the effects of educational choices on labor market,...
TRANSCRIPT
APPENDIX TO “THE EFFECTS OF EDUCATIONALCHOICES ON LABOR MARKET, HEALTH, AND SOCIAL
OUTCOMES” ∗
James J. HeckmanUniversity of Chicago,
University College Dublinand the American Bar Foundation
John Eric HumphriesUniversity of Chicago
Sergio UrzuaNorthwestern University, NBER
and IZA
Gregory VeramendiAarhus University
September 6, 2013
∗James Heckman: Department of Economics, University of Chicago, 1126 East 59th Street, Chicago, IL 60637;phone, 773-702-0634; fax, 773-702-8490; email, [email protected]. John Eric Humphries: Department of Economics,University of Chicago, 1126 East 59th Street, Chicago, IL 60637; phone, 773-980-6575; email, [email protected] Urzua, Department of Economics and Institute for Policy Research, Northwestern University, Handerson Hall,2001 Sheridan Road, Evanston, IL 60208; phone, 847-491-8213; email, [email protected]. Gregory Veramendi,Aarhus University, School of Economics and Management, Bartholins Alle 10, Building 1322, DK-8000 Aarhus C,Denmark; phone, 45-8942-1546; email, [email protected]. We thank Chris Taber for comments on this draft.This research was supported by NIH R01-HD32058-3, NSF SES-024158, and NSF SES-05-51089, the J.B. and M.K.Pritzker Foundation, NIH R01 HD054702, NIH R01-HD065072-01, an INET grant to the Milton Friedman Institute,an ERC grant to the University College Dublin, and the American Bar Foundation. The Web Appendix for this paperis http://jenni.uchicago.edu/effects-school-labor.
1
Abstract
Using a sequential model of educational choices, we investigate the effect of educational choiceson labor market, health, and social outcomes. Unobserved endowments drive the correlations inunobservables across choice and outcome equations. We proxy these endowments with numerousmeasurements and account for measurement error in the proxies. For each schooling level, weestimate outcomes for labor market, health, and social outcome. This allows us to generatecounter-factual outcomes for dynamic choices and a variety of policy and treatment effects. Inour framework, responses to treatment vary among observationally identical persons and agentsmay select into the treatment on the basis of their responses. We find important effects of earlycognitive and socio-emotional abilities on schooling choices, labor market outcomes, adult health,and social outcomes. Education at most levels causally produces gains on labor market, health,and social outcomes. We estimate the distribution of responses to education and find substantialheterogeneity on which agents act.
Keywords: education, early endowments, factor models, health, treatment effects.
JEL codes: C32, C38, I12, I14, I21
James Heckman John Eric HumphriesDepartment of Economics Department of EconomicsUniversity of Chicago University of Chicago1126 East 59th Street, Chicago, IL 60637 1126 East 59th Street, Chicago, IL 60637Phone: 773-702-0634 Phone: 773-980-6575Email: [email protected] Email: [email protected]
Sergio Urzua Gregory VeramendiDepartment of Economics Department of EconomicsUniversity of Maryland Arizona State University3114 Tydings Hall College Park, MD20742-7211
501 E Orange St., CPCOM 412A Tempe,AZ 85287-9801
Phone: 301-405- 3266 Phone: 480-965-0894Email: [email protected] Email: [email protected]
Contents
A Estimates of Models 4
A.1 Models of the Measurement System . . . . . . . . . . . . . . . . . . . . . . . . . . 4
A.2 Models of Labor Market Outcomes . . . . . . . . . . . . . . . . . . . . . . . . . . 11
A.3 Models of Physical Health Outcomes and Behaviors . . . . . . . . . . . . . . . . 16
A.4 Models of Mental Health Outcomes . . . . . . . . . . . . . . . . . . . . . . . . . . 21
B Variance Decompositions 26
C Goodness of Fit 31
D Evaluating the Number of Factors 36
D.1 Exploratory Factor Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
D.2 Robustness check: three factor models . . . . . . . . . . . . . . . . . . . . . . . . 38
D.2.1 Comparing simplified model to full model . . . . . . . . . . . . . . . . . . 39
D.2.2 Three-factor model with two cognitive factors . . . . . . . . . . . . . . . . 41
D.2.3 Three-factor model with two socio-emotional factors . . . . . . . . . . . . 43
D.2.4 Three-factor model with additional factor in outcomes . . . . . . . . . . . 45
E Treatment Effects: Comparison of Outcomes for different final schooling lev-
els 47
F Treatment Effects: Pairwise Comparing by Decision Node 54
G The Effect of Cognitive and Socio-emotional endowments 69
G.1 Sorting into Schooling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
G.2 Labor Market Outcomes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
G.3 Physical Health Outcomes and Behaviors . . . . . . . . . . . . . . . . . . . . . . 76
G.4 Mental Health Outcomes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
H The Effect of Endowments on Treatment Effects 85
H.1 Labor Market Outcomes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
H.2 Physical Health Outcomes and Behaviors . . . . . . . . . . . . . . . . . . . . . . 90
H.3 Mental Health Outcomes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
I Additional Description of the Data 97
I.1 Outcomes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
I.2 Schooling Levels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
I.3 Measurement System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
I.4 Control Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
J Identification of latent factor model 100
3
A Estimates of Models
A.1 Models of the Measurement System
4
Tab
le1:
Est
imat
esfo
rS
chool
ing
Ch
oice
Mod
el
Var
iab
leD
0,1
:G
rad
uate
HS
D0,2
:G
ED
D1,3
:E
nro
llC
oll
ege
D3,4
:4-y
ear
Coll
ege
Degre
evs.
Dro
pou
tof
HS
vs.
HS
Dro
pou
tvs.
HS
Gra
du
ate
vs.
Som
eC
oll
ege
βS
tdE
r.β
Std
Er.
βS
tdE
r.β
Std
Er.
Bla
ck0.0
75
0.1
29
-0.1
19
0.1
78
0.1
74
0.1
40
0.0
10
0.1
96
His
pan
ic0.6
49
0.1
79
-0.0
83
0.2
52
0.6
43
0.1
75
0.4
10
0.2
55
Bro
ken
Hom
e-0
.484
0.1
01
-0.2
40
0.1
40
-0.0
47
0.1
03
-0.2
78
0.1
41
Nu
mb
erof
Sib
lin
gs-0
.048
0.0
19
0.0
03
0.0
27
-0.0
53
0.0
19
-0.0
28
0.0
27
Mot
her
’sE
du
cati
on0.1
27
0.0
22
0.0
73
0.0
33
0.0
97
0.0
21
0.1
00
0.0
27
Fat
her
’sE
du
cati
on0.0
66
0.0
16
0.0
38
0.0
26
0.1
27
0.0
15
0.1
03
0.0
19
Fam
ily
Inco
me
0.0
20
0.0
05
0.0
18
0.0
08
0.0
12
0.0
04
0.0
13
0.0
04
Inte
rcep
t-1
.288
0.5
25
-0.9
31
0.8
74
-3.6
16
0.4
83
-2.8
63
0.6
46
Urb
an(a
ge17
)-0
.179
0.0
97
0.5
08
0.1
61
0.1
53
0.0
88
Sou
th(a
ge17
)-0
.412
0.1
21
0.3
31
0.1
93
0.1
70
0.1
17
Wes
t(ag
e17
)-0
.398
0.1
25
0.2
09
0.2
11
-0.1
72
0.1
28
Nor
thea
st(a
ge17
)0.2
13
0.1
24
0.1
36
0.2
22
0.4
00
0.1
07
Loca
lU
nem
plo
ym
ent
(age
17)
2.8
87
1.7
56
5.3
44
2.8
72
3.9
48
1.6
28
Loca
lL
ong-
run
Un
emp
loym
ent
-10.9
92
4.3
47
-2.2
17
6.8
59
-5.1
27
4.0
51
-4.0
70
4.8
28
Age
0.0
43
0.0
19
-0.0
46
0.0
31
0.0
51
0.0
19
0.0
12
0.0
23
GE
DP
assr
ate
-0.0
01
0.0
68
Loca
l4-
year
Col
lege
Tu
itio
n(a
ge17
)-0
.261
0.0
59
Loca
lU
nem
plo
ym
ent
(age
22)
-1.0
78
1.7
22
Loca
l4-
year
Col
lege
Tu
itio
n(a
ge22
)-0
.022
0.0
85
Urb
an(a
ge22
)0.0
58
0.1
31
Sou
th(a
ge22
)-0
.107
0.1
53
Wes
t(a
ge22
)-0
.469
0.1
80
Nor
thea
st(a
ge22
)0.0
31
0.1
54
Cog
nit
ive
0.8
23
0.0
93
1.0
11
0.1
38
0.8
46
0.0
92
0.8
33
0.1
14
Soci
o-em
otio
nal
0.9
83
0.0
90
0.1
69
0.1
44
0.4
93
0.0
82
0.5
87
0.1
15
N2242
522
1720
891
Not
es:
Th
enu
mb
ers
inth
ista
ble
rep
rese
nt
the
esti
mat
edco
effici
ents
and
Std
.E
rror
sas
soci
ated
wit
hin
div
idu
alb
inar
ych
oice
mod
els
ofth
ese
qu
enti
aled
uca
tion
model
.T
erm
inal
school
ing
leve
lsar
ehig
hligh
ted
inb
old.
(a)
Loca
lav
erag
eunem
plo
ym
ent
isth
eav
erag
elo
cal
unem
plo
ym
ent
leve
lov
erth
epre
vio
us
5ye
ars.
Loca
lunem
plo
ym
ent
isth
ecu
rren
tunem
plo
ym
ent
rate
.G
ED
diffi
cult
yis
the
esti
mat
ednum
ber
ofhig
hsc
hool
grad
uat
esab
leto
pas
sth
ete
ston
asi
ngl
etr
ygi
ven
the
stat
e’s
pas
sing
stan
dar
d.
Unem
plo
ym
ent
vari
able
s,tu
itio
n,
regi
ondum
mie
s,an
durb
anst
atus
are
atag
e17
for
hig
hsc
hool
grad
uat
ion,
GE
Dce
rtifi
cati
on,
and
colleg
een
rollm
ent
choi
ces.
Tuit
ion,
unem
plo
ym
ent
vari
able
s,re
gion
dum
mie
s,an
durb
anst
atus
are
atag
e22
for
the
choi
ceto
grad
uat
efr
omco
lleg
e.(h
sd=
Hig
hSch
ool
Dro
pou
t,hsg
=H
igh
Sch
ool
Gra
duat
e,sc
ol=
Som
eC
olle
ge,
colg
rad
=4-
year
colleg
egr
ad)
5
Tab
le2:
Mea
sure
men
tS
yst
em:
Est
imat
esfo
rG
rad
esin
9th
year
and
Ad
oles
cent
Ris
ky
or
Rec
kle
ssB
ehav
ior
Var
iab
leL
angu
age
Art
sG
PA
Soci
al
Stu
die
sG
PA
Sci
enc
GP
AM
ath
GP
AR
eckle
ss(H
S)
Rec
kle
ss(H
SG
rad
)β
Std
Er.
βS
tdE
r.β
Std
Er.
βS
tdE
r.β
Std
Er.
βS
tdE
r.B
lack
-0.2
580.
074
-0.2
70
0.0
78
-0.3
39
0.0
79
-0.2
60
0.0
76
0.0
89
0.1
22
-0.0
76
0.1
57
His
pan
ic0.
196
0.09
20.1
51
0.0
98
0.0
97
0.1
00
0.0
36
0.0
94
-0.2
01
0.1
56
-0.0
72
0.2
07
Bro
ken
Hom
e-0
.080
0.05
5-0
.072
0.0
59
-0.1
65
0.0
57
-0.0
57
0.0
57
0.2
30
0.0
90
-0.0
38
0.1
16
Nu
mb
erof
Sib
lin
gs-0
.035
0.01
0-0
.018
0.0
11
-0.0
10
0.0
11
-0.0
10
0.0
11
-0.0
02
0.0
17
0.0
70
0.0
22
Mot
her
’sE
du
cati
on0.
039
0.01
10.0
43
0.0
12
0.0
50
0.0
11
0.0
25
0.0
11
-0.0
18
0.0
19
-0.0
05
0.0
23
Fat
her
’sE
du
cati
on0.
030
0.00
80.0
29
0.0
09
0.0
31
0.0
08
0.0
24
0.0
08
0.0
21
0.0
15
0.0
20
0.0
16
Fam
ily
Inco
me
0.00
50.
002
0.0
08
0.0
02
0.0
05
0.0
02
0.0
03
0.0
02
0.0
05
0.0
04
-0.0
01
0.0
03
Inte
rcep
t-1
.001
0.13
3-0
.973
0.1
43
-0.9
18
0.1
37
-0.6
02
0.1
38
-1.5
26
3.4
46
7.9
56
10.3
23
Urb
an0.
018
0.04
6-0
.064
0.0
49
-0.1
09
0.0
46
-0.0
26
0.0
48
Sou
th0.
065
0.04
60.0
42
0.0
48
0.0
13
0.0
47
-0.0
04
0.0
48
Urb
an0.1
70
0.0
89
-0.0
32
0.1
04
Sou
th-0
.215
0.0
93
-0.2
10
0.1
13
Wes
t0.0
74
0.1
14
-0.0
08
0.1
32
Nor
thea
st-0
.095
0.1
15
0.1
46
0.1
22
Age
0.2
11
0.3
69
-0.6
77
0.9
87
Age
2-0
.007
0.0
10
0.0
14
0.0
24
Col
lege
Att
end
ance
0.0
95
0.1
20
Cog
nit
ive
0.45
20.
057
0.5
09
0.0
57
0.5
01
0.0
56
0.4
47
0.0
52
0.0
00
0.0
00
Soci
o-em
otio
nal
1.00
00.9
39
0.0
41
0.9
14
0.0
40
0.8
25
0.0
39
-0.0
76
0.1
57
-0.6
36
0.0
78
Std
.E
rror
0.53
50.
015
0.5
67
0.0
15
0.5
42
0.0
14
0.6
95
0.0
14
N17
721438
1493
1840
1249
897
Note
s:T
he
nu
mb
ers
inth
ista
ble
rep
rese
nts
the
esti
mate
dco
effici
ents
an
dS
td.
Err
ors
ass
oci
ate
dw
ith
the
lin
ear
regre
ssio
nm
od
els
of
gra
des
(colu
mn
)on
the
set
of
contr
ols
pre
sente
din
row
s.G
PA
refe
rsto
gra
des
rece
ived
in9th
gra
de
core
class
es.
Rec
kle
ssre
fers
toco
mit
tin
gm
inor
risk
yor
reck
less
beh
avio
r.
6
Tab
le3:
Mea
sure
men
tS
yst
em:
Cog
nit
ive
Tes
tS
core
s-<
12Y
ears
ofE
du
cati
on
atth
eti
me
ofth
ete
st
Var
iab
les
Ari
thm
etic
Word
Para
gra
ph
Nu
mer
ical
Math
Cod
ing
Rea
son
ing
Kn
owle
dge
Com
pre
hen
sion
Op
erato
ns
Kn
owle
dge
Sp
eed
βS
td.
Err
orβ
Std
.E
rror
βS
td.
Err
or
βS
td.
Err
or
βS
td.
Err
or
βS
td.
Err
or
Bla
ck-0
.636
0.07
8-0
.778
0.0
82
-0.6
35
0.0
89
-0.5
46
0.0
84
-0.4
62
0.0
75
-0.5
60
0.0
78
His
pan
ic-0
.138
0.10
2-0
.076
0.1
06
-0.0
10
0.1
16
-0.0
10
0.1
10
0.0
50
0.0
98
0.0
52
0.1
02
Bro
ken
Hom
e-0
.084
0.05
6-0
.035
0.0
59
-0.0
75
0.0
65
-0.0
36
0.0
61
-0.0
80
0.0
54
-0.0
32
0.0
57
Nu
mb
erof
Sib
lin
gs-0
.011
0.01
1-0
.040
0.0
11
-0.0
38
0.0
12
-0.0
24
0.0
12
-0.0
09
0.0
10
-0.0
16
0.0
11
Mot
her
’sE
du
cati
on0.
058
0.01
20.
082
0.0
13
0.0
79
0.0
14
0.0
41
0.0
13
0.0
66
0.0
12
0.0
38
0.0
12
Fat
her
’sE
du
cati
on0.
031
0.01
00.
040
0.0
10
0.0
43
0.0
11
0.0
30
0.0
10
0.0
43
0.0
09
0.0
20
0.0
10
Fam
ily
Inco
me
0.01
20.
003
0.009
0.0
03
0.0
09
0.0
03
0.0
11
0.0
03
0.0
12
0.0
03
0.0
12
0.0
03
Inte
rcep
t-0
.775
2.16
8-0
.773
2.2
94
-2.4
06
2.5
03
-0.8
02
2.3
86
-1.7
91
2.0
88
-3.9
81
2.2
36
Urb
an0.
079
0.05
60.
095
0.0
59
0.0
13
0.0
64
0.1
04
0.0
62
0.0
79
0.0
54
0.0
47
0.0
58
Sou
th-0
.037
0.05
9-0
.133
0.0
62
-0.0
71
0.0
68
-0.0
97
0.0
65
0.0
34
0.0
56
-0.0
21
0.0
61
Wes
t0.
057
0.07
10.
042
0.0
75
0.0
03
0.0
82
-0.0
05
0.0
78
0.0
22
0.0
68
-0.0
38
0.0
73
Nor
thea
st0.
031
0.07
3-0
.058
0.0
77
-0.0
02
0.0
84
-0.0
10
0.0
80
0.0
94
0.0
70
-0.0
73
0.0
75
Age
-0.0
890.
230
-0.1
38
0.2
44
0.0
14
0.2
66
-0.0
65
0.2
54
0.0
13
0.2
22
0.2
56
0.2
38
Age
20.
003
0.00
60.
005
0.0
06
0.0
02
0.0
07
0.0
02
0.0
07
-0.0
00
0.0
06
-0.0
06
0.0
06
Cog
nit
ive
1.00
00.
930
0.0
38
1.0
72
0.0
42
0.8
95
0.0
39
0.9
44
0.0
32
0.7
39
0.0
37
Soci
o-em
otio
nal
0.00
00.
000
0.0
00
0.0
00
0.0
00
0.0
00
Std
.E
rror
0.45
80.
014
0.560
0.0
15
0.5
80
0.0
17
0.6
17
0.0
16
0.4
55
0.0
14
0.6
19
0.0
15
N98
898
8988
988
988
988
Not
es:
The
num
ber
sin
this
table
repre
sents
the
esti
mat
edco
effici
ents
and
Std
.E
rror
sas
soci
ated
wit
hth
elinea
rre
gres
sion
model
sof
each
cogn
itiv
ete
stsc
ore
(colu
mn
)on
the
set
ofco
ntr
ols
pre
sente
din
row
s.
7
Tab
le4:
Mea
sure
men
tS
yst
em:
Cog
nit
ive
Tes
tS
core
s-
12Y
ears
ofE
du
cati
on
atth
eti
me
of
the
test
Var
iab
les
Ari
thm
etic
Word
Para
gra
ph
Nu
mer
ical
Math
Cod
ing
Rea
son
ing
Kn
owle
dge
Com
pre
hen
sion
Op
erato
ns
Kn
owle
dge
Sp
eed
βS
td.
Err
orβ
Std
.E
rror
βS
td.
Err
or
βS
td.
Err
or
βS
td.
Err
or
βS
td.
Err
or
Bla
ck-0
.855
0.09
8-0
.932
0.0
91
-0.7
39
0.0
96
-0.7
32
0.1
04
-0.4
02
0.0
95
-0.7
16
0.0
99
His
pan
ic-0
.111
0.13
5-0
.081
0.1
23
0.1
73
0.1
30
0.0
96
0.1
41
0.0
57
0.1
30
0.1
75
0.1
33
Bro
ken
Hom
e-0
.089
0.07
7-0
.077
0.0
70
-0.1
04
0.0
74
-0.0
21
0.0
80
-0.1
39
0.0
75
0.1
10
0.0
75
Nu
mb
erof
Sib
lin
gs-0
.003
0.01
4-0
.059
0.0
13
-0.0
47
0.0
13
-0.0
10
0.0
14
-0.0
31
0.0
13
-0.0
08
0.0
14
Mot
her
’sE
du
cati
on0.
046
0.01
50.
040
0.0
14
0.0
38
0.0
14
0.0
19
0.0
16
0.0
43
0.0
14
0.0
29
0.0
15
Fat
her
’sE
du
cati
on0.
048
0.01
10.
047
0.0
10
0.0
49
0.0
11
0.0
45
0.0
11
0.0
63
0.0
10
0.0
37
0.0
11
Fam
ily
Inco
me
0.00
50.
003
0.004
0.0
03
0.0
03
0.0
03
0.0
12
0.0
03
0.0
05
0.0
03
0.0
08
0.0
03
Inte
rcep
t7.
314
4.31
85.
868
3.9
72
8.8
67
4.1
90
5.5
50
4.5
37
4.2
76
4.1
71
-1.5
14
4.2
95
Urb
an-0
.038
0.06
20.
027
0.0
58
-0.0
09
0.0
61
-0.1
09
0.0
67
0.0
13
0.0
60
-0.0
52
0.0
63
Sou
th-0
.066
0.07
0-0
.058
0.0
65
-0.0
72
0.0
68
-0.1
23
0.0
74
-0.0
49
0.0
68
-0.1
87
0.0
71
Wes
t-0
.155
0.08
4-0
.049
0.0
78
-0.1
38
0.0
82
-0.1
30
0.0
89
-0.1
43
0.0
81
-0.1
92
0.0
85
Nor
thea
st0.
161
0.08
00.
097
0.0
74
0.0
69
0.0
78
-0.0
73
0.0
84
0.2
02
0.0
77
-0.0
02
0.0
80
Age
-0.8
670.
425
-0.7
28
0.3
92
-0.9
97
0.4
13
-0.6
40
0.4
47
-0.5
13
0.4
11
0.0
10
0.4
24
Age
20.
023
0.01
00.
020
0.0
10
0.0
26
0.0
10
0.0
16
0.0
11
0.0
12
0.0
10
0.0
01
0.0
10
Cog
nit
ive
1.10
70.
053
0.842
0.0
47
0.9
16
0.0
50
0.9
11
0.0
54
1.0
36
0.0
51
0.7
44
0.0
50
Soci
o-em
otio
nal
0.00
00.
000
0.0
00
0.0
00
0.0
00
0.0
00
Std
.E
rror
0.46
20.
016
0.540
0.0
16
0.5
55
0.0
16
0.6
44
0.0
18
0.4
72
0.0
16
0.6
60
0.0
17
N83
283
2832
832
832
832
Not
es:
The
num
ber
sin
this
table
repre
sents
the
esti
mat
edco
effici
ents
and
Std
.E
rror
sas
soci
ated
wit
hth
elinea
rre
gres
sion
model
sof
each
cogn
itiv
ete
stsc
ore
(colu
mn
)on
the
set
ofco
ntr
ols
pre
sente
din
row
s.
8
Tab
le5:
Mea
sure
men
tS
yst
em:
Cog
nit
ive
Tes
tS
core
s->
12Y
ears
ofE
du
cati
on
atth
eti
me
of
the
test
Var
iab
les
Ari
thm
etic
Word
Para
gra
ph
Nu
mer
ical
Math
Cod
ing
Rea
son
ing
Kn
owle
dge
Com
pre
hen
sion
Op
erato
ns
Kn
owle
dge
Sp
eed
βS
td.
Err
orβ
Std
.E
rror
βS
td.
Err
or
βS
td.
Err
or
βS
td.
Err
or
βS
td.
Err
or
Bla
ck-0
.836
0.10
5-0
.588
0.0
86
-0.3
77
0.0
90
-0.7
50
0.1
20
-0.6
88
0.1
15
-0.6
00
0.1
31
His
pan
ic-0
.366
0.11
4-0
.277
0.1
05
-0.2
24
0.1
11
0.0
68
0.1
47
-0.1
28
0.1
29
0.1
00
0.1
62
Bro
ken
Hom
e0.
007
0.08
5-0
.031
0.0
69
-0.1
63
0.0
72
-0.0
52
0.0
96
-0.1
69
0.0
93
-0.0
60
0.1
05
Nu
mb
erof
Sib
lin
gs-0
.016
0.01
9-0
.025
0.0
13
-0.0
27
0.0
14
-0.0
19
0.0
18
-0.0
12
0.0
20
-0.0
55
0.0
20
Mot
her
’sE
du
cati
on0.
054
0.01
40.
016
0.0
12
0.0
14
0.0
13
0.0
02
0.0
17
0.0
36
0.0
15
0.0
50
0.0
18
Fat
her
’sE
du
cati
on0.
029
0.00
80.
026
0.0
08
0.0
04
0.0
08
0.0
39
0.0
11
0.0
38
0.0
10
-0.0
07
0.0
12
Fam
ily
Inco
me
0.00
50.
003
0.002
0.0
02
0.0
03
0.0
02
0.0
01
0.0
03
0.0
08
0.0
03
0.0
04
0.0
03
Inte
rcep
t0.
934
8.50
9-2
.934
6.0
03
-0.4
96
6.3
18
-0.0
02
8.3
69
2.1
48
8.9
51
2.2
62
8.9
92
Urb
an0.
144
0.08
40.
129
0.0
64
0.0
87
0.0
68
0.1
55
0.0
90
0.1
39
0.0
91
0.0
64
0.0
97
Sou
th0.
037
0.07
40.
044
0.0
61
0.0
38
0.0
64
-0.1
53
0.0
85
0.0
08
0.0
81
-0.1
28
0.0
93
Wes
t-0
.046
0.10
8-0
.074
0.0
76
-0.1
14
0.0
79
0.0
15
0.1
05
-0.2
48
0.1
15
-0.1
52
0.1
13
Nor
thea
st0.
056
0.08
20.
093
0.0
65
0.0
39
0.0
69
-0.1
13
0.0
91
0.0
82
0.0
88
-0.0
67
0.1
00
Age
-0.1
840.
807
0.236
0.5
71
0.0
53
0.6
02
-0.0
45
0.7
97
-0.2
51
0.8
47
-0.2
92
0.8
57
Age
20.
005
0.01
9-0
.005
0.0
14
-0.0
01
0.0
14
0.0
02
0.0
19
0.0
06
0.0
20
0.0
08
0.0
20
Cog
nit
ive
1.14
60.
102
0.496
0.0
46
0.5
03
0.0
50
0.6
49
0.0
65
1.1
67
0.0
75
0.5
94
0.0
68
Soci
o-em
otio
nal
0.00
00.
000
0.0
00
0.0
00
0.0
00
0.0
00
Std
.E
rror
0.28
80.
049
0.417
0.0
16
0.4
45
0.0
16
0.5
96
0.0
22
0.3
76
0.0
23
0.6
71
0.0
24
N42
242
2422
422
422
422
Not
es:
The
num
ber
sin
this
table
repre
sents
the
esti
mat
edco
effici
ents
and
Std
.E
rror
sas
soci
ated
wit
hth
elinea
rre
gres
sion
model
sof
each
cogn
itiv
ete
stsc
ore
(colu
mn
)on
the
set
ofco
ntr
ols
pre
sente
din
row
s.
9
Tab
le6:
Ear
lyO
utc
omes
:E
stim
ates
for
“Ear
lyR
isky
Beh
avio
rs”
Var
iab
leT
ried
Mar
iju
anaa
Dail
yS
mokin
ga
Reg
ula
rD
rin
kin
ga
Inte
rcou
rsea
Vio
lent
in1979b
<12
yrs
sch
ool≥
12
yrs
sch
ool
βS
tdE
r.β
Std
Er.
βS
tdE
r.β
Std
Er.
βS
tdE
r.β
Std
Er.
Bla
ck-0
.327
0.10
0-0
.331
0.1
12
-0.2
42
0.1
08
0.6
00
0.0
99
-0.2
53
0.1
25
0.1
54
0.1
57
His
pan
ic-0
.152
0.12
4-0
.504
0.1
50
-0.0
09
0.1
30
-0.0
33
0.1
40
-0.3
16
0.1
61
-0.0
54
0.2
04
Bro
ken
Hom
e0.
444
0.07
20.4
11
0.0
81
0.2
43
0.0
77
0.3
66
0.0
80
0.2
20
0.0
94
0.1
30
0.1
16
Nu
mb
erof
Sib
lin
gs0.
024
0.01
40.0
35
0.0
15
0.0
27
0.0
15
0.0
11
0.0
16
0.0
09
0.0
18
0.0
13
0.0
21
Mot
her
’sE
du
cati
on0.
007
0.01
5-0
.020
0.0
17
-0.0
01
0.0
16
-0.0
22
0.0
17
0.0
30
0.0
19
-0.0
33
0.0
23
Fat
her
’sE
du
cati
on-0
.007
0.01
1-0
.037
0.0
13
-0.0
03
0.0
12
-0.0
27
0.0
13
-0.0
34
0.0
15
0.0
09
0.0
16
Fam
ily
Inco
me
0.00
00.
003
-0.0
02
0.0
03
-0.0
01
0.0
03
-0.0
03
0.0
03
-0.0
06
0.0
04
-0.0
06
0.0
03
Inte
rcep
t-0
.770
0.18
5-0
.487
0.2
13
-1.0
59
0.1
98
-0.8
60
0.2
16
-3.2
23
3.5
97
15.9
48
10.3
54
Urb
an0.
258
0.07
20.1
16
0.0
81
0.0
95
0.0
77
0.2
12
0.0
87
0.1
31
0.0
93
0.0
12
0.1
04
Sou
th-0
.103
0.06
6-0
.027
0.0
75
0.0
66
0.0
71
0.1
04
0.0
76
-0.2
04
0.0
98
0.0
50
0.1
13
Wes
t-0
.125
0.1
17
0.1
49
0.1
32
Nor
thea
st-0
.184
0.1
20
0.0
53
0.1
22
Age
0.5
17
0.3
84
-1.4
34
0.9
90
Age
2-0
.017
0.0
10
0.0
33
0.0
24
Col
lege
Att
end
ance
-0.1
32
0.1
24
Cog
nit
ive
-0.1
130.
048
-0.2
10
0.0
54
-0.1
42
0.0
51
-0.2
81
0.0
57
-0.1
60
0.0
62
-0.2
33
0.0
73
Soci
o-em
otio
nal
-0.6
040.
059
-0.5
26
0.0
64
-0.2
87
0.0
61
-0.4
01
0.0
66
-0.4
97
0.0
78
-0.2
62
0.0
90
N22
392176
2231
2218
1272
909
Note
s:T
he
num
ber
sin
this
table
repre
sent
the
esti
mate
dco
effici
ents
and
Std
.E
rrors
ass
oci
ate
dw
ith
bin
ary
choic
em
odel
sof
earl
yri
sky
beh
avio
rson
the
set
of
contr
ols
pre
sente
din
row
s.In
form
ati
on
ab
out
livin
gin
the
Wes
tand
Nort
hea
stis
only
available
in1979.
a)
The
dep
enden
tva
riable
take
sa
valu
eof
one
ifth
ein
div
idu
alh
asre
por
ted
the
beh
avio
rb
efor
eag
e15
,an
dze
root
her
wis
e.b
)T
he
vari
able
“Vio
lent”
take
sa
valu
eof
one
ifth
ein
div
idu
alp
arti
cip
ated
infi
ghti
ng
oras
sau
ltin
1979,
an
dis
esti
mate
dse
para
tely
by
edu
cati
on
level
toacc
ou
nt
for
age
diff
eren
ces
in1979.
