Randall Reback

Barnard College

哥伦比亚大学

3009 Broadway
纽约, 纽约 10027
rr2165@columbia.edu

NONINSTRUCTIONAL SPENDING

IMPROVES NONCOGNITIVE

OUTCOMES: DISCONTINUITY

EVIDENCE FROM A UNIQUE

ELEMENTARY SCHOOL

COUNSELOR FINANCING SYSTEM

抽象的
Children’s noncognitive skills, 心理健康, 并-
havior are important predictors of future earnings and
educational attainment. Their behavior in the classroom
also affects their peers’ behavior and achievement. 那里
is limited prior evidence, 然而, concerning the im-
pact of school resources on student behavior. 一些
elementary schools employ counselors whose primary
purpose is to help improve students’ behavior, 精神的
健康, and noncognitive skill acquisition. This article
estimates regression discontinuity models exploiting Al-
abama’s unique ﬁnancing system for school counselors.
Alabama fully subsidizes counselor appointments for
all elementary schools, with the number of appoint-
ments based on schools’ prior year enrollments using
discrete enrollment cutoffs. The results suggest that
greater counselor subsidies reduce the frequency of dis-
ciplinary incidents but do not strongly inﬂuence mean
student achievement test scores. Increases in counselors
moderate relatively severe behavioral problems without
necessarily improving systemic behavior affecting class-
room learning.

C(西德:2) 2010 American Education Finance Association

105

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
e
d
p
A
r
t
我
C
e
–
p
d

我

F
/

5
2
1
0
5
1
6
8
9
1
8
3
e
d
p
2
0
1
0
5
2
5
2
0
1
p
d

乙
y
G
你
e
s
t

哦
n
0
8
S
e
p
e
米
乙
e
r
2
0
2
3

ELEMENTARY SCHOOL COUNSELOR FINANCING

介绍

1.
Young children’s noncognitive traits, 行为, and mental health are ex-
tremely important, as they are closely linked to their chance of future success.
Ratings of children’s self-control and self-esteem are strong predictors of aca-
demic success as well as future earnings conditional on schooling (赫克曼,
Stixrud, and Urzua 2006). Young children’s behavioral disorders have strong
negative effects on their test scores (Currie and Stabile 2006, 2007) and are
related to future cases of juvenile delinquency (Nagin and Tremblay 1999). 迪斯-
ruptive students also negatively inﬂuence peers’ behavior and learning (Figlio
2007; Aizer 2008; Carrell and Hoekstra 2008; Neidell and Waldfogel 2010).
While internalized disorders (例如, anxiety, depression) do not necessarily have
a direct negative inﬂuence on peer achievement (Neidell and Waldfogel 2010),
these disorders can be very costly because they substantially increase the prob-
ability that a student will repeat a grade (Currie and Stabile 2007). Researchers
have attributed noncognitive skill differentials as a source of the pay gap be-
tween general educational development (GED) recipients and traditional high
school graduates (Heckman and Rubinstein 2001; 赫克曼, Stixrud, 和
Urzua 2006) and the gap in college enrollment between men and women
(雅各布 2002). Differences in noncognitive traits may also substantially con-
tribute to socioeconomic gaps in school achievement and in future earnings.1
In light of the potentially enormous social returns to investments improv-
ing children’s noncognitive traits, 行为, and mental health, 研究人员
have become increasingly interested in learning which educational interven-
tions inﬂuence these outcomes. Changes in noncognitive traits and behavior
have been cited as potentially important effects from various schooling op-
系统蒸发散, including charter schools (Imbermann 2007), public schools of choice
(Cullen, 雅各布, and Levitt 2006), private schools (Figlio and Ludwig 2000),
and prekindergarten schools (Neidell and Waldfogel 2010).2 Relatively little is
已知的, 然而, about how traditional public schools can improve students’

Examining data from the 1958 和 1970 British birth cohorts, Blanden, Gregg, and Macmillan
(2006) ﬁnd that a growing return to noncognitive traits can partially explain rising income inequality
在英国.
Imbermann (2007) analyzes data from a large urban school district and ﬁnds that students shifting
from traditional public schools to newly created charter schools increase their attendance and incur
fewer disciplinary infractions. Cullen, 雅各布, and Levitt (2006) ﬁnd that students winning lotteries
allowing them to attend Chicago public high schools of their choice report lower rates of disciplinary
incidents and are arrested less frequently than losers of these lotteries. Figlio and Ludwig (2000)
ﬁnd that most behavioral differences between public and private school children are related to
differences in personal and family characteristics, 尽管, controlling for these characteristics,
religious private school students are less likely to be sexually active but are less likely to use birth
control when they are sexually active. Neidell and Waldfogel (2010) ﬁnd that preschool attendance
is associated with worse noncognitive outcomes during elementary school, such as externalized
problem behaviors; even so, they ﬁnd that, for both individual elementary students and their peers,
the cognitive beneﬁts of preschool attendance outweigh the negative effects associated with more
frequent externalized problem behaviors.

