 2
summarizes the estimated effects of NCLB on NAEP scores
for 4th- and 8th-grade math and 4th-grade reading. The effects
are available at the aggregated school level and by race,
free-lunch eligibility and proficiency level.
The results at the aggregate school level show that by 2007
NCLB has a moderate positive effect for 4th-grade math (7.2
score points above control states) but smaller and statistically
insignificant effects for 8th-grath math (3.7 points) and
4th-grade reading (2.3 points). Since one of the primary
objectives of NCLB is to reduce the achievement differences
by race and socioeconomic status, we are primarily
interested in the effects of Title I/NCLB on subgroups of
students from disadvantaged backgrounds and not so much
on the effects at the aggregate school level. However, given
that the analysis by Dee and Jacob for subgroups of students
does not cover the same group of states, we can only interpret
them in terms of effects on a particular subgroup but we
cannot infer what the results mean with respect to changes in
gaps between disadvantaged and non-disadvantaged groups
of students (which is our outcome measure of interest). We
address this limitation by complementing Dee and Jacob
analysis with our own analysis for subgroups of students
covering the same group of states (more below). Dee and
Jacob results show moderate effects by race/ethnicity and
greater for blacks and Hispanics than for whites for 4th-grade
math; somewhat smaller effects for 8th-grade math, greater
for blacks but of a similar magnitude for whites and
Hispanics; and a combination of significant effects for
whites but insignificant effects for blacks and Hispanics in
4th-grade reading.
Tab le 2. Summary of Estimated Effects of NCLB in Dee and Jacob Study
4th-grade math scores 8th-grade math scores 4th-grade reading scores
Aggregate school level 7.244** 3.704 2.297
Number of states 39 38 37
Race
White 4.855** 1.828 5.362**
Number of states 39 38 37
Black 14.573** 8.826 -0.871
Number of states 30 27 32
Hispanic 9.793** 8.219** 0.242
Number of states 19 16 22
Free-lunch eligibility
Eligible 8.011** 15.761** 2.482
Not eligible 1.385 0.992 -4.79
Number of states 36 34 37
Proficiency level
10th percentile 9.046** 5.598** 3.611
90th percentile 5.205** 2.537 2.097**
Number of states 39 38 37
*p < .05. ** p < .01.

306 The Effectiveness of Title I: Synthesis of National-level Evidence from 1966 to 2013
A) Black-White gap, Grade 4 math
B) Free lunch eligibility gap, Grade 4 math
Figure 7. Gaps for grade 4 math by pre-NCLB Accountability Status.
Being particularly interested in the gap between whites
and other minority groups, we conducted a further analysis
of NAEP scores showing changes in the 4th grade
black-white gap in math, comparing the group of 39 states in
the Dee and Jacob study with all 50 states. These results are
shown in Figure 7, where “selected states” are the 39 states
with NAEP testing in 2000 and at least two scores between
1992 and 2000. The “All States” group includes the 11 states
not used in the Dee & Jacob analysis, which include
Colorado, Florida, Illinois, New Jersey, Pennsylvania,
Washington, Oregon (black scores only) and four other
states. “Free lunch eligibility gap” is the gap between those
eligible and not eligible for free/reduced lunch meals.