10
A.2 Models of Labor Market Outcomes
11
Tab
le7:
Ou
tcom
eM
od
el:
Est
imat
esfo
r(L
og)
Wag
esat
Age
30by
Sch
ooli
ng
Lev
el
Var
iab
les
All
HS
Dro
pou
tG
ED
HS
Gra
d.
Som
eC
oll
ege
Coll
ege
Gra
d.
βS
tdE
rrβ
Std
Err
βS
tdE
rrβ
Std
Err
βS
tdE
rrβ
Std
Err
Bla
ck-0
.211
0.03
2-0
.198
0.06
8-0
.268
0.09
1-0
.228
0.0
53
-0.2
40
0.0
81
-0.0
650.
079
His
pan
ic-0
.044
0.04
1-0
.175
0.07
7-0
.054
0.14
6-0
.008
0.0
65
-0.1
27
0.1
03
-0.0
000.
109
Bro
ken
Hom
e-0
.012
0.02
3-0
.020
0.04
70.
126
0.07
50.
060
0.0
38
-0.0
85
0.0
59
-0.0
670.
056
Nu
mb
erof
Sib
lin
gs-0
.006
0.00
4-0
.005
0.00
90.
005
0.01
4-0
.006
0.0
07
0.009
0.011
0.007
0.010
Mot
her
’sE
du
cati
on
0.0
14
0.00
50.
006
0.01
10.
008
0.01
8-0
.000
0.0
08
0.016
0.012
0.013
0.009
Fath
er’s
Ed
uca
tion
0.0
12
0.00
30.
007
0.00
90.
016
0.01
20.
010
0.0
06
-0.0
05
0.0
09
0.006
0.007
Fam
ily
Inco
me
0.0
07
0.00
10.
008
0.00
30.
007
0.00
40.
008
0.0
02
0.008
0.002
0.005
0.001
Inte
rcep
t2.1
21
0.06
62.
098
0.16
31.
920
0.27
92.
250
0.1
16
2.335
0.188
2.295
0.143
Loca
lU
nem
plo
ym
ent
-0.6
26
0.43
9-0
.951
1.09
30.
992
1.46
7-0
.191
0.6
99
-1.3
05
1.0
99
-0.7
780.
925
Nort
hea
st0.1
45
0.02
70.
288
0.07
30.
097
0.11
80.
079
0.0
41
0.135
0.070
0.182
0.049
Sou
th-0
.013
0.02
40.
082
0.05
80.
002
0.09
0-0
.050
0.0
39
0.019
0.062
-0.0
230.
049
Wes
t0.0
30
0.02
80.
054
0.07
60.
011
0.11
00.
044
0.0
45
0.069
0.066
-0.0
040.
056
Urb
an
0.1
18
0.02
30.
055
0.05
40.
103
0.08
50.
091
0.0
34
0.124
0.060
0.139
0.056
Cog
nit
ive
0.2
01
0.01
50.
097
0.05
20.
156
0.06
10.
155
0.0
25
0.047
0.042
0.235
0.042
Soci
o-e
mot
ion
al
0.0
52
0.01
8-0
.061
0.05
30.
068
0.07
4-0
.050
0.0
30
0.033
0.051
0.057
0.045
Std
.E
rror
0.4
01
0.00
60.
325
0.01
50.
436
0.02
30.
389
0.0
10
0.412
0.016
0.387
0.013
N1991
235
183
757
340
476
Not
es:
The
num
ber
sin
this
table
repre
sents
the
esti
mat
edco
effici
ents
and
Std
.E
rror
sas
soci
ated
wit
hth
elinea
rre
gres
sion
model
sof
(log
)w
ages
on
the
set
of
contr
ols
pre
sente
din
row
s.E
ach
colu
mn
conta
ins
the
resu
lts
obta
ined
for
ap
art
icu
lar
sch
ool
ing
level
.
12
Tab
le8:
Ou
tcom
eM
od
el:
Lab
orM
arke
tP
arti
cip
atio
nat
Age
30by
Sch
ool
ing
Lev
el
Var
iab
les
All
HS
Dro
pou
tG
ED
HS
Gra
d.
Som
eC
oll
ege
Coll
ege
Gra
d.
βS
tdE
rrβ
Std
Err
βS
tdE
rrβ
Std
Err
βS
tdE
rrβ
Std
Err
Bla
ck-0
.498
0.12
6-1
.283
0.30
9-0
.249
0.30
1-0
.581
0.2
65
-0.1
08
0.3
90
His
pan
ic-0
.037
0.19
0-0
.302
0.36
0-0
.167
0.49
90.
124
0.4
57
Bro
ken
Hom
e-0
.240
0.10
3-0
.004
0.24
1-0
.473
0.26
1-0
.113
0.2
13
-0.1
69
0.3
00
-0.4
860.
320
Nu
mb
erof
Sib
lin
gs-0
.012
0.02
0-0
.016
0.04
40.
087
0.05
60.
032
0.0
43
-0.0
53
0.0
55
-0.0
070.
071
Mot
her
’sE
du
cati
on
-0.0
07
0.02
30.
028
0.05
30.
010
0.06
3-0
.010
0.0
46
0.034
0.064
-0.1
110.
064
Fath
er’s
Ed
uca
tion
-0.0
07
0.01
70.
016
0.04
50.
015
0.04
90.
002
0.0
37
-0.0
30
0.0
46
-0.0
050.
048
Fam
ily
Inco
me
0.0
16
0.00
50.
028
0.01
70.
025
0.01
60.
026
0.0
12
0.002
0.012
0.014
0.010
Inte
rcep
t1.8
33
0.33
01.
152
0.81
50.
756
0.99
81.
733
0.6
67
2.447
0.977
2.696
0.953
Loca
lU
nem
plo
ym
ent
-4.6
38
2.13
4-8
.773
5.04
63.
477
5.62
6-6
.626
4.1
34
-6.8
75
6.1
13
-2.8
016.
166
Nort
hea
st0.1
60
0.14
3-0
.166
0.36
7-0
.292
0.43
81.
019
0.4
25
0.214
0.365
0.158
0.322
Sou
th0.1
20
0.11
90.
402
0.29
9-0
.117
0.35
10.
188
0.2
25
0.155
0.313
0.271
0.335
Wes
t-0
.095
0.13
5-0
.679
0.34
6-1
.036
0.39
20.
263
0.2
86
0.380
0.351
0.754
0.517
Urb
an
0.1
29
0.11
00.
872
0.26
5-0
.263
0.33
80.
024
0.2
02
-0.3
13
0.3
42
0.321
0.330
Cog
nit
ive
0.4
26
0.07
30.
734
0.29
10.
622
0.22
90.
389
0.1
50
0.187
0.216
0.549
0.251
Soci
o-e
mot
ion
al
0.0
43
0.08
6-0
.201
0.27
8-0
.248
0.29
3-0
.167
0.1
72
-0.1
74
0.2
61
0.263
0.347
N2132
278
214
791
357
492
Note
s:T
he
nu
mb
ers
inth
ista
ble
rep
rese
nts
the
esti
mate
dco
effici
ents
an
dS
td.
Err
ors
ass
oci
ate
dw
ith
bin
ary
choic
em
od
els
of
lab
or
mar
ket
par
tici
pati
on
on
the
set
of
contr
ols
pre
sente
din
row
s.E
ach
colu
mn
conta
ins
the
resu
lts
obta
ined
for
ap
arti
cula
rsc
hooli
ng
level
.
13
Tab
le9:
Ou
tcom
eM
od
el:
Wh
ite
Col
lar
Em
plo
ym
ent
(Age
30)
Var
iab
les
All
HS
Dro
pou
tG
ED
HS
Gra
d.
Som
eC
oll
ege
Coll
ege
Gra
d.
βS
tdE
rrβ
Std
Err
βS
tdE
rrβ
Std
Err
βS
tdE
rrβ
Std
Err
Bla
ck-0
.266
0.11
5-0
.653
0.39
9-0
.417
0.33
3-0
.359
0.1
96
-0.2
86
0.2
58
-0.1
930.
331
His
pan
ic0.5
05
0.14
5-0
.547
0.42
10.
458
0.2
21
0.458
0.328
0.674
0.562
Bro
ken
Hom
e0.0
49
0.08
10.
220
0.24
50.
040
0.25
00.
126
0.1
33
-0.0
13
0.1
86
0.144
0.264
Nu
mb
erof
Sib
lin
gs-0
.038
0.01
6-0
.007
0.04
9-0
.078
0.05
6-0
.017
0.0
25
-0.0
21
0.0
36
-0.0
090.
046
Mot
her
’sE
du
cati
on
0.0
72
0.01
70.
026
0.06
20.
067
0.06
00.
039
0.0
29
0.033
0.039
0.048
0.046
Fath
er’s
Ed
uca
tion
0.0
68
0.01
20.
010
0.05
30.
014
0.04
30.
056
0.0
21
0.015
0.027
0.067
0.034
Fam
ily
Inco
me
0.0
11
0.00
3-0
.011
0.01
40.
004
0.01
30.
006
0.0
06
-0.0
02
0.0
07
0.024
0.007
Inte
rcep
t-2
.198
0.24
0-2
.169
0.85
7-1
.282
0.95
8-1
.931
0.4
17
-0.4
95
0.5
97
-1.6
410.
683
Loca
lU
nem
plo
ym
ent
-1.4
49
1.54
49.
060
5.36
7-3
.751
5.54
5-0
.069
2.4
64
-4.3
99
3.4
42
-2.4
744.
339
Nort
hea
st0.2
38
0.09
31.
070
0.41
0-0
.366
0.39
60.
141
0.1
44
0.197
0.218
0.162
0.233
Sou
th0.2
06
0.08
40.
463
0.34
4-0
.375
0.29
70.
338
0.1
35
-0.0
13
0.1
95
0.229
0.230
Wes
t-0
.074
0.09
70.
995
0.40
3-0
.529
0.35
9-0
.008
0.1
58
-0.1
68
0.2
09
-0.2
300.
255
Urb
an
0.2
97
0.08
1-0
.284
0.29
80.
669
0.32
10.
157
0.1
20
0.320
0.190
0.250
0.240
Cog
nit
ive
0.6
67
0.05
40.
432
0.26
40.
579
0.21
10.
328
0.0
88
0.263
0.132
0.771
0.180
Soci
o-e
mot
ion
al
0.3
59
0.06
2-0
.339
0.28
3-0
.179
0.23
40.
090
0.1
06
0.176
0.162
0.725
0.211
N1953
226
177
746
333
471
Note
s:W
hit
e-co
llar
occ
up
ati
on
sare
(i)
pro
fess
ion
al,
tech
nic
al,
an
dkin
dre
d;
(ii)
man
ager
s,offi
cials
,an
dp
rop
riet
ors
;(i
ii)
sale
sw
ork
ers;
(iv)
farm
ers
an
dfa
rmm
an
ager
s;an
d(v
)cl
eric
al
an
dkin
dre
d.
Th
enu
mb
ers
inth
ista
ble
rep
rese
nts
the
esti
mate
dco
effici
ents
an
dS
td.
Err
ors
asso
ciat
edw
ith
bin
ary
model
for
whet
her
ornot
the
indiv
idual
wor
ks
ina
whit
e-co
llar
occ
upat
ion
atag
e30
onth
ese
tof
contr
ols
pre
sente
din
row
s.E
ach
colu
mn
conta
ins
the
resu
lts
obta
ined
for
ap
arti
cula
rsc
hool
ing
leve
l.
14
Tab
le10:
Ou
tcom
eM
od
el:
Est
imat
esfo
r(L
og)
Pre
sent
Val
ue
ofW
ages
by
Sch
ooli
ng
Lev
el
Vari
ab
les
All
HS
Dro
pou
tG
ED
HS
Gra
d.
Som
eC
oll
ege
Coll
ege
Gra
d.
βS
tdE
rrβ
Std
Err
βS
tdE
rrβ
Std
Err
βS
tdE
rrβ
Std
Err
Bla
ck-0
.364
0.0
38-0
.751
0.08
9-0
.345
0.11
4-0
.343
0.060
-0.3
180.0
91
-0.1
150.1
01
His
pan
ic0.0
400.0
51-0
.159
0.11
3-0
.101
0.19
20.
199
0.080
-0.3
220.1
08
0.295
0.136
Bro
ken
Hom
e-0
.119
0.0
28-0
.080
0.06
5-0
.056
0.09
6-0
.045
0.044
-0.1
220.0
66
-0.1
130.0
70
Nu
mb
erof
Sib
lin
gs-0
.014
0.0
05-0
.035
0.01
20.
029
0.01
8-0
.005
0.008
-0.0
050.0
13
-0.0
040.0
13
Mot
her
’sE
du
cati
on
0.0
310.0
060.
026
0.01
50.
030
0.02
20.
019
0.010
0.013
0.013
0.035
0.012
Fath
er’s
Ed
uca
tion
0.0
160.0
040.
015
0.01
20.
019
0.01
60.
019
0.007
-0.0
090.0
10
0.004
0.009
Fam
ily
Inco
me
0.0
110.0
010.
018
0.00
40.
018
0.00
50.
012
0.002
0.008
0.002
0.008
0.002
Inte
rcep
t11
.556
0.07
311
.549
0.19
111
.124
0.31
811
.644
0.1
22
12.
132
0.193
11.
813
0.168
Urb
an0.0
830.0
270.
220
0.07
00.
044
0.11
20.
043
0.039
0.175
0.062
0.059
0.060
Sou
th0.0
390.0
290.
071
0.07
60.
160
0.11
60.
024
0.045
0.032
0.068
0.029
0.058
Wes
t-0
.038
0.0
36-0
.153
0.10
2-0
.137
0.13
7-0
.027
0.052
0.116
0.076
-0.1
410.0
78
Nort
hea
st0.1
070.0
320.
159
0.09
7-0
.019
0.14
70.
037
0.048
0.100
0.075
0.159
0.058
Cog
nit
ive
0.2
770.0
180.
431
0.07
20.
410
0.07
50.
179
0.029
0.077
0.046
0.230
0.053
Soci
o-e
mot
ion
al
0.0
860.0
220.
014
0.07
70.
009
0.09
3-0
.088
0.035
-0.0
120.0
55
0.136
0.057
Std
.E
rror
0.4
960.0
080.
454
0.02
10.
584
0.03
00.
448
0.012
0.457
0.018
0.465
0.016
N198
625
420
474
4333
451
Not
es:
The
num
ber
sin
this
table
repre
sents
the
esti
mat
edco
effici
ents
and
Std
.E
rror
sas
soci
ated
wit
hth
elinea
rre
gres
sion
model
sof
(log
)w
ages
on
the
set
of
contr
ols
pre
sente
din
row
s.E
ach
colu
mn
conta
ins
the
resu
lts
obta
ined
for
ap
art
icu
lar
sch
ool
ing
level
.
15
A.3 Models of Physical Health Outcomes and Behaviors
16
Tab
le11:
Ou
tcom
eM
od
el:
Dai
lyS
mok
ing
(Age
30)
Vari
able
sA
llH
SD
rop
out
GE
DH
SG
rad
.S
ome
Col
lege
Col
lege
Gra
d.
βS
tdE
rrβ
Std
Err
βS
tdE
rrβ
Std
Err
βS
tdE
rrβ
Std
Err
Bla
ck-0
.006
0.1
06
-0.1
410.
274
-0.0
600.
267
0.21
60.
180
-0.3
19
0.2
85
-0.0
51
0.3
57
His
pan
ic-0
.597
0.1
46
-0.9
720.
349
-1.3
310.
470
-0.2
820.
244
-0.3
88
0.3
40
0.5
46
0.4
27
Bro
ken
Hom
e0.
245
0.0
770.
305
0.20
7-0
.232
0.22
00.
041
0.128
0.6
10
0.1
98
-0.0
27
0.2
53
Nu
mb
erof
Sib
lin
gs
0.03
10.0
150.
113
0.04
2-0
.026
0.04
5-0
.004
0.025
0.0
55
0.0
37
-0.0
17
0.0
42
Moth
er’s
Ed
uca
tion
-0.0
370.0
16
0.00
90.
046
-0.0
750.
054
0.01
90.
028
0.0
33
0.0
42
-0.0
76
0.0
40
Fat
her
’sE
du
cati
on
0.01
20.0
120.
045
0.03
90.
027
0.03
70.
045
0.021
0.0
09
0.0
29
0.0
75
0.0
32
Fam
ily
Inco
me
-0.0
070.0
03
0.01
10.
013
-0.0
040.
012
-0.0
090.
005
0.0
05
0.0
07
-0.0
04
0.0
05
Inte
rcep
t-0
.012
0.2
02
-1.0
110.
595
1.27
00.
760
-0.8
840.
364
-1.2
71
0.5
89
-1.0
69
0.6
02
Nor
thea
st-0
.123
0.0
93
0.05
70.
325
-0.3
460.
355
0.12
50.
139
-0.2
67
0.2
39
-0.1
95
0.2
15
Sou
th0.
018
0.0
800.
057
0.24
6-0
.463
0.27
90.
116
0.129
-0.1
12
0.2
07
0.0
14
0.1
98
Wes
t-0
.060
0.0
95
-0.3
030.
301
-0.0
450.
326
-0.2
130.
155
0.0
33
0.2
16
-0.3
97
0.2
64
Urb
an0.
035
0.0
77-0
.080
0.22
40.
107
0.25
50.
045
0.112
0.1
16
0.1
96
0.4
18
0.2
83
Cogn
itiv
e-0
.338
0.0
50
-0.3
390.
226
-0.3
560.
180
-0.0
280.
083
-0.0
58
0.1
34
-0.2
47
0.1
80
Soci
o-em
oti
on
al-0
.470
0.0
61
-0.4
840.
233
-0.1
800.
216
-0.1
350.
103
-0.1
44
0.1
64
-0.3
87
0.1
95
N18
82
236
191
708
315
432
Not
es:
The
num
ber
sin
this
table
repre
sents
the
esti
mat
edco
effici
ents
and
Std
.E
rror
sas
soci
ated
wit
hbin
ary
model
sof
dai
lysm
okin
g(a
ge30
)on
the
set
ofco
ntr
ols
pre
sente
din
row
s.E
ach
colu
mn
conta
ins
the
resu
lts
obta
ined
for
ap
arti
cula
rsc
hooli
ng
leve
l.
17
Tab
le12:
Ou
tcom
eM
od
el:
Hea
vy
Dri
nkin
g(A
ge30
)
Vari
able
sA
llH
SD
rop
out
GE
DH
SG
rad
.S
ome
Col
lege
Col
lege
Gra
d.
βS
tdE
rrβ
Std
Err
βS
tdE
rrβ
Std
Err
βS
tdE
rrβ
Std
Err
Bla
ck-0
.346
0.1
22
-0.5
040.
285
0.12
10.
311
-0.3
800.
206
-0.7
45
0.4
05
-0.4
92
0.4
19
His
pan
ic-0
.045
0.1
49
0.35
10.
319
0.05
80.
575
-0.0
260.
250
-0.6
84
0.4
24
-0.1
39
0.5
05
Bro
ken
Hom
e0.
099
0.0
840.
243
0.20
70.
019
0.26
70.
158
0.142
-0.1
99
0.2
37
0.1
98
0.2
34
Nu
mb
erof
Sib
lin
gs
0.01
20.0
160.
003
0.04
00.
112
0.05
4-0
.009
0.027
0.0
09
0.0
48
-0.0
41
0.0
42
Moth
er’s
Ed
uca
tion
-0.0
030.0
18
0.06
00.
048
0.13
70.
067
0.00
70.
032
-0.0
22
0.0
49
-0.0
78
0.0
41
Fat
her
’sE
du
cati
on
0.01
00.0
13-0
.065
0.04
00.
022
0.04
50.
024
0.024
0.0
19
0.0
33
0.0
47
0.0
31
Fam
ily
Inco
me
-0.0
030.0
03
0.01
10.
013
-0.0
000.
016
-0.0
060.
006
-0.0
08
0.0
08
-0.0
03
0.0
05
Inte
rcep
t-0
.861
0.2
19
-1.1
020.
594
-3.1
650.
983
-0.9
280.
400
-0.4
58
0.6
90
-0.0
15
0.5
99
Nor
thea
st-0
.027
0.0
99
0.11
90.
313
0.71
10.
430
-0.0
560.
157
-0.0
92
0.2
63
-0.2
17
0.2
12
Sou
th-0
.060
0.0
88
-0.0
660.
256
0.64
20.
342
-0.0
900.
144
-0.1
91
0.2
39
0.0
17
0.2
01
Wes
t-0
.066
0.1
04
-0.1
450.
321
0.20
80.
390
-0.2
270.
176
0.2
00
0.2
44
-0.0
54
0.2
44
Urb
an0.
164
0.0
850.
643
0.25
0-0
.003
0.30
30.
188
0.127
0.1
18
0.2
18
-0.0
38
0.2
35
Cogn
itiv
e-0
.020
0.0
56
0.12
00.
232
0.16
60.
217
0.10
30.
095
0.0
88
0.1
56
-0.2
34
0.1
76
Soci
o-em
oti
on
al-0
.245
0.0
67
0.19
70.
233
0.43
00.
255
-0.2
250.
115
-0.5
20
0.2
01
-0.5
62
0.1
89
N17
23
216
165
654
292
396
Not
es:
The
num
ber
sin
this
table
repre
sents
the
esti
mat
edco
effici
ents
and
Std
.E
rror
sas
soci
ated
wit
hbin
ary
model
sof
dai
lysm
okin
g(a
ge30
)on
the
set
ofco
ntr
ols
pre
sente
din
row
s.E
ach
colu
mn
conta
ins
the
resu
lts
obta
ined
for
ap
arti
cula
rsc
hooli
ng
leve
l.
18
Tab
le13:
Ou
tcom
eM
od
el:
Ob
esit
y(A
ge40
)
Vari
able
sA
llH
SD
rop
out
GE
DH
SG
rad
.S
ome
Col
lege
Col
lege
Gra
d.
βS
tdE
rrβ
Std
Err
βS
tdE
rrβ
Std
Err
βS
tdE
rrβ
Std
Err
Bla
ck0.
136
0.1
040.
143
0.26
70.
375
0.28
5-0
.059
0.178
0.4
95
0.2
82
0.7
56
0.2
93
His
pan
ic0.
288
0.1
370.
527
0.32
01.
006
0.52
0-0
.060
0.232
0.7
47
0.3
44
-0.0
27
0.4
49
Bro
ken
Hom
e0.
011
0.0
790.
202
0.20
3-0
.014
0.25
1-0
.010
0.134
-0.1
26
0.2
07
0.1
46
0.2
14
Nu
mb
erof
Sib
lin
gs
-0.0
140.0
15
-0.0
140.
039
-0.1
110.
051
0.01
40.
025
-0.0
37
0.0
44
0.0
32
0.0
38
Moth
er’s
Ed
uca
tion
-0.0
060.0
16
-0.0
580.
046
0.00
60.
056
0.00
00.
029
0.0
38
0.0
42
-0.0
24
0.0
38
Fat
her
’sE
du
cati
on
-0.0
160.0
12
-0.0
250.
038
-0.0
370.
040
0.01
10.
021
-0.0
02
0.0
31
-0.0
28
0.0
29
Fam
ily
Inco
me
-0.0
000.0
03
0.00
20.
013
0.00
60.
015
0.00
20.
006
-0.0
09
0.0
08
0.0
00
0.0
05
Inte
rcep
t-0
.222
0.2
01
0.43
10.
590
-0.0
670.
769
-0.6
270.
367
-0.5
89
0.6
10
-0.3
12
0.5
38
Nor
thea
st0.
009
0.0
96-0
.496
0.32
80.
554
0.35
70.
157
0.153
0.0
42
0.2
46
-0.1
60
0.2
01
Sou
th0.
002
0.0
82-0
.011
0.25
8-0
.130
0.29
00.
064
0.132
0.0
12
0.2
10
-0.0
59
0.1
86
Wes
t-0
.140
0.1
01
-0.7
690.
356
-0.1
150.
339
0.17
70.
164
-0.3
92
0.2
46
-0.0
98
0.2
30
Urb
an0.
040
0.0
70-0
.237
0.22
00.
155
0.23
60.
001
0.110
0.1
36
0.1
71
0.3
12
0.1
86
Cogn
itiv
e-0
.139
0.0
51
-0.1
500.
237
0.11
80.
196
-0.1
680.
086
-0.0
81
0.1
43
-0.1
58
0.1
65
Soci
o-em
oti
on
al0.
018
0.0
620.
148
0.22
80.
164
0.23
90.
115
0.104
-0.3
51
0.1
76
0.0
04
0.1
81
N17
29
223
170
649
293
394
Not
es:
Ob
esit
yh
asb
een
mor
ep
reci
sely
defi
ned
asa
BM
Iof
30an
dab
ove.
Th
enu
mb
ers
inth
ista
ble
rep
rese
nts
the
esti
mat
edco
effici
ents
and
Sta
nd
ard
Err
ors
asso
ciat
edw
ith
bin
ary
mod
els
ofob
esit
y(a
ge40
)on
the
set
ofco
ntr
ols
pre
sente
din
row
s.E
ach
colu
mn
conta
ins
the
resu
lts
ob
tain
edfo
ra
part
icu
lar
sch
ool
ing
leve
l.
19
Tab
le14:
Ou
tcom
eM
od
el:
Physi
cal
Hea
lth
atA
ge40
(PC
S-1
2)
Vari
able
sA
llH
SD
rop
out
GE
DH
SG
rad
.S
ome
Col
lege
Col
lege
Gra
d.
βS
tdE
rrβ
Std
Err
βS
tdE
rrβ
Std
Err
βS
tdE
rrβ
Std
Err
Bla
ck-0
.042
0.0
67
-0.2
800.
219
0.21
90.
207
0.06
60.
111
-0.2
57
0.1
79
-0.1
67
0.1
08
His
pan
ic0.
119
0.0
880.
129
0.27
80.
531
0.33
9-0
.054
0.141
0.1
12
0.2
13
0.1
85
0.1
39
Bro
ken
Hom
e-0
.054
0.0
50
-0.1
170.
168
0.25
20.
175
-0.0
330.
082
0.0
10
0.1
25
-0.0
20
0.0
72
Nu
mb
erof
Sib
lin
gs
-0.0
220.0
09
-0.0
520.
032
0.02
10.
033
-0.0
180.
015
-0.0
45
0.0
25
0.0
12
0.0
13
Moth
er’s
Ed
uca
tion
0.03
00.0
100.
006
0.03
80.
028
0.04
00.
031
0.018
0.0
22
0.0
26
0.0
24
0.0
12
Fat
her
’sE
du
cati
on
0.00
80.0
080.
034
0.03
10.
006
0.02
80.
006
0.013
-0.0
16
0.0
18
-0.0
09
0.0
09
Fam
ily
Inco
me
0.00
30.0
020.
005
0.01
00.
012
0.00
9-0
.002
0.003
0.0
07
0.0
04
0.0
02
0.0
02
Inte
rcep
t-0
.351
0.1
26
-0.3
580.
497
-1.0
240.
561
-0.2
130.
226
-0.0
55
0.3
63
0.1
36
0.1
73
Nor
thea
st-0
.007
0.0
60
-0.1
240.
259
0.03
50.
269
-0.0
150.
093
0.1
00
0.1
53
-0.0
34
0.0
65
Sou
th0.
009
0.0
510.
009
0.21
00.
178
0.20
9-0
.025
0.080
0.0
85
0.1
30
-0.0
56
0.0
60
Wes
t-0
.094
0.0
61
-0.2
300.
259
-0.2
380.
243
-0.1
230.
098
0.0
73
0.1
44
-0.0
02
0.0
72
Urb
an0.
011
0.0
440.
076
0.18
0-0
.022
0.16
8-0
.028
0.067
0.0
55
0.1
04
-0.0
49
0.0
57
Cogn
itiv
e0.
162
0.0
320.
241
0.18
50.
268
0.13
60.
068
0.053
0.0
60
0.0
87
0.0
32
0.0
53
Soci
o-em
oti
on
al0.
078
0.0
40-0
.154
0.19
3-0
.224
0.15
50.
024
0.067
0.0
97
0.1
03
0.0
41
0.0
56
Std
.E
rror
0.86
40.0
141.
172
0.05
31.
038
0.05
30.
830
0.022
0.8
47
0.0
34
0.4
82
0.0
16
N19
32
247
196
730
321
438
Note
s:T
he
PC
S-1
2sc
ale
isth
eP
hysi
cal
Com
ponen
tSum
mary
(mea
sure
sphysi
cal
hea
lth)
obta
ined
from
SF
-12.
SF
-12
isa
12-q
ues
tion
hea
lth
surv
eydes
igned
by
John
Ware
of
the
New
Engla
nd
Med
ical
Cen
ter
Hosp
ital
(Ware
,K
osi
nsk
i,and
Kel
ler,
1996).
The
MC
S-1
2is
des
igned
topro
vid
ea
mea
sure
ofth
ere
spon
den
t’s
men
tal
and
physi
cal
hea
lth
irre
spec
tive
ofth
eir
pro
cliv
ity
touse
form
alhea
lth
serv
ices
.R
esp
on
den
tsw
ith
asc
ore
ab
ove
(bel
ow)
50
hav
eb
ette
r(w
ors
e)h
ealt
hth
an
the
typ
ical
per
son
inth
egen
eral
U.S
.p
op
ula
tion
.E
ach
one-
poi
nt
diff
eren
ceab
ove
orb
elow
50co
rres
pon
ds
toa
one-
tenth
ofa
stan
dar
ddev
iati
on.
For
exam
ple
,a
per
son
wit
ha
scor
eof
30is
two
stan
dar
ddev
iati
ons
away
from
the
mea
n.
We
stan
dar
diz
edth
eSF
-12
scor
eto
hav
em
ean
zero
and
vari
ance
one
inth
eov
eral
lp
opula
tion
.
20
A.4 Models of Mental Health Outcomes
21
Tab
le15:
Ou
tcom
eM
od
el:
Men
tal
Hea
lth
atA
ge40
(MC
S-1
2)
Vari
able
sA
llH
SD
rop
out
GE
DH
SG
rad
.S
ome
Col
lege
Col
lege
Gra
d.
βS
tdE
rrβ
Std
Err
βS
tdE
rrβ
Std
Err
βS
tdE
rrβ
Std
Err
Bla
ck0.
048
0.0
71-0
.210
0.21
90.
428
0.20
9-0
.004
0.109
-0.1
95
0.1
70
0.3
14
0.1
74
His
pan
ic0.
092
0.0
920.
465
0.27
8-0
.176
0.34
10.
053
0.139
0.1
60
0.2
03
-0.1
54
0.2
23
Bro
ken
Hom
e-0
.170
0.0
52
-0.1
540.
