Predictors of Academic Performance 1
Running head: PREDICTORS OF ACADEMIC PERFORMANCE
Abstract
In this study, we examined the relative influence of physical and psychosocial variables on math
and reading achievement test scores. Between one and five months prior to taking annual
standardized reading and math tests, 1211 6th through 8th graders (53.7% girls; 57.2% White)
self-reported levels of physical activity, academic self-concept, self-esteem, and social support
and participated in objective testing to obtain measures of body composition (BMI) and
cardiorespiratory fitness. Socio-economic status (SES) and reading and math test scores were
provided by the school district at the end of the school year. Regression analyses revealed that,
after controlling for SES and academic self-concept, only cardiorespiratory fitness was a
consistent predictor of the students’ performance in reading and math; perceived social support
from family and friends was related to improvements in the boys’ reading scores. Our findings
support schools re-examining policies that have limited students’ involvement in physical
education classes.
KEYWORDS: Academic performance, physical fitness, social support, self-esteem, middle
school students
Predictors of Academic Performance 2
Physical and Psychological Predictors of Academic Performance
For Middle School Boys and Girls: A Longitudinal Investigation
Academic achievement, particularly among middle school students, is multiply
determined, being influenced not only by school-related variables, such as academic self-concept
(Green, Nelson, Martin, & Marsh, 2006; Osborne & Jones, 2011; Pintrich & De Groot, 1990;
Stringer & Heath, 2011), but by physical and psychosocial factors as well (e.g., Ahmed,
Minnaert, van der Werf, & Kuyper, 2010; Edwards, Mauch, & Winkelman, 2011; Kristjansson,
Sigfusdottir, & Allegrante, 2010; Rapport, Denney, Chung, & Hustace, 2001; Roberts, Freed, &
McCarthy, 2010). For example, better academic performances, measured through achievement
test scores and grades, have been related to higher levels of cardiorespiratory fitness and/or
physical activity (e.g., Castelli, Hillman, Buck, & Erwin, 2007; Chomitz et al., 2009; EvelandSayers, Farley, Fuller, Morgan, & Caputo, 2009; Singh Uijtdewilligen, Twisk, Mechelen, &
Chinapaw, 2012), lower body mass index (e.g., Cottrell, Northrup, & Wittberg, 2007; Roberts et
al., 2010; Welk, Jackson, Morrow, Haskell, Meredith, & Cooper, 2010), more social support
(e.g., Ahmed et al., 2010; Edwards et al., 2010; Rosenfeld, Richman & Bowen, 2000), and
greater self-esteem and less depression (e.g., Lundy, Silva, Kaemingk, Goodwin, & Quan, 2010;
Rapport et al., 2001). Research, however, has been limited in three important ways. First, many
studies have been cross-sectional (or retrospective) in design. Second, few studies have
examined simultaneously physical (e.g., body composition, cardiorespiratory fitness) and
psychosocial (e.g., self-esteem) variables to determine their relative importance in predicting
students’ academic performance. Third, although some studies have included for important
demographic variables (e.g., SES; Cottrell et al., 2007; Roberts et al., 2010), most have not
controlled for the effects of prior academic readiness. Thus, in the current study, we utilize a
Predictors of Academic Performance 3
prospective design to examine the potential influence of variables across multiple domains, while
controlling for the effects of relevant demographic (i.e., sex, SES) and academic factors.
Academic Self-Concept
Academic self-concept, which is defined as students’ perception of their proficiency (and
confidence) in general and/or specific academic areas, has been related positively to academic
performance (Green et al., 2006; Osborne & Jones, 2011; Pintrich & De Groot, 1990; Stringer &
Heath, 2008). For example, Pintrich and De Groot (1990) found, in a cross-sectional study of
male and female 7th graders, that academic self-efficacy (as defined by feelings of competence
and confidence in class work) was related to better performances on homework, quizzes/exams,
essays/reports, and overall course grades. In a longitudinal study, 4th and 5th grade boys’ and
girls’ academic self-perceptions of competence in both reading and mathematics, respectively,
were significant predictors of their scores on standardized reading and math tests one year later
(Stringer & Heath, 2008). Students with high levels of confidence in their academic abilities
may be more motivated to study (and put forth more effort doing so), set more challenging
academic goals, use more effective studying and test-taking strategies, and perform even better
on subsequent academic tasks. Thus, to determine accurately the extent to which physical and
psychosocial variables explain students’ academic performances, they need to control for either
prior academic achievement or, if not available, students’ current academic self-concept.