106

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
e
d
p
A
r
t
我
C
e
–
p
d

我

F
/

5
2
1
0
5
1
6
8
9
1
8
3
e
d
p
2
0
1
0
5
2
5
2
0
1
p
d

乙
y
G
你
e
s
t

哦
n
0
8
S
e
p
e
米
乙
e
r
2
0
2
3

Randall Reback

mental health and behavior. Lazear (2001) theorizes that small class sizes
may be particularly important because they reduce the frequency of episodes
of disruptive student behavior, and Dee and West (2008) ﬁnd evidence that
smaller eighth-grade class sizes improve students’ engagement in the subject.
Carrell and Carrell (2006) ﬁnd that intertemporal variation in the number
of counselors per student in elementary schools inﬂuenced rates of disci-
plinary infractions in a Florida school district. Aside from these studies, 那里
is remarkably little empirical evidence concerning the causal inﬂuence of edu-
cational policies on children’s noncognitive development and behavior. It may
be particularly important to better understand the impact of resources and
policies speciﬁcally designed to address students’ behavior and mental health.
There are more than forty thousand elementary school counselors em-
ployed in the United States, trained mental health professionals who are em-
ployed directly by public school districts and typically possess certiﬁcation
and graduate degrees in psychology, school psychology, or school counseling
(Reback 2009). Unlike high school counselors (who spend much of their time
advising students on their course selection, college applications, and career
choices), elementary counselors spend the vast majority of their time dealing
with students’ behavioral and mental health issues, usually interacting with
students either one on one or in small groups (Adelman and Taylor 2003).
Unlike special education psychologists, elementary school counselors serve
the general student population. A recent national study ﬁnds that there is
enormous variation in elementary schools’ employment of counselors and
that state-level policies strongly inﬂuence the provision of counseling services
(Reback 2009).

This article employs a regression discontinuity approach to investigate the
effects of elementary school counselor subsidies on students’ test scores and
行为. While past studies have used a regression discontinuity approach
to examine the importance of class size in schools (例如, Angrist and Lavy
1999; Hoxby 2000; Urquiola 2006),3 this is the ﬁrst regression discontinuity
study to examine the importance of student-to-staff ratios for noninstructional
职员. The analyses below exploit Alabama’s unique statewide policy of sub-
sidizing elementary counselors based on discrete enrollment cutoffs. 作为
2005, Alabama was one of only four states in the continental United States to
directly subsidize elementary school counselors and the only state to base these

3. Researchers have also employed regression discontinuity analyses to examine other topics related
to education. Some examples include Jacob and Lefgren’s (2004) study of the impact of summer
school and grade repetition, Ludwig and Miller’s (2007) study of the effects of the Head Start
程序, van der Klaauw’s (2002) and Kane’s (2003) studies of the impact of ﬁnancial aid on
college enrollment, and several studies examining the importance of primary school entry age (看
黑色的, Devereaux, and Salvanes 2008, McEwan and Shapiro 2008, and Elder and Lubotsky 2009
for recent examples and discussions of other entry age studies).

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
e
d
p
A
r
t
我
C
e
–
p
d

我

F
/

5
2
1
0
5
1
6
8
9
1
8
3
e
d
p
2
0
1
0
5
2
5
2
0
1
p
d

乙
y
G
你
e
s
t

哦
n
0
8
S
e
p
e
米
乙
e
r
2
0
2
3

107

ELEMENTARY SCHOOL COUNSELOR FINANCING

subsidies on discrete student enrollment ranges.4 For example, Alabama ele-
mentary schools automatically receive state funding for a half-time counselor if
their prior year enrollments were below 500 but receive funding for a full-time
counselor if enrollments were between 500 和 749. This enrollment cutoff
is not used for any other subsidy, and variation in school enrollments does not
appear to be inﬂuenced by the cutoffs for counselor subsidies; observations
just above or below the cutoffs are similar along observable dimensions, 和
schools are no more likely to be just above a cutoff than to be just below.
Using panel data for every Alabama elementary school, the analyses explore
whether student outcomes are inﬂuenced by the resulting arbitrary variation
in the number of subsidized counselors at the school. The results suggest that
additional counselor subsidies reduce the likelihood of disciplinary incidents,
such as weapon-related incidents and student suspensions, but do not strongly
affect mean student test scores. Increases in counselors moderate relatively
severe behavioral problems without necessarily improving systemic behavior
affecting classroom learning.

