taking the easy way out: how the ged testing program ...ftp.iza.org/dp3495.pdf · this paper...
TRANSCRIPT
DI
SC
US
SI
ON
P
AP
ER
S
ER
IE
S
Forschungsinstitut zur Zukunft der ArbeitInstitute for the Study of Labor
Taking the Easy Way Out: How the GED Testing Program Induces Students to Drop Out
IZA DP No. 3495
James J. HeckmanJohn Eric HumphriesPaul A. LaFontainePedro L. Rodríguez
May 2008(updated May 2011)
Taking the Easy Way Out: How the GED Testing Program Induces Students to Drop Out
James J. Heckman University of Chicago, American Bar Foundation,
University College Dublin and IZA
John Eric Humphries
University of Chicago
Paul A. LaFontaine American Bar Foundation
Pedro L. Rodríguez
IESA, Venezuela
Discussion Paper No. 3495 May 2008
*** updated May 2011 ***
IZA
P.O. Box 7240 53072 Bonn
Germany
Phone: +49-228-3894-0 Fax: +49-228-3894-180
E-mail: [email protected]
Any opinions expressed here are those of the author(s) and not those of IZA. Research published in this series may include views on policy, but the institute itself takes no institutional policy positions. The Institute for the Study of Labor (IZA) in Bonn is a local and virtual international research center and a place of communication between science, politics and business. IZA is an independent nonprofit organization supported by Deutsche Post World Net. The center is associated with the University of Bonn and offers a stimulating research environment through its international network, workshops and conferences, data service, project support, research visits and doctoral program. IZA engages in (i) original and internationally competitive research in all fields of labor economics, (ii) development of policy concepts, and (iii) dissemination of research results and concepts to the interested public. IZA Discussion Papers often represent preliminary work and are circulated to encourage discussion. Citation of such a paper should account for its provisional character. A revised version may be available directly from the author.
IZA Discussion Paper No. 3495 May 2008 (updated May 2011)
ABSTRACT
Taking the Easy Way Out: How the GED Testing Program Induces Students to Drop Out* The option to obtain a General Education Development (GED) certificate changes the incentives facing high school students. This paper evaluates the effect of three different GED policy innovations on high school graduation rates. A six point decrease in the GED pass rate due to an increase in national passing standards produced a 1.3 point decline in overall high school dropout rates. The introduction of a GED certification program in high schools in Oregon produced a four percent decrease in high school graduation rates. Introduction of GED certificates for civilians in California increased the high school dropout rate by 3 points. The GED program induces students to drop out of high school. JEL Classification: C61 Keywords: GED, dropout Corresponding author: James J. Heckman Department of Economics University of Chicago 1126 East 59th Street Chicago, IL 60637 USA E-mail: [email protected]
* This research was supported by the American Bar Foundation, NIH R01-HD043411, the Spencer Foundation, the Mellon Foundation, the JB and MK Pritzker Family Foundation, the Buffett Early Childhood Fund at the Susan T. Buffett Foundation, and an anonymous foundation. We would like to thank the California Demographic Research unit for helpful assistance. We thank the editor and an anonymous referee for helpful comments. We also thank participants at the Spencer Foundation GED Conference at the University of Chicago, April 2011, for helpful comments. Tim Kautz, Janice Laurence, Lois Quinn, Chris Taber and Rob Warren gave especially helpful commentary. The views expressed in this paper are those of the authors and not necessarily those of the funders listed here. A Web Appendix is available at http://jenni.uchicago.edu/GED_dropout/GED_incentives/
1 Introduction
This paper examines how changes in the availability and difficulty of the General Education
Development (GED) test affect high school dropout rates. GED certification allows dropouts
to earn a state-issued GED credential.
GED credentials account for approximately 12% of high school credentials issued in the
U.S. in 2008 (Figure 1). Test takers are required to pass a five part, 7.5 hour test to
certify their high school equivalence and earn a state-issued GED credential. Obtaining a
GED is easier for most students than graduating in the traditional fashion. The option
may be especially attractive for cognitively able students who lack credits or face other
challenges. The median study time of GED test takers that study is 32 hours (Zhang, Han,
and Patterson, 2009a).1 In the early years of the test, the minimum passing score was low
and on some sub-tests, passing scores were only slightly above what could be achieved by
chance (Quinn, 1997).
A large literature documents the small average labor market returns to GED certifi-
cation.2 Few papers have addressed whether the availability of the GED option induces
students to drop out of school rather than graduate. Chaplin (1999) and Lillard (2001)
estimate the effect of the availability of the GED on high school continuation and dropout
rates by exploiting cross-state variation in GED testing policies over time. Controlling for
state, year and age fixed effects, both studies find that state GED policies are statistically
significant predictors of high school dropout rates. Policies that provide exemptions to age
restrictions for GED testing or lower passing standards promote dropping out of high school.
States with lower requirements for the GED have higher GED test-taking rates.
The endogeneity of cross-state variation in GED requirements is a potential problem with
the identification strategy employed in previous studies. If states change GED requirements
1We do not know the amount of time spent studying by those who pass the exam. Data from 1989 (Boeselet al., 1998) shows that 56% of people reported studying 40 hours or less and 24% reported studying over100 hours. Classroom attendance for a typical school year is over 1000 hours per year, and most GEDs area year or more below twelfth grade when they drop out of school.
2See Heckman, Humphries, and Mader (2011) for a review of the literature.
2
Fig
ure
1:G
ED
Cre
den
tial
sIs
sued
asa
Per
centa
geof
All
Hig
hSch
ool
Com
ple
ters
,19
60-2
006
4%6%8%10%
12%
14%
16%
18%
20%
s as a Percent of High School Credentials
(2002) New
GED
Test In
trod
uced
(1988)
New
GED
Test Introd
uced
(1978)
New
GED
Test Introd
uced
(1966) Adu
lt Ba
sic
Education Act
0%2%
195519571959196119631965196719691971197319751977197919811983198519871989199119931995199719992001200320052007
GEDs
Year
GE
Ds
as A
Per
cent
of H
igh
Sch
ool C
rede
ntia
ls
SO
UR
CE
:P
ub
lic
an
dp
rivate
hig
hsc
hool
gra
du
ate
tota
lsfr
om
NC
ES
Dig
est
of
Ed
uca
tion
al
Sta
tist
ics
(2005),
Tab
le101;
GE
DC
reden
tials
Issu
edfr
om
GE
DS
tati
stic
al
Rep
ort
s
(Vari
ou
sY
ears
).N
OT
E:
Th
efi
gu
rep
lots
pro
port
ion
of
GE
Ds
issu
edea
chyea
rover
the
nu
mb
erh
igh
sch
ool
com
ple
ters
that
yea
r(r
egu
lar
hig
hsc
hool
gra
du
ate
san
dG
ED
reci
pie
nts
).
3
in response to trends in state-level dropout rates, estimates of the GED effect will be bi-
ased. States might respond to increased dropout rates by lowering the GED requirements.
Estimates that do not account for this response would tend to overstate the effect of lower
passing standards.
This paper presents three studies of the incentive effects of the GED program. The first
study uses an identification strategy based on a nationally mandated change in GED passing
standards imposed in 1997 by the GED Testing Service. All states were required to meet
new minimum and mean score requirements. This national mandate forced some states to
raise passing standards while other states were unaffected. This strategy addresses the endo-
geneity problem because the timing and magnitude of the one-time change in requirements
is exogenous to any state-specific trends or policy changes.
Students react strongly to the change in the difficulty of the GED test. Difference-in-
difference estimates show that a 6 percentage point decrease in the probability of passing
the GED causes a statistically significant 1.3 percentage point decline in the overall dropout
rate. The policy has its greatest effect on older students who are less restricted in their
GED testing and school leaving decisions. The percentage of students enrolled in the 12th
grade who do not graduate declines by 3 points more in states that were required to raise
GED requirements compared to those that were not required to do so. GED policy changes
have larger effects on minorities because at any grade they tend to be older and hence less
subject to minimum school leaving age requirements. They are also more likely to be behind
majority students in meeting graduation requirements.
In a second study, we examine the effect of introducing the “GED Option Program” in
Oregon. First introduced in 2001, these programs offer GED preparation and certification
in high schools. They target students perceived to be at risk of dropping out of high school
and guide them into GED certification. The program substantially reduces high school
graduation rates. Using panel data, we show that such programs lower high school graduation
rates in Oregon by 4%.
4
In a third study, we examine the impact of introducing the GED in California. In 1974,
California became the last state to award a high school equivalency credential to civilians who
passed the GED exam.3 Prior to recognizing the GED, California had higher graduation rates
than other states in the U.S. After adopting the GED program, California graduation rates
quickly fell to levels similar to those in other states. Difference-in-difference estimates show
that high school graduation rates fell by 3 percentage points more in California compared
to the rest of the U.S.
Our findings are consistent with previous studies that show that the GED induces youth
to drop out of school. We expand upon previous studies by showing that minorities and
males are more strongly affected by GED policy changes. We provide the first empirical
estimate of the effect of introducing the GED program on high school graduation rates.
The paper proceeds as follows. Section two presents a background discussion of the
relationship between GED policies and dropout rates. Section three analyzes the impact
of the 1997 GED policy change on the dropout rate. Section four estimates the effect on
the dropout rate of introducing the GED as an option for at risk high school students in
Oregon. Section five estimates the effect of introducing the GED program for civilians on
California dropout rates. Section six concludes with a discussion of our main findings and
their implications for policy.
2 Evidence on the Effects of GED Policies and Incen-
tives
The GED Testing Service (GEDTS) promotes its credential as being equivalent to a tradi-
tional high school diploma (Quinn, 1997). A recent NCES study shows that many people
view the GED credentials as an attractive alternative to graduating from high school. The
Education Longitudinal Study (ELS) follows a representative sample of 10th graders enrolled
3Prior to 1974, the GED program in California was restricted to veterans and military personnel.
5
in the spring of 2002 through graduation and beyond. In the spring of 2004, over 40% of
dropouts stated that they did not complete high school because they “thought it would be
easier to get a GED.” This was the second most cited reason behind “missed too many school
days” (43.5%). It also placed far above what are commonly believed to be primary reasons
for dropping out of school such as pregnancy (27.8%), work (27.8%) and marriage (6.8%)
(Table 1).4
There is a close relationship between trends in GED testing among school age youth and
the national dropout rate. Figure 2 plots the dropout rate both including and excluding
GED recipients as graduates. It also plots the percentage of GED test takers ages nineteen
or under in each year. Increases in the fraction of students who choose not to complete high
school are associated with rising GED test taking among secondary school-age youth. The
two time series move together in response to national GED policy changes. When GED
age requirements are lowered, GED testing rates increase for the young along with dropout
rates. When standards are increased, dropout rates fall and GED test taking by the young
declines.
The dropout rate that classifies GED recipients as dropouts reached historic lows in the
early 1970s and rose afterward (Heckman and LaFontaine, 2010). In contrast, the dropout
rate that counts GEDs as high school graduates, steadily declines over the entire period.
In the first few years depicted, the two measures are nearly equal. They begin to diverge
sharply after 1970, coinciding with the rapid expansion of the GED testing program shown
in Figure 1.
Expansion of the GED testing program is associated with a number of important policy
changes that made the GED more accessible to school-age youth. During the early 1970s,
states began to eliminate age restrictions on GED testing in an attempt to make GED cre-
dentials more accessible to young dropouts (Quinn, 1997). Previously, most states required
that individuals be at least 20 years old in order to take the GED. Additionally, in 1970 Adult
4Answers are not mutually exclusive and therefore percentages do not sum to one hundred.
6
Table 1: Percentage of Spring 2002 HS sophomores who had not Completed a HS Degree bySpring 2004, by Reason for Leaving School
Reason for leaving school Percent
Missed too many school days 43.5
Thought it would be easier to get GED 40.5
Getting poor grades/failing school 38.0
Did not like school 36.6
Could not keep up with schoolwork 32.1
Became pregnant* 27.8
Got a job 27.8
Thought could not complete course requirements 25.6
Could not get along with teachers 25.0
Could not work at same time 21.7
Had to support family 20.0
Did not feel belonged there 19.9
Could not get along with other students 18.7
Was suspended from school 16.9
Had to care for a member of family 15.5
Became father/mother of a baby 14.4
Had changed schools and did not like new one 11.2
Thought would fail competency test 10.5
Did not feel safe 10.0
Was expelled from school 9.9
Got married/planned to get married 6.8
Table 2. Percentage of Spring 2002 HS sophomores who had not
Completed a HS Degree by Spring 2004, by Reason for Leaving School
Note: This indicator shows the percentage of high school students in the spring
of their sophomore year who, in the spring 2 years later, were not in school and
had not graduated with a regular diploma or certificate of attendance. The 1
percent of sophomores who left school and earned a General Educational
Development (GED) certificate or other form of equivalency certificate as of
the spring 2 years later are counted as having left school without a regular
diploma or certificate of attendance.
Source: Reproduced from U.S. Department of Education, National Center for
Education Statistics, Education Longitudinal Study of 2002 (ELS:2002/04),
“First Follow-up, Student Survey, 2004,” previously unpublished tabulation
(January 2006).
*Percentage of female respondents only. The reason could only be selected by
female respondents.
Note: This indicator shows the percentage of high school students in the spring of their sophomore year who, in the spring 2years later, were not in school and had not graduated with a regular diploma or certificate of attendance. The 1 percent ofsophomores who left school and earned a General Educational Development (GED) certificate or other form of equivalency
certificate as of the spring 2 years later are counted as having left school without a regular diploma or certificate of attendance.Source: Reproduced from U.S. Department of Education, National Center for Education Statistics, Education LongitudinalStudy of 2002 (ELS:2002/04), “First Follow-up, Student Survey, 2004,” previously unpublished tabulation (January 2006).
∗Percentage of female respondents only. The reason could only be selected by female respondents.
7
Fig
ure
2:U
.S.
Hig
hSch
ool
Dro
pou
tR
ate
Incl
udin
gan
dE
xcl
udin
gG
ED
Rec
ipie
nts
,19
68-2
005
5%10%
15%
20%
25%
30%
35%
40%
45%
50%
10%
12%
14%
16%
18%
20%
22%
24%
26%
28%
30%
1968
1970
1972
1974
1976
1978
1980
1982
1984
1986
1988
1990
1992
1994
1996
1998
2000
2002
2004
% GED Testers Less Than 20Dropout Rate
Year
Dro
pout
Rat
e (In
c. G
EDs
as D
ropo
uts)
Dro
pout
Rat
e (E
xc. G
EDs
as D
ropo
uts)
% o
f GED
Tes
ters
<=1
9
Not
e: Th
e tru
e dro
pout
rate
is c
alcul
ated
from
NCE
S an
d CP
S da
ta a
s the
frac
tion
of p
ublic
and
priv
ate s
choo
l 8th
gra
ders
who
dono
t obt
ain a
regu
lar h
igh
scho
ol
dipl
oma.
The
NCE
S dr
opou
t rat
e is
one
min
us th
e st
atus
com
plet
ion
rate
. The
stat
us co
mpl
etio
n ra
te is
com
pute
d as
the
perc
enta
geof
18-
thro
ugh
24-y
ear-
olds
who
are
no
t enr
olle
d in
hig
h sc
hool
and
who
hav
e an
y ty
pe o
f hig
h sc
hool
cre
dent
ial. H
igh
scho
ol c
rede
ntial
incl
udes
a hi
gh sc
hool
dip
lom
a or e
quiv
alent
cred
entia
l suc
h as
a
Gen
eral
Edu
catio
nal D
evel
opm
ent (
GE
D) c
ertif
icat
e. Th
e st
atus
com
plet
ion
rate
is c
alcul
ated
from
CPS
Oct
ober
196
8-20
00 d
ata.
← (1
970)
Sta
tes b
egin
to re
duce
GE
D ag
e re
stric
tions
. Adu
lt Ba
sic E
duca
tion
low
ers
parti
cipa
tion
age f
rom
18
to 1
6.
(200
2) H
arde
r GE
D t
est i
ntro
duce
d→
Note
:T
he
tru
ed
rop
ou
tra
teis
calc
ula
ted
as
the
fract
ion
of
pu
blic
an
dp
rivate
sch
ool
8th
gra
der
sw
ho
do
not
ob
tain
are
gu
lar
hig
hsc
hool
dip
lom
a.
Pu
blic
sch
ool
enro
llm
ent
an
dp
ub
lic
an
dp
rivate
hig
hsc
hool
dip
lom
aco
unts
com
efr
om
the
NC
ES
Dig
est
of
Ed
uca
tion
Sta
tist
ics
(vari
ou
syea
rs).
An
nu
al
pri
vate
sch
ool
enro
llm
ent
ises
tim
ate
dfr
om
CP
SO
ctob
erd
ata
.T
he
NC
ES
dro
pou
tra
teis
on
em
inu
sth
est
atu
sco
mp
leti
on
rate
.T
he
statu
sco
mp
leti
on
rate
isco
mp
ute
dfr
om
CP
SO
ctob
erd
ata
as
the
per
centa
ge
of
18-
thro
ugh
24-y
ear-
old
sw
ho
are
not
enro
lled
inh
igh
sch
ool
an
dw
ho
have
any
typ
eof
hig
hsc
hool
cred
enti
al.
Hig
hsc
hool
cred
enti
al
incl
ud
esa
hig
hsc
hool
dip
lom
aor
equ
ivale
nt
cred
enti
al
such
as
aG
ED
.
8
Education (AE) programs began targeting younger populations by lowering the minimum
age requirement for participation from 18 to 16 (Heckman, Humphries, and Mader, 2011).
In the same year, Adult Secondary Education (ASE) programs were introduced targeting
those lacking secondary education. These programs produced many GED credentials. AE
programs issued 20% of GED credentials in 1972 and 40% by 1980 (Heckman, Humphries,
and Mader, 2011).
Following these changes, both the dropout rate and the percentage of young GED test
takers began to rise. Figure 3 shows that the average age of GED testing dropped precipi-
tously in the early 70s. The average age of GED test takers declined from 29 in 1970 to 25
in 1973.5 The average age has remained low since then except for a sharp increase in 1974
that coincides with the introduction of Pell grants financing higher education, which initially
required at least a GED to qualify.6
3 The Effect of the 1997 GED Policy Change
The survey and time series evidence suggests that GED test taking is related to youth
dropout behavior. Are the observed relationships causal? In the first of the three studies
reported in this paper, we address this question by exploiting exogenous variation in the
difficulty of passing the GED arising from a nationally mandated toughening of GED passing
score requirements in 1997. Prior to 1997, states fell into one of three groups: (1) 19 states
with a requirement of a minimum score of 40 on each sub-test and a mean score of 45 across
all sub-tests; (2) 26 states with a 35 minimum and 45 mean requirement and; (3) a group of
5 states where GED candidates had to achieve a 40 minimum on each test and/or a mean
score of 45 across all tests.
5Heckman and LaFontaine (2010) show that the baby boom and the subsequent baby bust accounts foronly a small portion of the variation in average age of GED test takers.
6The Federal Pell Grant Program provides need-based grants to low-income undergraduate and certainpost-baccalaureate students to promote access to postsecondary education. (U.S. Department of EducationWebsite, http://www.ed.gov/programs/fpg/index.html). The sharp rise in the average age in 1974 waspossibly due to a pent up demand for college among older dropouts.
9
Fig
ure
3:A
vera
geA
geof
GE
DT
est
Tak
ers
from
the
GE
DT
esti
ng
Ser
vic
e,19
60-2
006
252627282930erage Age of GED Test Takers
(1970)
Adu
lt Ed
ucation lowers age requ
irem
ent from
18
to 16 and introd
uces Adu
lt Second
ary Ed
cuation
for secon
dary edu
catio
n
(1974)
Pell Grants Introd
uced
2324
1960
1962
1964
1966
1968
1970
1972
1974
1976
1978
1980
1982
1984
1986
1988
1990
1992
1994
1996
1998
2000
2002
2004
2006
2008
Ave
Year
Aver
age
Age
Sou
rce:
Am
eric
an
Cou
nci
lon
Ed
uca
tion
,G
ener
al
Ed
uca
tion
al
Dev
elop
men
tT
esti
ng
Ser
vic
eS
tati
stic
al
Rep
ort
s.
Note
s:T
he
Fed
eral
Pel
lG
rant
Pro
gra
mp
rovid
esn
eed
-base
dgra
nts
tolo
w-i
nco
me
un
der
gra
du
ate
an
dce
rtain
post
bacc
ala
ure
ate
stu
den
tsto
pro
mote
acc
ess
top
ost
seco
nd
ary
edu
cati
on
.(U
.S.
Dep
art
men
tof
Ed
uca
tion
Web
site
).
10
Starting January 1st 1997, all states had to meet the new standard of a minimum score
of 40 on each test and a mean score of 45. This standard forced the second group of states
to raise their minimum score requirement on each test from 35 to 40 and the third group of
states to eliminate the and/or scoring option. The first group of states that already met the
new standards did not change their requirements. Figure 4 shows the geographic distribution
of the states by category.
According to a norming study conducted by the American Council on Education, only
67% of graduating high school seniors are able to meet a minimum score requirement of 40
and a mean score requirement of 45. A minimum of 35 and a mean of 45 was obtained
by 70% and 75% scored at the 40 or mean of 45 threshold (Table 2). Thus, the change in
difficulty of passing the GED was far greater in the third group relative to the other two.
Observed changes in pass rates in the three types of states before and after 1997 reflect this
difference (See the far right-hand column of Table 2).
The third group of 5 states serves as our “treatment” group. The states that did not
change their standards serve as the “control” group.7 In our analysis, we compare GED
testing and dropout rates in treatment and control states in the years 1994-1996 to the
same rates measured in 1998-2000. We exclude 1997 from our empirical analysis because the
change in GED requirements occurred in the middle of the school year. The reform could
cause some students to drop out and take the GED early in the year and others to stay in
school after the requirements were changed later in the same year.
We compute three measures of annual dropout rates using the Common Core of Data
(CCD) and a methodology similar to that developed by Kominski (1990).8 The measures
are: (1) the overall dropout rate, defined as the percentage of students enrolled in the 10th,
7In the Web Appendix to this paper, we use as our control group the states that were required to minimallyraise the difficulty of obtaining a GED to test the exogeneity assumption. The results from this analysis areconsistent with the results reported in the text.
8The Common Core of Data (CCD) are collected from state departments of education and contain thenumber of students enrolled in each grade level in a given year in each state, as well as the number of highschool diplomas issued in that year. From these annual counts, approximate annual exit rates from eachgrade can be computed. See the Web Appendix for more details on the construction of these measures.
11
Fig
ure
4:Sta
tes
That
Wer
eR
equir
edto
Rai
seG
ED
Pas
sing
Sta
ndar
ds
in19
97
Sta
tes
that
wer
e re
quir
ed t
o e
limin
ate
the
and
/o
r sc
ore
op
tio
n (
trea
tmen
t gr
oup
).
Sta
tes
that
wer
e re
quir
ed t
o r
aise
th
e G
ED
min
imum
sco
re r
equir
emen
t fr
om
35 t
o 4
0.
Sta
tes
that
wer
e n
ot
requir
ed t
o r
aise
sta
nd
ard
s in
1997 (
con
tro
l gr
oup
).
So
urc
e: G
ED
Tes
tin
g Ser
vic
e: 2
001 G
ED
Sta
tist
ical
Rep
ort
.
Fig
ure
4.
Sta
tes
Th
at W
ere
Req
uir
ed t
o R
aise
GE
D P
assi
ng
Sta
nd
ard
s in
1997
Sou
rce:
GE
DT
esti
ng
Ser
vic
e:2001
GE
DS
tati
stic
al
Rep
ort
.
12
Tab
le2:
Per
centa
geof
Hig
hSch
ool
Sen
iors
Mee
ting
Var
ious
GE
DSco
reR
equir
emen
tsan
dth
eA
ctual
Chan
gein
Pas
sR
ates
Pre
-an
dP
ost-
1997
GE
D S
core
Sta
nd
ard
Nu
mb
er o
f St
ates
P
rior
to
1997
C
han
ge
% o
f H
S Se
nio
rs
Mee
tin
g R
equ
irem
ents
*
Act
ual
C
han
ge in
P
ass
Rat
e**
Min
imum
40
or M
ean
455
73%
-7.4
3%
Min
imum
35
and
Mea
n 45
2669
%-1
.68%
Min
imum
40
and
Mea
n 45
1967
%-1
.26%
Sou
rce:
Th
ep
erce
nta
ge
of
hig
hsc
hool
sen
iors
inth
eG
ED
norm
ing
stu
dy
mee
tin
gth
egiv
ensc
ore
requ
irem
ent
isfr
om
the
1987
GE
Dst
ati
stic
al
rep
ort
.T
he
act
ual
chan
ge
in
pass
rate
sare
from
au
thors
’ca
lcu
lati
on
base
don
vari
ou
sG
ED
stati
stic
al
rep
ort
s.
∗G
ED
Norm
ing
stu
die
sare
base
don
the
per
form
an
ceof
are
pre
senta
tive
sam
ple
of
hig
hsc
hool
sen
iors
.D
epen
din
gon
thei
rp
erfo
rman
ceG
ED
score
sare
norm
alize
dto
ob
tain
an
orm
al
dis
trib
uti
on
of
mea
n50
an
dst
an
dard
dev
iati
on
10.
∗∗In
state
sth
at
wen
tfr
om
ind
icate
dre
qu
irem
ent
pre
-1997
toM
inim
um
40
an
dM
ean
45
post
-1997.
13
11th and 12th grades in year t who are not enrolled and have not graduated in year t + 1;
(2) the lower level dropout rate, defined as the percentage of students enrolled in 10th and
11th grades in year t who are not enrolled in year t+1; and (3) the upper level dropout rate,
given by the percentage of students enrolled in 12th grade in year t who did not graduate in
year t + 1.9 These rates are yearly exit rates from school and therefore differ in levels from
more commonly reported cohort dropout rates (see, e.g., Heckman and LaFontaine, 2010).
All dropout rate calculations are then weighted by the fraction of the U.S. 15-17 year old
population that resides in each state for our sample period. Figure 5 plots our measures
of GED test taking and dropout rates by year in treatment and control states during our
sample period. 10
We define dropout rates this way for the following reasons. First, we need to compute
yearly exit rates from schooling to capture the timing of the school leaving decision before
and after the GED policy change. Second, we seek to examine whether there are differential
effects by grade and age.11 If students drop out to take the GED, we would expect to find
larger effects for students enrolled in upper grade levels since they are older and, as a group,
less restricted by compulsory schooling laws and GED testing age requirements. Third,
these measures are less sensitive to migration than estimated cohort rates at the state level
because they are defined over shorter intervals. Cohort dropout and graduation rates are
generally calculated using up to 5 year lags of enrollment and diploma counts (e.g. diplomas
issued in the spring of year t over fall 8th grade enrollment in year t-5). Our exit rates are
lagged one year and therefore less sensitive to migration.12 Finally, we do not include 9th
9The labels “overall”, “lower” and “upper” are our own and are not based on any official definitions. Allformulas used to compute each of the dropout rate measures are included in the Web Appendix.
10The plots by race are available in the Web Appendix. Data on GED testing by age is from the 1994-2000GED Statistical reports (See GED Testing Service, Various). Population totals by age are obtained fromthe Census bureau.
11The age of students is not available in CCD data so we use the grade level as a proxy measure.12A 1997 immigration reform generally made it more difficult to migrate and reside in the U.S. To test the
sensitivity of our Hispanic estimates to this reform, we compare Hispanic dropout rates in high immigrationcontrol states to the large estimates we find in treatment states. We find no significant declines in dropoutrates in these control states suggesting that bias due to migration is minimal. See the Web Appendix TableG-38 for this analysis.
14
Fig
ure
5:G
ED
Tes
ting
and
Dro
pou
tR
ates
by
Yea
r,T
reat
men
tvs.
Con
trol
Sta
tes
0123456Percentage
1994
1995
1996
1997
1998
1999
2000
Yea
r
Age
s 16−
17 T
reat
men
t Sta
tes
Age
s 18−
19 T
reat
men
t Sta
tes
Age
s 16−
17 C
ontro
l Sta
tes
Age
s 18−
19 C
ontro
l Sta
tes
(a) G
ED
Tes
ting
Rate
s by
Age
, 199
4−20
00
789101112Percentage
1994
1995
1996
1997
1998
1999
2000
Yea
r
Trea
tmen
t Sta
tes
Con
trol S
tate
s
(b) O
vera
ll 10
th−
12th
Gra
de D
ropo
ut R
ate,
199
4−20
00789101112
Percentage
1994
1995
1996
1997
1998
1999
2000
Yea
r
Trea
tmen
t Sta
tes
Con
trol S
tate
s
(c) O
vera
ll 10
th−
11th
Gra
de D
ropo
ut R
ate,
199
4−20
00
3456789101112Percentage
1994
1995
1996
1997
1998
1999
2000
Yea
r
Trea
tmen
t Sta
tes
Con
trol S
tate
s
(d) O
vera
ll 12
th G
rade
Dro
pout
Rat
e, 1
994−
2000
Not
e: G
ED
test
ting
rate
s are
cal
cula
ted
from
yea
rly G
ED
Sta
tistic
al R
epor
ts a
s the
per
ceta
nge
of th
e st
ate
popu
latio
n in
the
give
n ag
e ra
nge
who
take
the
GE
D in
that
yea
r.D
ropo
ut ra
tes a
re c
alcu
late
d fr
om th
e C
omm
on C
ore
of D
ata
(CC
D) a
s the
exi
t rat
e fo
r tho
se in
the
indi
cate
d gr
ades
in th
e gi
ven
year
. See
the
appe
ndix
for f
urth
er d
etai
ls.St
ates
requ
ired
to ra
ise G
ED
pas
s req
uire
men
ts (t
reat
men
t sta
tes)
are
: LA
, MS,
NE
, NM
, TX
. Sta
tes t
hat d
id n
ot c
hang
e pa
ss re
quire
men
ts (c
ontro
l sta
tes)
are
: AR,
CA
, CO
,D
E, D
C, F
L, ID
, KY
, MD
, MO
, NJ,
NY
, ND
, OK
, OR,
SD
, UT,
WA
, WV
, WI.
NJ i
s exc
lude
d in
all
drop
out c
alcu
latio
ns d
ue to
dat
a er
rors
.
Figu
re 5
.G
ED
Tes
ting
and
Dro
pout
Rat
es b
y Y
ear,
Trea
tmen
t vs.
Con
trol S
tate
s
Note
:G
ED
test
ing
rate
sare
calc
ula
ted
from
yea
rly
GE
DS
tati
stic
al
Rep
ort
sas
the
per
centa
ge
of
the
state
pop
ula
tion
inth
egiv
enage
ran
ge
wh
ota
ke
the
GE
Din
that
yea
r.
Dro
pou
tra
tes
are
calc
ula
ted
from
the
Com
mon
Core
of
Data
(CC
D)
as
the
exit
rate
for
those
inth
ein
dic
ate
dgra
des
inth
egiv
enyea
r.S
eeth
eW
ebA
pp
end
ixfo
rfu
rth
er
det
ails.
Sta
tes
requ
ired
tora
ise
GE
Dp
ass
requ
irem
ents
(tre
atm
ent
state
s)are
:L
A,
MS
,N
E,
NM
,T
X.
Sta
tes
that
did
not
chan
ge
pass
requ
irem
ents
(contr
ol
state
s)are
:A
R,
CA
,C
O,
DE
,D
C,
FL
,ID
,K
Y,
MD
,M
O,
NJ,
NY
,N
D,
OK
,O
R,
SD
,U
T,
WA
,W
V,
WI.
NJ
isex
clu
ded
inall
dro
pou
tca
lcu
lati
on
sd
ue
tod
ata
erro
rs.
15
graders because high rates of retention at that grade make it difficult to calculate yearly exit
rates between that grade and 10th grade. Students enrolled in 9th grade are predominately
younger than 16 and therefore not allowed to take the GED test in any state.
Conley and Taber (2005) show that standard asymptotic results do not apply to many
difference-in-difference studies due to the small number of observed policy changes. They
develop a permutation test methodology that consistently estimates the asymptotic distri-
bution of the treatment effect under the null hypothesis of no treatment effect. Since our
sample is limited to only five states that were required to change GED testing policies, we
follow their methodology when computing our test statistics. We also report robust standard
errors clustered at the state level for purposes of comparison.
Figure 6 presents the average GED test taking rate by age pre- and post-1997. The
unadjusted mean difference-in-difference estimates and standard errors are also reported
for each age group at the top of the figure. For the control group, average GED testing
rates remained essentially flat over the two periods for all age groups. In contrast, treatment
group states exhibit a sharp decline in GED testing post-1997, especially for the older cohorts
(ages 18-19) that face fewer restrictions in both leaving school and taking the GED test. The
estimated change in the treatment group GED test taking rate for the older cohorts relative
to that of the control group is about 0.74 points and is statistically significant at the 5%
level. This is a 20% decline relative to the average GED test taking rate in treatment states
prior to the change. Also, before raising passing requirements, treatment group states had
much higher GED testing rates than did states in the control group. This difference in GED
testing levels is nearly eliminated once the treatment states increased their standards.
The overall dropout rates pre- and post-1997 across all races in both control and treatment
states are presented in Figure 7. Unadjusted difference-in-difference estimates and standard
errors are reported at the top of each set of figures. The overall dropout rate declines sharply
across all race groups in the treatment states, with the largest declines occurring for blacks
16
Fig
ure
6:A
vera
geP
re-
and
Pos
t-19
97G
ED
Tes
tT
akin
gR
ate
by
Age
Gro
up
2.0
%
1.4%
2.6
%
3.1
%
2.6
%
3.6
%
2.1
%
1.5%
2.7
%2.7
%2.5
%
2.9
%
0%
1%
2%
3%
4%
5%
GE
D T
est
Tak
ing
Ag
es
16-1
9G
ED
Test
Tak
ing
Ag
es
16-1
7G
ED
Test
Tak
ing
Ag
es
18-1
9
GED Test Taking RateF
igu
re 6
. A
vera
ge
Pre
-an
d P
ost
-1997 G
ED
Tes
t T
akin
g R
ate
by
Age
Gro
up
Co
ntr
ol G
roup
Pre
-97
Tre
atm
ent
Gro
up
Pre
-97
Co
ntr
ol G
roup
Po
st-9
7T
reat
men
t G
roup
Po
st-9
7
DiD
Est
imate
-0
.53%
(-1.
09%
, -
0.3
6%
)
No
te: G
ED
tes
t ta
kin
g ra
tes
are
def
ined
as
the
rati
o b
etw
een
to
tal n
um
ber
of
test
tak
ers
of
a gi
ven
age
an
d t
ota
l p
op
ula
tio
n o
f th
at a
ge. T
he
plo
t ab
ove
sho
ws
the
aver
age
GE
D t
est
takin
g ra
te f
or
the
per
iod
pre
-1997 (
i.e. 1994-1
996)
and
po
st-1
997 (
i.e. 1998-2
000).
A
ll es
tim
ates
are
wei
ghte
d b
y th
e15-1
7 y
ear
old
po
pula
tio
n b
y st
ate.
Co
nle
y-T
aber
ad
just
edco
nfi
den
ce i
nte
rval
s in
par
enth
eses
. T
he
trea
tmen
t gr
oup
co
nsi
sts
of
stat
es t
hat
wer
e re
quir
ed t
o e
limin
ate
the
and
/o
r sc
ore
op
tio
n. T
hes
e in
clud
e: L
A, M
S,
NE
, NM
, T
X. T
he
con
tro
l gr
oup
co
nsi
sts
of
stat
es t
hat
alr
ead
y h
ad h
igh
en
ough
sta
nd
ard
s b
y 1997. T
hes
e in
clud
e: A
R, C
A, C
O, D
E, D
C, F
L, ID
, K
Y, M
D, M
O, N
J, N
Y,
ND
, OK
, O
R, SD
, U
T, W
A, W
V, W
I. T
he
stat
e o
f N
J is
dro
pp
ed in
ord
er f
or
test
tak
ing
rate
s to
be
con
sist
ent
wit
h d
rop
out
rate
reg
ress
ion
s. S
ourc
e: G
ED
Tes
tin
g Ser
vic
e A
nn
ual
Rep
ort
s: "
Wh
o T
oo
k t
he
GE
D?"
(1994-2
000).
DiD
Est
imate
-
0.3
1%(-
1.14
%,
-0.1
2%
)
DiD
Est
imate
-
0.7
4%
(-1.
21%
, -
0.4
4%
)
Note
:G
ED
test
takin
gra
tes
are
defi
ned
as
the
rati
ob
etw
een
tota
lnu
mb
erof
test
taker
sof
agiv
enage
an
dto
tal
pop
ula
tion
of
that
age.
Th
ep
lot
ab
ove
show
sth
eaver
age
GE
D
test
takin
gra
tefo
rth
ep
erio
dp
re-1
997
(i.e
.1994-1
996)
an
dp
ost
-1997
(i.e
.1998-2
000).
All
esti
mate
sare
wei
ghte
dby
the
15-1
7yea
rold
pop
ula
tion
by
state
.C
on
ley-T
ab
er
ad
just
edco
nfi
den
cein
terv
als
inp
are
nth
eses
.T
he
trea
tmen
tgro
up
con
sist
sof
state
sth
at
wer
ere
qu
ired
toel
imin
ate
the
an
d/or
score
op
tion
.T
hes
ein
clu
de:
LA
,M
S,
NE
,N
M,
TX
.T
he
contr
ol
gro
up
con
sist
sof
state
sth
at
alr
ead
yh
ad
hig
hen
ou
gh
stan
dard
sby
1997.
Th
ese
incl
ud
e:A
R,
CA
,C
O,
DE
,D
C,
FL
,ID
,K
Y,
MD
,M
O,
NJ,
NY
,N
D,
OK
,O
R,
SD
,U
T,
WA
,W
V,
WI.
Th
est
ate
of
NJ
isd
rop
ped
inord
erfo
rte
stta
kin
gra
tes
tob
eco
nsi
sten
tw
ith
dro
pou
tra
tere
gre
ssio
ns.
Sou
rce:
GE
DT
esti
ng
Ser
vic
eA
nnu
al
Rep
ort
s:
“W
ho
Took
the
GE
D?”
(1994-2
000).
17
and Hispanics.13 The estimated change in the 10th-12th grade dropout rate across all races
combined is -1.3% and is statistically significant at the 5% level. The effect for whites is
-0.5 percentage points whereas for blacks and Hispanics it is -0.9 and -1.7 percentage points
respectively.
In contrast, younger students did not drop out at higher rates, likely because they did not
meet the minimum age requirements for the GED (see Figure C-1 in the Web Appendix).
Dropout rates at lower levels decline in both treatment and control states for whites, blacks
and Hispanics but none of the difference-in-difference estimates are statistically significantly
different from zero.
The estimated effect of the GED testing reform on school dropout rates is also much
larger for older students (See Figure 8). The estimated change in the 12th grade dropout
rate across all race groups combined is 3.1 points and is statistically significant at the 5%
level. Again, we observe larger effects among minority students. Whereas the 12th grade
dropout for whites decreases by 1.4 points, black and Hispanic dropout rates decline by 4.4
and 7 points more in treatment states, respectively. Declines in the 12th grade dropout rate
account for nearly all of the decline in the overall dropout rate in states that increased GED
standards.
Students enrolled in lower grade levels in treatment states effectively provide a second
control group in our analysis. Most of the students in this group are not affected by changes
in GED requirements because they are too young to take the GED test without obtaining a
special exemption.14 The greater decline in dropout rates for older students suggests that the
relationship between students’ behavior and the reform does not stem from a confounding
factor that would affect all students (e.g. increased spending per pupil or number of teachers
per pupil).
13The estimates by race are not directly comparable with the “all races” category since the former includefewer states as a result of missing enrollment data by race. All estimates by race are restricted to the samesub-sample of states.
14Special exemptions to age requirements vary by state and include such conditions as teenage pregnancy,residence in a juvenile detention facility and enrollment in Job Corps programs. The 2006 GED StatisticalReport contains additional information on this topic. (See GED Testing Service, Various.)
18
Fig
ure
7:A
vera
geP
re-
and
Pos
t-19
9710
th-1
2th
Gra
de
Dro
pou
tR
ate
for
Tre
atm
ent
and
Con
trol
Gro
up
10.5
%
8.6
%
15.8
%
15.8
%
10.5
%
7.9
%
14.0
%
14.2
%
10.2
%
8.2
%
15.2
%
14.3
%
9.0
%
7.0
%
12.4
%
10.9
%
0%
2%
4%
6%
8%
10%
12%
14%
16%
18%
20%
22%
All
Races
Wh
ites
Bla
ck
sH
isp
an
ics
Dropout RateF
igu
re 7
.A
vera
ge
Pre
-an
d P
ost
-1997 1
0th
-12th
Gra
de
Dro
po
ut
Rat
e fo
r T
reat
men
t an
d C
on
tro
l G
roup
Co
ntr
ol G
roup
Pre
-97
Tre
atm
ent
Gro
up
Pre
-97
Co
ntr
ol G
roup
Po
st-9
7T
reat
men
t G
roup
Po
st-9
7
No
te: T
he
dro
po
ut
rate
is
def
ined
as
the
rati
o o
f st
ud
ents
en
rolle
d in
a g
iven
gra
de(
s) in
yea
r t
and
th
e n
um
ber
of
stud
ents
en
rolle
d in
th
e p
revio
us
grad
e(s)
in
yea
r t-
1, w
her
e t
= 1
994-2
000. A
ll es
tim
ates
are
wei
ghte
d b
y th
e 15-1
7 y
ear
old
po
pula
tio
n in
th
e gi
ven
sta
te. T
he
plo
t ab
ove
sho
ws
the
aver
age
dro
po
ut
rate
rat
e fo
r th
e p
erio
d p
re-1
997 (
i.e.
1994-1
996)
and
po
st-1
997 (
i.e. 1998-2
000).
Co
nle
y-T
aber
ad
just
edco
nfi
den
ce i
nte
rval
s in
par
enth
eses
.T
reat
men
t st
ates
are
th
ose
sta
tes
that
wer
e re
quir
ed t
o e
limin
ate
the
and
/o
r sc
ore
op
tio
n. T
hes
e in
clud
e: L
A, M
S, N
E, N
M, T
X. C
on
tro
l st
ates
are
th
ose
th
at a
lrea
dy
had
hig
h e
no
ugh
sta
nd
ard
s b
y 1
997. T
hes
e in
clud
e: A
R, C
A, C
O, D
E, D
C,
FL
, ID
, K
Y, M
D, M
O, N
J, N
Y, N
D, O
K, O
R, S
D, U
T, W
A, W
V, W
I. S
tate
s w
ith
few
er t
han
tw
o o
bse
rvat
ion
s p
er p
erio
d a
re d
rop
ped
fo
r 'a
ll ra
ces'
cat
ego
ry. Sta
tes
wit
h
few
er t
han
tw
o o
bse
rvat
ion
s p
er p
erio
d f
or
any
of
the
dro
po
ut
rate
mea
sure
s b
y ra
ce a
re d
rop
ped
fo
r b
y ra
ce c
ateg
ori
es. C
on
tro
l st
ates
dro
pp
ed f
rom
'all
race
s' r
egre
ssio
ns
due
to m
issi
ng
and
neg
ativ
e d
rop
out
rate
s in
clud
e: N
J. C
on
tro
l st
ates
dro
pp
ed f
rom
reg
ress
ion
by
race
due
to m
issi
ng
and
neg
ativ
e d
rop
out
rate
s in
clud
e: A
R, ID
, K
Y, M
O,
ND
, NJ,
NY
, SD
, U
T, W
A, W
V. N
o t
reat
men
t st
ates
are
dro
pp
ed f
rom
an
y re
gres
sio
ns.