Body Composition
Body composition generally is represented by individuals’ body mass index (BMI), a
proportional measure of weight taking into account height. Studies examining its relationship to
academic performance have been equivocal, demonstrating either no association (e.g., Edwards
et al., 2011) or small effects (Cottrell et al., 2007; Kristjansson et al., 2010; Roberts et al., 2010;
Predictors of Academic Performance 4
Welk et al., 2010). Other researchers have reported zero-order, inverse relationships between
overweight status (as measured by BMI) and state-mandated achievement test scores in reading
and math, but also noted that these negative effects were attenuated when considered in
conjunction with cardiorespiratory fitness (Cottrell et al., 2007; Roberts et al., 2010). These
findings indicate that it is fitness levels, not obesity per se, which may be the primary influence
on children’s achievement. Cottrell et al. (2007) suggested that future research incorporate a
longitudinal design and examine not only body composition and fitness levels, but also social,
emotional, and environmental factors so as to understand their relative importance.
Physical Activity and Physical Fitness
Being physically active and fit has been associated with improved academic
performances in grades and test scores (Chomitz et al., 2009; Dwyer et al., 2001; Edwards et al.,
2011; Fox, Barr-Anderson, Neumark-Sztainer & Wall, 2010; Kristjansson et al., 2010; Roberts et
al., 2010). For example, Hillman et al. (2009) found, in an experimental study of male and
female preadolescent children, that a single 20-minute bout of treadmill walking led to improved
response accuracy in an incongruent visual stimuli task as well as higher scores on a
standardized test of reading comprehension. In a sample of male and female middle and high
school students, Fox et al. (2010) discovered that more time spent being physically active (mild,
moderate, or vigorous levels) was associated with higher grade point averages, across gender and
year in school. Singh et al. (2012), in their systematic review of the literature, concluded that
there was strong evidence to link physical activity to higher levels of academic performance.
Researchers also have identified several potential causal mechanisms linking physical activity to
improvements in academic performance, including increased blood flow to the brain, neural
activity (e.g., P3 amplitude), and response accuracy, increase cognitive controls (e.g., attentional
Predictors of Academic Performance 5
focus), and improved memory function (Aberg et al., 2009; Chaddock, Pontifex, Hillman, &
Kramer, 2011; Chomitz et al., 2009; Hillman et al., 2007; Pontifex et al., 2011).
Given the connection between physical activity and cardiorespiratory fitness, it is not
surprising that fitness levels also have been related to academic performance (e.g., Aberg et al.,
2009; Cottrell et al., 2007; Edwards et al., 2011; Wittberg, Cottrell, Davis, & Northrup, 2010;
Roberts et al., 2010). For example, in a cross-sectional study of 5th, 7th, and 9th grade boys and
girls, students whose one-mile run/walk times exceeded FITNESSGRAM® (The Cooper
Institute, 2007) standards performed significantly worse on state reading, math, and language
tests than those whose times met standards (Roberts et al., 2010). Using data supplied by the
California Department of Education, Grissom (2005) found that as male and female 5th, 7th, and
9th graders’ overall performance on the FITNESSGRAM tests improved, so did their scores on
state achievement tests in reading and mathematics. Among 5th grade boys and girls, overall
fitness level, as determined through FITNESSGRAM testing, was related positively to stateachievement test scores in social studies, reading, math, and science, even after controlling for
SES and body composition (Cottrell et al., 2007). Although physical activity and physical fitness
have been associated independently with better academic performances, these factors have not
been examined extensively in relation to other potential predictors to determine their relative
influence.
