Selection Into, and Academic Benefits From, Middle School Dance Elective Courses Among Urban Youth

Abstract

Although research shows associations between adolescents general arts involvement and academic performance, little research documents links between enrollment in middle school dance elective courses and academic achievement, especially within low-income, urban populations. Further, differences between adolescents who do and do not have access to, or self-select into, middle school dance electives have yet to be identified. We prospectively followed a large (n = 31,332), ethnically diverse sample of children from preschool through 8th grade in Miami, Florida. About 7% of adolescents enrolled in a dance elective course at some point in middle school (6th- 8th grade), with the majority of those (68.8%) taking dance for only one year. Black students were more likely than White and Latinx students to attend middle schools that did not offer dance. When dance courses were available, males and Black students were less likely to select into a dance elective. Students who took dance in middle school showed greater initial social skills at age four and better prior academic achievement in elementary school compared with those who did not take dance. Importantly, controlling for all preexisting selection effects and prior academic achievement, dance engagement in middle school was associated with higher grade point averages and standardized test scores, better school attendance, and a lower likelihood of suspension during middle school, with stronger positive effects observed for taking dance electives for multiple years. Implications for future research and educational policy are discussed.

Full text

Psychology of Aesthetics, Creativity,
and the Arts
Selection Into, and Academic Benefits From, Middle
School Dance Elective Courses Among Urban Youth
Taylor V. Gara and Adam Winsler
Online First Publication, April 11, 2019. http://dx.doi.org/10.1037/aca0000250
CITATION
Gara, T. V., & Winsler, A. (2019, April 11). Selection Into, and Academic Benefits From, Middle
School Dance Elective Courses Among Urban Youth. Psychology of Aesthetics, Creativity, and
the Arts. Advance online publication. http://dx.doi.org/10.1037/aca0000250

Selection Into, and Academic Benefits From, Middle School Dance
Elective Courses Among Urban Youth
Taylor V. Gara
University of California, Irvine
Adam Winsler
George Mason University
Although research shows associations between adolescents general arts involvement and academic
performance, little research documents links between enrollment in middle school dance elective courses
and academic achievement, especially within low-income, urban populations. Further, differences
between adolescents who do and do not have access to, or self-select into, middle school dance electives
have yet to be identified. We prospectively followed a large (n⫽31,332), ethnically diverse sample of
children from preschool through 8th grade in Miami, Florida. About 7% of adolescents enrolled in a
dance elective course at some point in middle school (6th– 8th grade), with the majority of those (68.8%)
taking dance for only one year. Black students were more likely than White and Latinx students to attend
middle schools that did not offer dance. When dance courses were available, males and Black students
were less likely to select into a dance elective. Students who took dance in middle school showed greater
initial social skills at age four and better prior academic achievement in elementary school compared with
those who did not take dance. Importantly, controlling for all preexisting selection effects and prior
academic achievement, dance engagement in middle school was associated with higher grade point
averages and standardized test scores, better school attendance, and a lower likelihood of suspension
during middle school, with stronger positive effects observed for taking dance electives for multiple
years. Implications for future research and educational policy are discussed.
Keywords: academic achievement, adolescence, dance, educational policy, selection
Supplemental materials: http://dx.doi.org/10.1037/aca0000250.supp
Engagement in the visual and performing arts has long been
thought to provide numerous benefits, not only to individuals but
also to society and human cultures (National Endowment for the
Arts [NEA], 2012; Winner, Goldstein, & Vincent-Lancrin, 2013).
However, in recent years there has been limited opportunities for
students in the United States (U.S.) public schools to participate in
different types of art forms, particularly dance (Parsad, Spiegel-
man, & Coopersmith, 2012). Access to the arts is particularly low
among public secondary schools in low-income neighborhoods
(Parsad et al., 2012), despite evidence that the arts may be partic-
ularly helpful for students of color in poverty (Catterall, 2009;
Catterall, Dumais, & Hampden-Thompson, 2012).
The middle school period is an important time when children
can begin (or continue) their engagement in the arts because
students have the opportunity to enroll in full elective courses in
subject areas like music, dance, visual art, or drama. Entrance into
the arts during adolescence becomes more of a closed system in
that a certain degree of skill is needed, (i.e., auditions; McNeal,
1998) compared with entrance into the arts earlier in life. Further,
adolescence is a sensitive period of brain development when
exposure to different external experiences, like the arts, can en-
hance students’ long-term achievement, motivation, and/or emo-
tional well-being (Dahl, 2004; Jensen, 2004).
The current study, with a large (n⬎30,000) and ethnically
diverse sample of predominantly low-income students followed
longitudinally from preschool through 8th grade, addresses three
research goals. First, we descriptively report how many students in
our sample enrolled in a dance elective during middle school, and
for how many years. Second, we examine differential demo-
graphic, school readiness, and elementary academic achievement
profiles for students who do and do not (a) have access to a middle
Taylor V. Gara, School of Education, University of California, Irvine;
Adam Winsler, Department of Psychology, George Mason University.
This project was supported in part by an award from the Research: Art
Works program at the National Endowment for the Arts, Grant 15-3800-
7015. The opinions expressed in this paper are those of the author(s) and
do not represent the views of the Office of Research and Analysis or the
National Endowment for the Arts. This work was also supported by the
Early Learning Coalition of Miami-Dade/Monroe and the Children’s Trust.
The Trust is a dedicated source of revenue established by voter referendum
to improve the lives of children and families in Miami-Dade County. We
thank all the participating children, families, schools, and agencies.
Correspondence concerning this article should be addressed to Taylor V.
Gara, School of Education, University of California, 3200 Education
Building, University of California, Irvine, CA 92697-5500. E-mail:
tgara@uci.edu
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Psychology of Aesthetics, Creativity, and the Arts
© 2019 American Psychological Association 2019, Vol. 1, No. 999, 000
1931-3896/19/$12.00 http://dx.doi.org/10.1037/aca0000250
1

school that offers dance and (b) self select into dance elective
courses in middle school when they are available. Finally, we
statistically control for all observed selection effects to determine
if taking dance courses in middle school is associated with better
academic performance later in middle school.
Research on Benefits of the Arts
Researchers from different theoretical perspectives have ex-
plored how different forms of artistic experience (i.e., music,
dance, drama, and visual art) influence child development (Gard-
ner, 1990; Goldstein, Lerner, & Winner, 2017). Vygotsky and John
Dewey held similar philosophies about learning and the arts (Con-
nery, John-Steiner, & Marjanovic-Shane, 2010; Jackson, 2000).
Dewey’s philosophy described human engagement with the arts as
learning by doing, or art as experience. Similarly, Vygotsky em-
phasized art activities as experiences in social context where
children participate in joint artistic endeavors and internalize the
cultural symbol systems involved in the art forms (i.e., language,
music, dance) and use them as tools for thinking, behaving, and
regulating the inner world of the child’s mind (Winsler, Ducenne,
& Koury, 2011).
Studies have shown associations between participation in the
arts (music, dance, visual art, drama, multiarts) and positive child
outcomes. Children’s exposure to high-quality arts classes is re-
lated to decreased stress in the classroom measured via cortisol
levels, and enhanced positive emotion (Brown, Garnett, Anderson,
& Laurenceau, 2017; Brown & Sax, 2013). During early child-
hood, music activities and instruction are linked to higher nu-
meracy skills, attention regulation, and prosocial behavior years
later, even after controlling for parent– child home reading activ-
ities and other covariates (Williams, Barrett, Welch, Abad, &
Broughton, 2015). Drama activities are also related to children’s
positive social development in ways not always observed for
music activities. For example, Schellenberg (2004) tested for ef-
fects of music lessons on children’s development and reported that
the drama-lesson comparison group showed pre-to-post increases
in prosocial behavior not observed among the music groups.
Theater activities can also help build children’s social and emo-
tional competencies via enhanced theory of mind, emotional per-
spective taking, and adaptive emotion regulation strategies (Gold-
stein & Lerner, 2017; Goldstein, Tamir, & Winner, 2013;
Goldstein & Winner, 2012). Associations between in- or-out-of-
school artistic involvement and academic performance and student
well-being have also been noted among adolescents, especially
among those in poverty (Catterall et al., 2012; Elpus, 2013a).
Methodological Challenges
Although there is no shortage of research showing positive
effects of the arts on student academic and social outcomes, the
literature suffers from numerous critical methodological chal-
lenges. First and most important, the research linking arts involve-
ment and increased academic or developmental outcomes is
largely correlational or quasi-experimental (somewhat better but
still lacking sufficient control of selection variables), which only
shows (often weak) associations between arts participation and
positive developmental outcomes. There exist a few true experi-
mental studies that have randomly assigned students to arts par-
ticipation and control groups (the design needed to infer causality)
that have shown positive effects of arts engagement (Greenfader &
Brouillette, 2017; Holochwost et al., 2017; Lobo & Winsler, 2006;
Moreno et al., 2011; Rabinowitch & Meltzoff, 2017; Schellenberg,
2004). However, the majority of research in this area compare an
arts-exposed group with a naturally occurring comparison group of
children not exposed to the art form in question without suffi-
ciently accounting for selection factors.
Naturally occurring comparison groups are problematic because
they introduce selection bias, or omitted variable bias, into the
sample. That is, many important family and child factors are
related to both selection into the arts and positive academic or
social outcomes (Elpus, 2013b). For example, it is well docu-
mented that children from families in poverty or with low parental
education receive less exposure to arts-related activities and they
do not perform as well in school compared with children with
more financial and educational capital (Catterall, 2009; Catterall et
al., 2012; Child Trends Database, 2015; Elpus & Abril, 2011;
Foster & Jenkins, 2017; Hall, 2015; O’Hagan, 2014; Rabkin &
Hedberg, 2011; Winner et al., 2013).
Children who enroll in art electives during secondary school
typically have prior exposure to the arts, and/or are from middle to
upper class families with the monetary resources and social capital
to provide the attire, instruments, or art supplies necessary for
engagement (Eccles, 2005; Fredricks et al., 2002; Lareau, 2002;
McNeal, 1998). Further, children from more affluent families
typically academically outperform their peers before they enter
middle and high school, and therefore it is likely that other indi-
vidual traits or family factors are responsible for both the arts
engagement and the higher achievement, rather than there being a
direct causal effect from arts involvement to enhanced outcomes.
Indeed, differences in family background characteristics such as
socioeconomic status or parent education level account for just as
much if not more variance in student academic outcomes com-
pared to the unique effect of the arts (Elpus, 2013b; Foster &
Jenkins, 2017).
Underrepresentation of certain groups of students in the arts is
reported by Elpus and Abril (2011), who used data from the
Educational Longitudinal Study and created demographic profiles
of music- and non-music-involved high school students. Males,
English Language Learners (ELLs), Hispanics, children of parents
with a high school diploma or less, and those in the lowest
socioeconomic quartile were significantly underrepresented in mu-
sic courses (Elpus & Abril, 2011). Further, most high school music
students were from the highest socioeconomic backgrounds and
had the highest academic performance. In a separate study, after
controlling for demographics, prior academic achievement, school
fixed effects, and other covariates, Elpus (2013b) found no signif-
icant difference in the Scholastic Aptitude Test (SAT) standard-
ized math tests of high school music and non music students. The
current paper extends this literature by first revealing demographic
profiles of adolescents who select into dance during middle school.
Then, controlling for all observed selection differences, we show
whether dance exposure in middle school is associated with en-
hanced academic outcomes.
Another limitation of the arts education literature is that the link
between the arts and academic performance is not replicated
consistently, even with experimental or high-quality, quasi-
experimental studies (Costa-Giomi, 2004; Elpus, 2013b; Mehr,
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
2GARA AND WINSLER