Sin
ce t
her
e ar
e m
ore
mis
sin
gs in
th
e d
rop
ou
t ra
tes
by
race
, th
e 'a
ll ra
ces'
cat
ego
ry is
no
t d
irec
tly
com
par
able
to
th
e ca
tego
ries
by
race
. So
urc
e: C
om
mo
n C
ore
of
Dat
a (C
CD
).
DiD
Est
imate
-1
.3%
(-2.5
0%
, -0
.08%
)
DiD
Est
imate
-0
.5%
(-1.
16%
, 1.
45%
)
DiD
Est
imate
-0
.9%
(-1.
93%
, 4.7
9%
)
DiD
Est
imate
-1
.7%
(-12
.20%
, 0.6
2%
)
Note
:T
he
dro
pou
tra
teis
defi
ned
as
the
rati
oof
stud
ents
enro
lled
ina
giv
engra
de(
s)in
yea
rt
an
dth
enu
mb
erof
stu
den
tsen
roll
edin
the
pre
vio
us
gra
de(
s)in
yea
rt−
1,
wh
ere
t=1994-2
000.
All
esti
mate
sare
wei
ghte
dby
the
15-1
7yea
rold
pop
ula
tion
inth
egiv
enst
ate
.T
he
plo
tab
ove
show
sth
eaver
age
dro
pou
tra
tefo
rth
ep
erio
dp
re-1
997
(i.e
.1994-1
996)
an
dp
ost
-1997
(i.e
.1998-2
000).
Con
ley-T
ab
erad
just
edco
nfi
den
cein
terv
als
inp
are
nth
eses
.T
reatm
ent
state
sare
those
state
sth
at
wer
ere
qu
ired
toel
imin
ate
the
an
d/or
score
op
tion
.T
hes
ein
clu
de:
LA
,M
S,
NE
,N
M,
TX
.C
ontr
ol
state
sare
those
that
alr
ead
yh
ad
hig
hen
ou
gh
stan
dard
sby
1997.
Th
ese
incl
ud
e:A
R,
CA
,C
O,
DE
,
DC
,F
L,
ID,
KY
,M
D,
MO
,N
J,
NY
,N
D,
OK
,O
R,
SD
,U
T,
WA
,W
V,
WI.
Sta
tes
wit
hfe
wer
than
two
ob
serv
ati
on
sp
erp
erio
dare
dro
pp
edfo
r’a
llra
ces’
cate
gory
.S
tate
sw
ith
few
erth
an
two
ob
serv
ati
ons
per
per
iod
for
any
of
the
dro
pou
tra
tem
easu
res
by
race
are
dro
pp
edfo
rby
race
cate
gori
es.
Contr
ol
state
sd
rop
ped
from
’all
race
s’re
gre
ssio
ns
du
e
tom
issi
ng
an
dn
egati
ve
dro
pou
tra
tes
incl
ud
e:N
J.
Contr
ol
state
sd
rop
ped
from
regre
ssio
nby
race
du
eto
mis
sin
gan
dn
egati
ve
dro
pou
tra
tes
incl
ud
e:A
R,
ID,
KY
,M
O,
ND
,
NJ,
NY
,S
D,
UT
,W
A,
WV
.N
otr
eatm
ent
state
sare
dro
pp
edfr
om
any
regre
ssio
ns.
Sin
ceth
ere
are
more
mis
sin
gs
inth
ed
rop
ou
tra
tes
by
race
,th
e’a
llra
ces’
cate
gory
isn
ot
dir
ectl
yco
mp
ara
ble
toth
eca
tegori
esby
race
.S
ou
rce:
Com
mon
Core
of
Data
(CC
D).
19
Fig
ure
8:A
vera
geP
re-
and
Pos
t-19
9712
thG
rade
Dro
pou
tR
ate
for
Tre
atm
ent
and
Con
trol
Gro
up
9.5
%8.6
%
12.4
%13
.8%
8.8
%
5.7
%
13.9
%
11.4
%
9.1
%
7.9
%
12.9
%14
.3%
5.2
%
3.6
%
9.9
%
4.9
%
0%
2%
4%
6%
8%
10%
12%
14%
16%
18%
All
Races
Wh
ites
Bla
ck
sH
isp
an
ics
Dropout RateF
igu
re 9
.A
vera
ge
Pre
-an
d P
ost
-1997 1
2th
Gra
de
Dro
po
ut
Rat
e fo
r T
reat
men
t an
d C
on
tro
l G
roup
Co
ntr
ol G
roup
Pre
-97
Tre
atm
ent
Gro
up
Pre
-97
Co
ntr
ol G
roup
Po
st-9
7T
reat
men
t G
roup
Po
st-9
7
No
te: T
he
dro
po
ut
rate
is
def
ined
as
the
rati
o o
f st
ud
ents
en
rolle
d in
a g
iven
gra
de(
s) in
yea
r t
and
th
e n
um
ber
of
stud
ents
en
rolle
d in
th
e p
revio
us
grad
e(s)
in
yea
r t-
1, w
her
e t
= 1
994-2
000. A
ll es
tim
ates
are
wei
ghte
d b
y th
e 15-1
7 y
ear
old
po
pula
tio
n i
n t
he
given
sta
te. T
he
plo
t ab
ove
sho
ws
the
aver
age
dro
po
ut
rate
rat
e fo
r th
e p
erio
d p
re-1
997 (
i.e.
1994-1
996)
and
po
st-1
997 (
i.e. 1998-2
000).
Co
nle
y-T
aber
ad
just
edco
nfi
den
ce i
nte
rval
s in
par
enth
eses
.T
reat
men
t st
ates
are
th
ose
sta
tes
that
wer
e re
quir
ed t
o e
limin
ate
the
and
/o
r sc
ore
op
tio
n. T
hes
e in
clud
e: L
A, M
S, N
E, N
M, T
X. C
on
tro
l st
ates
are
th
ose
th
at a
lrea
dy
had
hig
h e
no
ugh
sta
nd
ard
s b
y 1
997. T
hes
e in
clud
e: A
R, C
A, C
O, D
E, D
C,
FL
, ID
, K
Y, M
D, M
O, N
J, N
Y, N
D, O
K, O
R, S
D, U
T, W
A, W
V, W
I. S
tate
s w
ith
few
er t
han
tw
o o
bse
rvat
ion
s p
er p
erio
d a
re d
rop
ped
fo
r 'a
ll ra
ces'
cat
ego
ry. Sta
tes
wit
h
few
er t
han
tw
o o
bse
rvat
ion
s p
er p
erio
d f
or
any
of
the
dro
po
ut
rate
mea
sure
s b
y ra
ce a
re d
rop
ped
fo
r b
y ra
ce c
ateg
ori
es. C
on
tro
l st
ates
dro
pp
ed f
rom
'all
race
s' r
egre
ssio
ns
due
to m
issi
ng
and
neg
ativ
e d
rop
out
rate
s in
clud
e: N
J. C
on
tro
l st
ates
dro
pp
ed f
rom
reg
ress
ion
by
race
due
to m
issi
ng
and
neg
ativ
e d
rop
out
rate
s in
clud
e: A
R, ID
, K
Y, M
O,
ND
, NJ,
NY
, SD
, U
T, W
A, W
V. N
o t
reat
men
t st
ates
are
dro
pp
ed f
rom
an
y re
gres
sio
ns.
Sin
ce t
her
e ar
e m
ore
mis
sin
gs in
th
e d
rop
ou
t ra
tes
by
race
, th
e 'a
ll ra
ces'
cat
ego
ry is
no
t d
irec
tly
com
par
able
to
th
e ca
tego
ries
by
race
. So
urc
e: C
om
mo
n C
ore
of
Dat
a (C
CD
).
DiD
Est
imate
-3
.1%
(-5.0
0%
, -
1.06%
)
DiD
Est
imate
-1
.4%
(-3.0
9%
, 0.5
2%
)
DiD
Est
imate
-4
.4%
(-7.4
5%
, 2.7
6%
)
DiD
Est
imate
-7
.0%
(-8.4
9%
, -
0.8
5%
)
Note
:T
he
dro
pou
tra
teis
defi
ned
as
the
rati
oof
stud
ents
enro
lled
ina
giv
engra
de(
s)in
yea
rt
an
dth
enu
mb
erof
stu
den
tsen
roll
edin
the
pre
vio
us
gra
de(
s)in
yea
rt−
1,
wh
ere
t=1994-2
000.
All
esti
mate
sare
wei
ghte
dby
the
15-1
7yea
rold
pop
ula
tion
inth
egiv
enst
ate
.T
he
plo
tab
ove
show
sth
eaver
age
dro
pou
tra
tefo
rth
ep
erio
dp
re-1
997
(i.e
.1994-1
996)
an
dp
ost
-1997
(i.e
.1998-2
000).
Con
ley-T
ab
erad
just
edco
nfi
den
cein
terv
als
inp
are
nth
eses
.T
reatm
ent
state
sare
those
state
sth
at
wer
ere
qu
ired
toel
imin
ate
the
an
d/or
score
op
tion
.T
hes
ein
clu
de:
LA
,M
S,
NE
,N
M,
TX
.C
ontr
ol
state
sare
those
that
alr
ead
yh
ad
hig
hen
ou
gh
stan
dard
sby
1997.
Th
ese
incl
ud
e:A
R,
CA
,C
O,
DE
,
DC
,F
L,
ID,
KY
,M
D,
MO
,N
J,
NY
,N
D,
OK
,O
R,
SD
,U
T,
WA
,W
V,
WI.
Sta
tes
wit
hfe
wer
than
two
ob
serv
ati
on
sp
erp
erio
dare
dro
pp
edfo
r’a
llra
ces’
cate
gory
.S
tate
sw
ith
few
erth
an
two
ob
serv
ati
ons
per
per
iod
for
any
of
the
dro
pou
tra
tem
easu
res
by
race
are
dro
pp
edfo
rby
race
cate
gori
es.
Contr
ol
state
sd
rop
ped
from
’all
race
s’re
gre
ssio
ns
du
e
tom
issi
ng
an
dn
egati
ve
dro
pou
tra
tes
incl
ud
e:N
J.
Contr
ol
state
sd
rop
ped
from
regre
ssio
nby
race
du
eto
mis
sin
gan
dn
egati
ve
dro
pou
tra
tes
incl
ud
e:A
R,
ID,
KY
,M
O,
ND
,
NJ,
NY
,S
D,
UT
,W
A,
WV
.N
otr
eatm
ent
state
sare
dro
pp
edfr
om
any
regre
ssio
ns.
Sin
ceth
ere
are
more
mis
sin
gs
inth
ed
rop
ou
tra
tes
by
race
,th
e’a
llra
ces’
cate
gory
isn
ot
dir
ectl
yco
mp
ara
ble
toth
eca
tegori
esby
race
.S
ou
rce:
Com
mon
Core
of
Data
(CC
D).
20
Fig
ure
9:P
erce
nta
geof
HS
Stu
den
tsE
ighte
enor
Old
erby
Gra
de
and
Rac
e,C
PS
Oct
ober
1994
-200
0
1.3%
5.4
%
29.3
%
0.9
%
3.1
%
25.6
%
1.6%
9.7
%
37.7
%
2.8
%
11.5
%
38.7
%
0%
5%
10%
15%
20%
25%
30%
35%
40%
45%
50%
10th
Gra
de
11th
Gra
de
12th
Gra
de
% of Students
Fig
ure
10.
Per
cen
tage
of
HS S
tuden
ts E
igh
teen
or
Old
er b
y G
rade
and R
ace,
CP
S O
cto
ber
1994-2
000
Over
all
Wh
ites
Bla
cks
His
pan
ics
Auth
ors
' cal
cula
tio
ns
fro
m C
PS O
cto
ber
1994-2
000 d
ata.
Th
e sa
mp
le is
rest
rict
ed t
o t
ho
se w
ho
rep
ort
bei
ng
enro
lled
in
hig
h s
cho
ol b
etw
een
th
e ag
es o
f 12 a
nd
20.
Au
thors
’ca
lcu
lati
on
sfr
om
CP
SO
ctob
er1994-2
000
data
.T
he
sam
ple
isre
stri
cted
toth
ose
wh
ore
port
bei
ng
enro
lled
inh
igh
sch
ool
bet
wee
nth
eages
of
12
an
d20.
21
If high school students respond to changes in GED score requirements, GED testing rates
would likely increase immediately before increases in the standards and decline immediately
afterward, artificially increasing our estimate of the dropout effect. In order to check that
our estimates are not produced from a surge in test taking in 1996 and a subsequent decline
induced by the shift to higher standards in 1997, we delete the 1996 observations. This
barely effects our estimates.15
Figure 9 shows that differences in age between whites and minority students might explain
why the reform has greater impact on minority students. For whites, 25.6% of the students
are 18 and above in the fall of 12th grade. The corresponding figures for blacks and Hispanics
are 37.7% and 38.7%, respectively. Far more minority students are in the age group that is not
restricted by mandatory school leaving age requirements or GED minimum age requirements.
Thus, more minority students are at risk of being induced to drop out of school by the GED at
any given grade level. In addition, minority students have fewer credits than white students
at each grade level, making the GED a more attractive option for them (See Agodini and
Dynarski (1998)). Availability of the GED will induce more students to drop out as more
students both delay entry into school and are held back in school.16 Restricting the minimum
age of GED test taking is one way to prevent early exit from secondary education.17
4 The GED Option Program
In our second study we evaluate the effect of introducing school-sanctioned GED preparation
programs into high schools. As previously noted, a large and growing number of GED test
takers certify before age 20 and before their high school class graduates. This represents a
shift away from the “traditional” concept of the GED as a second chance for older dropouts.
15See Web Appendix D.4 for this and other sensitivity analyses. In Section D.5 of the web appendix weshow that the trends prior to 1994 and post-2000 are in line with the trends displayed in Figure 5. Thetrends in the time period studies are not anamalous.
16 Heckman and LaFontaine (2010).17The Web Appendix Section E (locate at http://jenni.uchicago.edu/GED_dropout/GED_
incentives/) presents a fixed effect analysis of the introduction of higher passing standards on dropoutrates which corroborate the analysis of Section 3.
22
As shown in Figure 10a, 16 to 19 year old GED test takers are the largest and fastest
growing group. Figure 10b shows that the bulk of the growth in the 16 to 19 age category
comes from 16 to 17 year olds. The GED may be inducing students to leave high school
rather than graduate.
4.1 The GED Option Program
The American Council on Education (ACE), the organization that operates the GED test,
allows some states to offer the “GED Option Program.” This program offers GED prepara-
tion and certification in high schools. It aims to target students at a high risk of dropping
out, and guide them into GED certification as an alternative. The definition of high risk
varies by state, but typically is defined to mean that the student is at risk of not graduating
with his class or is a year behind in credits.
Originally started in 1989, implementation of the Option Program varies greatly by state.
Virginia requires 15 hours of academic preparation per week and work- or career-based
training for 10 hours a week, for a median of 12 weeks. Virginia also requires scores of 450
on each subsection of the official practice test. This is higher than the passing standard. In
contrast, Oregon reports a median of 20 study hours and median enrollment of 75 days.18,19
Despite the stated goal of targeting students behind in credits and likely to dropout, the
majority of Option Program participants graduate before their high school class.20
The GED Option Program offers at-risk students a mixed bag. The program may help
teach valuable skills to students who would otherwise drop out. However, introducing the
GED directly into the high school may induce some students to GED certify rather than
graduate. It may do this in several ways. Its presence in regular high schools lowers the
information costs of learning about, preparing for, and taking the GED. Integrating it into
the school system may also give the GED credential credibility. Teachers and counselors
18GED Option Statistical Report (2009).19 Figures F-1 and F-2 in the Web Appendix show that the state option programs vary greatly in terms
of average days enrolled and average preparation.20See Figure F-3 in the Web Appendix.
23
Figure 10: Decomposing GED Testing Trends by Age
(a) Number of Test Takers by Age
50,000
100,000
150,000
200,000
250,000
300,000
350,000
400,000
450,000
GED
Test Takers
0
1974
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
Year
Age 16 to 19 Age 20 to 24 Age 25 to 29
Age 30 to 34 Age 35 to 39 Age 40 Plus
(b) Decomposing 16 to 19 year of Test Taking
30%
25%
t Takers
20%
t of G
ED Test
15%
Percen
t
10%
5%
1975 1980 1985 1990 1995 2000 2005 2010
Age 16 to 17 Age 18 to 19Age 16 to 17 Age 18 to 19
Source: GED Testing Services Annual Statistical Reports (1974-2009).
24
may encourage at-risk students to pursue the GED, not knowing the evidence of its minimal
beneficial impact. School districts can count GED Option students for funding purposes
while they are enrolled in the program, but remove them from the classroom. Administrators
may encourage disruptive students to take the GED Option. The GED Option may have
peer effects as those preparing for the GED are still present at their local high schools. The
credibility of the GED Option is further bolstered by the fact that in many states the GED
credential is semantically indistinguishable from a high school diploma.21 (See Table F-1 in
the Web Appendix.)
4.2 The Oregon GED Option Program
We evaluate the effect of the Oregon GED Option program on high school graduation rates
using administrative records. Oregon implemented a GED Option program in 2001. The
requirements of Oregon’s program are lower than those of many other states.
Oregon allows school districts to petition the state for permission to implement GED
Option programs. Once permitted, school districts typically implement the program in one
of the following ways: (1) district wide; (2) in specific high schools in the district; or (3)
only in non-traditional high schools or community colleges in the district. By 2005, 54% of
districts had some form of an option program, with 49% having option programs in some
schools and 28% having district-wide option programs.
4.3 The Effect of the GED Option Program on Cohort Completion
Rates
Using the Common Core Data, we construct 8th grade, 9th grade, and 10th grade cohort
completion rates, where the 8th grade cohort completion rate is the number of diplomas
issued in a year divided by the number of 8th graders enrolled four years earlier. 9th and
21However, No Child Left Behind legislation prohibits states from issuing actual diplomas based on GEDcertification.
25
10th grade cohort completion rates are constructed in the same manner, but using 9th grade
enrollment lagged three years or 10th grade enrollment lagged 2 years. We use all three
cohort completion rates to check the robustness of our estimates. Using Oregon Department
of Education Administrative data we construct three variables to capture the presence of a
GED Option Program: (1) a dummy for district-wide implementation, (2) a dummy if the
district has any regular high schools offering GED Option programs, and (3) a dummy for
implementation of a GED Option program, but not in regular high schools. Districts with
GED option programs outside of regular high schools typically have GED Option programs
in local community colleges, or second-chance schools for expelled students. School districts
with option programs in regular schools typically have a large number of schools offering the
program and resemble district-wide programs. We present evidence on the exogeneity of the
presence of these programs in districts in Section 4.4.
Using this data we regress one of our three cohort completion rates on one of the three
measures of the GED Option program, a set of district controls, and district and year fixed
effects:
Yi,t = αGEDi,t + β1Xi,t + β2Y eart + β3Districti + εi,t, i = 1, . . . , I; t = 1, . . . , T
where Yi,t is either 8th, 9th, or 10th grade cohort completion rates; GEDi,t is a binary vari-
able indicating the presence or absence of a GED Option program; Xi,t is a vector of time
variant district characteristics; Y eart is a year fixed effect and Districti is a district fixed
effect. We include among the Xi,t: percent black enrollment, percent Hispanic enrollment,
percent free lunch eligible, percent free or reduced lunch eligible, the pupil-teacher ratio,
total expenditure per pupil and total revenue per pupil. State and federally operated dis-
tricts, charter districts, vocational or special needs districts, and non-operating districts are
excluded from our analysis. We include data from 2000 through 2008 in our analysis, with
2002 being the first year programs were offered.
26
The presence of district-wide GED Option programs in schools decreases cohort comple-
tion rates. 8th, 9th, and 10th grade cohort completion rates decrease by 4.2%, 3.0%, and
4.2% after districts implement a district-wide program (Figure 11a). These estimates are
all statistically significant.22 We find similar results ranging from 3% to 4% for districts
that have any option program in a regular high school.23 Cohort completion rates are not
affected in districts with Option programs implemented outside of regular high schools. (See
Figure 11b.) This evidence supports the notion that information and availability play key
roles in the decision to GED certify.
Not all students induced to drop out of high school by the Option program GED certify.
We find an increase in GED certification rates24 across cohorts in districts that adopt a GED
Option program. The presence of a district-wide Option program is associated with a 1.7%
increase in the cohort GED certification rate for the 8th, 9th, and 10th grade cohorts.25 This
increase is only half of the estimated decrease in cohort diploma rates.26
4.4 Which Districts Adopt Option Programs?
Districts select into Option Programs. Such selection may bias our results. Using NCES
Common Core Data and district level 2000 Census data, we compare districts prior to the
2001 introduction of the GED Option. We find only small differences between districts that
adopt GED Option programs and those that do not, suggesting that selection does not play
a role (see Figure F-5 and Figure F-6 in the Web Appendix).
22Estimates are statistically significant at the 0.05 level for 8th and 10th grade cohort completion ratesand the 0.1 level for 9th grade rates.
23See Figure F-4 in the Web Appendix.24GED certification rates are measured by the number of “other completers” reported by a district, which
includes individuals that GED certify through school or state preparation programs.25These estimates are jointly statistically significant at the 0.10 level.26We find no statistically significant effect of the Option Program on dropout rates when it is placed in
alternative schools. This provides evidence that its presence in ordinary high school advertises its availabilityand possibly fosters iatrogenic peer effects.
27
Figure 11: The Effect of the GED Option Program on High School Cohort Completion Rates
(a) District-Wide GED Option Program
1.7% 1.7% 1.7%
2%
4%eter Rates
0%
2%
r Other Com
ple
-2%
hort Diploma o
‐3.0%
‐4.2%-6%
-4%
t Change in Coh
‐4.2%
-8%Diploma Rate (8th) Diploma Rate (9th) Diploma Rate (10th)
Percen
t
Diplomas Other Completers
(b) Option Program Outside of Regular High Schools
2.5%2.7%
5%
6%
eter Rates
1.8%
1.4%3%
4%
or Other Com
ple
0.4% 1.0%
1%
2%
hort Diploma o
1%
0%
1%
t Change in Coh
-2%
-1%
Diploma Rate (8th) Diploma Rate (9th) Diploma Rate (10th)
Percen
Diplomas Other Completers
Source: National Center for Education Statistics, Common Core Data and Oregon School Districts Administrative Data. Notes:
Cohort completion rates are defined as the number of diplomas issued divided by 8th, 9th, or 10th grade enrollment lagged
the appropriate number of years. The definition of other completers includes students who GED certify through a district or
state-sanctioned certification program, and thus should capture students who GED certify through the GED Option program.
Regressions include controls for percent black enrollment, percent Hispanic enrollment, percent free lunch eligible, percent free
or reduced lunch eligible, pupil teacher ratio, expenditures per pupil, revenue per pupil, and district and year fixed effects.
Regressions include 2001-2002 school year through 2007-2008 school year. The bars show standard errors.
28
5 Eliminating the GED Option in California
Our third study examines the effect of introducing the GED for civilians on dropout rates in
California. In 1974, California became the last state to offer a state recognized credential for
GEDs.27 The California legislature amended the state education code to include provisions
for the issuance of the California High School Equivalency Certificate to be given out on the
basis of GED test scores.28 According to the new law, this certificate would be “deemed to
be a high school diploma for the purpose of meeting the requirements of employment by all
state and local public agencies.”
In 1974, the GED testing program was neither as large as it is today nor as popular
among school age youth. However, even then it accounted for nearly 9% of all high school
credentials issued and 34% of the test takers were age 16 to 19.29
To assess the impact of introducing the GED program we compare the high school grad-
uation rate in California with that of all other states in the three years before and after
1974. Since enrollment counts by grade are not available on a state-by-state basis in this
period, we use an estimate of the 14 year old population as a proxy for the entering 9th
grade enrollment.30
Figure 12 displays the overall, male and female mean high school graduation rates pre-
and post-1974 in both California and the rest of the country.31 Difference-in-difference
estimates are also reported. Prior to the introduction of the GED program, California had a
high school graduation rate that was higher than that in the rest of the United States. Once
the GED was introduced, California graduation rates immediately fell to the levels of other
27See Allen and Jones (1992).28California Legislature (1973).29See Figures 1 and 2.30Population estimates for California were obtained from the California Demographic Research Unit. They
provide estimates of the state population by age for the resident population on July 1st of each year. Weuse the July 1st 15 year old population in the next year to proxy for the previous years fall 14 year oldpopulation. U.S. population estimates by age are from the Census Bureau and are also estimates of theresident population on July 1st.
31Figure G-1 in the Web Appendix displays completion rates by year for California and the rest of thecountry for the years 1971-1977.
29
states. While graduation rates fell both in California and the rest of the U.S. during this
time, difference-in-difference estimates show that introducing the GED program resulted in
a 3 percentage point greater drop in California relative to other states in the period from
1975-1977.
The adverse effect of introducing the GED program on graduation rates was larger for
males than females.32 Male graduation rates fell by 3.6 percentage points while the gradu-
ation rate for females declined by 2.6 points. One reason for this differential is that males
might have better immediate employment opportunities and would, therefore, find an early
exit from high school through GED certification a more attractive option. Additionally,
males are farther behind in school than females at any given age. This finding is also con-
sistent with the evidence reported in Heckman and LaFontaine (2010), who show that male
graduation rates have declined more than female rates since the early 1970s.
Evidence from the late introduction of the GED program in California further suggests
that the GED induces youth to drop out of school. Eliminating the GED option would
increase high school graduation rates. Arguably, estimates based on 1974 data understate
the effect we would observe today if the GED were not available to students. Since 1974,
the GED program has expanded and become more popular with adolescents and young
adults. In addition, high school standards, as measured by mandatory courses and high-
stakes testing requirements, have increased substantially since the mid-1980s (See Lillard
and DeCicca (2001)). These changes increase the cost of graduating from high school and
the attractiveness of the GED. For all of these reasons, it is plausible that our estimates
based on California in the mid-1970s understate the potential impact of enacting this reform
under current conditions.
32Data for this period are not available by race.
30
Fig
ure
12:
Gra
duat
ion
Rat
eB
efor
ean
dA
fter
Imple
men
ting
the
GE
DP
rogr
am,
Cal
ifor
nia
vs.
All
other
Sta
tes
75.5
%
73.3
%
77.8
%77.9
%
76.2
%
79.8
%
73.7
%
71.
3%
76.1
%
73.0
%
70.6
%
75.5
%
66%
68%
70%
72%
74%
76%
78%
80%
82%
84%
Ove
rall
Male
sF
em
ale
s
Graduation Rate
Fig
ure
11.
Gra
duat
ion
Rat
e B
efo
re a
nd A
fter
Im
ple
men
tin
g th
e G
ED
Pro
gra
m, C
alif
orn
ia v
s. A
ll o
ther
Sta
tes
U.S
. (E
xcl
. C
A)
Pre
-74
Cal
ifo
rnia
Pre
-74
U.S
. (E
xcl
. C
A)
Po
st-7
4C
alif
orn
ia P
ost
-74
DiD
Est
imate
-3
.1%
No
tes:
Auth
ors
' cal
cula
tio
ns
bas
ed o
n N
CE
S d
ata.
Th
e gr
aduat
ion
rat
e is
th
e n
um
ber
of
regu
lar
pub
lic a
nd
pri
vat
e h
igh
sch
oo
l d
iplo
mas
iss
ued
over
th
e 14 y
ear
old
p
op
ula
tio
n f
our
year
s p
revio
us.
Po
pula
tio
n t
ota
ls f
or
the
U.S
. w
ere
ob
tain
ed f
rom
th
e U
.S. C
ensu
s B
ure
au. C
alif
orn
ia p
op
ula
tio
n e
stim
ates
wer
e o
bta
ined
fro
m t
he
Cal
ifo
rnia
Dem
ogr
aph
ic R
esea
rch
Un
it. H
ub
er-W
hit
e ro
bust
sta
nd
ard
err
ors
in
par
enth
eses
. Sta
te 1
5 y
ear
old
po
pula
tio
n a
re u
sed
as
wei
ghts
. P
re-p
erio
d is
def
ined
as
1971-1
973 a
nd
Po
st-p
erio
d a
s 1975-1
977.
DiD
Est
imate
-3.6
%
DiD
Est
imate
-2.6
%
Note
s:A
uth
ors
’ca
lcu
lati
on
sb
ase
don
NC
ES
data
.T
he
gra
du
ati
on
rate
isth
enu
mb
erof
regu
lar
pu
blic
an
dp
rivate
hig
hsc
hool
dip
lom
as
issu
edover
the
14
yea
rold
pop
ula
tion
fou
ryea
rsp
revio
us.
Pop
ula
tion
tota
lsfo
rth
eU
.S.
wer
eob
tain
edfr
om
the
U.S
.C
ensu
sB
ure
au
.C
ali
forn
iap
op
ula
tion
esti
mate
sw
ere
ob
tain
edfr
om
the
Califo
rnia
Dem
ogra
ph
ic
Res
earc
hU
nit
.H
ub
er-W
hit
ero
bu
stst
an
dard
erro
rsin
pare
nth
eses
.S
tate
15
yea
rold
pop
ula
tion
are
use
das
wei
ghts
.P
re-p
erio
dis
defi
ned
as
1971-1
973
an
dP
ost
-per
iod
as
1975-1
977.
31
6 Conclusion
This paper presents three studies of the effect of the GED program on the high school
graduation rate. In the first study, we find that raising the difficulty of obtaining the GED,
through increasing passing requirements, reduces dropout rates. A nationally mandated
increase in GED passing standards in 1997 resulted in a 1.3 percentage point drop in the
overall dropout rate in states that were required to change their standards relative to those
that were not required to do so. The observed reduction in dropout rates was stronger for
older students enrolled in upper grade levels since these individuals are less restricted in
both school leaving and GED testing. The twelfth grade high school dropout rate fell by 3
percentage points following the 1997 reform.
Minorities are more sensitive to the availability of GED credentials than are whites. At
a given grade level, minority students tend to be older and further behind than majority
students. These factors make obtaining a GED credential more attractive than high school
graduation for minorities. Minority dropout rates exhibit the sharpest declines following
the increase in GED passing standards. Black 12th grade dropout rates declined by 4.8
percentage points, those for Hispanics by 6.2 points and those for whites by 1.3 percentage
points. There are smaller changes in dropout behavior for younger students not eligible to
drop out.
In a second study, we examine the effect of introducing the GED into the high school
setting. The GED Option program integrates GED test preparation and certification for
struggling students directly into high schools. The introduction of the GED Option program
in Oregon led to a four percent decrease in graduation rates.
In a third study, we show that introducing the GED produces substantial changes in
overall graduation rates. Prior to the introduction of the GED program in 1974, California
had higher graduation rates compared to those in the rest of the country. Our estimates show
that when the California legislature established credentials for civilian dropouts passing the
GED test, graduation rates fell by 3 points in California relative to the rest of the U.S., and
32
graduation levels dropped to those of the other states. Eliminating the GED option today
would likely have much larger effects given the wider acceptance of the program.
Taken together, these studies suggest that the GED program induces students to drop
out of school. The program has changed from its original intention of providing a second
chance for adults to becoming a primary vehicle for obtaining high school certification among
many students enrolled in secondary education.
This evidence should be a source of concern. The benefits of GED certification are slight.
GED recipients perform in the labor market, post-secondary schooling, the military, and,
in general, society at a level very close to that of dropouts and below that of high school
graduates (see Heckman, Humphries, and Mader, 2011). Given the poor performance of
GED recipients relative to high school graduates, the findings reported in this paper provide
evidence that states should adopt policies to eliminate the availability of the GED for school-
age children. Such a change in policy would not only raise high school graduation rates, but
could also improve the future prospects of disadvantaged students.
33
References
Agodini, R. and M. Dynarski (1998, June). Understanding the trend toward alternative certi-
fication for high school graduates. Document PR98-39, 37, Mathematica Policy Research,
Inc., Princeton, NJ.
Allen, C. A. and E. V. Jones (1992). GED Testing Program: The First Fifty Years. Wash-
ington, D.C.: American Council on Education.
Boesel, D., N. Alsalam, and T. M. Smith (1998). Educational and Labor Market Performance
of GED Recipients. Washington, DC: U.S. Dept. of Education, Office of Educational
Research and Improvement, National Library of Education.
California Legislature (1973). Chapter 6 of division 6 of the education code, article 9. In
West’s Education Code, Chapter 872, pp. 1594. Sacramento, CA: State of California.
Chaplin, D. (1999, November). GEDs for teenagers: Are there unintended consequences?
Technical report, The Urban Institute.
Conley, T. and C. Taber (2005, June). Inference with “difference in differences” with a small
number of policy changes. Technical Working Paper 312, NBER.
GED Testing Service (Various). Who Took the GED?: GED Statistical Report. Washington,
DC: American Council on Higher Education.
GED Testing Service, American Council on Education (2009, December). 2008–2009 GED
Option Statistical Report. Washington, DC: GED Testing Service, American Council on
Education.
Heckman, J. J., J. E. Humphries, and N. Mader (2011). The GED. In E. A. Hanushek,
S. Machin, and L. Woßmann (Eds.), Handbook of the Economics of Education, Volume 3,
Chapter 9, pp. 423–484. Amsterdam: North Holland, Elsevier.
34
Heckman, J. J. and P. A. LaFontaine (2010, May). The American high school graduation
rate: Trends and levels. Review of Economics and Statistics 92 (2), 244–262.
Kominski, R. (1990, May). Estimating the national high school dropout rate. Demogra-
phy 27 (2), 303–311.
Lane Community College (2008). Online GED preparation, GED 2002 changes.
https://teach.lanecc.edu/ged/ged2002.htm (accessed May 9, 2008).
Lillard, D. R. (2001). Do general educational development certificate policies induce youth
out of high school? Unpublished manuscript, Cornell University.
Lillard, D. R. and P. DeCicca (2001). Higher standards, more dropouts? evidence within
and across time. Economics of Education Review 20 (5), 459–473.
National Center for Education Statistics (2005). Education longitudinal study of 2002/2004:
Base-year to first follow-up data file documentation. Technical report, National Center for
Education Statistics, Washington, DC.
National Center for Education Statistics (2006). Education Longitudinal Study of 2002
(ELS:2002/04), First Follow-up, Student Survey, 2004. Washington, DC: U.S. Depart-
ment of Education, National Center for Education Statistics.
National Center for Education Statistics (Various). Digest of Education Statistics. Wash-
ington, DC: National Center for Education Statistics.
Quinn, L. M. (1997). An institutional history of the GED. Unpublished manuscript.
Van Slyke, C. (2005). GED 2005-2006 (Kaplan). New York: Simon and Schuster.
Zhang, J., M. Han, and M. Patterson (2009a). Young GED examinees and their performance
on the GED Tests. Research Studies 2009-1. Washington, DC: GED Testing Service.
35
Zhang, J., M. Y. Han, and M. B. Patterson (2009b). Young GED examinees and their
performance on the GED tests. GED Testing Service Research Studies 2009-1, GED
Testing Service, Center for Adult Learning and Educational Credentials, American Council
on Education.
36
Web Appendix for “Taking the Easy Way Out: How the
GED Testing Program Induces Students to Drop Out”
James J. Heckman, John Eric Humphries, Paul A. LaFontaine,
and Pedro L. Rodrıguez
First draft, December 2007
Revised, May 4, 2011
Contents
A Data Sources 7
B Definitions 8
B.1 Weighted GED Test Taking Rate Across Groups: . . . . . . . . . . . . . . . 8
B.2 Overall Dropout Rate (10th-12th Grade) in state i in year t: . . . . . . . . . 9
B.3 Upper Level Dropout Rate (12th Grade) in state i in year t: . . . . . . . . . 9
B.4 Lower Level Dropout Rate (10th-11th Grade) in state i in year t: . . . . . . 9
B.5 Cohort Completion Rates (8th, 9th, or 10th) in district i in year t: . . . . . . 10
C Supplementary Materials for the 1997 Increase in Passing Standards 11
D Robustness Checks for the Effect of the 1997 Increase in Passing Standards 16
D.1 Alternate Control Group . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
D.2 Southern States Only . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
D.3 Excluding States that Changed Minimum Age Requirements . . . . . . . . . 20
D.4 Excluding Additional Years from the Analysis . . . . . . . . . . . . . . . . . 20
D.5 Extending the GED Testing and Dropout Rate Trends . . . . . . . . . . . . 22
E Fixed Effect Estimates of the Effect of Increasing Passing Standards 24
F Supplementary Material for the Analysis of the GED Option Program 29
G Additional Supplementary Materials 43
List of Figures
C-1 Average Pre- and Post-1997 10th-11th Grade Dropout Rate for Treatment
and Control Group . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
C-2 White Dropout Rates by Year, Treatment vs. Control States . . . . . . . . . 13
1
C-3 Black Dropout Rates by Year, Treatment vs. Control States . . . . . . . . . 14
C-4 Hispanic Dropout Rates by Year, Treatment vs. Control States . . . . . . . 15
D-1 GED Testing and Dropout Rates By Year, Treatment vs. Control States
(extended years) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
F-1 Median Days of Enrollment in GED Option, by State: 2008–09 School Year . 30
F-2 Median Preparation Hours of GED Option Candidates, by State: 2008–09
School Year . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
F-3 Ninth Grade Cohort Graduation Status of GED Option Candidates, by State:
2008–09 School Year . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
F-4 The Effect of Regular Schools Option Program on High School Cohort Com-
pletion Rates. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
F-5 Descriptive Comparisons of Districts with and without GED Option Programs
(2000, prior to GED Option). . . . . . . . . . . . . . . . . . . . . . . . . . . 35
F-6 Descriptive Comparisons of Districts with and without GED Option Programs
(2000, prior to GED Option) . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
G-1 Graduation Rate Before and After Implementing the GED Program, Califor-
nia vs. All other States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
List of Tables
D-1 Summary of Robustness Checks . . . . . . . . . . . . . . . . . . . . . . . . . 17
D-1 Summary of Robustness Checks (Continued) . . . . . . . . . . . . . . . . . . 18
D-1 Summary of Robustness Checks (Continued) . . . . . . . . . . . . . . . . . . 19
D-2 Alternative Year Specification for Change in Test Difficulty . . . . . . . . . . 21
E-1 Changes in GED Testing and Mandatory Schooling Age Requirements by
Treatment Status, 1994-2000 . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2
E-2 Weighted OLS Fixed Effects Estimates of the Impact of the 1997 GED Reform
on Various Dropout Rate Measures . . . . . . . . . . . . . . . . . . . . . . . 27
E-3 Summary Statistics of Variables Used in the Analysis . . . . . . . . . . . . . 28
F-1 States Issuing Credentials Indistinguishable from High School Diplomas (2008)
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
F-2 The Effect of District-Wide Option Programs on Cohort Diploma Rates in
Oregon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
F-3 The Effect of District-Wide Option Programs on Cohort Other-Completer
Rates in Oregon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
F-4 The Effect of Option Programs only in Alternative Schools on Cohort Diploma
Rates in Oregon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
F-5 The Effect of Option Programs only in Alternative Schools on Cohort Other-
Completer Rates in Oregon . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
F-6 The Effect of Option Programs in Traditional Schools on Cohort Diploma
Rates in Oregon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
F-7 The Effect of Option Programs in Traditional Schools on Cohort Other-
Completer Rates in Oregon . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
G-1 Fixed Effects Estimates of the Effect of the Reform on GED Test Taking Rates
by Younger Cohorts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
G-2 Fixed Effects Estimates of the Effect of the Reform on GED Test Taking Rates
by Younger Cohorts Controlling for Age Requirements . . . . . . . . . . . . 46
G-3 Fixed Effects Estimates of the Effect of the Reform on Dropout Rates (All
Races) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
G-4 Fixed Effects Estimates of the Effect of the Reform on Dropout Rates (Whites) 48
G-5 Fixed Effects Estimates of the Effect of the Reform on Dropout Rates (Blacks) 49
G-6 Fixed Effects Estimates of the Effect of the Reform on Dropout Rates (Hispanics) 50
3
G-7 Fixed Effects Estimates of the Effect of the Reform on Dropout Rates Con-
trolling for Age Requirements (All Races) . . . . . . . . . . . . . . . . . . . 51
G-8 Fixed Effects Estimates of the Effect of the Reform on Dropout Rates Con-
trolling for Age Requirements (Whites) . . . . . . . . . . . . . . . . . . . . . 52
G-9 Fixed Effects Estimates of the Effect of the Reform on Dropout Rates Con-
trolling for Age Requirements (Blacks . . . . . . . . . . . . . . . . . . . . . . 53
G-10 Fixed Effects Estimates of the Effect of the Reform on Dropout Rates Con-
trolling for Age Requirements (Hispanics) . . . . . . . . . . . . . . . . . . . 54
G-11 Fixed Effects Estimates of the Effect of the Reform on Dropout Rates with
Minimum Score Changer States as Control Group (All Races) . . . . . . . . 55
G-12 Fixed Effects Estimates of the Effect of the Reform on Dropout Rates with
Minimum Score Changer States as Control Group (Whites) . . . . . . . . . . 56
G-13 Fixed Effects Estimates of the Effect of the Reform on Dropout Rates with
Minimum Score Changer States as Control Group (Blacks) . . . . . . . . . . 57
G-14 Fixed Effects Estimates of the Effect of the Reform on Dropout Rates with
Minimum Score Changer States as Control Group (Hispanics) . . . . . . . . 58
G-15 Fixed Effects Estimates of the Effect of the Reform on Dropout Rates Re-
stricting Sample to Southern States (All Races) . . . . . . . . . . . . . . . . 59
G-16 Fixed Effects Estimates of the Effect of the Reform on Dropout Rates Re-
stricting Sample to Southern States (Whites) . . . . . . . . . . . . . . . . . . 60
G-17 Fixed Effects Estimates of the Effect of the Reform on Dropout Rates Re-
stricting Sample to Southern States (Blacks) . . . . . . . . . . . . . . . . . . 61
G-18 Fixed Effects Estimates of the Effect of the Reform on Dropout Rates Re-
stricting Sample to Southern States (Hispanics) . . . . . . . . . . . . . . . . 62
G-19 Fixed Effects Estimates of the Effect of the Reform on Dropout Rates Ex-
cluding States that Changed Minimum Age Required to Drop Out (All Races) 63
4
G-20 Fixed Effects Estimates of the Effect of the Reform on Dropout Rates Ex-
cluding States that Changed Minimum Age Required to Drop Out (Whites) 64
G-21 Fixed Effects Estimates of the Effect of the Reform on Dropout Rates Ex-
cluding States that Changed Minimum Age Required to Drop Out (Blacks) . 65
G-22 Fixed Effects Estimates of the Effect of the Reform on Dropout Rates Ex-
cluding States that Changed Minimum Age Required to Drop Out (Hispanics) 66
G-23 Fixed Effects Estimates of the Effect of the Reform on Dropout Rate Exclud-
ing States that Changed the Minimum Age Required to either Drop Out or
Take the GED (All Races) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
G-24 Fixed Effects Estimates of the Effect of the Reform on Dropout Rate Exclud-
ing States that Changed the Minimum Age Required to either Drop Out or
Take the GED (Whites) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
G-25 Fixed Effects Estimates of the Effect of the Reform on Dropout Rate Exclud-
ing States that Changed the Minimum Age Required to either Drop Out or
Take the GED (Blacks) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
G-26 Fixed Effects Estimates of the Effect of the Reform on Dropout Rate Exclud-
ing States that Changed the Minimum Age Required to either Drop Out or
Take the GED (Hispanics) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
G-27 GLS Fixed Effects Estimates of the Effect of the Reform on Dropout Rates
Controlling for Age Requirements (All Races) . . . . . . . . . . . . . . . . . 71
G-28 GLS Fixed Effects Estimates of the Effect of the Reform on Dropout Rates
Controlling for Age Requirements (Whites) . . . . . . . . . . . . . . . . . . . 72
G-29 GLS Fixed Effects Estimates of the Effect of the Reform on Dropout Rates
Controlling for Age Requirements (Blacks) . . . . . . . . . . . . . . . . . . . 73
G-30 GLS Fixed Effects Estimates of the Effect of the Reform on Dropout Rates
Controlling for Age Requirements (Hispanics) . . . . . . . . . . . . . . . . . 74
5
G-31 GLS Fixed Effects Estimates of the Effect of the Reform on Dropout Rates
Controlling for Age Requirements and Using Panel Specific AR-1 Autocorre-
lation Structure (All Races) . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
G-32 GLS Fixed Effects Estimates of the Effect of the Reform on Dropout Rates
Controlling for Age Requirements and Using Panel Specific AR-1 Autocorre-
lation Structure (Whites) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
G-33 GLS Fixed Effects Estimates of the Effect of the Reform on Dropout Rates
Controlling for Age Requirements and Using Panel Specific AR-1 Autocorre-
lation Structure (Blacks) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
G-34 GLS Fixed Effects Estimates of the Effect of the Reform on Dropout Rates
Controlling for Age Requirements and Using Panel Specific AR-1 Autocorre-
lation Structure (Hispanics) . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
G-35 Difference-in-Difference Estimates of the Effect of the Reform on Dropout
Rates Controlling for Age Requirements and Restricting Control Group to
California and Florida (High Immigrant States) (All Races) . . . . . . . . . . 79
G-36 Difference-in-Difference Estimates of the Effect of the Reform on Dropout
Rates Controlling for Age Requirements and Restricting Control Group to
California and Florida (High Immigrant States) (Whites) . . . . . . . . . . . 80
G-37 Difference-in-Difference Estimates of the Effect of the Reform on Dropout
Rates Controlling for Age Requirements and Restricting Control Group to
California and Florida (High Immigrant States) (Blacks) . . . . . . . . . . . 81
G-38 Difference-in-Difference Estimates of the Effect of the Reform on Dropout
Rates Controlling for Age Requirements and Restricting Control Group to
California and Florida (High Immigrant States) (Hispanics) . . . . . . . . . . 82
6
A Data Sources
This article uses the Common Core of Data (CCD) to construct annual exit rates from
secondary schooling. The CCD data are collected each year by the National Center for Edu-
cation Statistics from state and local departments of education. The data provide aggregate
annual counts of enrollments and diplomas issued (excluding GED certificates, certificates of
completion and other alternative credentials) at the state, district and even school level. We
use the aggregate state counts to construct various dropout measures using the methodology
summarized in the next section. Many states do not report estimates by race for all years.