168
0.16
00.
176
-0.2
130.
081
-0.2
52
0.1
19
-0.1
48
0.1
17
Nu
mb
erof
Sib
lin
gs
-0.0
070.0
10
-0.0
700.
032
0.04
80.
033
0.00
00.
015
0.0
18
0.0
24
0.0
08
0.0
21
Moth
er’s
Ed
uca
tion
0.00
80.0
110.
004
0.03
8-0
.022
0.04
10.
006
0.018
0.0
44
0.0
24
-0.0
02
0.0
19
Fat
her
’sE
du
cati
on
-0.0
060.0
08
0.00
30.
031
0.04
00.
028
-0.0
000.
013
-0.0
31
0.0
17
-0.0
18
0.0
15
Fam
ily
Inco
me
0.00
10.0
020.
008
0.01
00.
002
0.00
90.
000
0.003
0.0
01
0.0
04
-0.0
01
0.0
02
Inte
rcep
t0.
227
0.1
320.
444
0.49
7-0
.575
0.56
50.
229
0.222
0.0
35
0.3
48
0.5
70
0.2
79
Nor
thea
st0.
012
0.0
63-0
.237
0.25
90.
138
0.27
2-0
.015
0.092
0.3
01
0.1
47
0.0
11
0.1
04
Sou
th-0
.059
0.0
53
-0.1
530.
211
0.07
50.
210
-0.0
630.
079
0.1
89
0.1
24
-0.0
45
0.0
97
Wes
t-0
.066
0.0
64
-0.2
720.
259
0.07
70.
245
-0.1
230.
097
0.2
22
0.1
37
-0.1
58
0.1
16
Urb
an-0
.062
0.0
46
0.10
30.
180
-0.1
430.
169
-0.0
340.
066
-0.1
97
0.0
99
-0.0
89
0.0
91
Cogn
itiv
e0.
072
0.0
340.
236
0.18
50.
158
0.13
70.
065
0.052
-0.0
30
0.0
82
-0.0
53
0.0
87
Soci
o-em
oti
on
al0.
088
0.0
420.
321
0.20
3-0
.254
0.17
50.
010
0.063
0.1
63
0.1
04
0.1
38
0.0
95
Std
.E
rror
0.90
60.0
151.
167
0.05
51.
045
0.05
40.
818
0.021
0.8
07
0.0
32
0.7
75
0.0
26
N19
32
247
196
730
321
438
Note
s:T
he
MC
S-1
2sc
ale
isth
eM
enta
lC
om
pon
ent
Su
mm
ary
(mea
sure
sm
enta
lh
ealt
h)
ob
tain
edfr
om
SF
-12.
SF
-12
isa
12-q
ues
tion
hea
lth
surv
eydes
igned
by
John
Ware
of
the
New
Engla
nd
Med
ical
Cen
ter
Hosp
ital
(Ware
,K
osi
nsk
i,and
Kel
ler,
1996).
The
MC
S-1
2is
des
igned
topro
vid
ea
mea
sure
ofth
ere
spon
den
t’s
men
tal
and
physi
cal
hea
lth
irre
spec
tive
ofth
eir
pro
cliv
ity
touse
form
alhea
lth
serv
ices
.R
esp
on
den
tsw
ith
asc
ore
ab
ove
(bel
ow)
50
hav
eb
ette
r(w
ors
e)h
ealt
hth
an
the
typ
ical
per
son
inth
egen
eral
U.S
.p
op
ula
tion
.E
ach
one-
poi
nt
diff
eren
ceab
ove
orb
elow
50co
rres
pon
ds
toa
one-
tenth
ofa
stan
dar
ddev
iati
on.
For
exam
ple
,a
per
son
wit
ha
scor
eof
30is
two
stan
dar
ddev
iati
ons
away
from
the
mea
n.
We
stan
dar
diz
edth
eSF
-12
scor
eto
hav
em
ean
zero
and
vari
ance
one
inth
eov
eral
lp
opula
tion
.
22
Tab
le16:
Ou
tcom
eM
od
el:
Th
eC
ente
rfo
rE
pid
emio
logi
cS
tud
ies
Dep
ress
ion
Sca
le(C
ES
-D)
(Age
40)
Vari
able
sA
llH
SD
rop
out
GE
DH
SG
rad
.S
ome
Col
lege
Col
lege
Gra
d.
βS
tdE
rrβ
Std
Err
βS
tdE
rrβ
Std
Err
βS
tdE
rrβ
Std
Err
Bla
ck-0
.094
0.0
69
-0.2
530.
226
0.20
40.
199
-0.0
870.
114
-0.7
16
0.1
48
0.1
61
0.1
44
His
pan
ic0.
032
0.0
910.
093
0.29
00.
218
0.32
50.
043
0.147
-0.2
72
0.1
76
0.0
38
0.1
84
Bro
ken
Hom
e-0
.159
0.0
51
-0.2
310.
173
0.30
20.
168
-0.1
690.
084
-0.1
96
0.1
03
-0.0
53
0.0
96
Nu
mb
erof
Sib
lin
gs
-0.0
240.0
10
-0.0
850.
033
0.01
40.
031
-0.0
240.
016
0.0
04
0.0
21
0.0
20
0.0
17
Moth
er’s
Ed
uca
tion
0.02
50.0
110.
038
0.04
00.
033
0.03
80.
009
0.018
0.0
36
0.0
21
0.0
03
0.0
16
Fat
her
’sE
du
cati
on
0.00
70.0
08-0
.030
0.03
20.
047
0.02
70.
017
0.014
-0.0
39
0.0
15
0.0
06
0.0
12
Fam
ily
Inco
me
0.00
10.0
020.
014
0.01
10.
012
0.00
9-0
.007
0.004
0.0
02
0.0
04
-0.0
00
0.0
02
Inte
rcep
t-0
.056
0.1
30
0.38
10.
516
-1.0
700.
534
0.21
60.
233
0.3
77
0.3
02
0.2
24
0.2
31
Nor
thea
st-0
.055
0.0
62
0.01
20.
266
0.03
60.
260
-0.1
820.
096
0.0
45
0.1
28
0.0
40
0.0
86
Sou
th-0
.116
0.0
52
-0.3
720.
219
-0.0
640.
200
-0.1
420.
083
0.1
50
0.1
08
-0.0
40
0.0
80
Wes
t-0
.190
0.0
63
-0.7
260.
272
-0.3
630.
232
-0.2
220.
102
0.1
21
0.1
19
-0.0
48
0.0
96
Urb
an-0
.020
0.0
45
0.09
10.
186
-0.2
520.
161
-0.0
000.
069
-0.1
41
0.0
86
0.0
11
0.0
76
Cogn
itiv
e0.
230
0.0
330.
501
0.19
60.
194
0.12
90.
154
0.054
0.1
58
0.0
72
-0.0
50
0.0
72
Soci
o-em
oti
on
al0.
016
0.0
42-0
.044
0.22
2-0
.218
0.16
9-0
.044
0.067
-0.0
46
0.0
87
0.0
20
0.0
76
Std
.E
rror
0.88
70.0
141.
201
0.05
50.
998
0.05
10.
853
0.022
0.7
01
0.0
28
0.6
42
0.0
22
N19
22
242
197
726
320
437
Note
s:C
ES
-Dis
on
eof
the
most
com
mon
scre
enin
gte
sts
for
hel
pin
gan
ind
ivid
ual
tod
eter
min
eh
isor
her
dep
ress
ion
qu
oti
ent.
Th
issc
ale
mea
sure
ssy
mpto
ms
of
dep
ress
ion,
dis
crim
inate
sb
etw
een
clin
ically
dep
ress
edin
div
iduals
and
oth
ers,
and
ishig
hly
corr
elate
dw
ith
oth
erd
epre
ssio
nra
tin
gsc
ale
s(s
eeR
ad
loff
,1977;
Ross
and
Mir
owsk
y,
1989).
We
form
the
scale
sum
min
gth
esc
ore
sfr
om
the
item
s:“I
did
not
feel
like
eati
ng;
my
ap
pet
ite
was
poor”
,“I
had
trou
ble
kee
pin
gm
ym
ind
on
wh
at
Iw
as
doin
g”,
“I
felt
dep
ress
ed”,
“I
felt
that
ever
yth
ing
Idid
was
aneff
ort”
,“M
ysl
eep
was
rest
less
”,“I
felt
sad”
and
“Ico
uld
not
get
goin
g”.
For
each
item
sth
ep
oten
tial
answ
ers
are:
“0R
arel
y/N
one
ofth
eti
me/
1D
ay”,
“1S
ome/
Alitt
leof
the
tim
e/1-
2D
ays”
,“2
Occ
asio
nal
ly/M
od
erat
eam
ount
ofth
eti
me/
3-4
Day
s”,
“3M
ost/
All
ofth
eti
me/
5-7
Day
s”.
We
stan
dar
diz
edth
esc
ores
tohav
em
ean
0an
dva
rian
ce1
inth
eov
eral
lp
opula
tion
.T
he
num
ber
sin
this
table
repre
sents
the
esti
mate
dco
effici
ents
and
Std
.E
rrors
ass
oci
ate
dw
ith
the
linea
rre
gre
ssio
nm
odel
sof
CE
S-D
on
the
set
of
contr
ols
pre
sente
din
row
s.E
ach
colu
mn
conta
ins
the
resu
lts
obta
ined
for
ap
arti
cula
rsc
hool
ing
leve
l.
23
Tab
le17:
Ou
tcom
eM
od
el:
Pea
rlin
’s“P
erso
nal
Mas
tery
Sca
le”
Vari
able
sA
llH
SD
rop
out
GE
DH
SG
rad
.S
ome
Col
lege
Col
lege
Gra
d.
βS
tdE
rrβ
Std
Err
βS
tdE
rrβ
Std
Err
βS
tdE
rrβ
Std
Err
Bla
ck0.
004
0.0
690.
080
0.15
50.
018
0.16
9-0
.113
0.121
-0.0
96
0.1
91
0.1
82
0.1
80
His
pan
ic0.
085
0.0
910.
229
0.18
90.
320
0.28
40.
023
0.153
-0.2
14
0.2
29
-0.0
98
0.2
52
Bro
ken
Hom
e-0
.029
0.0
52
-0.0
610.
118
-0.0
110.
144
0.06
50.
089
-0.1
28
0.1
38
0.0
69
0.1
28
Nu
mb
erof
Sib
lin
gs
-0.0
160.0
10
-0.0
210.
022
0.00
90.
028
-0.0
100.
016
-0.0
12
0.0
27
0.0
01
0.0
23
Moth
er’s
Ed
uca
tion
0.02
80.0
110.
055
0.02
70.
077
0.03
50.
004
0.019
-0.0
03
0.0
28
0.0
19
0.0
21
Fat
her
’sE
du
cati
on
0.01
80.0
080.
005
0.02
30.
034
0.02
40.
017
0.014
0.0
04
0.0
20
-0.0
09
0.0
16
Fam
ily
Inco
me
0.00
40.0
020.
003
0.00
80.
014
0.00
70.
002
0.004
-0.0
00
0.0
05
0.0
03
0.0
03
Inte
rcep
t-0
.562
0.1
31
-0.9
180.
355
-1.1
480.
484
-0.3
710.
242
0.2
22
0.4
06
0.2
86
0.3
08
Nor
thea
st0.
112
0.0
61-0
.105
0.18
70.
275
0.22
00.
178
0.097
0.1
20
0.1
62
-0.0
36
0.1
14
Sou
th0.
002
0.0
53-0
.029
0.14
0-0
.158
0.17
10.
039
0.090
0.0
21
0.1
43
-0.0
12
0.1
08
Wes
t0.
050
0.0
630.
040
0.17
9-0
.266
0.19
70.
086
0.104
0.1
41
0.1
54
-0.0
10
0.1
29
Urb
an0.
033
0.0
51-0
.025
0.12
9-0
.127
0.16
00.
085
0.078
0.0
54
0.1
37
-0.0
86
0.1
28
Cogn
itiv
e0.
288
0.0
330.
227
0.12
90.
420
0.11
20.
298
0.057
0.0
86
0.0
96
-0.0
63
0.0
94
Soci
o-em
oti
on
al0.
092
0.0
40-0
.125
0.13
60.
255
0.13
60.
042
0.068
-0.0
12
0.1
14
0.0
44
0.1
06
Std
.E
rror
0.94
30.0
150.
879
0.03
80.
903
0.04
50.
928
0.023
0.9
89
0.0
37
0.9
12
0.0
29
N21
30
277
217
790
359
487
Note
s:P
earl
ins
“P
erso
nal
Mast
ery
Sca
le”
con
sist
sof
7it
ems
wh
ich
are
an
swer
edon
a4-p
oin
t(4
stro
ngly
agre
e,3
agre
e,2
dis
agre
e,1
stro
ngly
dis
agre
e)sc
ale
and
has
bee
nsh
own
toex
hib
itre
aso
nable
inte
rnal
reliabilit
yand
good
const
ruct
validit
y(P
earl
inand
Sch
oole
r,19
78;
Pea
rlin
,M
enag
han
,L
ieb
erm
an,
and
Mull
an,
1981
).T
he
item
sar
e“t
her
eis
real
lyn
ow
ayi
can
solv
eso
me
ofth
ep
rob
lem
si
hav
e”,
“so
met
imes
ife
elth
at
i’m
bei
ng
pu
shed
aro
un
din
life
”,
“i
hav
elitt
leco
ntr
ol
over
the
thin
gs
that
hap
pen
tom
e”,
“i
can
do
just
ab
ou
tanyth
ing
ire
ally
set
my
min
dto
”,
“i
oft
enfe
elhel
ple
ssin
dea
ling
wit
hth
epro
ble
ms
of
life
”,
“w
hat
happ
ens
tom
ein
the
futu
rem
ost
lyd
epen
ds
onm
e”,
“th
ere
islitt
lei
can
do
toch
ange
man
yof
the
imp
orta
nt
thin
gsin
my
life
”.W
efo
rmth
esc
ale
sum
min
gth
esc
ores
from
the
item
s,an
dst
and
ard
izin
gth
esc
ores
toh
ave
mea
n0
and
vari
ance
1in
the
over
all
pop
ula
tion
.T
he
nu
mb
ers
inth
ista
ble
rep
rese
nts
the
esti
mat
edco
effici
ents
and
Std
.E
rror
sas
soci
ated
wit
hth
elinea
rre
gres
sion
model
son
the
set
ofco
ntr
ols
pre
sente
din
row
s.E
ach
colu
mn
conta
ins
the
resu
lts
ob
tain
edfo
ra
par
ticu
lar
school
ing
leve
l.
24
Tab
le18:
Ou
tcom
eM
od
el:
Ros
enb
erg’
sS
elf-
Est
eem
Sca
le
Vari
able
sA
llH
SD
rop
out
GE
DH
SG
rad
.S
ome
Col
lege
Col
lege
Gra
d.
βS
tdE
rrβ
Std
Err
βS
tdE
rrβ
Std
Err
βS
tdE
rrβ
Std
Err
Bla
ck0.
025
0.0
77-0
.062
0.19
10.
343
0.21
5-0
.146
0.122
0.2
01
0.2
02
0.1
91
0.2
13
His
pan
ic0.
118
0.1
040.
056
0.23
5-0
.166
0.38
70.
082
0.168
0.4
05
0.2
46
-0.0
89
0.2
98
Bro
ken
Hom
e0.
011
0.0
580.
046
0.14
50.
236
0.18
50.
059
0.092
0.1
20
0.1
48
-0.1
90
0.1
43
Nu
mb
erof
Sib
lin
gs
-0.0
240.0
11
-0.0
210.
028
-0.0
560.
035
-0.0
210.
017
0.0
20
0.0
31
-0.0
20
0.0
26
Moth
er’s
Ed
uca
tion
0.03
20.0
120.
037
0.03
4-0
.037
0.04
10.
034
0.021
0.0
42
0.0
29
0.0
24
0.0
25
Fat
her
’sE
du
cati
on
0.01
40.0
090.
028
0.02
70.
010
0.02
90.
012
0.015
-0.0
06
0.0
21
-0.0
05
0.0
19
Fam
ily
Inco
me
-0.0
000.0
02
0.00
10.
009
0.01
00.
011
0.00
30.
004
0.0
02
0.0
05
-0.0
07
0.0
03
Inte
rcep
t-0
.459
0.1
46
-0.4
960.
422
0.38
40.
588
-0.5
540.
257
-0.4
06
0.4
30
0.2
87
0.3
60
Nor
thea
st0.
182
0.0
690.
060
0.23
40.
300
0.27
00.
037
0.106
0.1
93
0.1
77
0.3
71
0.1
31
Sou
th0.
094
0.0
59-0
.056
0.18
6-0
.050
0.21
50.
103
0.090
0.0
07
0.1
51
0.2
36
0.1
23
Wes
t0.
195
0.0
720.
232
0.22
3-0
.082
0.24
70.
129
0.115
0.0
44
0.1
69
0.4
46
0.1
49
Urb
an-0
.073
0.0
51
-0.1
410.
158
-0.3
770.
175
-0.0
530.
076
-0.0
64
0.1
21
-0.1
01
0.1
16
Cogn
itiv
e0.
349
0.0
370.
561
0.16
30.
435
0.14
10.
391
0.060
0.0
72
0.1
01
0.0
50
0.1
11
Soci
o-em
oti
on
al0.
078
0.0
45-0
.048
0.16
5-0
.025
0.18
40.
037
0.072
0.2
43
0.1
23
-0.0
73
0.1
21
Std
.E
rror
0.93
60.0
160.
939
0.04
70.
972
0.05
50.
884
0.025
0.9
29
0.0
39
0.9
21
0.0
33
N16
94
210
162
648
286
388
Not
es:
Ros
enb
erg’
sSel
f-E
stee
mSca
leco
nsi
sts
of11
item
sw
hic
har
ean
swer
edon
a4-
poi
nt
(4st
rongl
yag
ree,
3ag
ree,
2dis
agre
e,1
stro
ngl
yd
isagre
e).
Th
eit
ems
are
“I
feel
that
I’m
ap
erso
nof
wort
h,
at
least
on
equ
al
basi
sw
ith
oth
ers”
,“I
feel
that
Ih
ave
anu
mb
erof
good
qual
itie
s”,
“All
inal
l,I
amin
clin
edto
feel
that
Iam
afa
ilure
”,“I
amab
leto
do
thin
gsas
wel
las
mos
tot
her
peo
ple
”,“I
feel
Ido
not
hav
em
uch
tob
epro
ud
of”,
“Ita
kea
pos
itiv
eat
titu
de
tow
ard
myse
lf”,
“On
the
whol
e,I
amsa
tisfi
edw
ith
myse
lf”,
“Iw
ish
Ico
uld
hav
em
ore
resp
ect
for
myse
lf”
,“I
cert
ainly
feel
use
less
atti
mes
”,“A
tti
mes
Ith
ink
Iam
no
good
atal
l”.
We
form
the
scal
esu
mm
ing
the
scor
esfr
omth
eit
ems,
and
stan
dar
diz
ing
the
scor
esto
hav
em
ean
0an
dva
rian
ce1
inth
eov
eral
lp
opu
lati
on.
Th
enu
mb
ers
inth
ista
ble
rep
rese
nts
the
esti
mat
edco
effici
ents
and
Std
.E
rror
sas
soci
ated
wit
hth
elinea
rre
gres
sion
model
son
the
set
ofco
ntr
ols
pre
sente
din
row
s.E
ach
colu
mn
conta
ins
the
resu
lts
ob
tain
edfo
ra
par
ticu
lar
school
ing
leve
l.
25
B Variance Decompositions
Table 19: Variance decomposition of educational decisions and grades
Observables Latent endowments Unobservables
D0,1: Graduate HS 0.25 0.35 0.40D0,2: GED 0.17 0.28 0.55D1,3: Enroll College 0.27 0.25 0.48D3,4: 4-year College Degree 0.24 0.28 0.49GPA Language 0.08 0.63 0.29GPA Social Sciences 0.09 0.59 0.32GPA Science 0.10 0.59 0.31GPA Math 0.04 0.47 0.49
Notes: The fraction of variance explained by each term of the models. The components are theobservables (X′β), latent endowments (αθ), and unobservables (e).
26
Table 20: Variance decomposition of ASVAB tests
Observables Latent endowments Unobservables
Arithmetic Reasoning (< 12) 0.25 0.52 0.23Word Knowledge (< 12) 0.32 0.38 0.30Paragraph Comprehension (< 12) 0.26 0.45 0.29Numerical Operations (< 12) 0.18 0.40 0.42Math Knowledge (< 12) 0.25 0.50 0.25Coding Speed (< 12) 0.18 0.33 0.49Arithmetic Reasoning (= 12) 0.23 0.56 0.21Word Knowledge (= 12) 0.32 0.36 0.32Paragraph Comprehension (= 12) 0.23 0.43 0.34Numerical Operations (= 12) 0.19 0.39 0.42Math Knowledge (= 12) 0.23 0.53 0.24Coding Speed (= 12) 0.18 0.30 0.51Arithmetic Reasoning (> 12) 0.21 0.69 0.09Word Knowledge (> 12) 0.27 0.29 0.44Paragraph Comprehension (> 12) 0.15 0.32 0.54Numerical Operations (> 12) 0.19 0.29 0.53Math Knowledge (> 12) 0.20 0.66 0.15Coding Speed (> 12) 0.16 0.22 0.62
Notes: The fraction of variance explained by each term of the models. The components are theobservables (X′β), latent endowments (αθ), and unobservables (e).
Table 21: Variance decomposition of early reckless and adverse behaviors
Observables Latent endowments Unobservables
Early Violence (9th-11th)b 0.11 0.10 0.79Early Violence (12h)b 0.15 0.05 0.79Early Reckless (9th-11th)b 0.05 0.03 0.92Early Reckless (12th)b 0.12 0.02 0.86Early marijuanac 0.05 0.14 0.81Early daily smokingc 0.06 0.13 0.81Early drinkingc 0.02 0.05 0.93Early intercoursec 0.11 0.10 0.79
Notes: The fraction of variance explained by each term of the models. The components are theobservables (X′β), latent endowments (αθ), and unobservables (e).
27
Table 22: Variance decomposition of labor market outcomes
Observables Latent endowments Unobservables
Log Wages (30) 0.18 0.10 0.72High school dropouts 0.27 0.03 0.70GED Recipients 0.15 0.06 0.79High school graduates 0.16 0.05 0.79Some college 0.17 0.01 0.82Four-year college graduate 0.12 0.15 0.74
PVLog Wages (30) 0.25 0.11 0.64High school dropouts 0.44 0.17 0.39GED Recipients 0.21 0.15 0.64High school graduates 0.26 0.05 0.69Some college 0.19 0.01 0.80Four-year college graduate 0.15 0.13 0.72
LF Participation 0.10 0.07 0.82High school dropouts 0.33 0.13 0.54GED Recipients 0.25 0.11 0.64High school graduates 0.24 0.05 0.72Some college 0.08 0.02 0.90Four-year college graduate 0.15 0.14 0.71
White Collar Employment 0.17 0.20 0.63High school dropouts 0.21 0.07 0.71GED Recipients 0.19 0.10 0.71High school graduates 0.11 0.05 0.84Some college 0.07 0.05 0.89Four-year college graduate 0.16 0.32 0.52
Notes: The fraction of variance explained by each term of the models. The components are theobservables (X′β), latent endowments (αθ), and unobservables (e).
28
Table 23: Variance decomposition of physical health outcomes
Observables Latent endowments Unobservables
SF12 Physical Health 0.03 0.02 0.95High school dropouts 0.05 0.02 0.93GED Recipients 0.05 0.04 0.91High school graduates 0.02 0.00 0.97Some college 0.05 0.01 0.94Four-year college graduate 0.04 0.01 0.95
Obesity 0.02 0.01 0.98High school dropouts 0.17 0.01 0.82GED Recipients 0.16 0.02 0.83High school graduates 0.01 0.01 0.98Some college 0.07 0.05 0.87Four-year college graduate 0.13 0.01 0.86
Smoking Age 30 0.05 0.15 0.81High school dropouts 0.13 0.14 0.73GED Recipients 0.13 0.07 0.80High school graduates 0.06 0.01 0.93Some college 0.10 0.01 0.89Four-year college graduate 0.09 0.09 0.82
Heavy Drinking Age 30 0.02 0.03 0.96High school dropouts 0.13 0.02 0.84GED Recipients 0.17 0.08 0.75High school graduates 0.04 0.02 0.94Some college 0.09 0.09 0.82Four-year college graduate 0.05 0.15 0.80
Notes: The fraction of variance explained by each term of the models. The components are theobservables (X′β), latent endowments (αθ), and unobservables (e).
29
Table 24: Variance decomposition of mental health outcomes
Observables Latent endowments Unobservables
SF12 Mental Health 0.01 0.01 0.98High school dropouts 0.05 0.06 0.89GED Recipients 0.04 0.03 0.93High school graduates 0.02 0.00 0.98Some college 0.07 0.02 0.92Four-year college graduate 0.04 0.01 0.94
CESD Depression 0.04 0.03 0.93High school dropouts 0.12 0.06 0.81GED Recipients 0.10 0.03 0.88High school graduates 0.03 0.01 0.96Some college 0.15 0.02 0.83Four-year college graduate 0.01 0.00 0.98
Pearlin Self Mastery 0.03 0.05 0.92High school dropouts 0.05 0.03 0.93GED Recipients 0.16 0.12 0.72High school graduates 0.02 0.05 0.93Some college 0.01 0.00 0.98Four-year college graduate 0.01 0.00 0.99
Rosenberg Self-Esteem 0.03 0.07 0.91High school dropouts 0.05 0.13 0.82GED Recipients 0.08 0.08 0.84High school graduates 0.03 0.08 0.88Some college 0.03 0.03 0.94Four-year college graduate 0.05 0.00 0.95
Notes: The fraction of variance explained by each term of the models. The components are theobservables (X′β), latent endowments (αθ), and unobservables (e).
30
C Goodness of Fit
Table 25: Goodness of Fit - Schooling Choice
Schooling Level Data Model p-value
High School Dropout 0.13 0.12 0.62GED (No College) 0.10 0.10 0.94High School Graduate 0.37 0.38 0.79Some College 0.17 0.18 0.68College Graduate 0.23 0.22 0.72
Notes: The simulated data (Model) contains one million observations generated from the Model’sestimates. The actual data (Actual) contains 2242 observations from the NLSY79 sample of Males.(a) Goodness of fit is tested using a χ2 test where the Null Hypothesis is Model=Data.
31
Table 26: Goodness of Fit - ASVAB and Grade models
mean(Actual) mean(Model) std(Actual) std(Model) p-valuea
ASVAB TestsArithmetic Reasoning (< 12) -0.29 0.03 0.97 0.95 0.00Word Knowledge (< 12) -0.45 -0.10 1.04 1.02 0.00Paragraph Comprehension (< 12) -0.51 -0.11 1.09 1.08 0.00Numerical Operations (< 12) -0.52 -0.26 0.98 0.96 0.00Math Knowledge (< 12) -0.32 -0.05 0.94 0.91 0.00Coding Speed (< 12) -0.60 -0.34 0.88 0.88 0.00Arithmetic Reasoning (= 12) 0.20 0.21 0.93 1.01 0.01Word Knowledge (= 12) 0.13 0.12 0.88 0.95 0.00Paragraph Comprehension (= 12) 0.04 0.08 0.89 0.95 0.00Numerical Operations (= 12) -0.01 0.01 0.94 0.99 0.00Math Knowledge (= 12) 0.00 0.06 0.90 0.96 0.00Coding Speed (= 12) -0.16 -0.21 0.88 0.92 0.00Arithmetic Reasoning (> 12) 0.94 0.31 0.82 0.94 0.00Word Knowledge (> 12) 0.74 0.33 0.57 0.63 0.00Paragraph Comprehension (> 12) 0.64 0.33 0.57 0.61 0.00Numerical Operations (> 12) 0.58 0.17 0.77 0.82 0.00Math Knowledge (> 12) 0.97 0.40 0.86 0.98 0.00Coding Speed (> 12) 0.47 0.07 0.81 0.86 0.009th Grade GPAGPA Language -0.12 -0.20 0.98 0.99 0.00GPA Social Sciences -0.01 -0.10 0.99 1.00 0.00GPA Science 0.03 -0.06 0.98 0.98 0.00GPA Math -0.01 -0.07 0.99 0.99 0.00
Notes: The simulated data (Model) contains one million observations generated from the Model’sestimates. The actual data (Actual) contains 2242 observations from the NLSY79 sample ofMales.The numbers inside the paranthesis describe the years of schooling at the time of the test.The ASVAB models are estimated separately for those with less than twelve years (< 12), thosewho are high school graduates (=12), and those who have attended college (> 12) at the time theytook the ASVAB tests. (a) The equality of the Model and Data distributions are tested using atwo-sample Kolmogorov-Smirnov test.
32
Table 27: Goodness of Fit - Early Risky and Reckless Behavior
Outcome Actual Model p-valuea
Early marijuanac 0.34 0.34 0.98Early daily smokingc 0.19 0.19 0.99Early drinkingc 0.19 0.19 1.00Early intercoursec 0.16 0.16 0.90Early Violence (9th-11th)b 0.67 0.58 0.01Early Violence (12h)b 0.50 0.64 0.00Early Reckless (9th-11th)b 0.61 0.57 0.33Early Reckless (12th)b 0.53 0.63 0.02
Notes: The simulated data (Model) contains one million observations generated from the Model’sestimates. The actual data (Actual) contains 2242 observations from the NLSY79 sample of Males.(a) Goodness of fit is tested using a χ2 test where the Null Hypothesis is Model=Data. (b) Thereckless and violent variables are taken from the NSLY 1980 Illegal Activities Supplement. (c) Earlyis defined as engaging in risky behavior before 15 years old.
Table 28: Goodness of Fit - Discrete Labor Outcomes
Outcome Actual Model p-valuea
LF Participation 0.92 0.94 0.17High school dropouts 0.83 0.84 0.73GED Recipients 0.84 0.85 0.78High school graduates 0.94 0.96 0.33Some college 0.94 0.95 0.55Four-year college graduate 0.96 0.97 0.53
White Collar Employment 0.44 0.43 0.78High school dropouts 0.15 0.15 0.96GED Recipients 0.24 0.26 0.85High school graduates 0.28 0.29 0.85Some college 0.48 0.49 0.94Four-year college graduate 0.87 0.85 0.76
Notes: The simulated data (Model) contains one million observations generated from the Model’sestimates. The actual data (Actual) contains 2242 observations from the NLSY79 sample of Males.(a) Goodness of fit is tested using a χ2 test where the Null Hypothesis is Model=Data.