Both the 39 selected states in the Dee and Jacob study and
the total population of 50 states show that the black-white
achievement gap in 4th grade math declined appreciably
0
5
10
15
20
25
30
35
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
Black-White Gap, 4th Grade NAEP Scale Scores
Selected States without Pre-NCLB Accountability
Selected States with Pre-NCLB Accountability
All States without Pre-NCLB Accountability
All States with Pre-NCLB Accountability
0
5
10
15
20
25
30
35
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
Gap in NAEP Scale Score Points
All States without Pre-NCLB Accountability
All States with Pre-NCLB Accountability

Universal Journal of Educational Research 4(1): 205-311, 2016 307
between 1996 and 2003, a time period that encompasses
most of the state and federal policy changes in accountability
(Figure 7A). The 4th grade math gap stood at about 33 points
in 1996, and fell in two steps to about 27 points by 2003, a
standardized effect size of about 0.2 sd. We note that the
accountability states falling more sharply between 1996 and
2000 than the non-accountability states. However, the gaps
changed very little after 2003, dropping to about 25 points
and remaining there until 2013. States without pre-NCLB
accountability drift upwards slightly, particularly those in the
selected states without scores in the year 2000. There are
some notable differences in the gaps when comparing the 39
states selected by Dee and Jacob versus all 50 states,
particularly when looking at the most recent NAEP scores in
2011 and 2013 (Dee and Jacob examined NAEP scores
through 2007). The selected samples of states without
pre-NCLB accountability trend upwards in 2013, while all
states remain relatively flat. Using data from all states, the
gap in the no-accountability states is only two points higher
than that in the accountability states (27 vs. 25); just one
point higher than it was in 1992 (34 vs. 33). Thus NCLB
appears to have reduced the black-white achievement gap
during the early years of implementation, but the gap has
remained stubbornly stable for the past 10 years or so.
Looking at Dee and Jacob results by poverty subgroups,
the authors found a major effect of accountability for
students in poverty, as indicated by free lunch status. Again,
only 36 states were used for this analysis, because there were
only two NAEP assessments before 2003 with information
on free lunch eligibility (1996 and 2000), and 14 states were
missing scores for one or both of these assessments (see
Table 2). However, our own analysis for all states by
pre-NCLB accountability status (Figure 7B) shows that the
missing states made less of a difference on the poverty gap
than it did for the black-white gap; no difference was greater
than a single scale score point. Figure 7B shows the 4th grade
math gap between those eligible for free/reduced lunch and
those not eligible, and the gap is compared for all states
without pre-NCLB accountability policies to all states with
accountability. The reduction in the gap attributed to NCLB
is more modest than in the black-white gap, and it is just
slightly greater for the accountability states. The gap for the
accountability states was 27 points in 1996, falling to 22
points by 2003, versus a reduction from 24 to 22 for the
non-accountability states. 2 Like the black-white gap, the
eligible vs. non-eligible gap remained constant for the next
10 years.
2 There are no NAEP math scores by free/reduced lunch status prior to 1996,
so it is possible that the gap was larger in the early 1990's. Additionally,
there are no reading scores prior to 1998.
In short, the study by Dee and Jacob suggests that the
NCLB policies reduced achievement gaps modestly for 4th
grade math but much less for 8th grade math and reading.
Perhaps most important, the NCLB has definitely not met its
objective of eliminating achievement gaps between white
and minority students, even though both white and minority
students have experienced increasing achievement levels in
math.
3.6. Supplemental Analysis: NAEP Trends Until 2013
We supplement the information from the major evaluation
studies with additional trend analyses of NAEP test scores
from 1992 to 2013 to bring the achievement trends shown in
the 2007 National Assessment up to date. For the poverty
trends, we shift to students’ poverty status rather than school
poverty status (Figure 8). We also update the trend data by
race/ethnicity groups, which are shown Figure 9. We used
data from [24] in both Figure 8 and Figure 9.
Figure 8 shows the trends in NAEP reading and math
scores by students’ individual poverty status. For reading,
between 1998 and 2013 4th grade scores rise modestly by 10
points for non-poverty students in a nearly linear fashion,
and they also rise by 9 points for free/reduced lunch students
but most of the gain was in the first half of the period. Thus
the reading gap between poverty and non-poverty 4th
graders has remained constant for 15 years. Reading also
shows overall improvement for 8th graders, although it is
slightly less at 9 points for non-poverty students and 8 points
for free/reduced students. The pattern differs, somewhat, in
that scores were relatively flat for both groups until 2007, so
most of the increases for 8th grade reading have come
between 2007 and 2013. Like 4th graders, the reading gap for
8th graders has not closed over this 15 year period. For math,
the gains have been much greater, and they continue rising
for both groups and both grades after 2007. For 4th graders,
the total increase between 1996 and 2013 is 23 points for
both poverty and non-poverty students, but most of the
increase occurred between 1996 and 2007, particularly for
the poverty students. NCLB might have been responsible for
the especially steep increase of 12 points for poverty students
between 2000 and 2003. For 8th graders, math scores rise 18
points for both groups in a nearly linear fashion. Like 4th
graders, the math gap between poverty and non-poverty
students remains constant over this 17 year period.