They tend to be states that do not have large minority populations and therefore our esti-
mates should not be overly biased due to their exclusion. In a very small number of cases,
the estimated dropout rate was negative and these were set to missing. We experimented
with a number of imputation procedures to correct for missing values. These were found
not to affect our results in any substantial manner. The final measures used in the paper
do not contain imputations and all estimates by race are restricted to the same sample of
states to make the estimates comparable across groups. To be included in the analysis, states
needed to have at least two observations for each dropout measure in both the pre and post
treatment periods. It was not necessary to drop any treatment states in the analysis by race.
The estimates by race should be considered more cautiously than the overall estimates due
to these data limitations. For the analysis not by race only a few state-year observations are
missing for the treatment and control states.
A summary of all the variables used in our analysis broken down by treatment status
and time period are listed in Table E-3. GED testing rates by age at the individual state
level are obtained from multiple years of the annual GED statistical reports published by the
American Council on Education (ACE). GED age requirements by state are also from this
source. Mandatory school leaving age for each state was obtained from various years of the
Digest of Education Statistics. Annual measures at the state level of unemployment rates
and per capita income were obtained from the Bureau of Labor Statistics and the Census
7
Bureau, respectively. Population estimates at the state level for each age are obtained from
the U.S. Census Bureau. For the California analysis, population estimates were obtained
from the California Demographic Research Unit due to a lack of data available on the state
level from the Census Bureau. Diplomas issued in California and the U.S. were obtained
from various years of the Digest of Education Statistics.
District level data on the implementation of the GED Option program were collected
from unpublished administrative records from the Oregon Department of Education. These
data include which districts implemented Option programs from its introduction in the 2001-
2002 school year, through 2008. Cohort Completion rates and additional district level data
were collected from the National Center for Educational Statistics Common Core Data.
Enrollment by grade, number of diplomas issued per year, number of other completers per
year, and district-level demographics were collected from 1998 through 2008. Additional
Data from the 2000 Census incorporated into the NCES Common Core Data on poverty
rates, median family income, and per-capita income by district were also extracted
B Definitions
B.1 Weighted GED Test Taking Rate Across Groups:
Let i denote state and t denote years. The rate is
51∑i=1
G(a)i,tP (a)i,t
,
with i = 1, ..., 51 and t = 1994, ..., 2000, where
G(a)i,t = Number of GED Test Takers Age a in state i in year t.
P (a)i,t = Population Age a in state i in year t.
8
The number of states included in each sum is the number of states in groups 1 and 3 as
defined in the text, dropping any states with fewer than two observations per period.
B.2 Overall Dropout Rate (10th-12th Grade) in state i in year t:
DOi,t =
P (15− 17)i,t51∑i=1
P (15− 17)i,t
·(E(10)i,t−1 + E(11)i,t−1 + E(12)i,t−1)− (E(11)i,t + E(12)i,t +Hi,t)
(E(10)i,t−1 + E(11)i,t−1 + E(12)i,t−1),
with i = 1, ..., 51 and t = 1994, ..., 2000, where
P (15− 17)i,t = Population Age 15-17 for i, t.
E(10)i,t = Enrollment in Grade 10 for i, t.
E(11)i,t = Enrollment in Grade 10 for i, t.
E(12)i,t = Enrollment in Grade 10 for i, t.
Hi,t is the number who graduate in state i at time t. These are people who were enrolled in
school in the previous year.
B.3 Upper Level Dropout Rate (12th Grade) in state i in year t:
DUi,t =
P (15− 17)i,t51∑i=1
P (15− 17)i,t
· E(12)i,t−1 −Hi,t
E(12)i,t−1
,
with i = 1, ..., 51 and t = 1994, ..., 2000.
B.4 Lower Level Dropout Rate (10th-11th Grade) in state i in
year t:
DLi,t =
P (15− 17)i,t51∑i=1
P (15− 17)i,t
· (E(10)i,t−1 + E(11)i,t−1)− (E(11)i,t + E(12)i,t)
(E(10)i,t−1 + E(11)i,t−1),
9
with i = 1, ..., 51 and t = 1994, ..., 2000.
Weighted dropout rates by group are obtained by summing across the states in each group.
B.5 Cohort Completion Rates (8th, 9th, or 10th) in district i in
year t:
CR =Diplomasi,t
Enrollment8th,i,t−4
CR9th,i,t =Diplomasi,t
Enrollment9th,i,t−3
CR10th,i,t =Diplomasi,t
Enrollment10th,i,t−2
with t = 2000,...,2008, where:
CRjth,i,t = Completion Rates Using Base Grade j for i, t.
Diplomasi,t = Number of Diplomas Issued for i, t.
Enrollmentjth,i,t = Number of Enrolled Students in Grade j for i, t.
10
C Supplementary Materials for the 1997 Increase in
Passing Standards
11
Fig
ure
C-1
:A
vera
geP
re-
and
Pos
t-19
9710
th-1
1th
Gra
de
Dro
pou
tR
ate
for
Tre
atm
ent
and
Con
trol
Gro
up
10.8
%
8.6
%
17.0
%
16.5
%
11.2
%
8.9
%
14.0
%
15.2
%
10.5
%
8.2
%
16.0
%
14.3
%
10.5
%
8.5
%
13.4
%13
.3%
0%
2%
4%
6%
8%
10%
12%
14%
16%
18%
20%
22%
All
Races
Wh
ites
Bla
ck
sH
isp
an
ics
Dropout RateF
igu
re 8
.A
vera
ge
Pre
-an
d P
ost
-1997 1
0th
-11th
Gra
de
Dro
po
ut
Rat
e fo
r T
reat
men
t an
d C
on
tro
l G
roup
Co
ntr
ol G
roup
Pre
-97
Tre
atm
ent
Gro
up
Pre
-97
Co
ntr
ol G
roup
Po
st-9
7T
reat
men
t G
roup
Po
st-9
7
No
te: T
he
dro
po
ut
rate
is
def
ined
as
the
rati
o o
f st
ud
ents
en
rolle
d in
a g
iven
gra
de(
s) in
yea
r t
and
th
e n
um
ber
of
stud
ents
en
rolle
d in
th
e p
revio
us
grad
e(s)
in
yea
r t-
1,
wh
ere
t =
1994-2
000. A
ll es
tim
ates
are
wei
ghte
d b
y th
e 15-1
7 y
ear
old
po
pula
tio
n in
th
e gi
ven
sta
te. T
he
plo
t ab
ove
sho
ws
the
aver
age
dro
po
ut
rate
rat
e fo
r th
e p
erio
d
pre
-1997 (
i.e. 1994-1
996)
and
po
st-1
997 (
i.e. 1998-2
000).
Co
nle
y-T
aber
ad
just
edco
nfi
den
ce in
terv
als
in p
aren
thes
es.T
reat
men
t st
ates
are
th
ose
sta
tes
that
wer
e re
quir
ed t
o
elim
inat
e th
e an
d/
or
sco
re o
pti
on
. T
hes
e in
clud
e: L
A, M
S, N
E, N
M, T
X. C
on
tro
l st
ates
are
th
ose
th
at a
lrea
dy
had
hig
h e
no
ugh
sta
nd
ard
s b
y 1997. T
hes
e in
clud
e: A
R,
CA
, CO
, DE
, D
C, F
L, ID
, K
Y, M
D, M
O, N
J, N
Y, N
D, O
K, O
R, S
D, U
T, W
A, W
V, W
I. S
tate
s w
ith
few
er t
han
tw
o o
bse
rvat
ion
s p
er p
erio
d a
re d
rop
ped
fo
r 'a
ll ra
ces'
ca
tego
ry. Sta
tes
wit
h f
ewer
th
an t
wo
ob
serv
atio
ns
per
per
iod
fo
r an
y o
f th
e d
rop
out
rate
mea
sure
s b
y ra
ce a
re d
rop
ped
fo
r b
y ra
ce c
ateg
ori
es. C
on
tro
l st
ates
dro
pp
ed f
rom
'a
ll ra
ces'
reg
ress
ion
s d
ue
to m
issi
ng
and
neg
ativ
e d
rop
out
rate
s in
clud
e: N
J. C
on
tro
l st
ates
dro
pp
ed f
rom
reg
ress
ion
by
race
due
to m
issi
ng
and
neg
ativ
e d
rop
out
rate
s in
clud
e: A
R, ID
, K
Y, M
O, N
D, N
J, N
Y, SD
, U
T, W
A, W
V. N
o t
reat
men
t st
ates
are
dro
pp
ed f
rom
an
y re
gres
sio
ns.
Sin
ce t
her
e ar
e m
ore
mis
sin
gs in
th
e d
rop
out
rate
s b
y ra
ce, th
e 'a
ll ra
ces'
cat
ego
ry is
no
t d
irec
tly
com
par
able
to
th
e ca
tego
ries
by
race
. So
urc
e: C
om
mo
n C
ore
of
Dat
a (C
CD
).
DiD
Est
imate
-0
.42%
(-1.
87%
,
0.7
4%
)
DiD
Est
imate
0.0
2%
( -1
.13%
, 1.
64%
)
DiD
Est
imate
0.3
6%
(-1.
88%
, 8.3
9%
)
DiD
Est
imate
0.2
7%
(-13
.38%
, 2.5
7%
)
Note
:T
he
dro
pou
tra
teis
defi
ned
as
the
rati
oof
stud
ents
enro
lled
ina
giv
engra
de(
s)in
yea
rt
an
dth
enu
mb
erof
stu
den
tsen
roll
edin
the
pre
vio
us
gra
de(
s)in
yea
rt−
1,
wh
eret=
1994-2
000.
All
esti
mate
sare
wei
ghte
dby
the
15-1
7yea
rold
pop
ula
tion
inth
egiv
enst
ate
.T
he
plo
tab
ove
show
sth
eaver
age
dro
pou
tra
tefo
rth
ep
erio
dp
re-1
997
(i.e
.1994-1
996)
an
dp
ost
-1997
(i.e
.1998-2
000).
Con
ley-T
ab
erad
just
edco
nfi
den
cein
terv
als
inp
are
nth
eses
.T
reatm
ent
state
sare
those
state
sth
at
wer
ere
qu
ired
toel
imin
ate
the
an
d/or
score
op
tion
.T
hes
ein
clu
de:
LA
,M
S,
NE
,N
M,
TX
.C
ontr
ol
state
sare
those
that
alr
ead
yh
ad
hig
hen
ou
gh
stan
dard
sby
1997.
Th
ese
incl
ud
e:A
R,
CA
,C
O,
DE
,
DC
,F
L,
ID,
KY
,M
D,
MO
,N
J,
NY
,N
D,
OK
,O
R,
SD
,U
T,
WA
,W
V,
WI.
Sta
tes
wit
hfe
wer
than
two
ob
serv
ati
on
sp
erp
erio
dare
dro
pp
edfo
r’a
llra
ces’
cate
gory
.S
tate
sw
ith
few
erth
an
two
ob
serv
ati
ons
per
per
iod
for
any
of
the
dro
pou
tra
tem
easu
res
by
race
are
dro
pp
edfo
rby
race
cate
gori
es.
Contr
ol
state
sd
rop
ped
from
’all
race
s’re
gre
ssio
ns
du
e
tom
issi
ng
an
dn
egati
ve
dro
pou
tra
tes
incl
ud
e:N
J.
Contr
ol
state
sd
rop
ped
from
regre
ssio
nby
race
du
eto
mis
sin
gan
dn
egati
ve
dro
pou
tra
tes
incl
ud
e:A
R,
ID,
KY
,M
O,
ND
,
NJ,
NY
,S
D,
UT
,W
A,
WV
.N
otr
eatm
ent
state
sare
dro
pp
edfr
om
any
regre
ssio
ns.
Sin
ceth
ere
are
more
mis
sin
gs
inth
ed
rop
ou
tra
tes
by
race
,th
e’a
llra
ces’
cate
gory
isn
ot
dir
ectl
yco
mp
ara
ble
toth
eca
tegori
esby
race
.S
ou
rce:
Com
mon
Core
of
Data
(CC
D).
12
Fig
ure
C-2
:W
hit
eD
rop
out
Rat
esby
Yea
r,T
reat
men
tvs.
Con
trol
Sta
tes
Note
:G
ED
test
ing
rate
sare
calc
ula
ted
from
yea
rly
GE
DS
tati
stic
al
Rep
ort
sas
the
per
centa
ge
of
the
state
pop
ula
tion
inth
egiv
enage
ran
ge
wh
ota
ke
the
GE
Din
that
yea
r.
Dro
pou
tra
tes
are
calc
ula
ted
from
the
Com
mon
Core
of
Data
(CC
D)
as
the
exit
rate
for
those
inth
ein
dic
ate
dgra
des
inth
egiv
enyea
r.S
eeth
eap
pen
dix
for
furt
her
det
ails.
Sta
tes
requ
ired
tora
ise
GE
Dp
ass
requ
irem
ents
(ch
an
ger
state
s)are
:L
A,
MS
,N
E,
NM
,T
X.
Sta
tes
that
did
not
chan
ge
pass
requ
irem
ents
(non
-ch
an
ger
state
s)are
:A
R,
CA
,
CO
,D
E,
DC
,F
L,
ID,
KY
,M
D,
MO
,N
J,
NY
,N
D,
OK
,O
R,
SD
,U
T,
WA
,W
V,
WI.
NJ
isex
clu
ded
inall
dro
pou
tca
lcu
lati
on
sd
ue
tod
ata
erro
rs.
13
Fig
ure
C-3
:B
lack
Dro
pou
tR
ates
by
Yea
r,T
reat
men
tvs.
Con
trol
Sta
tes
Note
:G
ED
test
ing
rate
sare
calc
ula
ted
from
yea
rly
GE
DS
tati
stic
al
Rep
ort
sas
the
per
centa
ge
of
the
state
pop
ula
tion
inth
egiv
enage
ran
ge
wh
ota
ke
the
GE
Din
that
yea
r.
Dro
pou
tra
tes
are
calc
ula
ted
from
the
Com
mon
Core
of
Data
(CC
D)
as
the
exit
rate
for
those
inth
ein
dic
ate
dgra
des
inth
egiv
enyea
r.S
eeth
eap
pen
dix
for
furt
her
det
ails.
Sta
tes
requ
ired
tora
ise
GE
Dp
ass
requ
irem
ents
(ch
an
ger
state
s)are
:L
A,
MS
,N
E,
NM
,T
X.
Sta
tes
that
did
not
chan
ge
pass
requ
irem
ents
(non
-ch
an
ger
state
s)are
:A
R,
CA
,
CO
,D
E,
DC
,F
L,
ID,
KY
,M
D,
MO
,N
J,
NY
,N
D,
OK
,O
R,
SD
,U
T,
WA
,W
V,
WI.
NJ
isex
clu
ded
inall
dro
pou
tca
lcu
lati
on
sd
ue
tod
ata
erro
rs.
14
Fig
ure
C-4
:H
ispan
icD
rop
out
Rat
esby
Yea
r,T
reat
men
tvs.
Con
trol
Sta
tes
Note
:G
ED
test
ing
rate
sare
calc
ula
ted
from
yea
rly
GE
DS
tati
stic
al
Rep
ort
sas
the
per
centa
ge
of
the
state
pop
ula
tion
inth
egiv
enage
ran
ge
wh
ota
ke
the
GE
Din
that
yea
r.
Dro
pou
tra
tes
are
calc
ula
ted
from
the
Com
mon
Core
of
Data
(CC
D)
as
the
exit
rate
for
those
inth
ein
dic
ate
dgra
des
inth
egiv
enyea
r.S
eeth
eap
pen
dix
for
furt
her
det
ails.
Sta
tes
requ
ired
tora
ise
GE
Dp
ass
requ
irem
ents
(ch
an
ger
state
s)are
:L
A,
MS
,N
E,
NM
,T
X.
Sta
tes
that
did
not
chan
ge
pass
requ
irem
ents
(non
-ch
an
ger
state
s)are
:A
R,
CA
,
CO
,D
E,
DC
,F
L,
ID,
KY
,M
D,
MO
,N
J,
NY
,N
D,
OK
,O
R,
SD
,U
T,
WA
,W
V,
WI.
NJ
isex
clu
ded
inall
dro
pou
tca
lcu
lati
on
sd
ue
tod
ata
erro
rs.
15
D Robustness Checks for the Effect of the 1997 In-
crease in Passing Standards
This section reports alternative specifications of our model in the main text to test the
robustness of the results (Table D-1). As in the main text, we only report the γ estimates
for each check. For the full set of parameter estimates please refer to the Web Appendix.
D.1 Alternate Control Group
As one check of the exogeneity of the policy change assumption, we re-estimate the model
using states that were required to raise the GED minimum score requirement rather than
states that did not change. These are the lightly shaded states in Figure 4 in the text.
These states were also required to change GED policies but the increase in difficulty was
much smaller.
The first row of Table D-1 summarizes our overall results and results by race using this
alternate control group. The estimated effect on the upper level dropout rate is in general
larger than the effect obtained from our main control group. On the other hand, the effect
on the lower level dropout rate is in general smaller except for whites. However, these results
are generally consistent with the results reported in the text.
D.2 Southern States Only
With the exception of Nebraska, all treatment group states are located in the South. This
suggests that while the timing of the score requirement change was exogenous, the states
that were required to change were not a random sample of states. States likely set GED
standards endogenously to reflect conditions in the state, i.e. states with traditionally higher
dropout rates have lower GED testing standards. As a further robustness check of our main
results, we estimate the model using only treatment and control states located in the South.
The estimates, shown in the second row of Table D-1, are very similar to those reported
16
Tab
leD
-1:
Sum
mar
yof
Rob
ust
nes
sC
hec
ks
All
Rac
esW
hit
esB
lack
sH
isp
anic
sTr
eatm
ent E
ffect
-1.5
3%-1
.08%
-1.9
0%-2
.67%
95%
CI (
Hub
er-W
hite
)(-2
.37%
, -0.
68%
)(-1
.92%
, -0.
24%
)(-4
.00%
, 0.2
1%)
(-4.8
7%, -
0.47
%)
95%
CI (
Conl
ey-T
aber
)(-2
.19%
, -0.
49%
)(-3
.40%
, 0.1
5%)
(-4.3
9%, 2
.06%
)(-6
.17%
, -0.
98%
)90
% C
I (Co
nley
-Tab
er)
(-2.0
5%, -
0.64
%)
(-3.0
2%, -
0.02
%)
(-3.8
6%, 1
.40%
)(-5
.68%
, -1.
33%
)Tr
eatm
ent E
ffect
-1.4
5%-0
.42%
-2.1
6%-3
.08%
95%
CI (
Hub
er-W
hite
)(-1
.91%
, -1.
00%
)(-0
.75%
, -0.
10%
)(-3
.23%
, -1.
09%
)(-5
.51%
, -0.
65%
)95
% C
I (Co
nley
-Tab
er)
(-2.0
4%, 0
.06%
)(-0
.79%
, 2.8
2%)
(-3.3
9%, -
1.18
%)
(-7.5
3%, -
0.72
%)
90%
CI (
Conl
ey-T
aber
)(-1
.91%
, -0.
14%
)(-0
.65%
, 1.5
2%)
(-2.9
6%, -
1.29
%)
(-7.2
3%, -
0.99
%)
Trea
tmen
t Effe
ct-1
.30%
-0.4
2%-1
.25%
-2.5
8%95
% C
I (H
uber
-Whi
te)
(-2.0
6%, -
0.53
%)
(-0.7
7%, -
0.07
%)
(-3.4
2%, 0
.92%
)(-3
.47%
, -1.
70%
)95
% C
I (Co
nley
-Tab
er)
(-2.5
6%, -
0.37
%)
(-0.9
0%, 0
.15%
)(-3
.00%
, 3.4
7%)
(-3.7
2%, 0
.48%
)90
% C
I (Co
nley
-Tab
er)
(-2.3
0%, -
0.56
%)
(-0.8
0%, 0
.03%
)(-2
.69%
, 2.6
0%)
(-3.5
6%, -
0.06
%)
Not
e: Th
e fu
ll re
gres
sion
spec
ifica
tions
are
pre
sent
ed in
the
web
app
endi
x. U
nles
s oth
erw
ise st
ated
the
mod
el is
estim
ated
usin
g O
LS. S
tate
15-
17 y
ear o
ld p
opul
atio
ns
(by
race
for r
eleva
nt re
gres
sions
) are
use
d as
weig
hts.
The
estim
ate
repo
rted
abov
e is
the
inte
ract
ion
betw
een
the
treat
men
t sta
te d
umm
y an
d th
e po
st p
erio
d du
mm
y, w
here
the
treat
men
t sta
te d
umm
y is
equa
l to
1 fo
r tre
atm
ent s
tate
s and
the
post
per
iod
dum
my
is eq
ual t
o 1
for t
he y
ears
199
8-20
00, o
ther
wise
bot
h du
mm
ies a
re e
qual
to 0
. Sta
tes w
ith fe
wer
than
two
obse
rvat
ions
per
per
iod
are
drop
ped
for "
all ra
ces"
regr
essio
ns. S
tate
s with
few
er th
an tw
o ob
serv
atio
ns p
er p
erio
d fo
r any
of t
he
drop
out r
ate
mea
sure
s by
race
are
dro
pped
for b
y ra
ce re
gres
sions
. Tre
atm
ent s
tate
s are
thos
e st
ates
that
wer
e re
quire
d to
elim
inat
e th
e an
d/or
scor
e op
tion.
The
se
inclu
de: L
A, M
S, N
E, N
M, T
X. U
nles
s oth
erw
ise st
ated
con
trol s
tate
s are
thos
e th
at w
ere
not r
equi
red
to ra
ise th
eir G
ED
min
imum
scor
e re
quire
men
t. Th
ese
inclu
de:
AR,
CA
, CO
, DE
, DC,
FL,
ID, K
Y, M
D, M
O, N
J, N
Y, N
D, O
K, O
R, S
D, U
T, W
A, W
V, W
I. Fr
om th
ese
stat
es th
e fo
llow
ing
had
to b
e dr
oppe
d fr
om th
e "b
y ra
ce"
regr
essio
ns d
ue to
miss
ing
and
nega
tive
drop
out r
ates
: AR,
ID, K
Y, M
O, N
D, N
J, N
Y, S
D, U
T, W
A, W
V. T
he st
ate
of N
J is a
lso d
ropp
ed fr
om th
e "a
ll ra
ces"
re
gres
sions
. Sco
re c
hang
er st
ates
are
thos
e st
ates
that
wer
e re
quire
d to
raise
their
min
imum
scor
e re
quire
men
t fro
m 3
5 to
40
in 1
997.
The
se in
clude
: AK
, AL,
AZ
, CT,
G
A, H
I, IA
, IL,
IN, K
S, M
A, M
E, M
I, M
N, M
T, N
C, N
H, N
V, O
H, P
A, R
I, SC
, TN
, VA
, VT,
WY
. Fro
m th
ese
stat
es th
e fo
llow
ing
are
drop
ped
due
to m
issin
g an
d ne
gativ
e dr
opou
t rat
es: A
L, A
Z, I
A, M
E, M
N, M
T, N
H, S
C, T
N, V
T, W
Y. N
o tre
atm
ent s
tate
s are
dro
pped
from
any
of t
he re
gres
sions
. Sta
tes t
hat c
hang
ed th
e m
inim
um a
ge re
quire
d to
dro
p ou
t inc
lude
: DC
(from
orig
inal
cont
rol g
roup
), M
S an
d N
M (b
oth
from
trea
tmen
t gro
up).
Stat
es th
at c
hang
ed e
ither
scho
ol le
avin
g or
G
ED
age
requ
irem
ents
inclu
de:
AR,
DC,
KY
, MO
, OK
, OR,
SD
, UT,
WI (
from
orig
inal
cont
rol g
roup
) and
MS,
NE
and
NM
(fro
m tr
eatm
ent g
roup
).
Exc
ludi
ng st
ates
that
ch
ange
d m
inim
um a
ge
requ
ired
to d
rop
out
10th
-12t
h G
rad
e D
rop
out
Rat
e
Sout
hern
stat
es o
nly
Scor
e ch
ange
r sta
tes a
s co
ntro
l gro
up
Note
:T
he
full
regre
ssio
nsp
ecifi
cati
on
sare
pre
sente
din
the
Web
Ap
pen
dix
.U
nle
ssoth
erw
ise
state
dth
em
od
elis
esti
mate
du
sin
gO
LS
.S
tate
15-1
7yea
rold
pop
ula
tion
s(b
y
race
for
rele
vant
regre
ssio
ns)
are
use
das
wei
ghts
.T
he
esti
mate
rep
ort
edab
ove
isth
ein
tera
ctio
nb
etw
een
the
trea
tmen
tst
ate
du
mm
yan
dth
ep
ost
per
iod
du
mm
y,w
her
eth
e
trea
tmen
tst
ate
du
mm
yis
equ
al
to1
for
trea
tmen
tst
ate
san
dth
ep
ost
per
iod
du
mm
yis
equ
al
to1
for
the
yea
rs1998-2
000,
oth
erw
ise
both
du
mm
ies
are
equ
al
to0.
Sta
tes
wit
h
few
erth
an
two
ob
serv
ati
on
sp
erp
erio
dare
dro
pp
edfo
r“all
race
s”re
gre
ssio
ns.
Sta
tes
wit
hfe
wer
than
two
ob
serv
ati
on
sp
erp
erio
dfo
rany
of
the
dro
pou
tra
tem
easu
res
by
race
are
dro
pp
edfo
rby
race
regre
ssio
ns.
Tre
atm
ent
state
sare
those
state
sth
at
wer
ere
qu
ired
toel
imin
ate
the
an
d/or
score
op
tion
.T
hes
ein
clu
de:
LA
,M
S,
NE
,N
M,
TX
.U
nle
ss
oth
erw
ise
state
dco
ntr
ol
state
sare
those
that
wer
en
ot
requ
ired
tora
ise
thei
rG
ED
min
imu
msc
ore
requ
irem
ent.
Th
ese
incl
ud
e:A
R,
CA
,C
O,
DE
,D
C,
FL
,ID
,K
Y,
MD
,M
O,
NJ,
NY
,N
D,
OK
,O
R,
SD
,U
T,
WA
,W
V,
WI.
Fro
mth
ese
state
sth
efo
llow
ing
had
tob
ed
rop
ped
from
the
”by
race
”re
gre
ssio
ns
du
eto
mis
sin
gan
dn
egati
ve
dro
pou
tra
tes:
AR
,ID
,K
Y,
MO
,N
D,
NJ,
NY
,S
D,
UT
,W
A,
WV
.T
he
state
of
NJ
isals
od
rop
ped
from
the
”all
race
s”re
gre
ssio
ns.
Sco
rech
an
ger
state
sare
those
state
sth
at
wer
ere
qu
ired
to
rais
eth
eir
min
imu
msc
ore
requ
irem
ent
from
35
to40
in1997.
Thes
ein
clu
de:
AK
,A
L,
AZ
,C
T,
GA
,H
I,IA
,IL
,IN
,K
S,
MA
,M
E,
MI,
MN
,M
T,
NC
,N
H,
NV
,O
H,
PA
,R
I,S
C,
TN
,V
A,
VT
,W
Y.
Fro
mth
ese
state
sth
efo
llow
ing
are
dro
pp
edd
ue
tom
issi
ng
an
dn
egati
ve
dro
pou
tra
tes:
AL
,A
Z,
IA,
ME
,M
N,
MT
,N
H,
SC
,T
N,
VT
,W
Y.
No
trea
tmen
t
state
sare
dro
pp
edfr
om
any
of
the
regre
ssio
ns.
Sta
tes
that
chan
ged
the
min
imu
mage
requ
ired
tod
rop
ou
tin
clu
de:
DC
(fro
mori
gin
al
contr
ol
gro
up
),M
San
dN
M(b
oth
from
trea
tmen
tgro
up
).S
tate
sth
at
chan
ged
eith
ersc
hool
leavin
gor
GE
Dage
requ
irem
ents
incl
ud
e:A
R,
DC
,K
Y,
MO
,O
K,
OR
,S
D,
UT
,W
I(f
rom
ori
gin
al
contr
ol
gro
up
)an
dM
S,
NE
an
dN
M(f
rom
trea
tmen
tgro
up
).
17
Tab
leD
-1:
Sum
mar
yof
Rob
ust
nes
sC
hec
ks
(Con
tinued
)
All
Rac
esW
hit
esB
lack
sH
isp
anic
sTr
eatm
ent E
ffect
-1.5
3%-1
.08%
-1.9
0%-2
.67%
95%
CI (
Hub
er-W
hite
)(-2
.37%
, -0.
68%
)(-1
.92%
, -0.
24%
)(-4
.00%
, 0.2
1%)
(-4.8
7%, -
0.47
%)
95%
CI (
Conl
ey-T
aber
)(-2
.19%
, -0.
49%
)(-3
.40%
, 0.1
5%)
(-4.3
9%, 2
.06%
)(-6
.17%
, -0.
98%
)90
% C
I (Co
nley
-Tab
er)
(-2.0
5%, -
0.64
%)
(-3.0
2%, -
0.02
%)
(-3.8
6%, 1
.40%
)(-5
.68%
, -1.
33%
)Tr
eatm
ent E
ffect
-1.4
5%-0
.42%
-2.1
6%-3
.08%
95%
CI (
Hub
er-W
hite
)(-1
.91%
, -1.
00%
)(-0
.75%
, -0.
10%
)(-3
.23%
, -1.
09%
)(-5
.51%
, -0.
65%
)
10th
-12t
h G
rad
e D
rop
out
Rat
e
Sout
hern
stat
es o
nly
Scor
e ch
ange
r sta
tes a
s co
ntro
l gro
up
95%
C(H
beW
te)
(.9
%,
.00%
)(0
.75%
,0.
0%)
(3.
3%,
.09%
)(5
.5%
,0.
65%
)95
% C
I (Co
nley
-Tab
er)
(-2.0
4%, 0
.06%
)(-0
.79%
, 2.8
2%)
(-3.3
9%, -
1.18
%)
(-7.5
3%, -
0.72
%)
90%
CI (
Conl
ey-T
aber
)(-1
.91%
, -0.
14%
)(-0
.65%
, 1.5
2%)
(-2.9
6%, -
1.29
%)
(-7.2
3%, -
0.99
%)
Trea
tmen
t Effe
ct-1
.30%
-0.4
2%-1
.25%
-2.5
8%95
% C
I (H
uber
-Whi
te)
(-2.0
6%, -
0.53
%)
(-0.7
7%, -
0.07
%)
(-3.4
2%, 0
.92%
)(-3
.47%
, -1.
70%
)95
% C
I (Co
nley
-Tab
er)
(-2.5
6%, -
0.37
%)
(-0.9
0%, 0
.15%
)(-3
.00%
, 3.4
7%)
(-3.7
2%, 0
.48%
)90
% C
I (Co
nley
-Tab
er)
(-2.3
0%, -
0.56
%)
(-0.8
0%, 0
.03%
)(-2
.69%
, 2.6
0%)
(-3.5
6%, -
0.06
%)
Not
e: Th
e fu
ll re
gres
sion
spec
ifica
tions
are
pre
sent
ed in
the
web
app
endi
x. U
nles
s oth
erw
ise st
ated
the
mod
el is
estim
ated
usin
g O
LS. S
tate
15-
17 y
ear o
ld p
opul
atio
ns
(by
race
for r
eleva
nt re
gres
sions
) are
use
d as
weig
hts.
The
estim
ate
repo
rted
abov
e is
the
inte
ract
ion
betw
een
the
treat
men
t sta
te d
umm
y an
d th
e po
st p
erio
d du
mm
y, w
here
the
treat
men
t sta
te d
umm
y is
equa
l to
1 fo
r tre
atm
ent s
tate
s and
the
post
per
iod
dum
my
is eq
ual t
o 1
for t
he y
ears
199
8-20
00, o
ther
wise
bot
h du
mm
ies a
re e
qual
0S
ihf
hb
ii
dd
df
"ll
"i
Sih
fh
bi
id
ff
h
Exc
ludi
ng st
ates
that
ch
ange
d m
inim
um a
ge
requ
ired
to d
rop
out
to 0
. Sta
tes w
ith fe
wer
than
two
obse
rvat
ions
per
per
iod
are
drop
ped
for "
all ra
ces"
regr
essio
ns. S
tate
s with
few
er th
an tw
o ob
serv
atio
ns p
er p
erio
d fo
r any
of t
he
drop
out r
ate
mea
sure
s by
race
are
dro
pped
for b
y ra
ce re
gres
sions
. Tre
atm
ent s
tate
s are
thos
e st
ates
that
wer
e re
quire
d to
elim
inat
e th
e an
d/or
scor
e op
tion.
The
se
inclu
de: L
A, M
S, N
E, N
M, T
X. U
nles
s oth
erw
ise st
ated
con
trol s
tate
s are
thos
e th
at w
ere
not r
equi
red
to ra
ise th
eir G
ED
min
imum
scor
e re
quire
men
t. Th
ese
inclu
de:
AR,
CA
, CO
, DE
, DC,
FL,
ID, K
Y, M
D, M
O, N
J, N
Y, N
D, O
K, O
R, S
D, U
T, W
A, W
V, W
I. Fr
om th
ese
stat
es th
e fo
llow
ing
had
to b
e dr
oppe
d fr
om th
e "b
y ra
ce"
regr
essio
ns d
ue to
miss
ing
and
nega
tive
drop
out r
ates
: AR,
ID, K
Y, M
O, N
D, N
J, N
Y, S
D, U
T, W
A, W
V. T
he st
ate
of N
J is a
lso d
ropp
ed fr
om th
e "a
ll ra
ces"
re
gres
sions
. Sco
re c
hang
er st
ates
are
thos
e st
ates
that
wer
e re
quire
d to
raise
their
min
imum
scor
e re
quire
men
t fro
m 3
5 to
40
in 1
997.
The
se in
clude
: AK
, AL,
AZ
, CT,
G
A, H
I, IA
, IL,
IN, K
S, M
A, M
E, M
I, M
N, M
T, N
C, N
H, N
V, O
H, P
A, R
I, SC
, TN
, VA
, VT,
WY
. Fro
m th
ese
stat
es th
e fo
llow
ing
are
drop
ped
due
to m
issin
g an
d ne
gativ
e dr
opou
t rat
es: A
L, A
Z, I
A, M
E, M
N, M
T, N
H, S
C, T
N, V
T, W
Y. N
o tre
atm
ent s
tate
s are
dro
pped
from
any
of t
he re
gres
sions
. Sta
tes t
hat c
hang
ed th
e m
inim
um a
ge re
quire
d to
dro
p ou
t inc
lude
: DC
(from
orig
inal
cont
rol g
roup
), M
S an
d N
M (b
oth
from
trea
tmen
t gro
up).
Stat
es th
at c
hang
ed e
ither
scho
ol le
avin
g or
G
ED
age
requ
irem
ents
inclu
de:
AR,
DC,
KY
, MO
, OK
, OR,
SD
, UT,
WI (
from
orig
inal
cont
rol g
roup
) and
MS,
NE
and
NM
(fro
m tr
eatm
ent g
roup
).
10th
-11t
hG
rad
eD
rop
out
Rat
e
All
Rac
esW
hit
esB
lack
sH
isp
anic
s
Trea
tmen
t Effe
ct-0
.23%
-0.3
9%0.
25%
-1.0
9%95
% C
I (H
uber
-Whi
te)
(-0.9
1%, 0
.47%
)(-1
.10%
, 0.3
2%)
(-2.7
4%, 3
.25%
)(-4
.33%
, 2.1
5%)
95%
CI (
Conl
ey-T
aber
)(-1
.47%
, 0.6
6%)
(-3.2
2%, 0
.52%
)(-2
.43%
, 6.1
2%)
(-4.8
2%, 0
.92%
)90
% C
I (Co
nley
-Tab
er)
(-1.3
1%, 0
.47%
)(-2
.75%
, 0.3
9%)
(-2.0
1%, 4
.83%
)(-4
.34%
, 0.4
8%)
Trea
tmen
t Effe
ct-0
.95%
-0.5
3%-1
.59%
-1.8
9%95
% C
I (H
uber
-Whi
te)
(-1.5
2%, -
0.38
%)
(-1.5
6%, 0
.50%
)(-3
.67%
, 0.4
8%)
(-5.1
6%, 1
.39%
)95
%CI
(Cl
Tb
)(1
83%
086
%)
(126
%1
77%
)(3
59%
015
%)
(906
%0
14%
)
Scor
e ch
ange
r sta
tes a
s co
ntro
l gro
up
Sout
hern
stat
es o
nly
10th
-11t
h G
rad
e D
rop
out
Rat
e
95%
CI (
Conl
ey-T
aber
)(-1
.83%
, 0.8
6%)
(-1.2
6%, 1
.77%
)(-3
.59%
, -0.
15%
)(-9
.06%
, 0.1
4%)
90%
CI (
Conl
ey-T
aber
)(-1
.71%
, 0.5
6%)
(-1.1
1%, 1
.11%
)(-3
.09%
, -0.
52%
)(-8
.18%
, -0.
34%
)Tr
eatm
ent E
ffect
-0.5
6%0.
01%
0.13
%-1
.25%
95%
CI (
Hub
er-W
hite
)(-1
.68%
, 0.5
7%)
(-1.0
3%, 1
.05%
)(-3
.49%
, 3.7
4%)
(-2.4
0%, -
0.09
%)
95%
CI (
Conl
ey-T
aber
)(-1
.91%
, 0.6
11%
)(-0
.95%
, 1.3
9%)
(-2.7
6%, 7
.41%
)-3
.73%
, 2.6
6%)
90%
CI (
Conl
ey-T
aber
)(-1
.73%
, 0.3
9%)
(-0.8
8%, 0
.99%
)(-2
.50%
, 5.0
0%)
(-3.3
1%, 1
.63%
)N
ote:
The
full
regr
essio
n sp
ecifi
catio
ns a
re p
rese
nted
in th
e w
eb a
ppen
dix.
Unl
ess o
ther
wise
stat
ed th
e m
odel
is es
timat
ed u
sing
OLS
. Sta
te 1
5-17
yea
r old
pop
ulat
ions
(b
y ra
ce fo
r rele
vant
regr
essio
ns) a
re u
sed
as w
eight
s. Th
e es
timat
e re
porte
d ab
ove
is th
e in
tera
ctio
n be
twee
n th
e tre
atm
ent s
tate
dum
my
and
the
post
per
iod
dum
my,
whe
re th
e tre
atm
ent s
tate
dum
my
is eq
ual t
o 1
for t
reat
men
t sta
tes a
nd th
e po
st p
erio
d du
mm
y is
equa
l to
1 fo
r the
yea
rs 1
998-
2000
, oth
erw
ise b
oth
dum
mies
are
equ
al
Exc
ludi
ng st
ates
that
ch
ange
d m
inim
um a
ge
requ
ired
to d
rop
out
to 0
. Sta
tes w
ith fe
wer
than
two
obse
rvat
ions
per
per
iod
are
drop
ped
for "
all ra
ces"
regr
essio
ns. S
tate
s with
few
er th
an tw
o ob
serv
atio
ns p
er p
erio
d fo
r any
of t
he
drop
out r
ate
mea
sure
s by
race
are
dro
pped
for b
y ra
ce re
gres
sions
. Tre
atm
ent s
tate
s are
thos
e st
ates
that
wer
e re
quire
d to
elim
inat
e th
e an
d/or
scor
e op
tion.
The
se
inclu
de: L
A, M
S, N
E, N
M, T
X. U
nles
s oth
erw
ise st
ated
con
trol s
tate
s are
thos
e th
at w
ere
not r
equi
red
to ra
ise th
eir G
ED
min
imum
scor
e re
quire
men
t. Th
ese
inclu
de:
AR,
CA
, CO
, DE
, DC,
FL,
ID, K
Y, M
D, M
O, N
J, N
Y, N
D, O
K, O
R, S
D, U
T, W
A, W
V, W
I. Fr
om th
ese
stat
es th
e fo
llow
ing
had
to b
e dr
oppe
d fr
om th
e "b
y ra
ce"
regr
essio
ns d
ue to
miss
ing
and
nega
tive
drop
out r
ates
: AR,
ID, K
Y, M
O, N
D, N
J, N
Y, S
D, U
T, W
A, W
V. T
he st
ate
of N
J is a
lso d
ropp
ed fr
om th
e "a
ll ra
ces"
re
gres
sions
. Sco
re c
hang
er st
ates
are
thos
e st
ates
that
wer
e re
quire
d to
raise
their
min
imum
scor
e re
quire
men
t fro
m 3
5 to
40
in 1
997.