33
Table 29: Goodness of Fit - Discrete Health Outcomes
Outcome Actual Model p-valuea
Obesity 0.32 0.32 0.98High school dropouts 0.30 0.30 0.95GED Recipients 0.29 0.30 0.99High school graduates 0.37 0.38 0.93Some college 0.35 0.35 0.93Four-year college graduate 0.22 0.23 0.89
Smoking Age 30 0.38 0.39 0.95High school dropouts 0.67 0.66 0.89GED Recipients 0.63 0.63 0.95High school graduates 0.39 0.39 0.92Some college 0.34 0.34 0.98Four-year college graduate 0.15 0.14 0.97
Heavy Drinking Age 30 0.23 0.23 0.98High school dropouts 0.27 0.28 0.97GED Recipients 0.23 0.23 0.96High school graduates 0.25 0.25 0.90Some college 0.24 0.23 0.96Four-year college graduate 0.18 0.18 0.99
Notes: The simulated data (Model) contains one million observations generated from the Model’sestimates. The actual data (Actual) contains 2242 observations from the NLSY79 sample of Males.(a) Goodness of fit is tested using a χ2 test where the Null Hypothesis is Model=Data.
34
Table 30: Goodness of Fit - Health Outcomes
Average Std Dev p-valuea
Actual Model Actual ModelSF12 Physical Health 0.07 0.07 0.89 0.89 0.00
High school dropouts -0.29 -0.28 1.21 1.21 0.00GED Recipients -0.17 -0.16 1.08 1.09 0.00High school graduates 0.05 0.05 0.84 0.84 0.00Some college 0.13 0.14 0.87 0.87 0.00Four-year college graduate 0.36 0.36 0.49 0.49 0.00
SF12 Mental Health 0.14 0.14 0.91 0.92 0.00High school dropouts -0.10 -0.08 1.21 1.22 0.00GED Recipients 0.02 0.00 1.09 1.09 0.00High school graduates 0.19 0.19 0.83 0.83 0.00Some college 0.23 0.23 0.84 0.84 0.00Four-year college graduate 0.18 0.21 0.79 0.80 0.00
CESD Depression 0.13 0.13 0.92 0.92 0.00High school dropouts -0.36 -0.33 1.29 1.30 0.00GED Recipients -0.02 -0.01 1.05 1.06 0.00High school graduates 0.13 0.13 0.87 0.87 0.00Some college 0.24 0.22 0.76 0.77 0.00Four-year college graduate 0.38 0.39 0.65 0.65 0.00
Pearlin Self Mastery 0.05 0.05 0.99 0.98 0.00High school dropouts -0.46 -0.45 0.91 0.91 0.00GED Recipients -0.04 -0.01 0.98 0.98 0.00High school graduates -0.04 -0.03 0.95 0.95 0.00Some college 0.24 0.24 1.00 0.99 0.00Four-year college graduate 0.40 0.40 0.92 0.92 0.00
Rosenberg Self-Esteem 0.06 0.06 0.99 0.98 0.00High school dropouts -0.32 -0.31 0.99 1.01 0.00GED Recipients -0.12 -0.13 1.04 1.05 0.01High school graduates -0.06 -0.05 0.93 0.92 0.00Some college 0.24 0.26 0.95 0.95 0.00Four-year college graduate 0.40 0.41 0.95 0.94 0.00
Notes: The simulated data (Model) contains one million observations generated from the Model’sestimates. The actual data (Actual) contains 2242 observations from the NLSY79 sample of Males.(a) The equality of the Model and Data distributions are tested using a two-sample Kolmogorov-Smirnov test.
35
Table 31: Goodness of Fit - Continuous Labor Market Outcomes
Average Std Dev p-valuea
Actual Model Actual Model
Log Wages (30) 2.61 2.60 0.48 0.47 0.15High school dropouts 2.29 2.28 0.37 0.37 0.48GED Recipients 2.43 2.43 0.48 0.47 0.43High school graduates 2.53 2.53 0.43 0.43 0.92Some college 2.67 2.67 0.45 0.45 0.90Four-year college graduate 2.93 2.90 0.43 0.43 0.04
PVLog Wages (30) 12.32 12.31 0.63 0.62 0.00High school dropouts 11.79 11.78 0.61 0.66 0.07GED Recipients 11.95 11.98 0.68 0.67 0.04High school graduates 12.27 12.28 0.52 0.52 0.02Some college 12.42 12.43 0.51 0.50 0.03Four-year college graduate 12.76 12.74 0.52 0.52 0.30
Notes: The simulated data (Model) contains one million observations generated from the Model’sestimates. The actual data (Actual) contains 2242 observations from the NLSY79 sample of Males.(a) The equality of the Model and Data distributions are tested using a two-sample Kolmogorov-Smirnov test.
D Evaluating the Number of Factors
D.1 Exploratory Factor Analysis
Tables 32 and 33 provide results from exploratory factor analysis on our measurement system.
All measurer are adjusted for the control variables listed in the table footnotes. As shown in
Table 32, principle component factor analysis finds three factors using a scree plot test. As an
alternative more robust, we also implement Horn’s test (Horn, 1965). As shown in Table 33,
Horn’s test finds 2 factors for principle component factor analysis. These exploratory results
support the use of two factors, but with a marginal support for three factors from the scree plot
test. In the next section, various three-factor models are estimated as a robustness check.
36
Table 32: Testing the Number of Factors: Results from Eigenvalue or Scree Plot Test
Factor Eigenvalue Difference Proportion Cumulative1 4.160 1.902 0.378 0.3782 2.257 1.251 0.205 0.5833 1.005 0.253 0.091 0.6744 0.752 0.040 0.068 0.7435 0.712 0.169 0.064 0.8086 0.542 0.071 0.049 0.8577 0.471 0.119 0.042 0.9008 0.352 0.084 0.032 0.9329 0.268 0.007 0.024 0.95610 0.260 0.044 0.023 0.98011 0.215 0.019 1.000
Notes: Results are from exploratory principle component factor analysis. Criterion: retain factors with
eigenvalues greater than 1. Measures include six subtests of ASVAB, GPA in 9th grade in math, language,
social studies, and science, and early reckless behavior. All measures are adjusted for race, parent’s education,
family income, urban status and region in 1980, age in 1980, and age squared.
Table 33: Testing the Number of Factors: Results from Horn’s Test
Factor Adjusted-Eigen Eigenvalue1 3.061 4.1602 1.175 2.2573 -.064 1.0054 -.283 .7525 -.303 .7126 -.457 .5427 -.513 .4718 -.619 .3529 -.670 .26810 -.664 .26011 -.660 .215
Notes: Results are from Horn’s parallel analysis on principle-component factor analysis. Criterion: retain
adjusted factors > 0 Measures include six subtests of ASVAB, GPA in 9th grade in math, language, social
studies, and science, and early reckless behavior. All measures are adjusted for race, parent’s education,
family income, urban status and region in 1980, age in 1980, and age squared.
37
D.2 Robustness check: three factor models
Given the weak evidence for a third factor in exploratory factor analysis, models with an
additional factor are considered as robustness tests. Three different three-factor models are
considered. Since estimating three-factors as opposed to two is computationally costly, we simply
the model along two dimensions. First, we assume a normal distribution for the factors rather
than a mixture of normals. Second, the factor distributions are integrated using 8 quadrature
points rather than 16 points. Estimates of the default model with these simplifications yields
small changes in the estimates of the different treatment effects compared to the full model.
Three alternative models are considered:
1. Additional socio-emotional factor: We include a third factor that loads only on the
early-adverse behavior outcomes (smoking – cigarettes and marijuana –, drinking, and
intercourse before age 15). The measurement system consists of the ASVAB tests (COG),
school grades (COG,SE1) and early-adverse behaviors (COG, SE1, SE2). All three factors
are included in the education models and outcomes.
2. Additional cognitive factor: Include a third factor that loads only on the verbal sub-
tests of the ASVAB. The measurement system consists of the two verbal ASVAB tests
(COG1), the four numerical ASVAB tests (COG1, COG2) and the school grades (COG1,
COG2, SE). All three factores are included in the education models and outcomes.
3. Additional factor in outcomes Include third factor in labor market outcomes. In this
case, the model is estimated in one stage. The measurement system consists of ASVAB
tests (COG), school grades (COG, SE) and labor market outcomes (COG, SE, FAC3). The
COG and SE factors are included in the education model, but not the labor market factor.
The estimates of the treatment effects for each alternate model is compared to the two
factor model presented in the paper. To understand whether the size of the differences are
important, t-statistics are formed where the differences are divided by the standard error. We use
standard error as a metric since differences that are less than a standard error are statistically
indistinguishable from our results.
Section D.2.1 compares the default simplified model to the full model. The largest deviation
is 0.85 standard errors from the full model, where more than 85% of the estimates are less than a
third of a standard error away from the full model. The simplified model is a good approximation
of the full model. Sections D.2.2, D.2.3 and D.2.4 compare the alternate three-factor models.
Most of the differences are less than one standard error away from the default 2-factor model.
38
The largest difference is 1.19 standard errors away from the default measurement. We conclude
that if there are any differences in the treatment effects from including a third factor, they are
not observable given the standard errors on the measurements.
D.2.1 Comparing simplified model to full model
The TE results for the simplified two-factor default model compared to the full two-factor model
are:
Table 34: Node-specific TE comparison: wage
ATE* ATE TT TUT AMTE
HS Dropout-HS Graduate 0.14 0.14 0.13 0.13 -0.27HS Dropout-GED -0.18 0.04 -0.04 0.10 0.22HS Graduate-Enroll College 0.18 0.27 0.41 0.05 0.85Enroll College-4-yr College Grad 0.34 -0.10 -0.23 0.13 -0.23
Table 35: Node-specific TE comparison: PVwage
ATE* ATE TT TUT AMTE
HS Dropout-HS Graduate 0.20 0.20 0.19 0.11 0.07HS Dropout-GED 0.12 0.15 0.25 0.07 0.17HS Graduate-Enroll College 0.35 0.46 0.41 0.38 0.52Enroll College-4-yr College Grad 0.27 0.02 -0.04 0.09 0.01
Table 36: Node-specific TE comparison: whitecollar
ATE* ATE TT TUT AMTE
HS Dropout-HS Graduate 0.08 0.08 0.05 0.27 0.12HS Dropout-GED 0.16 0.22 0.23 0.16 0.10HS Graduate-Enroll College 0.17 0.13 0.13 0.14 0.31Enroll College-4-yr College Grad 0.06 0.01 0.00 0.02 0.04
39
Table 37: Final schooling TE comparison: wage
ATE* ATE TT TUT
HS Dropout-GED -0.19 0.04 -0.04 0.10HS Dropout-HS Graduate -0.23 -0.15 -0.22 0.20HS Dropout-Some College -0.10 -0.08 -0.05 -0.07HS Dropout-4-yr College Grad 0.11 0.06 -0.14 0.53
Table 38: Final schooling TE comparison: PVwage
ATE* ATE TT TUT
HS Dropout-GED 0.12 0.15 0.25 0.07HS Dropout-HS Graduate 0.22 0.11 0.10 0.08HS Dropout-Some College 0.36 0.41 0.47 0.02HS Dropout-4-yr College Grad 0.50 0.51 0.33 0.45
Table 39: Final schooling TE comparison: whitecollar
ATE* ATE TT TUT
HS Dropout-GED 0.16 0.22 0.23 0.16HS Dropout-HS Graduate 0.17 0.17 0.14 0.19HS Dropout-Some College 0.25 0.25 0.22 0.19HS Dropout-4-yr College Grad 0.26 0.25 0.25 0.21
40
D.2.2 Three-factor model with two cognitive factors
The TE results for the three-factor model with two cognitive factors:
Table 40: Node-specific TE comparison: wage
ATE* ATE TT TUT AMTE
HS Dropout-HS Graduate -0.12 -0.12 -0.08 -0.31 0.28HS Dropout-GED 0.42 -0.09 0.35 -0.41 -0.08HS Graduate-Enroll College 0.16 0.02 -0.35 0.43 -0.47Enroll College-4-yr College Grad 0.17 0.30 0.27 0.30 0.19
Table 41: Node-specific TE comparison: PVwage
ATE* ATE TT TUT AMTE
HS Dropout-HS Graduate 0.35 0.35 0.34 0.16 0.01HS Dropout-GED 0.70 0.44 0.75 0.20 0.50HS Graduate-Enroll College -0.09 -0.20 -0.36 0.05 0.07Enroll College-4-yr College Grad 0.35 0.35 0.10 0.49 0.61
Table 42: Node-specific TE comparison: whitecollar
ATE* ATE TT TUT AMTE
HS Dropout-HS Graduate -0.29 -0.29 -0.24 -0.36 -0.28HS Dropout-GED -0.48 -0.55 -0.67 -0.29 -0.43HS Graduate-Enroll College -0.03 -0.03 -0.16 0.08 0.02Enroll College-4-yr College Grad 0.46 0.24 -0.03 0.45 0.00
41
Table 43: Final schooling TE comparison: wage
ATE* ATE TT TUT
HS Dropout-GED 0.43 -0.09 0.35 -0.41HS Dropout-HS Graduate 0.66 0.46 0.62 -0.41HS Dropout-Some College 0.58 0.59 0.58 0.24HS Dropout-4-yr College Grad 0.64 0.53 0.56 0.11
Table 44: Final schooling TE comparison: PVwage
ATE* ATE TT TUT
HS Dropout-GED 0.70 0.44 0.75 0.20HS Dropout-HS Graduate 1.04 1.15 1.09 0.19HS Dropout-Some College 0.96 0.96 0.91 0.35HS Dropout-4-yr College Grad 1.09 0.95 0.81 0.44
Table 45: Final schooling TE comparison: whitecollar
ATE* ATE TT TUT
HS Dropout-GED -0.48 -0.55 -0.67 -0.29HS Dropout-HS Graduate -0.65 -0.66 -0.61 -0.27HS Dropout-Some College -0.71 -0.63 -0.69 -0.11HS Dropout-4-yr College Grad -0.22 -0.50 -0.78 0.42
42
D.2.3 Three-factor model with two socio-emotional factors
The TE results for the three-factor model with two socio-emotional factors:
Table 46: Node-specific TE comparison: wage
ATE* ATE TT TUT AMTE
HS Dropout-HS Graduate -0.16 -0.16 -0.21 0.36 0.86HS Dropout-GED 0.23 0.03 -0.02 0.06 0.06HS Graduate-Enroll College 0.13 0.11 0.08 0.11 -0.39Enroll College-4-yr College Grad -0.17 0.20 0.41 -0.16 0.35
Table 47: Node-specific TE comparison: PVwage
ATE* ATE TT TUT AMTE
HS Dropout-HS Graduate -0.81 -0.81 -0.71 -1.02 -1.13HS Dropout-GED 0.60 0.02 0.05 0.00 0.00HS Graduate-Enroll College 0.05 -0.17 -0.23 -0.05 -0.11Enroll College-4-yr College Grad -0.24 -0.34 -0.30 -0.27 -0.28
Table 48: Node-specific TE comparison: whitecollar
ATE* ATE TT TUT AMTE
HS Dropout-HS Graduate -0.32 -0.32 -0.27 -0.39 -0.15HS Dropout-GED 0.14 0.00 0.01 0.00 0.11HS Graduate-Enroll College 0.05 -0.10 -0.17 -0.02 -0.12Enroll College-4-yr College Grad -0.20 0.00 0.16 -0.19 -0.03
43
Table 49: Final schooling TE comparison: wage
ATE* ATE TT TUT
HS Dropout-GED 0.24 0.03 -0.02 0.06HS Dropout-HS Graduate 0.07 0.11 0.08 0.19HS Dropout-Some College 0.04 0.12 -0.00 0.23HS Dropout-4-yr College Grad -0.06 -0.04 0.06 -0.23
Table 50: Final schooling TE comparison: PVwage
ATE* ATE TT TUT
HS Dropout-GED 0.60 0.02 0.05 0.00HS Dropout-HS Graduate 0.03 -0.09 0.07 -0.87HS Dropout-Some College 0.11 -0.00 0.11 -0.22HS Dropout-4-yr College Grad -0.05 0.02 0.14 -0.28
Table 51: Final schooling TE comparison: whitecollar
ATE* ATE TT TUT
HS Dropout-GED 0.14 0.00 0.01 0.00HS Dropout-HS Graduate -0.07 -0.15 -0.09 -0.37HS Dropout-Some College -0.07 0.02 -0.04 0.13HS Dropout-4-yr College Grad -0.21 -0.17 -0.08 -0.24
44
D.2.4 Three-factor model with additional factor in outcomes
The TE results for the three-factor model with a factor estimated from the outcomes:
Table 52: Node-specific TE comparison: wage
ATE* ATE TT TUT AMTE
HS Dropout-HS Graduate 0.36 0.36 0.23 1.06 1.19HS Dropout-GED -0.94 -0.80 -0.86 -0.64 -1.01HS Graduate-Enroll College 0.64 0.69 0.45 0.80 0.36Enroll College-4-yr College Grad -0.57 -0.61 -0.49 -0.73 -0.29
Table 53: Node-specific TE comparison: PVwage
ATE* ATE TT TUT AMTE
HS Dropout-HS Graduate 0.28 0.28 0.29 0.08 0.04HS Dropout-GED 0.06 -0.06 -0.10 -0.03 -0.11HS Graduate-Enroll College 0.44 0.46 0.33 0.47 0.50Enroll College-4-yr College Grad -0.92 -0.75 -0.48 -0.81 -0.67
Table 54: Node-specific TE comparison: whitecollar
ATE* ATE TT TUT AMTE
HS Dropout-HS Graduate 0.02 0.02 0.02 0.03 0.12HS Dropout-GED -0.04 -0.36 -0.22 -0.46 -0.41HS Graduate-Enroll College -0.01 0.01 0.03 -0.02 -0.07Enroll College-4-yr College Grad -0.21 -0.14 -0.08 -0.15 -0.17
45
Table 55: Final schooling TE comparison: wage
ATE* ATE TT TUT
HS Dropout-GED -0.95 -0.80 -0.86 -0.64HS Dropout-HS Graduate -0.85 -0.48 -0.70 0.69HS Dropout-Some College -0.35 -0.10 -0.54 0.62HS Dropout-4-yr College Grad -0.67 -0.86 -1.06 0.25
Table 56: Final schooling TE comparison: PVwage
ATE* ATE TT TUT
HS Dropout-GED 0.06 -0.06 -0.10 -0.03HS Dropout-HS Graduate 0.21 0.17 0.19 -0.10HS Dropout-Some College 0.53 0.52 0.46 0.33HS Dropout-4-yr College Grad -0.11 0.02 0.27 -0.57
Table 57: Final schooling TE comparison: whitecollar
ATE* ATE TT TUT
HS Dropout-GED -0.04 -0.36 -0.22 -0.46HS Dropout-HS Graduate 0.04 0.00 0.03 -0.16HS Dropout-Some College 0.07 0.04 0.10 -0.09HS Dropout-4-yr College Grad -0.11 -0.07 0.06 -0.29
46
E Treatment Effects: Comparison of Outcomes for differ-
ent final schooling levels
We now compare the outcomes from a particular schooling level s with those associated with the
HS dropout status. In other words, we use HS dropout as our baseline comparison group. For
each the outcomes, the numbers under the column “Observed” displays the observed differences in
the data. The column “ATE” displays the average treatment effect obtained from the comparison
of the outcomes associated with a particular schooling level s relative to the HS dropout status.
ATE is computed using the overall population. The column “TT” displays the average treatment
effects associated with a particular schooling level s relative to the HS dropout status but
computed from those individuals selecting s as their final schooling decision. Finally, the column
“TUT” displays the average treatment effects associated with a particular schooling level s relative
to the HS dropout status but computed only for those individuals selecting “HS dropout” as
their final schooling decision.
Table 58: The Effects of Education on Log Wages, by Final Schooling Level using High SchoolDropouts as Baseline
Observed ATE† ATE TT TUT
GED vs. HS Dropout 0.17 0.11 0.04 0.06 0.03( 0.08) ( 0.05) ( 0.05) ( 0.05)
HS Graduate vs. HS Dropout 0.28 0.13* 0.12** 0.12* 0.11**( 0.06) ( 0.04) ( 0.05) ( 0.04)
Some College vs. HS Dropout 0.54 0.21** 0.21** 0.21** 0.21**( 0.06) ( 0.05) ( 0.07) ( 0.07)
Four Year College Degree vs. HS Dropout 0.90 0.26** 0.27** 0.34** 0.15**( 0.07) ( 0.07) ( 0.10) ( 0.07)
Notes: Standard errors (in parenthesis) and significance levels (*=p< 0.05, **=p< 0.01) are calculatedusing 200 bootstrap samples. Each row compares the outcomes from a particular schooling level j and theHS dropout status. The column “Observed” displays the observed differences in the data. The column“ATE” displays the average treatment effect obtained from the comparison of the outcomes associated with aparticular schooling level j relative to the HS dropout status. ATE† is evaluated over the whole population,whereas ATE is evaluated for everyone who reaches a certain decision node. The column “TT” displays theaverage treatment effects associated with a particular schooling level j relative to the HS dropout statusbut computed from those individuals selecting j as their final schooling decision. Finally, the column “TUT”displays the average treatment effects associated with a particular schooling level j relative to the HS dropoutstatus but computed only for those individuals selecting “HS dropout” as their final schooling decision.
47
Table 59: The Effects of Education on Log PV of wages, by Final Schooling Level using HighSchool Dropouts as Baseline
Observed ATE† ATE TT TUT
GED vs. HS Dropout 0.18 -0.21* -0.11 -0.15* -0.08( 0.10) ( 0.06) ( 0.06) ( 0.07)
HS Graduate vs. HS Dropout 0.62 -0.03 0.07 -0.00 0.30**( 0.08) ( 0.06) ( 0.08) ( 0.05)
Some College vs. HS Dropout 0.76 0.04 0.16* -0.05 0.46**( 0.08) ( 0.07) ( 0.10) ( 0.08)
Four Year College Degree vs. HS Dropout 1.16 0.10 0.07 -0.08 0.33**( 0.09) ( 0.09) ( 0.13) ( 0.10)
Notes: Standard errors (in parenthesis) and significance levels (*=p< 0.05, **=p< 0.01) are calculatedusing 200 bootstrap samples. Each row compares the outcomes from a particular schooling level j and theHS dropout status. The column “Observed” displays the observed differences in the data. The column“ATE” displays the average treatment effect obtained from the comparison of the outcomes associated with aparticular schooling level j relative to the HS dropout status. ATE† is evaluated over the whole population,whereas ATE is evaluated for everyone who reaches a certain decision node. The column “TT” displays theaverage treatment effects associated with a particular schooling level j relative to the HS dropout statusbut computed from those individuals selecting j as their final schooling decision. Finally, the column “TUT”displays the average treatment effects associated with a particular schooling level j relative to the HS dropoutstatus but computed only for those individuals selecting “HS dropout” as their final schooling decision.
Table 60: The Effects of Education on Physical Health (PCS-12), by Final Schooling Level usingHigh School Dropouts as Baseline
Observed ATE† ATE TT TUT
GED vs. HS Dropout 0.19 -0.19 -0.03 -0.07 0.00( 0.24) ( 0.13) ( 0.16) ( 0.14)
HS Graduate vs. HS Dropout 0.39 0.08 0.14 0.11 0.23*( 0.21) ( 0.15) ( 0.19) ( 0.12)
Some College vs. HS Dropout 0.50 0.10 0.13 0.08 0.19( 0.22) ( 0.17) ( 0.25) ( 0.16)
Four Year College Degree vs. HS Dropout 0.68 0.31 0.28 0.14 0.52**( 0.21) ( 0.23) ( 0.35) ( 0.11)
Notes: Standard errors (in parenthesis) and significance levels (*=p< 0.05, **=p< 0.01) are calculatedusing 200 bootstrap samples. Each row compares the outcomes from a particular schooling level j and theHS dropout status. The column “Observed” displays the observed differences in the data. The column“ATE” displays the average treatment effect obtained from the comparison of the outcomes associated with aparticular schooling level j relative to the HS dropout status. ATE† is evaluated over the whole population,whereas ATE is evaluated for everyone who reaches a certain decision node. The column “TT” displays theaverage treatment effects associated with a particular schooling level j relative to the HS dropout statusbut computed from those individuals selecting j as their final schooling decision. Finally, the column “TUT”displays the average treatment effects associated with a particular schooling level j relative to the HS dropoutstatus but computed only for those individuals selecting “HS dropout” as their final schooling decision.
48
Table 61: The Effects of Education on Mental Health (MCS-12), by Final Schooling Level usingHigh School Dropouts as Baseline
Observed ATE† ATE TT TUT
GED vs. HS Dropout 0.27 -0.48* -0.03 -0.08 0.02( 0.23) ( 0.12) ( 0.14) ( 0.14)
HS Graduate vs. HS Dropout 0.68 -0.18 -0.08 -0.15 0.17( 0.20) ( 0.15) ( 0.18) ( 0.10)
Some College vs. HS Dropout 0.76 -0.16 -0.07 -0.24 0.19( 0.20) ( 0.15) ( 0.24) ( 0.14)
Four Year College Degree vs. HS Dropout 1.05 -0.14 -0.21 -0.50 0.32( 0.21) ( 0.23) ( 0.38) ( 0.19)
Notes: Standard errors (in parenthesis) and significance levels (*=p< 0.05, **=p< 0.01) are calculatedusing 200 bootstrap samples. Each row compares the outcomes from a particular schooling level j and theHS dropout status. The column “Observed” displays the observed differences in the data. The column“ATE” displays the average treatment effect obtained from the comparison of the outcomes associated with aparticular schooling level j relative to the HS dropout status. ATE† is evaluated over the whole population,whereas ATE is evaluated for everyone who reaches a certain decision node. The column “TT” displays theaverage treatment effects associated with a particular schooling level j relative to the HS dropout statusbut computed from those individuals selecting j as their final schooling decision. Finally, the column “TUT”displays the average treatment effects associated with a particular schooling level j relative to the HS dropoutstatus but computed only for those individuals selecting “HS dropout” as their final schooling decision.
Table 62: The Effects of Education on Depression (CES-D), by Final Schooling Level using HighSchool Dropouts as Baseline
Observed ATE† ATE TT TUT
GED vs. HS Dropout 0.11 -0.19 0.12 0.03 0.20( 0.25) ( 0.12) ( 0.13) ( 0.14)
HS Graduate vs. HS Dropout 0.31 -0.07 0.04 -0.05 0.33**( 0.21) ( 0.15) ( 0.20) ( 0.11)
Some College vs. HS Dropout 0.33 -0.01 0.10 -0.10 0.40**( 0.20) ( 0.16) ( 0.25) ( 0.13)
Four Year College Degree vs. HS Dropout 0.29 0.18 0.12 -0.23 0.75**( 0.22) ( 0.26) ( 0.39) ( 0.16)
Notes: Standard errors (in parenthesis) and significance levels (*=p< 0.05, **=p< 0.01) are calculatedusing 200 bootstrap samples. Each row compares the outcomes from a particular schooling level j and theHS dropout status. The column “Observed” displays the observed differences in the data. The column“ATE” displays the average treatment effect obtained from the comparison of the outcomes associated with aparticular schooling level j relative to the HS dropout status. ATE† is evaluated over the whole population,whereas ATE is evaluated for everyone who reaches a certain decision node. The column “TT” displays theaverage treatment effects associated with a particular schooling level j relative to the HS dropout statusbut computed from those individuals selecting j as their final schooling decision. Finally, the column “TUT”displays the average treatment effects associated with a particular schooling level j relative to the HS dropoutstatus but computed only for those individuals selecting “HS dropout” as their final schooling decision.
49
Table 63: The Effects of Education on Pearlin, by Final Schooling Level using High SchoolDropouts as Baseline
Observed ATE† ATE TT TUT
GED vs. HS Dropout -0.27 0.59** 0.21* 0.28* 0.14( 0.23) ( 0.11) ( 0.13) ( 0.12)
HS Graduate vs. HS Dropout -0.44 0.25 0.24 0.25 0.20**( 0.21) ( 0.16) ( 0.19) ( 0.09)
Some College vs. HS Dropout -0.59 0.46** 0.49** 0.43* 0.57**( 0.23) ( 0.18) ( 0.26) ( 0.15)
Four Year College Degree vs. HS Dropout -0.68 0.65** 0.61** 0.47 0.88**( 0.24) ( 0.25) ( 0.35) ( 0.17)
Notes: Standard errors (in parenthesis) and significance levels (*=p< 0.05, **=p< 0.01) are calculatedusing 200 bootstrap samples. Each row compares the outcomes from a particular schooling level j and theHS dropout status. The column “Observed” displays the observed differences in the data. The column“ATE” displays the average treatment effect obtained from the comparison of the outcomes associated with aparticular schooling level j relative to the HS dropout status. ATE† is evaluated over the whole population,whereas ATE is evaluated for everyone who reaches a certain decision node. The column “TT” displays theaverage treatment effects associated with a particular schooling level j relative to the HS dropout statusbut computed from those individuals selecting j as their final schooling decision. Finally, the column “TUT”displays the average treatment effects associated with a particular schooling level j relative to the HS dropoutstatus but computed only for those individuals selecting “HS dropout” as their final schooling decision.
Table 64: The Effects of Education on Self-Esteem (Rosenberg), by Final Schooling Level usingHigh School Dropouts as Baseline
Observed ATE† ATE TT TUT
GED vs. HS Dropout 0.42 -0.19 -0.07 -0.14 -0.01( 0.25) ( 0.12) ( 0.15) ( 0.15)
HS Graduate vs. HS Dropout 0.41 -0.15 -0.11 -0.13 -0.05( 0.22) ( 0.15) ( 0.19) ( 0.09)
Some College vs. HS Dropout 0.76 0.05 0.13 -0.02 0.35*( 0.21) ( 0.18) ( 0.26) ( 0.17)
Four Year College Degree vs. HS Dropout 0.91 0.26 0.18 -0.10 0.71**( 0.24) ( 0.27) ( 0.37) ( 0.21)
Notes: Standard errors (in parenthesis) and significance levels (*=p< 0.05, **=p< 0.01) are calculatedusing 200 bootstrap samples. Each row compares the outcomes from a particular schooling level j and theHS dropout status. The column “Observed” displays the observed differences in the data. The column“ATE” displays the average treatment effect obtained from the comparison of the outcomes associated with aparticular schooling level j relative to the HS dropout status. ATE† is evaluated over the whole population,whereas ATE is evaluated for everyone who reaches a certain decision node. The column “TT” displays theaverage treatment effects associated with a particular schooling level j relative to the HS dropout statusbut computed from those individuals selecting j as their final schooling decision. Finally, the column “TUT”displays the average treatment effects associated with a particular schooling level j relative to the HS dropoutstatus but computed only for those individuals selecting “HS dropout” as their final schooling decision.