308 The Effectiveness of Title I: Synthesis of National-level Evidence from 1966 to 2013
Figure 8. NAEP Reading and Math Scores for Public School Students by Poverty Status (Free Lunch Eligibility), 1996 to 2013
Figure 9 shows reading and math trends by race and
ethnicity. Not surprisingly, like the poverty trends these
results also reveal a generally upward trend for all groups
starting in 1992 for reading and 1990 for math. For 4th grade
reading, the trend for white students is flatter than for black
or Hispanic students, increasing just 8 points over the 21 year
period – less than half a point per year. This compares to an
increase of 14 points for black students and 13 points for
Hispanics. So the 4th grade black-white reading gap reduced
by 6 points for black students (or 0.2 sd) and the
Hispanic-white gap reduced by 5 points (or 0.02 sd) over this
21-year period. The trends for 8th grade reading are similar,
with a closure of the black – white gap of 4 points (0.1 sd)
and a closure of the Hispanic-white gap of 7 points (0.2 sd).
The growth in math scores has been greater for all groups in
both grades, with the 4th grade gains being larger than 8th
grade gains (30+ points vs. 20+ points). Black students at
both grade levels gain somewhat more than white students,
so the black-white gap is reduced by 6 points for 4th graders
(0.1 sd) but just 2 points (0.05 sd) for 8th graders. The
Hispanic-white gap remains unchanged for Hispanic 4th
graders and reduced by 2 points (0.03 sd) for 8th graders.
198
195
202
201
203
205
206
207
207
246
249
246
247
247
249
251
254
226
227
229
229
230
232
232
234
236
269
271
271
270
271
273
275
278
160
170
180
190
200
210
220
230
240
250
260
270
280
290
300
1998 2000 2002 2003 2005 2007 2009 2011 2013 1998 2002 2003 2005 2007 2009 2011 2013
4th Graders 8th Graders
Mean NAEP Scale Scores
Eligible
Not eligible
207
210
222
225
227
228
229
230
252
255
258
261
265
266
269
270
231
236
244
248
249
250
252
254
279
285
287
288
291
293
295
297
160
170
180
190
200
210
220
230
240
250
260
270
280
290
300
1996 2000 2003 2005 2007 2009 2011 2013 1996 2000 2003 2005 2007 2009 2011 2013
4th Graders 8th Graders
Mean NAEP Scale Scores
Eligible
Not eligible

Universal Journal of Educational Research 4(1): 205-311, 2016 309
*p < .05
Figure 9. NAEP Reading and Math Scores for Public School Students by Race/ethnicity, 1996 to 2013
The overall impression given by the long term trends in
NAEP scores is that there is an improvement in achievement
for all grades, groups, and subject matters, but with greater
gains in math than reading and somewhat greater gains for
black and Hispanic students than white students. This
suggests that Title I and NCLB reforms have succeeded
more in raising achievement levels for all students than in
closing achievement gaps for disadvantaged children.
4. Conclusions
The review of multiple national-level evaluations of Title
I/NCLB covering the period from 1966 to 2013 offers very
little evidence that the Title I compensatory education
program has significantly improved the academic
achievement of disadvantaged students nationwide. The
earliest national study, the meta analysis of Borman and
D'Agostino covering studies between 1966 and 1993, did
show modest Title I effects on math during elementary
grades (about 0.2 sd) but much lower effects in higher grades
and for reading in all grades (0.1 sd). The Prospects study
covering 1991 to 1994 showed no significant reduction of
the achievement gaps between Title I participants and
nonparticipants. The final early study, covering the period
1988 to 1999, was an evaluation carried out by the
Department of Education which compared 4th grade reading
and math scores for highest poverty schools to lowest
poverty schools. Generally, the achievement gap widened
between lowest and highest poverty schools over this period.