The
se in
clude
: AK
, AL,
AZ
, CT,
G
A, H
I, IA
, IL,
IN, K
S, M
A, M
E, M
I, M
N, M
T, N
C, N
H, N
V, O
H, P
A, R
I, SC
, TN
, VA
, VT,
WY
. Fro
m th
ese
stat
es th
e fo
llow
ing
are
drop
ped
due
to m
issin
g an
d ne
gativ
e dr
opou
t rat
es: A
L, A
Z, I
A, M
E, M
N, M
T, N
H, S
C, T
N, V
T, W
Y. N
o tre
atm
ent s
tate
s are
dro
pped
from
any
of t
he re
gres
sions
. Sta
tes t
hat c
hang
ed th
e m
inim
um a
ge re
quire
d to
dro
p ou
t inc
lude
: DC
(from
orig
inal
cont
rol g
roup
), M
S an
d N
M (b
oth
from
trea
tmen
t gro
up).
Stat
es th
at c
hang
ed e
ither
scho
ol le
avin
g or
G
ED
age
requ
irem
ents
inclu
de:
AR,
DC,
KY
, MO
, OK
, OR,
SD
, UT,
WI (
from
orig
inal
cont
rol g
roup
) and
MS,
NE
and
NM
(fro
m tr
eatm
ent g
roup
).
Note
:T
he
full
regre
ssio
nsp
ecifi
cati
on
sare
pre
sente
din
the
Web
Ap
pen
dix
.U
nle
ssoth
erw
ise
state
dth
em
od
elis
esti
mate
du
sin
gO
LS
.S
tate
15-1
7yea
rold
pop
ula
tion
s(b
y
race
for
rele
vant
regre
ssio
ns)
are
use
das
wei
ghts
.T
he
esti
mate
rep
ort
edab
ove
isth
ein
tera
ctio
nb
etw
een
the
trea
tmen
tst
ate
du
mm
yan
dth
ep
ost
per
iod
du
mm
y,w
her
eth
e
trea
tmen
tst
ate
du
mm
yis
equ
al
to1
for
trea
tmen
tst
ate
san
dth
ep
ost
per
iod
du
mm
yis
equ
al
to1
for
the
yea
rs1998-2
000,
oth
erw
ise
both
du
mm
ies
are
equ
al
to0.
Sta
tes
wit
h
few
erth
an
two
ob
serv
ati
on
sp
erp
erio
dare
dro
pp
edfo
r“all
race
s”re
gre
ssio
ns.
Sta
tes
wit
hfe
wer
than
two
ob
serv
ati
on
sp
erp
erio
dfo
rany
of
the
dro
pou
tra
tem
easu
res
by
race
are
dro
pp
edfo
rby
race
regre
ssio
ns.
Tre
atm
ent
state
sare
those
state
sth
at
wer
ere
qu
ired
toel
imin
ate
the
an
d/or
score
op
tion
.T
hes
ein
clu
de:
LA
,M
S,
NE
,N
M,
TX
.U
nle
ss
oth
erw
ise
state
dco
ntr
ol
state
sare
those
that
wer
en
ot
requ
ired
tora
ise
thei
rG
ED
min
imu
msc
ore
requ
irem
ent.
Th
ese
incl
ud
e:A
R,
CA
,C
O,
DE
,D
C,
FL
,ID
,K
Y,
MD
,M
O,
NJ,
NY
,N
D,
OK
,O
R,
SD
,U
T,
WA
,W
V,
WI.
Fro
mth
ese
state
sth
efo
llow
ing
had
tob
ed
rop
ped
from
the
”by
race
”re
gre
ssio
ns
du
eto
mis
sin
gan
dn
egati
ve
dro
pou
tra
tes:
AR
,ID
,K
Y,
MO
,N
D,
NJ,
NY
,S
D,
UT
,W
A,
WV
.T
he
state
of
NJ
isals
od
rop
ped
from
the
”all
race
s”re
gre
ssio
ns.
Sco
rech
an
ger
state
sare
those
state
sth
at
wer
ere
qu
ired
to
rais
eth
eir
min
imu
msc
ore
requ
irem
ent
from
35
to40
in1997.
Thes
ein
clu
de:
AK
,A
L,
AZ
,C
T,
GA
,H
I,IA
,IL
,IN
,K
S,
MA
,M
E,
MI,
MN
,M
T,
NC
,N
H,
NV
,O
H,
PA
,R
I,S
C,
TN
,V
A,
VT
,W
Y.
Fro
mth
ese
state
sth
efo
llow
ing
are
dro
pp
edd
ue
tom
issi
ng
an
dn
egati
ve
dro
pou
tra
tes:
AL
,A
Z,
IA,
ME
,M
N,
MT
,N
H,
SC
,T
N,
VT
,W
Y.
No
trea
tmen
t
state
sare
dro
pp
edfr
om
any
of
the
regre
ssio
ns.
Sta
tes
that
chan
ged
the
min
imu
mage
requ
ired
tod
rop
ou
tin
clu
de:
DC
(fro
mori
gin
al
contr
ol
gro
up
),M
San
dN
M(b
oth
from
trea
tmen
tgro
up
).S
tate
sth
at
chan
ged
eith
ersc
hool
leavin
gor
GE
Dage
requ
irem
ents
incl
ud
e:A
R,
DC
,K
Y,
MO
,O
K,
OR
,S
D,
UT
,W
I(f
rom
ori
gin
al
contr
ol
gro
up
)an
dM
S,
NE
an
dN
M(f
rom
trea
tmen
tgro
up
).
18
Tab
leD
-1:
Sum
mar
yof
Rob
ust
nes
sC
hec
ks
(Con
tinued
)
All
Rac
esW
hit
esB
lack
sH
isp
anic
s
Trea
tmen
t Effe
ct-4
.83%
-2.1
1%-6
.10%
-6.6
6%95
% C
I (H
uber
-Whi
te)
(-7.0
5%, -
2.59
%)
(-4.5
2%, 0
.29%
)(-1
5.98
%, 3
.78%
)(-9
.55%
, -3.
78%
)
12th
Gra
de
Dro
pou
t R
ate
Scor
e ch
ange
r sta
tes a
s co
ntro
l gro
up
95%
CI (
Conl
ey-T
aber
)(-6
.62%
, -0.
79%
)(-4
.72%
, 1.5
4%)
(-1.2
3%, 3
.87%
)(-1
1.18
%, -
1.62
%)
90%
CI (
Conl
ey-T
aber
)(-6
.12%
, -1.
28%
)(-4
.33%
, 1.0
8%)
(-1.0
6%, 2
.49%
)(-1
0.15
%, -
2.22
%)
Trea
tmen
t Effe
ct-2
.17%
0.07
%-3
.30%
-5.8
2%95
% C
I (H
uber
-Whi
te)
(-4.2
9%, 0
.05%
)(-2
.76%
, 2.9
0%)
(-8.4
8%, 1
.87%
)(-8
.04%
, -3.
61%
)95
% C
I (Co
nley
-Tab
er)
(-4.3
7%, 0
.13%
)(-3
.65%
, 1.9
5%)
(-7.5
6%, 4
.04%
)(-7
.69%
, 6.7
0%)
90%
CI (
Conl
ey-T
aber
)(-3
.94%
, -0.
13%
)(-2
.06%
, 1.7
7%)
(-5.9
7%, 2
.53%
)(-7
.47%
, 1.2
7%)
Trea
tmen
t Effe
ct-2
.93%
-1.2
8%-4
.76%
-5.9
8%95
% C
I (H
uber
-Whi
te)
(-4.2
9%, -
1.57
%)
(-3.3
0%, 0
.74%
)(-8
.63%
, -0.
90%
)(-7
.05%
, -4.
92%
)
Sout
hern
stat
es o
nly
Exc
ludi
ng st
ates
that
ch
ange
d m
inim
um a
ge
id
d95
% C
I (H
uber
Whi
te)
(4.2
9%,
1.57
%)
(3.3
0%, 0
.74%
)(8
.63%
, 0.
90%
)(7
.05%
, 4.
92%
)95
% C
I (Co
nley
-Tab
er)
(-6.4
1%, -
0.46
%)
(-3.8
5%, 1
.67%
)(-9
.84%
, 1.5
2%)
(-9.9
7%, 1
.32%
)90
% C
I (Co
nley
-Tab
er)
(-5.5
7%, -
0.91
%)
(-3.2
1%, 1
.07%
)(-8
.86%
, 1.1
5%)
(-8.4
5%, 0
.29%
)
requ
ired
to d
rop
out
Not
e: Th
e fu
ll re
gres
sion
spec
ifica
tions
are
pre
sent
ed in
the
web
app
endi
x. U
nles
s oth
erw
ise st
ated
the
mod
el is
estim
ated
usin
g O
LS. S
tate
15-
17 y
ear o
ld p
opul
atio
ns
(by
race
for r
eleva
nt re
gres
sions
) are
use
d as
weig
hts.
The
estim
ate
repo
rted
abov
e is
the
inte
ract
ion
betw
een
the
treat
men
t sta
te d
umm
y an
d th
e po
st p
erio
d du
mm
y, w
here
the
treat
men
t sta
te d
umm
y is
equa
l to
1 fo
r tre
atm
ent s
tate
s and
the
post
per
iod
dum
my
is eq
ual t
o 1
for t
he y
ears
199
8-20
00, o
ther
wise
bot
h du
mm
ies a
re e
qual
to 0
. Sta
tes w
ith fe
wer
than
two
obse
rvat
ions
per
per
iod
are
drop
ped
for "
all ra
ces"
regr
essio
ns. S
tate
s with
few
er th
an tw
o ob
serv
atio
ns p
er p
erio
d fo
r any
of t
he
drop
out r
ate
mea
sure
s by
race
are
dro
pped
for b
y ra
ce re
gres
sions
. Tre
atm
ent s
tate
s are
thos
e st
ates
that
wer
e re
quire
d to
elim
inat
e th
e an
d/or
scor
e op
tion.
The
se
inclu
de: L
A, M
S, N
E, N
M, T
X. U
nles
s oth
erw
ise st
ated
con
trol s
tate
s are
thos
e th
at w
ere
not r
equi
red
to ra
ise th
eir G
ED
min
imum
scor
e re
quire
men
t. Th
ese
inclu
de:
AR,
CA
, CO
, DE
, DC,
FL,
ID, K
Y, M
D, M
O, N
J, N
Y, N
D, O
K, O
R, S
D, U
T, W
A, W
V, W
I. Fr
om th
ese
stat
es th
e fo
llow
ing
had
to b
e dr
oppe
d fr
om th
e "b
y ra
ce"
regr
essio
ns d
ue to
miss
ing
and
nega
tive
drop
out r
ates
: AR,
ID, K
Y, M
O, N
D, N
J, N
Y, S
D, U
T, W
A, W
V. T
he st
ate
of N
J is a
lso d
ropp
ed fr
om th
e "a
ll ra
ces"
g
gg
p,
,,
,,
J,,
,,
,J
ppre
gres
sions
. Sco
re c
hang
er st
ates
are
thos
e st
ates
that
wer
e re
quire
d to
raise
their
min
imum
scor
e re
quire
men
t fro
m 3
5 to
40
in 1
997.
The
se in
clude
: AK
, AL,
AZ
, CT,
G
A, H
I, IA
, IL,
IN, K
S, M
A, M
E, M
I, M
N, M
T, N
C, N
H, N
V, O
H, P
A, R
I, SC
, TN
, VA
, VT,
WY
. Fro
m th
ese
stat
es th
e fo
llow
ing
are
drop
ped
due
to m
issin
g an
d ne
gativ
e dr
opou
t rat
es: A
L, A
Z, I
A, M
E, M
N, M
T, N
H, S
C, T
N, V
T, W
Y. N
o tre
atm
ent s
tate
s are
dro
pped
from
any
of t
he re
gres
sions
. Sta
tes t
hat c
hang
ed th
e m
inim
um a
ge re
quire
d to
dro
p ou
t inc
lude
: DC
(from
orig
inal
cont
rol g
roup
), M
S an
d N
M (b
oth
from
trea
tmen
t gro
up).
Stat
es th
at c
hang
ed e
ither
scho
ol le
avin
g or
G
ED
age
requ
irem
ents
inclu
de:
AR,
DC,
KY
, MO
, OK
, OR,
SD
, UT,
WI (
from
orig
inal
cont
rol g
roup
) and
MS,
NE
and
NM
(fro
m tr
eatm
ent g
roup
).
Note
:T
he
full
regre
ssio
nsp
ecifi
cati
on
sare
pre
sente
din
the
Web
Ap
pen
dix
.U
nle
ssoth
erw
ise
state
dth
em
od
elis
esti
mate
du
sin
gO
LS
.S
tate
15-1
7yea
rold
pop
ula
tion
s(b
y
race
for
rele
vant
regre
ssio
ns)
are
use
das
wei
ghts
.T
he
esti
mate
rep
ort
edab
ove
isth
ein
tera
ctio
nb
etw
een
the
trea
tmen
tst
ate
du
mm
yan
dth
ep
ost
per
iod
du
mm
y,w
her
eth
e
trea
tmen
tst
ate
du
mm
yis
equ
al
to1
for
trea
tmen
tst
ate
san
dth
ep
ost
per
iod
du
mm
yis
equ
al
to1
for
the
yea
rs1998-2
000,
oth
erw
ise
both
du
mm
ies
are
equ
al
to0.
Sta
tes
wit
h
few
erth
an
two
ob
serv
ati
on
sp
erp
erio
dare
dro
pp
edfo
r“all
race
s”re
gre
ssio
ns.
Sta
tes
wit
hfe
wer
than
two
ob
serv
ati
on
sp
erp
erio
dfo
rany
of
the
dro
pou
tra
tem
easu
res
by
race
are
dro
pp
edfo
rby
race
regre
ssio
ns.
Tre
atm
ent
state
sare
those
state
sth
at
wer
ere
qu
ired
toel
imin
ate
the
an
d/or
score
op
tion
.T
hes
ein
clu
de:
LA
,M
S,
NE
,N
M,
TX
.U
nle
ss
oth
erw
ise
state
dco
ntr
ol
state
sare
those
that
wer
en
ot
requ
ired
tora
ise
thei
rG
ED
min
imu
msc
ore
requ
irem
ent.
Th
ese
incl
ud
e:A
R,
CA
,C
O,
DE
,D
C,
FL
,ID
,K
Y,
MD
,M
O,
NJ,
NY
,N
D,
OK
,O
R,
SD
,U
T,
WA
,W
V,
WI.
Fro
mth
ese
state
sth
efo
llow
ing
had
tob
ed
rop
ped
from
the
”by
race
”re
gre
ssio
ns
du
eto
mis
sin
gan
dn
egati
ve
dro
pou
tra
tes:
AR
,ID
,K
Y,
MO
,N
D,
NJ,
NY
,S
D,
UT
,W
A,
WV
.T
he
state
of
NJ
isals
od
rop
ped
from
the
”all
race
s”re
gre
ssio
ns.
Sco
rech
an
ger
state
sare
those
state
sth
at
wer
ere
qu
ired
to
rais
eth
eir
min
imu
msc
ore
requ
irem
ent
from
35
to40
in1997.
Thes
ein
clu
de:
AK
,A
L,
AZ
,C
T,
GA
,H
I,IA
,IL
,IN
,K
S,
MA
,M
E,
MI,
MN
,M
T,
NC
,N
H,
NV
,O
H,
PA
,R
I,S
C,
TN
,V
A,
VT
,W
Y.
Fro
mth
ese
state
sth
efo
llow
ing
are
dro
pp
edd
ue
tom
issi
ng
an
dn
egati
ve
dro
pou
tra
tes:
AL
,A
Z,
IA,
ME
,M
N,
MT
,N
H,
SC
,T
N,
VT
,W
Y.
No
trea
tmen
t
state
sare
dro
pp
edfr
om
any
of
the
regre
ssio
ns.
Sta
tes
that
chan
ged
the
min
imu
mage
requ
ired
tod
rop
ou
tin
clu
de:
DC
(fro
mori
gin
al
contr
ol
gro
up
),M
San
dN
M(b
oth
from
trea
tmen
tgro
up
).S
tate
sth
at
chan
ged
eith
ersc
hool
leavin
gor
GE
Dage
requ
irem
ents
incl
ud
e:A
R,
DC
,K
Y,
MO
,O
K,
OR
,S
D,
UT
,W
I(f
rom
ori
gin
al
contr
ol
gro
up
)an
dM
S,
NE
an
dN
M(f
rom
trea
tmen
tgro
up
).
19
in the text for nearly all groups. The one exception is that the white upper level dropout
estimate is now very small and statistically insignificant.
D.3 Excluding States that Changed Minimum Age Requirements
A number of states in both our treatment and control groups either raised or lowered the
minimum age for GED testing or the minimum age for school leaving during the period
under study. Our fixed effect model controls for these changes. Alternatively, we estimate
the model excluding these states to be sure that these changes are not driving our results.
In row 3 of Table D-1, we drop all states that changed the minimum school leaving age and
find that our estimates are robust to this alternate specification.
D.4 Excluding Additional Years from the Analysis
The change in test difficulty was implemented at the beginning of 1997, in the middle of the
1996-1997 school year. All regressions exclude 1996-1997 dropout rates as these would be the
number enrolled in 1996 who were not still enrolled or graduated in 1997. If the change was
well publicized, it may have lead to a rush of test takers trying to pass before the increase
in test difficulty. As a robustness check we exclude 1996 through 1998 from our regressions.
As shown in Figure D-2, excluding these years has little effect on the estimates. Similarly,
the 12th grade dropout rate is notably higher in 1994 than the other years. As shown in
Figure D-2, excluding 1994 has little effect on the estimates.
20
Tab
leD
-2:
Alt
ernat
ive
Yea
rSp
ecifi
cati
onfo
rC
han
gein
Tes
tD
ifficu
lty
Dep
end
ent
Var
iab
leA
ll R
aces
Wh
ites
Bla
cks
His
pan
ics
All
Rac
esW
hit
esB
lack
sH
isp
anic
sA
ll R
aces
Wh
ites
Bla
cks
His
pan
ics
-1.2
9%-0
.43%
-1.2
9%-2
.74%
-1.4
2%-0
.43%
-1.6
0%-3
.20%
-1.1
7%-0
.33%
-1.4
9%-2
.39%
(.37)
(.15)
(.98)
(.40)
(.50)
(.30)
(1.0
6)(.6
7)(.2
6)(.1
2)(.8
4)(.4
2)-0
.55%
0.00
%0.
09%
-1.3
8%-0
.57%
0.12
%-0
.02%
-1.8
1%-0
.64%
-0.1
1%-0
.62%
-0.9
9%(.5
4)(.4
5)(1
.61)
(.50)
(.65)
(.54)
(1.8
6)(.8
2)(.4
3)(.4
1)(1
.23)
(.58)
-2.9
5%-1
.32%
-4.8
4%-6
.16%
-3.3
0%-1
.58%
-5.7
0%-6
.67%
-2.3
9%-0
.72%
-3.6
3%-5
.95%
(.65)
(.88)
(1.8
2)(.4
6)(.8
3)(1
.14)
(2.3
5)(.8
3)(.6
5)(.9
4)(1
.63)
(.91)
-0.5
7%…
……
-0.5
5%-0
.53%
(.18)
……
…(.1
9)(.1
7)-0
.36%
……
…-0
.32%
-0.3
7%(.1
7)…
……
(.21)
(.18)
-0.7
7%…
……
-0.7
6%-0
.68%
(.25)
……
…(.2
7)(.2
2)
10th
-12t
h G
rade
D
ropo
ut R
ate
GE
D T
estin
g Ra
te
Age
s 18-
19
The
year
199
7 is
exclu
ded
from
all
regr
essio
ns a
s was
don
e in
the
orig
inal
pape
r.
Exc
lud
ing
1994
Not
e: H
uber
-Whi
te ro
bust
stan
dard
err
ors a
re in
par
enth
eses
(clu
ster
ed b
y st
ate)
. Mod
el is
estim
ated
usin
g O
LS. S
tate
15-
17 y
ear o
ld p
opul
atio
ns b
y ra
ce a
re u
sed
as w
eight
s. Th
e es
timat
e re
porte
d ab
ove
is th
e in
tera
ctio
n be
twee
n th
e tre
atm
ent s
tate
dum
my
and
the
post
199
7 du
mm
y, w
here
the
treat
men
t sta
te d
umm
y is
equa
l to
1 fo
r tre
atm
ent s
tate
s and
th
e po
st 1
997
dum
my
is eq
ual t
o 1
for t
he y
ears
199
8-20
00, o
ther
wise
bot
h du
mm
ies a
re e
qual
to 0
. Sta
tes w
ith fe
wer
than
two
obse
rvat
ions
per
per
iod
are
drop
ped
for '
all ra
ces'
regr
essio
ns. S
tate
s with
few
er th
an tw
o ob
serv
atio
ns p
er p
erio
d fo
r any
of t
he d
ropo
ut ra
te m
easu
res b
y ra
ce a
re d
ropp
ed fo
r by
race
regr
essio
ns. T
reat
men
t sta
tes a
re th
ose
stat
es
that
wer
e re
quire
d to
elim
inat
e th
e an
d/or
scor
e op
tion.
The
se in
clude
: LA
, MS,
NE
, NM
, TX
. Con
trol s
tate
s are
thos
e th
at a
lread
y ha
d hi
gh e
noug
h st
anda
rds b
y 19
97. T
hese
in
clude
: AR,
CA
, CO
, DE
, DC,
FL,
ID, K
Y, M
D, M
O, N
J, N
Y, N
D, O
K, O
R, S
D, U
T, W
A, W
V, W
I. Co
ntro
l sta
tes d
ropp
ed fr
om 'a
ll ra
ces'
regr
essio
ns d
ue to
miss
ing
and
nega
tive
drop
out r
ates
inclu
de: N
J. Co
ntro
l sta
tes d
ropp
ed fr
om re
gres
sion
by ra
ce d
ue to
miss
ing
and
nega
tive
drop
out r
ates
inclu
de: A
R, ID
, KY
, MO
, ND
, NJ,
NY
, SD
, UT,
W
A, W
V. N
o tre
atm
ent s
tate
s are
dro
pped
.
10th
-11t
h G
rade
D
ropo
ut R
ate
12th
Gra
de
Dro
pout
Rat
eG
ED
Tes
ting
Rate
A
ges 1
6-19
GE
D T
estin
g Ra
te
Age
s 16-
17
Incl
ud
ing
all Y
ears
Exc
lud
ing
1996
an
d 1
998
21
D.5 Extending the GED Testing and Dropout Rate Trends
When we extend the times series of GED testing and dropout rates on both sides of 1997, we
obtain results that support the analysis in the main text. Prior to the increase in standards,
the GED test-taking rate is higher in treatment states compared to control states. For
the higher grade levels, dropout rates are higher in control states compared to treatment
states prior to the introduction of the new standards but are lower afterwards. The effect is
particularly strong for the 12th grade dropout rate. The breaks in the trends around 1997
are evident for the student pools more eligible to take the GED.
22
Fig
ure
D-1
:G
ED
Tes
ting
and
Dro
pou
tR
ates
By
Yea
r,T
reat
men
tvs.
Con
trol
Sta
tes
(exte
nded
year
s)
23
E Fixed Effect Estimates of the Effect of Increasing
Passing Standards
One difficulty in isolating the effect of changes in GED passing standards on dropout rates
is that both minimum school leaving age requirements and GED testing age requirements
changed in the sample period under study (See Table E-1). During our sample period, three
of the five states in our treatment group both raised and lowered their GED minimum age
requirement. Two of these three states also raised the minimum age at which students can
drop out of school. States included in our control group also made changes in their age
requirements.
To control for these potentially confounding changes in age requirements and other
sources of variation across states, we estimate a state fixed-effect regression. The model
is
Yi,t = γ(Dtreat ·Dpost 97) + πDpost 97 + θi + ψXi,t + εi,t
where Yi,t is the dropout rate for state i in year t and Dtreat and Dpost 97 are defined as
Dtreat = {1 if the state eliminated the and/or GED score option in 19970 if the state was not required to raise GED standards in 1997
Dpost 97 = {1 if 1998 ≤ year ≤ 20000 otherwise .
The θi are time-invariant state level fixed effects and the Xi,t are control variables that
vary by states over time. These include dummy indicators for both the minimum age required
to take the GED and the minimum age required to drop out of school, as well as measures
of state level unemployment rates and per capita income to control for changes in labor
market conditions during the sample period.1 The parameter of interest is γ, which is the
conditional difference-in-difference estimate of the treatment effect of the reform in GED
1See Table E-3 for the summary statistics of all variables used in these models. We do not control for highstakes testing because no treatment or control states implemented or changed testing requirements duringthe sample period.
24
Tab
leE
-1:
Chan
ges
inG
ED
Tes
ting
and
Man
dat
ory
Sch
ool
ing
Age
Req
uir
emen
tsby
Tre
atm
ent
Sta
tus,
1994
-200
0
Pol
icy
Ch
ange
Tre
atm
ent
Gro
up
Con
trol
Gro
up
Raise
d M
inim
um G
ED
Age
Re
quire
men
tM
S (1
7 to
18,
199
7), N
E (1
6 to
18,
199
8),
NM
(16
to 1
7, 2
000)
.
AR
(16
to 1
8, 2
000)
, KY
(16
to 1
9 ,1
997
and
200
0),
MO
(16
to 1
8, 1
997)
, OK
(16
to 1
8, 1
997
and
2000
), O
R (1
6 to
18,
200
0), S
D (1
7 to
19,
199
7 an
d 16
to 1
8,
1999
), U
T (1
7 to
18,
199
7 an
d 20
00),
WI (
18 to
18.
5,
1999
).
Low
ered
Min
imum
GE
D A
ge
Requ
irem
ent
MS
(18
to 1
6, 2
000)
, NE
(17
to 1
6, 1
995)
, N
M (1
8 to
16,
199
9).
KY
(19
to 1
6, 1
999)
, MO
(18
to 1
6, 1
995)
, OK
(18
to
16, 1
995
and
1999
), O
R (1
8 to
16,
199
9), S
D (1
8 to
17
, 199
5 an
d 19
to 1
6, 1
998)
, UT
(18
to 1
7, 1
995
and
1999
), W
I (18
.5 to
18,
199
5).
Raise
d M
inim
um S
choo
l Le
avin
g A
ge R
equi
rem
ent
MS
(16
to 1
7, 1
997)
, NM
(16
to 1
8, 1
997)
.D
C (1
7 to
18,
199
7).
Low
ered
Min
imum
Lea
ving
A
ge R
equi
rem
ent
Non
e.N
one.
Not
e: Th
e ye
ar o
f cha
nge
as w
ell a
s the
initi
al an
d fin
al va
lue
for t
he a
ge re
quirm
ent a
re re
porte
d in
par
enth
eses
.
Sour
ce: G
ED
Tes
ting
Serv
ice A
nnua
l Rep
orts
: "W
ho T
ook
the
GE
D?"
(var
ious
yea
rs) a
nd D
iges
t of E
duca
tion
Stat
istics
(var
ious
ye
ars)
.
25
standards on the high school dropout rate.
Weighted OLS estimates of γ from the full model both controlling and not controlling for
changes in minimum age requirements are summarized in Table E-2.2 The other parameter
estimates are available in Table G-27. Using the full specification, the overall effect of
the reform is a 1.3 percentage point reduction in the dropout rate in treatment states.
The estimated effect on the upper level dropout rate remains close to 3 percentage points
and is statistically significant. The estimated effect on the overall lower level dropout rate
remains small and statistically insignificant. In general, the estimates including state level
fixed effects but not controlling for changes in minimum age requirements are smaller than
estimates based on the full specification. The regression-adjusted dropout and GED testing
rate estimates are for the most part smaller but in close agreement with the unadjusted
estimates reported in the text.
The fixed effects estimates by race are consistent with the unadjusted estimates as well.
Again, the estimated treatment effect is greater for minorities compared to whites. As is true
of the estimates reported in the text, the largest effect is on the upper level dropout rate.
Increasing GED passing requirements decreased the upper level dropout rate in treatment
states by 1.3 percentage points for whites, 4.8 percentage points for blacks and 6.2 percentage
points for Hispanics.
2GLS estimates of the model are also available in the Web Appendix and match those reported in thetext. The results also hold with serially correlated errors.
26
Tab
leE
-2:
Wei
ghte
dO
LS
Fix
edE
ffec
tsE
stim
ates
ofth
eIm
pac
tof
the
1997
GE
DR
efor
mon
Var
ious
Dro
pou
tR
ate
Mea
sure
s
Dep
en
den
t V
ari
ab
leA
ll R
aces
Wh
ites
Bla
ck
sH
isp
an
ics
All
Races
Wh
ites
Bla
ck
sH
isp
an
ics
Tre
atm
ent
Eff
ect
-1.2
1%
-0.4
7%
-0.8
3%
-2.6
8%
-1.2
9%
-0.4
3%
-1.2
9%
-2.7
4%
95%
CI
Hub
er-W
hit
e Sta
nd
ard
Err
ors
(-1.9
4%
, -0
.48%
)(-
0.8
0%
, -0
.13%
)(-
2.8
6%
, 1.1
2%
)(-
3.6
1%
, -1
.75%
)(-
2.0
6%
, -0
.53%
)(-
0.7
6%
, -0
.11%
)(-
3.4
0%
, 0.8
3%
)(-
3.6
0%
, -1
.89%
)
95%
CI
Co
nle
y-T
aber
Sta
nd
ard
Err
ors
(-2.0
6%
, 0.2
5%
)(-
0.7
5%
, 1.7
4%
)(-
2.0
2%
, 4.7
7%
)(-
10.7
7%
, 1.1
0%
)(-
2.1
4%
, -0
.37%
)(-
0.7
5%
, 0.8
4%
)(-
2.8
0%
, 1.5
5%
)(-
5.6
9%
, -1
.09%
)
90%
CI
Co
nle
y-T
aber
Sta
nd
ard
Err
ors
(-1.9
1%
, -0
.01%
)(-
0.6
8%
, 1.3
6%
)(-
1.6
1%
, 3.6
9%
)(-
9.4
2%
, 0.4
3%
)(-
1.9
9%
, -0
.51%
)(-
0.6
8%
, 0.6
4%
)(-
2.5
4%
, 1.0
6%
)(-
5.2
1%
, -1
.46%
)
Tre
atm
ent
Eff
ect
-0.4
6%
-0.0
2%
0.3
7%
-1.3
8%
-0.5
5%
0.0
0%
0.0
9%
-1.3
8%
95%
CI
Hub
er-W
hit
e Sta
nd
ard
Err
ors
(-1.5
3%
, 0.6
0%
)(-
0.9
6%
, 0.9
1%
)(-
2.8
3%
, 3.5
7%
)(-
2.6
7%
, -0
.095%
)(-
1.6
7%
, 0.5
7%
)(-
0.9
7%
, 0.9
7%
)(-
3.3
8%
, 3.5
7%
)(-
2.4
6%
, -0
.31%
)
95%
CI
Co
nle
y-T
aber
Sta
nd
ard
Err
ors
(-1.5
4%
, 1.1
1%
)(-
0.7
6%
, 1.8
9%
)(-
2.0
8%
, 8.9
9%
)(-
11.3
3%
, 2.9
5%
)(-
1.6
1%
, 0.5
9%
)(-
0.7
6%
, 1.0
0%
)(-
2.3
7%
, 4.7
4%
)(-
6.0
7%
, 0.3
9%
)
90%
CI
Co
nle
y-T
aber
Sta
nd
ard
Err
ors
(-1.3
8%
, 0.8
5%
)(-
0.6
3%
, 1.5
1%
)(-
1.5
1%
, 7.3
4%
)(-
9.9
0%
, 2.2
0%
)(-
1.4
1%
, 0.3
9%
)(-
0.6
3%
, 0.8
1%
)(-
2.0
9%
, 3.8
4%
)(-
5.3
1%
, -0
.02%
)
Tre
atm
ent
Eff
ect
-2.8
6%
-1.3
8%
-4.0
2%
-5.9
9%
-2.9
5%
-1.3
2%
-4.8
4%
-6.1
6%
95%
CI
Hub
er-W
hit
e Sta
nd
ard
Err
ors
(-4.2
5%
, -1
.47%
)(-
3.1
4%
, 0.3
8%
)(-
7.8
3%
, -0
.21%
)(-
6.6
4%
, -5
.34%
)(-
4.3
0%
, -1
.60%
)(-
3.2
3%
, 0.5
9%
)(-
8.7
6%
, -0
.91%
)(-
7.5
2%
, -5
.16%
)
95%
CI
Co
nle
y-T
aber
Sta
nd
ard
Err
ors
(-5.1
9%
, -1
.04%
)(-
2.8
2%
, 0.8
4%
)(-
7.9
1%
, 2.4
7%
)(-
7.9
0%
, 0.0
0%
)(-
5.1
8%
, -1
.14%
)(-
2.8
5%
, 0.4
5%
)(-
9.0
1%
, 0.2
9%
)(-
7.5
6%
, 0.3
4%
)
90%
CI
Co
nle
y-T
aber
Sta
nd
ard
Err
ors
(-4.8
0%
, -1
.31%
)(-
2.5
4%
, 0.4
8%
)(-
7.0
0%
, 1.5
6%
)(-
7.3
9%
, -0
.98%
)(-
4.8
6%
, -1
.40%
)(-
2.5
4%
, 0.2
1%
)(-
8.1
9%
, -0
.50%
)(-
7.1
1%
, -0
.52%
)
Tre
atm
ent
Eff
ect
-0.5
5%
……
…-0
.57%
……
…
95%
CI
Hub
er-W
hit
e Sta
nd
ard
Err
ors
(-0.8
7%
, -0
.23%
)(-
0.9
4%
, -0
.21%
)
95%
CI
Co
nle
y-T
aber
Sta
nd
ard
Err
ors
(-1.0
0%
, -
0.3
8%
)…
……
(-1.0
2%
, -0
.41%
)…
……
90%
CI
Co
nle
y-T
aber
Sta
nd
ard
Err
ors
(-0.9
4%
, -
0.4
2%
)…
……
(-0.9
5%
, -0
.46%
)…
……
Tre
atm
ent
Eff
ect
-0.3
4%
……
…-0
.36%
……
…
95%
CI
Hub
er-W
hit
e Sta
nd
ard
Err
ors
(-0.6
6%
, -0
.02%
)(-
0.7
1%
, -0
.023%
)
95%
CI
Co
nle
y-T
aber
Sta
nd
ard
Err
ors
(-1.0
7%
, -
0.1
4%
)…
……
(-1.0
7%
, -
0.1
7%
)…
……
90%
CI
Co
nle
y-T
aber
Sta
nd
ard
Err
ors
(-1.0
0%
, -
0.2
0%
)…
……
(-0.9
8%
, -
0.2
4%
)…
……
Tre
atm
ent
Eff
ect
-0.7
6%
……
…-0
.77%
……
…
95%
CI
Hub
er-W
hit
e Sta
nd
ard
Err
ors
(-1.2
0%
, -0
.32%
)(-
1.2
8%
, -0
.26%
)
95%
CI
Co
nle
y-T
aber
Sta
nd
ard
Err
ors
(-1.1
3%
, -
0.4
2%
)…
……
(-1.1
7%
, -
0.4
7%
)…
……
90%
CI
Co
nle
y-T
aber
Sta
nd
ard
Err
ors
(-1.0
5%
, -
0.4
7%
)…
……
(-1.0
8%
, -
0.5
1%
)…
……
No
te: C
on
ley-
Tab
er a
dju
sted
co
nfi
den
ce in
terv
als
are
in p
aren
thes
es. T
he
full r
egre
ssio
n s
pec
ific
atio
ns
are
pre
sen
ted
in
th
e w
eb a
pp
end
ix. U
nle
ss o
ther
wis
e st
ated
th
e m
od
el is
esti
mat
ed u
sin
g O
LS. Sta
te 1
5-1
7 y
ear
old
po
pula
tio
ns
(by
race
fo
r re
levan
t re
gres
sio
ns)
are
use
d a
s w
eigh
ts. T
he
esti
mat
e re
po
rted
ab
ove
is t
he
inte
ract
ion
bet
wee
n t
he
trea
tmen
t st
ate
dum
my
and
th
e p
ost
per
iod
dum
my,
wh
ere
the
trea
tmen
t st
ate
dum
my
is e
qual
to
1 f
or
trea
tmen
t st
ates
an
d t
he
po
st p
erio
d d
um
my
is e
qual
to
1 f
or
the
year
s 1998-2
000, o
ther
wis
e b
oth
dum
mie
s ar
e eq
ual
to
0. Sta
tes
wit
h f
ewer
th
an t
wo
ob
serv
atio
ns
per
per
iod
are
dro
pp
ed f
or
"all
race
s" r
egre
ssio
ns.
Sta
tes
wit
h f
ewer
th
an t
wo
ob
serv
atio
ns
per
per
iod
fo
r an
y o
f th
e d
rop
out
rate
mea
sure
s b
y ra
ce a
re d
rop
ped
fo
r b
y ra
ce r
egre
ssio
ns.
Tre
atm
ent
stat
es a
re t
ho
se s
tate
s th
at w
ere
requir
ed t
o e
lim
inat
e
the
and
/o
r sc
ore
op
tio
n. T
hes
e in
clud
e: L
A, M
S, N
E, N
M, T
X. U
nle
ss o
ther
wis
e st
ated
co
ntr
ol st
ates
are
th
ose
th
at w
ere
no
t re
quir
ed t
o r
aise
th
eir
GE
D m
inim
um
sco
re r
equir
emen
t. T
hes
e in
clud
e: A
R, C
A, C
O,
DE
, D
C, F
L, ID
, K
Y, M
D, M
O, N
J, N
Y, N
D, O
K, O
R, SD
, U
T, W
A, W
V, W
I. F
rom
th
ese
stat
es t
he
follo
win
g h
ad t
o b
e d
rop
ped
fro
m t
he
"by
race
" re
gres
sio
ns
due
to m
issi
ng
and
neg
ativ
e d
rop
out
rate
s: A
R, ID
,
KY
, M
O, N
D, N
J, N
Y, SD
, U
T, W
A, W
V. T
he
stat
e o
f N
J is
als
o d
rop
ped
fro
m t
he
"all r
aces
" re
gres
sio
ns.
Sco
re c
han
ger
stat
es a
re t
ho
se s
tate
s th
at w
ere
requir
ed t
o r
aise
th
eir
min
imum
sco
re r
equir
emen
t fr
om
35
to 4
0 in
1997. T
hes
e in
clud
e: A
K, A
L, A
Z, C
T, G
A, H
I, I
A, IL
, IN
, K
S, M
A, M
E, M
I, M
N, M
T, N
C, N
H, N
V, O
H, P
A, R
I, S
C, T
N, V
A, V
T, W
Y. F
rom
th
ese
stat
es t
he
follo
win
g ar
e d
rop
ped
due
to m
issi
ng
and
neg
ativ
e d
rop
out
rate
s: A
L, A
Z, IA
, M
E, M
N, M
T, N
H, SC
, T
N, V
T, W
Y. N
o t
reat
men
t st
ates
are
dro
pp
ed f
rom
an
y o
f th
e re
gres
sio
ns.
Sta
tes
that
ch
ange
d t
he
min
imum
age
req
uir
ed t
o d
rop
out
incl
ud
e: D
C (
fro
m
ori
gin
al c
on
tro
l gr
oup
), M
S a
nd
NM
(b
oth
fro
m t
reat
men
t gr
oup
). S
tate
s th
at c
han
ged
eit
her
sch
oo
l le
avin
g o
r G
ED
age
req
uir
emen
ts in
clud
e:
AR
, D
C, K
Y, M
O, O
K, O
R, SD
, U
T, W
I (f
rom
ori
gin
al c
on
tro
l gr
oup
)
and
MS, N
E a
nd
NM
(fr
om
tre
atm
ent
gro
up
).
Tab
le 5
. W
eigh
ted
OL
S F
ixed
Eff
ects
Est
imat
es o
f th
e Im
pac
t o
f th
e 1997 G
ED
Ref
orm
on
Var
ious
Dro
po
ut
Rat
e M
easu
res
No
t C
on
tro
llin
g f
or
Min
imu
m A
ge R
eq
uir
em
en
tsC
on
tro
llin
g f
or
Min
imu
m A
ge R
eq
uir
em
en
ts
10th
-12th
Gra
de
Dro
po
ut
Rat
e
10th
-11th
Gra
de
Dro
po
ut
Rat
e
12th
Gra
de
Dro
po
ut
Rat
e
GE
D T
esti
ng
Rat
e
Age
s 16-1
9
GE
D T
esti
ng
Rat
e
Age
s 16-1
7
GE
D T
esti
ng
Rat
e
Age
s 18-1
9
Note
:C
on
ley-T
ab
erad
just
edco
nfi
den
cein
terv
als
are
inp
are
nth
eses
.T
he
full
regre
ssio
nsp
ecifi
cati
on
sare
pre
sente
din
the
Web
Ap
pen
dix
.U
nle
ssoth
erw
ise
state
dth
em
od
el
ises
tim
ate
du
sin
gO
LS
.S
tate
15-1
7yea
rold
pop
ula
tion
s(b
yra
cefo
rre
levant
regre
ssio
ns)
are
use
das
wei
ghts
.T
he
esti
mate
rep
ort
edab
ove
isth
ein
tera
ctio
nb
etw
een
the
trea
tmen
tst
ate
du
mm
yan
dth
ep
ost
per
iod
du
mm
y,w
her
eth
etr
eatm
ent
state
du
mm
yis
equ
al
to1
for
trea
tmen
tst
ate
san
dth
ep
ost
per
iod
du
mm
yis
equ
al
to1
for
the
yea
rs1998-2
000,
oth
erw
ise
both
du
mm
ies
are
equ
al
to0.
Sta
tes
wit
hfe
wer
than
two
ob
serv
ati
on
sp
erp
erio
dare
dro
pp
edfo
r“all
race
s”re
gre
ssio
ns.
Sta
tes
wit
hfe
wer
than
two
ob
serv
ati
on
sp
erp
erio
dfo
rany
of
the
dro
pou
tra
tem
easu
res
by
race
are
dro
pp
edfo
rby
race
regre
ssio
ns.
Tre
atm
ent
state
sare
those
state
sth
at
wer
ere
qu
ired
toel
imin
ate
the
an
d/or
score
op
tion
.T
hes
ein
clu
de:
LA
,M
S,
NE
,N
M,
TX
.U
nle
ssoth
erw
ise
state
dco
ntr
ol
state
sare
those
that
wer
en
ot
requ
ired
tora
ise
thei
rG
ED
min
imu
msc
ore
requ
irem
ent.
Th
ese
incl
ud
e:A
R,
CA
,C
O,
DE
,D
C,
FL
,ID
,K
Y,
MD
,M
O,
NJ,
NY
,N
D,
OK
,O
R,
SD
,U
T,
WA
,W
V,
WI.
Fro
mth
ese
state
sth
efo
llow
ing
had
tob
ed
ropp
ed
from
the
”by
race
”re
gre
ssio
ns
du
eto
mis
sin
gan
dn
egati
ve
dro
pou
tra
tes:
AR
,ID
,K
Y,
MO
,N
D,
NJ,
NY
,S
D,
UT
,W
A,
WV
.T
he
state
of
NJ
isals
od
rop
ped
from
the
”all
race
s”re
gre
ssio
ns.