50
Table 65: The Effects of Education on Participation, by Final Schooling Level using High SchoolDropouts as Baseline
Observed ATE† ATE TT TUT
GED vs. HS Dropout 0.25 -0.06 -0.06 -0.07* -0.06( 0.04) ( 0.03) ( 0.03) ( 0.05)
HS Graduate vs. HS Dropout 0.34 0.03* 0.04* 0.03 0.09**( 0.02) ( 0.03) ( 0.03) ( 0.03)
Some College vs. HS Dropout 0.64 0.02 0.04 0.00 0.09*( 0.03) ( 0.03) ( 0.03) ( 0.04)
Four Year College Degree vs. HS Dropout 0.80 0.00 0.01 -0.01 0.04( 0.04) ( 0.04) ( 0.02) ( 0.09)
Notes: Standard errors (in parenthesis) and significance levels (*=p< 0.05, **=p< 0.01) are calculatedusing 200 bootstrap samples. Each row compares the outcomes from a particular schooling level j and theHS dropout status. The column “Observed” displays the observed differences in the data. The column“ATE” displays the average treatment effect obtained from the comparison of the outcomes associated with aparticular schooling level j relative to the HS dropout status. ATE† is evaluated over the whole population,whereas ATE is evaluated for everyone who reaches a certain decision node. The column “TT” displays theaverage treatment effects associated with a particular schooling level j relative to the HS dropout statusbut computed from those individuals selecting j as their final schooling decision. Finally, the column “TUT”displays the average treatment effects associated with a particular schooling level j relative to the HS dropoutstatus but computed only for those individuals selecting “HS dropout” as their final schooling decision.
51
Table 66: The Effects of Education on White Collar, by Final Schooling Level using High SchoolDropouts as Baseline
Observed ATE† ATE TT TUT
GED vs. HS Dropout 0.11 0.14 0.03 0.06 0.01( 0.09) ( 0.04) ( 0.05) ( 0.04)
HS Graduate vs. HS Dropout 0.22 0.14* 0.10* 0.12* 0.04( 0.07) ( 0.05) ( 0.06) ( 0.03)
Some College vs. HS Dropout 0.25 0.27** 0.25** 0.29** 0.18**( 0.07) ( 0.06) ( 0.08) ( 0.06)
Four Year College Degree vs. HS Dropout 0.26 0.41** 0.46** 0.65** 0.10( 0.08) ( 0.09) ( 0.11) ( 0.10)
Notes: Standard errors (in parenthesis) and significance levels (*=p< 0.05, **=p< 0.01) are calculatedusing 200 bootstrap samples. Each row compares the outcomes from a particular schooling level j and theHS dropout status. The column “Observed” displays the observed differences in the data. The column“ATE” displays the average treatment effect obtained from the comparison of the outcomes associated with aparticular schooling level j relative to the HS dropout status. ATE† is evaluated over the whole population,whereas ATE is evaluated for everyone who reaches a certain decision node. The column “TT” displays theaverage treatment effects associated with a particular schooling level j relative to the HS dropout statusbut computed from those individuals selecting j as their final schooling decision. Finally, the column “TUT”displays the average treatment effects associated with a particular schooling level j relative to the HS dropoutstatus but computed only for those individuals selecting “HS dropout” as their final schooling decision.
Table 67: The Effects of Education on Obesity, by Final Schooling Level using High SchoolDropouts as Baseline
Observed ATE† ATE TT TUT
GED vs. HS Dropout 0.05 0.12 0.02 0.05 -0.01( 0.09) ( 0.05) ( 0.06) ( 0.06)
HS Graduate vs. HS Dropout 0.11 0.15* 0.12* 0.14* 0.08*( 0.08) ( 0.06) ( 0.07) ( 0.05)
Some College vs. HS Dropout 0.21 0.14* 0.15* 0.15 0.15*( 0.08) ( 0.06) ( 0.09) ( 0.08)
Four Year College Degree vs. HS Dropout 0.37 0.08 0.07 0.06 0.11( 0.08) ( 0.08) ( 0.11) ( 0.10)
Notes: Standard errors (in parenthesis) and significance levels (*=p< 0.05, **=p< 0.01) are calculatedusing 200 bootstrap samples. Each row compares the outcomes from a particular schooling level j and theHS dropout status. The column “Observed” displays the observed differences in the data. The column“ATE” displays the average treatment effect obtained from the comparison of the outcomes associated with aparticular schooling level j relative to the HS dropout status. ATE† is evaluated over the whole population,whereas ATE is evaluated for everyone who reaches a certain decision node. The column “TT” displays theaverage treatment effects associated with a particular schooling level j relative to the HS dropout statusbut computed from those individuals selecting j as their final schooling decision. Finally, the column “TUT”displays the average treatment effects associated with a particular schooling level j relative to the HS dropoutstatus but computed only for those individuals selecting “HS dropout” as their final schooling decision.
52
Table 68: The Effects of Education on Daily Smoking, by Final Schooling Level using HighSchool Dropouts as Baseline
Observed ATE† ATE TT TUT
GED vs. HS Dropout -0.03 0.05 0.01 0.01 0.01( 0.10) ( 0.05) ( 0.06) ( 0.05)
HS Graduate vs. HS Dropout 0.04 -0.14 -0.18** -0.15* -0.25**( 0.08) ( 0.06) ( 0.08) ( 0.05)
Some College vs. HS Dropout -0.01 -0.20* -0.21** -0.17 -0.27**( 0.09) ( 0.07) ( 0.10) ( 0.07)
Four Year College Degree vs. HS Dropout -0.10 -0.35** -0.34** -0.32* -0.38**( 0.09) ( 0.10) ( 0.14) ( 0.09)
Notes: Standard errors (in parenthesis) and significance levels (*=p< 0.05, **=p< 0.01) are calculatedusing 200 bootstrap samples. Each row compares the outcomes from a particular schooling level j and theHS dropout status. The column “Observed” displays the observed differences in the data. The column“ATE” displays the average treatment effect obtained from the comparison of the outcomes associated with aparticular schooling level j relative to the HS dropout status. ATE† is evaluated over the whole population,whereas ATE is evaluated for everyone who reaches a certain decision node. The column “TT” displays theaverage treatment effects associated with a particular schooling level j relative to the HS dropout statusbut computed from those individuals selecting j as their final schooling decision. Finally, the column “TUT”displays the average treatment effects associated with a particular schooling level j relative to the HS dropoutstatus but computed only for those individuals selecting “HS dropout” as their final schooling decision.
Table 69: The Effects of Education on Heavy Drinker, by Final Schooling Level using HighSchool Dropouts as Baseline
Observed ATE† ATE TT TUT
GED vs. HS Dropout -0.05 -0.03 -0.07 -0.07 -0.08( 0.12) ( 0.06) ( 0.08) ( 0.05)
HS Graduate vs. HS Dropout -0.22 -0.08 -0.07 -0.08 -0.02( 0.09) ( 0.07) ( 0.08) ( 0.05)
Some College vs. HS Dropout -0.30 -0.10 -0.07 -0.13 0.00( 0.10) ( 0.08) ( 0.11) ( 0.08)
Four Year College Degree vs. HS Dropout -0.43 -0.07 -0.09 -0.22 0.14( 0.11) ( 0.13) ( 0.17) ( 0.12)
Notes: Standard errors (in parenthesis) and significance levels (*=p< 0.05, **=p< 0.01) are calculatedusing 200 bootstrap samples. Each row compares the outcomes from a particular schooling level j and theHS dropout status. The column “Observed” displays the observed differences in the data. The column“ATE” displays the average treatment effect obtained from the comparison of the outcomes associated with aparticular schooling level j relative to the HS dropout status. ATE† is evaluated over the whole population,whereas ATE is evaluated for everyone who reaches a certain decision node. The column “TT” displays theaverage treatment effects associated with a particular schooling level j relative to the HS dropout statusbut computed from those individuals selecting j as their final schooling decision. Finally, the column “TUT”displays the average treatment effects associated with a particular schooling level j relative to the HS dropoutstatus but computed only for those individuals selecting “HS dropout” as their final schooling decision.
53
F Treatment Effects: Pairwise Comparing by Decision
Node
After discussing the treatment effects obtained for the analysis of final schooling levels, we now
analyze the treatment effects by decision node.
Each table presents the average effects of education on the outcome of interest. The effects
are presented in the different panels in each table and they are defined as the differences in
the outcome associated with two schooling levels (not necessarily final or terminal schooling
levels). For each panel, let Y0 and Y1 denotes the outcomes associated with schooling levels 0
and 1, respectively. Let Y0 and Y1 denotes the outcomes associated with schooling levels 0 and
1, respectively. Importantly, each schooling level might provide the option to pursuing higher
schooling levels.
For example, the choice to graduate from high school opens up the possibility of enrolling in
college. Let s indicate the level of final schooling, where 0 corresponds to dropping out of high
school, 1 to graduating high school, 2 to attaining a GED, 3 to attaining some college, and 4 for
graduating college. Then let D0,1 represent the decision to graduate from high school and D0,2
represent the decision to get the GED once an individual has chosen to drop out (D0,1 = 0). The
decision-specific treatment effect from graduating from high school (D0,1 = 1) can be written as:
TE(Y |D0,1 = 1) = E(Y |D0,1 = 1)− E(Y |D0,1 = 0)
= Pr(s = 1|D0,1 = 1)× Y1 + Pr(s = 3|D0,1 = 1)× Y3 + Pr(s = 4|D0,1 = 1)× Y4
− Pr(s = 0|D0,1 = 0)× Y0 − Pr(s = 2|D0,1 = 0)× Y2
where Pr() is the probability that an individual has a given final educational level and the
wage Y depends on the final schooling level. Of course, Pr(s = 1|D0,1 = 1) + Pr(s = 3|D0,1 =
1) + Pr(s = 4|D0,1 = 1) = 1 and Pr(s = 0|D0,1 = 0) + Pr(s = 2|D0,1 = 0) = 1.
Both outcomes and transition probabilities depend on an agent’s observable and unobservable
54
Table 70: The Effects of Education on Log Wages, by Decision Node
% ATE† ATE TT TUT AMTE
A. Dropping from HS vs. Graduating from HSAll 0.10* 0.10* 0.11* 0.10** 0.10**
( 0.05) ( 0.05) ( 0.06) ( 0.03) ( 0.03)Low Ability 0.31 0.10** 0.10** 0.10** 0.11**
( 0.04) ( 0.04) ( 0.04) ( 0.04)High Ability 0.31 0.11 0.11 0.11 0.07
( 0.10) ( 0.10) ( 0.10) ( 0.07)B. HS Dropout vs. Getting a GED
All 0.11 0.04 0.06 0.03 0.04( 0.08) ( 0.05) ( 0.05) ( 0.05) ( 0.05)
Low Ability 0.61 0.00 -0.00 0.00 -0.01( 0.06) ( 0.06) ( 0.06) ( 0.06)
High Ability 0.06 0.21 0.22* 0.22* 0.22*( 0.15) ( 0.11) ( 0.11) ( 0.11)
C. HS Graduate vs. College EnrollmentAll 0.12** 0.13** 0.14** 0.12** 0.11**
( 0.03) ( 0.03) ( 0.03) ( 0.03) ( 0.02)Low Ability 0.22 0.09* 0.09** 0.08* 0.10**
( 0.05) ( 0.04) ( 0.04) ( 0.05)High Ability 0.38 0.17** 0.17** 0.18** 0.15**
( 0.03) ( 0.03) ( 0.04) ( 0.03)D. Some College vs. 4-year college degree
All 0.05 0.11** 0.14** 0.08** 0.10**( 0.04) ( 0.04) ( 0.04) ( 0.03) ( 0.04)
Low Ability 0.14 -0.07 -0.04 -0.04 -0.05( 0.07) ( 0.05) ( 0.05) ( 0.05)
High Ability 0.51 0.18** 0.19** 0.19** 0.18**( 0.04) ( 0.05) ( 0.06) ( 0.04)
Notes: Standard errors (in parenthesis) and significance levels (*=p< 0.05, **=p< 0.01) are calculated using200 bootstrap samples. Each column presents the average effect of an educational decision. Importantly,each schooling level might provide the option to pursue higher schooling levels, while final schooling levels donot provide an option. Final schooling levels are highlighted using bold letters. ATE† represents the averageeffect for the full population, while ATE presents the average effect for those who visit the decision node.The TT column presents the average effect for those who chose a higher level of schooling (Dj,j′′ = 1), andTUT presents the average effect for those who do not choose a higher level of schooling (Dj,j′′ = 0). Finally,AMTE presents the average effect for those who are indifferent between choosing a higher level of schoolingor not. The table also presents the estimated treatment effects conditional upon endowment levels. Thehigh (low) ability group is defined as those individuals with cognitive and socio-emotional endowment above(below) the overall median. For each decision node we display the fraction of individuals with low and highability levels visiting each node.
55
Table 71: The Effects of Education on Log PV of wages, by Decision Node
% ATE† ATE TT TUT AMTE
A. Dropping from HS vs. Graduating from HSAll 0.18** 0.18** 0.15** 0.30** 0.28**
( 0.06) ( 0.06) ( 0.07) ( 0.04) ( 0.04)Low Ability 0.31 0.28** 0.28** 0.23** 0.35**
( 0.04) ( 0.04) ( 0.06) ( 0.04)High Ability 0.31 0.10 0.10 0.10 0.11
( 0.12) ( 0.12) ( 0.12) ( 0.08)B. HS Dropout vs. Getting a GED
All -0.21* -0.11 -0.15* -0.08 -0.12*( 0.10) ( 0.06) ( 0.06) ( 0.07) ( 0.06)
Low Ability 0.61 -0.19* -0.13 -0.18* -0.10( 0.08) ( 0.08) ( 0.08) ( 0.09)
High Ability 0.06 -0.22 -0.04 -0.06 0.02( 0.19) ( 0.15) ( 0.16) ( 0.17)
C. HS Graduate vs. College EnrollmentAll 0.14** 0.14** 0.14** 0.14** 0.11**
( 0.03) ( 0.03) ( 0.03) ( 0.03) ( 0.03)Low Ability 0.22 0.09* 0.06 0.00 0.08*
( 0.06) ( 0.05) ( 0.05) ( 0.05)High Ability 0.38 0.21** 0.21** 0.21** 0.20**
( 0.04) ( 0.04) ( 0.04) ( 0.04)D. Some College vs. 4-year college degree
All 0.06 0.17** 0.22** 0.11* 0.14**( 0.06) ( 0.04) ( 0.05) ( 0.05) ( 0.04)
Low Ability 0.14 -0.11 -0.01 0.04 -0.03( 0.10) ( 0.07) ( 0.07) ( 0.07)
High Ability 0.51 0.23** 0.26** 0.28** 0.22**( 0.05) ( 0.05) ( 0.05) ( 0.05)
Notes: Standard errors (in parenthesis) and significance levels (*=p< 0.05, **=p< 0.01) are calculated using200 bootstrap samples. Each column presents the average effect of an educational decision. Importantly,each schooling level might provide the option to pursue higher schooling levels, while final schooling levels donot provide an option. Final schooling levels are highlighted using bold letters. ATE† represents the averageeffect for the full population, while ATE presents the average effect for those who visit the decision node.The TT column presents the average effect for those who chose a higher level of schooling (Dj,j′′ = 1), andTUT presents the average effect for those who do not choose a higher level of schooling (Dj,j′′ = 0). Finally,AMTE presents the average effect for those who are indifferent between choosing a higher level of schoolingor not. The table also presents the estimated treatment effects conditional upon endowment levels. Thehigh (low) ability group is defined as those individuals with cognitive and socio-emotional endowment above(below) the overall median. For each decision node we display the fraction of individuals with low and highability levels visiting each node.
56
Table 72: The Effects of Education on Physical Health (PCS-12), by Decision Node
% ATE† ATE TT TUT AMTE
A. Dropping from HS vs. Graduating from HSAll 0.27** 0.27** 0.29** 0.20** 0.18**
( 0.09) ( 0.09) ( 0.11) ( 0.09) ( 0.09)Low Ability 0.31 0.20** 0.20** 0.19* 0.20*
( 0.10) ( 0.10) ( 0.11) ( 0.11)High Ability 0.31 0.36* 0.36* 0.36* 0.23
( 0.18) ( 0.18) ( 0.18) ( 0.15)B. HS Dropout vs. Getting a GED
All -0.19 -0.03 -0.07 0.00 -0.05( 0.24) ( 0.13) ( 0.16) ( 0.14) ( 0.14)
Low Ability 0.61 -0.16 -0.06 -0.12 -0.02( 0.15) ( 0.14) ( 0.15) ( 0.14)
High Ability 0.06 -0.21 0.10 0.06 0.21( 0.43) ( 0.35) ( 0.35) ( 0.35)
C. HS Graduate vs. College EnrollmentAll 0.08 0.10* 0.12* 0.07 0.11*
( 0.06) ( 0.05) ( 0.06) ( 0.07) ( 0.05)Low Ability 0.22 0.01 0.04 0.07 0.03
( 0.12) ( 0.10) ( 0.09) ( 0.10)High Ability 0.38 0.14* 0.14* 0.15* 0.13*
( 0.08) ( 0.08) ( 0.08) ( 0.07)D. Some College vs. 4-year college degree
All 0.22** 0.15** 0.13* 0.18** 0.17**( 0.07) ( 0.05) ( 0.06) ( 0.06) ( 0.06)
Low Ability 0.14 0.26 0.18 0.15 0.19( 0.13) ( 0.10) ( 0.10) ( 0.10)
High Ability 0.51 0.17** 0.14* 0.12 0.17**( 0.07) ( 0.08) ( 0.09) ( 0.07)
Notes: Standard errors (in parenthesis) and significance levels (*=p< 0.05, **=p< 0.01) are calculated using200 bootstrap samples. Each column presents the average effect of an educational decision. Importantly,each schooling level might provide the option to pursue higher schooling levels, while final schooling levels donot provide an option. Final schooling levels are highlighted using bold letters. ATE† represents the averageeffect for the full population, while ATE presents the average effect for those who visit the decision node.The TT column presents the average effect for those who chose a higher level of schooling (Dj,j′′ = 1), andTUT presents the average effect for those who do not choose a higher level of schooling (Dj,j′′ = 0). Finally,AMTE presents the average effect for those who are indifferent between choosing a higher level of schoolingor not. The table also presents the estimated treatment effects conditional upon endowment levels. Thehigh (low) ability group is defined as those individuals with cognitive and socio-emotional endowment above(below) the overall median. For each decision node we display the fraction of individuals with low and highability levels visiting each node.
57
Table 73: The Effects of Education on Mental Health (MCS-12), by Decision Node
% ATE† ATE TT TUT AMTE
A. Dropping from HS vs. Graduating from HSAll 0.20* 0.20* 0.21* 0.15* 0.10
( 0.11) ( 0.11) ( 0.14) ( 0.08) ( 0.07)Low Ability 0.31 0.17* 0.17* 0.16 0.20*
( 0.09) ( 0.09) ( 0.10) ( 0.09)High Ability 0.31 0.29 0.29 0.30 0.12
( 0.22) ( 0.22) ( 0.22) ( 0.17)B. HS Dropout vs. Getting a GED
All -0.48* -0.03 -0.08 0.02 -0.07( 0.23) ( 0.12) ( 0.14) ( 0.14) ( 0.12)
Low Ability 0.61 -0.12 0.08 0.02 0.12( 0.16) ( 0.14) ( 0.16) ( 0.16)
High Ability 0.06 -0.83* -0.39 -0.41 -0.34( 0.38) ( 0.29) ( 0.32) ( 0.31)
C. HS Graduate vs. College EnrollmentAll 0.01 0.01 -0.01 0.04 0.03
( 0.06) ( 0.05) ( 0.06) ( 0.05) ( 0.05)Low Ability 0.22 -0.01 -0.02 -0.04 -0.01
( 0.10) ( 0.08) ( 0.08) ( 0.09)High Ability 0.38 0.03 0.02 0.00 0.07
( 0.06) ( 0.06) ( 0.07) ( 0.06)D. Some College vs. 4-year college degree
All 0.02 -0.04 -0.07 -0.01 -0.01( 0.10) ( 0.07) ( 0.08) ( 0.08) ( 0.07)
Low Ability 0.14 0.05 -0.05 -0.09 -0.03( 0.18) ( 0.15) ( 0.15) ( 0.14)
High Ability 0.51 -0.01 -0.04 -0.06 0.00( 0.07) ( 0.07) ( 0.08) ( 0.09)
Notes: Standard errors (in parenthesis) and significance levels (*=p< 0.05, **=p< 0.01) are calculated using200 bootstrap samples. Each column presents the average effect of an educational decision. Importantly,each schooling level might provide the option to pursue higher schooling levels, while final schooling levels donot provide an option. Final schooling levels are highlighted using bold letters. ATE† represents the averageeffect for the full population, while ATE presents the average effect for those who visit the decision node.The TT column presents the average effect for those who chose a higher level of schooling (Dj,j′′ = 1), andTUT presents the average effect for those who do not choose a higher level of schooling (Dj,j′′ = 0). Finally,AMTE presents the average effect for those who are indifferent between choosing a higher level of schoolingor not. The table also presents the estimated treatment effects conditional upon endowment levels. Thehigh (low) ability group is defined as those individuals with cognitive and socio-emotional endowment above(below) the overall median. For each decision node we display the fraction of individuals with low and highability levels visiting each node.
58
Table 74: The Effects of Education on Depression (CES-D), by Decision Node
% ATE† ATE TT TUT AMTE
A. Dropping from HS vs. Graduating from HSAll 0.16 0.16 0.13 0.23** 0.18**
( 0.10) ( 0.10) ( 0.11) ( 0.08) ( 0.08)Low Ability 0.31 0.14* 0.14* 0.05 0.24*
( 0.08) ( 0.08) ( 0.10) ( 0.10)High Ability 0.31 0.20 0.20 0.20 0.26
( 0.18) ( 0.18) ( 0.18) ( 0.15)B. HS Dropout vs. Getting a GED
All -0.19 0.12 0.03 0.20 0.09( 0.25) ( 0.12) ( 0.13) ( 0.14) ( 0.11)
Low Ability 0.61 0.08 0.20 0.13 0.25( 0.14) ( 0.15) ( 0.14) ( 0.17)
High Ability 0.06 -0.48 -0.18 -0.19 -0.14( 0.40) ( 0.32) ( 0.32) ( 0.35)
C. HS Graduate vs. College EnrollmentAll 0.11* 0.11** 0.10* 0.12** 0.12*
( 0.06) ( 0.05) ( 0.06) ( 0.06) ( 0.05)Low Ability 0.22 0.10 0.11 0.11 0.11
( 0.11) ( 0.09) ( 0.08) ( 0.10)High Ability 0.38 0.11 0.11 0.10 0.13*
( 0.07) ( 0.08) ( 0.08) ( 0.07)D. Some College vs. 4-year college degree
All 0.19* 0.13** 0.10* 0.17* 0.12**( 0.09) ( 0.06) ( 0.06) ( 0.07) ( 0.06)
Low Ability 0.14 0.27 0.24* 0.22* 0.25*( 0.15) ( 0.12) ( 0.12) ( 0.12)
High Ability 0.51 0.09 0.08 0.07 0.11( 0.07) ( 0.07) ( 0.07) ( 0.08)
Notes: Standard errors (in parenthesis) and significance levels (*=p< 0.05, **=p< 0.01) are calculated using200 bootstrap samples. Each column presents the average effect of an educational decision. Importantly,each schooling level might provide the option to pursue higher schooling levels, while final schooling levels donot provide an option. Final schooling levels are highlighted using bold letters. ATE† represents the averageeffect for the full population, while ATE presents the average effect for those who visit the decision node.The TT column presents the average effect for those who chose a higher level of schooling (Dj,j′′ = 1), andTUT presents the average effect for those who do not choose a higher level of schooling (Dj,j′′ = 0). Finally,AMTE presents the average effect for those who are indifferent between choosing a higher level of schoolingor not. The table also presents the estimated treatment effects conditional upon endowment levels. Thehigh (low) ability group is defined as those individuals with cognitive and socio-emotional endowment above(below) the overall median. For each decision node we display the fraction of individuals with low and highability levels visiting each node.
59
Table 75: The Effects of Education on Pearlin, by Decision Node
% ATE† ATE TT TUT AMTE
A. Dropping from HS vs. Graduating from HSAll -0.14 -0.14 -0.21 0.12* 0.14**
( 0.09) ( 0.09) ( 0.12) ( 0.08) ( 0.08)Low Ability 0.31 0.02 0.02 -0.06 0.11
( 0.08) ( 0.08) ( 0.08) ( 0.08)High Ability 0.31 -0.37* -0.37* -0.39* 0.08
( 0.18) ( 0.18) ( 0.19) ( 0.15)B. HS Dropout vs. Getting a GED
All 0.59** 0.21* 0.28* 0.14 0.21*( 0.23) ( 0.11) ( 0.13) ( 0.12) ( 0.12)
Low Ability 0.61 0.28* 0.10 0.17 0.06( 0.13) ( 0.12) ( 0.12) ( 0.13)
High Ability 0.06 0.92** 0.59* 0.62* 0.52*( 0.40) ( 0.28) ( 0.30) ( 0.27)
C. HS Graduate vs. College EnrollmentAll 0.26** 0.23** 0.16** 0.29** 0.25**
( 0.06) ( 0.06) ( 0.07) ( 0.06) ( 0.06)Low Ability 0.22 0.39** 0.37** 0.33** 0.39**
( 0.13) ( 0.10) ( 0.09) ( 0.11)High Ability 0.38 0.11 0.11 0.08 0.16*
( 0.08) ( 0.08) ( 0.09) ( 0.08)D. Some College vs. 4-year college degree
All 0.19* 0.13* 0.11 0.17* 0.14*( 0.09) ( 0.07) ( 0.08) ( 0.08) ( 0.07)
Low Ability 0.14 0.24 0.19 0.18 0.20( 0.18) ( 0.13) ( 0.13) ( 0.14)
High Ability 0.51 0.13 0.11 0.09 0.14( 0.09) ( 0.10) ( 0.10) ( 0.09)
Notes: Standard errors (in parenthesis) and significance levels (*=p< 0.05, **=p< 0.01) are calculated using200 bootstrap samples. Each column presents the average effect of an educational decision. Importantly,each schooling level might provide the option to pursue higher schooling levels, while final schooling levels donot provide an option. Final schooling levels are highlighted using bold letters. ATE† represents the averageeffect for the full population, while ATE presents the average effect for those who visit the decision node.The TT column presents the average effect for those who chose a higher level of schooling (Dj,j′′ = 1), andTUT presents the average effect for those who do not choose a higher level of schooling (Dj,j′′ = 0). Finally,AMTE presents the average effect for those who are indifferent between choosing a higher level of schoolingor not. The table also presents the estimated treatment effects conditional upon endowment levels. Thehigh (low) ability group is defined as those individuals with cognitive and socio-emotional endowment above(below) the overall median. For each decision node we display the fraction of individuals with low and highability levels visiting each node.
60
Table 76: The Effects of Education on Self-Esteem (Rosenberg), by Decision Node
% ATE† ATE TT TUT AMTE
A. Dropping from HS vs. Graduating from HSAll 0.08 0.08 0.09 0.02 0.02
( 0.10) ( 0.10) ( 0.12) ( 0.07) ( 0.08)Low Ability 0.31 0.05 0.05 0.08 0.01
( 0.10) ( 0.10) ( 0.11) ( 0.10)High Ability 0.31 0.08 0.08 0.08 0.03
( 0.18) ( 0.18) ( 0.19) ( 0.14)B. HS Dropout vs. Getting a GED
All -0.19 -0.07 -0.14 -0.01 -0.11( 0.25) ( 0.12) ( 0.15) ( 0.15) ( 0.13)
Low Ability 0.61 -0.13 -0.07 -0.14 -0.02( 0.15) ( 0.15) ( 0.16) ( 0.17)
High Ability 0.06 -0.26 -0.08 -0.12 0.00( 0.40) ( 0.28) ( 0.28) ( 0.29)
C. HS Graduate vs. College EnrollmentAll 0.23** 0.20** 0.13** 0.28** 0.21**
( 0.07) ( 0.07) ( 0.07) ( 0.07) ( 0.06)Low Ability 0.22 0.31* 0.28** 0.25* 0.29**
( 0.14) ( 0.12) ( 0.10) ( 0.13)High Ability 0.38 0.11* 0.10 0.06 0.19*
( 0.07) ( 0.07) ( 0.08) ( 0.08)D. Some College vs. 4-year college degree
All 0.21* 0.12 0.07 0.18* 0.15*( 0.12) ( 0.09) ( 0.10) ( 0.10) ( 0.09)
Low Ability 0.14 0.40** 0.32** 0.26* 0.35**( 0.20) ( 0.18) ( 0.19) ( 0.17)
High Ability 0.51 0.03 0.01 -0.00 0.05( 0.11) ( 0.11) ( 0.12) ( 0.11)
Notes: Standard errors (in parenthesis) and significance levels (*=p< 0.05, **=p< 0.01) are calculated using200 bootstrap samples. Each column presents the average effect of an educational decision. Importantly,each schooling level might provide the option to pursue higher schooling levels, while final schooling levels donot provide an option. Final schooling levels are highlighted using bold letters. ATE† represents the averageeffect for the full population, while ATE presents the average effect for those who visit the decision node.The TT column presents the average effect for those who chose a higher level of schooling (Dj,j′′ = 1), andTUT presents the average effect for those who do not choose a higher level of schooling (Dj,j′′ = 0). Finally,AMTE presents the average effect for those who are indifferent between choosing a higher level of schoolingor not. The table also presents the estimated treatment effects conditional upon endowment levels. Thehigh (low) ability group is defined as those individuals with cognitive and socio-emotional endowment above(below) the overall median. For each decision node we display the fraction of individuals with low and highability levels visiting each node.
61
Table 77: The Effects of Education on Participation, by Decision Node
% ATE† ATE TT TUT AMTE
A. Dropping from HS vs. Graduating from HSAll 0.06** 0.06** 0.05** 0.09** 0.09**
( 0.02) ( 0.02) ( 0.03) ( 0.02) ( 0.02)Low Ability 0.31 0.07** 0.07** 0.05** 0.09**
( 0.02) ( 0.02) ( 0.02) ( 0.03)High Ability 0.31 0.04* 0.04* 0.04* 0.07**
( 0.04) ( 0.04) ( 0.04) ( 0.04)B. HS Dropout vs. Getting a GED
All -0.06 -0.06 -0.07* -0.06 -0.07*( 0.04) ( 0.03) ( 0.03) ( 0.05) ( 0.04)
Low Ability 0.61 -0.06* -0.06 -0.07* -0.06( 0.03) ( 0.05) ( 0.04) ( 0.05)
High Ability 0.06 -0.05 -0.06 -0.06 -0.05( 0.05) ( 0.06) ( 0.06) ( 0.07)
C. HS Graduate vs. College EnrollmentAll -0.01 -0.01 -0.01 -0.01 -0.01
( 0.01) ( 0.01) ( 0.01) ( 0.01) ( 0.01)Low Ability 0.22 -0.01 -0.02 -0.03 -0.02
( 0.03) ( 0.02) ( 0.02) ( 0.02)High Ability 0.38 -0.01 -0.01 -0.00 -0.01
( 0.01) ( 0.01) ( 0.01) ( 0.01)D. Some College vs. 4-year college degree
All -0.01 0.00 0.01 -0.00 0.00( 0.03) ( 0.02) ( 0.02) ( 0.02) ( 0.02)
Low Ability 0.14 -0.06 -0.05 -0.05 -0.05( 0.07) ( 0.05) ( 0.04) ( 0.05)
High Ability 0.51 0.03 0.02 0.02 0.03( 0.02) ( 0.02) ( 0.02) ( 0.02)
Notes: Standard errors (in parenthesis) and significance levels (*=p< 0.05, **=p< 0.01) are calculated using200 bootstrap samples. Each column presents the average effect of an educational decision. Importantly,each schooling level might provide the option to pursue higher schooling levels, while final schooling levels donot provide an option. Final schooling levels are highlighted using bold letters. ATE† represents the averageeffect for the full population, while ATE presents the average effect for those who visit the decision node.The TT column presents the average effect for those who chose a higher level of schooling (Dj,j′′ = 1), andTUT presents the average effect for those who do not choose a higher level of schooling (Dj,j′′ = 0). Finally,AMTE presents the average effect for those who are indifferent between choosing a higher level of schoolingor not. The table also presents the estimated treatment effects conditional upon endowment levels. Thehigh (low) ability group is defined as those individuals with cognitive and socio-emotional endowment above(below) the overall median. For each decision node we display the fraction of individuals with low and highability levels visiting each node.