The next section reviews the literature related to children’s noncognitive
skills and mental health needs, their impact on classmates’ performance, 和
schools’ interventions targeting these areas. 部分 3 provides background
information on Alabama’s school counseling policies, 部分 4 描述了
school-level Alabama panel data set, 部分 5 describes the empirical models,
部分 6 presents the results, and section 7 concludes.

2. RELATED LITERATURE
Researchers have estimated that as many as 20 percent of young children “have
mental disorders with at least mild functional impairment” (USDHHS 1999)
but that 80 percent of children needing mental health services fail to receive
他们 (Kataoka, 张, and Wells 2002). This potential underprovision of
mental health services is not simply due to the lack of universal private health
insurance; young children and adolescents who lack private health insurance
are just as likely to receive mental health services as those with private insur-
安斯 (Glied et al. 1998). Children’s mental health problems and noncognitive
traits can have a profound impact on short-run and long-run outcomes. 考试-
ining data from the National Longitudinal Survey of Youth–1979, 赫克曼,
Stixrud, and Urzua (2006) ﬁnd that noncognitive skills are important predic-
tors of both future educational attainment and future earnings. Controlling
for educational attainment, they ﬁnd that noncognitive skill measures are even
better than cognitive measures for predicting the future wages of all but the

4. The other three states were Delaware, 乔治亚州, and Nevada (see Reback 2009).

108

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
e
d
p
A
r
t
我
C
e
–
p
d

我

F
/

5
2
1
0
5
1
6
8
9
1
8
3
e
d
p
2
0
1
0
5
2
5
2
0
1
p
d

乙
y
G
你
e
s
t

哦
n
0
8
S
e
p
e
米
乙
e
r
2
0
2
3

Randall Reback

least educated women (high school dropouts) and the most educated men
(four-year college graduates). Recent studies of attention-deﬁcit/hyperactivity
紊乱 (ADHD) among children in the United States and Canada reveal
substantial negative effects of the disorder on both test scores and educa-
tional attainment (Currie and Stabile 2006, 2007). Currie and Stabile (2007)
ﬁnd that untreated externalized disorders (例如, hyperactivity) have substantial
effects on test scores, while untreated internalized disorders (例如, anxiety, 的-
压力) are associated with as much as a one percentage point increase in the
probability that a student will repeat a grade. Using longitudinal data tracking
boys who attended schools in low socioeconomic areas of Montreal, 加拿大,
Nagin and Tremblay (1999) ﬁnd that physical aggression and “oppositional
behavior” during one’s youth are relatively strong predictors of future cases of
juvenile delinquency.

Reducing disruptive behavior could also be very important if a student’s
disruptive behavior substantially affects peers’ achievement and behavior. Four
recent studies ﬁnd compelling evidence that disruptive student behavior has
substantial negative peer effects. These studies all ﬁnd large negative effects
on peers’ academic achievement, yet each study uses a very different underly-
ing source of variation in student disruptions: students’ ADD (attention-deﬁcit
紊乱) diagnoses, their disruptive home lives, their nomenclatural disadvan-
塔盖斯, or their teachers’ survey responses concerning behavioral problems in
the classroom. Neidell and Waldfogel (2010) ﬁnd that elementary school teach-
ers’ negative ratings of their students’ social skills and behavior, 特别
externalized behavior problems, are negatively related to classmates’ academic
growth in both reading and math. Examining data from a large county in
Florida, Figlio (2007) identiﬁes peer effects of problematic behavior caused by
males with ﬁrst names that are relatively more common among females (例如,
as in the Johnny Cash song, boys named Sue). While these boys do not have sys-
tematically different outcomes during elementary grades, boys with typically
female names get into more disciplinary trouble than other boys during sixth
年级. A greater representation of these types of boys in sixth-grade classes
leads to lower academic performance for other students in the classroom and
leads to increased disciplinary incidents among other boys. Examining data
from Alachua County, Florida, public schools, Carrell and Hoekstra (2008)
ﬁnd that greater exposure to grade mates whose families have reported do-
mestic violence incidents adversely affects both students’ test scores and their
行为. They ﬁnd that exposure to students from these families is especially
harmful to the behavior of students from lower income families, male students,
and black female students. Examining a nationally representative sample of
elementary students and exploiting variation in the timing of the expansion of
insurance programs that help promote ADD diagnoses, Aizer (2008) ﬁnds that