Sco
rech
an
ger
state
sare
those
state
sth
at
wer
ere
qu
ired
tora
ise
thei
rm
inim
um
score
requ
irem
ent
from
35
to40
in1997.
Thes
ein
clu
de:
AK
,A
L,
AZ
,C
T,
GA
,H
I,IA
,IL
,IN
,K
S,
MA
,M
E,
MI,
MN
,M
T,
NC
,N
H,
NV
,O
H,
PA
,R
I,S
C,
TN
,V
A,
VT
,W
Y.
Fro
mth
ese
state
sth
efo
llow
ing
are
dro
pp
edd
ue
tom
issi
ng
an
dn
egati
ve
dro
pou
tra
tes:
AL
,A
Z,
IA,
ME
,M
N,
MT
,N
H,
SC
,T
N,
VT
,W
Y.
No
trea
tmen
tst
ate
sare
dro
pp
edfr
om
any
of
the
regre
ssio
ns.
Sta
tes
that
chan
ged
the
min
imu
mage
requ
ired
tod
rop
ou
tin
clu
de:
DC
(fro
mori
gin
al
contr
ol
gro
up
),M
San
dN
M(b
oth
from
trea
tmen
tgro
up
).S
tate
sth
at
chan
ged
eith
ersc
hool
leavin
gor
GE
Dage
requ
irem
ents
incl
ud
e:
AR
,D
C,
KY
,M
O,
OK
,O
R,
SD
,U
T,
WI
(fro
mori
gin
al
contr
ol
gro
up
)an
dM
S,
NE
an
dN
M(f
rom
trea
tmen
tgro
up
).
27
Table E-3: Summary Statistics of Variables Used in the Analysis
Mean N Mean N Mean N Mean N
10th-12th Grade Dropout Rate
Overall .105 .090 .100 .098
(.019) (.021) (.033) (.032)
Whites .079 .070 .086 .082
(.019) (.019) (.029) (.030)
Blacks .140 .124 .158 .152
(.012) (.018) (.041) (.044)
Hispanics .142 .109 .158 .143
(.020) (.019) (.018) (.014)
10th-11th Grade Dropout Rate
Overall .112 .105 .105 .102
(.022) (.023) (.041) (.043)
Whites .089 .085 .086 .082
(.024) (.023) (.037) (.042)
Blacks .140 .134 .170 .160
(.019) (.015) (.042) (.051)
Hispanics .152 .133 .165 .143
(.025) (.020) (.025) (.021)
12th Grade Dropout Rate
Overall .088 .052 .095 .088
(.016) (.025) (.034) (.035)
Whites .057 .036 .086 .079
(.016) (.018) (.034) (.032)
Blacks .139 .099 .124 .129
(.019) (.035) (.058) (.039)
Hipsanics .114 .049 .138 .143
(.019) (.024) (.033) (.036)
GED Testing Rate
Ages 16-19 0.031 0.027 0.020 0.021
(.007) (.005) (.007) (.008)
Ages 16-17 0.026 0.025 0.014 0.015
(.006) (.005) (.009) (.009)
Ages 18-19 .036 .029 .026 .027
(.009) (.007) (.007) (.009)
Time Varying Covariates
Minimum dropout age 16.78 17.00 16.92 16.92
(.425) (.347) (.991) (.993)
Minimum GED testing age 17.65 17.73 17.71 17.88
(.569) (.564) (1.031) (.937)
Log per capita income 9.91 10.12 10.07 10.27
(.087) (.114) (.133) (.143)
Local unemployment rate .062 .047 .062 .046
(.010) (.007) (.014) (.009)
15
57
57
57
57
57
57
57
57
57
14
14
15
15
15
15
15
15
14
15
15
57
57
15
15 57
57
15 26 26
14 15 27 26
57
26
14 15 27 27
15 15 56 56
15 15 26
26
15 15 27 27
15 15
15 15
27
57
27
27
27
57 57
15
15
27
27
27
Note: Standard deviations in parentheses. The dropout rate is defined as the ratio of students enrolled in a given
grade(s) in year t and the number of students enrolled in the previous grade(s) in year t-1, where t = 1994-2000.
Enrollment data by grade are from the Common Core of Data (CCD). Minimum age required to drop out of school are
from the Digest of Education Statistics (several years). Minimum age required to take the GED are from the yearly
reports published by the GED Testing Service: "Who Took the GED?" (several years). Local unemployment rates are
computed using monthly data from the Bureau of Labor Statistics. Log of per capita income are computed using data
from the Census Bureau. All calculations are weighted by the 15-17 year old population in each state.
Table A-1. Summary Statistics of Variables Used in the Anlaysis
Treatment Group States Control Group States
Pre-1997
(N=15)
Post-1997
(N=15)
Pre-1997
(N=57)
Post-1997
(N=57)
57
15
15
15
15
15
15
15
Note: Standard deviations in parentheses. The dropout rate is defined as the ratio of students enrolled in a given grade(s) in
year t and the number of students enrolled in the previous grade(s) in year t−1, where t= 1994-2000. Enrollment data by grade
are from the Common Core of Data (CCD). Minimum age required to drop out of school are from the Digest of Education
Statistics (several years). Minimum age required to take the GED are from the yearly reports published by the GED Testing
Service: ”Who Took the GED?” (several years). Local unemployment rates are computed using monthly data from the Bureau
of Labor Statistics. Log of per capita income are computed using data from the Census Bureau. All calculations are weighted
by the 15-17 year old population in each state. 28
F Supplementary Material for the Analysis of the GED
Option Program
29
Fig
ure
F-1
:M
edia
nD
ays
ofE
nro
llm
ent
inG
ED
Opti
on,
by
Sta
te:
2008
–09
Sch
ool
Yea
r
75
68
101
87
New
Ham
pshire (N
=0)
Oregon (N=2,276)
Tenn
essee (N=375)
Texas (N=773)
Virginia (N
=3,044)
Wisconsin (N
=o)
r of GED Option Candidates)
Med
ian Days of Enrollm
ent in GED
Option, by State: 2008‐09
Schoo
l Year
Med
ian Num
ber o
f Days
of Enrollm
ent o
f GED
Option Cand
idates: 105
NA NA
168
72
217
99
94
050
100
150
200
250
Florida (N=3,755)
Kentucky (N
=70)
Louisiana (N=1,186)
Mississippi (N
=1,018)
Missouri (N=2,117)
Med
ian Num
ber o
f Days
State (Numbe
Sou
rce:
GE
DO
pti
onS
tati
stic
alR
epor
t(2
009)
.
30
Fig
ure
F-2
:M
edia
nP
repar
atio
nH
ours
ofG
ED
Opti
onC
andid
ates
,by
Sta
te:
2008
–09
Sch
ool
Yea
r
Med
ian Prep
aration Hou
rs of G
ED Option Ca
ndidates, by State: 2008‐09
Scho
ol Year
60
50
Virginia (N
=2,663
)
Wisconsin (N
=2,565
)Med
ian Num
ber of
Prep
arationHou
rs of G
ED
Option Cand
idates: 80
50
60
Tenn
essee (N=426
)
Texas (N=495
)
Candidates)
45
20
New
Ham
pshire (N
=65)
Oregon (N=1,927
)
r of GED Option C
150
120
137
Louisiana(N
111
7)
Mississippi (N
=1,053
)
Missouri (N=2,150
)
State (Numbe
79
75
150
Florida(N=2
614)
Kentucky (N
=61)
Louisiana (N=1,117
)
79
020
4060
8010
012
014
016
0
Florida (N
2,61
4)
Med
ian Num
ber o
f Hou
rs
Sou
rce:
GE
DO
pti
onS
tati
stic
alR
epor
t(2
009)
.
31
Fig
ure
F-3
:N
inth
Gra
de
Coh
ort
Gra
duat
ion
Sta
tus
ofG
ED
Opti
onC
andid
ates
,by
Sta
te:
2008
–09
Sch
ool
Yea
r
Ninth Grade
Coh
ort G
radu
ation Status of G
ED Option Ca
ndidates, by
State: 2008‐09
Schoo
l Year
83.8%
84.2%
Wisconsin (N
=3,612
)
GED
Option Total
95.5%
78.1%
91.8%
Tenn
essee (N=443
)
Texas (N=517
)
Virginia (N
=3,187
)
Candidates)
82.7%
New
Ham
pshire (N
=NA)
Oregon (N=2,409
)
()
er of GED Option
NA
86.0%
88.6%
72.5%
Louisiana (N=1,258
)
Mississippi (N
=1,132
)
Missouri (N=2,425
)
State (Numbe
84.0%89
.6%
Florida (N=3,767
)
Kentucky (N
=67)
0%20
%40
%60
%80
%10
0%12
0%
Percen
t Did Not Gradu
ate Ye
t
Sou
rce:
GE
DO
pti
onS
tati
stic
alR
epor
t(2
009)
.
32
Tab
leF
-1:
Sta
tes
Issu
ing
Cre
den
tial
sIn
dis
tingu
ishab
lefr
omH
igh
Sch
ool
Dip
lom
as(2
008)
State
Cred
ential Title
Arkansas
Arkansas High Scho
ol Diploma
Conn
ecticut
Conn
ecticut State High Scho
ol Diploma
Florida
State of Florida
High Scho
ol Diploma
Haw
aii
High Scho
ol Diploma
Kansas
Kansas State High Scho
ol Diploma
Maryland
Maryland High Scho
ol Diploma
New
Jersey
New
Jersey State Issued
High Scho
ol Diploma
New
Mexico
New
Mexico High Scho
ol Diploma
Oklahom
aOklahom
a High Scho
ol Diploma
Penn
sylvania
Common
wealth
Secon
dary Schoo
l Diploma
Source: G
ED Statistical Rep
ort 2
008
Not
es:
Incl
ud
esye
aran
dd
istr
ict
bac
kgr
oun
dch
ara
cter
isti
csan
dco
vari
ate
s.
33
Fig
ure
F-4
:T
he
Eff
ect
ofR
egula
rSch
ool
sO
pti
onP
rogr
amon
Hig
hSch
ool
Coh
ort
Com
ple
tion
Rat
es.
1.1%
1.0%
1.5%
2%3%4%eter Rates
0%1%2%
or Other Comple
-3%
-2%
-1%
hort Diploma o
‐3.0%
-5%
-4%
-3%
t Change in Coh
‐3.8%
‐3.6%
-7%
-6%
Dip
lom
a R
ate
(8th
)D
iplo
ma
Rat
e (9
th)
Dip
lom
a R
ate
(10t
h)
Percen
Dip
lom
as (O
ptio
n In
Nor
mal
Sch
ools
)O
ther
Com
plet
ers
(Opt
ion
In N
orm
al S
choo
ls)
Sou
rce:
Nati
on
al
Cen
ter
for
Ed
uca
tion
Sta
tist
ics,
Com
mon
Core
Data
an
dO
regon
Sch
ool
Dis
tric
tsA
dm
inis
trati
ve
Data
.N
ote
s:C
oh
ort
com
ple
tion
rate
sare
defi
ned
as
the
nu
mb
erof
dip
lom
as
issu
edd
ivid
edby
8th
,9th
,or
10th
gra
de
enro
llm
ent
lagged
the
ap
pro
pri
ate
nu
mb
erof
yea
rs.
We
show
coh
ort
com
ple
tion
rate
sfo
r8th
,9th
,an
d10th
gra
de
as
ach
eck
of
rob
ust
nes
s.T
he
defi
nit
ion
of
oth
erco
mp
lete
rsin
clu
des
stu
den
tsw
ho
GE
Dce
rtif
yth
rou
gh
ad
istr
ict
or
state
san
ctio
ned
cert
ifica
tion
pro
gra
m,
an
dth
us
shou
ldca
ptu
rest
ud
ents
wh
oG
ED
cert
ify
thro
ugh
the
GE
DO
pti
on
pro
gra
m.
Ore
gon
was
gra
nte
dp
erm
issi
on
tooff
erth
eG
ED
Op
tion
by
the
Am
eric
an
Cou
nci
lon
Ed
uca
tion
in2001,
bu
tn
osc
hools
imp
lem
ente
dp
rogra
ms
unti
lth
e2001-2
002
sch
ool
yea
r.R
egre
ssio
ns
incl
ud
eco
ntr
ols
for
per
cent
bla
cken
rollm
ent,
per
cent
His
pan
icen
rollm
ent,
per
cent
free
lun
chel
igib
le,
per
cent
free
or
red
uce
dlu
nch
elig
ible
,p
up
ilte
ach
erra
tio,
exp
end
itu
res
per
pu
pil,
reven
ue
per
pu
pil,
an
dd
istr
ict
an
dyea
rfi
xed
effec
ts.
Reg
ress
ion
sin
clu
de
2001-2
002
sch
ool
yea
rth
rou
gh
2007-2
008
sch
ool
yea
r.A
nalt
ern
ati
ve
Op
tion
pro
gra
mis
defi
ned
as
on
eth
at
was
not
pre
sent
intr
ad
itio
nal
hig
hsc
hools
,b
ut
on
lyin
com
mu
nit
y
colleg
esor
oth
erin
stit
uti
on
s.E
rror
bars
show
stan
dard
erro
rs.
34
Fig
ure
F-5
:D
escr
ipti
veC
ompar
ison
sof
Dis
tric
tsw
ith
and
wit
hou
tG
ED
Opti
onP
rogr
ams
(200
0,pri
orto
GE
DO
pti
on).
35%
40%
45%
20%
25%
30% 5%10%
15% 0%
Dis
trict
Opt
ion
No
Dis
trict
Opt
ion
Any
GE
D O
ptio
nN
o O
ptio
n
Sou
rce:
Nati
on
al
Cen
ter
for
Ed
uca
tion
Sta
tist
ics,
Com
mon
Core
Data
an
dO
regon
Sch
ool
Dis
tric
tsA
dm
inis
trati
ve
Data
.N
ote
s:A
llm
easu
res
are
from
yea
r2000,
pri
or
toth
e
availab
ilit
yof
the
GE
DO
pti
on
pro
gra
m.
35
Fig
ure
F-6
:D
escr
ipti
veC
ompar
ison
sof
Dis
tric
tsw
ith
and
wit
hou
tG
ED
Opti
onP
rogr
ams
(200
0,pri
orto
GE
DO
pti
on)
101520253035404550 05
Dip
lom
a R
ate
(9th
)To
tal E
nrol
lmen
tP
upil
Teac
her
Rat
ioR
even
ue P
er
Pup
ilE
xpen
ditu
res
Per
Pup
ilP
er C
apita
In
com
eM
edia
n Fa
mily
In
com
e
Dis
trict
Opt
ion
No
Dis
trict
Opt
ion
Any
GE
D O
ptio
nN
o O
ptio
n
Sou
rce:
Nati
on
alC
ente
rfo
rE
du
cati
on
Sta
tist
ics,
Com
mon
Core
Data
an
dO
regon
Sch
oolD
istr
icts
Ad
min
istr
ati
ve
Data
.N
ote
s:T
ota
len
rollm
ent,
reven
ue
per
pu
pil,ex
pen
dit
ure
s
per
pu
pil,
per
-cap
ita
inco
me,
an
dm
edia
nfa
mily
inco
me
are
inth
ou
san
ds
of
yea
r2000
dollars
.A
llm
easu
res
are
from
yea
r2000,
pri
or
toth
eavailab
ilit
yof
the
GE
DO
pti
on
pro
gra
m.
36
Tab
leF
-2:
The
Eff
ect
ofD
istr
ict-
Wid
eO
pti
onP
rogr
ams
onC
ohor
tD
iplo
ma
Rat
esin
Ore
gon
Dip
lom
aR
ate
(8th
)D
iplo
ma
Rat
e(9
th)
Dip
lom
aR
ate
(10t
h)
Per
c.B
lack
Enro
llm
ent
1.02
1-0
.598
0.33
1(1
.071
)(0
.909
)(0
.712
)
Per
c.H
ispan
icE
nro
llm
ent
-0.8
53∗
0.07
840.
210
(0.5
14)
(0.2
93)
(0.2
45)
Per
cF
ree
Lunch
-0.3
45-0
.077
1-0
.210
(0.7
09)
(0.3
21)
(0.2
77)
Per
cR
educe
dor
Fre
eL
unch
-0.2
03-0
.237
-0.1
55(0
.452
)(0
.216
)(0
.200
)
Pupil
Tea
cher
Rat
io-0
.000
216
-0.0
0013
7-0
.000
342
(0.0
0024
4)(0
.000
244)
(0.0
0029
9)
Exp
endit
ure
sP
erP
upil
-0.0
0605
∗∗-0
.003
290.
0005
67(0
.002
79)
(0.0
0232
)(0
.001
98)
Rev
enue
Per
Pupil
-0.0
157
-0.0
0526
-0.0
0666
(0.0
123)
(0.0
108)
(0.0
110)
Dis
t.W
ide
Opti
on-0
.042
4∗∗
-0.0
297∗
-0.0
420∗
∗∗
(0.0
211)
(0.0
154)
(0.0
151)
Con
stan
t1.
214∗
∗∗0.
886∗
∗∗0.
916∗
∗∗
(0.1
81)
(0.1
29)
(0.1
23)
Obse
rvat
ions
1552
1552
1552
R2
0.54
20.
562
0.35
7A
dju
sted
R2
0.47
70.
500
0.26
6
Yea
ran
dD
istr
ict
Fix
edE
ffec
tsN
ot
Sh
own
En
dogen
ou
sva
riab
leis
aco
hort
gra
du
ati
on
mea
sure
∗p<
0.10
,∗∗
p<
0.0
5,∗∗
∗p<
0.01
37
Tab
leF
-3:
The
Eff
ect
ofD
istr
ict-
Wid
eO
pti
onP
rogr
ams
onC
ohor
tO
ther
-Com
ple
ter
Rat
esin
Ore
gon
Oth
er-C
omple
ter
Rat
e(8
th)
Oth
er-C
omple
ter
Rat
e(9
th)
Oth
er-C
omple
ter
Rat
e(1
0th)
Per
c.B
lack
Enro
llm
ent
0.67
40.
258
0.18
1(0
.833
)(1
.181
)(1
.087
)
Per
c.H
ispan
icE
nro
llm
ent
-0.1
38-0
.036
1-0
.133
(0.2
12)
(0.2
17)
(0.1
94)
Per
cF
ree
Lunch
-0.3
99-0
.285
-0.0
678
(0.5
03)
(0.4
85)
(0.2
97)
Per
cR
educe
dor
Fre
eL
unch
0.33
00.
262
0.06
59(0
.339
)(0
.327
)(0
.205
)
Pupil
Tea
cher
Rat
io0.
0040
7∗0.
0057
1∗∗
0.00
313
(0.0
0245
)(0
.002
85)
(0.0
0304
)
Exp
endit
ure
sP
erP
upil
-0.0
0020
30.
0001
990.
0020
3(0
.002
12)
(0.0
0219
)(0
.002
37)
Rev
enue
Per
Pupil
0.00
0921
0.00
506
0.00
380
(0.0
0788
)(0
.009
36)
(0.0
0924
)
Dis
t.W
ide
Opti
on0.
0173
∗0.
0174
∗0.
0174
∗
(0.0
104)
(0.0
101)
(0.0
0954
)
Con
stan
t0.
0310
-0.0
744
0.01
21(0
.111
)(0
.103
)(0
.111
)
Obse
rvat
ions
1134
1134
1134
R2
0.35
20.
330
0.32
5A
dju
sted
R2
0.22
00.
195
0.18
8
Yea
ran
dD
istr
ict
Fix
edE
ffec
tsN
otS
how
nE
nd
ogen
ou
sva
riab
leis
aco
hort
gra
du
ati
on
mea
sure
∗p<
0.10
,∗∗
p<
0.05
,∗∗
∗p<
0.0
1
38
Tab
leF
-4:
The
Eff
ect
ofO
pti
onP
rogr
ams
only
inA
lter
nat
ive
Sch
ool
son
Coh
ort
Dip
lom
aR
ates
inO
rego
n
Dip
lom
aR
ate
(8th
)D
iplo
ma
Rat
e(9
th)
Dip
lom
aR
ate
(10t
h)
Per
c.B
lack
Enro
llm
ent
0.98
5-0
.619
0.28
8(1
.068
)(0
.912
)(0
.709
)
Per
c.H
ispan
icE
nro
llm
ent
-0.8
330.
0883
0.24
0(0
.512
)(0
.291
)(0
.243
)
Per
cF
ree
Lunch
-0.3
58-0
.087
0-0
.223
(0.7
10)
(0.3
22)
(0.2
78)
Per
cR
educe
dor
Fre
eL
unch
-0.1
93-0
.231
-0.1
45(0
.452
)(0
.216
)(0
.200
)
Pupil
Tea
cher
Rat
io-0
.000
240
-0.0
0015
2-0
.000
368
(0.0
0024
4)(0
.000
247)
(0.0
0030
5)
Exp
endit
ure
sP
erP
upil
-0.0
0600
∗∗-0
.003
270.
0006
51(0
.002
80)
(0.0
0233
)(0
.001
99)
Rev
enue
Per
Pupil
-0.0
153
-0.0
0497
-0.0
0626
(0.0
122)
(0.0
108)
(0.0
110)
Opti
onP
rogr
amon
lyin
Alt
.In
stit
uti
ons
0.02
490.
0269
0.00
393
(0.0
222)
(0.0
191)
(0.0
205)
Con
stan
t1.
219∗
∗∗0.
890∗
∗∗0.
920∗
∗∗
(0.1
82)
(0.1
30)
(0.1
24)
Obse
rvat
ions
1552
1552
1552
R2
0.54
20.
562
0.35
6A
dju
sted
R2
0.47
70.
499
0.26
5
Yea
ran
dD
istr
ict
Fix
edE
ffec
tsN
otS
how
nE
nd
ogen
ou
sva
riab
leis
aco
hort
gra
du
ati
on
mea
sure
∗p<
0.1
0,∗∗
p<
0.05
,∗∗
∗p<
0.0
1
39
Tab
leF
-5:
The
Eff
ect
ofO
pti
onP
rogr
ams
only
inA
lter
nat
ive
Sch
ool
son
Coh
ort
Oth
er-C
omple
ter
Rat
esin
Ore
gon
Oth
er-C
omple
ter
Rat
e(8
th)
Oth
er-C
omple
ter
Rat
e(9
th)
Oth
er-C
omple
ter
Rat
e(1
0th)
Per
c.B
lack
Enro
llm
ent
0.70
70.
289
0.21
0(0
.832
)(1
.179
)(1
.085
)
Per
c.H
ispan
icE
nro
llm
ent
-0.1
67-0
.062
7-0
.157
(0.2
06)
(0.2
11)
(0.1
90)
Per
cF
ree
Lunch
-0.3
92-0
.278
-0.0
607
(0.5
04)
(0.4
86)
(0.2
97)
Per
cR
educe
dor
Fre
eL
unch
0.32
60.
259
0.06
19(0
.340
)(0
.327
)(0
.205
)
Pupil
Tea
cher
Rat
io0.
0041
4∗0.
0057
8∗∗
0.00
320
(0.0
0245
)(0
.002
84)
(0.0
0303
)
Exp
endit
ure
sP
erP
upil
-0.0
0028
00.
0001
410.
0019
9(0
.002
14)
(0.0
0220
)(0
.002
38)
Rev
enue
Per
Pupil
0.00
0701
0.00
483
0.00
355
(0.0
0787
)(0
.009
34)
(0.0
0923
)
Opti
onP
rogr
amon
lyin
Alt
.In
stit
uti
ons
0.01
760.
0136
0.01
02(0
.013
6)(0
.013
3)(0
.013
6)
Con
stan
t0.
0309
-0.0
749
0.01
13(0
.111
)(0
.103
)(0
.111
)
Obse
rvat
ions
1134
1134
1134
R2
0.35
10.
330
0.32
4A
dju
sted
R2
0.22
00.
194
0.18
7
Yea
ran
dD
istr
ict
Fix
edE
ffec
tsN
otS
how
nE
nd
ogen
ou
sva
riab
leis
aco
hort
gra
du
ati
on
mea
sure
∗p<
0.1
0,∗∗
p<
0.05
,∗∗
∗p<
0.01
40
Tab
leF
-6:
The
Eff
ect
ofO
pti
onP
rogr
ams
inT
radit
ional
Sch
ool
son
Coh
ort
Dip
lom
aR
ates
inO
rego
n
Dip
lom
aR
ate
(8th
)D
iplo
ma
Rat
e(9
th)
Dip
lom
aR
ate
(10t
h)
Per
c.B
lack
Enro
llm
ent
1.18
8-0
.464
0.48
3(1
.098
)(0
.924
)(0
.705
)
Per
c.H
ispan
icE
nro
llm
ent
-0.8
050.
113
0.25
7(0
.511
)(0
.291
)(0
.241
)
Per
cF
ree
Lunch
-0.3
43-0
.075
0-0
.210
(0.7
07)
(0.3
20)
(0.2
77)
Per
cR
educe
dor
Fre
eL
unch
-0.2
06-0
.240
-0.1
57(0
.450
)(0
.216
)(0
.201
)
Pupil
Tea
cher
Rat
io-0
.000
233
-0.0
0014
8-0
.000
359
(0.0
0024
2)(0
.000
243)
(0.0
0029
9)
Exp
endit
ure
sP
erP
upil
-0.0
0572
∗∗-0
.003
040.
0008
77(0
.002
82)
(0.0
0234
)(0
.001
98)
Rev
enue
Per
Pupil
-0.0
162
-0.0
0567
-0.0
0708
(0.0
122)
(0.0
108)
(0.0
110)
Opti
onP
rog.
inR
egula
rSch
ool
s-0
.038
4∗-0
.030
0∗-0
.035
5∗
(0.0
218)
(0.0
170)
(0.0
192)
Con
stan
t1.
206∗
∗∗0.
880∗
∗∗0.
910∗
∗∗
(0.1
79)
(0.1
28)
(0.1
22)
Obse
rvat
ions
1552
1552
1552
R2
0.54
20.
562
0.35
7A
dju
sted
R2
0.47
70.
500
0.26
6
Yea
ran
dD
istr
ict
Fix
edE
ffec
tsN
ot
Sh
own
En
dogen
ou
sva
riab
leis
aco
hort
gra
du
ati
on
mea
sure
∗p<
0.1
0,∗∗
p<
0.05
,∗∗
∗p<
0.01
41
Tab
leF
-7:
The
Eff
ect
ofO
pti
onP
rogr
ams
inT
radit
ional
Sch
ool
son
Coh
ort
Oth
er-C
omple
ter
Rat
esin
Ore
gon
Oth
er-C
omple
ter
Rat
e(8
th)
Oth
er-C
omple
ter
Rat
e(9
th)
Oth
er-C
omple
ter
Rat
e(1
0th)
Per
c.B
lack
Enro
llm
ent
0.63
90.
229
0.12
0(0
.843
)(1
.187
)(1
.097
)
Per
c.H
ispan
icE
nro
llm
ent
-0.1
61-0
.058
4-0
.156
(0.2
05)
(0.2
11)
(0.1
89)
Per
cF
ree
Lunch
-0.3
96-0
.281
-0.0
669
(0.5
03)
(0.4
85)
(0.2
97)
Per
cR
educe
dor
Fre
eL
unch
0.32
90.
261
0.06
64(0
.339
)(0
.326
)(0
.204
)
Pupil
Tea
cher
Rat
io0.
0040
5∗0.
0057
0∗∗
0.00
308
(0.0
0244
)(0
.002
84)
(0.0
0305
)
Exp
endit
ure
sP
erP
upil
-0.0
0024
20.
0001
650.
0019
7(0
.002
13)
(0.0
0219
)(0
.002
36)
Rev
enue
Per
Pupil
0.00
0786
0.00
491
0.00
372
(0.0
0788
)(0
.009
35)
(0.0
0922
)
Opti
onP
rog.
inR
egula
rSch
ool
s0.
0107
0.00
956
0.01
51(0
.011
2)(0
.011
0)(0
.010
8)
Con
stan
t0.
0341
-0.0
718
0.01
71(0
.110
)(0
.102
)(0
.111
)
Obse
rvat
ions
1134
1134
1134
R2
0.35
10.
330
0.32
5A
dju
sted
R2
0.22
00.
194
0.18
8
Yea
ran
dD
istr
ict
Fix
edE
ffec
tsN
otS
how
nE
nd
ogen
ou
sva
riab
leis
aco
hort
gra
du
ati
on
mea
sure
∗p<
0.1
0,∗∗
p<
0.05
,∗∗
∗p<
0.01
42
G Additional Supplementary Materials
43
Fig
ure
G-1
:G
raduat
ion
Rat
eB
efor
ean
dA
fter
Imple
men
ting
the
GE
DP
rogr
am,
Cal
ifor
nia
vs.
All
other
Sta
tes
78.0
%78.4
%
77.3
%
74.3
%74.6
%
72.7
%
71.
8%
76.1
%
75.8
%
74.7
%74.7
%
73.5
%73.7
%73.8
%
65%
67%
69%
71%
73%
75%
77%
79%
81%
83%
85%
1971
1972
1973
1974
1975
1976
1977
Graduation Rate
Year
Fig
ure
A-1
. G
rad
uat
ion
Rat
e B
efo
re a
nd
Aft
er I
mp
lem
enti
ng
the
GE
D P
rogra
m, C
alif
orn
ia v
s. A
ll o
ther
Sta
tes
Cal
ifo
rnia
U.S
. (E
xclu
din
g C
alif
orn
ia)
No
tes:
Auth
ors
' cal
cula
tio
ns
bas
ed o
n N
CE
S d
ata.
Th
e gr
aduat
ion
rat
e is
th
e n
um
ber
of
regu
lar
pub
lic a
nd
pri
vat
e h
igh
sch
oo
l d
iplo
mas
iss
ued
over
th
e 14 y
ear
old
po
pula
tio
n
four
year
s p
revio
us.
Po
pula
tio
n t
ota
ls f
or
the
U.S
. w
ere
ob
tain
ed f
rom
th
e U
.S. C
ensu
s B
ure
au. C
alif
orn
ia p
op
ula
tio
n e
stim
ates
wer
e o
bta
ined
fro
m t
he
Cal
ifo
rnia
Dem
ogr
aph
ic
Res
earc
h U
nit
.
Note
s:A
uth
ors
’ca
lcu
lati
on
sb
ase
don
NC
ES
data
.T
he
gra
du
ati
on
rate
isth
enu
mb
erof
regu
lar
pu
blic
an
dp
rivate
hig
hsc
hool
dip
lom
as
issu
edover
the
14
yea
rold
pop
ula
tion
fou
ryea
rsp
revio
us.
Pop
ula
tion
tota
lsfo
rth
eU
.S.
wer
eob
tain
edfr
om
the
U.S
.C
ensu
sB
ure
au
.C
ali
forn
iap
op
ula
tion
esti
mate
sw
ere
ob
tain
edfr
om
the
Califo
rnia
Dem
ogra
ph
ic
Res
earc
hU
nit
.
44
Tab
leG
-1:
Fix
edE
ffec
tsE
stim
ates
ofth
eE
ffec
tof
the
Ref
orm
onG
ED
Tes
tT
akin
gR
ates
by
You
nge
rC
ohor
ts
Ind
epen
den
t V
aria
ble
sG
ED
Tes
t T
akin
g R
ate
(Age
s 16
-19)
GE
D T
est
Tak
ing
Rat
e (A
ges
16-1
7)G
ED
Tes
t T
akin
g R
ate
(Age
s 18
-19)
Post
199
7 du
mm
y-0
.000
20.
0007
-0.0
012
(0.0
007)
(0.0
006)
(0.0
011)
Scor
e op
tion
chan
ger s
tate
pos
t 199
7 (tr
eatm
ent e
ffect
)-0
.005
5-0
.003
4-0
.007
6(0
.001
6)(0
.001
5)(0
.002
1)Lo
cal u
nem
ploy
men
t rat
e0.
0020
0.00
190.
0021
(0.0
005)
(0.0
007)
(0.0
008)
Log
of p
er c
apita
inco
me
0.02
130.
0184
0.02
40(0
.006
1)(0
.006
4)(0
.008
4)O
bser
vatio
ns14
314
314
3N
umbe
r of S
tate
s24
2424
Not
e: H
uber
-Whi
te ro
bust
stan
dard
err
ors a
re in
par
enth
eses
(clu
ster
ed b
y st
ate)
. Mod
el is
estim
ated
usin
g O
LS. S
tate
15-
17 y
ear o
ld p
opul
atio
ns a
re
used
as w
eight
s. Th
e tre
atm
ent e
ffect
repo
rted
abov
e is
the
inte
ract
ion
betw
een
the
treat
men
t sta
te d
umm
y an
d th
e po
st 1
997
dum
my,
whe
re th
e tre
atm
ent s
tate
dum
my
is eq
ual t
o 1
for t
reat
men
t sta
tes a
nd th
e po
st 1
997
dum
my
is eq
ual t
o 1
for t
he y
ears
199
8-20
00, o
ther
wise
bot
h du
mm
ies
are
equa
l to
0. S
tate
s with
few
er th
an tw
o ob
serv
atio
ns p
er p
erio
d ar
e dr
oppe
d. T
reat
men
t sta
tes a
re th
ose
stat
es th
at w
ere
requ
ired
to e
limin
ate
the
and/
or sc
ore
optio
n. T
hese
inclu
de: L
A, M
S, N
E, N
M, T
X. C
ontro
l sta
tes a
re th
ose
that
alre
ady
had
high
eno
ugh
stan
dard
s by
1997
. The
se in
clude
: A
R, C
A, C
O, D
E, D
C, F
L, ID
, KY
, MD
, MO
, NY
, ND
, OK
, OR,
SD
, UT,
WA
, WV
, WI.
Con
trol s
tate
s dro
pped
due
to m
issin
g an
d ne
gativ
e dr
opou
t rat
es in
clude
: NJ.
No
treat
men
t sta
tes a
re d
ropp
ed.
45
Tab
leG
-2:
Fix
edE
ffec
tsE
stim
ates
ofth
eE
ffec
tof
the
Ref
orm
onG
ED
Tes
tT
akin
gR
ates
by
You
nge
rC
ohor
tsC
ontr
olling
for
Age
Req
uir
emen
ts
Ind
epen
den
t V
aria
ble
sG
ED
Tes
t T
akin
g R
ate
(Age
s 16
-19)
GE
D T
est
Tak
ing
Rat
e (A
ges
16-1
7)G
ED
Tes
t T
akin
g R
ate
(Age
s 18
-19)
Post
199
7 du
mm
y-0
.000
10.
0007
-0.0
011
(0.0
008)
(0.0
007)
(0.0
011)
Scor
e op
tion
chan
ger s
tate
pos
t 199
7 (tr
eatm
ent e
ffect
)-0
.005
7-0
.003
6-0
.007
7(0
.001
8)(0
.001
7)(0
.002
5)Lo
cal u
nem
ploy
men
t rat
e0.
0020
0.00
190.
0021
(0.0
006)
(0.0
007)
(0.0
009)
Log
of p
er c
apita
inco
me
0.02
140.
0185
0.02
41(0
.006
5)(0
.007
0)(0
.008
9)M
inim
um G
ED
age
requ
irem
ent i
s 16
0.00
090.
0007
0.00
13(d
umm
y fo
r min
imum
GE
D a
ge re
quire
men
t set
abo
ve 1
8 is
left o
ut)
(0.0
016)
(0.0
019)
(0.0
015)
Min
imum
GE
D a
ge re
quire
men
t is 1
7 -0
.001
-0.0
013
-0.0
005
(0.0
017)
(0.0
020)
(0.0
020)
Min
imum
GE
D a
ge re
quire
men
t is 1
8 -0
.000
20.
0001
-0.0
002
(0.0
014)
(0.0
015)
(0.0
013)
Min
imum
scho
ol le
avin
g ag
e is
16-0
.000
4-0
.000
70.
0001
(dum
my
for m
inim
um sc
hool
leav
ing
age
set a
bove
17
is lef
t out
)(0
.001
9)(0
.000
9)(0
.003
2)M
inim
um sc
hool
leav
ing
age
is 17
0.00
050.
0000
0.00
11(0
.001
7)(0
.001
2)(0
.003
6)O
bser
vatio
ns14
314
314
3N
umbe
r of S
tate
s24
2424
Not
e: H
uber
-Whi
te ro
bust
stan
dard
err
ors a
re in
par
enth
eses
(clu
ster
ed b
y st
ate)
. Mod
el is
estim
ated
usin
g O
LS. S
tate
15-
17 y
ear o
ld p
opul
atio
ns a
re
used
as w
eight
s. Th
e tre
atm
ent e
ffect
repo
rted
abov
e is
the
inte
ract
ion
betw
een
the
treat
men
t sta
te d
umm
y an
d th
e po
st 1
997
dum
my,
whe
re th
e tre
atm
ent s
tate
dum
my
is eq
ual t
o 1
for t
reat
men
t sta
tes a
nd th
e po
st 1
997
dum
my
is eq
ual t
o 1
for t
he y
ears
199
8-20
00, o
ther
wise
bot
h du
mm
ies
are
equa
l to
0. S
tate
s with
few
er th
an tw
o ob
serv
atio
ns p
er p
erio
d ar
e dr
oppe
d. T
reat
men
t sta
tes a
re th
ose
stat
es th
at w
ere
requ
ired
to e
limin
ate
the
and/
or sc
ore
optio
n. T
hese
inclu
de: L
A, M
S, N
E, N
M, T
X. C
ontro
l sta
tes a
re th
ose
that
alre
ady
had
high
eno
ugh
stan
dard
s by
1997
. The
se in
clude
: A
R, C
A, C
O, D
E, D
C, F
L, ID
, KY
, MD
, MO
, NY
, ND
, OK
, OR,
SD
, UT,
WA
, WV
, WI.
Con
trol s
tate
s dro
pped
due
to m
issin
g an
d ne
gativ
e dr
opou
t rat
es in
clude
: NJ.
No
treat
men
t sta
tes a
re d
ropp
ed.
46
Tab
leG
-3:
Fix
edE
ffec
tsE
stim
ates
ofth
eE
ffec
tof
the
Ref
orm
onD
rop
out
Rat
es(A
llR
aces
)
Ind
epen
den
t V
aria
ble
s10
th-1
2th
Gra
de
Dro
pou
t R
ate
10th
-11t
h G
rad
e D
rop
out
Rat
e12
th G
rad
e D
rop
out
Rat
e
Post
199
7 du
mm
y0.
0089
0.01
130.
003
(0.0
039)
(0.0
047)
(0.0
066)
Scor
e op
tion
chan
ger s
tate
pos
t 199
7 (tr
eatm
ent e
ffect
)-0
.012
1-0
.004
6-0
.028
6(0
.003
5)(0
.005
1)(0
.006
7)Lo
cal u
nem
ploy
men
t rat
e0.
0041
0.00
83-0
.007
8(0
.003
0)(0
.005
7)(0
.005
6)Lo
g of
per
cap
ita in
com
e-0
.028
5-0
.007
3-0
.099
4(0
.027
3)(0
.037
1)(0
.039
1)O
bser
vatio
ns14
414
414
2N
umbe
r of S
tate
s24
2424
Not
e: H
uber
-Whi
te ro
bust
stan
dard
err
ors a
re in
par
enth
eses
(clu
ster
ed b
y st
ate)
. Mod
el is
estim
ated
usin
g O
LS. S
tate
15-
17 y
ear o
ld p
opul
atio
ns a
re
used
as w
eight
s. Th
e tre
atm
ent e
ffect
repo
rted
abov
e is
the
inte
ract
ion
betw
een
the
treat
men
t sta
te d
umm
y an
d th
e po
st 1
997
dum
my,
whe
re th
e tre
atm
ent s
tate
dum
my
is eq
ual t
o 1
for t
reat
men
t sta
tes a
nd th
e po
st 1
997
dum
my
is eq
ual t
o 1
for t
he y
ears
199
8-20
00, o
ther
wise
bot
h du
mm
ies
are
equa
l to
0. S
tate
s with
few
er th
an tw
o ob
serv
atio
ns p
er p
erio
d ar
e dr
oppe
d. T
reat
men
t sta
tes a
re th
ose
stat
es th
at w
ere
requ
ired
to e
limin
ate
the
and/
or sc
ore
optio
n. T
hese
inclu
de: L
A, M
S, N
E, N
M, T
X. C
ontro
l sta
tes a
re th
ose
that
alre
ady
had
high
eno
ugh
stan
dard
s by
1997
. The
se in
clude
: A
R, C
A, C
O, D
E, D
C, F
L, ID
, KY
, MD
, MO
, NJ,
NY
, ND
, OK
, OR,
SD
, UT,
WA
, WV
, WI.
Con
trol s
tate
s dro
pped
due
to m
issin
g an
d ne
gativ
e dr
opou
t rat
es in
clude
: NJ.
No
treat
men
t sta
tes a
re d
ropp
ed.
47
Tab
leG
-4:
Fix
edE
ffec
tsE
stim
ates
ofth
eE
ffec
tof
the
Ref
orm
onD
rop
out
Rat
es(W
hit
es)
Ind
epen
den
t V
aria
ble
s10
th-1
2th
Gra
de
Dro
pou
t R
ate
10th
-11t
h G
rad
e D
rop
out
Rat
e12
th G
rad
e D
rop
out
Rat
e
Post
199
7 du
mm
y0.
0073
0.00
820.
0042
(0.0
031)
(0.0
054)
(0.0
064)
Scor
e op
tion
chan
ger s
tate
pos
t 199
7 (tr
eatm
ent e
ffect
)-0
.004
7-0
.000
2-0
.013
8(0
.001
5)(0
.004
3)(0
.008
2)Lo
cal u
nem
ploy
men
t rat
e0.
0037
0.00
61-0
.001
7(0
.001
5)(0
.004
0)(0
.005
0)Lo
g of
per
cap
ita in
com
e-0
.030
9-0
.014
9-0
.066
1(0
.019
1)(0
.029
1)(0
.024
7)O
bser
vatio
ns84
8382
Num
ber o
f Sta
tes
1414
14N
ote:
Hub
er-W
hite
robu
st st
anda
rd e
rror
s are
in p
aren
thes
es (c
lust
ered
by
stat
e). M
odel
is es
timat
ed u
sing
OLS
. Sta
te 1
5-17
yea
r old
pop
ulat
ions
are
us
ed a
s weig
hts.
The
treat
men
t effe
ct re
porte
d ab
ove
is th
e in
tera
ctio
n be
twee
n th
e tre
atm
ent s
tate
dum
my
and
the
post
199
7 du
mm
y, w
here
the
treat
men
t sta
te d
umm
y is
equa
l to
1 fo
r tre
atm
ent s
tate
s and
the
post
199
7 du
mm
y is
equa
l to
1 fo
r the
yea
rs 1
998-
2000
, oth
erw
ise b
oth
dum
mies
ar
e eq
ual t
o 0.
Sta
tes w
ith fe
wer
than
two
obse
rvat
ions
per
per
iod
for a
ny o
f the
dro
pout
rate
mea
sure
s by
race
are
dro
pped
. Tre
atm
ent s
tate
s are
th
ose
stat
es th
at w
ere
requ
ired
to e
limin
ate
the
and/
or sc
ore
optio
n. T
hese
inclu
de: L
A, M
S, N
E, N
M, T
X. C
ontro
l sta
tes a
re th
ose
that
alre
ady
had
high
eno
ugh
stan
dard
s by
1997
. The
se in
clude
: AR,
CA
, CO
, DE
, DC,
FL,
ID, K
Y, M
D, M
O, N
J, N
Y, N
D, O
K, O
R, S
D, U
T, W
A, W
V, W
I.