Schachner, Katz, & Spelke, 2013). The meta-analysis, the Review-
ing Education and the Arts Project (REAP; Winner et al., 2013)
revealed that although many studies showed positive effects, other
empirical studies reported inconsistent, inverse, or no effects of
student engagement in multiarts in-or-out-of-school activities on
academic achievement (Winner et al., 2013). Other meta-analyses
sometimes report no causal relations between arts engagement and
academic performance (See & Kokotsaki, 2015; Winner & Coo-
per, 2000).
Although part of the replication inconsistency is likely due to
the selection bias problem discussed above, difficulty in replica-
tion may also be due to studies using very different definitions of
arts involvement (i.e., not differentiating between in- or out-of-
school involvement), or studies reporting the effects of general
(multi) arts engagement, without differentiating between artistic
disciplines. Each art form (music, visual art, drama, dance) is
likely linked with specific neurological, cognitive, social, and
behavioral outcomes, which cannot be captured without examining
each art form as its own entity (Winner et al., 2013). Student
achievement related to arts involvement also likely varies in di-
rection and magnitude depending on the age of the student and on
the academic domain examined (i.e., math, science, reading; Smi-
thrim & Upitis, 2005).
Given that it is impossible to randomly assign students to
arts-elective courses in middle school, a strong methodological
approach to examine outcomes associated with arts engagement in
middle school is to use a large-scale, prospective longitudinal
design that first identifies the many selection factors that differ-
entiate those that do and do not take arts elective courses in middle
school and then statistically control for these selection factors to
examine whether exposure to arts is associated with enhanced later
academic outcomes. We did that with a sample of more than
30,000 ethnically diverse, almost entirely low-income middle
school students who were followed longitudinally since preschool
in the context of the Miami School Readiness Project (MSRP;
Winsler et al., 2008; Winsler, Gara, Alegrado, Castro, & Tavas-
solie, 2019). We found large and important differences between
students that did and did not enroll in arts-related elective courses
(a combination of music, dance, visual art, and drama) in 6th, 7th,
or 8th grade. Black students, males, students with disabilities,
those previously retained in grade, and those not English proficient
were less likely to take arts electives in middle school. Impor-
tantly, students exposed to arts electives in middle school already
showed enhanced social and behavioral skills and stronger lan-
guage, motor, and cognitive skills 7 years earlier in preschool at
age 4 and stronger prior academic performance in fifth grade
(Winsler et al., 2019).
After controlling for the strong selection differences, however,
we found that students exposed to arts electives in middle school
had significantly higher grade point average (GPA), standardized
math and reading test scores, and decreased odds of school sus-
pension later in middle school, compared with students not ex-
posed to the arts during school. Unfortunately, only a multiarts
composite variable was examined in that study (Winsler et al.,
2019). As noted above, it is critical to examine specific types of
arts exposure given likely differential outcomes associated with
each art form (Winner et al., 2013). In the current study, we follow
the same approach with this large and ethnically diverse sample, to
examine and systematically control for selection factors to explore
the potential positive academic outcomes associated specifically
with dance exposure in middle school.
Research on Dance
Dance instruction is thought to foster an environment where
children can engage in higher cognitive processes such as problem
solving, creativity, and critical thinking when choreographing or
interacting with peers (Gilbert, 2006; Hanna, 2008). A dance
enriched environment incorporates multisensory curriculum
through auditory (music) and physical stimuli (body movement).
Such sensory and bodily experiences are believed to be essential to
children’s concept formation and repertoire of learned experiences
(Eisner, 2002). Dance may also be an exceptional artistic disci-
pline for children to engage in compared with drama, music, or the
visual arts because of the integration of music and movement—
which requires the activation of multiple senses at once (Brown,
Martinez, & Parsons, 2006; Metcalfe, 2016; Walker, 2016).
Early childhood movement and music classes give children
practice in controlling and regulating their motor behavior through
music (i.e., fast-slow, high-low; start-stop), and such experience
appears to enhance children’s behavioral self-control (Winsler et
al., 2011). Using a double-blind, randomized control design, Lobo
and Winsler (2006) assigned 40 Head-Start preschoolers to a dance
program or an attention control group. Preschoolers in the treat-
ment group improved significantly in their social competence and
decreased in internalizing and externalizing behavior problems as
rated by both parents and teachers, compared with a randomly
assigned, demographically similar control group.
Synchronized movement activities during early and middle
childhood also appear to be advantageous to children’s coopera-
tion, helping behavior, and positive affect. In a randomized control
experiment, Rabinowitch and Meltzoff (2017) found that joint
synchronized movement between two people (i.e., dancing) via a
swing-like apparatus increased cooperation among 4-year olds,
compared with a control group that experienced asynchronized
movement. Joint music activities also provide opportunities for
synchronized movement (tapping, singing, and dancing) that is
shown to increase cooperation and willingness to help others,
among 4-year-olds, compared with children in nonmusical joint
task conditions (Kirschner & Tomasello, 2010). Other well-
controlled studies have noted movement as a medium for syn-
chrony that leads to more helping behavior, smiling, eye contact
among, and social bonding among older children and for seniors
(61–108 years; Rossberg-Gempton, Dickinson, & Poole, 1999;
Tunçgenç & Cohen, 2018, 2016). Rossberg-Gempton et al.
showed that an innovative, highly structured intergenerational
12-week dance intervention for 30 min twice per week increased
or maintained children’s cooperation, direction following, leader-
ship skills, communication, sense of belonging, and awareness of
others (Rossberg-Gempton et al., 1999).
Children whose first language is not English may particularly
use music, dance, and drama as a system of nonverbal symbolism
to explore and exercise their creative and cognitive abilities (Con-
nery et al., 2010; Hanna, 2008). When strategically coupled with a
literacy curriculum, English learners in grades K–2 who partici-
pated in an in-school drama and dance intervention integrating
movement, gesture, and expression into classroom-based early
English literacy lessons reported greater gains in language acqui-
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
3
SELECTION INTO DANCE AND OUTCOMES

sition and science vocabulary, compared with a control group
(Greenfader & Brouillette, 2017; Morgan & Stengel-Mohr, 2014).
Another experimental study by McMahon, Rose, and Parks (2003)
found that a dance and literacy intervention improved the reading
skills of low-income, African American first graders in Chicago
public schools, compared with a control group that received tra-
ditional reading instruction.
Other than the social, emotional, and literacy benefits related to
dance engagement among younger and older individuals, much
less empirical evidence exists about how exposure to in-school
dance courses might enhance student’s academic and learning
outcomes during middle school. As mentioned earlier, this is
largely because of the difficulty in assigning randomized groups to
capture the isolated effects of dance engagement on other domains
(See & Kokotsaki, 2015; Winner et al., 2013). The empirical
evidence that dance engagement, in particular, positively affects
student academic outcomes is weak (Lanfredi, 2013; See & Kokot-
saki, 2015; Winner et al., 2013).
Yet, exposure to dance has been associated with greater creative
thinking skills among adolescents (Minton, 2003). Two experi-
mental studies reported that tailored dance instruction significantly
increased middle and high school students’ problem solving and
critical thinking skills, compared with a control group (Kim, 2007;
Park, 2007). However, generalizability to students in the U.S. is
limited in that both Kim (2007) and Park (2007) sampled Korean
students. Finally, small meta-analyses conducted by Keinänen,
Hetland, and Winner (2000) examined eight experimental and
quasi-experimental studies on dance engagement and reading and
nonverbal reasoning skills, finding little support that dance instruc-
tion increased student’s reading or nonverbal reasoning skills.
Access to, and Selection Into, Dance Education
Compared with levels of music, visual art, and theater programs,
dance education has been the least accessible form of arts pro-
gramming in U.S. public secondary schools (Parsad et al., 2012).
One factor that restricts access to dance electives in public sec-
ondary schools is limited funding for arts education programs. The
National Endowment for the Arts’ budget for arts education in the
U.S. has decreased considerably over the years (Hall, 2015). This
limits the resources school districts have to provide dance courses
(i.e., studio room, music equipment) and hire certified dance
instructors or teaching artists (Bonbright, 1999). Neighborhood
variation in the amount of tax revenue collected by local school
districts also limits students’ access to dance electives. Schools
located in lower-income areas receive less tax revenue which
affects the schools’ budget and the range of elective programs that
can be offered.
Another prominent factor influencing the accessibility of dance
programs in public secondary schools in the United States is that
arts programs—particularly dance education— have not been in-
cluded or rewarded within the evaluation and accountability sys-
tem of public schools (Kisida, Morrison, & Tuttle, 2017; Peterson,
2007). In 2003, the United States’ reauthorization of the Elemen-
tary and Secondary Education Act, The No Child Left Behind Act,
required that funding of public schools be dependent upon stu-
dents’ demonstrated proficiency on standardized assessments of
math and reading (Beveridge, 2009). Access to dance elective
courses in secondary schools at this time was largely overrun by
other programs focused on helping low-performing students pass
the annual proficiency exams. At some schools, this meant the
removal of certain arts education electives, like theatre. At other
schools with more resources, where dance and theater programs
still existed, such arts electives were often not an option for the
low-performing students who were forced to take remedial math
and reading tutoring electives (Beveridge, 2009; Peterson, 2007).
Thus, an important aim of the current study was to explore the
extent to which students have access to dance education programs
in middle school, and for those who attend a school that offers
dance, what individual selection factors determine which students
do and do not sign up for dance courses.
Adolescent participation in school-based dance courses is likely
influenced by a variety of individual, peer, family, and school
factors (Eccles, 2005; Fredricks et al., 2002; McCarthy, Ondaatje,
Zakaras, & Brooks, 2004; NEA, 2012). Opportunities for social-
ization with selected peer groups, for example, is often an impor-
tant reason why adolescents choose one elective over another, in
addition to alignment with both family and religious values
(Fredricks et al., 2002). Middle school is also an important devel-
opmental period when adolescents experiment with different social
and personal identities (i.e., “I am a dancer”), which can influence
their middle-and high-school trajectory and long-term career
choice (Wigfield & Eccles, 2002). The middle school years are
critical in determining adolescents’ identity development and long-
term trajectories in the arts because these are the first academic
years (6th– 8th) when students can enroll in full arts elective
courses with limited skill.
When funding for arts programs is available, the public-school
setting is an ideal environment where children from all social
backgrounds can receive access to the arts (Beveridge, 2009;
Chapman, 2004). According to a national survey administered by
Child Trends Database (2015), among adolescents enrolled in
schools with arts programs between 1991 and 2013, about half
(48% to 55%) engaged in a performing arts program in any given
year, and participation declined in the higher grades. According to
a nationwide survey of adults, fewer than 5% of adults in the
United States (18 or older) reported receiving dance instruction in
a school setting (NEA, 2013) and the proportion of adults who
received dance instruction as a child varied from 9.6% to 17%
(Hall, 2015; Rabkin & Hedberg, 2011).
Despite the small proportion of adults who reported dancing
during childhood (NEA, 2013), researchers (O’Neill, Pate, &
Liese, 2011) have reported that dance is a common activity among
adolescents according to self-report measures. O’Neill et al. (2011)
sampled 3,598 adolescents, ages 12–19 years and found that ap-
proximately 21% of U.S. adolescents participated in out-of-school
dance activities in the month prior to data collection. Self-selected
dancing experiences were most prevalent among adolescents from
Black and Hispanic backgrounds, compared with non-Hispanic
Whites (O’Neill et al., 2011). Interestingly, O’Neill et al. (2011)
also reported that adolescents from the lowest-income families
reported engaging in more dance than adolescents from higher-
income families.
The findings reported by O’Neill et al. (2011), using self-report
measures of dance engagement largely outside of school, run
counter to the literature stating children from higher-income fam-
ilies are more likely to engage in the arts compared with children
from lower-income backgrounds (Elpus & Abril, 2011; Foster &
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
4GARA AND WINSLER

Jenkins, 2017; Hall, 2015). Perhaps when examining dance as a
single subject (rather than an any-arts composite), and adolescents
are given the autonomy to choose the type of activity they want to
engage in, adolescents’ selection into dance activities does not
follow the notion that arts engagement is limited to children from
higher-income families. Further, ethnic and income differences in
dance participation during school may not be observed in ethni-
cally diverse, lower-income, urban communities— given the di-
verse and culturally rich heritage of many forms of dance (other
than ballet) prevalent within such communities (Hanna, 2015;
NEA, 2013). The current study adds to this literature by reporting
the frequency of, and selection factors involved in, adolescents’
participation in middle school dance courses overall, and during
6th, 7th, and 8th grade, a period, according to O’Neill et al. (2011),
in which adolescents are actively choosing to engage in dance
activities.
Large gender differences favoring females are prevalent in
dance participation during early childhood, adolescence, and adult-
hood (Bucknavage & Worrell, 2005; Dumais, 2006; Hall, 2015;
O’Neill et al., 2011). Male participation in dance may be stigma-
tized by stereotypes that challenge male identity and masculinity,
especially among males who partake in Western European dance
(i.e., ballet). Noteworthy, however, is that gender stereotypes
about dance may not carryover to Afro-Caribbean and Latinx
students, where male and partner dance are normalized in social
engagement and cultural sharing (NEA, 2013). In addition to
gender and ethnicity, our study adds to the literature by examining
other factors, including disability/special education status, English
proficiency, ELL status, initial school readiness (cognitive, social,
behavioral, and motor skills) at kindergarten entry (7 years before
middle school), and prior academic performance as potential self-
selection factors into dance during middle school.
The Current Study
We assessed more than 30,000 ethnically diverse and largely
low-income preschoolers for school readiness at age four, and then
followed them longitudinally as they progressed through public
middle school (Winsler et al., 2008, 2019). We had three goals.
First, we descriptively report the number of students enrolling in
dance electives during middle school (6th, 7th, and 8th grade).
Second, we carefully examine numerous child and family factors
(age-4 school readiness, academic performance in 5th grade, gen-
der, ethnicity, special education, poverty, and ELL status associ-
ated with both access to dance electives (for all middle schools)
and selection into dance electives when they were available (only
in middle schools that offered dance). Our ultimate goal was to
statistically control for all observed selection effects to test
whether in-school dance courses are linked to a variety of authen-
tic, school-based measures of student academic performance
(GPA, retention, suspension, standardized math and reading test
scores, and attendance) during 6th, 7th, and 8th grade.
We add to the existing literature in several ways. First, we
examine multiple, novel selection factors associated with middle
school dance (i.e., age-4 school readiness, academic performance
in 5th grade, gender, ethnicity, special education, poverty, and
ELL status). Second, we isolate potential effects of in-school
dance engagement on student academic outcomes. (Rather than a
combination of both in- and out-of-school participation, or dance
combined with sports or other forms or art). Third, we have a large
ethnically diverse and low-income sample, including many Latinx
students, an understudied group (Garcia & Jensen, 2009). Fourth,
our indicator of dance engagement is from official school data as
opposed to less reliable student or parent retrospective reports.
Finally, our strong quasi-experimental, prospective longitudinal
design carefully controls for observed selection factors and has the
potential to provide some of the strongest scientific evidence to
date (short of random assignment) for examining associations
between dance and academic performance. We ask the following
research questions:
1. What proportion of our sample enrolls in dance electives
during middle school (6th, 7th and 8th grade), and what
proportion of the middle schools attended by children offer
dance electives?
2. What are the preexisting selection factors that predict enroll-
ment in dance electives in middle school? That is, how do
students who do and do not take dance electives in middle
school differ in terms of demographic variables, school read-
iness skills at kindergarten entry, and prior elementary
school academic performance? This question is examined
from two perspectives:
A. Access – For the entire sample, who is accessing dance
courses in middle school?
B. Selection/Choice – Limiting the sample to only students
who attend a middle school that offers dance, which
students are actively choosing to take dance electives
when available?
3. Controlling for all preexisting selection factors, are dance
classes associated with enhanced concurrent and later aca-
demic outcomes (i.e., GPA, retention, suspension, standard-
ized math and reading scores, and attendance) for students in
middle school?
Method
Participants
We used data from the Miami School Readiness Project (MSRP;
Winsler et al., 2008, 2019), a large-scale, university-community
partnership using a prospective, cohort-sequential, longitudinal
design. Research for this project received institutional review
board approval. Five cohorts of ethnically diverse, predominantly
low-income, 4-year-old children were comprehensively and indi-
vidually assessed for school readiness in prekindergarten between
academic years 2002–2003 to 2006 –2007. Children make up es-
sentially the entire Miami-Dade County population (92%) of those
who received subsidies to pay for childcare or attended public
school pre-K programs at age four. Thus, the sample does not
include (a) low-income children who were in Head Start programs
only, (b) children of any income who received only parental care
at age four, or (c) wealthier families who paid fully for their own
childcare at age four. Our research team was responsible for
administering the school readiness assessments to the children at
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
5
SELECTION INTO DANCE AND OUTCOMES