The lack of meaningful gap reductions in the evaluations
undertaken before 2000 is contrasted somewhat by evidence
provided in later studies, including an evaluation by the U.S.
Department of Education in 2007 and a study by Dee and
Jacob in 2011. Both of these studies suggest that No Child
Left Behind had modest effects on 4th grade test scores,
especially in math, and these gains were somewhat stronger
for disadvantages students. According to the Department of
Education study, the gains for disadvantaged students and
schools took place primarily between 2000 and 2002, which

310 The Effectiveness of Title I: Synthesis of National-level Evidence from 1966 to 2013
corresponds to the implementation of NCLB.
Stronger evidence that NCLB accountability improved the
achievement of disadvantaged students was provided by the
Dee and Jacob study, which used a quasi-experimental
method to compare states that had NCLB-type accountability
reforms prior to the national law. Focusing on the
achievement gap per se, the authors conducted a further
analysis of the states selected for the Dee & Jacob study. This
analysis showed that states with pre-NCLB accountability
reduced the black-white gap 7 points by 2003 compared to 5
points for states without accountability.
In order to summarize the progress made on closing
achievement gaps, a final analysis was carried out by the
authors using NAEP data that covers 1990 to 2013, more
than two decades during which various Title I programs and
policies were in place. The overall progress is disappointing,
particularly for the poverty gap. The achievement gaps
between students eligible for free/reduced lunch vs those not
eligible have remained virtually constant for reading and
math at both grade levels. The picture is more positive for
black-white and Hispanic-white gaps, particularly in 4th
grade. At that grade level, both of these gaps have been
reduced by about 6 scale score points, which is a
standardized effect size of slightly less than 0.2 sd.
Reduction in the 8th grade reading and math black-white
gaps are only 4 and 3 points, respectively.
The different progress in gap reductions before and after
2000 could reflect the very different policy approaches of
these periods. The original concept behind Title I was to
establish compensatory programs for disadvantages students,
on the assumption that extra remedial instruction would
allow these children to catch up. In retrospect, it might be
fairly argued that the level of funding, which rarely exceeded
$1500 per student (in 2012 dollars), could not be expected to
close achievement gaps, given the difficulty of this task.
The national approach after 2000 was quite different, it
was to adopt accountability practices which had proven
effective in some states during the late 1990s. Thus NCLB
was a systemic reform that aimed to raise achievement by
standardizing curriculum, adopting uniform standards, and
publishing results by demographic group, all at the state level.
It is not completely clear why policymakers assumed this
would raise achievement levels of disadvantages students
rather than raise achievement for everyone; at least that has
not been clearly articulated. To some extent, this same
thinking is behind the most recent attempt at standards
reform, which is to adopt a common core of curriculum and
standards that would be adopted nationwide. Based on the
failure of No Child Left Behind to close achievement gaps, it
is unlikely that Common Core will do so either, whatever it
does for overall achievement levels.
REFERENCES
Education Department Budget History, Online available from [1]
http://www2.ed.gov/about/overview/budget/history/index.ht
ml
ED Data Express, Online available from [2]
http://eddataexpress.ed.gov/index.cfm
H. Cooper, L. V. Hedges, J. C. Valentine (eds.). The [3]
handbook of research synthesis and meta-analysis. Russel
Sage Foundation, 2009.
S. W Raudenbush. New directions in the Evaluation of Title I. [4]
American Educational Research Association annual meeting,
New Orleans, 2002.
U.S. Department of Education. Final report on the national [5]
assessment of Title I: Summary of key findings, U.S.
Government Printing Office, Washington. DC, 2007.