Cont
rol s
tate
s dro
pped
due
to m
issin
g an
d ne
gativ
e dr
opou
t rat
es in
clude
: AR,
ID, K
Y, M
O, N
D, N
J, N
Y, S
D, U
T, W
A, W
V. N
o tre
atm
ent s
tate
s ar
e dr
oppe
d.
48
Tab
leG
-5:
Fix
edE
ffec
tsE
stim
ates
ofth
eE
ffec
tof
the
Ref
orm
onD
rop
out
Rat
es(B
lack
s)
Ind
epen
den
t V
aria
ble
s10
th-1
2th
Gra
de
Dro
pou
t R
ate
10th
-11t
h G
rad
e D
rop
out
Rat
e12
th G
rad
e D
rop
out
Rat
e
Post
199
7 du
mm
y0.
0145
0.01
380.
0156
(0.0
098)
(0.0
141)
(0.0
251)
Scor
e op
tion
chan
ger s
tate
pos
t 199
7 (tr
eatm
ent e
ffect
)-0
.008
30.
0037
-0.0
402
(0.0
094)
(0.0
148)
(0.0
176)
Loca
l une
mpl
oym
ent r
ate
-0.0
032
-0.0
004
-0.0
109
(0.0
067)
(0.0
118)
(0.0
160)
Log
of p
er c
apita
inco
me
-0.1
364
-0.1
245
-0.1
584
(0.0
688)
(0.0
760)
(0.1
409)
Obs
erva
tions
8484
82N
umbe
r of S
tate
s14
1414
Not
e: H
uber
-Whi
te ro
bust
stan
dard
err
ors a
re in
par
enth
eses
(clu
ster
ed b
y st
ate)
. Mod
el is
estim
ated
usin
g O
LS. S
tate
15-
17 y
ear o
ld p
opul
atio
ns a
re
used
as w
eight
s. Th
e tre
atm
ent e
ffect
repo
rted
abov
e is
the
inte
ract
ion
betw
een
the
treat
men
t sta
te d
umm
y an
d th
e po
st 1
997
dum
my,
whe
re th
e tre
atm
ent s
tate
dum
my
is eq
ual t
o 1
for t
reat
men
t sta
tes a
nd th
e po
st 1
997
dum
my
is eq
ual t
o 1
for t
he y
ears
199
8-20
00, o
ther
wise
bot
h du
mm
ies
are
equa
l to
0. S
tate
s with
few
er th
an tw
o ob
serv
atio
ns p
er p
erio
d fo
r any
of t
he d
ropo
ut ra
te m
easu
res b
y ra
ce a
re d
ropp
ed. T
reat
men
t sta
tes a
re
thos
e st
ates
that
wer
e re
quire
d to
elim
inat
e th
e an
d/or
scor
e op
tion.
The
se in
clude
: LA
, MS,
NE
, NM
, TX
. Con
trol s
tate
s are
thos
e th
at a
lread
y ha
d hi
gh e
noug
h st
anda
rds b
y 19
97. T
hese
inclu
de: A
R, C
A, C
O, D
E, D
C, F
L, ID
, KY
, MD
, MO
, NJ,
NY
, ND
, OK
, OR,
SD
, UT,
WA
, WV
, WI.
Co
ntro
l sta
tes d
ropp
ed d
ue to
miss
ing
and
nega
tive
drop
out r
ates
inclu
de: A
R, ID
, KY
, MO
, ND
, NJ,
NY
, SD
, UT,
WA
, WV
. No
treat
men
t sta
tes
are
drop
ped.
49
Tab
leG
-6:
Fix
edE
ffec
tsE
stim
ates
ofth
eE
ffec
tof
the
Ref
orm
onD
rop
out
Rat
es(H
ispan
ics)
Ind
epen
den
t V
aria
ble
s10
th-1
2th
Gra
de
Dro
pou
t R
ate
10th
-11t
h G
rad
e D
rop
out
Rat
e12
th G
rad
e D
rop
out
Rat
e
Post
199
7 du
mm
y0.
0136
0.00
970.
0233
(0.0
062)
(0.0
064)
(0.0
074)
Scor
e op
tion
chan
ger s
tate
pos
t 199
7 (tr
eatm
ent e
ffect
)-0
.026
8-0
.013
8-0
.059
9(0
.004
3)(0
.006
0)(0
.003
0)Lo
cal u
nem
ploy
men
t rat
e0.
0120
0.02
02(0
.011
4)(0
.004
6)(0
.006
9)(0
.006
3)Lo
g of
per
cap
ita in
com
e-0
.004
90.
0674
-0.2
041
(0.0
562)
(0.0
677)
(0.0
611)
Obs
erva
tions
8483
82N
umbe
r of S
tate
s14
1414
Not
e: H
uber
-Whi
te ro
bust
stan
dard
err
ors a
re in
par
enth
eses
(clu
ster
ed b
y st
ate)
. Mod
el is
estim
ated
usin
g O
LS. S
tate
15-
17 y
ear o
ld p
opul
atio
ns a
re
used
as w
eight
s. Th
e tre
atm
ent e
ffect
repo
rted
abov
e is
the
inte
ract
ion
betw
een
the
treat
men
t sta
te d
umm
y an
d th
e po
st 1
997
dum
my,
whe
re th
e tre
atm
ent s
tate
dum
my
is eq
ual t
o 1
for t
reat
men
t sta
tes a
nd th
e po
st 1
997
dum
my
is eq
ual t
o 1
for t
he y
ears
199
8-20
00, o
ther
wise
bot
h du
mm
ies
are
equa
l to
0. S
tate
s with
few
er th
an tw
o ob
serv
atio
ns p
er p
erio
d fo
r any
of t
he d
ropo
ut ra
te m
easu
res b
y ra
ce a
re d
ropp
ed. T
reat
men
t sta
tes a
re
thos
e st
ates
that
wer
e re
quire
d to
elim
inat
e th
e an
d/or
scor
e op
tion.
The
se in
clude
: LA
, MS,
NE
, NM
, TX
. Con
trol s
tate
s are
thos
e th
at a
lread
y ha
d hi
gh e
noug
h st
anda
rds b
y 19
97. T
hese
inclu
de: A
R, C
A, C
O, D
E, D
C, F
L, ID
, KY
, MD
, MO
, NJ,
NY
, ND
, OK
, OR,
SD
, UT,
WA
, WV
, WI.
Cont
rol s
tate
s dro
pped
due
to m
issin
g an
d ne
gativ
e dr
opou
t rat
es in
clude
: AR,
ID, K
Y, M
O, N
D, N
J, N
Y, S
D, U
T, W
A, W
V. N
o tre
atm
ent s
tate
s ar
e dr
oppe
d.
50
Tab
leG
-7:
Fix
edE
ffec
tsE
stim
ates
ofth
eE
ffec
tof
the
Ref
orm
onD
rop
out
Rat
esC
ontr
olling
for
Age
Req
uir
emen
ts(A
llR
aces
)
Ind
epen
den
t V
aria
ble
s10
th-1
2th
Gra
de
Dro
pou
t R
ate
10th
-11t
h G
rad
e D
rop
out
Rat
e12
th G
rad
e D
rop
out
Rat
e
Post
199
7 du
mm
y0.
0093
0.01
250.
0023
(0.0
040)
(0.0
050)
(0.0
066)
Scor
e op
tion
chan
ger s
tate
pos
t 199
7 (tr
eatm
ent e
ffect
)-0
.012
9-0
.005
5-0
.029
5(0
.003
7)(0
.005
4)(0
.006
5)Lo
cal u
nem
ploy
men
t rat
e0.
0043
0.00
90-0
.008
4(0
.003
1)(0
.006
0)(0
.005
7)Lo
g of
per
cap
ita in
com
e-0
.027
6-0
.003
7-0
.101
8(0
.028
5)(0
.039
2)(0
.039
2)M
inim
um G
ED
age
requ
irem
ent i
s 16
0.00
470.
0145
-0.0
092
(dum
my
for m
inim
um G
ED
age
requ
irem
ent s
et a
bove
18
is lef
t out
)(0
.004
3)(0
.003
7)(0
.005
7)M
inim
um G
ED
age
requ
irem
ent i
s 17
0.00
650.
0155
-0.0
101
(0.0
054)
(0.0
053)
(0.0
117)
Min
imum
GE
D a
ge re
quire
men
t is 1
8 0.
0035
0.01
14-0
.008
6(0
.003
8)(0
.002
3)(0
.004
8)M
inim
um sc
hool
leav
ing
age
is 16
0.00
70.
0073
0.00
39(d
umm
y fo
r min
imum
scho
ol le
avin
g ag
e se
t abo
ve 1
7 is
left o
ut)
(0.0
058)
(0.0
071)
(0.0
060)
Min
imum
scho
ol le
avin
g ag
e is
170.
0187
0.01
520.
0187
(0.0
091)
(0.0
099)
(0.0
139)
Obs
erva
tions
144
144
142
Num
ber o
f Sta
tes
2424
24N
ote:
Hub
er-W
hite
robu
st st
anda
rd e
rror
s are
in p
aren
thes
es (c
lust
ered
by
stat
e). M
odel
is es
timat
ed u
sing
OLS
. Sta
te 1
5-17
yea
r old
pop
ulat
ions
are
us
ed a
s weig
hts.
The
treat
men
t effe
ct re
porte
d ab
ove
is th
e in
tera
ctio
n be
twee
n th
e tre
atm
ent s
tate
dum
my
and
the
post
199
7 du
mm
y, w
here
the
treat
men
t sta
te d
umm
y is
equa
l to
1 fo
r tre
atm
ent s
tate
s and
the
post
199
7 du
mm
y is
equa
l to
1 fo
r the
yea
rs 1
998-
2000
, oth
erw
ise b
oth
dum
mies
ar
e eq
ual t
o 0.
Sta
tes w
ith fe
wer
than
two
obse
rvat
ions
per
per
iod
are
drop
ped.
Tre
atm
ent s
tate
s are
thos
e st
ates
that
wer
e re
quire
d to
elim
inat
e th
e an
d/or
scor
e op
tion.
The
se in
clude
: LA
, MS,
NE
, NM
, TX
. Con
trol s
tate
s are
thos
e th
at a
lread
y ha
d hi
gh e
noug
h st
anda
rds b
y 19
97. T
hese
inclu
de:
AR,
CA
, CO
, DE
, DC,
FL,
ID, K
Y, M
D, M
O, N
Y, N
D, O
K, O
R, S
D, U
T, W
A, W
V, W
I. C
ontro
l sta
tes d
ropp
ed d
ue to
miss
ing
and
nega
tive
drop
out r
ates
inclu
de: N
J. N
o tre
atm
ent s
tate
s are
dro
pped
.
51
Tab
leG
-8:
Fix
edE
ffec
tsE
stim
ates
ofth
eE
ffec
tof
the
Ref
orm
onD
rop
out
Rat
esC
ontr
olling
for
Age
Req
uir
emen
ts(W
hit
es)
Ind
epen
den
t V
aria
ble
s10
th-1
2th
Gra
de
Dro
pou
t R
ate
10th
-11t
h G
rad
e D
rop
out
Rat
e12
th G
rad
e D
rop
out
Rat
e
Post
199
7 du
mm
y0.
0076
0.00
870.
0042
(0.0
033)
(0.0
056)
(0.0
066)
Scor
e op
tion
chan
ger s
tate
pos
t 199
7 (tr
eatm
ent e
ffect
)-0
.004
30.
0000
-0.0
132
(0.0
015)
(0.0
045)
(0.0
088)
Loca
l une
mpl
oym
ent r
ate
0.00
370.
0065
-0.0
026
(0.0
016)
(0.0
044)
(0.0
054)
Log
of p
er c
apita
inco
me
-0.0
320
-0.0
132
-0.0
741
(0.0
200)
(0.0
313)
(0.0
275)
Min
imum
GE
D a
ge re
quire
men
t is 1
6 0.
0014
0.00
59-0
.008
1(d
umm
y fo
r min
imum
GE
D a
ge re
quire
men
t set
abo
ve 1
8 is
left o
ut)
(0.0
031)
(0.0
053)
(0.0
040)
Min
imum
GE
D a
ge re
quire
men
t is 1
7 0.
0019
0.00
92-0
.016
8(0
.005
9)(0
.006
0)(0
.018
8)M
inim
um G
ED
age
requ
irem
ent i
s 18
0.00
230.
0051
-0.0
04(0
.000
6)(0
.001
1)(0
.001
6)M
inim
um sc
hool
leav
ing
age
is 16
0.01
350.
0126
0.01
15(d
umm
y fo
r min
imum
scho
ol le
avin
g ag
e se
t abo
ve 1
7 is
left o
ut)
(0.0
044)
(0.0
037)
(0.0
061)
Min
imum
scho
ol le
avin
g ag
e is
170.
0148
0.01
76-0
.001
(0.0
092)
(0.0
054)
(0.0
213)
Obs
erva
tions
8483
82N
umbe
r of S
tate
s14
1414
Not
e: H
uber
-Whi
te ro
bust
stan
dard
err
ors a
re in
par
enth
eses
(clu
ster
ed b
y st
ate)
. Mod
el is
estim
ated
usin
g O
LS. S
tate
15-
17 y
ear o
ld p
opul
atio
ns a
re
used
as w
eight
s. Th
e tre
atm
ent e
ffect
repo
rted
abov
e is
the
inte
ract
ion
betw
een
the
treat
men
t sta
te d
umm
y an
d th
e po
st 1
997
dum
my,
whe
re th
e tre
atm
ent s
tate
dum
my
is eq
ual t
o 1
for t
reat
men
t sta
tes a
nd th
e po
st 1
997
dum
my
is eq
ual t
o 1
for t
he y
ears
199
8-20
00, o
ther
wise
bot
h du
mm
ies
are
equa
l to
0. S
tate
s with
few
er th
an tw
o ob
serv
atio
ns p
er p
erio
d fo
r any
of t
he d
ropo
ut ra
te m
easu
res b
y ra
ce a
re d
ropp
ed. T
reat
men
t sta
tes a
re
thos
e st
ates
that
wer
e re
quire
d to
elim
inat
e th
e an
d/or
scor
e op
tion.
The
se in
clude
: LA
, MS,
NE
, NM
, TX
. Con
trol s
tate
s are
thos
e th
at a
lread
y ha
d hi
gh e
noug
h st
anda
rds b
y 19
97. T
hese
inclu
de: C
A, C
O, D
E, D
C, F
L, M
D,
OK
, OR,
WI.
Con
trol s
tate
s dro
pped
due
to m
issin
g an
d ne
gativ
e dr
opou
t rat
es in
clude
: AR,
ID, K
Y, M
O, N
D, N
J, N
Y, S
D, U
T, W
A, W
V. N
o tre
atm
ent s
tate
s are
dro
pped
.
52
Tab
leG
-9:
Fix
edE
ffec
tsE
stim
ates
ofth
eE
ffec
tof
the
Ref
orm
onD
rop
out
Rat
esC
ontr
olling
for
Age
Req
uir
emen
ts(B
lack
s
Ind
epen
den
t V
aria
ble
s10
th-1
2th
Gra
de
Dro
pou
t R
ate
10th
-11t
h G
rad
e D
rop
out
Rat
e12
th G
rad
e D
rop
out
Rat
e
Post
199
7 du
mm
y0.
0179
0.01
850.
0143
(0.0
101)
(0.0
145)
(0.0
258)
Scor
e op
tion
chan
ger s
tate
pos
t 199
7 (tr
eatm
ent e
ffect
)-0
.012
90.
0009
-0.0
484
(0.0
098)
(0.0
161)
(0.0
182)
Loca
l une
mpl
oym
ent r
ate
-0.0
026
0.00
08-0
.012
1(0
.006
7)(0
.012
5)(0
.016
7)Lo
g of
per
cap
ita in
com
e-0
.136
2-0
.123
9-0
.158
9(0
.065
2)(0
.073
3)(0
.140
5)M
inim
um G
ED
age
requ
irem
ent i
s 16
0.06
300.
0697
0.04
10(d
umm
y fo
r min
imum
GE
D a
ge re
quire
men
t set
abo
ve 1
8 is
left o
ut)
(0.0
080)
(0.0
138)
(0.0
156)
Min
imum
GE
D a
ge re
quire
men
t is 1
7 0.
0296
0.03
8-0
.025
7(0
.014
6)(0
.013
9)(0
.020
9)M
inim
um G
ED
age
requ
irem
ent i
s 18
0.05
190.
055
0.04
17(0
.003
8)(0
.005
1)(0
.008
8)M
inim
um sc
hool
leav
ing
age
is 16
0.05
650.
0666
0.03
01(d
umm
y fo
r min
imum
scho
ol le
avin
g ag
e se
t abo
ve 1
7 is
left o
ut)
(0.0
139)
(0.0
149)
(0.0
122)
Min
imum
scho
ol le
avin
g ag
e is
170.
0413
0.04
63-0
.003
(0.0
090)
(0.0
154)
(0.0
148)
Obs
erva
tions
8484
82N
umbe
r of S
tate
s14
1414
Not
e: H
uber
-Whi
te ro
bust
stan
dard
err
ors a
re in
par
enth
eses
(clu
ster
ed b
y st
ate)
. Mod
el is
estim
ated
usin
g O
LS. S
tate
15-
17 y
ear o
ld p
opul
atio
ns a
re
used
as w
eight
s. Th
e tre
atm
ent e
ffect
repo
rted
abov
e is
the
inte
ract
ion
betw
een
the
treat
men
t sta
te d
umm
y an
d th
e po
st 1
997
dum
my,
whe
re th
e tre
atm
ent s
tate
dum
my
is eq
ual t
o 1
for t
reat
men
t sta
tes a
nd th
e po
st 1
997
dum
my
is eq
ual t
o 1
for t
he y
ears
199
8-20
00, o
ther
wise
bot
h du
mm
ies
are
equa
l to
0. S
tate
s with
few
er th
an tw
o ob
serv
atio
ns p
er p
erio
d fo
r any
of t
he d
ropo
ut ra
te m
easu
res b
y ra
ce a
re d
ropp
ed. T
reat
men
t sta
tes a
re
thos
e st
ates
that
wer
e re
quire
d to
elim
inat
e th
e an
d/or
scor
e op
tion.
The
se in
clude
: LA
, MS,
NE
, NM
, TX
. Con
trol s
tate
s are
thos
e th
at a
lread
y ha
d hi
gh e
noug
h st
anda
rds b
y 19
97. T
hese
inclu
de: C
A, C
O, D
E, D
C, F
L, M
D,
OK
, OR,
WI.
Con
trol s
tate
s dro
pped
due
to m
issin
g an
d ne
gativ
e dr
opou
t rat
es in
clude
: AR,
ID, K
Y, M
O, N
D, N
J, N
Y, S
D, U
T, W
A, W
V. N
o tre
atm
ent s
tate
s are
dro
pped
.
53
Tab
leG
-10:
Fix
edE
ffec
tsE
stim
ates
ofth
eE
ffec
tof
the
Ref
orm
onD
rop
out
Rat
esC
ontr
olling
for
Age
Req
uir
emen
ts(H
ispan
ics)
Ind
epen
den
t V
aria
ble
s10
th-1
2th
Gra
de
Dro
pou
t R
ate
10th
-11t
h G
rad
e D
rop
out
Rat
e12
th G
rad
e D
rop
out
Rat
e
Post
199
7 du
mm
y0.
0141
0.01
070.
0224
(0.0
067)
(0.0
068)
(0.0
079)
Scor
e op
tion
chan
ger s
tate
pos
t 199
7 (tr
eatm
ent e
ffect
)-0
.027
4-0
.013
8-0
.061
6(0
.004
0)(0
.005
0)(0
.004
6)Lo
cal u
nem
ploy
men
t rat
e0.
0122
0.02
04-0
.011
0(0
.004
5)(0
.007
1)(0
.007
0)Lo
g of
per
cap
ita in
com
e-0
.003
30.
0664
-0.1
958
(0.0
582)
(0.0
690)
(0.0
714)
Min
imum
GE
D a
ge re
quire
men
t is 1
6 0.
0265
0.03
270.
0088
(dum
my
for m
inim
um G
ED
age
requ
irem
ent s
et a
bove
18
is lef
t out
)(0
.015
0)(0
.021
8)(0
.008
9)M
inim
um G
ED
age
requ
irem
ent i
s 17
0.03
010.
0414
0.00
16(0
.011
4)(0
.017
4)(0
.010
4)M
inim
um G
ED
age
requ
irem
ent i
s 18
0.04
150.
0521
0.01
31(0
.003
5)(0
.005
1)(0
.004
7)M
inim
um sc
hool
leav
ing
age
is 16
-0.0
051
-0.0
001
-0.0
166
(dum
my
for m
inim
um sc
hool
leav
ing
age
set a
bove
17
is lef
t out
)(0
.007
6)(0
.010
5)(0
.006
4)M
inim
um sc
hool
leav
ing
age
is 17
0.09
120.
113
-0.0
194
(0.0
186)
(0.0
281)
(0.0
216)
Obs
erva
tions
8483
82N
umbe
r of S
tate
s14
1414
Not
e: H
uber
-Whi
te ro
bust
stan
dard
err
ors a
re in
par
enth
eses
(clu
ster
ed b
y st
ate)
. Mod
el is
estim
ated
usin
g O
LS. S
tate
15-
17 y
ear o
ld p
opul
atio
ns a
re
used
as w
eight
s. Th
e tre
atm
ent e
ffect
repo
rted
abov
e is
the
inte
ract
ion
betw
een
the
treat
men
t sta
te d
umm
y an
d th
e po
st 1
997
dum
my,
whe
re th
e tre
atm
ent s
tate
dum
my
is eq
ual t
o 1
for t
reat
men
t sta
tes a
nd th
e po
st 1
997
dum
my
is eq
ual t
o 1
for t
he y
ears
199
8-20
00, o
ther
wise
bot
h du
mm
ies
are
equa
l to
0. S
tate
s with
few
er th
an tw
o ob
serv
atio
ns p
er p
erio
d fo
r any
of t
he d
ropo
ut ra
te m
easu
res b
y ra
ce a
re d
ropp
ed. T
reat
men
t sta
tes a
re
thos
e st
ates
that
wer
e re
quire
d to
elim
inat
e th
e an
d/or
scor
e op
tion.
The
se in
clude
: LA
, MS,
NE
, NM
, TX
. Con
trol s
tate
s are
thos
e th
at a
lread
y ha
d hi
gh e
noug
h st
anda
rds b
y 19
97. T
hese
inclu
de: C
A, C
O, D
E, D
C, F
L, M
D,
OK
, OR,
WI.
Con
trol s
tate
s dro
pped
due
to m
issin
g an
d ne
gativ
e dr
opou
t rat
es in
clude
: AR,
ID, K
Y, M
O, N
D, N
J, N
Y, S
D, U
T, W
A, W
V. N
o tre
atm
ent s
tate
s are
dro
pped
.
54
Tab
leG
-11:
Fix
edE
ffec
tsE
stim
ates
ofth
eE
ffec
tof
the
Ref
orm
onD
rop
out
Rat
esw
ith
Min
imum
Sco
reC
han
ger
Sta
tes
asC
ontr
olG
roup
(All
Rac
es)
Ind
epen
den
t V
aria
ble
s10
th-1
2th
Gra
de
Dro
pou
t R
ate
10th
-11t
h G
rad
e D
rop
out
Rat
e12
th G
rad
e D
rop
out
Rat
e
Post
199
7 du
mm
y0.
0182
0.01
460.
0274
(0.0
071)
(0.0
064)
(0.0
161)
Scor
e op
tion
chan
ger s
tate
pos
t 199
7 (tr
eatm
ent e
ffect
)-0
.015
3-0
.002
3-0
.048
3(0
.004
4)(0
.003
5)(0
.011
4)Lo
cal u
nem
ploy
men
t rat
e0.
0069
0.00
060.
0193
(0.0
036)
(0.0
042)
(0.0
073)
Log
of p
er c
apita
inco
me
-0.0
420
-0.0
897
0.06
40(0
.035
5)(0
.054
0)(0
.058
6)M
inim
um G
ED
age
requ
irem
ent i
s 16
0.00
07-0
.001
00.
0034
(dum
my
for m
inim
um G
ED
age
requ
irem
ent s
et a
bove
18
is lef
t out
)(0
.002
7)(0
.004
2)(0
.007
6)M
inim
um G
ED
age
requ
irem
ent i
s 17
0.00
460.
0037
0.00
31(0
.007
7)(0
.008
6)(0
.023
6)M
inim
um G
ED
age
requ
irem
ent i
s 18
0.00
450.
0088
-0.0
103
(0.0
076)
(0.0
087)
(0.0
099)
Min
imum
scho
ol le
avin
g ag
e is
160.
0031
-0.0
016
0.01
7(d
umm
y fo
r min
imum
scho
ol le
avin
g ag
e se
t abo
ve 1
7 is
left o
ut)
(0.0
036)
(0.0
026)
(0.0
083)
Min
imum
scho
ol le
avin
g ag
e is
170.
0101
-0.0
002
0.03
79(0
.005
2)(0
.005
1)(0
.023
7)O
bser
vatio
ns18
618
618
4N
umbe
r of S
tate
s31
3131
Not
e: H
uber
-Whi
te ro
bust
stan
dard
err
ors a
re in
par
enth
eses
(clu
ster
ed b
y st
ate)
. Mod
el is
estim
ated
usin
g O
LS. S
tate
15-
17 y
ear o
ld p
opul
atio
ns a
re
used
as w
eight
s. Th
e tre
atm
ent e
ffect
repo
rted
abov
e is
the
inte
ract
ion
betw
een
the
treat
men
t sta
te d
umm
y an
d th
e po
st 1
997
dum
my,
whe
re th
e tre
atm
ent s
tate
dum
my
is eq
ual t
o 1
for t
reat
men
t sta
tes a
nd th
e po
st 1
997
dum
my
is eq
ual t
o 1
for t
he y
ears
199
8-20
00, o
ther
wise
bot
h du
mm
ies
are
equa
l to
0. S
tate
s with
few
er th
an tw
o ob
serv
atio
ns p
er p
erio
d ar
e dr
oppe
d. T
reat
men
t sta
tes a
re th
ose
stat
es th
at w
ere
requ
ired
to e
limin
ate
the
and/
or sc
ore
optio
n. T
hese
inclu
de: L
A, M
S, N
E, N
M, T
X. C
ontro
l sta
tes a
re th
ose
that
wer
e re
quire
d to
raise
their
min
imum
scor
e re
quire
men
t fr
om 3
5 to
40
in 1
997.
The
se in
clude
: AK
, AL,
AZ
, CT,
GA
, HI,
IA, I
L, IN
, KS,
MA
, ME
, MI,
MN
, MT,
NC,
NH
, NV
, OH
, PA
, RI,
SC, T
N, V
A,
VT,
WY
. No
stat
es h
ad b
e dr
oppe
d as
a re
sult
of m
issin
g or
neg
ativ
e dr
opou
t rat
es.
55
Tab
leG
-12:
Fix
edE
ffec
tsE
stim
ates
ofth
eE
ffec
tof
the
Ref
orm
onD
rop
out
Rat
esw
ith
Min
imum
Sco
reC
han
ger
Sta
tes
asC
ontr
olG
roup
(Whit
es)
Ind
epen
den
t V
aria
ble
s10
th-1
2th
Gra
de
Dro
pou
t R
ate
10th
-11t
h G
rad
e D
rop
out
Rat
e12
th G
rad
e D
rop
out
Rat
e
Post
199
7 du
mm
y0.
0071
0.00
150.
001
(0.0
099)
(0.0
044)
(0.0
193)
Scor
e op
tion
chan
ger s
tate
pos
t 199
7 (tr
eatm
ent e
ffect
)-0
.010
8-0
.003
9-0
.021
1(0
.004
0)(0
.003
4)(0
.011
5)Lo
cal u
nem
ploy
men
t rat
e-0
.002
3-0
.004
80.
0024
(0.0
040)
(0.0
040)
(0.0
090)
Log
of p
er c
apita
inco
me
-0.0
403
-0.0
386
0.01
32(0
.027
8)(0
.025
3)(0
.044
8)M
inim
um G
ED
age
requ
irem
ent i
s 16
-0.0
061
-0.0
009
-0.0
174
(dum
my
for m
inim
um G
ED
age
requ
irem
ent s
et a
bove
18
is lef
t out
)(0
.004
3)(0
.003
4)(0
.009
4)M
inim
um G
ED
age
requ
irem
ent i
s 17
-0.0
084
-0.0
066
-0.0
089
(0.0
080)
(0.0
049)
(0.0
228)
Min
imum
GE
D a
ge re
quire
men
t is 1
8 -0
.007
6-0
.002
4-0
.015
4(0
.005
5)(0
.004
6)(0
.011
1)M
inim
um sc
hool
leav
ing
age
is 16
0.00
450.
0037
0.00
02(d
umm
y fo
r min
imum
scho
ol le
avin
g ag
e se
t abo
ve 1
7 is
left o
ut)
(0.0
048)
(0.0
033)
(0.0
106)
Min
imum
scho
ol le
avin
g ag
e is
170.
0059
0.00
350.
0078
(0.0
069)
(0.0
037)
(0.0
229)
Obs
erva
tions
117
116
118
Num
ber o
f Sta
tes
2020
20N
ote:
Hub
er-W
hite
robu
st st
anda
rd e
rror
s are
in p
aren
thes
es (c
lust
ered
by
stat
e). M
odel
is es
timat
ed u
sing
OLS
. Sta
te 1
5-17
yea
r old
pop
ulat
ions
are
us
ed a
s weig
hts.
The
treat
men
t effe
ct re
porte
d ab
ove
is th
e in
tera
ctio
n be
twee
n th
e tre
atm
ent s
tate
dum
my
and
the
post
199
7 du
mm
y, w
here
the
treat
men
t sta
te d
umm
y is
equa
l to
1 fo
r tre
atm
ent s
tate
s and
the
post
199
7 du
mm
y is
equa
l to
1 fo
r the
yea
rs 1
998-
2000
, oth
erw
ise b
oth
dum
mies
ar
e eq
ual t
o 0.
Sta
tes w
ith fe
wer
than
two
obse
rvat
ions
per
per
iod
for a
ny o
f the
dro
pout
rate
mea
sure
s by
race
are
dro
pped
. Tre
atm
ent s
tate
s are
th
ose
stat
es th
at w
ere
requ
ired
to e
limin
ate
the
and/
or sc
ore
optio
n. T
hese
inclu
de: L
A, M
S, N
E, N
M, T
X. C
ontro
l sta
tes a
re th
ose
that
wer
e re
quire
d to
raise
their
min
imum
scor
e re
quire
men
t fro
m 3
5 to
40
in 1
997.
The
se in
clude
: AK
, AL,
AZ
, CT,
GA
, HI,
IA, I
L, IN
, KS,
MA
, ME
, MI,
MN
, MT,
NC,
NH
, NV
, OH
, PA
, RI,
SC, T
N, V
A, V
T, W
Y. C
ontro
l sta
tes d
ropp
ed d
ue to
miss
ing
and
nega
tive
drop
out r
ates
inclu
de:A
L, A
Z, I
A,
ME
, MN
, MT,
NH
, SC,
TN
, VT,
WY
. No
treat
men
t sta
tes w
ere
drop
ped.
56
Tab
leG
-13:
Fix
edE
ffec
tsE
stim
ates
ofth
eE
ffec
tof
the
Ref
orm
onD
rop
out
Rat
esw
ith
Min
imum
Sco
reC
han
ger
Sta
tes
asC
ontr
olG
roup
(Bla
cks)
Ind
epen
den
t V
aria
ble
s10
th-1
2th
Gra
de
Dro
pou
t R
ate
10th
-11t
h G
rad
e D
rop
out
Rat
e12
th G
rad
e D
rop
out
Rat
e
Post
199
7 du
mm
y0.
0274
0.00
630.
0414
(0.0
193)
(0.0
098)
(0.0
587)
Scor
e op
tion
chan
ger s
tate
pos
t 199
7 (tr
eatm
ent e
ffect
)-0
.019
0.00
25-0
.061
(0.0
101)
(0.0
143)
(0.0
472)
Loca
l une
mpl
oym
ent r
ate
-0.0
047
-0.0
190
0.03
39(0
.007
8)(0
.017
0)(0
.031
7)Lo
g of
per
cap
ita in
com
e-0
.171
9-0
.234
90.
1443
(0.1
152)
(0.1
908)
(0.2
572)
Min
imum
GE
D a
ge re
quire
men
t is 1
6 -0
.027
10.
0037
-0.1
088
(dum
my
for m
inim
um G
ED
age
requ
irem
ent s
et a
bove
18
is lef
t out
)(0
.009
4)(0
.016
3)(0
.048
5)M
inim
um G
ED
age
requ
irem
ent i
s 17
-0.0
415
-0.0
04-0
.138
2(0
.014
8)(0
.021
0)(0
.074
3)M
inim
um G
ED
age
requ
irem
ent i
s 18
-0.0
357
0.00
08-0
.123
6(0
.012
4)(0
.021
7)(0
.063
1)M
inim
um sc
hool
leav
ing
age
is 16
0.01
95-0
.006
10.
0753
(dum
my
for m
inim
um sc
hool
leav
ing
age
set a
bove
17
is lef
t out
)(0
.009
7)(0
.015
1)(0
.045
0)M
inim
um sc
hool
leav
ing
age
is 17
0.02
16-0
.001
60.
0708
(0.0
136)
(0.0
200)
(0.0
693)
Obs
erva
tions
118
117
112
Num
ber o
f Sta
tes
2020
20N
ote:
Hub
er-W
hite
robu
st st
anda
rd e
rror
s are
in p
aren
thes
es (c
lust
ered
by
stat
e). M
odel
is es
timat
ed u
sing
OLS
. Sta
te 1
5-17
yea
r old
pop
ulat
ions
are
us
ed a
s weig
hts.
The
treat
men
t effe
ct re
porte
d ab
ove
is th
e in
tera
ctio
n be
twee
n th
e tre
atm
ent s
tate
dum
my
and
the
post
199
7 du
mm
y, w
here
the
treat
men
t sta
te d
umm
y is
equa
l to
1 fo
r tre
atm
ent s
tate
s and
the
post
199
7 du
mm
y is
equa
l to
1 fo
r the
yea
rs 1
998-
2000
, oth
erw
ise b
oth
dum
mies
ar
e eq
ual t
o 0.
Sta
tes w
ith fe
wer
than
two
obse
rvat
ions
per
per
iod
for a
ny o
f the
dro
pout
rate
mea
sure
s by
race
are
dro
pped
. Tre
atm
ent s
tate
s are
th
ose
stat
es th
at w
ere
requ
ired
to e
limin
ate
the
and/
or sc
ore
optio
n. T
hese
inclu
de: L
A, M
S, N
E, N
M, T
X. C
ontro
l sta
tes a
re th
ose
that
wer
e re
quire
d to
raise
their
min
imum
scor
e re
quire
men
t fro
m 3
5 to
40
in 1
997.
The
se in
clude
: AK
, AL,
AZ
, CT,
GA
, HI,
IA, I
L, IN
, KS,
MA
, ME
, MI,
MN
, MT,
NC,
NH
, NV
, OH
, PA
, RI,
SC, T
N, V
A, V
T, W
Y. C
ontro
l sta
tes d
ropp
ed d
ue to
miss
ing
and
nega
tive
drop
out r
ates
inclu
de:A
L, A
Z, I
A,
ME
, MN
, MT,
NH
, SC,
TN
, VT,
WY
. No
treat
men
t sta
tes w
ere
drop
ped.
57
Tab
leG
-14:
Fix
edE
ffec
tsE
stim
ates
ofth
eE
ffec
tof
the
Ref
orm
onD
rop
out
Rat
esw
ith
Min
imum
Sco
reC
han
ger
Sta
tes
asC
ontr
olG
roup
(His
pan
ics)
Ind
epen
den
t V
aria
ble
s10
th-1
2th
Gra
de
Dro
pou
t R
ate
10th
-11t
h G
rad
e D
rop
out
Rat
e12
th G
rad
e D
rop
out
Rat
e
Post
199
7 du
mm
y0.
0356
0.02
820.
0426
(0.0
081)
(0.0
102)
(0.0
162)
Scor
e op
tion
chan
ger s
tate
pos
t 199
7 (tr
eatm
ent e
ffect
)-0
.026
7-0
.010
9-0
.066
6(0
.010
5)(0
.015
5)(0
.013
8)Lo
cal u
nem
ploy
men
t rat
e0.
0251
0.01
140.
0511
(0.0
106)
(0.0
148)
(0.0
178)
Log
of p
er c
apita
inco
me
-0.0
177
-0.0
866
0.14
86(0
.091
5)(0
.130
9)(0
.135
4)M
inim
um G
ED
age
requ
irem
ent i
s 16
0.02
15-0
.022
70.
1092
(dum
my
for m
inim
um G
ED
age
requ
irem
ent s
et a
bove
18
is lef
t out
)(0
.036
3)(0
.015
6)(0
.087
9)M
inim
um G
ED
age
requ
irem
ent i
s 17
0.01
29-0
.032
40.
1062
(0.0
348)
(0.0
186)
(0.0
826)
Min
imum
GE
D a
ge re
quire
men
t is 1
8 0.
0062
-0.0
227
0.06
81(0
.031
1)(0
.015
7)(0
.074
1)M
inim
um sc
hool
leav
ing
age
is 16
0.00
950.
0078
0.00
38(d
umm
y fo
r min
imum
scho
ol le
avin
g ag
e se
t abo
ve 1
7 is
left o
ut)
(0.0
095)
(0.0
156)
(0.0
128)
Min
imum
scho
ol le
avin
g ag
e is
170.
1285
0.15
02-0
.009
(0.0
162)
(0.0
221)
(0.0
183)
Obs
erva
tions
118
116
115
Num
ber o
f Sta
tes
2020
20N
ote:
Hub
er-W
hite
robu
st st
anda
rd e
rror
s are
in p
aren
thes
es (c
lust
ered
by
stat
e). M
odel
is es
timat
ed u
sing
OLS
. Sta
te 1
5-17
yea
r old
pop
ulat
ions
are
us
ed a
s weig
hts.
The
treat
men
t effe
ct re
porte
d ab
ove
is th
e in
tera
ctio
n be
twee
n th
e tre
atm
ent s
tate
dum
my
and
the
post
199
7 du
mm
y, w
here
the
treat
men
t sta
te d
umm
y is
equa
l to
1 fo
r tre
atm
ent s
tate
s and
the
post
199
7 du
mm
y is
equa
l to
1 fo
r the
yea
rs 1
998-
2000
, oth
erw
ise b
oth
dum
mies
ar
e eq
ual t
o 0.
Sta
tes w
ith fe
wer
than
two
obse
rvat
ions
per
per
iod
for a
ny o
f the
dro
pout
rate
mea
sure
s by
race
are
dro
pped
. Tre
atm
ent s
tate
s are
th
ose
stat
es th
at w
ere
requ
ired
to e
limin
ate
the
and/
or sc
ore
optio
n. T
hese
inclu
de: L
A, M
S, N
E, N
M, T
X. C
ontro
l sta
tes a
re th
ose
that
wer
e re
quire
d to
raise
their
min
imum
scor
e re
quire
men
t fro
m 3
5 to
40
in 1
997.
The
se in
clude
: AK
, AL,
AZ
, CT,
GA
, HI,
IA, I
L, IN
, KS,
MA
, ME
, MI,
MN
, MT,
NC,
NH
, NV
, OH
, PA
, RI,
SC, T
N, V
A, V
T, W
Y. C
ontro
l sta
tes d
ropp
ed d
ue to
miss
ing
and
nega
tive
drop
out r
ates
inclu
de:A
L, A
Z, I
A,
ME
, MN
, MT,
NH
, SC,
TN
, VT,
WY
. No
treat
men
t sta
tes w
ere
drop
ped.
58
Tab
leG
-15:
Fix
edE
ffec
tsE
stim
ates
ofth
eE
ffec
tof
the
Ref
orm
onD
rop
out
Rat
esR
estr
icti
ng
Sam
ple
toSou
ther
nSta
tes
(All
Rac
es)
Ind
epen
den
t V
aria
ble
s10
th-1
2th
Gra
de
Dro
pou
t R
ate
10th
-11t
h G
rad
e D
rop
out
Rat
e12
th G
rad
e D
rop
out
Rat
e
Post
199
7 du
mm
y0.
0117
0.01
120.
013
(0.0
055)
(0.0
072)
(0.0
079)
Scor
e op
tion
chan
ger s
tate
pos
t 199
7 (tr
eatm
ent e
ffect
)-0
.014
5-0
.009
5-0
.021
7(0
.002
1)(0
.002
6)(0
.010
1)Lo
cal u
nem
ploy
men
t rat
e-0
.002
3-0
.006
90.
0047
(0.0
030)
(0.0
033)
(0.0
053)
Log
of p
er c
apita
inco
me
-0.0
798
-0.0
906
-0.1
029
(0.0
359)
(0.0
308)
(0.0
601)
Min
imum
GE
D a
ge re
quire
men
t is 1
6 -0
.001
90.
0128
-0.0
198
(dum
my
for m
inim
um G
ED
age
requ
irem
ent s
et a
bove
18
is lef
t out
)(0
.001
9)(0
.001
9)(0
.006
1)M
inim
um G
ED
age
requ
irem
ent i
s 17
-0.0
054
0.01
39-0
.045
3(0
.012
4)(0
.013
9)(0
.006
9)M
inim
um G
ED
age
requ
irem
ent i
s 18
-0.0
032
0.01
23-0
.020
9(0
.001
6)(0
.002
0)(0
.004
2)M
inim
um sc
hool
leav
ing
age
is 16
0.00
860.
010.
0005
(dum
my
for m
inim
um sc
hool
leav
ing
age
set a
bove
17
is lef
t out
)(0
.008
5)(0
.010
0)(0
.004
5)M
inim
um sc
hool
leav
ing
age
is 17
0.01
70.
0205
-0.0
11(0
.021
1)(0
.024
1)(0
.006
7)O
bser
vatio
ns72
7270
Num
ber o
f Sta
tes
1212
12N
ote:
Hub
er-W
hite
robu
st st
anda
rd e
rror
s are
in p
aren
thes
es (c
lust
ered
by
stat
e). M
odel
is es
timat
ed u
sing
OLS
. Sta
te 1
5-17
yea
r old
pop
ulat
ion
are
used
as w
eight
s. St
ates
with
few
er th
an tw
o ob
serv
atio
ns p
er p
erio
d ar
e dr
oppe
d. T
he tr
eatm
ent e
ffect
repo
rted
abov
e is
the
inte
ract
ion
betw
een
the
treat
men
t sta
te d
umm
y an
d th
e po
st p
erio
d du
mm
y, w
here
the
treat
men
t sta
te d
umm
y is
equa
l to
1 fo
r tre
atm
ent s
tate
s and
the
post
perio
d du
mm
y is
equa
l to
1 fo
r the
yea
rs 1
998-
2000
, oth
erw
ise b
oth
dum
mies
are
equ
al to
0. T
reat
men
t sta
tes a
re th
ose
stat
es in
the
sout
h th
at w
ere
requ
ired
to
elim
inat
e th
e an
d/or
scor
e op
tion.