age 4 (training the assessors and teachers, and distributing and
processing the parent and teacher surveys etc.; see Winsler et al.,
2008, 2012). Once the children got to the public school system in
kindergarten, we received secondary administrative, de-identified,
data sets for the annual school outcomes and eventual elective
course selections for students.
Students who later entered the local public school system were
followed longitudinally every year as they progressed through
school. With the help of the school system, children were carefully
matched/linked according to their unique IDs and followed even if
they moved to another school as long as they didn’t leave the
school district. The research team received de-identified, confiden-
tial administrative school record data for the children who were
still in the school system each year. The sample was largely
progressing through middle school (6th– 8th grade) at the time of
this study. Attrition (or failure to initially link/match) for this
mobile, low-income sample from prekindergarten (pre-K) to kin-
dergarten was about 20%, but after reaching the public school
system, regular longitudinal attrition was only about 3% to 5% per
year.
We had data on 31,332 total children who had completed either
6th, 7th, or 8th grade by the academic year 2013–2014. Because of
cohort-based attrition (i.e., some of the children were not old
enough to have reached 7th or 8th grade), our sample had 6th grade
data for all five cohorts, 7th grade data for four cohorts, and 8th
grade data for three cohorts. Thus, we had 16,392 students with
8th-grade data, 23,788 with 7th-grade data, and 30,413 with 6th-
grade data.
The background characteristics, average school readiness, and
elementary school performance of our sample is shown in Table 1.
Children were 51.8% male, and the ethnic breakdown was 60.9%
Latinx, 32.1% Black, 6.4% White, and .06% Asian/Other/Mixed.
Slightly more than half were considered ELL by the district
(defined by having reported speaking another language than Eng-
lish at home and testing below English for Speakers of Other
Languages (ESOL) exit levels on an English proficiency test). By
5th grade, only 5% were still considered limited in their English
proficiency. The sample was largely in poverty (81% received
free/reduced lunch at school in 6th grade), and 16% received some
kind of special education services in 6th grade.
Measures
Predictor variables/covariates. Background information and
the other predictor variables in the section below were collected
via administrative school records with the exception of the school
readiness assessments which were given directly by the research-
ers.
ELL status in kindergarten. ELL status was acquired from
parent-reported home language at kindergarten entry. Those who
reported predominantly speaking another language at home were
considered ELLs during the kindergarten year by the school sys-
tem.
English proficient in 6th grade. Students classified as ELL
by the school district are assessed each year for English profi-
ciency with the Comprehensive English Learner Assessment
(CELLA; Educational Testing Service, 2005). The CELLA as-
sesses aural/oral, writing, and reading skills in English. Raw scores
place children in one of the five ordinal ESOL levels. The ESOL
levels are marked 1–5, with levels 1 and 2 indicating beginning
English learners who still have much difficulty, levels 3 and 4
being advanced stages of English learning, and level 5 considered
sufficiently proficient in English to exit the ESOL program. Stu-
dents must also reach a minimum threshold of performance on the
high stakes (English) reading Florida Comprehensive Achieve-
ment Test (FCAT; Human Resources Research Organization &
Harcourt Assessment, 2007), in addition to reaching ESOL Level
5, to exit the ESOL program (Miami-Dade County Public Schools,
2008). We examined students’ ESOL level in 6th grade, and those
at level 5 (and those who were never considered ELLs) were
considered English proficient (“1”) and those with a value less
than 5 received a “0.”
Poverty status in 6th grade. Free/reduced lunch (FRL) re-
ceipt in 6th grade served as a proxy for poverty status. Children
from low-income families are eligible for free or reduced-price
lunch (130% of the Federal Poverty Line and 185% of the FPL,
respectively) in the public-school system. Children who received
free or reduced-price lunch received a “1” versus “0.”
Table 1
Demographic Information of the Sample (N ⫽31,322)
Characteristic N(%)
Grade
Sixth grade 30,413
Seventh grade 23,788
Eighth grade 16,392
Gender
Male 16,144 (51.8)
Female 15,021 (48.2)
Ethnicity
White/Other 1,989 (6.40)
Hispanic 18,971 (60.90)
Black 10,000 (32.10)
Asian/Pacific Islander 185 (.60)
Poverty status in sixth grade
Received free/reduced lunch 24,636 (81)
Did not receive free/reduced lunch 5,775 (19)
English language learner (ELL) in kindergarten
ELL 17,978 (57.40)
Not ELL 13,348 (42.60)
English proficiency in sixth grade
English proficient 28,300 (95)
Not English proficient 1,477 (5)
Disability status in sixth grade
Disability 4,921 (16.20)
Nondisabled 25,502 (83.80)
Elementary school retention
Retained at least once 4,206 (13.40)
Never retained 27,126 (86.60)
School readiness skills (nat. percentiles) M(SD)
DECA
Social skills (TPF) 59.40 (28.00)
Behavioral concerns 46.34 (29.51)
LAP-D
Fine motor 56.62 (28.87)
Gross motor 68.15 (28.90)
Cognitive 52.31 (30.31)
Language 44.55 (30.69)
Fifth grade academic achievement
GPA (5-pt scale) 4.09 (.57)
Math (100–500 scale) 284.95 (68.69)
Reading (100–500 scale) 268.45 (61.30)
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
6GARA AND WINSLER

Disability status in 6th grade. Students were coded for
whether they had a primary exceptionality in 6th grade. Codes
included the following groups: intellectual disability, speech/lan-
guage disorder, visually impaired, deaf or hard of hearing, specific
learning disabled, dual-sensory impaired, autistic, severely emo-
tionally disturbed, traumatic brain injured, or other health im-
paired. If any of these codes were present in 6th grade, children
were coded a “1” ⫽yes (vs. “0”). Gifted students were coded as
a “0.”
Retention in elementary school. Four criteria had to be met
for a child to be considered retained at some point during elemen-
tary school. First, the child had to enter kindergarten (or their first
grade in the district) on time according to their birth date. Second,
the child had to complete a grade, as demonstrated by having final
grades for that grade level. Third, the child had to appear in the
same grade for a second time the following academic year accord-
ing to the school system. Last, the child had to have grades for the
second, subsequent academic year in that repeated grade. Children
who were ever retained during elementary school (kindergarten–
5th grade) received a “1” (vs. 0).
5th grade GPA. We used students’ 5th grade GPA as an
indicator of prior academic competence (before middle school). At
the end of the academic year, students received grades from their
teachers for all subject areas. Fifth grade subjects might include,
for example, science, social studies, music, reading, language arts,
English as a second language, math, and physical education. Sub-
ject enrollment varied by student. Grades were changed to a
5-point scale, where 5 ⫽A, 4 ⫽B, 3 ⫽C, 2 ⫽D, 1 ⫽F. A
continuous composite GPA score was created by averaging all
grades received across subjects in the 5th grade, resulting in an
overall GPA for 5th grade.
5th grade math and reading scores. In 5th grade, students
were required to take the state-wide, high-stakes Florida Compre-
hensive Achievement Test (FCAT; Human Resources Research
Organization & Harcourt Assessment, 2007). Questions are in both
multiple-choice and short-answer formats. Total scale scores for
math are included in the present study (range ⫽100 –500; Cron-
bach’s alpha ⫽.90). Just math scores were used for multivariate
regression analyses because math and reading were highly corre-
lated (r⫽.88) to avoid multicollinearity.
Cognitive, language, and motor skills at school entry.
Children’s cognitive, language, and fine and gross motor skills
were measured at age four with the Learning Accomplishment
Profile–Diagnostic (LAP-D; Nehring, Nehring, Bruni, & Ran-
dolph, 1992), a norm-referenced, standardized developmental as-
sessment administered individually to children in their pre-K ac-
ademic year. Children were typically assessed once in the
beginning (September/October) of the 4-year-old year and once at
the end (April/May). For certain years/cohorts, 3-year-olds were
also assessed in the middle of the academic year. We used chil-
dren’s latest available assessment for children who were assessed
multiple times. For children in subsidized childcare, MA-level
bilingual assessors administered the LAP-D assessment in either
Spanish or English (whichever was the child’s strongest language),
and for those in public school pre-K programs, the child’s teacher
administered the assessment after receiving the same rigorous
training from the publishers. Internal consistency reliability for the
LAP-D within the sample ranged from .92 to .95 (Winsler et al.,
2008). National percentiles scores are used to increase interpret-
ability.
Social skills and behavior problems at school entry.
Children’s social skills and behavior problems were measured
(same time periods as above) using the Devereux Early Childhood
Assessment (DECA: LeBuffe & Naglieri, 1999). The DECA sub-
scales, initiative, self-control, and attachment are combined and
referred to as total socioemotional protective factors (TPF), for
which larger numbers indicate greater social skills. The DECA
behavior concerns scale, with higher numbers indicating more
problems, was also used. Teachers and parents reported (identical
forms) the frequency of children’s behaviors using a 5-point Likert
scale to indicate how often within the past 4 weeks the child had
exhibited a variety of behaviors (0 ⫽never,4⫽very frequently).
Parents/teachers chose whether they wanted to complete the form
in English or Spanish. Sixteen percent of teachers and 34% of
parents completed the Spanish form. National percentiles scores
are used. Internal consistency reliability within this diverse sample
was .91 (parent) and .94 (teacher) for TPF, and .72 (parent) and .81
(teacher) for behavior concerns. Reliability did not vary as a
function of language of form (Spanish, English) or rater (Crane,
Mincic, & Winsler, 2011).
Dance Exposure in Middle School
Child-level. Included in the administrative data we received
each academic year, for each student for all grades, were the
course subjects taken (i.e., math, social studies, science, art) with
an end-of-the-year teacher-assigned grade for each course. Using
whether certain course names appeared on the student’s final
report card each year, we created variables denoting whether, and
when, students took a dance elective course in middle school.
Dance course names included “Dance 1, 2, 3, or 4,” and “Dance
Conditioning 1 or 2”. If any of these appeared in a given grade
(6th, 7th, or 8th), the student was flagged as having enrolled in
dance (i.e., 6th grade Dance, 1 ⫽yes and 0 ⫽no). This was done
again for 7th and 8th grade. These grade-based variables were
aggregated across all grades (6th, 7th, and 8th) to make another
variable indicating yes (1) or no whether the child enrolled in
dance at least once during middle school. A continuous variable
indicating the total number of years a child enrolled in dance was
also generated and ranged from 0 to 3 years of dance based on
dance elective enrollment during the 6th, 7th, and 8th grades.
School-level. The variables described above that flag stu-
dents’ exposure to dance in middle school were created at the child
level, however, not all schools that children attended offered dance
courses. To address this, middle schools (N⫽202) were each
flagged for offering in-school dance electives (yes ⫽1/no ⫽0;
during the years that students attended them) using the following
three sources of information: (a) The presence of a student who
had taken a dance class based on the school IDs associated with
each student. For any student who took a dance course in 6th, 7th,
or 8th grade, that child’s school was flagged as yes ⫽1 for offering
dance. (b) The school’s website often indicated course availability.
Schools were flagged as offering dance electives if a dance elec-
tive was listed on their 6th, 7th, or 8th grade online curriculum
sheet, or the school listed a dance instructor on the staff page of the
school’s website. (c) Phone calls were made to administrators. For
schools with no publically available data about dance electives on
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
7
SELECTION INTO DANCE AND OUTCOMES