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
e
d
p
A
r
t
我
C
e
–
p
d

我

F
/

5
2
1
0
5
1
6
8
9
1
8
3
e
d
p
2
0
1
0
5
2
5
2
0
1
p
d

乙
y
G
你
e
s
t

哦
n
0
8
S
e
p
e
米
乙
e
r
2
0
2
3

109

ELEMENTARY SCHOOL COUNSELOR FINANCING

greater exposure to undiagnosed students who later become diagnosed with
ADD has a negative effect on students’ academic performance. The majority
of students diagnosed with ADD receive medications to treat their conditions,
and many of them also receive counseling in clinical or nonclinical settings,
so the beneﬁcial peer effects of diagnoses might arise from medication, 县-
seling, or integrated treatments. Given that ADD diagnoses generally improve
the diagnosed student’s behavior but not the quality of this student’s academic
工作, Aizer’s ﬁndings suggest that it is the disruptive behavior of the student
that leads to negative peer effects for academic achievement.

Schools may be a relatively easy place to target children’s noncognitive
skills and mental health needs (Atkins et al. 2003; Weist, 埃文斯, and Lever
2003), and young children’s noncognitive skills may be particularly malleable
(赫克曼 2000). There is limited evidence, 然而, concerning whether
school-based interventions actually improve student behavior. Empirical stud-
ies of the impact of teachers, class size, and school expenditures tend to esti-
mate effects on test scores rather than behavioral outcomes, and there have
been relatively few studies examining noninstructional resources.5 There is a
growing literature discussing the merits of speciﬁc types of emotional learn-
ing programs (see Zins et al. 2004), but reviews of the impact of elementary
school counseling on students’ academic outcomes reveal that it is difﬁcult
to make strong, generalizable conclusions (Prout and Prout 1998; Whitson
and Sexton 1998). It is challenging to identify the impact of counseling on
student outcomes because students in these studies are typically not randomly
assigned to receive counseling. Recent studies showing connections between
student test score outcomes and speciﬁc elementary counseling programs in
the states of Florida (Brigman and Campbell 2003) and Washington (Sink and
Stroh 2003) base their ﬁndings on comparisons of students in treated and
nontreated schools, even though the intervention was not randomly assigned.
Even the best studies utilizing random assignment generally suffer from other
limitations such as small sample sizes, treatment via the researchers rather
than counselors, potentially biased survey responses determining treatment
成功, short duration of the intervention (often two weeks or less), 和/或

5. A few notable recent studies have identiﬁed the causal effects of public school resources on be-
havioral outcomes and/or examined the impact of noninstructional resources. Examining evidence
from the Project STAR class size experiment and from a nationally representative data set, Dee
and West (2008) ﬁnd evidence that elementary school class size reductions can improve short-run
measures of students’ initiative, and eighth-grade class size reductions can improve students’ en-
gagement with the relevant subject matter. Using a dynamic regression discontinuity model, Cellini,
费雷拉, and Rothstein (2008) ﬁnd evidence that the passage of bonds funding capital construction
in California school districts leads to increases in local house values without large effects on student
test scores. As described below, Carrell and Carrell (2006) ﬁnd a reduction in disciplinary incidents
given a greater supply of counselors per pupil in elementary schools in Alachua County, Florida.

110

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
e
d
p
A
r
t
我
C
e
–
p
d

我

F
/

5
2
1
0
5
1
6
8
9
1
8
3
e
d
p
2
0
1
0
5
2
5
2
0
1
p
d

乙
y
G
你
e
s
t

哦
n
0
8
S
e
p
e
米
乙
e
r
2
0
2
3

Randall Reback

tracking outcomes only shortly after the conclusion of the program (看, 例如,
Gerler, Kinney, and Anderson 1985; Larkin and Thyer 1999; 曼宁 1988;
Russell and Roberts 1979).