The
se in
clude
: LA
, MS,
NM
, TX
. Con
trol s
tate
s are
thos
e st
ates
in th
e so
uth
that
alre
ady
had
high
eno
ugh
stan
dard
s by
1997
. The
se in
clude
: AR,
DE
, DC,
FL,
KY
, MD
, OK
and
WV
.
59
Tab
leG
-16:
Fix
edE
ffec
tsE
stim
ates
ofth
eE
ffec
tof
the
Ref
orm
onD
rop
out
Rat
esR
estr
icti
ng
Sam
ple
toSou
ther
nSta
tes
(Whit
es)
Ind
epen
den
t V
aria
ble
s10
th-1
2th
Gra
de
Dro
pou
t R
ate
10th
-11t
h G
rad
e D
rop
out
Rat
e12
th G
rad
e D
rop
out
Rat
e
Post
199
7 du
mm
y0.
0075
0.00
80.
004
(0.0
052)
(0.0
093)
(0.0
078)
Scor
e op
tion
chan
ger s
tate
pos
t 199
7 (tr
eatm
ent e
ffect
)-0
.004
2-0
.005
30.
0007
(0.0
014)
(0.0
045)
(0.0
123)
Loca
l une
mpl
oym
ent r
ate
0.00
10-0
.001
70.
0082
(0.0
032)
(0.0
050)
(0.0
035)
Log
of p
er c
apita
inco
me
-0.0
490
-0.0
403
-0.0
606
(0.0
408)
(0.0
546)
(0.0
476)
Min
imum
GE
D a
ge re
quire
men
t is 1
6 0.
5726
…0.
6661
(dum
my
for m
inim
um G
ED
age
requ
irem
ent s
et a
bove
18
is lef
t out
)(0
.411
0)…
(0.5
179)
Min
imum
GE
D a
ge re
quire
men
t is 1
7 0.
5632
-0.0
021
0.63
02(0
.413
4)(0
.005
2)(0
.518
2)M
inim
um G
ED
age
requ
irem
ent i
s 18
0.57
23-0
.000
30.
6664
(0.4
115)
(0.0
040)
(0.5
199)
Min
imum
scho
ol le
avin
g ag
e is
160.
0129
0.01
230.
0068
(dum
my
for m
inim
um sc
hool
leav
ing
age
set a
bove
17
is lef
t out
)(0
.006
9)(0
.006
2)(0
.006
6)M
inim
um sc
hool
leav
ing
age
is 17
0.00
560.
0139
-0.0
345
(0.0
160)
(0.0
107)
(0.0
042)
Obs
erva
tions
5453
52N
umbe
r of S
tate
s9
99
Not
e: H
uber
-Whi
te ro
bust
stan
dard
err
ors a
re in
par
enth
eses
(clu
ster
ed b
y st
ate)
. Mod
el is
estim
ated
usin
g O
LS. S
tate
15-
17 y
ear o
ld p
opul
atio
n by
ra
ce a
re u
sed
as w
eight
s. S
tate
s with
few
er th
an tw
o ob
serv
atio
ns p
er p
erio
d fo
r any
of t
he d
ropo
ut ra
te m
easu
res b
y ra
ce a
re d
ropp
ed. T
he
treat
men
t effe
ct re
porte
d ab
ove
is th
e in
tera
ctio
n be
twee
n th
e tre
atm
ent s
tate
dum
my
and
the
post
per
iod
dum
my,
whe
re th
e tre
atm
ent s
tate
dum
my
is eq
ual t
o 1
for t
reat
men
t sta
tes a
nd th
e po
st p
erio
d du
mm
y is
equa
l to
1 fo
r the
yea
rs 1
998-
2000
, oth
erw
ise b
oth
dum
mies
are
equ
al to
0.
Trea
tmen
t sta
tes a
re th
ose
stat
es in
the
sout
h th
at w
ere
requ
ired
to e
limin
ate
the
and/
or sc
ore
optio
n. T
hese
inclu
de: L
A, M
S, N
M, T
X. C
ontro
l st
ates
are
thos
e st
ates
in th
e so
uth
that
alre
ady
had
high
eno
ugh
stan
dard
s by
1997
. The
se in
clude
: AR,
DE
, DC,
FL,
KY
, MD
, OK
and
WV
. Con
trol
stat
es d
ropp
ed d
ue to
miss
ings
inclu
de: A
R, K
Y a
nd W
V. N
o tre
atm
ent s
tate
s wer
e dr
oppe
d as
a re
sult
of m
issin
gs.
60
Tab
leG
-17:
Fix
edE
ffec
tsE
stim
ates
ofth
eE
ffec
tof
the
Ref
orm
onD
rop
out
Rat
esR
estr
icti
ng
Sam
ple
toSou
ther
nSta
tes
(Bla
cks)
Ind
epen
den
t V
aria
ble
s10
th-1
2th
Gra
de
Dro
pou
t R
ate
10th
-11t
h G
rad
e D
rop
out
Rat
e12
th G
rad
e D
rop
out
Rat
e
Post
199
7 du
mm
y0.
0233
0.01
810.
0347
(0.0
121)
(0.0
184)
(0.0
219)
Scor
e op
tion
chan
ger s
tate
pos
t 199
7 (tr
eatm
ent e
ffect
)-0
.021
6-0
.015
9-0
.033
(0.0
046)
(0.0
090)
(0.0
224)
Loca
l une
mpl
oym
ent r
ate
-0.0
061
-0.0
112
0.00
84(0
.008
2)(0
.009
8)(0
.010
5)Lo
g of
per
cap
ita in
com
e-0
.144
6-0
.129
4-0
.172
3(0
.069
8)(0
.054
4)(0
.169
0)M
inim
um G
ED
age
requ
irem
ent i
s 16
……
1.79
31(d
umm
y fo
r min
imum
GE
D a
ge re
quire
men
t set
abo
ve 1
8 is
left o
ut)
……
(1.7
685)
Min
imum
GE
D a
ge re
quire
men
t is 1
7 -0
.013
8-0
.011
81.
7195
(0.0
052)
(0.0
069)
(1.7
593)
Min
imum
GE
D a
ge re
quire
men
t is 1
8 -0
.006
4-0
.007
41.
7908
(0.0
031)
(0.0
055)
(1.7
627)
Min
imum
scho
ol le
avin
g ag
e is
160.
0506
0.06
550.
0323
(dum
my
for m
inim
um sc
hool
leav
ing
age
set a
bove
17
is lef
t out
)(0
.004
3)(0
.005
5)(0
.015
4)M
inim
um sc
hool
leav
ing
age
is 17
0.05
260.
0656
-0.0
177
(0.0
033)
(0.0
093)
(0.0
178)
Obs
erva
tions
5454
52N
umbe
r of S
tate
s9
99
Not
e: H
uber
-Whi
te ro
bust
stan
dard
err
ors a
re in
par
enth
eses
(clu
ster
ed b
y st
ate)
. Mod
el is
estim
ated
usin
g O
LS. S
tate
15-
17 y
ear o
ld p
opul
atio
n by
ra
ce a
re u
sed
as w
eight
s. St
ates
with
few
er th
an tw
o ob
serv
atio
ns p
er p
erio
d fo
r any
of t
he d
ropo
ut ra
te m
easu
res b
y ra
ce a
re d
ropp
ed. T
he
treat
men
t effe
ct re
porte
d ab
ove
is th
e in
tera
ctio
n be
twee
n th
e tre
atm
ent s
tate
dum
my
and
the
post
per
iod
dum
my,
whe
re th
e tre
atm
ent s
tate
dum
my
is eq
ual t
o 1
for t
reat
men
t sta
tes a
nd th
e po
st p
erio
d du
mm
y is
equa
l to
1 fo
r the
yea
rs 1
998-
2000
, oth
erw
ise b
oth
dum
mies
are
equ
al to
0.
Trea
tmen
t sta
tes a
re th
ose
stat
es in
the
sout
h th
at w
ere
requ
ired
to e
limin
ate
the
and/
or sc
ore
optio
n. T
hese
inclu
de: L
A, M
S, N
M, T
X. C
ontro
l st
ates
are
thos
e st
ates
in th
e so
uth
that
alre
ady
had
high
eno
ugh
stan
dard
s by
1997
. The
se in
clude
: AR,
DE
, DC,
FL,
KY
, MD
, OK
and
WV
. Con
trol
stat
es d
ropp
ed d
ue to
miss
ings
inclu
de: A
R, K
Y a
nd W
V. N
o tre
atm
ent s
tate
s wer
e dr
oppe
d as
a re
sult
of m
issin
gs.
61
Tab
leG
-18:
Fix
edE
ffec
tsE
stim
ates
ofth
eE
ffec
tof
the
Ref
orm
onD
rop
out
Rat
esR
estr
icti
ng
Sam
ple
toSou
ther
nSta
tes
(His
pan
ics)
Ind
epen
den
t V
aria
ble
s10
th-1
2th
Gra
de
Dro
pou
t R
ate
10th
-11t
h G
rad
e D
rop
out
Rat
e12
th G
rad
e D
rop
out
Rat
e
Post
199
7 du
mm
y0.
0221
0.01
750.
0321
(0.0
176)
(0.0
222)
(0.0
080)
Scor
e op
tion
chan
ger s
tate
pos
t 199
7 (tr
eatm
ent e
ffect
)-0
.030
8-0
.018
9-0
.058
2(0
.010
5)(0
.014
2)(0
.009
6)Lo
cal u
nem
ploy
men
t rat
e0.
0069
0.00
580.
0108
(0.0
189)
(0.0
279)
(0.0
121)
Log
of p
er c
apita
inco
me
-0.0
579
-0.0
358
-0.1
109
(0.1
369)
(0.1
830)
(0.1
271)
Min
imum
GE
D a
ge re
quire
men
t is 1
6 …
…1.
1960
(dum
my
for m
inim
um G
ED
age
requ
irem
ent s
et a
bove
18
is lef
t out
)…
…(1
.327
7)M
inim
um G
ED
age
requ
irem
ent i
s 17
-0.0
015
0.00
001.
197
(0.0
068)
(0.0
090)
(1.3
271)
Min
imum
GE
D a
ge re
quire
men
t is 1
8 0.
0103
0.01
751.
1906
(0.0
189)
(0.0
278)
(1.3
340)
Min
imum
scho
ol le
avin
g ag
e is
160.
0003
0.00
52-0
.012
1(d
umm
y fo
r min
imum
scho
ol le
avin
g ag
e se
t abo
ve 1
7 is
left o
ut)
(0.0
100)
(0.0
122)
(0.0
060)
Min
imum
scho
ol le
avin
g ag
e is
170.
0961
0.11
97-0
.019
6(0
.018
4)(0
.024
4)(0
.012
5)O
bser
vatio
ns54
5352
Num
ber o
f Sta
tes
99
9N
ote:
Hub
er-W
hite
robu
st st
anda
rd e
rror
s are
in p
aren
thes
es (c
lust
ered
by
stat
e). M
odel
is es
timat
ed u
sing
OLS
. Sta
te 1
5-17
yea
r old
pop
ulat
ion
by
race
are
use
d as
weig
hts.
Stat
es w
ith fe
wer
than
two
obse
rvat
ions
per
per
iod
for a
ny o
f the
dro
pout
rate
mea
sure
s by
race
are
dro
pped
. The
tre
atm
ent e
ffect
repo
rted
abov
e is
the
inte
ract
ion
betw
een
the
treat
men
t sta
te d
umm
y an
d th
e po
st p
erio
d du
mm
y, w
here
the
treat
men
t sta
te d
umm
y is
equa
l to
1 fo
r tre
atm
ent s
tate
s and
the
post
per
iod
dum
my
is eq
ual t
o 1
for t
he y
ears
199
8-20
00, o
ther
wise
bot
h du
mm
ies a
re e
qual
to 0
. Tr
eatm
ent s
tate
s are
thos
e st
ates
in th
e so
uth
that
wer
e re
quire
d to
elim
inat
e th
e an
d/or
scor
e op
tion.
The
se in
clude
: LA
, MS,
NM
, TX
. Con
trol
stat
es a
re th
ose
stat
es in
the
sout
h th
at a
lread
y ha
d hi
gh e
noug
h st
anda
rds b
y 19
97. T
hese
inclu
de: A
R, D
E, D
C, F
L, K
Y, M
D, O
K an
d W
V. C
ontro
l st
ates
dro
pped
due
to m
issin
gs in
clude
: AR,
KY
and
WV
. No
treat
men
t sta
tes w
ere
drop
ped
as a
resu
lt of
miss
ings
.
62
Tab
leG
-19:
Fix
edE
ffec
tsE
stim
ates
ofth
eE
ffec
tof
the
Ref
orm
onD
rop
out
Rat
esE
xcl
udin
gSta
tes
that
Chan
ged
Min
imum
Age
Req
uir
edto
Dro
pO
ut
(All
Rac
es)
Ind
epen
den
t V
aria
ble
s10
th-1
2th
Gra
de
Dro
pou
t R
ate
10th
-11t
h G
rad
e D
rop
out
Rat
e12
th G
rad
e D
rop
out
Rat
e
Post
199
7 du
mm
y0.
0094
0.01
260.
0023
(0.0
040)
(0.0
050)
(0.0
066)
Scor
e op
tion
chan
ger s
tate
pos
t 199
7 (tr
eatm
ent e
ffect
)-0
.013
-0.0
056
-0.0
293
(0.0
037)
(0.0
054)
(0.0
065)
Loca
l une
mpl
oym
ent r
ate
0.00
420.
0089
-0.0
086
(0.0
032)
(0.0
062)
(0.0
058)
Log
of p
er c
apita
inco
me
-0.0
288
-0.0
043
-0.1
038
(0.0
298)
(0.0
413)
(0.0
409)
Min
imum
GE
D a
ge re
quire
men
t is 1
6 0.
0048
0.01
49-0
.009
8(d
umm
y fo
r min
imum
GE
D a
ge re
quire
men
t set
abo
ve 1
8 is
left o
ut)
(0.0
044)
(0.0
041)
(0.0
058)
Min
imum
GE
D a
ge re
quire
men
t is 1
7 0.
0073
0.01
42-0
.002
2(0
.005
0)(0
.005
2)(0
.008
9)M
inim
um G
ED
age
requ
irem
ent i
s 18
0.00
340.
0112
-0.0
084
(0.0
038)
(0.0
023)
(0.0
045)
Obs
erva
tions
126
126
125
Num
ber o
f Sta
tes
2121
21N
ote:
Hub
er-W
hite
robu
st st
anda
rd e
rror
s are
in p
aren
thes
es (c
lust
ered
by
stat
e). M
odel
is es
timat
ed u
sing
OLS
. Sta
te 1
5-17
yea
r old
pop
ulat
ion
are
used
as w
eight
s. St
ates
with
few
er th
an tw
o ob
serv
atio
ns p
er p
erio
d ar
e dr
oppe
d. T
he tr
eatm
ent e
ffect
repo
rted
abov
e is
the
inte
ract
ion
betw
een
the
treat
men
t sta
te d
umm
y an
d th
e po
st p
erio
d du
mm
y, w
here
the
treat
men
t sta
te d
umm
y is
equa
l to
1 fo
r tre
atm
ent s
tate
s and
the
post
perio
d du
mm
y is
equa
l to
1 fo
r the
yea
rs 1
998-
2000
, oth
erw
ise b
oth
dum
mies
are
equ
al to
0. T
reat
men
t sta
tes a
re th
ose
stat
es th
at w
ere
requ
ired
to e
limin
ate
the
and/
or sc
ore
optio
n. T
hese
inclu
de: L
A, M
S, N
E, N
M, T
X. T
reat
men
t sta
tes d
ropp
ed in
this
regr
essio
n ar
e M
S an
d N
M. C
ontro
l sta
tes a
re th
ose
that
alre
ady
had
high
eno
ugh
stan
dard
s by
1997
. The
se in
clude
: AR,
CA
, CO
, DE
, DC,
FL,
ID, K
Y, M
D, M
O, N
J, N
Y, N
D, O
K, O
R, S
D, U
T,
WA
, WV
, WI.
Was
hing
ton
D.C
. is t
he o
nly
one
from
the
cont
rol g
roup
dro
pped
due
to c
hang
es in
min
imum
age
requ
ired
to d
rop
out.
Cont
rol
stat
es d
ropp
ed d
ue to
miss
ing
and
nega
tive
drop
out r
ates
inclu
de: N
J. N
o tre
atm
ent s
tate
s are
dro
pped
.
63
Tab
leG
-20:
Fix
edE
ffec
tsE
stim
ates
ofth
eE
ffec
tof
the
Ref
orm
onD
rop
out
Rat
esE
xcl
udin
gSta
tes
that
Chan
ged
Min
imum
Age
Req
uir
edto
Dro
pO
ut
(Whit
es)
Ind
epen
den
t V
aria
ble
s10
th-1
2th
Gra
de
Dro
pou
t R
ate
10th
-11t
h G
rad
e D
rop
out
Rat
e12
th G
rad
e D
rop
out
Rat
e
Post
199
7 du
mm
y0.
0079
0.00
930.
0039
(0.0
033)
(0.0
057)
(0.0
068)
Scor
e op
tion
chan
ger s
tate
pos
t 199
7 (tr
eatm
ent e
ffect
)-0
.004
20.
0001
-0.0
128
(0.0
016)
(0.0
047)
(0.0
091)
Loca
l une
mpl
oym
ent r
ate
0.00
320.
0058
-0.0
024
(0.0
018)
(0.0
046)
(0.0
056)
Log
of p
er c
apita
inco
me
-0.0
374
-0.0
217
-0.0
708
(0.0
210)
(0.0
330)
(0.0
301)
Min
imum
GE
D a
ge re
quire
men
t is 1
6 0.
0002
0.00
47-0
.008
8(d
umm
y fo
r min
imum
GE
D a
ge re
quire
men
t set
abo
ve 1
8 is
left o
ut)
(0.0
031)
(0.0
058)
(0.0
045)
Min
imum
GE
D a
ge re
quire
men
t is 1
7 0.
0041
0.00
75-0
.002
5(0
.003
5)(0
.007
7)(0
.009
0)M
inim
um G
ED
age
requ
irem
ent i
s 18
0.00
20.
0047
-0.0
039
(0.0
007)
(0.0
011)
(0.0
016)
Obs
erva
tions
6666
66N
umbe
r of S
tate
s11
1111
No
treat
men
t sta
tes a
re d
ropp
ed.
Not
e: H
uber
-Whi
te ro
bust
stan
dard
err
ors a
re in
par
enth
eses
(clu
ster
ed b
y st
ate)
. Mod
el is
estim
ated
usin
g O
LS. S
tate
15-
17 y
ear o
ld p
opul
atio
n ar
e us
ed a
s weig
hts.
Sta
tes w
ith fe
wer
than
two
obse
rvat
ions
per
per
iod
for a
ny o
f the
dro
pout
rate
mea
sure
s by
race
are
dro
pped
. The
trea
tmen
t effe
ct
repo
rted
abov
e is
the
inte
ract
ion
betw
een
the
treat
men
t sta
te d
umm
y an
d th
e po
st p
erio
d du
mm
y, w
here
the
treat
men
t sta
te d
umm
y is
equa
l to
1 fo
r tre
atm
ent s
tate
s and
the
post
per
iod
dum
my
is eq
ual t
o 1
for t
he y
ears
199
8-20
00, o
ther
wise
bot
h du
mm
ies a
re e
qual
to 0
. Tre
atm
ent s
tate
s are
thos
e st
ates
that
wer
e re
quire
d to
elim
inat
e th
e an
d/or
scor
e op
tion.
The
se in
clude
: LA
, MS,
NE
, NM
, TX
. Tre
atm
ent s
tate
s dro
pped
in th
is re
gres
sion
due
to c
hang
es in
min
imum
age
requ
ired
to d
rop
out o
f sch
ool a
re M
S an
d N
M. C
ontro
l sta
tes a
re th
ose
that
alre
ady
had
high
eno
ugh
stan
dard
s by
1997
. The
se in
clude
: AR,
CA
, CO
, DE
, DC,
FL,
ID, K
Y, M
D, M
O, N
J, N
Y, N
D, O
K, O
R, S
D, U
T, W
A, W
V, W
I. W
ashi
ngto
n D
.C. i
s the
onl
y on
e fr
om th
e co
ntro
l gro
up d
ropp
ed d
ue to
cha
nges
in m
inim
um a
ge re
quire
d to
dro
p ou
t. Co
ntro
l sta
tes d
ropp
ed d
ue to
miss
ing
and
nega
tive
drop
out r
ates
inclu
de: A
R, ID
, KY
, MO
, ND
, NJ,
NY
, SD
, UT,
WA
, WV
.
64
Tab
leG
-21:
Fix
edE
ffec
tsE
stim
ates
ofth
eE
ffec
tof
the
Ref
orm
onD
rop
out
Rat
esE
xcl
udin
gSta
tes
that
Chan
ged
Min
imum
Age
Req
uir
edto
Dro
pO
ut
(Bla
cks)
Ind
epen
den
t V
aria
ble
s10
th-1
2th
Gra
de
Dro
pou
t R
ate
10th
-11t
h G
rad
e D
rop
out
Rat
e12
th G
rad
e D
rop
out
Rat
e
Post
199
7 du
mm
y0.
0168
0.01
770.
012
(0.0
098)
(0.0
145)
(0.0
264)
Scor
e op
tion
chan
ger s
tate
pos
t 199
7 (tr
eatm
ent e
ffect
)-0
.012
50.
0013
-0.0
476
(0.0
097)
(0.0
162)
(0.0
173)
Loca
l une
mpl
oym
ent r
ate
-0.0
057
-0.0
014
-0.0
173
(0.0
065)
(0.0
137)
(0.0
181)
Log
of p
er c
apita
inco
me
-0.1
582
-0.1
385
-0.1
952
(0.0
692)
(0.0
823)
(0.1
532)
Min
imum
GE
D a
ge re
quire
men
t is 1
6 0.
0693
0.08
820.
0153
(dum
my
for m
inim
um G
ED
age
requ
irem
ent s
et a
bove
18
is lef
t out
)(0
.016
7)(0
.024
0)(0
.021
1)M
inim
um G
ED
age
requ
irem
ent i
s 17
0.00
860.
0304
-0.0
44(0
.014
9)(0
.025
7)(0
.041
3)M
inim
um G
ED
age
requ
irem
ent i
s 18
0.04
990.
0537
0.03
82(0
.003
1)(0
.005
0)(0
.010
0)O
bser
vatio
ns66
6666
Num
ber o
f Sta
tes
1111
11
No
treat
men
t sta
tes a
re d
ropp
ed.
Not
e: H
uber
-Whi
te ro
bust
stan
dard
err
ors a
re in
par
enth
eses
(clu
ster
ed b
y st
ate)
. Mod
el is
estim
ated
usin
g O
LS. S
tate
15-
17 y
ear o
ld p
opul
atio
n ar
e us
ed a
s weig
hts.
Sta
tes w
ith fe
wer
than
two
obse
rvat
ions
per
per
iod
for a
ny o
f the
dro
pout
rate
mea
sure
s by
race
are
dro
pped
. The
trea
tmen
t effe
ct
repo
rted
abov
e is
the
inte
ract
ion
betw
een
the
treat
men
t sta
te d
umm
y an
d th
e po
st p
erio
d du
mm
y, w
here
the
treat
men
t sta
te d
umm
y is
equa
l to
1 fo
r tre
atm
ent s
tate
s and
the
post
per
iod
dum
my
is eq
ual t
o 1
for t
he y
ears
199
8-20
00, o
ther
wise
bot
h du
mm
ies a
re e
qual
to 0
. Tre
atm
ent s
tate
s are
th
ose
stat
es th
at w
ere
requ
ired
to e
limin
ate
the
and/
or sc
ore
optio
n. T
hese
inclu
de: L
A, M
S, N
E, N
M, T
X. T
reat
men
t sta
tes d
ropp
ed in
this
regr
essio
n du
e to
cha
nges
in m
inim
um a
ge re
quire
d to
dro
p ou
t of s
choo
l are
MS
and
NM
. Con
trol s
tate
s are
thos
e th
at a
lread
y ha
d hi
gh e
noug
h st
anda
rds b
y 19
97. T
hese
inclu
de: A
R, C
A, C
O, D
E, D
C, F
L, ID
, KY
, MD
, MO
, NJ,
NY
, ND
, OK
, OR,
SD
, UT,
WA
, WV
, WI.
Was
hing
ton
D.C
. is
the
only
one
from
the
cont
rol g
roup
dro
pped
due
to c
hang
es in
min
imum
age
requ
ired
to d
rop
out.
Cont
rol s
tate
s dro
pped
due
to m
issin
g an
d ne
gativ
e dr
opou
t rat
es in
clude
: AR,
ID, K
Y, M
O, N
D, N
J, N
Y, S
D, U
T, W
A, W
V.
65
Tab
leG
-22:
Fix
edE
ffec
tsE
stim
ates
ofth
eE
ffec
tof
the
Ref
orm
onD
rop
out
Rat
esE
xcl
udin
gSta
tes
that
Chan
ged
Min
imum
Age
Req
uir
edto
Dro
pO
ut
(His
pan
ics)
Ind
epen
den
t V
aria
ble
s10
th-1
2th
Gra
de
Dro
pou
t R
ate
10th
-11t
h G
rad
e D
rop
out
Rat
e12
th G
rad
e D
rop
out
Rat
e
Post
199
7 du
mm
y0.
0142
0.01
070.
0225
(0.0
064)
(0.0
065)
(0.0
076)
Scor
e op
tion
chan
ger s
tate
pos
t 199
7 (tr
eatm
ent e
ffect
)-0
.025
8-0
.012
5-0
.059
8(0
.004
0)(0
.005
2)(0
.004
8)Lo
cal u
nem
ploy
men
t rat
e0.
0105
0.01
89-0
.012
9(0
.004
6)(0
.007
6)(0
.007
2)Lo
g of
per
cap
ita in
com
e-0
.021
30.
0507
-0.2
160
(0.0
596)
(0.0
739)
(0.0
728)
Min
imum
GE
D a
ge re
quire
men
t is 1
6 0.
0226
0.02
620.
0057
(dum
my
for m
inim
um G
ED
age
requ
irem
ent s
et a
bove
18
is lef
t out
)(0
.013
2)(0
.019
2)(0
.013
1)M
inim
um G
ED
age
requ
irem
ent i
s 17
0.03
440.
0407
0.00
79(0
.011
9)(0
.018
3)(0
.016
3)M
inim
um G
ED
age
requ
irem
ent i
s 18
0.04
010.
0509
0.01
15(0
.003
5)(0
.005
4)(0
.005
0)O
bser
vatio
ns66
6664
Num
ber o
f Sta
tes
1111
11
No
treat
men
t sta
tes a
re d
ropp
ed.
Not
e: H
uber
-Whi
te ro
bust
stan
dard
err
ors a
re in
par
enth
eses
(clu
ster
ed b
y st
ate)
. Mod
el is
estim
ated
usin
g O
LS. S
tate
15-
17 y
ear o
ld p
opul
atio
n ar
e us
ed a
s weig
hts.
Sta
tes w
ith fe
wer
than
two
obse
rvat
ions
per
per
iod
for a
ny o
f the
dro
pout
rate
mea
sure
s by
race
are
dro
pped
. The
trea
tmen
t effe
ct
repo
rted
abov
e is
the
inte
ract
ion
betw
een
the
treat
men
t sta
te d
umm
y an
d th
e po
st p
erio
d du
mm
y, w
here
the
treat
men
t sta
te d
umm
y is
equa
l to
1 fo
r tre
atm
ent s
tate
s and
the
post
per
iod
dum
my
is eq
ual t
o 1
for t
he y
ears
199
8-20
00, o
ther
wise
bot
h du
mm
ies a
re e
qual
to 0
. Tre
atm
ent s
tate
s are
thos
e st
ates
that
wer
e re
quire
d to
elim
inat
e th
e an
d/or
scor
e op
tion.
The
se in
clude
: LA
, MS,
NE
, NM
, TX
. Tre
atm
ent s
tate
s dro
pped
in th
is re
gres
sion
due
to c
hang
es in
min
imum
age
requ
ired
to d
rop
out o
f sch
ool a
re M
S an
d N
M. C
ontro
l sta
tes a
re th
ose
that
alre
ady
had
high
eno
ugh
stan
dard
s by
1997
. The
se in
clude
: AR,
CA
, CO
, DE
, DC,
FL,
ID, K
Y, M
D, M
O, N
J, N
Y, N
D, O
K, O
R, S
D, U
T, W
A, W
V, W
I. W
ashi
ngto
n D
.C. i
s the
onl
y on
e fr
om th
e co
ntro
l gro
up d
ropp
ed d
ue to
cha
nges
in m
inim
um a
ge re
quire
d to
dro
p ou
t. Co
ntro
l sta
tes d
ropp
ed d
ue to
miss
ing
and
nega
tive
drop
out r
ates
inclu
de: A
R, ID
, KY
, MO
, ND
, NJ,
NY
, SD
, UT,
WA
, WV
.
66
Tab
leG
-23:
Fix
edE
ffec
tsE
stim
ates
ofth
eE
ffec
tof
the
Ref
orm
onD
rop
out
Rat
eE
xcl
udin
gSta
tes
that
Chan
ged
the
Min
imum
Age
Req
uir
edto
eith
erD
rop
Out
orT
ake
the
GE
D(A
llR
aces
)
Ind
epen
den
t V
aria
ble
s10
th-1
2th
Gra
de
Dro
pou
t R
ate
10th
-11t
h G
rad
e D
rop
out
Rat
e12
th G
rad
e D
rop
out
Rat
e
Post
199
7 du
mm
y0.
0125
0.01
610.
003
(0.0
048)
(0.0
059)
(0.0
083)
Scor
e op
tion
chan
ger s
tate
pos
t 199
7 (tr
eatm
ent e
ffect
)-0
.015
1-0
.007
8-0
.032
1(0
.004
5)(0
.006
8)(0
.006
3)Lo
cal u
nem
ploy
men
t rat
e0.
0072
0.01
42-0
.010
1(0
.003
9)(0
.007
7)(0
.007
0)Lo
g of
per
cap
ita in
com
e-0
.011
90.
0316
-0.1
167
(0.0
422)
(0.0
601)
(0.0
543)
Obs
erva
tions
7272
72N
umbe
r of S
tate
s12
1212
Not
e: H
uber
-Whi
te ro
bust
stan
dard
err
ors a
re in
par
enth
eses
(clu
ster
ed b
y st
ate)
. Mod
el is
estim
ated
usin
g O
LS. S
tate
15-
17 y
ear o
ld p
opul
atio
n ar
e us
ed a
s weig
hts.
Stat
es w
ith fe
wer
than
two
obse
rvat
ions
per
per
iod
are
drop
ped.
The
trea
tmen
t effe
ct re
porte
d ab
ove
is th
e in
tera
ctio
n be
twee
n th
e tre
atm
ent s
tate
dum
my
and
the
post
per
iod
dum
my,
whe
re th
e tre
atm
ent s
tate
dum
my
is eq
ual t
o 1
for t
reat
men
t sta
tes a
nd th
e po
st pe
riod
dum
my
is eq
ual t
o 1
for t
he y
ears
199
8-20
00, o
ther
wise
bot
h du
mm
ies a
re e
qual
to 0
. Tre
atm
ent s
tate
s are
thos
e st
ates
that
wer
e re
quire
d to
elim
inat
e th
e an
d/or
scor
e op
tion
and
did
not c
hang
e ag
e re
quire
men
ts d
urin
g pe
riod
unde
r stu
dy.
Thes
e in
clude
: LA
, TX
. Con
trol s
tate
s are
thos
e th
at a
lread
y ha
d hi
gh e
noug
h st
anda
rds b
y 19
97 a
nd d
id n
ot c
hang
e ag
e re
quire
men
ts d
urin
g pe
riod
unde
r stu
dy. T
hese
inclu
de: C
A, C
O, D
E, F
L, ID
, MD
, NJ,
NY
, ND
, WA
, WV
. Con
trol s
tate
s dro
pped
due
to m
issin
g an
d ne
gativ
e dr
opou
t rat
es in
clude
: NJ.
No
treat
men
t sta
tes a
re d
ropp
ed.
67
Tab
leG
-24:
Fix
edE
ffec
tsE
stim
ates
ofth
eE
ffec
tof
the
Ref
orm
onD
rop
out
Rat
eE
xcl
udin
gSta
tes
that
Chan
ged
the
Min
imum
Age
Req
uir
edto
eith
erD
rop
Out
orT
ake
the
GE
D(W
hit
es)
Ind
epen
den
t V
aria
ble
s10
th-1
2th
Gra
de
Dro
pou
t R
ate
10th
-11t
h G
rad
e D
rop
out
Rat
e12
th G
rad
e D
rop
out
Rat
e
Post
199
7 du
mm
y0.
0106
0.01
210.
0061
(0.0
032)
(0.0
069)
(0.0
103)
Scor
e op
tion
chan
ger s
tate
pos
t 199
7 (tr
eatm
ent e
ffect
)-0
.003
80.
0016
-0.0
145
(0.0
010)
(0.0
045)
(0.0
114)
Loca
l une
mpl
oym
ent r
ate
0.00
340.
0065
-0.0
035
(0.0
026)
(0.0
056)
(0.0
058)
Log
of p
er c
apita
inco
me
-0.0
481
-0.0
305
-0.0
856
(0.0
290)
(0.0
430)
(0.0
301)
Obs
erva
tions
4242
42N
umbe
r of S
tate
s12
1212
Not
e: H
uber
-Whi
te ro
bust
stan
dard
err
ors a
re in
par
enth
eses
(clu
ster
ed b
y st
ate)
. Mod
el is
estim
ated
usin
g O
LS. S
tate
15-
17 y
ear o
ld p
opul
atio
n ar
e us
ed a
s weig
hts.
Stat
es w
ith fe
wer
than
two
obse
rvat
ions
per
per
iod
are
drop
ped.
The
trea
tmen
t effe
ct re
porte
d ab
ove
is th
e in
tera
ctio
n be
twee
n th
e tre
atm
ent s
tate
dum
my
and
the
post
per
iod
dum
my,
whe
re th
e tre
atm
ent s
tate
dum
my
is eq
ual t
o 1
for t
reat
men
t sta
tes a
nd th
e po
st pe
riod
dum
my
is eq
ual t
o 1
for t
he y
ears
199
8-20
00, o
ther
wise
bot
h du
mm
ies a
re e
qual
to 0
. Tre
atm
ent s
tate
s are
thos
e st
ates
that
wer
e re
quire
d to
elim
inat
e th
e an
d/or
scor
e op
tion
and
did
not c
hang
e ag
e re
quire
men
ts d
urin
g pe
riod
unde
r stu
dy.
Thes
e in
clude
: LA
, TX
. Con
trol s
tate
s are
thos
e th
at a
lread
y ha
d hi
gh e
noug
h st
anda
rds b
y 19
97 a
nd d
id n
ot c
hang
e ag
e re
quire
men
ts d
urin
g pe
riod
unde
r stu
dy. T
hese
inclu
de: C
A, C
O, D
E, F
L, ID
, MD
, NJ,
NY
, ND
, WA
, WV
. Con
trol s
tate
s dro
pped
due
to m
issin
g an
d ne
gativ
e dr
opou
t rat
es in
clude
: NJ,
NY
, WA
, WV
. No
treat
men
t sta
tes a
re d
ropp
ed.
68
Tab
leG
-25:
Fix
edE
ffec
tsE
stim
ates
ofth
eE
ffec
tof
the
Ref
orm
onD
rop
out
Rat
eE
xcl
udin
gSta
tes
that
Chan
ged
the
Min
imum
Age
Req
uir
edto
eith
erD
rop
Out
orT
ake
the
GE
D(B
lack
s)
Ind
epen
den
t V
aria
ble
s10
th-1
2th
Gra
de
Dro
pou
t R
ate
10th
-11t
h G
rad
e D
rop
out
Rat
e12
th G
rad
e D
rop
out
Rat
e
Post
199
7 du
mm
y0.
0178
0.01
760.
0146
(0.0
107)
(0.0
163)
(0.0
297)
Scor
e op
tion
chan
ger s
tate
pos
t 199
7 (tr
eatm
ent e
ffect
)-0
.012
80.
0016
-0.0
488
(0.0
108)
(0.0
176)
(0.0
188)
Loca
l une
mpl
oym
ent r
ate
-0.0
025
0.00
24-0
.017
1(0
.006
6)(0
.014
0)(0
.020
5)Lo
g of
per
cap
ita in
com
e-0
.131
4-0
.103
1-0
.202
3(0
.073
7)(0
.074
0)(0
.169
6)O
bser
vatio
ns42
4242
Num
ber o
f Sta
tes
1212
12N
ote:
Hub
er-W
hite
robu
st st
anda
rd e
rror
s are
in p
aren
thes
es (c
lust
ered
by
stat
e). M
odel
is es
timat
ed u
sing
OLS
. Sta
te 1
5-17
yea
r old
pop
ulat
ion
are
used
as w
eight
s. St
ates
with
few
er th
an tw
o ob
serv
atio
ns p
er p
erio
d ar
e dr
oppe
d. T
he tr
eatm
ent e
ffect
repo
rted
abov
e is
the
inte
ract
ion
betw
een
the
treat
men
t sta
te d
umm
y an
d th
e po
st p
erio
d du
mm
y, w
here
the
treat
men
t sta
te d
umm
y is
equa
l to
1 fo
r tre
atm
ent s
tate
s and
the
post
perio
d du
mm
y is
equa
l to
1 fo
r the
yea
rs 1
998-
2000
, oth
erw
ise b
oth
dum
mies
are
equ
al to
0. T
reat
men
t sta
tes a
re th
ose
stat
es th
at w
ere
requ
ired
to e
limin
ate
the
and/
or sc
ore
optio
n an
d di
d no
t cha
nge
age
requ
irem
ents
dur
ing
perio
d un
der s
tudy
. Th
ese
inclu
de: L
A, T
X. C
ontro
l sta
tes a
re th
ose
that
alre
ady
had
high
eno
ugh
stan
dard
s by
1997
and
did
not
cha
nge
age
requ
irem
ents
dur
ing
perio
d un
der s
tudy
. The
se in
clude
: CA
, CO
, DE
, FL,
ID, M
D, N
J, N
Y, N
D, W
A, W
V. C
ontro
l sta
tes d
ropp
ed d
ue to
miss
ing
and
nega
tive
drop
out r
ates
inclu
de: N
J, N
Y, W
A, W
V. N
o tre
atm
ent s
tate
s are
dro
pped
.
69
Tab
leG
-26:
Fix
edE
ffec
tsE
stim
ates
ofth
eE
ffec
tof
the
Ref
orm
onD
rop
out
Rat
eE
xcl
udin
gSta
tes
that
Chan
ged
the
Min
imum
Age
Req
uir
edto
eith
erD
rop
Out
orT
ake
the
GE
D(H
ispan
ics)
Ind
epen
den
t V
aria
ble
s10
th-1
2th
Gra
de
Dro
pou
t R
ate
10th
-11t
h G
rad
e D
rop
out
Rat
e12
th G
rad
e D
rop
out
Rat
e
Post
199
7 du
mm
y0.
0147
0.01
160.
0219
(0.0
070)
(0.0
078)
(0.0
069)
Scor
e op
tion
chan
ger s
tate
pos
t 199
7 (tr
eatm
ent e
ffect
)-0
.027
4-0
.015
3-0
.057
5(0
.004
7)(0
.006
0)(0
.005
9)Lo
cal u
nem
ploy
men
t rat
e0.
0123
0.02
21-0
.015
5(0
.005
4)(0
.008
7)(0
.008
2)Lo
g of
per
cap
ita in
com
e-0
.004
80.
0809
-0.2
417
(0.0
663)
(0.0
811)
(0.0
803)
Obs
erva
tions
4242
41N
umbe
r of S
tate
s12
1212
Not
e: H
uber
-Whi
te ro
bust
stan
dard
err
ors a
re in
par
enth
eses
(clu
ster
ed b
y st
ate)
. Mod
el is
estim
ated
usin
g O
LS. S
tate
15-
17 y
ear o
ld p
opul
atio
n ar
e us
ed a
s weig
hts.
Stat
es w
ith fe
wer
than
two
obse
rvat
ions
per
per
iod
are
drop
ped.
The
trea
tmen
t effe
ct re
porte
d ab
ove
is th
e in
tera
ctio
n be
twee
n th
e tre
atm
ent s
tate
dum
my
and
the
post
per
iod
dum
my,
whe
re th
e tre
atm
ent s
tate
dum
my
is eq
ual t
o 1
for t
reat
men
t sta
tes a
nd th
e po
st pe
riod
dum
my
is eq
ual t
o 1
for t
he y
ears
199
8-20
00, o
ther
wise
bot
h du
mm
ies a
re e
qual
to 0
. Tre
atm
ent s
tate
s are
thos
e st
ates
that
wer
e re
quire
d to
elim
inat
e th
e an
d/or
scor
e op
tion
and
did
not c
hang
e ag
e re
quire
men
ts d
urin
g pe
riod
unde
r stu
dy.
Thes
e in
clude
: LA
, TX
. Con
trol s
tate
s are
thos
e th
at a
lread
y ha
d hi
gh e
noug
h st
anda
rds b
y 19
97 a
nd d
id n
ot c
hang
e ag
e re
quire
men
ts d
urin
g pe
riod
unde
r stu
dy. T
hese
inclu
de: C
A, C
O, D
E, F
L, ID
, MD
, NJ,
NY
, ND
, WA
, WV
. Con
trol s
tate
s dro
pped
due
to m
issin
g an
d ne
gativ
e dr
opou
t rat
es in
clude
: NJ,
NY
, WA
, WV
. No
treat
men
t sta
tes a
re d
ropp
ed.
70
Tab
leG
-27:
GL
SF
ixed
Eff
ects
Est
imat
esof
the
Eff
ect
ofth
eR
efor
mon
Dro
pou
tR
ates
Con
trol
ling
for
Age
Req
uir
emen
ts(A
llR
aces
)
Ind
epen
den
t V
aria
ble
s10
th-1
2th
Gra
de
Dro
pou
t R
ate
10th
-11t
h G
rad
e D
rop
out
Rat
e12
th G
rad
e D
rop
out
Rat
e
Post
199
7 du
mm
y0.
0079
0.01
110.
0027
(0.0
022)
(0.0
026)
(0.0
033)
Scor
e op
tion
chan
ger s
tate
pos
t 199
7 (tr
eatm
ent e
ffect
)-0
.012
7-0
.007
7-0
.027
(0.0
021)
(0.0
021)
(0.0
049)
Loca
l une
mpl
oym
ent r
ate
0.00
220.
0039
-0.0
034
(0.0
011)
(0.0
015)
(0.0
017)
Log
of p
er c
apita
inco
me
-0.0
370
-0.0
285
-0.0
680
(0.0
118)
(0.0
135)
(0.0
180)
Min
imum
GE
D a
ge re
quire
men
t is 1
6 0.
0072
0.01
26-0
.002
0(d
umm
y fo
r min
imum
GE
D a
ge re
quire
men
t set
abo
ve 1
8 is
left o
ut)
(0.0
030)
(0.0
041)
(0.0
035)
Min
imum
GE
D a
ge re
quire
men
t is 1
7 0.
0021
0.00
77-0
.013
4(0
.004
4)(0
.005
0)(0
.008
2)M
inim
um G
ED
age
requ
irem
ent i
s 18
0.00
590.