the website, we contacted by telephone the relevant person who
was listed as head of curriculum or student advising and asked
what, if any, dance elective courses were options for students to
take during the academic years our students attended middle
school. For some schools, current year (2015) course offerings had
to be used to flag schools rather than retroactive information about
what was offered in the academic years 2009 –2014. Schools for
which we could not find information on dance offerings were
treated as missing data. Of the middle schools included in our
study (N⫽202), we did not have information on availability of
dance elective courses for 11.4% (n⫽23).
Outcome Variables
Each outcome variable was recorded for both the concurrent
year in which the dance exposure took place (i.e., 6th grade dance
exposure and 6th grade GPA) and the following year after dance
exposure took place (i.e., 6th grade dance exposure and 7th grade
GPA).
Retention in middle school. Similar criteria mentioned ear-
lier for elementary school retention were used to define retention
in middle school. The student had to complete the middle school
year (6th, 7th, or 8th) and have end-of-the-year grades for that
academic year. Then, the student had to appear the following
academic year for a second time in the same grade. Also, the child
had to have remained long enough to have final grades at the end
of the repeated academic year. For each grade (6th, 7th, or 8th),
children who were retained received a “1” (vs. 0).
GPA in middle school. Middle school (6th, 7th, and 8th
grade) academic subjects included English/language arts/ESOL,
mathematics, social science, science, physical education, music/
art/theater/arts/dance, foreign language, and career and technical
education. As was done for elementary school GPA, we averaged
the student’s grades across all courses to create a GPA score for
each grade (6th, 7th, and 8th).
Standardized math and reading. Students in the state are
required to take the high-stakes FCAT (Human Resources Re-
search Organization & Harcourt Assessment, 2007) in both math
and reading each academic year in middle school (6th, 7th, and
8th). Questions are in both multiple-choice and short-answer for-
mats. Total scale scores are included in the present study (range ⫽
100 –500; Cronbach’s alpha ⫽.90).
Attendance. Participants’ school attendance information was
collected from school records each academic year. Teachers sub-
mitted attendance reports daily, and administrative records listed
the total number of days absent. These totals represent a combi-
nation of both excused and unexcused absences. We created a
continuous variable for analyses as the number of days absent.
School suspension. Administrative records also listed the to-
tal number of days a student was suspended for a behavioral
offense each year in school. This is a combination of both in-
school suspensions (e.g., removes students from regular class
periods and places them in an alternative supervised room at the
school during school hours) and out-of school suspensions (e.g.,
students are not authorized to be on the school grounds for a period
of time). We categorized this dichotomously for analyses—indi-
cating yes/no whether the child was suspended at least once during
that grade. For each grade (6th, 7th, or 8th), children who were
every suspended received a “1” (vs. 0).
Analysis Strategy
The first descriptive research question, having to do with pro-
portions of students who took dance and schools that offered dance
electives, was examined using descriptive statistics. For question
2, which addresses the differences between those who did and did
not take dance in middle school, we first examined this in a
bivariate fashion by running chi-square analyses for categorical
predictors, and independent samples ttests for continuous predic-
tors. Then, entering all of the significant predictors, we tested a
series of developmentally informative, hierarchical, multivariate,
logistic regression analyses to predict the types of students who
enroll in dance courses during middle school. These models were
run twice, first including all children regardless of whether they
attended a school that offered dance (estimating access to dance),
and then restricted to students who went to a school that actually
offered dance (estimating selection into a dance elective when
available). For our final research question on academic outcomes
associated with dance engagement, we ran multiple multilevel
regression models (logistic regression for categorical outcomes)
predicting concurrent and future outcomes as a function of dance
exposure among children who went to a middle school that offered
dance, controlling for the selection factors associated with taking
dance observed in our earlier selection model. The series of mul-
tilevel regression models, run in STATA/SE 14.2 using the
“mixed” or “melogit” commands, allowed us to account for stu-
dent nesting at the school level—the common variance within each
middle school that also accounts for some proportion of each
outcome—and adjusts standard errors accordingly.
Results
Research Question 1. What proportion of our sample enroll
in dance electives during middle school (6th, 7th and 8th grade),
and what proportion of the middle schools attended by the children
offer dance electives?
The number and proportion of our sample who ever took dance,
broken down by 6th, 7th, and 8th grade, is shown in Table 2. Of
the 31,332 adolescents with middle school data, 6.7% took dance
at some point during middle school. Per grade we see that 3.4% of
students enrolled in dance in 6th grade, 5% in 7th grade, and 4.7%
in 8th grade. Of students who took dance at least once, the majority
(68%) enrolled in dance for only one year, 22% took dance for two
years, and only 10% of students took dance all three years (see
Table 2).
Table 2
Proportion of Middle School Students Enrolled in a Dance
Elective by Grade and Cumulative Number of Years
Measure Frequency Total NPercent
Grade
6 1,031 30,413 3.4%
7 1,195 23,788 5%
8 769 16,392 4.7%
Ever 2,113 31,332 6.7%
Number of years
1 year 1,436 2,113 68.0%
2 years 471 2,113 22.3%
3 years 206 2,113 9.7%
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
8GARA AND WINSLER

The 31,332 students in our sample attended 202 different
schools serving 6th through 8th graders in the school district. Of
the 179 middle schools that had dance-relevant data, 36.9% of-
fered dance electives in 6th, 7th, or 8th grade. Only about half of
our sample went to a school that offered dance. Of the students
who went to a school that offered dance (and therefore had the
option of taking dance), 13.3% took a dance elective.
Research Question 2. What are the preexisting selection fac-
tors that predict enrollment in dance courses in middle school?
That is, how do students who do or do not take dance courses in
middle school differ in terms of demographic variables, school
readiness skills at kindergarten entry, and prior elementary school
academic performance?
Bivariate Analyses
Demographics. We addressed predictors of dance engage-
ment one variable at a time via ttests (continuous predictors) and
chi-square (categorical predictors). Table 3 shows how the cate-
gorical variables (gender, ethnicity, poverty, disability, and ELL,
English proficiency, and retention status in elementary school)
differed depending on whether the student later enrolled in dance
in middle school. Latinx students had the highest enrollment rates
(8.4%), followed by White (7.7%) and Asian (6.5%) students.
Black students had the lowest enrollment rates compared with all
other ethnic groups with only 3.5% taking a dance class, ␹
2
(3) ⫽
247.23, p⬍.001. For males, an extremely small percentage
(0.8%) took dance during middle school, compared to 13.2% of
females, ␹
2
(1) ⫽1,890.07, p⬍.001.
Students without disabilities enrolled in dance more than double
the rate of students who received special education services (7.8%
vs. 2.2%), ␹
2
(1) ⫽199.98, p⬍.001. ELL’s showed higher
enrollment in dance (7.8%) compared to native English speakers
(5.3%), ␹
2
(1) ⫽78.39, p⬍.001. Children who were proficient in
English (including native speakers) in 6th grade were more than
twice as likely to take dance (7.2%), compared with students who
had not reached full English proficiency (2.6%), ␹
2
(1) ⫽44.61,
p⬍.001. Those who were retained at least once in elementary
school were much less likely to take dance in middle school
(2.9%), compared to those who were never retained (7.3%),
␹
2
(1) ⫽112.69, p⬍.001. There was no difference in dance
enrollment between students who did or did not receive free/
reduced lunch in 6th grade.
School readiness. Table 4 shows how students who took
dance in middle school were different from those who did not on
continuous predictor variables (social skills, behavior problems,
cognitive, language, motor skills at school entry, GPA and math/
reading scores in 5th grade) years before they arrived to middle
school. At age four, children who enrolled in dance seven years
later during middle school had higher social skills compared to
students who never took dance, t(2,290.72) ⫽⫺18.15, p⬍.001,
d⫽.41. Children who enrolled in dance during middle school also
had fewer behavioral concerns at age four compared to children
who did not, t(2,231.72) ⫽14.67, p⬍.001, d⫽.34. Similarly, at
age four, children who later enrolled in a dance elective had
stronger fine motor, t(1,795.32) ⫽⫺15.27, p⬍.001, d⫽.39)
cognitive, t(1,753.64) ⫽⫺10.03, p⬍.001, d⫽.26, and language
skills, t(19,705) ⫽⫺8.09, p⬍.001, d⫽.22. Gross motor skills
at school entry did not differ between groups. Importantly, stu-
dents who took dance in 6th, 7th, or 8th grade were already
performing better academically in 5th grade compared with those
who did not take dance. Students who later went on to take
dance in middle school had higher 5th grade GPAs,
t(2,579.26) ⫽⫺27.68, p⬍.001, d⫽.57, 5th grade math,
t(2,406.89) ⫽⫺12.92, p⬍.001, d⫽.29, and 5th grade reading
test scores, t(2,408.70) ⫽⫺16.51, p⬍.001, d⫽.37, compared
with those who did not go on to enroll in dance.
Multivariate Analyses
The above analyses were conducted in a bivariate fashion, one
variable at a time to see the individual correlates of taking dance
to compare with other studies that did not have as many covariates.
Here we report the results of hierarchical, multivariate, logistic
regression analyses that predict dance enrollment during middle
school (yes/no ever) from the combination of all of our preexisting
selection variables. Importantly, these models control for intercor-
relations between the predictor variables and tell us their unique
and combined effects on dance exposure. The first block of the
regression model examined the contributions of demographic vari-
ables (ethnicity, gender, poverty, disability status, ELL status,
English proficiency) and child school readiness at age four as they
come together to predict later dance enrollment. Next, the elemen-
tary school academic performance indicators (5th grade GPA and
math scores, retention status) were entered into Block 2. Model 2
informs us not only as to whether 5th grade achievement is related
to middle school dance enrollment, controlling for earlier child
competence and demographic variables, but also whether the de-
mographic variables in Step 1 remain associated with dance en-
rollment after elementary school achievement is entered. That is, if
Table 3
Proportion of Students Who Enroll in Dance Electives in Middle
School by Demographic Variables – Bivariate Analyses
Variable
Enrolled
n%
Gender
ⴱⴱ
Male 128 .80%
Female 1,977 13.2%
Ethnicity
ⴱⴱ
White/Asian/Other 154 7.7%
Hispanic 1,587 8.4%
Black 352 3.5%
Asian/Pacific Islander 12 6.5%
Poverty in sixth grade
No free or reduced lunch 428 7.4%
Free or reduced lunch 1,656 6.7%
Disability in sixth grade
ⴱⴱ
Nondisabled 1,979 7.8%
Has a disability 108 2.2%
ELL in kindergarten
ⴱⴱ
Non-ELL 706 5.3%
ELL 1,407 7.8%
English proficiency in sixth grade
ⴱⴱ
Not English proficient 39 2.6%
English proficient 2,030 7.2%
Prior retention
Not retained 1,990 7.3%
Retained 123 2.9%
ⴱⴱ
p⬍.00l.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
9
SELECTION INTO DANCE AND OUTCOMES

the coefficient for a demographic variable is no longer significant
after children’s academic performance were included in the model,
it suggests that the variable was only indirectly related to middle-
school dance enrollment and its effect is better explained by
academic skill. As explained earlier, we examined these models
from two perspectives: (a) access –the entire sample, who is
accessing dance in middle school (including all students regardless
of whether they attend a school that offers dance)?, and (b)
selection/choice – limited to students who attend a school that
offered dance which gets at which students choose to take dance
electives when they are available.
Access – Step 1. Results of the logistic regression models for
access to dance in middle school are shown in Table 5. When the
demographic and school readiness variables were entered to pre-
dict middle school dance enrollment, ethnicity, gender, disability,
English proficiency, and social skills at age four were significant
predictors of who took dance in middle school. Odds ratios (OR)
are provided which indicate the extent to which the odds of taking
dance anytime in middle school increase (a number being greater
than 1) or decrease (less than 1) as a function of being one level of
the variable (i.e., male) compared to the other (female). For con-
tinuous predictors, the OR indicates how much the odds of dance
enrollment change with a 1-point increase in the predictor (i.e.,
moving from the 39th to the 40th percentile in cognitive skills).
Black students were much less likely (about half) to take a dance
elective compared with White and Latinx students. Females had
nearly twice the odds of enrolling in dance compared with males.
Students with a disability had 27% decreased odds of taking dance
compared with those without a disability. Last, students who were
proficient in English by 6th grade had almost double the odds of
enrolling in dance compared with those who were not yet English
proficient. ELL and poverty status were not significant unique
predictors of taking dance in middle school. It is important to point
out that these are adjusted effects after controlling for other vari-
ables in the model. Thus, for example, even controlling for pov-
erty, disability status, and school readiness, Black students still had
significantly less access to dance.
Children’s teacher-rated social skills also uniquely predicted the
odds of taking dance seven years later in middle school. For each
1-point increase in social skills at age four, the odds of taking
dance increased by .006. So, for example, a child at the 75th
percentile in school entry social skills compared to a child at the
25th percentile (a 50-point difference) has a 30% increased odds
(50 ⫻.006) of taking dance seven years later. Children’s behav-
ioral concerns, fine and gross motor skills, and cognitive and
language skills at age four were not significant unique predictors
of enrolling in dance seven years later when other demographic
variables were included.
Access – Step 2. In Step 2, when children’s 5th grade aca-
demic performance was entered, controlling for other covariates,
we see that each of the prior performance measures was associated
with taking dance during middle school. Students who were re-
tained at least once in elementary school had 36% decreased odds
of taking dance in middle school compared with those never
Table 4
Independent Samples t Test of Enrollment in a Dance Elective in Middle School
Variable
Enrolled Not enrolled
Cohen’s dn M SD n M SD
School readiness
Social skills
ⴱ
1,888 69.39 24.56 24,098 58.61 28.06 .41
Behavioral concerns
ⴱ
1,888 37.24 27.88 24,098 47.06 29.52 .34
Gross motor skills 1,313 66.77 28.11 16,121 68.26 28.95 .05
Fine motor skills
ⴱ
1,479 66.53 25.69 18,332 55.83 28.97 .39
Cognitive skills
ⴱ
1,474 59.51 28.49 18,321 51.74 30.38 .26
Language
ⴱ
1,463 50.78 30.14 18,244 44.05 30.68 .22
Elementary academic performance
Fifth grade GPA
ⴱ
2,039 4.36 .44 26,915 4.07 .57 .57
Fifth grade math
ⴱ
2,047 303.03 65.52 26,526 283.55 68.73 .29
Fifth grade reading
ⴱ
2,047 289.02 58.30 26,568 266.87 61.24 .37
ⴱ
p⬍.001.
Table 5
Multiple Regression Predicting Access to Dance Courses in
Middle School
Variable
Step 1 Step 2
Odds
ratio SE (B)
Odds
ratio SE (B)
Demographics
Hispanic/White 1.19 .15 1.25 .15
Black/White .44
ⴱⴱⴱ
.16 .51
ⴱⴱⴱ
.16
Asian/White .99 .44 .95 .44
Black/Hispanic .37
ⴱⴱⴱ
.10 .41
ⴱⴱⴱ
.11
Asian/Hispanic .83 .42 .76 .42
Male .05
ⴱⴱⴱ
.14 .05
ⴱⴱⴱ
.14
Poverty 1.04 .09 1.19 .09
Disability .73
ⴱ
.16 1.03 .17
ELL 1.03 .09 .99 .09
English proficient 1.85
ⴱ
.25 1.43 .25
School readiness at age 4
Total S-E protective factors 1.006
ⴱⴱⴱ
.002 1.19 .09
Behavioral concerns 1.001 .001 1.005
ⴱⴱ
.002
Gross motor skills 1.00 .001 1.00 .001
Fine motor skills 1.003 .002 1.002 .002
Cognitive skills 1.001 .002 .99 .002
Language skills 1.002 .001 1.001 .001
Elementary academic performance
Retention in elementary school .64
ⴱⴱ
.152
GPA in fifth grade 1.50
ⴱⴱⴱ
.083
Math score in fifth Grade 1.002
ⴱⴱⴱ
.001
ⴱ
p⬍.05.
ⴱⴱ
p⬍.01.
ⴱⴱⴱ
p⬍.001.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
10 GARA AND WINSLER