62
Table 78: The Effects of Education on White Collar, by Decision Node
% ATE† ATE TT TUT AMTE
A. Dropping from HS vs. Graduating from HSAll 0.15** 0.15** 0.17** 0.05* 0.05
( 0.05) ( 0.05) ( 0.06) ( 0.03) ( 0.03)Low Ability 0.31 0.06 0.06 0.08* 0.03
( 0.04) ( 0.04) ( 0.05) ( 0.04)High Ability 0.31 0.26** 0.26** 0.26** 0.16**
( 0.09) ( 0.09) ( 0.09) ( 0.06)B. HS Dropout vs. Getting a GED
All 0.14 0.03 0.06 0.01 0.04( 0.09) ( 0.04) ( 0.05) ( 0.04) ( 0.05)
Low Ability 0.61 0.08 0.03 0.05 0.01( 0.05) ( 0.05) ( 0.05) ( 0.05)
High Ability 0.06 0.20 0.05 0.06 0.03( 0.16) ( 0.11) ( 0.11) ( 0.10)
C. HS Graduate vs. College EnrollmentAll 0.22** 0.24** 0.28** 0.21** 0.23**
( 0.03) ( 0.03) ( 0.04) ( 0.04) ( 0.03)Low Ability 0.22 0.13** 0.14** 0.15** 0.14**
( 0.06) ( 0.05) ( 0.05) ( 0.06)High Ability 0.38 0.33** 0.33** 0.34** 0.31**
( 0.04) ( 0.04) ( 0.04) ( 0.04)D. Some College vs. 4-year college degree
All 0.14** 0.25** 0.29** 0.20** 0.27**( 0.06) ( 0.04) ( 0.05) ( 0.05) ( 0.04)
Low Ability 0.14 -0.02 0.12 0.19** 0.08( 0.12) ( 0.10) ( 0.09) ( 0.11)
High Ability 0.51 0.29** 0.31** 0.32** 0.29**( 0.05) ( 0.05) ( 0.06) ( 0.05)
Notes: Standard errors (in parenthesis) and significance levels (*=p< 0.05, **=p< 0.01) are calculated using200 bootstrap samples. Each column presents the average effect of an educational decision. Importantly,each schooling level might provide the option to pursue higher schooling levels, while final schooling levels donot provide an option. Final schooling levels are highlighted using bold letters. ATE† represents the averageeffect for the full population, while ATE presents the average effect for those who visit the decision node.The TT column presents the average effect for those who chose a higher level of schooling (Dj,j′′ = 1), andTUT presents the average effect for those who do not choose a higher level of schooling (Dj,j′′ = 0). Finally,AMTE presents the average effect for those who are indifferent between choosing a higher level of schoolingor not. The table also presents the estimated treatment effects conditional upon endowment levels. Thehigh (low) ability group is defined as those individuals with cognitive and socio-emotional endowment above(below) the overall median. For each decision node we display the fraction of individuals with low and highability levels visiting each node.
63
Table 79: The Effects of Education on Obesity, by Decision Node
% ATE† ATE TT TUT AMTE
A. Dropping from HS vs. Graduating from HSAll 0.01 0.01 -0.01 0.08* 0.10**
( 0.06) ( 0.06) ( 0.07) ( 0.04) ( 0.04)Low Ability 0.31 0.11** 0.11** 0.11* 0.11**
( 0.04) ( 0.04) ( 0.05) ( 0.05)High Ability 0.31 -0.10 -0.10 -0.11 -0.02
( 0.10) ( 0.10) ( 0.11) ( 0.08)B. HS Dropout vs. Getting a GED
All 0.12 0.02 0.05 -0.01 0.03( 0.09) ( 0.05) ( 0.06) ( 0.06) ( 0.06)
Low Ability 0.61 0.07 0.01 0.05 -0.01( 0.06) ( 0.06) ( 0.06) ( 0.06)
High Ability 0.06 0.18 0.03 0.05 -0.00( 0.16) ( 0.13) ( 0.14) ( 0.15)
C. HS Graduate vs. College EnrollmentAll -0.03 -0.06* -0.09** -0.03 -0.04
( 0.03) ( 0.03) ( 0.04) ( 0.04) ( 0.03)Low Ability 0.22 0.04 0.01 -0.02 0.03
( 0.06) ( 0.05) ( 0.05) ( 0.05)High Ability 0.38 -0.09* -0.10* -0.11** -0.07*
( 0.04) ( 0.04) ( 0.04) ( 0.04)D. Some College vs. 4-year college degree
All -0.06 -0.08* -0.08* -0.07 -0.07( 0.06) ( 0.04) ( 0.05) ( 0.05) ( 0.04)
Low Ability 0.14 -0.12 -0.17* -0.19* -0.17*( 0.10) ( 0.07) ( 0.07) ( 0.08)
High Ability 0.51 -0.01 -0.03 -0.05 -0.01( 0.05) ( 0.05) ( 0.06) ( 0.05)
Notes: Standard errors (in parenthesis) and significance levels (*=p< 0.05, **=p< 0.01) are calculated using200 bootstrap samples. Each column presents the average effect of an educational decision. Importantly,each schooling level might provide the option to pursue higher schooling levels, while final schooling levels donot provide an option. Final schooling levels are highlighted using bold letters. ATE† represents the averageeffect for the full population, while ATE presents the average effect for those who visit the decision node.The TT column presents the average effect for those who chose a higher level of schooling (Dj,j′′ = 1), andTUT presents the average effect for those who do not choose a higher level of schooling (Dj,j′′ = 0). Finally,AMTE presents the average effect for those who are indifferent between choosing a higher level of schoolingor not. The table also presents the estimated treatment effects conditional upon endowment levels. Thehigh (low) ability group is defined as those individuals with cognitive and socio-emotional endowment above(below) the overall median. For each decision node we display the fraction of individuals with low and highability levels visiting each node.
64
Table 80: The Effects of Education on Daily Smoking, by Decision Node
% ATE† ATE TT TUT AMTE
A. Dropping from HS vs. Graduating from HSAll -0.25** -0.25** -0.25** -0.25** -0.23**
( 0.05) ( 0.05) ( 0.07) ( 0.03) ( 0.03)Low Ability 0.31 -0.28** -0.28** -0.28** -0.28**
( 0.04) ( 0.04) ( 0.05) ( 0.04)High Ability 0.31 -0.23* -0.23* -0.24* -0.14*
( 0.11) ( 0.11) ( 0.11) ( 0.07)B. HS Dropout vs. Getting a GED
All 0.05 0.01 0.01 0.01 0.00( 0.10) ( 0.05) ( 0.06) ( 0.05) ( 0.05)
Low Ability 0.61 -0.00 -0.01 -0.01 -0.01( 0.06) ( 0.05) ( 0.05) ( 0.05)
High Ability 0.06 0.10 0.06 0.06 0.06( 0.18) ( 0.14) ( 0.15) ( 0.13)
C. HS Graduate vs. College EnrollmentAll -0.12** -0.14** -0.18** -0.10** -0.13**
( 0.03) ( 0.03) ( 0.03) ( 0.03) ( 0.03)Low Ability 0.22 -0.07 -0.09* -0.11** -0.08
( 0.06) ( 0.05) ( 0.04) ( 0.05)High Ability 0.38 -0.18** -0.19** -0.21** -0.14**
( 0.04) ( 0.04) ( 0.04) ( 0.03)D. Some College vs. 4-year college degree
All -0.15** -0.17** -0.18** -0.16** -0.17**( 0.05) ( 0.04) ( 0.05) ( 0.04) ( 0.04)
Low Ability 0.14 -0.10 -0.11* -0.12 -0.10( 0.09) ( 0.08) ( 0.08) ( 0.08)
High Ability 0.51 -0.20** -0.20** -0.20** -0.20**( 0.05) ( 0.05) ( 0.06) ( 0.04)
Notes: Standard errors (in parenthesis) and significance levels (*=p< 0.05, **=p< 0.01) are calculated using200 bootstrap samples. Each column presents the average effect of an educational decision. Importantly,each schooling level might provide the option to pursue higher schooling levels, while final schooling levels donot provide an option. Final schooling levels are highlighted using bold letters. ATE† represents the averageeffect for the full population, while ATE presents the average effect for those who visit the decision node.The TT column presents the average effect for those who chose a higher level of schooling (Dj,j′′ = 1), andTUT presents the average effect for those who do not choose a higher level of schooling (Dj,j′′ = 0). Finally,AMTE presents the average effect for those who are indifferent between choosing a higher level of schoolingor not. The table also presents the estimated treatment effects conditional upon endowment levels. Thehigh (low) ability group is defined as those individuals with cognitive and socio-emotional endowment above(below) the overall median. For each decision node we display the fraction of individuals with low and highability levels visiting each node.
65
characteristics. Final schooling levels do not provide the option for additional schooling.
66
Table 81: The Effects of Education on Heavy Drinker, by Decision Node
% ATE† ATE TT TUT AMTE
A. Dropping from HS vs. Graduating from HSAll -0.10* -0.10* -0.13* 0.01 0.02
( 0.06) ( 0.06) ( 0.06) ( 0.03) ( 0.04)Low Ability 0.31 0.03 0.03 0.02 0.05
( 0.04) ( 0.04) ( 0.05) ( 0.04)High Ability 0.31 -0.25** -0.25** -0.25** -0.16*
( 0.10) ( 0.10) ( 0.11) ( 0.07)B. HS Dropout vs. Getting a GED
All -0.03 -0.07 -0.07 -0.08 -0.07( 0.12) ( 0.06) ( 0.08) ( 0.05) ( 0.06)
Low Ability 0.61 -0.07 -0.08 -0.08 -0.09( 0.07) ( 0.05) ( 0.06) ( 0.05)
High Ability 0.06 0.02 -0.03 -0.02 -0.06( 0.21) ( 0.16) ( 0.17) ( 0.15)
C. HS Graduate vs. College EnrollmentAll -0.02 -0.03 -0.06* -0.01 -0.02
( 0.03) ( 0.03) ( 0.03) ( 0.03) ( 0.03)Low Ability 0.22 0.04 0.04 0.03 0.04
( 0.06) ( 0.06) ( 0.05) ( 0.06)High Ability 0.38 -0.09** -0.09** -0.10** -0.07*
( 0.03) ( 0.04) ( 0.04) ( 0.04)D. Some College vs. 4-year college degree
All 0.03 -0.01 -0.03 0.01 -0.01( 0.06) ( 0.04) ( 0.04) ( 0.05) ( 0.04)
Low Ability 0.14 0.10 0.07 0.06 0.08( 0.13) ( 0.09) ( 0.08) ( 0.09)
High Ability 0.51 -0.03 -0.04 -0.04 -0.03( 0.03) ( 0.04) ( 0.04) ( 0.04)
Notes: Standard errors (in parenthesis) and significance levels (*=p< 0.05, **=p< 0.01) are calculated using200 bootstrap samples. Each column presents the average effect of an educational decision. Importantly,each schooling level might provide the option to pursue higher schooling levels, while final schooling levels donot provide an option. Final schooling levels are highlighted using bold letters. ATE† represents the averageeffect for the full population, while ATE presents the average effect for those who visit the decision node.The TT column presents the average effect for those who chose a higher level of schooling (Dj,j′′ = 1), andTUT presents the average effect for those who do not choose a higher level of schooling (Dj,j′′ = 0). Finally,AMTE presents the average effect for those who are indifferent between choosing a higher level of schoolingor not. The table also presents the estimated treatment effects conditional upon endowment levels. Thehigh (low) ability group is defined as those individuals with cognitive and socio-emotional endowment above(below) the overall median. For each decision node we display the fraction of individuals with low and highability levels visiting each node.
67
68
G The Effect of Cognitive and Socio-emotional endow-
ments
G.1 Sorting into Schooling
Figure 1: Distribution of Cognitive and Socio-emotional Endowment by Decision Node
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional
12
34
56
78
910
Fra
ctio
n of
Pop
ulat
ion
0.004
0.006
0.008
0.01
0.012
0.014
0.016
0.018
(A)
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional
12
34
56
78
910
Fra
ctio
n of
Pop
ulat
ion
0
0.005
0.01
0.015
0.02
0.025
0.03
0.035
0.04
0.045
(B)
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional
12
34
56
78
910
Fra
ctio
n of
Pop
ulat
ion
0.0040.0060.008
0.010.0120.0140.0160.0180.02
0.0220.024
(C)
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional
12
34
56
78
910
Fra
ctio
n of
Pop
ulat
ion
0
0.005
0.01
0.015
0.02
0.025
0.03
0.035
(D)
Note: Each panel presents the distribution of individuals by levels of endowments for a particulardecision node schooling level. We define as “decile 1” the decile with the lowest values of endowmentsand “decile 10” as the decile with the highest levels of endowments. The deciles are computed usingthe overall marginal distributions of endowments (instead of the joint distribution). This explainswhy panel (A) does not present bars of the same height.
69
70
G.2 Labor Market Outcomes
Figure 2: The Effect of Cognitive and Socio-emotional endowments on Labor MarketParticipation (age 30)
A. All B. High School Dropout
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
Par
ticip
atio
n
0.60.650.7
0.750.8
0.850.9
0.951
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Par
ticip
atio
n
0.6
0.7
0.8
0.9
1 Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
Participation
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
Par
ticip
atio
n
0.6
0.7
0.8
0.9
1 Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
Participation
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
Par
ticip
atio
n
0.60.650.7
0.750.8
0.850.9
0.951
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Par
ticip
atio
n
0.6
0.7
0.8
0.9
1 Fra
ctio
n
0
0.1
0.2
0.3
0.4
0.5
0.6Participation
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
Par
ticip
atio
n
0.6
0.7
0.8
0.9
1 Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45Participation
C. GED D. High School Graduate
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
Par
ticip
atio
n
0.60.650.7
0.750.8
0.850.9
0.951
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Par
ticip
atio
n
0.6
0.7
0.8
0.9
1 Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25Participation
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
Par
ticip
atio
n
0.6
0.7
0.8
0.9
1 Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45Participation
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
Par
ticip
atio
n0.6
0.650.7
0.750.8
0.850.9
0.951
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Par
ticip
atio
n
0.6
0.7
0.8
0.9
1 Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22Participation
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
Par
ticip
atio
n
0.6
0.7
0.8
0.9
1 Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22
Participation
E. Some College F. Four-Year College
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
Par
ticip
atio
n
0.60.650.7
0.750.8
0.850.9
0.951
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Par
ticip
atio
n
0.6
0.7
0.8
0.9
1 Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
Participation
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
Par
ticip
atio
n
0.6
0.7
0.8
0.9
1 Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22Participation
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
Par
ticip
atio
n
0.60.650.7
0.750.8
0.850.9
0.951
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Par
ticip
atio
n
0.6
0.7
0.8
0.9
1 Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45Participation
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
Par
ticip
atio
n
0.6
0.7
0.8
0.9
1 Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35Participation
Note: For each schooling level we present three figures. The first figure (top) displays the levels of the outcome as a functionof cognitive and socio-emotional endowments. In particular, we present the average level of outcomes for different deciles ofcognitive and socio-emotional endowments. Notice that we define as “decile 1” the decile with the lowest values of endowmentsand “decile 10” as the decile with the highest levels of endowments. The second figure (bottom left) displays the average levelsof endowment across deciles of cognitive endowments. The bars in this figure indicates the fraction of individuals reporting therespective schooling level for each decile of cognitive endowment. The last figure (bottom right) mimics the structure of thesecond one but now for the socio-emotional endowment.
71
Figure 3: The Effect of Cognitive and Socio-emotional endowments on Probability of White-collaroccupation (age 30)
A. All B. High School Dropout
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
Whi
teco
llar
0.10.20.30.40.50.60.70.80.9
1
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Whi
teco
llar
0.2
0.4
0.6
0.8
1
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
Whitecollar
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
Whi
teco
llar
0.2
0.4
0.6
0.8
1
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
Whitecollar
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
Whi
teco
llar
0.10.20.30.40.50.60.70.80.9
1
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Whi
teco
llar
0.2
0.4
0.6
0.8
1
Fra
ctio
n
0
0.1
0.2
0.3
0.4
0.5
0.6Whitecollar
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
Whi
teco
llar
0.2
0.4
0.6
0.8
1
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45Whitecollar
C. GED D. High School Graduate
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
Whi
teco
llar
0.10.20.30.40.50.60.70.80.9
1
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Whi
teco
llar
0.2
0.4
0.6
0.8
1
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25Whitecollar
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
Whi
teco
llar
0.2
0.4
0.6
0.8
1
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45Whitecollar
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
Whi
teco
llar
0.10.20.30.40.50.60.70.80.9
1
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Whi
teco
llar
0.2
0.4
0.6
0.8
1
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22Whitecollar
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
Whi
teco
llar
0.2
0.4
0.6
0.8
1
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22
Whitecollar
E. Some College F. Four-Year College
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
Whi
teco
llar
0.10.20.30.40.50.60.70.80.9
1
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Whi
teco
llar
0.2
0.4
0.6
0.8
1
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
Whitecollar
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
Whi
teco
llar
0.2
0.4
0.6
0.8
1
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22Whitecollar
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
Whi
teco
llar
0.10.20.30.40.50.60.70.80.9
1
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Whi
teco
llar
0.2
0.4
0.6
0.8
1
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45Whitecollar
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
Whi
teco
llar
0.2
0.4
0.6
0.8
1
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35Whitecollar
Note: For each schooling level we present three figures. The first figure (top) displays the levels of the outcome as a functionof cognitive and socio-emotional endowments. In particular, we present the average level of outcomes for different deciles ofcognitive and socio-emotional endowments. Notice that we define as “decile 1” the decile with the lowest values of endowmentsand “decile 10” as the decile with the highest levels of endowments. The second figure (bottom left) displays the average levelsof endowment across deciles of cognitive endowments. The bars in this figure indicates the fraction of individuals reporting therespective schooling level for each decile of cognitive endowment. The last figure (bottom right) mimics the structure of thesecond one but now for the socio-emotional endowment.
72
Figure 4: The Effect of Cognitive and Socio-emotional endowments on (log) Wages (age 30)
A. All B. High School Dropout
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
Log-
wag
es
2.2
2.4
2.6
2.8
3
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Log-
wag
es
2.2
2.4
2.6
2.8
3
3.2
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
Log-wages
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
Log-
wag
es
2.2
2.4
2.6
2.8
3
3.2
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
Log-wages
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
Log-
wag
es
2.2
2.4
2.6
2.8
3
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Log-
wag
es
2.2
2.4
2.6
2.8
3
3.2
Fra
ctio
n
0
0.1
0.2
0.3
0.4
0.5
0.6Log-wages
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
Log-
wag
es
2.2
2.4
2.6
2.8
3
3.2
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45Log-wages
C. GED D. High School Graduate
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
Log-
wag
es
2.2
2.4
2.6
2.8
3
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Log-
wag
es
2.2
2.4
2.6
2.8
3
3.2
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25Log-wages
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
Log-
wag
es
2.2
2.4
2.6
2.8
3
3.2
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45Log-wages
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
Log-
wag
es
2.2
2.4
2.6
2.8
3
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Log-
wag
es
2.2
2.4
2.6
2.8
3
3.2
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22Log-wages
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
Log-
wag
es
2.2
2.4
2.6
2.8
3
3.2
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22
Log-wages
E. Some College F. Four-Year College
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
Log-
wag
es
2.2
2.4
2.6
2.8
3
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Log-
wag
es
2.2
2.4
2.6
2.8
3
3.2
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
Log-wages
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
Log-
wag
es
2.2
2.4
2.6
2.8
3
3.2
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22Log-wages
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
Log-
wag
es
2.2
2.4
2.6
2.8
3
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Log-
wag
es
2.2
2.4
2.6
2.8
3
3.2
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45Log-wages
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
Log-
wag
es
2.2
2.4
2.6
2.8
3
3.2
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35Log-wages
Note: For each schooling level we present three figures. The first figure (top) displays the levels of the outcome as a functionof cognitive and socio-emotional endowments. In particular, we present the average level of outcomes for different deciles ofcognitive and socio-emotional endowments. Notice that we define as “decile 1” the decile with the lowest values of endowmentsand “decile 10” as the decile with the highest levels of endowments. The second figure (bottom left) displays the average levelsof endowment across deciles of cognitive endowments. The bars in this figure indicates the fraction of individuals reporting therespective schooling level for each decile of cognitive endowment. The last figure (bottom right) mimics the structure of thesecond one but now for the socio-emotional endowment.
73
Figure 5: The Effect of Cognitive and Socio-emotional endowments on the Present Value ofWages (age 20-40)
A. All B. High School Dropout
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
PV
Wag
es
11.611.8
1212.212.4
12.612.8
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
PV
Wag
es
11.5
12
12.5
13
13.5
14
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
PV Wages
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
PV
Wag
es
11.5
12
12.5
13
13.5
14
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
PV Wages
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
PV
Wag
es
11.611.8
1212.212.4
12.612.8
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
PV
Wag
es
11.5
12
12.5
13
13.5
14
Fra
ctio
n
0
0.1
0.2
0.3
0.4
0.5
0.6PV Wages
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
PV
Wag
es
11.5
12
12.5
13
13.5
14
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45PV Wages
C. GED D. High School Graduate
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
PV
Wag
es
11.611.8
1212.212.4
12.612.8
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
PV
Wag
es
11.5
12
12.5
13
13.5
14
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25PV Wages
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
PV
Wag
es
11.5
12
12.5
13
13.5
14
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45PV Wages
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
PV
Wag
es
11.611.8
1212.212.4
12.612.8
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
PV
Wag
es
11.5
12
12.5
13
13.5
14
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22PV Wages
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
PV
Wag
es
11.5
12
12.5
13
13.5
14
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22
PV Wages
E. Some College F. Four-Year College
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
PV
Wag
es
11.611.8
1212.212.4
12.612.8
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
PV
Wag
es
11.5
12
12.5
13
13.5
14
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
PV Wages
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
PV
Wag
es
11.5
12
12.5
13
13.5
14
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22PV Wages
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
PV
Wag
es
11.611.8
1212.212.4
12.612.8
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
PV
Wag
es
11.5
12
12.5
13
13.5
14
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45PV Wages
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
PV
Wag
es
11.5
12
12.5
13
13.5
14
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35PV Wages
Note: For each schooling level we present three figures. The first figure (top) displays the levels of the outcome as a functionof cognitive and socio-emotional endowments. In particular, we present the average level of outcomes for different deciles ofcognitive and socio-emotional endowments. Notice that we define as “decile 1” the decile with the lowest values of endowmentsand “decile 10” as the decile with the highest levels of endowments. The second figure (bottom left) displays the average levelsof endowment across deciles of cognitive endowments. The bars in this figure indicates the fraction of individuals reporting therespective schooling level for each decile of cognitive endowment. The last figure (bottom right) mimics the structure of thesecond one but now for the socio-emotional endowment.
74
75
G.3 Physical Health Outcomes and Behaviors
Figure 6: The Effect of Cognitive and Socio-emotional endowments on Obesity during Adulthood
A. All B. High School Dropout
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
Obe
sity
0.2
0.250.3
0.350.4
0.45
0.50.55
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Obe
sity
0.2
0.25
0.3
0.35
0.4
0.45
0.5
0.55
0.6
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
Obesity
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
Obe
sity
0.2
0.25
0.3
0.35
0.4
0.45
0.5
0.55
0.6
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
Obesity
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
Obe
sity
0.2
0.250.3
0.350.4
0.45
0.50.55
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Obe
sity
0.2
0.25
0.3
0.35
0.4
0.45
0.5
0.55
0.6
Fra
ctio
n
0
0.1
0.2
0.3
0.4
0.5
0.6Obesity
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
Obe
sity
0.2
0.25
0.3
0.35
0.4
0.45
0.5
0.55
0.6
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45Obesity
C. GED D. High School Graduate
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
Obe
sity
0.2
0.250.3
0.350.4
0.45
0.50.55
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Obe
sity
0.2
0.25
0.3
0.35
0.4
0.45
0.5
0.55
0.6
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25Obesity
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
Obe
sity
0.2
0.25
0.3
0.35
0.4
0.45
0.5
0.55
0.6
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45Obesity
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
Obe
sity
0.2
0.250.3
0.350.4
0.45
0.50.55
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Obe
sity
0.2
0.25
0.3
0.35
0.4
0.45
0.5
0.55
0.6
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22Obesity
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
Obe
sity
0.2
0.25
0.3
0.35
0.4
0.45
0.5
0.55
0.6
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22
Obesity
E. Some College F. Four-Year College
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
Obe
sity
0.2
0.250.3
0.350.4
0.45
0.50.55
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Obe
sity
0.2
0.25
0.3
0.35
0.4
0.45
0.5
0.55
0.6
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
Obesity
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
Obe
sity
0.2
0.25
0.3
0.35
0.4
0.45
0.5
0.55
0.6
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22Obesity
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
Obe
sity
0.2
0.250.3
0.350.4
0.45
0.50.55
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Obe
sity
0.2
0.25
0.3
0.35
0.4
0.45
0.5
0.55
0.6
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45Obesity
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
Obe
sity
0.2
0.25
0.3
0.35
0.4
0.45
0.5
0.55
0.6
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35Obesity
Note: For each schooling level we present three figures. The first figure (top) displays the levels of the outcome as a functionof cognitive and socio-emotional endowments. In particular, we present the average level of outcomes for different deciles ofcognitive and socio-emotional endowments. Notice that we define as “decile 1” the decile with the lowest values of endowmentsand “decile 10” as the decile with the highest levels of endowments. The second figure (bottom left) displays the average levelsof endowment across deciles of cognitive endowments. The bars in this figure indicates the fraction of individuals reporting therespective schooling level for each decile of cognitive endowment. The last figure (bottom right) mimics the structure of thesecond one but now for the socio-emotional endowment.
76
Figure 7: The Effect of Cognitive and Socio-emotional endowments on Smoking during Adulthood
A. All B. High School Dropout
Decile of Cognitive
12345678910
Decile of Socio-Emotional
1 2 34 5 6
7 8 910
Sm
oker
0.10.20.30.40.50.60.70.8
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Sm
oker
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
Sm
oker
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
Sm
oker
0.10.20.30.40.50.60.70.8
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Sm
oker
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Fra
ctio
n
0
0.1
0.2
0.3
0.4
0.5
0.6Smoker
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
Sm
oker
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45Smoker
C. GED D. High School Graduate
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
Sm
oker
0.10.20.30.40.50.60.70.8
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Sm
oker
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25Smoker
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
Sm
oker
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45Smoker
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
Sm
oker
0.10.20.30.40.50.60.70.8
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Sm
oker
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22Smoker
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
Sm
oker
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22
Smoker
E. Some College F. Four-Year College
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
Sm
oker
0.10.20.30.40.50.60.70.8
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Sm
oker
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
Smoker
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
Sm
oker
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22Smoker
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
Sm
oker
0.10.20.30.40.50.60.70.8
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Sm
oker
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45Smoker
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
Sm
oker
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35Smoker
Note: For each schooling level we present three figures. The first figure (top) displays the levels of the outcome as a functionof cognitive and socio-emotional endowments. In particular, we present the average level of outcomes for different deciles ofcognitive and socio-emotional endowments. Notice that we define as “decile 1” the decile with the lowest values of endowmentsand “decile 10” as the decile with the highest levels of endowments. The second figure (bottom left) displays the average levelsof endowment across deciles of cognitive endowments. The bars in this figure indicates the fraction of individuals reporting therespective schooling level for each decile of cognitive endowment. The last figure (bottom right) mimics the structure of thesecond one but now for the socio-emotional endowment.
77
Figure 8: The Effect of Cognitive and Socio-emotional endowments on Heavy Drinking duringAdulthood
A. All B. High School Dropout
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
Hea
vy D
rinke
r
0.10.150.2
0.250.3
0.350.4
0.450.5
0.55
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Hea
vy D
rinke
r
0.1
0.2
0.3
0.4
0.5 Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
Heavy Drinker
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
Hea
vy D
rinke
r
0.1
0.2
0.3
0.4
0.5 Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
Heavy Drinker
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
Hea
vy D
rinke
r
0.10.150.2
0.250.3
0.350.4
0.450.5
0.55
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Hea
vy D
rinke
r
0.1
0.2
0.3
0.4
0.5 Fra
ctio
n
0
0.1
0.2
0.3
0.4
0.5
0.6Heavy Drinker
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
Hea
vy D
rinke
r
0.1
0.2
0.3
0.4
0.5 Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45Heavy Drinker
C. GED D. High School Graduate
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
Hea
vy D
rinke
r
0.10.150.2
0.250.3
0.350.4
0.450.5
0.55
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Hea
vy D
rinke
r
0.1
0.2
0.3
0.4
0.5 Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25Heavy Drinker
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
Hea
vy D
rinke
r
0.1
0.2
0.3
0.4
0.5 Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45Heavy Drinker
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
Hea
vy D
rinke
r
0.10.150.2
0.250.3
0.350.4
0.450.5
0.55
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Hea
vy D
rinke
r
0.1
0.2
0.3
0.4
0.5 Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22Heavy Drinker
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
Hea
vy D
rinke
r
0.1
0.2
0.3
0.4
0.5 Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22
Heavy Drinker
E. Some College F. Four-Year College
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
Hea
vy D
rinke
r
0.10.150.2
0.250.3
0.350.4
0.450.5
0.55
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Hea
vy D
rinke
r
0.1
0.2
0.3
0.4
0.5 Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
Heavy Drinker
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
Hea
vy D
rinke
r
0.1
0.2
0.3
0.4
0.5 Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22Heavy Drinker
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
Hea
vy D
rinke
r
0.10.150.2
0.250.3
0.350.4
0.450.5
0.55
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Hea
vy D
rinke
r
0.1
0.2
0.3
0.4
0.5 Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45Heavy Drinker
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
Hea
vy D
rinke
r
0.1
0.2
0.3
0.4
0.5 Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35Heavy Drinker
Note: For each schooling level we present three figures. The first figure (top) displays the levels of the outcome as a functionof cognitive and socio-emotional endowments. In particular, we present the average level of outcomes for different deciles ofcognitive and socio-emotional endowments. Notice that we define as “decile 1” the decile with the lowest values of endowmentsand “decile 10” as the decile with the highest levels of endowments. The second figure (bottom left) displays the average levelsof endowment across deciles of cognitive endowments. The bars in this figure indicates the fraction of individuals reporting therespective schooling level for each decile of cognitive endowment. The last figure (bottom right) mimics the structure of thesecond one but now for the socio-emotional endowment.