01-0
.004
8(0
.002
5)(0
.003
6)(0
.003
0)M
inim
um sc
hool
leav
ing
age
is 16
0.00
230.
0009
0.00
47(d
umm
y fo
r min
imum
scho
ol le
avin
g ag
e se
t abo
ve 1
7 is
left o
ut)
(0.0
056)
(0.0
041)
(0.0
106)
Min
imum
scho
ol le
avin
g ag
e is
170.
0059
0.00
290.
0097
(0.0
076)
(0.0
073)
(0.0
129)
Obs
erva
tions
144
144
142
Num
ber o
f Sta
tes
2424
24N
ote:
Hub
er-W
hite
robu
st st
anda
rd e
rror
s are
in p
aren
thes
es (c
lust
ered
by
stat
e). M
odel
is es
timat
ed u
sing
GLS
. Sta
te 1
5-17
yea
r old
pop
ulat
ions
are
us
ed a
s weig
hts.
The
treat
men
t effe
ct re
porte
d ab
ove
is th
e in
tera
ctio
n be
twee
n th
e tre
atm
ent s
tate
dum
my
and
the
post
199
7 du
mm
y, w
here
the
treat
men
t sta
te d
umm
y is
equa
l to
1 fo
r tre
atm
ent s
tate
s and
the
post
199
7 du
mm
y is
equa
l to
1 fo
r the
yea
rs 1
998-
2000
, oth
erw
ise b
oth
dum
mies
ar
e eq
ual t
o 0.
Sta
tes w
ith fe
wer
than
two
obse
rvat
ions
per
per
iod
are
drop
ped.
Tre
atm
ent s
tate
s are
thos
e st
ates
that
wer
e re
quire
d to
elim
inat
e th
e an
d/or
scor
e op
tion.
The
se in
clude
: LA
, MS,
NE
, NM
, TX
. Con
trol s
tate
s are
thos
e th
at a
lread
y ha
d hi
gh e
noug
h st
anda
rds b
y 19
97. T
hese
inclu
de:
AR,
CA
, CO
, DE
, DC,
FL,
ID, K
Y, M
D, M
O, N
Y, N
D, O
K, O
R, S
D, U
T, W
A, W
V, W
I. C
ontro
l sta
tes d
ropp
ed d
ue to
miss
ing
and
nega
tive
drop
out r
ates
inclu
de: N
J. N
o tre
atm
ent s
tate
s are
dro
pped
.
71
Tab
leG
-28:
GL
SF
ixed
Eff
ects
Est
imat
esof
the
Eff
ect
ofth
eR
efor
mon
Dro
pou
tR
ates
Con
trol
ling
for
Age
Req
uir
emen
ts(W
hit
es)
Ind
epen
den
t V
aria
ble
s10
th-1
2th
Gra
de
Dro
pou
t R
ate
10th
-11t
h G
rad
e D
rop
out
Rat
e12
th G
rad
e D
rop
out
Rat
e
Post
199
7 du
mm
y0.
007
0.00
860.
0008
(0.0
019)
(0.0
022)
(0.0
035)
Scor
e op
tion
chan
ger s
tate
pos
t 199
7 (tr
eatm
ent e
ffect
)-0
.004
20.
0005
-0.0
118
(0.0
015)
(0.0
017)
(0.0
041)
Loca
l une
mpl
oym
ent r
ate
0.00
380.
0051
-0.0
007
(0.0
010)
(0.0
016)
(0.0
018)
Log
of p
er c
apita
inco
me
-0.0
297
-0.0
201
-0.0
528
(0.0
100)
(0.0
124)
(0.0
181)
Min
imum
GE
D a
ge re
quire
men
t is 1
6 0.
0039
0.00
87-0
.004
4(d
umm
y fo
r min
imum
GE
D a
ge re
quire
men
t set
abo
ve 1
8 is
left o
ut)
(0.0
024)
(0.0
033)
(0.0
033)
Min
imum
GE
D a
ge re
quire
men
t is 1
7 0.
0054
0.00
92-0
.013
8(0
.005
1)(0
.007
1)(0
.010
8)M
inim
um G
ED
age
requ
irem
ent i
s 18
0.00
230.
0046
-0.0
036
(0.0
012)
(0.0
021)
(0.0
022)
Min
imum
scho
ol le
avin
g ag
e is
160.
0159
0.01
590.
0114
(dum
my
for m
inim
um sc
hool
leav
ing
age
set a
bove
17
is lef
t out
)(0
.008
9)(0
.012
9)(0
.013
6)M
inim
um sc
hool
leav
ing
age
is 17
0.01
950.
0194
0.00
1(0
.010
6)(0
.015
1)(0
.018
3)O
bser
vatio
ns84
8382
Num
ber o
f Sta
tes
1414
14N
ote:
Hub
er-W
hite
robu
st st
anda
rd e
rror
s are
in p
aren
thes
es (c
lust
ered
by
stat
e). M
odel
is es
timat
ed u
sing
GLS
. Sta
te 1
5-17
yea
r old
pop
ulat
ions
are
us
ed a
s weig
hts.
The
treat
men
t effe
ct re
porte
d ab
ove
is th
e in
tera
ctio
n be
twee
n th
e tre
atm
ent s
tate
dum
my
and
the
post
199
7 du
mm
y, w
here
the
treat
men
t sta
te d
umm
y is
equa
l to
1 fo
r tre
atm
ent s
tate
s and
the
post
199
7 du
mm
y is
equa
l to
1 fo
r the
yea
rs 1
998-
2000
, oth
erw
ise b
oth
dum
mies
ar
e eq
ual t
o 0.
Sta
tes w
ith fe
wer
than
two
obse
rvat
ions
per
per
iod
for a
ny o
f the
dro
pout
rate
mea
sure
s by
race
are
dro
pped
. Tre
atm
ent s
tate
s are
th
ose
stat
es th
at w
ere
requ
ired
to e
limin
ate
the
and/
or sc
ore
optio
n. T
hese
inclu
de: L
A, M
S, N
E, N
M, T
X. C
ontro
l sta
tes a
re th
ose
that
alre
ady
had
high
eno
ugh
stan
dard
s by
1997
. The
se in
clude
: AR,
CA
, CO
, DE
, DC,
FL,
ID, K
Y, M
D, M
O, N
Y, N
D, O
K, O
R, S
D, U
T, W
A, W
V, W
I. C
ontro
l st
ates
dro
pped
due
to m
issin
g an
d ne
gativ
e dr
opou
t rat
es in
clude
: AR,
ID, K
Y, M
O, N
D, N
J, N
Y, S
D, U
T, W
A, W
V. N
o tre
atm
ent s
tate
s are
dr
oppe
d.
72
Tab
leG
-29:
GL
SF
ixed
Eff
ects
Est
imat
esof
the
Eff
ect
ofth
eR
efor
mon
Dro
pou
tR
ates
Con
trol
ling
for
Age
Req
uir
emen
ts(B
lack
s)
Ind
epen
den
t V
aria
ble
s10
th-1
2th
Gra
de
Dro
pou
t R
ate
10th
-11t
h G
rad
e D
rop
out
Rat
e12
th G
rad
e D
rop
out
Rat
e
Post
199
7 du
mm
y0.
0172
0.01
40.
0136
(0.0
056)
(0.0
082)
(0.0
116)
Scor
e op
tion
chan
ger s
tate
pos
t 199
7 (tr
eatm
ent e
ffect
)-0
.016
-0.0
071
-0.0
386
(0.0
043)
(0.0
061)
(0.0
107)
Loca
l une
mpl
oym
ent r
ate
-0.0
065
-0.0
025
0.00
30(0
.003
3)(0
.005
2)(0
.005
3)Lo
g of
per
cap
ita in
com
e-0
.157
7-0
.097
6-0
.043
7(0
.031
1)(0
.044
1)(0
.054
4)M
inim
um G
ED
age
requ
irem
ent i
s 16
0.05
740.
0676
0.04
93(d
umm
y fo
r min
imum
GE
D a
ge re
quire
men
t set
abo
ve 1
8 is
left o
ut)
(0.0
143)
(0.0
217)
(0.0
122)
Min
imum
GE
D a
ge re
quire
men
t is 1
7 0.
023
0.04
06-0
.008
3(0
.022
2)(0
.035
2)(0
.019
9)M
inim
um G
ED
age
requ
irem
ent i
s 18
0.05
0.05
520.
0498
(0.0
133)
(0.0
203)
(0.0
107)
Min
imum
scho
ol le
avin
g ag
e is
160.
0395
0.06
210.
0334
(dum
my
for m
inim
um sc
hool
leav
ing
age
set a
bove
17
is lef
t out
)(0
.016
0)(0
.029
1)(0
.020
6)M
inim
um sc
hool
leav
ing
age
is 17
0.02
420.
0494
-0.0
04(0
.024
8)(0
.034
3)(0
.024
0)O
bser
vatio
ns84
8482
Num
ber o
f Sta
tes
1414
14N
ote:
Hub
er-W
hite
robu
st st
anda
rd e
rror
s are
in p
aren
thes
es (c
lust
ered
by
stat
e). M
odel
is es
timat
ed u
sing
GLS
. Sta
te 1
5-17
yea
r old
pop
ulat
ions
are
us
ed a
s weig
hts.
The
treat
men
t effe
ct re
porte
d ab
ove
is th
e in
tera
ctio
n be
twee
n th
e tre
atm
ent s
tate
dum
my
and
the
post
199
7 du
mm
y, w
here
the
treat
men
t sta
te d
umm
y is
equa
l to
1 fo
r tre
atm
ent s
tate
s and
the
post
199
7 du
mm
y is
equa
l to
1 fo
r the
yea
rs 1
998-
2000
, oth
erw
ise b
oth
dum
mies
ar
e eq
ual t
o 0.
Sta
tes w
ith fe
wer
than
two
obse
rvat
ions
per
per
iod
for a
ny o
f the
dro
pout
rate
mea
sure
s by
race
are
dro
pped
. Tre
atm
ent s
tate
s are
th
ose
stat
es th
at w
ere
requ
ired
to e
limin
ate
the
and/
or sc
ore
optio
n. T
hese
inclu
de: L
A, M
S, N
E, N
M, T
X. C
ontro
l sta
tes a
re th
ose
that
alre
ady
had
high
eno
ugh
stan
dard
s by
1997
. The
se in
clude
: AR,
CA
, CO
, DE
, DC,
FL,
ID, K
Y, M
D, M
O, N
Y, N
D, O
K, O
R, S
D, U
T, W
A, W
V, W
I. C
ontro
l st
ates
dro
pped
due
to m
issin
g an
d ne
gativ
e dr
opou
t rat
es in
clude
: AR,
ID, K
Y, M
O, N
D, N
J, N
Y, S
D, U
T, W
A, W
V. N
o tre
atm
ent s
tate
s are
dr
oppe
d.
73
Tab
leG
-30:
GL
SF
ixed
Eff
ects
Est
imat
esof
the
Eff
ect
ofth
eR
efor
mon
Dro
pou
tR
ates
Con
trol
ling
for
Age
Req
uir
emen
ts(H
ispan
ics)
Ind
epen
den
t V
aria
ble
s10
th-1
2th
Gra
de
Dro
pou
t R
ate
10th
-11t
h G
rad
e D
rop
out
Rat
e12
th G
rad
e D
rop
out
Rat
e
Post
199
7 du
mm
y0.
013
0.01
480.
0258
(0.0
048)
(0.0
063)
(0.0
086)
Scor
e op
tion
chan
ger s
tate
pos
t 199
7 (tr
eatm
ent e
ffect
)-0
.026
5-0
.015
5-0
.064
4(0
.004
6)(0
.005
5)(0
.006
5)Lo
cal u
nem
ploy
men
t rat
e0.
0118
0.02
31-0
.004
6(0
.002
7)(0
.003
6)(0
.003
8)Lo
g of
per
cap
ita in
com
e-0
.007
50.
0709
-0.1
568
(0.0
293)
(0.0
367)
(0.0
421)
Min
imum
GE
D a
ge re
quire
men
t is 1
6 0.
0290
0.04
540.
0143
(dum
my
for m
inim
um G
ED
age
requ
irem
ent s
et a
bove
18
is lef
t out
)(0
.017
4)(0
.025
2)(0
.017
1)M
inim
um G
ED
age
requ
irem
ent i
s 17
0.03
420.
0526
0.00
86(0
.019
5)(0
.027
7)(0
.018
8)M
inim
um G
ED
age
requ
irem
ent i
s 18
0.04
070.
054
0.01
73(0
.015
5)(0
.023
0)(0
.014
7)M
inim
um sc
hool
leav
ing
age
is 16
-0.0
027
0.00
62-0
.018
1(d
umm
y fo
r min
imum
scho
ol le
avin
g ag
e se
t abo
ve 1
7 is
left o
ut)
(0.0
114)
(0.0
136)
(0.0
101)
Min
imum
scho
ol le
avin
g ag
e is
170.
0706
0.08
77-0
.018
9(0
.014
7)(0
.022
9)(0
.042
6)O
bser
vatio
ns84
8382
Num
ber o
f Sta
tes
1414
14N
ote:
Hub
er-W
hite
robu
st st
anda
rd e
rror
s are
in p
aren
thes
es (c
lust
ered
by
stat
e). M
odel
is es
timat
ed u
sing
GLS
. Sta
te 1
5-17
yea
r old
pop
ulat
ions
are
us
ed a
s weig
hts.
The
treat
men
t effe
ct re
porte
d ab
ove
is th
e in
tera
ctio
n be
twee
n th
e tre
atm
ent s
tate
dum
my
and
the
post
199
7 du
mm
y, w
here
the
treat
men
t sta
te d
umm
y is
equa
l to
1 fo
r tre
atm
ent s
tate
s and
the
post
199
7 du
mm
y is
equa
l to
1 fo
r the
yea
rs 1
998-
2000
, oth
erw
ise b
oth
dum
mies
ar
e eq
ual t
o 0.
Sta
tes w
ith fe
wer
than
two
obse
rvat
ions
per
per
iod
for a
ny o
f the
dro
pout
rate
mea
sure
s by
race
are
dro
pped
. Tre
atm
ent s
tate
s are
th
ose
stat
es th
at w
ere
requ
ired
to e
limin
ate
the
and/
or sc
ore
optio
n. T
hese
inclu
de: L
A, M
S, N
E, N
M, T
X. C
ontro
l sta
tes a
re th
ose
that
alre
ady
had
high
eno
ugh
stan
dard
s by
1997
. The
se in
clude
: AR,
CA
, CO
, DE
, DC,
FL,
ID, K
Y, M
D, M
O, N
Y, N
D, O
K, O
R, S
D, U
T, W
A, W
V, W
I. C
ontro
l st
ates
dro
pped
due
to m
issin
g an
d ne
gativ
e dr
opou
t rat
es in
clude
: AR,
ID, K
Y, M
O, N
D, N
J, N
Y, S
D, U
T, W
A, W
V. N
o tre
atm
ent s
tate
s are
dr
oppe
d.
74
Tab
leG
-31:
GL
SF
ixed
Eff
ects
Est
imat
esof
the
Eff
ect
ofth
eR
efor
mon
Dro
pou
tR
ates
Con
trol
ling
for
Age
Req
uir
emen
tsan
dU
sing
Pan
elSp
ecifi
cA
R-1
Auto
corr
elat
ion
Str
uct
ure
(All
Rac
es)
Ind
epen
den
t V
aria
ble
s10
th-1
2th
Gra
de
Dro
pou
t R
ate
10th
-11t
h G
rad
e D
rop
out
Rat
e12
th G
rad
e D
rop
out
Rat
e
Post
199
7 du
mm
y0.
0076
0.01
220.
0018
(0.0
019)
(0.0
019)
(0.0
024)
Scor
e op
tion
chan
ger s
tate
pos
t 199
7 (tr
eatm
ent e
ffect
)-0
.013
7-0
.009
-0.0
283
(0.0
017)
(0.0
019)
(0.0
038)
Loca
l une
mpl
oym
ent r
ate
0.00
270.
0049
-0.0
068
(0.0
010)
(0.0
012)
(0.0
016)
Log
of p
er c
apita
inco
me
-0.0
283
-0.0
207
-0.0
925
(0.0
106)
(0.0
106)
(0.0
135)
Min
imum
GE
D a
ge re
quire
men
t is 1
6 0.
0077
0.01
50-0
.003
2(d
umm
y fo
r min
imum
GE
D a
ge re
quire
men
t set
abo
ve 1
8 is
left o
ut)
(0.0
026)
(0.0
019)
(0.0
023)
Min
imum
GE
D a
ge re
quire
men
t is 1
7 0.
0036
0.01
29-0
.008
5(0
.004
2)(0
.003
6)(0
.007
0)M
inim
um G
ED
age
requ
irem
ent i
s 18
0.00
520.
0119
-0.0
062
(0.0
024)
(0.0
011)
(0.0
012)
Min
imum
scho
ol le
avin
g ag
e is
160.
005
0.00
320.
0067
(dum
my
for m
inim
um sc
hool
leav
ing
age
set a
bove
17
is lef
t out
)(0
.005
4)(0
.005
0)(0
.010
2)M
inim
um sc
hool
leav
ing
age
is 17
0.01
210.
0098
0.02
18(0
.007
5)(0
.008
2)(0
.012
5)O
bser
vatio
ns14
414
414
2N
umbe
r of S
tate
s24
2424
Not
e: H
uber
-Whi
te ro
bust
stan
dard
err
ors a
re in
par
enth
eses
(clu
ster
ed b
y st
ate)
. Mod
el is
estim
ated
usin
g G
LS u
sing
pane
l spe
cific
AR-
1 au
toco
rrela
tion
stru
ctur
e. St
ate
15-1
7 ye
ar o
ld p
opul
atio
ns a
re u
sed
as w
eight
s. Th
e tre
atm
ent e
ffect
repo
rted
abov
e is
the
inte
ract
ion
betw
een
the
treat
men
t sta
te d
umm
y an
d th
e po
st 1
997
dum
my,
whe
re th
e tre
atm
ent s
tate
dum
my
is eq
ual t
o 1
for t
reat
men
t sta
tes a
nd th
e po
st 1
997
dum
my
is eq
ual t
o 1
for t
he y
ears
199
8-20
00, o
ther
wise
bot
h du
mm
ies a
re e
qual
to 0
. Sta
tes w
ith fe
wer
than
two
obse
rvat
ions
per
per
iod
are
drop
ped.
Tr
eatm
ent s
tate
s are
thos
e st
ates
that
wer
e re
quire
d to
elim
inat
e th
e an
d/or
scor
e op
tion.
The
se in
clude
: LA
, MS,
NE
, NM
, TX
. Con
trol s
tate
s are
th
ose
that
alre
ady
had
high
eno
ugh
stan
dard
s by
1997
. The
se in
clude
: AR,
CA
, CO
, DE
, DC,
FL,
ID, K
Y, M
D, M
O, N
Y, N
D, O
K, O
R, S
D, U
T,
WA
, WV
, WI.
Con
trol s
tate
s dro
pped
due
to m
issin
g an
d ne
gativ
e dr
opou
t rat
es in
clude
: NJ.
No
treat
men
t sta
tes a
re d
ropp
ed.
75
Tab
leG
-32:
GL
SF
ixed
Eff
ects
Est
imat
esof
the
Eff
ect
ofth
eR
efor
mon
Dro
pou
tR
ates
Con
trol
ling
for
Age
Req
uir
emen
tsan
dU
sing
Pan
elSp
ecifi
cA
R-1
Auto
corr
elat
ion
Str
uct
ure
(Whit
es)
Ind
epen
den
t V
aria
ble
s10
th-1
2th
Gra
de
Dro
pou
t R
ate
10th
-11t
h G
rad
e D
rop
out
Rat
e12
th G
rad
e D
rop
out
Rat
e
Post
199
7 du
mm
y0.
0076
0.00
930.
0042
(0.0
017)
(0.0
019)
(0.0
035)
Scor
e op
tion
chan
ger s
tate
pos
t 199
7 (tr
eatm
ent e
ffect
)-0
.005
1-0
.000
5-0
.017
(0.0
012)
(0.0
015)
(0.0
033)
Loca
l une
mpl
oym
ent r
ate
0.00
340.
0041
-0.0
030
(0.0
010)
(0.0
015)
(0.0
016)
Log
of p
er c
apita
inco
me
-0.0
312
-0.0
251
-0.0
638
(0.0
091)
(0.0
113)
(0.0
177)
Min
imum
GE
D a
ge re
quire
men
t is 1
6 0.
0042
0.00
93-0
.005
1(d
umm
y fo
r min
imum
GE
D a
ge re
quire
men
t set
abo
ve 1
8 is
left o
ut)
(0.0
021)
(0.0
030)
(0.0
037)
Min
imum
GE
D a
ge re
quire
men
t is 1
7 0.
0049
0.00
93-0
.025
1(0
.004
4)(0
.005
5)(0
.009
5)M
inim
um G
ED
age
requ
irem
ent i
s 18
0.00
220.
0045
-0.0
033
(0.0
009)
(0.0
020)
(0.0
027)
Min
imum
scho
ol le
avin
g ag
e is
160.
0179
0.01
930.
0122
(dum
my
for m
inim
um sc
hool
leav
ing
age
set a
bove
17
is lef
t out
)(0
.009
4)(0
.011
8)(0
.007
5)M
inim
um sc
hool
leav
ing
age
is 17
0.02
110.
0228
-0.0
076
(0.0
106)
(0.0
129)
(0.0
127)
Obs
erva
tions
8483
82N
umbe
r of S
tate
s14
1414
Not
e: H
uber
-Whi
te ro
bust
stan
dard
err
ors a
re in
par
enth
eses
(clu
ster
ed b
y st
ate)
. Mod
el is
estim
ated
usin
g G
LS u
sing
pane
l spe
cific
AR-
1 au
toco
rrela
tion
stru
ctur
e. St
ate
15-1
7 ye
ar o
ld p
opul
atio
ns a
re u
sed
as w
eight
s. Th
e tre
atm
ent e
ffect
repo
rted
abov
e is
the
inte
ract
ion
betw
een
the
treat
men
t sta
te d
umm
y an
d th
e po
st 1
997
dum
my,
whe
re th
e tre
atm
ent s
tate
dum
my
is eq
ual t
o 1
for t
reat
men
t sta
tes a
nd th
e po
st 1
997
dum
my
is eq
ual t
o 1
for t
he y
ears
199
8-20
00, o
ther
wise
bot
h du
mm
ies a
re e
qual
to 0
. Sta
tes w
ith fe
wer
than
two
obse
rvat
ions
per
per
iod
for a
ny o
f the
dr
opou
t rat
e m
easu
res b
y ra
ce a
re d
ropp
ed. T
reat
men
t sta
tes a
re th
ose
stat
es th
at w
ere
requ
ired
to e
limin
ate
the
and/
or sc
ore
optio
n. T
hese
inclu
de:
LA, M
S, N
E, N
M, T
X. C
ontro
l sta
tes a
re th
ose
that
alre
ady
had
high
eno
ugh
stan
dard
s by
1997
. The
se in
clude
: AR,
CA
, CO
, DE
, DC,
FL,
ID, K
Y,
MD
, MO
, NY
, ND
, OK
, OR,
SD
, UT,
WA
, WV
, WI.
Cont
rol s
tate
s dro
pped
due
to m
issin
g an
d ne
gativ
e dr
opou
t rat
es in
clude
: AR,
ID, K
Y, M
O,
ND
, NJ,
NY
, SD
, UT,
WA
, WV
. No
treat
men
t sta
tes a
re d
ropp
ed.
76
Tab
leG
-33:
GL
SF
ixed
Eff
ects
Est
imat
esof
the
Eff
ect
ofth
eR
efor
mon
Dro
pou
tR
ates
Con
trol
ling
for
Age
Req
uir
emen
tsan
dU
sing
Pan
elSp
ecifi
cA
R-1
Auto
corr
elat
ion
Str
uct
ure
(Bla
cks)
Ind
epen
den
t V
aria
ble
s10
th-1
2th
Gra
de
Dro
pou
t R
ate
10th
-11t
h G
rad
e D
rop
out
Rat
e12
th G
rad
e D
rop
out
Rat
e
Post
199
7 du
mm
y0.
0169
0.01
140.
013
(0.0
041)
(0.0
073)
(0.0
111)
Scor
e op
tion
chan
ger s
tate
pos
t 199
7 (tr
eatm
ent e
ffect
)-0
.015
6-0
.008
9-0
.040
4(0
.003
4)(0
.005
8)(0
.009
9)Lo
cal u
nem
ploy
men
t rat
e-0
.007
5-0
.006
50.
0014
(0.0
031)
(0.0
049)
(0.0
053)
Log
of p
er c
apita
inco
me
-0.1
653
-0.1
084
-0.0
613
(0.0
278)
(0.0
415)
(0.0
540)
Min
imum
GE
D a
ge re
quire
men
t is 1
6 0.
0622
0.07
320.
0457
(dum
my
for m
inim
um G
ED
age
requ
irem
ent s
et a
bove
18
is lef
t out
)(0
.012
5)(0
.018
9)(0
.012
5)M
inim
um G
ED
age
requ
irem
ent i
s 17
0.03
090.
0401
-0.0
16(0
.020
4)(0
.031
4)(0
.018
5)M
inim
um G
ED
age
requ
irem
ent i
s 18
0.05
320.
0603
0.04
73(0
.011
3)(0
.017
7)(0
.010
6)M
inim
um sc
hool
leav
ing
age
is 16
0.04
950.
0772
0.02
92(d
umm
y fo
r min
imum
scho
ol le
avin
g ag
e se
t abo
ve 1
7 is
left o
ut)
(0.0
174)
(0.0
279)
(0.0
146)
Min
imum
scho
ol le
avin
g ag
e is
170.
0388
0.06
06-0
.008
3(0
.022
1)(0
.026
8)(0
.019
5)O
bser
vatio
ns84
8482
Num
ber o
f Sta
tes
1414
14N
ote:
Hub
er-W
hite
robu
st st
anda
rd e
rror
s are
in p
aren
thes
es (c
lust
ered
by
stat
e). M
odel
is es
timat
ed u
sing
GLS
usin
g pa
nel s
pecif
ic A
R-1
auto
corr
elatio
n st
ruct
ure.
Stat
e 15
-17
year
old
pop
ulat
ions
are
use
d as
weig
hts.
The
treat
men
t effe
ct re
porte
d ab
ove
is th
e in
tera
ctio
n be
twee
n th
e tre
atm
ent s
tate
dum
my
and
the
post
199
7 du
mm
y, w
here
the
treat
men
t sta
te d
umm
y is
equa
l to
1 fo
r tre
atm
ent s
tate
s and
the
post
199
7 du
mm
y is
equa
l to
1 fo
r the
yea
rs 1
998-
2000
, oth
erw
ise b
oth
dum
mies
are
equ
al to
0. S
tate
s with
few
er th
an tw
o ob
serv
atio
ns p
er p
erio
d fo
r any
of t
he
drop
out r
ate
mea
sure
s by
race
are
dro
pped
. Tre
atm
ent s
tate
s are
thos
e st
ates
that
wer
e re
quire
d to
elim
inat
e th
e an
d/or
scor
e op
tion.
The
se
inclu
de: L
A, M
S, N
E, N
M, T
X. C
ontro
l sta
tes a
re th
ose
that
alre
ady
had
high
eno
ugh
stan
dard
s by
1997
. The
se in
clude
: AR,
CA
, CO
, DE
, DC,
FL,
ID
, KY
, MD
, MO
, NY
, ND
, OK
, OR,
SD
, UT,
WA
, WV
, WI.
Cont
rol s
tate
s dro
pped
due
to m
issin
g an
d ne
gativ
e dr
opou
t rat
es in
clude
: AR,
ID,
KY
, MO
, ND
, NJ,
NY
, SD
, UT,
WA
, WV
. No
treat
men
t sta
tes a
re d
ropp
ed.
77
Tab
leG
-34:
GL
SF
ixed
Eff
ects
Est
imat
esof
the
Eff
ect
ofth
eR
efor
mon
Dro
pou
tR
ates
Con
trol
ling
for
Age
Req
uir
emen
tsan
dU
sing
Pan
elSp
ecifi
cA
R-1
Auto
corr
elat
ion
Str
uct
ure
(His
pan
ics)
Ind
epen
den
t V
aria
ble
s10
th-1
2th
Gra
de
Dro
pou
t R
ate
10th
-11t
h G
rad
e D
rop
out
Rat
e12
th G
rad
e D
rop
out
Rat
e
Post
199
7 du
mm
y0.
0093
0.01
0.02
44(0
.004
3)(0
.005
6)(0
.008
3)Sc
ore
optio
n ch
ange
r sta
te p
ost 1
997
(trea
tmen
t effe
ct)
-0.0
249
-0.0
115
-0.0
645
(0.0
036)
(0.0
047)
(0.0
058)
Loca
l une
mpl
oym
ent r
ate
0.01
120.
0198
-0.0
057
(0.0
022)
(0.0
030)
(0.0
035)
Log
of p
er c
apita
inco
me
-0.0
025
0.05
17-0
.153
9(0
.026
1)(0
.032
7)(0
.041
5)M
inim
um G
ED
age
requ
irem
ent i
s 16
0.02
670.
0356
0.01
27(d
umm
y fo
r min
imum
GE
D a
ge re
quire
men
t set
abo
ve 1
8 is
left o
ut)
(0.0
163)
(0.0
189)
(0.0
169)
Min
imum
GE
D a
ge re
quire
men
t is 1
7 0.
0314
0.04
790.
0047
(0.0
185)
(0.0
217)
(0.0
185)
Min
imum
GE
D a
ge re
quire
men
t is 1
8 0.
0418
0.05
770.
0173
(0.0
147)
(0.0
167)
(0.0
146)
Min
imum
scho
ol le
avin
g ag
e is
16-0
.008
6-0
.005
3-0
.019
1(d
umm
y fo
r min
imum
scho
ol le
avin
g ag
e se
t abo
ve 1
7 is
left o
ut)
(0.0
092)
(0.0
106)
(0.0
098)
Min
imum
scho
ol le
avin
g ag
e is
170.
0713
0.09
11-0
.037
4(0
.008
7)(0
.015
8)(0
.035
3)O
bser
vatio
ns84
8382
Num
ber o
f Sta
tes
1414
14N
ote:
Hub
er-W
hite
robu
st st
anda
rd e
rror
s are
in p
aren
thes
es (c
lust
ered
by
stat
e). M
odel
is es
timat
ed u
sing
GLS
usin
g pa
nel s
pecif
ic A
R-1
auto
corr
elatio
n st
ruct
ure.
Stat
e 15
-17
year
old
pop
ulat
ions
are
use
d as
weig
hts.
The
treat
men
t effe
ct re
porte
d ab
ove
is th
e in
tera
ctio
n be
twee
n th
e tre
atm
ent s
tate
dum
my
and
the
post
199
7 du
mm
y, w
here
the
treat
men
t sta
te d
umm
y is
equa
l to
1 fo
r tre
atm
ent s
tate
s and
the
post
199
7 du
mm
y is
equa
l to
1 fo
r the
yea
rs 1
998-
2000
, oth
erw
ise b
oth
dum
mies
are
equ
al to
0. S
tate
s with
few
er th
an tw
o ob
serv
atio
ns p
er p
erio
d fo
r any
of t
he
drop
out r
ate
mea
sure
s by
race
are
dro
pped
. Tre
atm
ent s
tate
s are
thos
e st
ates
that
wer
e re
quire
d to
elim
inat
e th
e an
d/or
scor
e op
tion.
The
se in
clude
: LA
, MS,
NE
, NM
, TX
. Con
trol s
tate
s are
thos
e th
at a
lread
y ha
d hi
gh e
noug
h st
anda
rds b
y 19
97. T
hese
inclu
de: A
R, C
A, C
O, D
E, D
C, F
L, ID
, KY
, M
D, M
O, N
Y, N
D, O
K, O
R, S
D, U
T, W
A, W
V, W
I. Co
ntro
l sta
tes d
ropp
ed d
ue to
miss
ing
and
nega
tive
drop
out r
ates
inclu
de: A
R, ID
, KY
, MO
, N
D, N
J, N
Y, S
D, U
T, W
A, W
V. N
o tre
atm
ent s
tate
s are
dro
pped
.
78
Tab
leG
-35:
Diff
eren
ce-i
n-D
iffer
ence
Est
imat
esof
the
Eff
ect
ofth
eR
efor
mon
Dro
pou
tR
ates
Con
trol
ling
for
Age
Req
uir
emen
tsan
dR
estr
icti
ng
Con
trol
Gro
up
toC
alif
ornia
and
Flo
rida
(Hig
hIm
mig
rant
Sta
tes)
(All
Rac
es)
Ind
epen
den
t V
aria
ble
s10
th-1
2th
Gra
de
Dro
pou
t R
ate
10th
-11t
h G
rad
e D
rop
out
Rat
e12
th G
rad
e D
rop
out
Rat
e
Scor
e op
tion
chan
ger s
tate
-0.0
16-0
.016
6-0
.013
6(0
.012
7)(0
.014
6)(0
.009
1)Sc
ore
optio
n ch
ange
r sta
te p
ost 1
997
(trea
tmen
t effe
ct)
-0.0
069
0.00
56-0
.036
5(0
.006
3)(0
.012
2)(0
.013
2)Po
st 1
997
dum
my
-0.0
086
-0.0
126
0.00
07(0
.005
5)(0
.012
1)(0
.011
8)Co
nsta
nt0.
1214
0.12
900.
1014
(0.0
111)
(0.0
132)
(0.0
050)
Obs
erva
tions
4242
42N
umbe
r of S
tate
s7
77
Not
e: H
uber
-Whi
te ro
bust
stan
dard
err
ors a
re in
par
enth
eses
(clu
ster
ed b
y st
ate)
. Mod
el is
estim
ated
usin
g O
LS. S
tate
15-
17 y
ear o
ld p
opul
atio
ns a
re
used
as w
eight
s. Th
e tre
atm
ent e
ffect
repo
rted
abov
e is
the
inte
ract
ion
betw
een
the
treat
men
t sta
te d
umm
y an
d th
e po
st 1
997
dum
my,
whe
re th
e tre
atm
ent s
tate
dum
my
is eq
ual t
o 1
for t
reat
men
t sta
tes a
nd th
e po
st 1
997
dum
my
is eq
ual t
o 1
for t
he y
ears
199
8-20
00, o
ther
wise
bot
h du
mm
ies
are
equa
l to
0. S
tate
s with
few
er th
an tw
o ob
serv
atio
ns p
er p
erio
d ar
e dr
oppe
d. T
reat
men
t sta
tes a
re th
ose
stat
es th
at w
ere
requ
ired
to e
limin
ate
the
and/
or sc
ore
optio
n. T
hese
inclu
de: L
A, M
S, N
E, N
M, T
X. C
ontro
l sta
tes a
re th
ose
that
alre
ady
had
high
eno
ugh
stan
dard
s by
1997
and
hav
e lar
ge
imm
igra
nt p
opul
atio
ns. T
hese
inclu
de: C
A a
nd F
L.
79
Tab
leG
-36:
Diff
eren
ce-i
n-D
iffer
ence
Est
imat
esof
the
Eff
ect
ofth
eR
efor
mon
Dro
pou
tR
ates
Con
trol
ling
for
Age
Req
uir
emen
tsan
dR
estr
icti
ng
Con
trol
Gro
up
toC
alif
ornia
and
Flo
rida
(Hig
hIm
mig
rant
Sta
tes)
(Whit
es)
Ind
epen
den
t V
aria
ble
s10
th-1
2th
Gra
de
Dro
pou
t R
ate
10th
-11t
h G
rad
e D
rop
out
Rat
e12
th G
rad
e D
rop
out
Rat
e
Scor
e op
tion
chan
ger s
tate
-0.0
216
-0.0
192
-0.0
26(0
.020
5)(0
.025
1)(0
.009
9)Sc
ore
optio
n ch
ange
r sta
te p
ost 1
997
(trea
tmen
t effe
ct)
-0.0
020.
0043
-0.0
146
(0.0
038)
(0.0
108)
(0.0
148)
Post
199
7 du
mm
y-0
.007
2-0
.008
0-0
.006
5(0
.003
7)(0
.010
8)(0
.014
2)Co
nsta
nt0.
1008
0.10
770.
0830
(0.0
193)
(0.0
238)
(0.0
073)
Obs
erva
tions
4242
42N
umbe
r of S
tate
s7
77
Not
e: H
uber
-Whi
te ro
bust
stan
dard
err
ors a
re in
par
enth
eses
(clu
ster
ed b
y st
ate)
. Mod
el is
estim
ated
usin
g O
LS. S
tate
15-
17 y
ear o
ld p
opul
atio
ns a
re
used
as w
eight
s. Th
e tre
atm
ent e
ffect
repo
rted
abov
e is
the
inte
ract
ion
betw
een
the
treat
men
t sta
te d
umm
y an
d th
e po
st 1
997
dum
my,
whe
re th
e tre
atm
ent s
tate
dum
my
is eq
ual t
o 1
for t
reat
men
t sta
tes a
nd th
e po
st 1
997
dum
my
is eq
ual t
o 1
for t
he y
ears
199
8-20
00, o
ther
wise
bot
h du
mm
ies
are
equa
l to
0. S
tate
s with
few
er th
an tw
o ob
serv
atio
ns p
er p
erio
d ar
e dr
oppe
d. T
reat
men
t sta
tes a
re th
ose
stat
es th
at w
ere
requ
ired
to e
limin
ate
the
and/
or sc
ore
optio
n. T
hese
inclu
de: L
A, M
S, N
E, N
M, T
X. C
ontro
l sta
tes a
re th
ose
that
alre
ady
had
high
eno
ugh
stan
dard
s by
1997
and
hav
e lar
ge
imm
igra
nt p
opul
atio
ns. T
hese
inclu
de: C
A a
nd F
L.
80
Tab
leG
-37:
Diff
eren
ce-i
n-D
iffer
ence
Est
imat
esof
the
Eff
ect
ofth
eR
efor
mon
Dro
pou
tR
ates
Con
trol
ling
for
Age
Req
uir
emen
tsan
dR
estr
icti
ng
Con
trol
Gro
up
toC
alif
ornia
and
Flo
rida
(Hig
hIm
mig
rant
Sta
tes)
(Bla
cks)
Ind
epen
den
t V
aria
ble
s10
th-1
2th
Gra
de
Dro
pou
t R
ate
10th
-11t
h G
rad
e D
rop
out
Rat
e12
th G
rad
e D
rop
out
Rat
e
Scor
e op
tion
chan
ger s
tate
-0.0
31-0
.041
1-0
.001
(0.0
104)
(0.0
071)
(0.0
223)
Scor
e op
tion
chan
ger s
tate
pos
t 199
7 (tr
eatm
ent e
ffect
)-0
.011
90.
0003
-0.0
43(0
.015
5)(0
.026
6)(0
.026
7)Po
st 1
997
dum
my
-0.0
038
-0.0
065
0.00
35(0
.013
6)(0
.026
1)(0
.022
3)Co
nsta
nt0.
1707
0.18
120.
1398
(0.0
098)
(0.0
058)
(0.0
212)
Obs
erva
tions
4242
42N
umbe
r of S
tate
s7
77
Not
e: H
uber
-Whi
te ro
bust
stan
dard
err
ors a
re in
par
enth
eses
(clu
ster
ed b
y st
ate)
. Mod
el is
estim
ated
usin
g O
LS. S
tate
15-
17 y
ear o
ld p
opul
atio
ns a
re
used
as w
eight
s. Th
e tre
atm
ent e
ffect
repo
rted
abov
e is
the
inte
ract
ion
betw
een
the
treat
men
t sta
te d
umm
y an
d th
e po
st 1
997
dum
my,
whe
re th
e tre
atm
ent s
tate
dum
my
is eq
ual t
o 1
for t
reat
men
t sta
tes a
nd th
e po
st 1
997
dum
my
is eq
ual t
o 1
for t
he y
ears
199
8-20
00, o
ther
wise
bot
h du
mm
ies
are
equa
l to
0. S
tate
s with
few
er th
an tw
o ob
serv
atio
ns p
er p
erio
d ar
e dr
oppe
d. T
reat
men
t sta
tes a
re th
ose
stat
es th
at w
ere
requ
ired
to e
limin
ate
the
and/
or sc
ore
optio
n. T
hese
inclu
de: L
A, M
S, N
E, N
M, T
X. C
ontro
l sta
tes a
re th
ose
that
alre
ady
had
high
eno
ugh
stan
dard
s by
1997
and
hav
e lar
ge
imm
igra
nt p
opul
atio
ns. T
hese
inclu
de: C
A a
nd F
L.
81
Tab
leG
-38:
Diff
eren
ce-i
n-D
iffer
ence
Est
imat
esof
the
Eff
ect
ofth
eR
efor
mon
Dro
pou
tR
ates
Con
trol
ling
for
Age
Req
uir
emen
tsan
dR
estr
icti
ng
Con
trol
Gro
up
toC
alif
ornia
and
Flo
rida
(Hig
hIm
mig
rant
Sta
tes)
(His
pan
ics)
Ind
epen
den
t V
aria
ble
s10
th-1
2th
Gra
de
Dro
pou
t R
ate
10th
-11t
h G
rad
e D
rop
out
Rat
e12
th G
rad
e D
rop
out
Rat
e
Scor
e op
tion
chan
ger s
tate
-0.0
189
-0.0
171
-0.0
214
(0.0
076)
(0.0
095)
(0.0
055)
Scor
e op
tion
chan
ger s
tate
pos
t 199
7 (tr
eatm
ent e
ffect
)-0
.016
0.00
53-0
.072
5(0
.003
7)(0
.006
8)(0
.005
9)Po
st 1
997
dum
my
-0.0
161
-0.0
246
0.00
79(0
.003
6)(0
.006
3)(0
.004
3)Co
nsta
nt0.
1604
0.16
930.
1351
(0.0
024)
(0.0
014)
(0.0
050)
Obs
erva
tions
4242
42N
umbe
r of S
tate
s7
77
Not
e: H
uber
-Whi
te ro
bust
stan
dard
err
ors a
re in
par
enth
eses
(clu
ster
ed b
y st
ate)
. Mod
el is
estim
ated
usin
g O
LS. S
tate
15-
17 y
ear o
ld p
opul
atio
ns a
re
used
as w
eight
s. Th
e tre
atm
ent e
ffect
repo
rted
abov
e is
the
inte
ract
ion
betw
een
the
treat
men
t sta
te d
umm
y an
d th
e po
st 1
997
dum
my,
whe
re th
e tre
atm
ent s
tate
dum
my
is eq
ual t
o 1
for t
reat
men
t sta
tes a
nd th
e po
st 1
997
dum
my
is eq
ual t
o 1
for t
he y
ears
199
8-20
00, o
ther
wise
bot
h du
mm
ies
are
equa
l to
0. S
tate
s with
few
er th
an tw
o ob
serv
atio
ns p
er p
erio
d ar
e dr
oppe
d. T
reat
men
t sta
tes a
re th
ose
stat
es th
at w
ere
requ
ired
to e
limin
ate
the
and/
or sc
ore
optio
n. T
hese
inclu
de: L
A, M
S, N
E, N
M, T
X. C
ontro
l sta
tes a
re th
ose
that
alre
ady
had
high
eno
ugh
stan
dard
s by
1997
and
hav
e lar
ge
imm
igra
nt p
opul
atio
ns. T
hese
inclu
de: C
A a
nd F
L.
82