retained. A 1-point increase in 5th grade GPA (i.e., moving from
a ‘B’ to an ‘A’) was associated with a 50% increased odds of
taking dance. A 1-point increase in 5th grade math scores was
associated with .002% increased odds of later exposure to dance
(i.e., a 100-point increase on the 500-point FCAT scale would
mean a 20% increased odds of taking dance).
Ethnicity and gender remained significant unique predictors of
taking dance during middle school, after adding 5th grade aca-
demic performance. Disability status, English proficiency, and
social skills at age four (significant effects in Step 1) were no
longer significantly related to dance enrollment after factoring in
5th grade academic performance in Step 2. Finally, behavioral
concerns at age four emerged as a significant unique predictor of
taking dance in middle school after including 5th grade academic
performance. Interestingly, each 1-point increase in behavior
problems at age 4 was associated with a .005% increased odds of
taking dance in middle school.
Selection/Choice. The above analyses inform us about who is
taking dance classes in middle school, and thus tell us about which
types of students have access to dance in their school, but these
models do not account for whether or not the student attended a
school that offers dance. Only 37% of middle schools attended by
our sample offered dance electives. To account for the availability
of dance courses among middle schools, we estimated the same
regression analyses as above, except we limited the sample to
students who attended a middle school where dance was actually
offered. These analyses are more representative of which students
actively choose or select dance classes when they are available at
their school.
Selection – Step 1. Results of the logistic regression model
predicting selection/choice into dance in middle school are shown
in Table 6. Selection into middle school dance significantly varied
as a function of ethnicity, gender, disability, and social skills at age
four. Similar to the access model reported above, Black students
were still less likely to take dance, 40% decreased odds compared
with White and Hispanic students, controlling for other covariates.
This indicates that underrepresentation of Black students in dance
is not simply explained by them being more likely to attend
schools that do not offer dance (true, but not the whole story).
Even when available, Black students are not choosing to take
dance classes as much as other groups. The gender difference
remained stable, with females being twice as likely to take dance
compared to males, and students with a disability had half the odds
of enrolling in dance, than those without a disability. Among the
school readiness predictors, only social skills were associated with
selecting dance in middle school. The odds of taking dance in-
creased by .007% for every one-percentile-point increase in social
skills at age four.
Selection – Step 2. Each of the 5th grade performance mea-
sures was associated with choosing to take a dance class in middle
school. Students who were retained during elementary school had
35% decreased odds of taking dance in middle school. Each
one-point increase in 5th grade GPA yielded a 44% increased odds
of taking dance, and a one-point increase on the 5th grade math
was associated with a .002% increase in the odds of taking dance.
Ethnicity remained a predictor of dance enrollment, such that
Black students had less than half the odds of taking dance than
Hispanic students. Interestingly, Black students were no longer
less likely to take dance compared to White students, once prior
academic performance was included. This suggests that a large
part of the reason why Black students enroll less in dance is
attributable to ethnic differences in academic performance—a cer-
tain level of academic performance might be needed to take dance.
Similarly, the effect of disability on selection into dance disap-
peared after entering 5th grade academic performance. This means
that having a disability is correlated with academic performance,
and it is performance in 5th grade that is more important for
predicting dance participation. Males were still significantly un-
derrepresented in taking dance, with 95% decreased odds com-
pared with females. Social skills at age four remained a significant
unique predictor of taking dance. Each one-percentile-point in-
crease in social skills was associated with a .006% increased odds
of taking dance seven years later. Last, poverty emerged as a
unique predictor of taking dance when 5th grade performance was
included. The relation between poverty and selection into dance
was positive—among students with similar academic performance,
those in poverty were slightly more likely to choose dance when it
was available, compared to those not in poverty.
Research Question 3. Controlling for all preexisting selection
factors, are dance classes linked to enhanced concurrent and later
academic outcomes (i.e., GPA, retention, suspension, standardized
math and reading test scores, and attendance) for students in
middle school?
Now that we know that significant preexisting differences exist
between those that do and do not select into middle school dance
courses, our next goal was to control for these variables to examine
whether student’s enrollment in dance classes was associated with
better concurrent and later academic outcomes in middle school.
We ran multilevel multiple regression analyses including, each
Table 6
Multiple Regression Predicting Selection to Dance Courses in
Middle School
Variable
Step 1 Step 2
Odds
ratio SE (B)
Odds
ratio SE (B)
Demographics
Hispanic/White 1.03 .16 1.09 .16
Black/White .61
ⴱⴱ
.17 .73 .17
Asian/White .93 .47 .91 .47
Black/Hispanic .59
ⴱⴱⴱ
.12 .67
ⴱⴱ
.12
Asian/Hispanic .90 .45 .83 .45
Male .05
ⴱⴱⴱ
.14 .05
ⴱⴱⴱ
.14
Poverty 1.12 .10 1.26
ⴱ
.10
Disability .62
ⴱⴱ
.16 .87 .17
ELL 1.103 .10 1.07 .10
English proficient 1.49 .26 1.12 .27
School readiness at age 4
Social skills 1.007
ⴱⴱⴱ
.002 1.006
ⴱⴱ
.002
Behavioral concerns 1.00 .001 1.00 .001
Gross Motor skills 1.00 .001 1.00 .001
Fine Motor skills 1.00 .002 1.00 .002
Cognitive skills 1.00 .002 .99 .002
Language skills 1.00 .002 1.00 .002
Elementary academic performance
Retention in elementary school .65
ⴱ
.16
GPA in fifth grade 1.56
ⴱⴱⴱ
.09
Math score in fifth grade 1.002
ⴱ
.001
ⴱ
p⬍.05.
ⴱⴱ
p⬍.01.
ⴱⴱⴱ
p⬍.001.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
11
SELECTION INTO DANCE AND OUTCOMES

time, the relevant selection factors that were associated with taking
dance (from Step 2 of the selection regressions reported above)
with exposure to dance as the main predictor of interest (1 ⫽took
dance, 0 ⫽did not), among students who went to a school that
offered dance electives. Retention in elementary school was not
included as a covariate— despite its significance in Table 6 as a
predictor of selection into dance— because of lack of variance (no
one who took dance had been retained in elementary school). Our
findings were similar with and without including this variable in
the model. We ran models both for concurrent outcomes during
the same year that the dance class was taken, and for the subse-
quent year, the year after the dance class was taken. The unstan-
dardized beta coefficients and adjusted standard errors (accounting
for nesting at the school level) for concurrent and subsequent year
outcomes associated with having taken dance classes in 6th grade,
7th grade, and 8th grade are shown in Table 7.
6th grade. Controlling for prior academic achievement and
other selection factors distinguishing those who did and did not
take dance, those who took dance in 6th grade had significantly
better grades in 6th grade compared to those who did not (see
Table 7). There were no other significant differences on the other
outcomes in 6th or 7th grade between those that did and did not
take dance in 6th grade.
7th grade. Students who took dance in 7th grade had signif-
icantly higher math scores and were less likely to be suspended in
7th grade, compared with those who did not enroll in dance,
controlling for covariates (see Table 7). Those who took dance in
7th grade also had significantly higher GPAs and missed signifi-
cantly fewer days of school in 8th grade, compared with nondance
students.
8th grade. Students who took dance in 8th grade went on to
earn significantly higher GPAs in 8th grade than those who did not
take dance, controlling for covariates (see Table 7). Those who
took dance in 8th grade also missed significantly fewer days of
school in 8th grade than nondance students. There were no differ-
ences on standardized test scores nor on retention or suspension in
8th grade.
Supplemental dosage analyses. The above outcome analyses
only examined whether dichotomous dance exposure (yes/no in X
grade) was related to later outcomes. We also, however, had some
information (albeit limited because not all of our students had a
chance to get to 8th grade yet) on the number of years that the
students took dance. At the helpful suggestion of an anonymous
reviewer, we also ran preliminary, supplemental analyses to see
whether dosage, or the number of years a student took dance,
mattered for academic achievement. This question was analyzed
with a similar regression approach controlling for the same cova-
riates, as above, however without accounting for nesting, and only
including students who went to a school that offered dance and
who had reached eighth grade and, thus, had an opportunity to take
dance for all 3 years (n⫽8,427). Also, to have time to see a
cumulative effect of multiple years of dance, we restricted the
analyses to estimate only 8th grade academic outcomes.
We first estimated multiple regression models (logit for dichot-
omous outcomes) with our continuous independent variable of
years of dance engagement in middle school (range 0 –3 years) in
place of the prior Y/N dichotomous predictor used for the results
included in Table 7. Using the continuous measure of years of
dance taken, we replicated the effects observed in Table 7 on 8th
grade GPA and attendance, but here showing that more years of
dance was even better. In addition, a new positive effect on 8th
grade math scores was now observed using the continuous predic-
tor of years of dance exposure. These results can be seen in Table
S1 in the online supplementary materials.
Next, we estimated identical regression models but instead we
used dummy-coded indicators of the years taken (using one year as
the reference group) to examine, for example, whether two or three
years of dance enrollment was more advantageous to students’
academic achievement than enrolling in just one year of dance. For
this model, we limited the analysis to only those who took dance
at least once to see whether more than one year of dance exposure
mattered. These results can be seen in Table S2 in the online
supplementary materials.
Interestingly, even more positive associations appeared when
students took multiple years of dance electives in middle school.
Students’ 8th grade GPAs were significantly higher for students
who had enrolled in two or three years of dance, versus one year
of dance during middle school. Eighth grade math scores also were
significantly higher for those students who had taken two years of
dance electives (vs. one year). Students who enrolled in two or
years three of dance electives missed significantly less days of
school in 8th grade than students with only one year of dance.
Table 7
Concurrent and Subsequent-Year Outcomes From Taking Dance
in Middle School
Enrolled in dance
6th grade
6th grade outcomes 7th grade outcomes
BSEBSE
GPA .058
ⴱ
.018 .041 .023
Math 1.650 1.555 .255 1.615
Reading ⫺.315 1.629 ⫺1.941 1.757
Days absent ⫺.301 .284 ⫺.381 .368
Retained
a
2.337 1.237 2.461 1.423
Suspended
a
.761 .117 .942 .139
Enrolled in dance
7th grade 7th grade outcomes 8th grade outcomes
GPA .042 .019 .067
ⴱ
.022
Math 3.763
ⴱ
1.353 1.546 .649
Reading 1.996 1.453 .724 .618
Days absent ⫺.594 .305 ⫺1.029
ⴱ
.385
Retained
a
1 n/a .786 .634
Suspension
a
.582
ⴱⴱ
.080 .863 .140
Enrolled in dance
8th grade 8th grade outcomes
GPA .088
ⴱⴱ
.022
Math .997 .655
Reading .457 .607
Days absent ⫺1.289
ⴱⴱ
.379
Retained
a
.995 .792
Suspension
a
.700 .118
Note. Unstandardized beta coefficients are shown. Estimates displayed
are the concurrent and subsequent outcomes of taking a dance elective in
sixth, seventh, or eighth grade. Each variable in the left-most column
represents a unique outcome estimated by our models. Race, gender, FRL,
behavioral concerns at age four, fifth grade GPA, and fifth grade math
scores are controlled for in each model.
a
Odds ratios.
ⴱ
p⬍.01.
ⴱⴱ
p⬍.001.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
12 GARA AND WINSLER