78
Figure 9: The Effect of Cognitive and Socio-emotional endowments on Physical Health at age 40(PCS-12)
A. All B. High School Dropout
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
SF
-12
Phy
sica
l
-0.6
-0.4
-0.2
0
0.2
0.4
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
SF
-12
Phy
sica
l
-0.6
-0.4
-0.2
0
0.2
0.4
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
SF-12 Physical
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
SF
-12
Phy
sica
l
-0.6
-0.4
-0.2
0
0.2
0.4
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
SF-12 Physical
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
SF
-12
Phy
sica
l
-0.6
-0.4
-0.2
0
0.2
0.4
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
SF
-12
Phy
sica
l
-0.6
-0.4
-0.2
0
0.2
0.4
Fra
ctio
n
0
0.1
0.2
0.3
0.4
0.5
0.6SF-12 Physical
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
SF
-12
Phy
sica
l
-0.6
-0.4
-0.2
0
0.2
0.4
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45SF-12 Physical
C. GED D. High School Graduate
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
SF
-12
Phy
sica
l
-0.6
-0.4
-0.2
0
0.2
0.4
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
SF
-12
Phy
sica
l
-0.6
-0.4
-0.2
0
0.2
0.4
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25SF-12 Physical
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
SF
-12
Phy
sica
l
-0.6
-0.4
-0.2
0
0.2
0.4
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45SF-12 Physical
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
SF
-12
Phy
sica
l
-0.6
-0.4
-0.2
0
0.2
0.4
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
SF
-12
Phy
sica
l
-0.6
-0.4
-0.2
0
0.2
0.4
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22SF-12 Physical
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
SF
-12
Phy
sica
l
-0.6
-0.4
-0.2
0
0.2
0.4
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22
SF-12 Physical
E. Some College F. Four-Year College
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
SF
-12
Phy
sica
l
-0.6
-0.4
-0.2
0
0.2
0.4
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
SF
-12
Phy
sica
l
-0.6
-0.4
-0.2
0
0.2
0.4
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
SF-12 Physical
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
SF
-12
Phy
sica
l
-0.6
-0.4
-0.2
0
0.2
0.4
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22SF-12 Physical
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
SF
-12
Phy
sica
l
-0.6
-0.4
-0.2
0
0.2
0.4
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
SF
-12
Phy
sica
l
-0.6
-0.4
-0.2
0
0.2
0.4
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45SF-12 Physical
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
SF
-12
Phy
sica
l
-0.6
-0.4
-0.2
0
0.2
0.4
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35SF-12 Physical
Note: For each schooling level we present three figures. The first figure (top) displays the levels of the outcome as a functionof cognitive and socio-emotional endowments. In particular, we present the average level of outcomes for different deciles ofcognitive and socio-emotional endowments. Notice that we define as “decile 1” the decile with the lowest values of endowmentsand “decile 10” as the decile with the highest levels of endowments. The second figure (bottom left) displays the average levelsof endowment across deciles of cognitive endowments. The bars in this figure indicates the fraction of individuals reporting therespective schooling level for each decile of cognitive endowment. The last figure (bottom right) mimics the structure of thesecond one but now for the socio-emotional endowment.
79
80
G.4 Mental Health Outcomes
Figure 10: The Effect of Cognitive and Socio-emotional endowments on Pearlin’s “PersonalMastery Scale”
A. All B. High School Dropout
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
Pea
rlin
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Pea
rlin
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
Pearlin
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
Pea
rlin
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
Pearlin
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
Pea
rlin
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Pea
rlin
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
Fra
ctio
n
0
0.1
0.2
0.3
0.4
0.5
0.6Pearlin
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
Pea
rlin
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45Pearlin
C. GED D. High School Graduate
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
Pea
rlin
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Pea
rlin
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25Pearlin
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
Pea
rlin
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45Pearlin
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
Pea
rlin
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Pea
rlin
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22Pearlin
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
Pea
rlin
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22
Pearlin
E. Some College F. Four-Year College
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
Pea
rlin
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Pea
rlin
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
Pearlin
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
Pea
rlin
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22Pearlin
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
Pea
rlin
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Pea
rlin
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45Pearlin
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
Pea
rlin
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35Pearlin
Note: For each schooling level we present three figures. The first figure (top) displays the levels of the outcome as a functionof cognitive and socio-emotional endowments. In particular, we present the average level of outcomes for different deciles ofcognitive and socio-emotional endowments. Notice that we define as “decile 1” the decile with the lowest values of endowmentsand “decile 10” as the decile with the highest levels of endowments. The second figure (bottom left) displays the average levelsof endowment across deciles of cognitive endowments. The bars in this figure indicates the fraction of individuals reporting therespective schooling level for each decile of cognitive endowment. The last figure (bottom right) mimics the structure of thesecond one but now for the socio-emotional endowment.
81
Figure 11: The Effect of Cognitive and Socio-emotional endowments on Self-Esteem duringAdulthood
A. All B. High School Dropout
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
Ros
enbe
rg
-0.8
-0.6
-0.4
-0.2
0
0.20.4
0.6
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Ros
enbe
rg
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
Rosenberg
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
Ros
enbe
rg
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
Rosenberg
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
Ros
enbe
rg
-0.8
-0.6
-0.4
-0.2
0
0.20.4
0.6
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Ros
enbe
rg
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
Fra
ctio
n
0
0.1
0.2
0.3
0.4
0.5
0.6Rosenberg
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
Ros
enbe
rg
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45Rosenberg
C. GED D. High School Graduate
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
Ros
enbe
rg
-0.8
-0.6
-0.4
-0.2
0
0.20.4
0.6
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Ros
enbe
rg
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25Rosenberg
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
Ros
enbe
rg
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45Rosenberg
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
Ros
enbe
rg
-0.8
-0.6
-0.4
-0.2
0
0.20.4
0.6
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Ros
enbe
rg
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22Rosenberg
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
Ros
enbe
rg
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22
Rosenberg
E. Some College F. Four-Year College
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
Ros
enbe
rg
-0.8
-0.6
-0.4
-0.2
0
0.20.4
0.6
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Ros
enbe
rg
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
Rosenberg
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
Ros
enbe
rg
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22Rosenberg
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
Ros
enbe
rg
-0.8
-0.6
-0.4
-0.2
0
0.20.4
0.6
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Ros
enbe
rg
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45Rosenberg
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
Ros
enbe
rg
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35Rosenberg
Note: For each schooling level we present three figures. The first figure (top) displays the levels of the outcome as a functionof cognitive and socio-emotional endowments. In particular, we present the average level of outcomes for different deciles ofcognitive and socio-emotional endowments. Notice that we define as “decile 1” the decile with the lowest values of endowmentsand “decile 10” as the decile with the highest levels of endowments. The second figure (bottom left) displays the average levelsof endowment across deciles of cognitive endowments. The bars in this figure indicates the fraction of individuals reporting therespective schooling level for each decile of cognitive endowment. The last figure (bottom right) mimics the structure of thesecond one but now for the socio-emotional endowment.
82
Figure 12: The Effect of Cognitive and Socio-emotional endowments on Mental Health at age 40(MCS-12)
A. All B. High School Dropout
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
SF
-12
Men
tal
-0.4
-0.2
0
0.2
0.4
0.6
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
SF
-12
Men
tal
-0.4
-0.2
0
0.2
0.4
0.6
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
SF-12 Mental
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
SF
-12
Men
tal
-0.4
-0.2
0
0.2
0.4
0.6
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
SF-12 Mental
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
SF
-12
Men
tal
-0.4
-0.2
0
0.2
0.4
0.6
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
SF
-12
Men
tal
-0.4
-0.2
0
0.2
0.4
0.6
Fra
ctio
n
0
0.1
0.2
0.3
0.4
0.5
0.6SF-12 Mental
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
SF
-12
Men
tal
-0.4
-0.2
0
0.2
0.4
0.6
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45SF-12 Mental
C. GED D. High School Graduate
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
SF
-12
Men
tal
-0.4
-0.2
0
0.2
0.4
0.6
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
SF
-12
Men
tal
-0.4
-0.2
0
0.2
0.4
0.6
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25SF-12 Mental
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
SF
-12
Men
tal
-0.4
-0.2
0
0.2
0.4
0.6
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45SF-12 Mental
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
SF
-12
Men
tal
-0.4
-0.2
0
0.2
0.4
0.6
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
SF
-12
Men
tal
-0.4
-0.2
0
0.2
0.4
0.6
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22SF-12 Mental
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
SF
-12
Men
tal
-0.4
-0.2
0
0.2
0.4
0.6
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22
SF-12 Mental
E. Some College F. Four-Year College
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
SF
-12
Men
tal
-0.4
-0.2
0
0.2
0.4
0.6
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
SF
-12
Men
tal
-0.4
-0.2
0
0.2
0.4
0.6
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
SF-12 Mental
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
SF
-12
Men
tal
-0.4
-0.2
0
0.2
0.4
0.6
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22SF-12 Mental
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
SF
-12
Men
tal
-0.4
-0.2
0
0.2
0.4
0.6
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
SF
-12
Men
tal
-0.4
-0.2
0
0.2
0.4
0.6
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45SF-12 Mental
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
SF
-12
Men
tal
-0.4
-0.2
0
0.2
0.4
0.6
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35SF-12 Mental
Note: For each schooling level we present three figures. The first figure (top) displays the levels of the outcome as a functionof cognitive and socio-emotional endowments. In particular, we present the average level of outcomes for different deciles ofcognitive and socio-emotional endowments. Notice that we define as “decile 1” the decile with the lowest values of endowmentsand “decile 10” as the decile with the highest levels of endowments. The second figure (bottom left) displays the average levelsof endowment across deciles of cognitive endowments. The bars in this figure indicates the fraction of individuals reporting therespective schooling level for each decile of cognitive endowment. The last figure (bottom right) mimics the structure of thesecond one but now for the socio-emotional endowment.
83
Figure 13: The Effect of Cognitive and Socio-emotional endowments on Depression at age 40(CES-D - Reverse Score)
A. All B. High School Dropout
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
CE
SD
-0.8
-0.6-0.4
-0.20
0.20.40.6
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
CE
SD
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
CESD
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
CE
SD
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
CESD
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
CE
SD
-0.8
-0.6-0.4
-0.20
0.20.40.6
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
CE
SD
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
Fra
ctio
n
0
0.1
0.2
0.3
0.4
0.5
0.6CESD
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
CE
SD
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45CESD
C. GED D. High School Graduate
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
CE
SD
-0.8
-0.6-0.4
-0.20
0.20.40.6
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
CE
SD
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25CESD
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
CE
SD
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45CESD
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
CE
SD
-0.8
-0.6-0.4
-0.20
0.20.40.6
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
CE
SD
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22CESD
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
CE
SD
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22
CESD
E. Some College F. Four-Year College
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
CE
SD
-0.8
-0.6-0.4
-0.20
0.20.40.6
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
CE
SD
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
CESD
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
CE
SD
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22CESD
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
CE
SD
-0.8
-0.6-0.4
-0.20
0.20.40.6
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
CE
SD
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45CESD
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
CE
SD
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35CESD
Note: For each schooling level we present three figures. The first figure (top) displays the levels of the outcome as a functionof cognitive and socio-emotional endowments. In particular, we present the average level of outcomes for different deciles ofcognitive and socio-emotional endowments. Notice that we define as “decile 1” the decile with the lowest values of endowmentsand “decile 10” as the decile with the highest levels of endowments. The second figure (bottom left) displays the average levelsof endowment across deciles of cognitive endowments. The bars in this figure indicates the fraction of individuals reporting therespective schooling level for each decile of cognitive endowment. The last figure (bottom right) mimics the structure of thesecond one but now for the socio-emotional endowment.
84
H The Effect of Endowments on Treatment Effects
These figures complement the results from the treatment effect tables.Each figure analyzes the average effects of education on the outcome of interest. For a
particular outcome, the effect is defined as the difference in the outcomes associated with twoschooling levels (not necessarily final or terminal schooling levels). For each pairwise comparisonof outcomes, let Y0 and Y1 denotes the outcomes associated with schooling levels 0 and 1,respectively. Importantly, each schooling level might provide the option to pursuing higherschooling levels. Final schooling levels do not allow for further options. Notice that in the figuresfinal schooling levels are highlighted using bold letters. For each pair of schooling levels 0 and 1,the first figure (top) presents E(Y1 − Y0|dC , dSE) where dC and dSE denote the cognitive andsocio-emotional deciles computed from the marginal distributions of cognitive and socio-emotionalendowments. E(Y1 − Y0|dC , dSE) is computed for those who reach the decision node involving adecision between levels 0 and 1. The second figure (bottom left) presents E(Y1 − Y0|dC) so thatthe socio-emotional factor is integrated out. The bars in this figure displays, for a given decile ofcognitive endowment, the fraction of individuals visiting the node leading to the educationaldecision involving levels 0 and 1. The last figure (bottom right) presents E(Y1 − Y0|dSE) as wellas, for a given decile of socio-emotional endowment, the fraction of individuals visiting the nodeleading to the educational decision involving levels 0 and 1.
85
H.1 Labor Market Outcomes
Figure 14: Average Treatment Effect of Education on Labor Market Participation, by DecisionNode and Endowment Levels
A. Dropping from HS vs. Graduating from HS B. HS Dropout vs. Getting a GED
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
(P
artic
ipat
ion)
0 -
Y1
Y
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
(P
artic
ipat
ion)
0 -
Y1
Y
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
(Participation)0 - Y1Y
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
(P
artic
ipat
ion)
0 -
Y1
Y
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
(Participation)0 - Y1Y
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
(P
artic
ipat
ion)
0 -
Y1
Y
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
(P
artic
ipat
ion)
0 -
Y1
Y
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4 (Participation)0 - Y1Y
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
(P
artic
ipat
ion)
0 -
Y1
Y
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45 (Participation)0 - Y1Y
C. HS Graduate vs. College Enrollment D. Some College vs. 4-year college degree
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
(P
artic
ipat
ion)
0 -
Y1
Y
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
(P
artic
ipat
ion)
0 -
Y1
Y
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22
0.24 (Participation)0 - Y1Y
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
(P
artic
ipat
ion)
0 -
Y1
Y
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22
0.24 (Participation)0 - Y1Y
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910 (
Par
ticip
atio
n)0
- Y
1Y
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
(P
artic
ipat
ion)
0 -
Y1
Y
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35 (Participation)0 - Y1Y
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
(P
artic
ipat
ion)
0 -
Y1
Y
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3 (Participation)0 - Y1Y
Notes: Each panel in this figure studies the average effect of an educational decision for those individualsvisting the decision node. Importantly, each schooling level might provide the option to pursue higherschooling levels, while final schooling levels do not provide an option. Final schooling levels are highlightedusing bold letters. For each educational decision node, the first figure (top) presents ∆ATE
j,j′′
(θ ∈ (dC , dSE)
)where dC and dSE denote the cognitive and socio-emotional deciles computed from the marginal distributionsof cognitive and socio-emotional endowments for the full population. The second figure (bottom left) presents∆ATE
j,j′′
(θC ∈ dC
)so that the socio-emotional factor is integrated out. The bars in this figure display the
fraction of individuals visiting the node in each decile of cognitive endowment. The last figure (bottomright) presents ∆ATE
j,j′′
(θSE ∈ dSE
)and the fraction of individuals visiting the node in a given decile of
socio-emotional endowment.
86
Figure 15: Average Treatment Effect of Probability of White-collar Employment at Age 30, byDecision Node and Endowment Levels
A. Dropping from HS vs. Graduating from HS B. HS Dropout vs. Getting a GED
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
(W
hite
colla
r)0
- Y
1Y 0
0.1
0.2
0.3
0.4
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
(W
hite
colla
r)0
- Y
1Y
0
0.1
0.2
0.3
0.4
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
(Whitecollar)0 - Y1Y
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
(W
hite
colla
r)0
- Y
1Y
0
0.1
0.2
0.3
0.4
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
(Whitecollar)0 - Y1Y
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
(W
hite
colla
r)0
- Y
1Y 0
0.1
0.2
0.3
0.4
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
(W
hite
colla
r)0
- Y
1Y
0
0.1
0.2
0.3
0.4
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4 (Whitecollar)0 - Y1Y
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
(W
hite
colla
r)0
- Y
1Y
0
0.1
0.2
0.3
0.4
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45 (Whitecollar)0 - Y1Y
C. HS Graduate vs. College Enrollment D. Some College vs. 4-year college degree
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
(W
hite
colla
r)0
- Y
1Y 0
0.1
0.2
0.3
0.4
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
(W
hite
colla
r)0
- Y
1Y
0
0.1
0.2
0.3
0.4
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22
0.24 (Whitecollar)0 - Y1Y
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
(W
hite
colla
r)0
- Y
1Y
0
0.1
0.2
0.3
0.4
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22
0.24 (Whitecollar)0 - Y1Y
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
(W
hite
colla
r)0
- Y
1Y 0
0.1
0.2
0.3
0.4
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
(W
hite
colla
r)0
- Y
1Y
0
0.1
0.2
0.3
0.4
Fra
ctio
n0
0.05
0.1
0.15
0.2
0.25
0.3
0.35 (Whitecollar)0 - Y1Y
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
(W
hite
colla
r)0
- Y
1Y
0
0.1
0.2
0.3
0.4
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3 (Whitecollar)0 - Y1Y
Notes: Each panel in this figure studies the average effect of an educational decision for those individualsvisting the decision node. Importantly, each schooling level might provide the option to pursue higherschooling levels, while final schooling levels do not provide an option. Final schooling levels are highlightedusing bold letters. For each educational decision node, the first figure (top) presents ∆ATE
j,j′′
(θ ∈ (dC , dSE)
)where dC and dSE denote the cognitive and socio-emotional deciles computed from the marginal distributionsof cognitive and socio-emotional endowments for the full population. The second figure (bottom left) presents∆ATE
j,j′′
(θC ∈ dC
)so that the socio-emotional factor is integrated out. The bars in this figure display the
fraction of individuals visiting the node in each decile of cognitive endowment. The last figure (bottomright) presents ∆ATE
j,j′′
(θSE ∈ dSE
)and the fraction of individuals visiting the node in a given decile of
socio-emotional endowment.
87
Figure 16: Average Treatment Effect of Education on (Log) Wages at Age 30, by Decision Nodeand Endowment Levels
A. Dropping from HS vs. Graduating from HS B. HS Dropout vs. Getting a GED
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
(Lo
g-w
ages
)0
- Y
1Y
-0.2
-0.1
0
0.1
0.2
0.3
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
(Lo
g-w
ages
)0
- Y
1Y
-0.2
-0.1
0
0.1
0.2
0.3
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
(Log-wages)0 - Y1Y
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
(Lo
g-w
ages
)0
- Y
1Y
-0.2
-0.1
0
0.1
0.2
0.3
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
(Log-wages)0 - Y1Y
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
(Lo
g-w
ages
)0
- Y
1Y
-0.2
-0.1
0
0.1
0.2
0.3
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
(Lo
g-w
ages
)0
- Y
1Y
-0.2
-0.1
0
0.1
0.2
0.3
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4 (Log-wages)0 - Y1Y
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
(Lo
g-w
ages
)0
- Y
1Y
-0.2
-0.1
0
0.1
0.2
0.3
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45 (Log-wages)0 - Y1Y
C. HS Graduate vs. College Enrollment D. Some College vs. 4-year college degree
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
(Lo
g-w
ages
)0
- Y
1Y
-0.2
-0.1
0
0.1
0.2
0.3
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
(Lo
g-w
ages
)0
- Y
1Y
-0.2
-0.1
0
0.1
0.2
0.3
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22
0.24 (Log-wages)0 - Y1Y
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
(Lo
g-w
ages
)0
- Y
1Y
-0.2
-0.1
0
0.1
0.2
0.3
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22
0.24 (Log-wages)0 - Y1Y
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
(Lo
g-w
ages
)0
- Y
1Y
-0.2
-0.1
0
0.1
0.2
0.3
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
(Lo
g-w
ages
)0
- Y
1Y
-0.2
-0.1
0
0.1
0.2
0.3
Fra
ctio
n0
0.05
0.1
0.15
0.2
0.25
0.3
0.35 (Log-wages)0 - Y1Y
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
(Lo
g-w
ages
)0
- Y
1Y
-0.2
-0.1
0
0.1
0.2
0.3
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3 (Log-wages)0 - Y1Y
Notes: Each panel in this figure studies the average effect of an educational decision for those individualsvisting the decision node. Importantly, each schooling level might provide the option to pursue higherschooling levels, while final schooling levels do not provide an option. Final schooling levels are highlightedusing bold letters. For each educational decision node, the first figure (top) presents ∆ATE
j,j′′
(θ ∈ (dC , dSE)
)where dC and dSE denote the cognitive and socio-emotional deciles computed from the marginal distributionsof cognitive and socio-emotional endowments for the full population. The second figure (bottom left) presents∆ATE
j,j′′
(θC ∈ dC
)so that the socio-emotional factor is integrated out. The bars in this figure display the
fraction of individuals visiting the node in each decile of cognitive endowment. The last figure (bottomright) presents ∆ATE
j,j′′
(θSE ∈ dSE
)and the fraction of individuals visiting the node in a given decile of
socio-emotional endowment.
88
Figure 17: Average Treatment Effect of Education on Present Value of Wages, by Decision Nodeand Endowment Levels
A. Dropping from HS vs. Graduating from HS B. HS Dropout vs. Getting a GED
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
(P
V W
ages
)0
- Y
1Y
-0.1
0
0.1
0.2
0.3
0.4
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
(P
V W
ages
)0
- Y
1Y
-0.1
0
0.1
0.2
0.3
0.4
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
(PV Wages)0 - Y1Y
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
(P
V W
ages
)0
- Y
1Y
-0.1
0
0.1
0.2
0.3
0.4
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
(PV Wages)0 - Y1Y
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
(P
V W
ages
)0
- Y
1Y
-0.1
0
0.1
0.2
0.3
0.4
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
(P
V W
ages
)0
- Y
1Y
-0.1
0
0.1
0.2
0.3
0.4
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4 (PV Wages)0 - Y1Y
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
(P
V W
ages
)0
- Y
1Y
-0.1
0
0.1
0.2
0.3
0.4
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45 (PV Wages)0 - Y1Y
C. HS Graduate vs. College Enrollment D. Some College vs. 4-year college degree
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
(P
V W
ages
)0
- Y
1Y
-0.1
0
0.1
0.2
0.3
0.4
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
(P
V W
ages
)0
- Y
1Y
-0.1
0
0.1
0.2
0.3
0.4
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22
0.24 (PV Wages)0 - Y1Y
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
(P
V W
ages
)0
- Y
1Y
-0.1
0
0.1
0.2
0.3
0.4
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22
0.24 (PV Wages)0 - Y1Y
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
(P
V W
ages
)0
- Y
1Y
-0.1
0
0.1
0.2
0.3
0.4
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
(P
V W
ages
)0
- Y
1Y
-0.1
0
0.1
0.2
0.3
0.4
Fra
ctio
n0
0.05
0.1
0.15
0.2
0.25
0.3
0.35 (PV Wages)0 - Y1Y
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
(P
V W
ages
)0
- Y
1Y
-0.1
0
0.1
0.2
0.3
0.4
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3 (PV Wages)0 - Y1Y
Notes: Each panel in this figure studies the average effect of an educational decision for those individualsvisting the decision node. Importantly, each schooling level might provide the option to pursue higherschooling levels, while final schooling levels do not provide an option. Final schooling levels are highlightedusing bold letters. For each educational decision node, the first figure (top) presents ∆ATE
j,j′′
(θ ∈ (dC , dSE)
)where dC and dSE denote the cognitive and socio-emotional deciles computed from the marginal distributionsof cognitive and socio-emotional endowments for the full population. The second figure (bottom left) presents∆ATE
j,j′′
(θC ∈ dC
)so that the socio-emotional factor is integrated out. The bars in this figure display the
fraction of individuals visiting the node in each decile of cognitive endowment. The last figure (bottomright) presents ∆ATE
j,j′′
(θSE ∈ dSE
)and the fraction of individuals visiting the node in a given decile of
socio-emotional endowment.
89
H.2 Physical Health Outcomes and Behaviors
Figure 18: Average Treatment Effect of Education on Probability of Being Obese (2006), byDecision Node and Endowment Levels
A. Dropping from HS vs. Graduating from HS B. HS Dropout vs. Getting a GED
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
(O
besi
ty)
0 -
Y1
Y
-0.25-0.2
-0.15-0.1
-0.050
0.050.1
0.15
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
(O
besi
ty)
0 -
Y1
Y
-0.25
-0.2
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
(Obesity)0 - Y1Y
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
(O
besi
ty)
0 -
Y1
Y
-0.25
-0.2
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
(Obesity)0 - Y1Y
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
(O
besi
ty)
0 -
Y1
Y
-0.25-0.2
-0.15-0.1
-0.050
0.050.1
0.15
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
(O
besi
ty)
0 -
Y1
Y
-0.25
-0.2
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4 (Obesity)0 - Y1Y
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
(O
besi
ty)
0 -
Y1
Y
-0.25
-0.2
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45 (Obesity)0 - Y1Y
C. HS Graduate vs. College Enrollment D. Some College vs. 4-year college degree
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
(O
besi
ty)
0 -
Y1
Y
-0.25-0.2
-0.15-0.1
-0.050
0.050.1
0.15
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
(O
besi
ty)
0 -
Y1
Y
-0.25
-0.2
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22
0.24 (Obesity)0 - Y1Y
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
(O
besi
ty)
0 -
Y1
Y
-0.25
-0.2
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22
0.24 (Obesity)0 - Y1Y
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
(O
besi
ty)
0 -
Y1
Y-0.25-0.2
-0.15-0.1
-0.050
0.050.1
0.15
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
(O
besi
ty)
0 -
Y1
Y
-0.25
-0.2
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35 (Obesity)0 - Y1Y
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
(O
besi
ty)
0 -
Y1
Y
-0.25
-0.2
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3 (Obesity)0 - Y1Y
Notes: Each panel in this figure studies the average effect of an educational decision for those individualsvisting the decision node. Importantly, each schooling level might provide the option to pursue higherschooling levels, while final schooling levels do not provide an option. Final schooling levels are highlightedusing bold letters. For each educational decision node, the first figure (top) presents ∆ATE
j,j′′
(θ ∈ (dC , dSE)
)where dC and dSE denote the cognitive and socio-emotional deciles computed from the marginal distributionsof cognitive and socio-emotional endowments for the full population. The second figure (bottom left) presents∆ATE
j,j′′
(θC ∈ dC
)so that the socio-emotional factor is integrated out. The bars in this figure display the
fraction of individuals visiting the node in each decile of cognitive endowment. The last figure (bottomright) presents ∆ATE
j,j′′
(θSE ∈ dSE
)and the fraction of individuals visiting the node in a given decile of
socio-emotional endowment.
90
Figure 19: Average Treatment Effect of Education on Probability of Being a Smoker, by DecisionNode and Endowment Levels
A. Dropping from HS vs. Graduating from HS B. HS Dropout vs. Getting a GED
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
(S
mok
er)
0 -
Y1
Y
-0.3
-0.2
-0.1
0
0.1
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
(S
mok
er)
0 -
Y1
Y
-0.3
-0.2
-0.1
0
0.1
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
(Smoker)0 - Y1Y
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
(S
mok
er)
0 -
Y1
Y
-0.3
-0.2
-0.1
0
0.1
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
(Smoker)0 - Y1Y
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
(S
mok
er)
0 -
Y1
Y
-0.3
-0.2
-0.1
0
0.1
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
(S
mok
er)
0 -
Y1
Y
-0.3
-0.2
-0.1
0
0.1
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4 (Smoker)0 - Y1Y
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
(S
mok
er)
0 -
Y1
Y
-0.3
-0.2
-0.1
0
0.1
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45 (Smoker)0 - Y1Y
C. HS Graduate vs. College Enrollment D. Some College vs. 4-year college degree
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
(S
mok
er)
0 -
Y1
Y
-0.3
-0.2
-0.1
0
0.1
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
(S
mok
er)
0 -
Y1
Y
-0.3
-0.2
-0.1
0
0.1
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22
0.24 (Smoker)0 - Y1Y
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
(S
mok
er)
0 -
Y1
Y
-0.3
-0.2
-0.1
0
0.1
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22
0.24 (Smoker)0 - Y1Y
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
(S
mok
er)
0 -
Y1
Y
-0.3
-0.2
-0.1
0
0.1
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
(S
mok
er)
0 -
Y1
Y
-0.3
-0.2
-0.1
0
0.1
Fra
ctio
n0
0.05
0.1
0.15
0.2
0.25
0.3
0.35 (Smoker)0 - Y1Y
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
(S
mok
er)
0 -
Y1
Y
-0.3
-0.2
-0.1
0
0.1
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3 (Smoker)0 - Y1Y
Notes: Each panel in this figure studies the average effect of an educational decision for those individualsvisting the decision node. Importantly, each schooling level might provide the option to pursue higherschooling levels, while final schooling levels do not provide an option. Final schooling levels are highlightedusing bold letters. For each educational decision node, the first figure (top) presents ∆ATE
j,j′′
(θ ∈ (dC , dSE)
)where dC and dSE denote the cognitive and socio-emotional deciles computed from the marginal distributionsof cognitive and socio-emotional endowments for the full population. The second figure (bottom left) presents∆ATE
j,j′′
(θC ∈ dC
)so that the socio-emotional factor is integrated out. The bars in this figure display the
fraction of individuals visiting the node in each decile of cognitive endowment. The last figure (bottomright) presents ∆ATE
j,j′′
(θSE ∈ dSE
)and the fraction of individuals visiting the node in a given decile of
socio-emotional endowment.