J. Torgesen, A. Schirm, L. Castner, S. Vartivarian, W. Myers, [6]
F. Stancavage, D. Durno, R. Javorsky and C. Haan. National
Assessment of Title I, Final Report: Volume II: Closing the
Reading Gap, Findings from a Randomized Trial of Four
Reading Interventions for Striving Readers (NCEE
2008-4013), U.S. Government Printing Office, Washington.
DC, 2007.
U.S. Department of Education. State and Local [7]
Implementation of the No Child Left Behind Act Volume
I—Title I School Choice, Supplemental Educational Services,
and Student Achievement, U.S. Government Printing Office,
Washington. DC, 2007.
U.S. Department of Education. State and Local [8]
Implementation of the No Child Left Behind Act, Volume
IV—Title I School Choice and Supplemental Educational
Services: Interim Report, U.S. Government Printing Office,
Washington. DC, 2008.
U.S. Department of Education. State and Local [9]
Implementation of the No Child Left Behind Act, Volume
VIII—Teacher Quality Under NCLB: Final Report, U.S.
Government Printing Office, Washington. DC, 2009.
U.S. Department of Education. State and Local [10]
Implementation of the No Child Left Behind Act, Volume
IX—Accountability Under NCLB: Final Report, U.S.
Government Printing Office, Washington. DC, 2010.
E. Hanushek, M. E. Raymond. Does School Accountability [11]
Lead to Improved Student Performance?, Journal of Policy
Analysis and Management, Vol.24, No.2, 297-327, 2005.
J. D. Matsudaira, A. Hosek, E. Walsh. An integrated [12]
assessment of the effects of Title I on school behavior,
resources, and student achievement, Economics of Education
Review, Vol.31, 1–14, 2012.
W. Van der Klaauw, Breaking the link between poverty and [13]
low student achievement: and evaluation of Title I, Journal of
Econometrics, Vol.142, 731–756, 2008.
E. U. Cascio, N. E. Gordon, S.J. Reber. Federal aid and [14]
equality of educational opportunity: evidence from the
introduction of title I in the south, NBER Working Paper
Series No. 17155, National Bureau of Economic Research,
2011.
M. Wong, T. D. Cook, P. M. Steiner. No Child Left Behind: [15]
An interim evaluation of its effects on learning using two
interrupted time series each with its own non-equivalent
comparison series, Institute for Policy Research, Documento
de trabajo 09-11, No.15, 2009.

Universal Journal of Educational Research 4(1): 205-311, 2016 311
G. D. Borman, J. V. D’Agostino. Title I and student [16]
achievement: a meta-analysis of Federal evaluation results,
Educational Evaluation and Policy Analysis, Vol.18, 309–
326, 1996.
M. J. Puma, N. Karweit, C. Price, A. Ricciuti, W. Thompson, [17]
M. Vaden-Kiernan. Prospects: student outcomes, Final report.
Abt Associates, Inc., Cambridge, MA, 1997.
U.S. Department of Education. Promising results, continuing [18]
challenges: The final report of the national assessment of Title
I. U.S. Government Printing Office, Washington, DC, 1999.
U.S. Department of Education. High standards for all students: [19]
A report from the national assessment of Title I on progress
and challenges since the 1994 Reauthorization, Government
Printing Office, Washington, DC, 2001.
S. Stullich, E. Eisner, J. McCrary, Policy and Program Studies [20]
Service. National assessment of Title I. Final report. Volume I:
Implementation, U.S. Government Printing Office,
Washington, DC, 2007.
U.S. Department of Education. Title I Implementation: [21]
Update on recent evaluation findings, U.S. Government
Printing Office, Washington, DC, 2009.
T. Dee, B. Jacob. The impact of no child left behind on [22]
student achievement. Journal of Policy Analysis and
Management, Vol.30, 418–446, 2011.
E. McDill, E, G. Natriello. The effectiveness of the Title I [23]
compensatory education Program: 1965 – 1997, Journal of
Education for Students Placed at Risk , Vol.3, 317–335, 1998.
Online available from: [24]
http://nces.ed.gov/nationsreportcard/naepdata/dataset.aspx