Two new outcomes emerged when examining multiple years of
taking dance, compared with our nondosage models. Those who
took dance for two years instead of one had significantly higher
standardized reading test scores in 8th grade. Finally, the small
number of students who took dance electives for all three years in
middle school had significantly lower rates of suspension in 8th
grade, compared with students with only one year of dance. These
dosage models support our original models reported in Table 7 and
extend these findings to show that taking multiple years of dance
electives during middle school appears to be advantageous to
academic achievement.
Discussion
Although many studies have shown associations between par-
ticipation in the arts in general and enhanced child cognitive,
social, and academic outcomes (Brown et al., 2017; Brown & Sax,
2013; Catterall, 2009; Catterall et al., 2012; Elpus, 2013a) and
some have explored positive benefits for dance engagement in
particular (Gilbert, 2006; Greenfader & Brouillette, 2017; Kim,
2007; Lobo & Winsler, 2006; Minton, 2003; Morgan & Stengel-
Mohr, 2014; Park, 2007; Rossberg-Gempton et al., 1999; Seham,
1997), much of the research is correlational and does not ade-
quately control for preexisting selection effects—the many ways
that students who do and do not get arts experiences are initially
different. Further, studies often examine rather global arts experi-
ences (Catterall, 2009; Catterall et al., 2012; Winsler et al., 2019)
when outcomes associated with arts participation are likely to vary
by the particular art form in question (Winner et al., 2013).
Dance is an art form that has been relatively understudied in the
literature, and opportunities for dance are becoming fewer in
today’s public school systems (Parsad et al., 2012). With a large
(n⫽31,332) and ethnically diverse sample followed longitudi-
nally from pre-K through 8th grade, we carefully identified (a) the
frequency with which adolescents enroll in dance electives during
middle school and (b) the differences across students who do and
do not (a) have access to and (b) select into dance electives in
middle school when available. We further used this information to
then statistically control for all observed selection factors to ex-
amine whether taking middle school dance classes was linked to
better academic outcomes for students later in middle school.
Access and Selection Into Dance
Overall, only 6.7% of our students took dance in middle school,
with the majority taking a dance class for only a single year. This
relatively small figure for dance enrollment is similar, although
slightly less than the 10% to 17% of adults who report taking
dance in or out of school as a child (Hall, 2015), yet higher than the
3.6% of a large sample of high school students who enrolled in
dance coursework as indicated by high school transcripts (Elpus,
2013a). Our low rate of dance participation is not surprising given
that dance is the least accessible form of arts programming (com-
pared to music, visual art, and drama) offered in middle schools in
the nation, at 12% (Parsad et al., 2012). Dance participation was
low in our sample largely because many schools did not appear to
offer it as an elective course. At the school level, only 37% of
middle schools in the area offered dance in 6th through 8th grade.
After limiting the sample to only students who went to a middle
school that offered dance, we found that 13% of students enrolled
in dance electives. Our 13% figure is very similar to that reported
for the state average for in-school dance enrollment (14%); how-
ever, these percentages are slightly lower than the Southeastern
regional average for uptake (20%) when dance was available (Bell,
2014).
Access to dance classes in middle school, however, is far from
equal. Black students were significantly underrepresented in mid-
dle school dance courses compared with both White and Latinx
students. This was partly attributable to access—Black students
were more likely to attend a middle school that did not offer dance
courses. Inequities in school funding, especially in schools located
in low-income neighborhoods, is a strong explanation for why
Black students, who are more likely to reside in low-income
neighborhoods, attend schools without dance programming (Du-
mais, 2002; NEA, 2012). Low-income neighborhoods are typically
highly segregated, as is the case in this community, where one in
three Black students are enrolled in “isolated” schools—“compris-
[ing] 85% of one racial group” (Moore, 2004; Smiley, 2014).
Schools without funding for arts such as dance will not have
adequate resources for dance infrastructure such as credentialed
dance teachers or classrooms designed for dance and movement
(Chapman, 2004).
Access to a school that offers dance is not the whole story,
however. Even when attending a school that offers dance electives,
Black students were still less likely to enroll in dance compared
with Latinx students. It is important to note that academic skills
and poverty status were controlled for in our analyses, so this
ethnic difference cannot be explained by those factors. This ethnic
discrepancy may be related to the ethnic make-up of the commu-
nity, which is 66.8% Hispanic/Latinx descent (U.S. Census Bu-
reau, 2015). As a part of Hispanic culture, social dancing is a
widely accepted cultural norm (NEA, 2013), and Hispanic youth
may be exposed to dance outside of school, which encourages their
decision to enroll in dance electives. Intrinsic disinterest, stigma,
or peer/social motivators (i.e., not feeling welcome in dance
classes or clashes with ‘Black’ identity) may also influence Black
students’ decision to not enroll in dance electives when they are
available (Eccles, 2005; Fredricks et al., 2002).
Interestingly, the difference between Black and White students
selecting into dance when it was available was no longer signifi-
cant after including academic performance in the model. (Black
students continued to be underrepresented in dance relative to
Latinx students, however). This suggests that a certain level of
academic performance may be needed before certain groups of
students either feel free to, or are allowed to, sign up for dance
classes. As discussed in the introduction, schools with a heavy
emphasis on standardized tests often require students who are low
performing in math and reading to use their elective options in the
form of remediation courses to help increase their chance of
passing the high stakes, standardized tests (Beveridge, 2009).
Numerous scholars are concerned that high-stakes standardized
tests disproportionally disadvantage students of color (Greene &
Winters, 2009; Horn, 2003; Penfield, 2010; Tavassolie & Winsler,
in press). Perhaps another, yet unexplored, side effect of high
stakes tests is that required or encouraged remedial courses de-
signed to help struggling students pass such tests are taking away
valuable elective opportunities for certain groups of students to
experience the arts.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
13
SELECTION INTO DANCE AND OUTCOMES

Females students were nearly twice as likely to enroll in dance
classes compared with males. Male participation in Western Eu-
ropean dance (i.e., ballet) is often stigmatized by social norms or
gender stereotypes favoring female engagement (Gard, 2008; Ris-
ner, 2007). Dance engagement in Western societies is primarily
female dominant, which might challenge male identity or mascu-
linity (Risner, 2007). Challenges to male identity often result in
decreased likelihood of males engaging in dance, especially during
adolescence when youth are exploring a sense of self and securing
friend groups. However, among Afro-Caribbean and Latin cul-
tures, male dancing is a normalized form of social engagement and
cultural sharing (NEA, 2013). Within the Miami area where the
current study took place, Caribbean and Latin American dances
such as Salsa, Tango, Cha Cha, and Mambo are popular, so one
might have thought that male dancing would be more normalized
in the present study, but we still saw a large gender difference. It
is possible that traditional ballet or modern dance rather than Latin
dance are what is offered in schools, but we did not have infor-
mation on the exact type of dance found in the curriculum.
Most arts education research has found that arts participation is
more likely among individuals of middle to high socioeconomic
status (Catterall et al., 2012; Elpus & Abril, 2011; Foster &
Jenkins, 2017). Yet, within our predominantly low-income and
urban sample, we found that when dance courses were offered and
other factors such as prior academic competence were controlled,
students whose family income reached the threshold for receiving
free/reduced-price lunch were actually 26% more likely to enroll
in dance than students not in poverty. This suggests that students
in poverty in this urban community may find dance electives as an
attractive outlet for expression, identity, and/or learning in middle
school, supporting findings reported by O’Neill et al. (2011) that,
when given the autonomy, adolescents from low-income families
will select into dance at higher rates than adolescents from higher-
income families.
According to the current study, students with disabilities were
less likely to take dance electives than those without, but interest-
ingly, this association was not significant after controlling for
students’ prior academic performance. This suggests that the rea-
son why children with disabilities are less likely to enroll in dance
is because of their increased likelihood of struggling in school,
perhaps combined with the need to allocate more time to academ-
ics rather than the arts. If this is the case, it is unfortunate given the
positive socioemotional and cognitive benefits dance engagement
has shown for students with disabilities and other underperforming
youth (Albin, 2016; Duggan, Stratton-Gonzalez, & Gallant, 2009;
Skoning, 2008; Seham, 1997).
We also examined whether ELL status and English proficiency
were related to dance enrollment in middle school. Although prior
work has proposed that students who do not speak English as a
first language may actively utilize dance as a form of self-
expression through nonverbal symbolism (Connery et al., 2010),
we found both ELL status and English proficiency in 6th grade
were not associated with taking dance electives in middle school
when all other variables were included in the model. There was,
however, relatively little variance in English proficiency for this
particular sample of older students since they had been in the
school system since pre-K.
A strong contribution of the present longitudinal study was that
we were able to examine whether different domains of children’s
school readiness skills at age four were associated with students’
selection into dance elective courses seven years later in middle
school. The bivariate analyses revealed that children who took
dance electives in middle school began kindergarten with stronger
social, language, fine motor, cognitive, and behavioral skills com-
pared with students who did not choose dance electives. However,
after demographic variables and later elementary school achieve-
ment were added to the models, the only school readiness domain
that remained uniquely and positively associated with selection
into dance was social skills. This finding has critical implications
for researchers attempting to show that dance education and joint
movement activities promote social skills, collaboration, and co-
operation among children (Gilbert, 2006; Kirschner & Tomasello,
2010; Lobo & Winsler, 2006; Rabinowitch & Meltzoff, 2017;
Rossberg-Gempton et al., 1999; Tunçgenç & Cohen, 2018, 2016)
because we show that children already had stronger social skills
way before the middle school dance exposure took place. Clearly,
dance education researchers must deal with the possibility of
bidirectional relations— dance might improve children’s social
skills, but it also appears to be the case from our results that social
skills might influence children’s selection into dance. Researchers
must understand and appropriately control for such selection and
bidirectional effects when attempting to show positive links be-
tween dance education and children’s outcomes.
The strongest and most robust selection factor identified in the
present study, however, was prior academic achievement. Before
even getting to middle school, students who went on to enroll in
dance in 6th through 8th grade already had better grades, scored
higher on standardized math and reading tests in 5th grade, and
were less likely to have been retained in elementary school com-
pared to students who did not enroll in dance in middle school.
These findings are consistent with prior research showing that high
school students who elect into music coursework are also more
academically competent than nonmusic students (Elpus & Abril,
2011). The finding that students naturally exposed to dance are
already higher functioning academically than those who are not,
even before the measured dance exposure begins, is critical for
researchers to consider when trying to make causal claims about
the effects of dance (or other arts for that matter) on cognition and
academic performance.
Indeed, research has documented that children who engage in
the arts typically have parents with higher cognitive skills, more
years of education, greater family assets, and spend more time with
their children than children not exposed to the arts (Elpus & Abril,
2011; Foster & Jenkins, 2017). Such increased family social cap-
ital leads to enhanced academic development through increased
parent– child scaffolding, learning activities, and educational in-
vestment, and also leads to enhanced exposure to the arts. Foster
and Jenkins (2017) report that these preexisting family background
characteristics explain more of the variance in children’s academic
achievement than direct causation between arts exposure and later
outcomes. We did include initial school readiness skills and prior
academic achievement in our models—factors that are presumably
the consequence of family capital—as covariates to capture such
early family influences, but we were limited in our availability of
rich family covariates to examine this further. Another possibility
to explain our findings is that many years of bidirectional causal
influences have occurred—where enhanced family capital lead to
early dance/arts exposure, and over time that exposure to dance
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
14 GARA AND WINSLER