91
Figure 20: Average Treatment Effect of Education on Probability of Being a Heavy Drinker, byDecision Node and Endowment Levels
A. Dropping from HS vs. Graduating from HS B. HS Dropout vs. Getting a GED
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
(H
eavy
Drin
ker)
0 -
Y1
Y
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
(H
eavy
Drin
ker)
0 -
Y1
Y
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2 (Heavy Drinker)0 - Y1Y
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
(H
eavy
Drin
ker)
0 -
Y1
Y
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2 (Heavy Drinker)0 - Y1Y
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
(H
eavy
Drin
ker)
0 -
Y1
Y
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
(H
eavy
Drin
ker)
0 -
Y1
Y
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4 (Heavy Drinker)0 - Y1Y
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
(H
eavy
Drin
ker)
0 -
Y1
Y
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45 (Heavy Drinker)0 - Y1Y
C. HS Graduate vs. College Enrollment D. Some College vs. 4-year college degree
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
(H
eavy
Drin
ker)
0 -
Y1
Y
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
(H
eavy
Drin
ker)
0 -
Y1
Y
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22
0.24 (Heavy Drinker)0 - Y1Y
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
(H
eavy
Drin
ker)
0 -
Y1
Y
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22
0.24 (Heavy Drinker)0 - Y1Y
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
(H
eavy
Drin
ker)
0 -
Y1
Y
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
(H
eavy
Drin
ker)
0 -
Y1
Y
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
Fra
ctio
n0
0.05
0.1
0.15
0.2
0.25
0.3
0.35 (Heavy Drinker)0 - Y1Y
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
(H
eavy
Drin
ker)
0 -
Y1
Y
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3 (Heavy Drinker)0 - Y1Y
Notes: Each panel in this figure studies the average effect of an educational decision for those individualsvisting the decision node. Importantly, each schooling level might provide the option to pursue higherschooling levels, while final schooling levels do not provide an option. Final schooling levels are highlightedusing bold letters. For each educational decision node, the first figure (top) presents ∆ATE
j,j′′
(θ ∈ (dC , dSE)
)where dC and dSE denote the cognitive and socio-emotional deciles computed from the marginal distributionsof cognitive and socio-emotional endowments for the full population. The second figure (bottom left) presents∆ATE
j,j′′
(θC ∈ dC
)so that the socio-emotional factor is integrated out. The bars in this figure display the
fraction of individuals visiting the node in each decile of cognitive endowment. The last figure (bottomright) presents ∆ATE
j,j′′
(θSE ∈ dSE
)and the fraction of individuals visiting the node in a given decile of
socio-emotional endowment.
92
Figure 21: Average Treatment Effect of Education on Physical Health (PCS-12) at Age 40, byDecision Node and Endowment Levels
A. Dropping from HS vs. Graduating from HS B. HS Dropout vs. Getting a GED
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
(S
F-1
2 P
hysi
cal)
0 -
Y1
Y -0.2-0.1
00.10.20.30.40.50.60.7
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
(S
F-1
2 P
hysi
cal)
0 -
Y1
Y
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
(SF-12 Physical)0 - Y1Y
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
(S
F-1
2 P
hysi
cal)
0 -
Y1
Y
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
(SF-12 Physical)0 - Y1Y
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
(S
F-1
2 P
hysi
cal)
0 -
Y1
Y -0.2-0.1
00.10.20.30.40.50.60.7
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
(S
F-1
2 P
hysi
cal)
0 -
Y1
Y
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4 (SF-12 Physical)0 - Y1Y
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
(S
F-1
2 P
hysi
cal)
0 -
Y1
Y
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45 (SF-12 Physical)0 - Y1Y
C. HS Graduate vs. College Enrollment D. Some College vs. 4-year college degree
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
(S
F-1
2 P
hysi
cal)
0 -
Y1
Y -0.2-0.1
00.10.20.30.40.50.60.7
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
(S
F-1
2 P
hysi
cal)
0 -
Y1
Y
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22
0.24 (SF-12 Physical)0 - Y1Y
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
(S
F-1
2 P
hysi
cal)
0 -
Y1
Y
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22
0.24 (SF-12 Physical)0 - Y1Y
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
(S
F-1
2 P
hysi
cal)
0 -
Y1
Y -0.2-0.1
00.10.20.30.40.50.60.7
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
(S
F-1
2 P
hysi
cal)
0 -
Y1
Y
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Fra
ctio
n0
0.05
0.1
0.15
0.2
0.25
0.3
0.35 (SF-12 Physical)0 - Y1Y
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
(S
F-1
2 P
hysi
cal)
0 -
Y1
Y
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3 (SF-12 Physical)0 - Y1Y
Notes: Each panel in this figure studies the average effect of an educational decision for those individualsvisting the decision node. Importantly, each schooling level might provide the option to pursue higherschooling levels, while final schooling levels do not provide an option. Final schooling levels are highlightedusing bold letters. For each educational decision node, the first figure (top) presents ∆ATE
j,j′′
(θ ∈ (dC , dSE)
)where dC and dSE denote the cognitive and socio-emotional deciles computed from the marginal distributionsof cognitive and socio-emotional endowments for the full population. The second figure (bottom left) presents∆ATE
j,j′′
(θC ∈ dC
)so that the socio-emotional factor is integrated out. The bars in this figure display the
fraction of individuals visiting the node in each decile of cognitive endowment. The last figure (bottomright) presents ∆ATE
j,j′′
(θSE ∈ dSE
)and the fraction of individuals visiting the node in a given decile of
socio-emotional endowment.
93
H.3 Mental Health Outcomes
Figure 22: Average Treatment Effect of Education on Pearlin’s “Personal Mastery Scale” (1992),by Decision Node and Endowment Levels
A. Dropping from HS vs. Graduating from HS B. HS Dropout vs. Getting a GED
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
(P
earli
n)0
- Y
1Y
-0.6-0.4-0.2
00.20.40.60.8
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
(P
earli
n)0
- Y
1Y
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
(Pearlin)0 - Y1Y
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
(P
earli
n)0
- Y
1Y
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
(Pearlin)0 - Y1Y
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
(P
earli
n)0
- Y
1Y
-0.6-0.4-0.2
00.20.40.60.8
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
(P
earli
n)0
- Y
1Y
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4 (Pearlin)0 - Y1Y
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
(P
earli
n)0
- Y
1Y
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45 (Pearlin)0 - Y1Y
C. HS Graduate vs. College Enrollment D. Some College vs. 4-year college degree
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
(P
earli
n)0
- Y
1Y
-0.6-0.4-0.2
00.20.40.60.8
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
(P
earli
n)0
- Y
1Y
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22
0.24 (Pearlin)0 - Y1Y
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
(P
earli
n)0
- Y
1Y
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22
0.24 (Pearlin)0 - Y1Y
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910 (
Pea
rlin)
0 -
Y1
Y-0.6-0.4-0.2
00.20.40.60.8
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
(P
earli
n)0
- Y
1Y
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35 (Pearlin)0 - Y1Y
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
(P
earli
n)0
- Y
1Y
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3 (Pearlin)0 - Y1Y
Notes: Each panel in this figure studies the average effect of an educational decision for those individualsvisting the decision node. Importantly, each schooling level might provide the option to pursue higherschooling levels, while final schooling levels do not provide an option. Final schooling levels are highlightedusing bold letters. For each educational decision node, the first figure (top) presents ∆ATE
j,j′′
(θ ∈ (dC , dSE)
)where dC and dSE denote the cognitive and socio-emotional deciles computed from the marginal distributionsof cognitive and socio-emotional endowments for the full population. The second figure (bottom left) presents∆ATE
j,j′′
(θC ∈ dC
)so that the socio-emotional factor is integrated out. The bars in this figure display the
fraction of individuals visiting the node in each decile of cognitive endowment. The last figure (bottomright) presents ∆ATE
j,j′′
(θSE ∈ dSE
)and the fraction of individuals visiting the node in a given decile of
socio-emotional endowment.
94
Figure 23: Average Treatment Effect of Education on Self-Esteem (2006), by Decision Node andEndowment Levels
A. Dropping from HS vs. Graduating from HS B. HS Dropout vs. Getting a GED
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
(R
osen
berg
)0
- Y
1Y
-0.2
0
0.2
0.4
0.6
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
(R
osen
berg
)0
- Y
1Y
-0.2
0
0.2
0.4
0.6 Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
(Rosenberg)0 - Y1Y
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
(R
osen
berg
)0
- Y
1Y
-0.2
0
0.2
0.4
0.6 Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
(Rosenberg)0 - Y1Y
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
(R
osen
berg
)0
- Y
1Y
-0.2
0
0.2
0.4
0.6
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
(R
osen
berg
)0
- Y
1Y
-0.2
0
0.2
0.4
0.6 Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4 (Rosenberg)0 - Y1Y
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
(R
osen
berg
)0
- Y
1Y
-0.2
0
0.2
0.4
0.6 Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45 (Rosenberg)0 - Y1Y
C. HS Graduate vs. College Enrollment D. Some College vs. 4-year college degree
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
(R
osen
berg
)0
- Y
1Y
-0.2
0
0.2
0.4
0.6
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
(R
osen
berg
)0
- Y
1Y
-0.2
0
0.2
0.4
0.6 Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22
0.24 (Rosenberg)0 - Y1Y
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
(R
osen
berg
)0
- Y
1Y
-0.2
0
0.2
0.4
0.6 Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22
0.24 (Rosenberg)0 - Y1Y
Decile of Cognitive
12345678910
Decile of Socio-Emotional
1 2 34 5 6
7 8 910
(R
osen
berg
)0
- Y
1Y
-0.2
0
0.2
0.4
0.6
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
(R
osen
berg
)0
- Y
1Y
-0.2
0
0.2
0.4
0.6 Fra
ctio
n0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
(R
osen
berg
)0
- Y
1Y
-0.2
0
0.2
0.4
0.6 Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
Notes: Each panel in this figure studies the average effect of an educational decision for those individualsvisting the decision node. Importantly, each schooling level might provide the option to pursue higherschooling levels, while final schooling levels do not provide an option. Final schooling levels are highlightedusing bold letters. For each educational decision node, the first figure (top) presents ∆ATE
j,j′′
(θ ∈ (dC , dSE)
)where dC and dSE denote the cognitive and socio-emotional deciles computed from the marginal distributionsof cognitive and socio-emotional endowments for the full population. The second figure (bottom left) presents∆ATE
j,j′′
(θC ∈ dC
)so that the socio-emotional factor is integrated out. The bars in this figure display the
fraction of individuals visiting the node in each decile of cognitive endowment. The last figure (bottomright) presents ∆ATE
j,j′′
(θSE ∈ dSE
)and the fraction of individuals visiting the node in a given decile of
socio-emotional endowment.
95
Figure 24: Average Treatment Effect of Education on Mental Health (MCS-12) at Age 40, byDecision Node and Endowment Levels
A. Dropping from HS vs. Graduating from HS B. HS Dropout vs. Getting a GED
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
(S
F-1
2 M
enta
l)0
- Y
1Y -0.6
-0.4
-0.2
0
0.2
0.4
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
(S
F-1
2 M
enta
l)0
- Y
1Y
-0.6
-0.4
-0.2
0
0.2
0.4
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2 (SF-12 Mental)0 - Y1Y
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
(S
F-1
2 M
enta
l)0
- Y
1Y
-0.6
-0.4
-0.2
0
0.2
0.4
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2 (SF-12 Mental)0 - Y1Y
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
(S
F-1
2 M
enta
l)0
- Y
1Y -0.6
-0.4
-0.2
0
0.2
0.4
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
(S
F-1
2 M
enta
l)0
- Y
1Y
-0.6
-0.4
-0.2
0
0.2
0.4
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4 (SF-12 Mental)0 - Y1Y
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
(S
F-1
2 M
enta
l)0
- Y
1Y
-0.6
-0.4
-0.2
0
0.2
0.4
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45 (SF-12 Mental)0 - Y1Y
C. HS Graduate vs. College Enrollment D. Some College vs. 4-year college degree
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
(S
F-1
2 M
enta
l)0
- Y
1Y -0.6
-0.4
-0.2
0
0.2
0.4
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
(S
F-1
2 M
enta
l)0
- Y
1Y
-0.6
-0.4
-0.2
0
0.2
0.4
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22
0.24 (SF-12 Mental)0 - Y1Y
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
(S
F-1
2 M
enta
l)0
- Y
1Y
-0.6
-0.4
-0.2
0
0.2
0.4
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22
0.24 (SF-12 Mental)0 - Y1Y
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
(S
F-1
2 M
enta
l)0
- Y
1Y -0.6
-0.4
-0.2
0
0.2
0.4
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
(S
F-1
2 M
enta
l)0
- Y
1Y
-0.6
-0.4
-0.2
0
0.2
0.4
Fra
ctio
n0
0.05
0.1
0.15
0.2
0.25
0.3
0.35 (SF-12 Mental)0 - Y1Y
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
(S
F-1
2 M
enta
l)0
- Y
1Y
-0.6
-0.4
-0.2
0
0.2
0.4
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3 (SF-12 Mental)0 - Y1Y
Notes: Each panel in this figure studies the average effect of an educational decision for those individualsvisting the decision node. Importantly, each schooling level might provide the option to pursue higherschooling levels, while final schooling levels do not provide an option. Final schooling levels are highlightedusing bold letters. For each educational decision node, the first figure (top) presents ∆ATE
j,j′′
(θ ∈ (dC , dSE)
)where dC and dSE denote the cognitive and socio-emotional deciles computed from the marginal distributionsof cognitive and socio-emotional endowments for the full population. The second figure (bottom left) presents∆ATE
j,j′′
(θC ∈ dC
)so that the socio-emotional factor is integrated out. The bars in this figure display the
fraction of individuals visiting the node in each decile of cognitive endowment. The last figure (bottomright) presents ∆ATE
j,j′′
(θSE ∈ dSE
)and the fraction of individuals visiting the node in a given decile of
socio-emotional endowment.
96
Figure 25: Average Treatment Effect of Education on Depression (Reverse Score), by DecisionNode and Endowment Levels
A. Dropping from HS vs. Graduating from HS B. HS Dropout vs. Getting a GED
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
(C
ES
D)
0 -
Y1
Y
-0.4
-0.2
0
0.2
0.4
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
(C
ES
D)
0 -
Y1
Y
-0.4
-0.2
0
0.2
0.4
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
(CESD)0 - Y1Y
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
(C
ES
D)
0 -
Y1
Y
-0.4
-0.2
0
0.2
0.4
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
(CESD)0 - Y1Y
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
(C
ES
D)
0 -
Y1
Y
-0.4
-0.2
0
0.2
0.4
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
(C
ES
D)
0 -
Y1
Y
-0.4
-0.2
0
0.2
0.4
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4 (CESD)0 - Y1Y
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
(C
ES
D)
0 -
Y1
Y
-0.4
-0.2
0
0.2
0.4
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45 (CESD)0 - Y1Y
C. HS Graduate vs. College Enrollment D. Some College vs. 4-year college degree
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
(C
ES
D)
0 -
Y1
Y
-0.4
-0.2
0
0.2
0.4
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
(C
ES
D)
0 -
Y1
Y
-0.4
-0.2
0
0.2
0.4
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22
0.24 (CESD)0 - Y1Y
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
(C
ES
D)
0 -
Y1
Y
-0.4
-0.2
0
0.2
0.4
Fra
ctio
n
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22
0.24 (CESD)0 - Y1Y
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
Decile of Socio-Emotional12
34567
8910
(C
ES
D)
0 -
Y1
Y
-0.4
-0.2
0
0.2
0.4
Decile of Cognitive
1 2 3 4 5 6 7 8 9 10
(C
ES
D)
0 -
Y1
Y
-0.4
-0.2
0
0.2
0.4
Fra
ctio
n0
0.05
0.1
0.15
0.2
0.25
0.3
0.35 (CESD)0 - Y1Y
Decile of Socio-Emotional
1 2 3 4 5 6 7 8 9 10
(C
ES
D)
0 -
Y1
Y
-0.4
-0.2
0
0.2
0.4
Fra
ctio
n
0
0.05
0.1
0.15
0.2
0.25
0.3 (CESD)0 - Y1Y
Notes: Each panel in this figure studies the average effect of an educational decision for those individualsvisting the decision node. Importantly, each schooling level might provide the option to pursue higherschooling levels, while final schooling levels do not provide an option. Final schooling levels are highlightedusing bold letters. For each educational decision node, the first figure (top) presents ∆ATE
j,j′′
(θ ∈ (dC , dSE)
)where dC and dSE denote the cognitive and socio-emotional deciles computed from the marginal distributionsof cognitive and socio-emotional endowments for the full population. The second figure (bottom left) presents∆ATE
j,j′′
(θC ∈ dC
)so that the socio-emotional factor is integrated out. The bars in this figure display the
fraction of individuals visiting the node in each decile of cognitive endowment. The last figure (bottomright) presents ∆ATE
j,j′′
(θSE ∈ dSE
)and the fraction of individuals visiting the node in a given decile of
socio-emotional endowment.
I Additional Description of the Data
This analysis uses the 1979 National Longitudinal Survey of Youth (NLSY79), a nationallyrepresentative sample of men and women born in the years 1957-64.The NLSY79 includes botha randomly chosen sample of 6,111 U.S. youth and a supplemental sample of 5,295 randomlychosen Black, Hispanic, and non-Black non-Hispanic economically disadvantaged youths. Bothof these samples are drawn from the civilian population. In addition, there is a small sampleof individuals (1,280) who were enrolled in the military in 1979. The respondents were firstinterviewed in 1979 when they were 14-22 years of age. The NLSY surveyed its participantsannually from 1979 to 1992, and has surveyed biennially since 1992. The NLSY measures a
97
variety of later-life outcomes including income and health outcomes. The survey also measuresmany other aspects of the respondents’ lives, such as educational achievement tests, fertility,scores on achievement, high school grades, and demographic information. This paper uses thecore sample of males, which, after removing observations with missing covariates, contains 2242individuals.1
I.1 Outcomes
As a measure of physical health, we use the PCS-12 scale. The PCS-12 scale is the PhysicalComponent Summary obtained from SF-12.2 The SF-12 is designed to provide a measure ofthe respondent’s mental and physical health irrespective of their proclivity to use formal healthservices.Smoking at age 30 is considered as an additional measure of physical health . Smokingat age 30 is self reported and is defined as a binary variable for if the individual smoked daily atage 30.As a measure of mental health, the effect of education on self-esteem is considered. Selfesteem is measured using the Rosenberg’s Self-esteem scale (collected in 2006, when individualswere in their 40s). Respondents with a score above (below) 50 have better (worse) health thanthe typical person in the general U.S. population. Each one-point difference above or below 50corresponds to one-tenth of a standard deviation. For example, a person with a score of 30 istwo standard deviations away from the mean. We standardize the PCS-12 score to have meanzero and variance one in the overall population.
Rosenberg’s Self-Esteem Scale consists of 11 items which are answered on a 4-point scale(4 strongly agree, 3 agree, 2 disagree, 1 strongly disagree). An index score is constructed bysumming the scores from the items and standardizing the scores to have mean 0 and variance 1in the overall population (Rosenberg, 1965).
As a measure of physical health, we construct an obesity indicator based on BMI. BMI iscalculated as BMI=(Weight in Pounds * 703)/(Height in inches)2, and the obesity indicatortakes a value of one if the BMI is 30 and above, and zero otherwise.
I.2 Schooling Levels
We consider four different transitions and five final schooling levels. The transitions studiedare (i) enrolled in high school deciding between graduating from high school and dropping outfrom high school, (ii) high school dropouts deciding whether or not to get the GED, (iii) highschool graduates deciding whether or not to enroll in college, and (iv) college students decidingwhether or not to earn a 4-year degree. Consequently, the final schooling levels are (I) highschool dropout, (II) GED, (III) High school graduate, (IV) some college and (V) four-year collegedegree. Education at age 30 is treated as respondent’s final schooling level.3 Figure 1 shows thefive possible educational choices and their conditional structure. Thus, following the notationintroduced in Section 2.1, the indicator variable for a college graduate is defined as
H4 =
{1 if D0,1 = D1,3 = D3,4 = 10 otherwise
. (1)
1Respondents were dropped from the analysis if they did not have valid ASVAB scores, missed multiple rounds ofinterview, had infeasible educational histories where true education could not be interpreted, were missing controlvariables which could not be imputed, or had extreme and infeasible labor market histories. A number of imputationswhere made as necessary. Previous years’ covariates were used when covariates where not available for a needed year(such as region of residence). Responses from adjacent years were used for some outcomes when outcome variableswere missing at the age of interest. Mother’s education and father’s education were imputed when missing.
2SF-12 is a 12-question health survey designed by John Ware of the New England Medical Center Hospital (see(Ware, Kosinski, and Keller, 1996), and (Gandek, Ware, Aaronson, Apolone, Bjorner, Brazier, Bullinger, Kaasa,Leplege, Prieto, and Sullivan, 1998)).
3A negligible fraction of individuals change schooling levels after age 30.
98
I.3 Measurement System
The cognitive and socio-emotional factors in the model are identified from the educational choicesof agents as well as a supplemental measurement system of tests and other early-life outcomes.
Sub-tests from the Armed Services Vocational Aptitude Battery (ASVAB) are used asmeasures of cognitive ability. Specifically, we consider the scores from Arithmetic Reasoning,Coding Speed, Paragraph Comprehension, World Knowledge, Math Knowledge, and NumericalOperations.4
Many psychologists use a socio-emotional taxonomy called the Big Five (John, Robins,and Pervin, 2008). This is an organizing framework that categorizes personality traits into 5categories. The five traits are extraversion, agreeableness, conscientiousness, neuroticism, andopenness. A growing body of work suggests that these traits and other non-cognitive traitsplay key roles in academic success. Duckworth and Seligman (2005) find that self-disciplinepredicts GPA in 8th graders better than IQ. Duckworth, Quinn, and Tsukayama (2010) use threeunique studies to show that self-control predicts grades earned in middle school better than IQacross racial and socio-economic demographics. Farsides and Woodfield (2003), Conard (2006),and Noftle and Robins (2007) find that Big 5 traits positively predict grades and academicsuccess. See also Borghans, Golsteyn, Heckman, and Humphries (2011). These studies findpredictive power after controlling for previous grades or test scores. In these studies, the benefitsof personality traits are mediated through behaviors such as increased attendance or increasedacademic effort. A meta-analysis by Crede and Kuncel (2008) finds that study habits, skills, andattitudes have similar predictive power as standardized tests and previous grades in predictingcollege performance. They find that study skills are largely independent of high school GPA andstandardized admissions tests, but do have moderate correlations with personality traits.
Academic success (such as GPA) depends on cognitive ability, but also depends strongly onsocio-emotional traits such as conscientiousness, self-control, and self-discipline. This motivatesour identification strategy of including both a cognitive and socio-emotional factor in 9th gradeGPA, as much of the variance not explained through test scores has been shown to be relatedto socio-emotional traits. Thus, socio-emotional ability is measured by the socio-emotionalcontributions towards 9th grade GPA in reading, social studies, science, and math.5
To identify the non-cognitive factor, we include participation in minor risky or recklessactivity in 1979 in the measurement system for the socio-emotional endowment.6 In order toidentify the distribution of correlated factors, risky behavior is restricted to not load on thecognitive factor.
As a robustness check, we include five additional measures of adverse adolescent behaviorto check our interpreting of the non-cognitive factor.7 We consider violent behavior in 1979(fighting at school or work and hitting or threatening to hit someone), tried marijuana before
4A subset of these tests are used to construct the Armed Forces Qualification Test (AFQT) score, which iscommonly used as a measure of cognitive ability. AFQT scores are often interpreted as proxies for cognitive ability(Herrnstein and Murray, 1994). See the discussion in Almlund, Duckworth, Heckman, and Kautz (2011).
5As noted by Borghans, Golsteyn, Heckman, and Humphries (2011) and Almlund, Duckworth, Heckman, andKautz (2011), the principal determinants of the grade point average are personality traits and not cognition. Similarly,Duckworth and Seligman (2005) find that self-discipline predicts GPA in 8th graders better than IQ. See alsoDuckworth, Quinn, and Tsukayama (2010).
6Preliminary data analysis suggested this measure was the least dependent on cognitive endowments. This variableis a binary variable which is unity if an agent answers yes to any of the following questions in 1980: “Taken somethingfrom the store without paying for it,” “Purposely destroyed or damaged property that did not belong to you?,” “Otherthan from a store, taken something that did not belong to you worth under $50?,” and “Tried to get something bylying to a person about what you would do for him, that is, tried to con someone?”
7Gullone and Moore (2000) present a line of research which studies the relationship between personality traits andadolescent risk-behavior. Duckworth and Urzua (2009) study the relationship between personality and the number ofarrests between 14 and 17 years old and find that it is correlated with conscientiousness, agreeableness, and IQ. Basedon literature relating early behavior to non-cognitive traits, our five additional measures of early adverse behaviorhelp demonstrate that our socio-emotional factor is capturing traits that then explain these observed behaviors in anexpected manner.
99
age 15, daily smoking before age 15, regular drinking before age 15, and any intercourse beforeage 15. For violent behavior, we control for the potential effect of schooling.
I.4 Control Variables
The variables used to control for observable characteristics depend on the timing and natureof the decision being made. In every outcome, measure, and educational choice, we control forrace, broken home status, number of siblings, mother’s education, father’s education, and familyincome in 1979. We additionally control for region of residence and urban status at the time therelevant measure, decision, or outcome was determined.8 For log wages at age 30, we additionallycontrol for local economic conditions at age 30.
The educational choice models include additional choice-specific covariates. FollowingCarneiro, Heckman, and Vytlacil (2011), we control for both long run economic conditions,and current deviations from those conditions. By controlling for the long-run local economicenvironment, local unemployment deviations capture current economic shocks. The choice to geta GED additionally controls for the difficulty of the GED within the state of residence in 1988.9
The choices to enroll in college and graduate from college control for local 4-year college tuitionat age 17 and 22 respectively.10
GPA controls for the standard controls, except for region dummies which are not availableprior to 1979. The GPA model alternatively controls for urban status at age 14 and southernresidence at age 14. The ASVAB test scores control for the standard controls, age, and agesquared. As noted above, the ASVAB tests are estimated separately by education at the timeof the test. The risky behavior in 1979 model controls for the standard controls, age and agesquared. The risky behavior measure is also estimated by educational group, but due to datalimitations pools high school graduates and those enrolled in college in 1979.
Log wages at age 30 controls for race, parents’ education, broken home status, number ofsiblings, region of residence at age 30, and local unemployment rates at age 30. Smoking at age30 includes the same controls, but excludes unemployment rates. Physical health and Rosenbergself-esteem at age 40 control for the same variables as smoking, but use region of residence atage 40 rather than 30.
J Identification of latent factor model
Identification of our factor model follows from a more general identification of correlated factormodels presented in Williams (2011). He shows that the general system of five measures Y1, ..., Y5and two correlated factors θ = (θC , θSE) can be identified nonparametrically given one dedicatedmeasure of θC and one dedicated measure of θSE . This system of equations can be expressed as:
Y1 = α1θC + ε1
Y2 = β2θSE + ε2
Y3 = α3θC + β3θSE + ε3
Y4 = α4θC + β4θSE + ε4
Y5 = α5θC + β5θSE + ε5
8Based on the data, we assume that high school, GED certification, and college enrollment decisions occur at age17 while the choice to graduate from college is made at age 22.
9GED difficulty is proxied by the percent of high school graduates able to pass the test in one try given the state’schosen average and minimum score requirements.
10The cost of college, or the difficulty of earning a GED may affect an individual’s choice to graduate from highschool. In preliminary models, we found these “forward looking” variables to be statistically insignificant in the choiceto graduate from high school and they are excluded from the high school choice.
100
The proof for the non-parametric identification of the factor distribution can be found inCunha, Heckman, and Schennach (2010).
101
References
Almlund, M., A. Duckworth, J. J. Heckman, and T. Kautz (2011): “PersonalityPsychology and Economics,” in Handbook of the Economics of Education, ed. by E. A.Hanushek, S. Machin, and L. Woßmann, vol. 4. Elsevier, Amsterdam, Forthcoming.
Borghans, L., B. H. H. Golsteyn, J. J. Heckman, and J. E. Humphries (2011):“Identification Problems in Personality Psychology,” Personality and Individual Differences,51(Special Issue on Personality and Economics), 315–320, E. Ferguson, J.J. Heckman, and P.Corr, editors.
Carneiro, P., J. J. Heckman, and E. J. Vytlacil (2011): “Estimating Marginal Returnsto Education,” American Economic Review, 101(6), 2754–2781.
Conard, M. A. (2006): “Aptitude is not enough: How personality and behavior predictacademic performance,” Journal of Research in Personality, 40(3), 339–346.
Crede, M., and N. R. Kuncel (2008): “Study Habits, Skills, and Attitudes: The Third PillarSupporting Collegiate Academic Performance,” Prospectives on Psychological Science, 3(6),425–453.
Cunha, F., J. J. Heckman, and S. M. Schennach (2010): “Estimating the Technology ofCognitive and Noncognitive Skill Formation,” Econometrica, 78(3), 883–931.
Duckworth, A., and S. Urzua (2009): “Determinants of Success in Early Adulthood:Comparing the Effects of Intelligence and Big Five Personality Traits,” Unpublished manuscript,University of Pennsylvania, Department of Psychology.
Duckworth, A. L., P. D. Quinn, and E. Tsukayama (2010): “What No Child Left BehindLeaves Behind: The Roles of IQ and Self-Control in Predicting Standardized AchievementTest Scores and Report Card Grades,” Under review, Journal of Educational Psychology.
Duckworth, A. L., and M. E. P. Seligman (2005): “Self-Discipline Outdoes IQ in PredictingAcademic Performance of Adolescents,” Psychological Science, 16(12), 939–944.
Farsides, T., and R. Woodfield (2003): “Individual differences and undergraduate academicsuccess: The roles of personality, intelligence, and application,” Personality and IndividualDifferences, 34(7), 1225–1243.
Gandek, B., J. E. Ware, N. K. Aaronson, G. Apolone, J. B. Bjorner, J. E. Brazier,M. Bullinger, S. Kaasa, A. Leplege, L. Prieto, and M. Sullivan (1998): “Cross-Validation of Item Selection and Scoring for the SF-12 Health Survey in Nine Countries:Results from the IQOLA Project,” Journal of Clinical Epidemiology, 51(11), 1171–1178.
Gullone, E., and S. Moore (2000): “Adolescent risk-taking and the five-factor model ofpersonality,” Journal of Adolescence, 23(4), 393–407.
Herrnstein, R. J., and C. A. Murray (1994): The Bell Curve: Intelligence and ClassStructure in American Life. Free Press, New York.
Horn, J. L. (1965): “A Rationale and Test for the Number of Factors in Factor Analysis,”Psychometrika, 30(2), 179–185.
John, O. P., R. W. Robins, and L. A. Pervin (eds.) (2008): Handbook of Personality:Theory and Research. Guilford Press, New York, NY, third edn.
Noftle, E. E., and R. W. Robins (2007): “Personality Predictors of Academic Outcomes:Big Five Correlates of GPA and SAT Scores,” Personality Processes and Individual Differences,93(1), 116–139.
102
Pearlin, L. I., E. G. Menaghan, M. A. Lieberman, and J. T. Mullan (1981): “TheStress Process,” Journal of Health and Social Behavior, 22(4), 337–356.
Pearlin, L. I., and C. Schooler (1978): “The Structure of Coping,” Journal of Health andSocial Behavior, 19(1), 2–21.
Radloff, L. S. (1977): “The CES-D Scale,” Applied Psychological Measurement, 1(3), 385–401.
Rosenberg, M. (1965): Society and the Adolescent Self-Image. Princeton University Press,Princeton, NJ.
Ross, C. E., and J. Mirowsky (1989): “Explaining the Social Patterns of Depression: Controland Problem-solving or Support and Talking,” Journal of Health and Social Behavior, 30(2),206–219.
Ware, John E., J., M. Kosinski, and S. D. Keller (1996): “A 12-Item Short-Form HealthSurvey: Construction of Scales and Preliminary Tests of Reliability and Validity,” MedicalCare, 34(3), 220–233.
Williams, B. (2011): “Identification of Factor Models,” Unpublished manuscript, University ofChicago, Department of Economics.
103