Carrell and Carrell (2006) conduct one of the most rigorous nonexper-
imental studies of the impact of counselors and the only prior study that
has speciﬁcally examined the effects of counselor-student ratios. 正在检查
school-level variation in student-counselor ratios in a Florida school district,
variation partially driven by the internship and practicum assignments of
counselors in training at the University of Florida, Carrell and Carrell ﬁnd
that greater elementary school counselor availability reduced rates of student
disciplinary problems. Exploiting within-school variation in counselor-student
ratios over various semesters, they ﬁnd that greater availability of counselors
reduces both the share of students involved in any disciplinary incident and
the likelihood of recidivism among students previously involved in a disci-
plinary incident, especially among black male students and students from
low-income families. Carrell and Carrell’s study may be viewed as comple-
mentary to the analyses below: they were able to identify student-level effects
for the recurrence of major disciplinary incidents of an unspeciﬁed nature,
whereas the analyses below identify school-level effects but can do so for a
variety of outcomes.

In another recent study (Reback 2009), I examine how states’ adoptions
of policies subsidizing or requiring elementary school counselors are related
to student outcomes and teachers’ survey responses concerning school cli-
mate. Cross-state evidence suggests that, controlling for students’ prior test
scores and a host of other factors, ﬁfth-grade students earn higher test scores
and self-report lower rates of emotional/behavioral problems if their states’
policies increase the provision of elementary school counseling. Triple dif-
ferences estimates exploiting both the timing and the targeted grade level of
counseling policy changes suggest that more aggressive statewide elementary
counseling policies reduce teachers’ concerns about students’ behavior. 那些
analyses are also complementary to the analyses below, as they are based
on nationally representative samples but either rely on cross-state evidence
or examine changes in subjective attitudes rather than changes in concrete
结果.

3. BACKGROUND ON ALABAMA’S FINANCING AND REQUIREMENTS

FOR ELEMENTARY SCHOOL COUNSELORS

The origins of Alabama’s current counselor subsidy policy date back to the
state’s foundation aid program for “instructional support units” (principals,
assistant principals, counselors, and librarians), which began in the 1988–89
school year. The state began to reimburse all public schools for the salaries

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
e
d
p
A
r
t
我
C
e
–
p
d

我

F
/

5
2
1
0
5
1
6
8
9
1
8
3
e
d
p
2
0
1
0
5
2
5
2
0
1
p
d

乙
y
G
你
e
s
t

哦
n
0
8
S
e
p
e
米
乙
e
r
2
0
2
3

111

ELEMENTARY SCHOOL COUNSELOR FINANCING

桌子 1. Prior Year Average Daily Membership (ADM) Ranges for State Subsidies of Instructional Support
Units in Alabama Elementary Schools, 2000–2007

Prior Year ADM

Principals

Assistant Principals

Counselors

Librarians

<263 0.5 263–439.99 440–499.99 500–659.99 660–749.99 750–879.99 880–999.99 1000–1099.99 1100–1249.99 1250–1319.99 1320–1499.99 ≥1500 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0.5 0.5 1 1 1.5 1.5 2 2 0.5 0.5 0.5 1 1 1.5 1.5 2 2 2.5 2.5 3 0.5 1 1.25 1.25 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 of enough of these staff members so that each school could satisfy the rec- ommended staff-student ratios Southern Association Colleges and Schools (SACS). The SACS requires as part its standards for cer- tifying schools. Throughout this article’s sample period, Alabama maintained foundation funding at SACS’s recommended levels year 2000, even though has more recently loosened some standards. Speciﬁc schools receive allocations, districts cover full cost these positions based on a statewide salary schedule, and state fully reimburses districts expenses. If hires experienced member, the district will greater funds from because will have to larger member. To allow plan upcoming year, instructional support units are awarded schools’ average daily membership (ADM) from ﬁrst forty days prior year. Table 1 displays ADM ranges elementary associated with various instructional support units, which translate into full-time-equivalent positions. a school earns than one unit librarians, is allowed use these additional librarian units hire either part-time librarians or aides. For the other three types positions—principals, assistant principals, and counselors—the can only use money toward salaries that particular type member; if fewer counselors its entitled counselor simply forfeits opportunity receive those funds. 112 l D o w n o a d e d f r o m h t t p : >
Randall Reback image

下载pdf