also stimulated cognitive and academic development among the
students. It could be that the students who enrolled in middle
school dance in our sample came from families where dance had
been encouraged early on (i.e., ballet lessons since age 5?) and that
such early exposure to dance may have already had a positive
influence on children’s elementary school achievement by 5th
grade.
Outcomes From Dance Participation
Our final goal was to control for the selection factors identified
in the current study to see whether taking dance as an elective
course in 6th, 7th, or 8th grade was still associated with enhanced
academic performance concurrently and later in middle school. It
is important to note that our quasi-experimental research design
does not allow us to make causal claims between dance participa-
tion and student academic performance. However, in the absence
of a randomized control trial, our large-scale, prospective, longi-
tudinal design with multilevel analyses, using strong statistical
controls for multiple, known, preexisting selection factors that
make participation groups initially different is among the next best
methodological options available (Shadish, Cook, & Campbell,
2002).
Controlling for selection factors, engagement in middle school
dance electives was associated with enhanced academic outcomes
for students either in their concurrent academic year or in the
following year in middle school. Students who took dance in either
6th, 7th, or 8th grade attained higher GPAs and/or higher stan-
dardized math scores in either the concurrent or the following
academic year. These findings, that dance students scored higher
in math, are in contrast to prior research that failed to find a
relation between dance engagement and math (Lanfredi, 2013).
However, Lanfredi (2013) examined a small sample of females
who engaged in dance training outside of school and they used a
visual-spatial thinking measure of math ability; our study esti-
mated the association between in-school dance and a school-
administrated comprehensive math assessment. Prior research has
noted that learning the kinesthetic of dance outside of the academic
classroom (i.e., symmetrical movement and rhythm) can comple-
ment the learning of math concepts, and vice versa (Belcastro &
Schaffer, 2011).
Children who took dance in 7th or 8th grade also had better
school attendance (fewer days absent) in the concurrent year, and
taking dance in 7th grade was also linked to lower rates of
suspension from school that year. Elpus (2013a) also reported high
school dance students having lower rates of suspension (6.8%)
compared with nondancers (8%), but for them the contrast was not
statistically significant. Lower rates of suspension and attending
school more often among dancers, compared with nondancers,
may be attributable to motivational factors in which students may
be more engaged in the school environment as a result of their
interest in the dance activity setting (Eccles, 2005).
In our primary analyses examining dichotomous (yes/no) dance
enrollment, dance was not associated with reading achievement in
middle school, despite other research supporting early childhood
dance involvement and literacy skills (Greenfader, Brouillette,
Farkas, 2015; Morgan & Stengel-Mohr, 2014). It is possible that
dance only enhances literacy when the dance curriculum is cou-
pled with a targeted intervention, as opposed to general dance
education electives as was examined here. It is notable, however,
that in our supplemental analyses examining dosage, cumulative
positive ‘effects’ of dance were observed for standardized reading
scores in 8th grade for those who had taken two years of dance.
Indeed, we found that several positive associations between taking
dance electives and academic outcomes in 8th grade were either
stronger or only emerged for students after multiple years of dance
enrollment during middle school. Students’ eighth grade reading
and math scores, for example, were higher for those who enrolled
in two years of dance, compared with 1-year takers. Also, reduced
likelihood of being suspended from school in 8th grade was found,
but only when students took three years of dance. These findings
related to dosage effects of dance support other work showing that
intense, sustained, and/or multiyear participation with the perform-
ing arts is more beneficial than brief exposure (Elpus, 2013a;
Holochwost et al., 2017; Jaschke, Honing, & Scherder, 2018).
The positive findings here, with our prospective longitudinal
design with statistical control for numerous selection effects, are
some of the strongest empirical data available to date showing
positive academic outcomes from dance engagement in middle
school. This adds to the growing literature showing positive effects
and increased academic performance associated with arts engage-
ment (Catterall, 2009; Catterall et al., 2012; Elpus, 2013a; Hardi-
man, Magsamen, McKhann, & Eilber, 2009; Holochwost et al.,
2017; Holochwost, Wolf, Fisher, & O’Grady, 2016; McCarthy et
al., 2004; Posner & Patoine, 2009). Our study contributes in
particular by examining a large, low-income and ethnically diverse
sample including many Latinx students who have been underrep-
resented in the literature, by following longitudinally the sample
for 10 years, by identifying and statistically controlling for numer-
ous selection factors, by focusing specifically on dance, and on
in-school dance courses, and by examining a wide array of au-
thentic school outcomes.
Limitations and Future Research
Our data, although strong in many ways, were limited. For
instance, a student’s decision to enroll in dance courses may have
been stunted due to the style of dance offered in middle school
(i.e., ballet vs. salsa). Some adolescents may have wanted to enroll
in dance, but their interest in dance may not have matched the
particular course(s) at their school, or that fit into their course
schedule. Unfortunately, we did not have data on the type of dance
offered in the curriculum. Also, for some schools, we could not
find information on whether dance was offered. We also did not
have data about the amount of out-of-school dance experiences
adolescents received while enrolled in middle-school or at an
earlier age. This particular concern— exposure to dance outside of
school—may or may not have affected our reported outcomes
associated with in-school dance engagement. Such gaps are clearly
of interest for future research. Further, future longitudinal research
should address the potential of bidirectional effects of exposure to
dance and cognitive development, overtime. In addition, although
we were able to account for nesting within schools, we did not
have the necessary data to account for nesting at the classroom
level. Last, although a strength of this study is our large and
diverse Miami community sample that describes a particular com-
munity (Jager, Putnick, & Bornstein, 2017), it also limits the
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
15
SELECTION INTO DANCE AND OUTCOMES

external validity for generalizing our findings to students of
medium- to high-income families, and those in other communities.
Conclusion/Implications
The present study showed that in Miami, a moderate proportion
of middle schools offered dance, and a relatively small proportion
of children enrolled in dance electives. Future research should
examine the barriers to dance enrollment and potential reasons
why adolescents do not enroll and/or persist in dance electives
during middle school, particularly among Black students, those
who are disabled, and those who are low performing in terms of
academics. In addition, school administrators and art educators
may wish to enhance their efforts at increasing access for dance
electives for students in 6th, 7th, and 8th grade in this community,
and likely other communities as well,.
There are also serious selection differences between students
who do or do not choose to take dance, favoring White and
Hispanic students, females, and students who were more socially
and academically competent before entering middle school. It is
imperative that future research on the effects of dance understand
and control for selection effects before documenting associations
between dance and enhanced development, if not using a true
experimental design. Finally, we showed, with a rigorous quasi-
experimental design, that taking dance in middle school is linked
with enhanced academic performance among low-income, ethni-
cally diverse students. These findings suggest that increasing op-
portunities for students to engage in dance in middle school would
not only be good for art’s sake, but also useful for other academic
goals as well.
References
Albin, C. M. (2016). The benefit of movement: Dance/movement therapy
and down syndrome. Journal of Distance Education, 16, 58 – 61. http://
dx.doi.org/10.1080/15290824.2015.1061196
Belcastro, S. M., & Schaffer, K. (2011). Dancing mathematics and the
mathematics of dance. Math Horizons, 18, 16 –20. http://dx.doi.org/10
.4169/194762111X12954578042939
Bell, A. (May, 2014). Arts education in the south phase I: Public school
data and principals’ perspective. Summary Report. Atlanta, GA: South
Arts, Inc.
Beveridge, T. (2009). No child left behind and fine arts classes. Arts
Education Policy Review, 111, 4 –7. http://dx.doi.org/10.1080/
10632910903228090
Bonbright, J. (1999). Dance education 1999: Status, challenges, and rec-
ommendations. Arts Education Policy Review, 101, 33–39.
Brown, E. D., Garnett, M. L., Anderson, K. E., & Laurenceau, J. (2017).
Can the arts get under the skin? Arts and cortisol for economically
disadvantaged children. Child Development, 88, 1368 –1381.
Brown, E. D., & Sax, K. L. (2013). Arts enrichment and preschool
emotions for low-income children at risk. Early Childhood Research
Quarterly, 28, 337–346. http://dx.doi.org/10.1016/j.ecresq.2012.08.002
Brown, S., Martinez, M. J., & Parsons, L. M. (2006). The neural basis of
human dance. Cerebral Cortex, 16, 1157–1167. http://dx.doi.org/10
.1093/cercor/bhj057
Bucknavage, L. B., & Worrell, F. C. (2005). A study of academically
talented students’ participation in extracurricular activities. Journal of
Secondary Gifted Education, 16, 74 – 86. http://dx.doi.org/10.4219/jsge-
2005-474
Catterall, J. S. (2009). Doing well and doing good by doing art: The effects
of education in the visual and performing arts on the achievements and
values of young adults. Los Angeles, CA: Imagination Group/I-Group
books.
Catterall, J. S., Dumais, S. A., & Hampden-Thompson, G. (2012). The arts
and achievement in at-risk youth: Findings from four longitudinal stud-
ies. Research Report #55. Washington, DC: National Endowment for the
Arts.
Chapman, L. H. (2004). No child left behind in art? Arts Education Policy
Review, 106, 3–20. http://dx.doi.org/10.3200/AEPR.106.2.3-20
Child Trends Database. (November, 2015). Participation in school music
or other performing arts. Bethesda, MD: Child Trends
Connery, M. C., John-Steiner, V. P., & Marjanovic-Shane, A. (2010).
Vygotsky and creativity: A cultural-historical approach to play, meaning
making, and the arts. New York, NY: Peter Lang. http://dx.doi.org/10
.3726/978-1-4539-0060-4
Costa-Giomi, E. (2004). Effects of three years of piano instruction on
children’s academic achievement, school performance, and self-esteem.
Psychology of Music, 32, 139 –152. http://dx.doi.org/10.1177/
0305735604041491
Crane, J., Mincic, M., & Winsler, A. (2011). Parent-teacher agreement and
reliability on the Devereux Early Childhood Assessment (DECA) in
English and Spanish for ethnically diverse children living in poverty.
Early Education and Development, 22, 520 –547. http://dx.doi.org/10
.1080/10409289.2011.565722
Dahl, R. E. (2004). Adolescent brain development: A period of vulnera-
bilities and opportunities. Annals of the New York Academy of Sciences,
1021, 1–22. http://dx.doi.org/10.1196/annals.1308.001
Duggan, D., Stratton-Gonzalez, S., & Gallant, C. (2009). Dance education
for diverse learners: A special education supplement to the blueprint for
teaching and learning in dance. New York, NY: Department of Educa-
tion.
Dumais, S. A. (2002). Cultural capital, gender, and school success: The
role of habitus. Sociology of Education, 75, 44 – 68. http://dx.doi.org/10
.2307/3090253
Dumais, S. A. (2006). Elementary school students’ extracurricular activi-
ties: The effects of participation on achievement and teachers’ evalua-
tions. Sociological Spectrum: Mid- South Sociological Association, 26,
117–147.
Eccles, J. S. (2005). The present and future of research on activity settings
as developmental contexts. In J. L. Mahoney, R. W. Larson, & J. S.
Eccles (Eds.), Organized activities as contexts of development (pp.
353–371). Mahwah, NJ: Erlbaum.
Educational Testing Service. (2005). Comprehensive English Language
Learning Assessment (CELLA): Technical summary report. Princeton,
NJ: Author. Retrieved from http://www.accountabilityworks.org/
photos/CELLA_Technical_Summary_Report.pdf
Eisner, E. W. (2002). The arts and the creation of mind. New Haven, CT:
Yale University Press.
Elpus, K. (2013a). Arts education and positive youth development: Cog-
nitive, behavioral, and social outcomes for adolescents who study the
arts. Washington, DC: National Endowment for the Arts. Retrieved from
https://www.arts.gov/sites/default/files/Research-Art-Works-Maryland
.pdf
Elpus, K. (2013b). Is it the music or is it selection bias? A nationwide
analysis of music and non-music students’ SAT scores. Journal of
Research in Music Education, 61, 175–194. http://dx.doi.org/10.1177/
0022429413485601
Elpus, K., & Abril, C. R. (2011). High school music ensemble students in
the United States: A demographic profile. Journal of Research in Music
Education, 59, 128 –145. http://dx.doi.org/10.1177/0022429411405207
Foster, M., & Jenkins, J. (2017). Does participation in music and perform-
ing arts influence child development? American Educational Research
Journal, 54, 399 – 443. http://dx.doi.org/10.3102/0002831217701830
Fredricks, J. A., Alfeld-Liro, C. J., Hruda, L. Z., Eccles, J. S., Patrick, S.,
& Ryan, A. M. (2002). A qualitative exploration of adolescents’ com-
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
16 GARA AND WINSLER

mitment to athletics and the arts. Journal of Adolescent Research, 17,
68 –97. http://dx.doi.org/10.1177/0743558402171004
Garcia, E., & Jensen, B. (2009). Early education opportunities for children
of Hispanic origins. Social Policy Report, 23, 1–20. http://dx.doi.org/10
.1002/j.2379-3988.2009.tb00059.x
Gard, M. (2008). When a boy’s gotta dance: New masculinites, old
pleasures. Sport, Education and Society, 13, 181–193. http://dx.doi.org/
10.1080/13573320801957087
Gardner, H. (1990). Arts education and human development. Los Angeles,
CA: Getty Center for Education in the Arts.
Gilbert, A. G. (2006). Brain-compatible dance education. Reston, VA:
American Alliance for Health, Physical Education, Recreation, and
Dance.
Goldstein, T. R., & Lerner, M. D. (2017). Dramatic pretend play games
uniquely improve emotional control in young children. Developmental
Science, 21, e12603.
Goldstein, T. R., Lerner, M. D., & Winner, E. (2017). The arts as a venue
for developmental science: Realizing a latent opportunity. Child Devel-
opment, 88, 1505–1512. http://dx.doi.org/10.1111/cdev.12884
Goldstein, T. R., Tamir, M., & Winner, E. (2013). Expressive suppression
and acting classes. Psychology of Aesthetics, Creativity, and the Arts, 7,
191–196. http://dx.doi.org/10.1037/a0030209
Goldstein, T. R., & Winner, E. (2012). Enhancing empathy and theory of
mind. Journal of Cognition and Development, 13, 19 –37. http://dx.doi
.org/10.1080/15248372.2011.573514
Greene, J. P., & Winters, M. A. (2009). The effects of exemptions to
Florida’s test-based promotion policy: Who is retained? Economics of
Education Review, 28, 135–142. http://dx.doi.org/10.1016/j.econedurev
.2008.02.002
Greenfader, C. M., & Brouillette, L. (2017). The arts, the Common Core,
and English language development in the primary grades. Teachers
College Record, 119, 10.
Greenfader, C. M., Brouillette, L., & Farkas, G. (2015). The effect of a
performing arts program on the oral language skills of young English
learners. Reading Research Quarterly, 50, 185–203.
Hall, T. W. (2015, August). Determinants, effects, and funding of arts
consumption (Working paper). Prepared for the National Endowment for
the Arts.
Hanna, J. L. (2008). A nonverbal language for imagining and learning:
Dance education in K-12 curriculum. Educational Researcher, 37, 491–
506. http://dx.doi.org/10.3102/0013189X08326032
Hanna, J. L. (2015). Dancing to learn: The brain’s cognition, emotion, and
movement. Lanham, MD: The Rowman & Littlefield Publishing Group.
Hardiman, M., Magsamen, S., McKhann, G., & Eilber, J. (2009). Neuro-
education: Learning, arts, and the brain: Findings and challenges for
educators and researchers from the 2009 Johns Hopkins University
summit. New York, NY: Dana Press.
Holochwost, S. J., Propper, C. B., Wolf, D. P., Willoughby, M. T., Fisher,
K. R., Zolacz, J.,...,Jaffee, S. R. (2017). Music education, academic
achievement, and executive functions. Journal of Aesthetics, Creativity,
and the Arts, 11, 147–166.
Holochwost, S. J., Wolf, D. P., Fisher, K. R., & O’Grady, K. (2016). The
socioemotional benefits of the arts: A new mandate for arts education.
Philadelphia, PA: The William Penn Foundation.
Horn, C. (2003). High-stakes testing and students: Stopping or perpetuating
a cycle of failure? Theory Into Practice, 42, 30 – 41. http://dx.doi.org/
10.1207/s15430421tip4201_5
Human Resources Research Organization & Harcourt Assessment. (2007).
Reading and mathematics technical report for 2006 FCAT test admin-
istrations. San Antonio, TX: Author.
Jackson, P. W. (2000). John Dewey and the lessons of art. New Haven, CT:
Yale University Press.
Jager, J., Putnick, D. L., & Bornstein, M. H. (2017). II. More than just
convenient: The scientific merits of homogeneous convenience samples
[monograph]. Monographs of the Society for Research in Child Devel-
opment, 82, 13–30. http://dx.doi.org/10.1111/mono.12296
Jaschke, A. C., Honing, H., & Scherder, E. J. A. (2018). Longitudinal
analysis of music education on executive functions in primary school
children. Frontiers in Neuroscience, 12, 103. http://dx.doi.org/10.3389/
fnins.2018.00103
Jensen, E. (2004). Art with the brain in mind. Alexandria, VA: Association
for Supervision and Curriculum Development.
Keinänen, M., Hetland, L., & Winner, E. (2000). Teaching cognitive skills
through dance: Evidence for near but not far transfer. Journal of Aes-
thetic Education, 34, 295–306. http://dx.doi.org/10.2307/3333646
Kim, Y. O. (2007). Effects of self-directed dance learning on high school
girl students’ academic motivation and problem-solving ability. Journal
of Korean Physical Education Association for Women, 21, 99 –111.
Kirschner, S., & Tomasello, M. (2010). Joint music making promotes
prosocial behavior in 4-year-old children. Evolution and Human Behav-
ior, 31, 354 –364. http://dx.doi.org/10.1016/j.evolhumbehav.2010.04
.004
Kisida, B., Morrison, B., & Tuttle, L. (2017, May). To elevate the role of
arts education, measure it. Washington, DC: Brookings Institute. Re-
trieved from https://www.brookings.edu/research/to-elevate-the-role-of-
arts-education-measure-it/
Lanfredi, S. C. (2013). Formal dance training, cognitive ability, and
academic performance of adolescent females (Doctoral dissertation).
Fordham University, Bronx, NY.
Lareau, A. (2002). Invisible Inequality: Social class and childrearing in
black families and white families. American Sociological Review, 67,
747–776. http://dx.doi.org/10.2307/3088916
LeBuffe, P., & Naglieri, J. A. (1999). The Devereux Early Childhood
Assessment. Lewisville, NC: Kaplan.
Lobo, Y. B., & Winsler, A. (2006). The effects of a creative dance and
movement program on the social competence of head start preschoolers.
Social Development, 15, 501–519.
McCarthy, K. F., Ondaatje, E. H., Zakaras, L., & Brooks, A. (2004). Gifts
of the muse: Reframing the debate about the benefits of the arts. Santa
Monica, CA: RAND Cooperation.
McMahon, S. D., Rose, D. S., & Parks, M. (2003). Basic reading through
dance program: The impact on first-grade students’ basic reading skills.
Evaluation Review, 27, 104 –125. http://dx.doi.org/10.1177/0193841X0
2239021
McNeal, R. B., Jr. (1998). High school extracurricular activities: Closed
structures and stratifying patterns of participation. The Journal of Edu-
cational Research, 91, 183–191. http://dx.doi.org/10.1080/0022067980
9597539
Mehr, S. A., Schachner, A., Katz, R. C., & Spelke, E. S. (2013). Two
randomized trials provide no consistent evidence for nonmusical cogni-
tive benefits of brief preschool music enrichment. PLoS ONE, 8, e82007.
http://dx.doi.org/10.1371/journal.pone.0082007
Metcalfe, T. (2016). The contribution of different musical variables to the
effect of background music on motor behavior. Psychomusicology:
Music, Mind, and Brain, 26, 132–138. http://dx.doi.org/10.1037/
pmu0000137
Miami-Dade County Public Schools. (2008). District ELL plan (2008 –
2009). Retrieved from http://bilingual.dadeschools.net/BEWL/pdfs/
ELL_Plan_08-09.pdf
Minton, S. (2003). Assessment of high school students’ creative thinking
skills: A comparison of the effects of dance and non-dance classes.
Research in Dance Education, 4, 31– 49. http://dx.doi.org/10.1080/
14647890308307
Moore, J. R. (2004). 50 years after Brown: Segregation in the Miami-Dade
county public schools. Equity & Excellence in Education, 37, 289 –301.
http://dx.doi.org/10.1080/10665680490491029
Moreno, S., Bialystok, E., Barac, R., Schellenberg, E. G., Cepeda, N. J., &
Chau, T. (2011). Short-term music training enhances verbal intelligence
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
17
SELECTION INTO DANCE AND OUTCOMES

and executive function. Psychological Science, 22, 1425–1433. http://
dx.doi.org/10.1177/0956797611416999
Morgan, C., & Stengel-Mohr, J. (2014). Language at the core: Developing
meaningful connections between the arts and classroom instruction. NYS
TESOL Journal, 1, 90 –99.
National Endowment for the Arts. (2012). How arts works: The National
Endowment for the arts’ five-year research agenda, with a system map
and measurement model. Washington, DC: Author.
National Endowment for the Arts. (2013). How a nation engages with art:
Highlights from the 2012 survey of public participation in the arts.
Research Report #57. Washington, DC: Author.
Nehring, A. D., Nehring, E. F., Bruni, J. R., & Randolph, P. L. (1992).
Learning Accomplishment Profile Diagnostic (LAPD). Lewisville, NC:
Kaplan Press.
O’Hagan, J. W. (2014). Attendance at/participation in the arts by educa-
tional level: Evidence and issues. Homo Oeconomicus, 31, 411– 429.
O’Neill, J. R., Pate, R. R., & Liese, A. D. (2011). Descriptive epidemiology
of dance participation in adolescents. Research Quarterly for Exercise
and Sport, 82, 373–380. http://dx.doi.org/10.1080/02701367.2011
.10599769
Park, J. G. (2007). The effects of dance classes with problem based
learning on middle school students’ critical thinking and problem-
solving ability. Korean Journal of Sport Psychology, 18, 105–122.
Parsad, B., Spiegelman, M., & Coopersmith, J. (2012). Arts education in
public elementary and secondary schools: 1999 –2000 and 2009 –10
(NCES 2012– 014). Washington, DC: National Center for Education
Statistics, Institute of Education Sciences, U.S. Department of Educa-
tion.
Penfield, R. D. (2010). Test-based grade retention: Does it stand up to
professional standards for fair and appropriate test use? Educational
Researcher, 39, 110 –119. http://dx.doi.org/10.3102/0013189X
10363007
Peterson, P. V. (2007). What is measured is treasured: The impact of the
No Child Left Behind Act on nonassessed subjects. The Clearinghouse:
A Journal of Educational Strategies and Issues, 80, 287–291.
Posner, M., & Patoine, B. (2009). How arts training improves attention
and cognition. New York, NY: The Dana Foundation.
Rabinowitch, T. C., & Meltzoff, A. N. (2017). Synchronized movement
experience enhances peer cooperation in preschool children. Journal of
Experimental Child Psychology, 160, 21–32. http://dx.doi.org/10.1016/
j.jecp.2017.03.001
Rabkin, N., & Hedberg, E. C. (2011). Arts education in America: What the
declines means for arts participation. Research Report #52. Washing-
ton, DC: National Endowment for the Arts.
Risner, D. (2007). Rehearsing masculinity: Challenging the ‘boy code’ in
dance education. Research in Dance Education, 8, 139 –153. http://dx
.doi.org/10.1080/14647890701706107
Rossberg-Gempton, I. E., Dickinson, J., & Poole, G. (1999). Creative
dance: Potentiality for enhancing social functioning in frail seniors and
young children. The Arts in Psychotherapy, 26, 313–327. http://dx.doi
.org/10.1016/S0197-4556(99)00036-2
Schellenberg, E. G. (2004). Music lessons enhance IQ. Psychological
Science, 15, 511–514. http://dx.doi.org/10.1111/j.0956-7976.2004
.00711.x
See, B. H., & Kokotsaki, D. (2015). Impact of arts education on the
cognitive and non-cognitive outcomes of school-aged children: A review
of evidence. Project Report. Durham, NC: Education Endowment Foun-
dation.
Seham, J. (1997). The effects of at-risk children of an in-school dance
program (Doctoral dissertation) Adephi University, Garden City, NY.
Shadish, W. R., Cook, T. D., & Campbell, D. T. (2002). Experimental and
quasi-experimental designs for generalized causal inference. Boston,
MA: Houghton Mifflin.
Skoning, S. N. (2008). Movement and dance in the inclusive classroom.
Teaching Exceptional Children Plus, 4, 2–11.
Smiley, D. (2014, May 19). Re-segregation’ trend: 60 years after ruling,
dozens of Miami-Dade schools remain isolated’. Miami Herald. Re-
trieved from http://www.miamiherald.com/news/local/community/
miami-dade/article1964672.html
Smithrim, K., & Upitis, R. (2005). Learning through the arts: Lessons of
engagement. Canadian Journal of Education, 28, 109 –127. http://dx.doi
.org/10.2307/1602156
Tavassolie, T., & Winsler, A. (in press). Predictors of mandatory 3rd grade
retention from high-stakes test performance for low-income, ethnically
diverse children. Early Childhood Research Quarterly.
Tunçgenç, B., & Cohen, E. (2016). Movement synchrony forges social
bonds across group divides. Frontiers in Psychology, 7, 782. http://dx
.doi.org/10.3389/fpsyg.2016.00782
Tunçgenç, B., & Cohen, E. (2018). Interpersonal movement synchrony
facilitates pro-social behavior in children’s peer-play. Developmental
Science, 21, 1–9. http://dx.doi.org/10.1111/desc.12505
U.S Census Bureau. (2015). Quick facts: Miami-Dade County, Florida-
Race and Hispanic origin. Retrieved from https://www.census.gov/
quickfacts/table/RHI125215/12086
Walker, P. (2016). Cross-sensory correspondences: A theoretical frame-
work and their relevance to music. Psychomusicology: Music, Mind, and
Brain, 26, 103–116. http://dx.doi.org/10.1037/pmu0000130
Wigfield, A., & Eccles, J. S. (2002). Students’ motivation during the
middle school years. In J. Aronson (Ed.), Improving academic achieve-
ment: Impact of psychological factors on education (pp. 159 –184). New
York, NY: Academic Press. http://dx.doi.org/10.1016/B978-012064455-
1/50011-7
Williams, K. E., Barrett, M. S., Welch, F. G., Abad, V., & Broughton, M.
(2015). Associations between early shared music activities in the home
and later child outcomes: Finds from the Longitudinal Study of Austra-
lian Children. Early Childhood Research Quarterly, 31, 113–124. http://
dx.doi.org/10.1016/j.ecresq.2015.01.004
Winner, E., & Cooper, M. (2000). Mute those claims: No evidence (yet) for
a causal link between arts study and academic achievement. Journal of
Aesthetic Education, 34, 11–75. http://dx.doi.org/10.2307/3333637
Winner, E., Goldstein, T. R., & Vincent-Lancrin, S. (2013). Art for art’s
sake? East Jerusalem, Israel: OECD Publishing. http://dx.doi.org/10
.1787/9789264180789-en
Winsler, A., Ducenne, L., & Koury, A. (2011). Singing one’s way to self
regulation: The role of early music and movement curricula and private
speech. Early Education and Development, 22, 274 –304. http://dx.doi
.org/10.1080/10409280903585739
Winsler, A., Gara, T. V., Alegrado, A., Castro, S., & Tavassolie, T. (2019).
Selection Into, and academic benefits from, arts-related courses in mid-
dle school among low-income, ethnically diverse youth. Psychology of
Aesthetics, Creativity, and the Arts. Advance online publication. http://
dx.doi.org/10.1037/aca0000222
Winsler, A., Hutchison, L. A., De Feyter, J. J., Manfra, L., Bleiker, C.,
Hartman, S. C., & Levitt, J. (2012). Child, family, and childcare pre-
dictors of delayed school entry and kindergarten retention among lin-
guistically and ethnically diverse children. Developmental Psychology,
48, 1299 –1314. http://dx.doi.org/10.1037/a0026985
Winsler, A., Tran, H., Hartman, S., Madigan, A. L., Manfra, L., & Bleiker,
C. (2008). School readiness gains made by ethnically-diverse children in
poverty attending center-based childcare and public school pre- kinder-
garten programs. Early Childhood Research Quarterly, 23, 314 –329.
http://dx.doi.org/10.1016/j.ecresq.2008.02.003
Received January 23, 2018
Revision received March 4, 2019
Accepted March 5, 2019 䡲
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
18 GARA AND WINSLER