Social Support and Self-Esteem
Although different psychosocial factors have been examined as potential predictors of
academic performance, social support and self-esteem have received considerable support
(Ahmed et al., 2008; Edwards et al., 2011; Rosenfeld et al., 2000). Social support has been
conceptualized as the degree to which individuals are satisfied with the different types of support
Predictors of Academic Performance 6
they receive, such as problem-solving and emotional assistance, and how much they can rely on
different people in their lives, such as from family and friends, fore such support (Zimet,
Dahlem, Zimet, & Farley, 1988). Middle and high school students with stronger perceptions of
social support from peers, parents, and teachers not only earned better grades in their classes, but
also had better attendance, higher levels of engagement in classes, and higher levels of
satisfaction with their school experience (Rosenfeld et al., 2000). Social support is likely to help
students be more confident in themselves and their abilities and have more control in challenging
academic situations, which in turn can make their attitude more positive, their motivation
stronger, and their academic performance better (Ahmed et al., 2008; Rosenfeld et al., 2000).
Self-esteem has been conceptualized as representing the extent to which individuals view
themselves as effective and capable, and are satisfied with and proud of themselves as they
currently are (Marsh, 1992). In a nationally-based, cross-sectional sample of male and female
adolescents, self-esteem was related to higher self-reported grades in core subject areas
(Kristjansson et al., 2008). Self-esteem is associated with lower levels of anxiety and depression,
as well as higher levels of optimism, all of which may help students pay better attention in
classes and be more motivated toward their studies (Lundy et al., 2010; Rapport et al., 2001).
Students who feel positively about themselves and their lives and believe they generally are
capable and effective may bring the necessary cognitive processing skills and maintain the
needed attentional focus and effort to improve their academic performances. However, the
influence of social support and self-esteem on academic performance, relative to other factors,
such as cardiorespiratory fitness, needs further study (Cottrell et al., 2007).
The Current Study
Predictors of Academic Performance 7
Using a longitudinal design, our purpose was to examine the relative influences of body
composition, cardiorespiratory fitness, physical activity, self-esteem, and social support on
middle school students’ achievement on state administered reading and mathematics
examinations. Given the strong relationships that exist between academic performance and
socio-economic status (Cottrell et al., 2007) and academic self-concept (e.g., Green et al., 2006),
we included these two variables to control for their effects. In doing so, we could determine the
extent to which the physical and psychosocial variables influenced academic performance above
and beyond that explained by the students’ belief in their academic abilities and the opportunities
(e.g., financial, educational) that may have been afforded to them by coming from a more
affluent family. Also, because there is evidence that the relationships between these variables
and academic achievement are different for boys and girls (e.g., Eveland-Sayers et al., 2009;
Grissom, 2005), we conducted the analyses separately for the male and female students to
examine this possibility. We hypothesized that, after controlling for SES and academic selfconcept, cardiorespiratory fitness, social support, and self-esteem would be positive predictors in
each area of academic achievement. We expected that any bivariate relationships of physical
activity levels and body composition to achievement test scores would be attenuated in the
multivariate model.
Method
Participants
Participants were 1211 middle school students (650 girls) who were drawn from the five
middle schools in a suburban school district in Texas. The boy’s mean age was 12.45 years (SD
= 1.00); 38.1% were in 6th grade, 37.4% in 7th grade, and 24.4% in 8th grade. In terms of
race/ethnicity, 57.2% were Caucasian, 24.2% were Mexican-American, 9.1% were African-
Predictors of Academic Performance 8
American, 1.1% were Asian, and 1.2% were American Indian. Consistent with past research
(Cottrell et al., 2007), SES was based on federal guidelines for determining students’ status for
free or reduced lunch based on family income: 24.1% received free lunch, 6.2% received
reduced lunch, and 69.7% did not receive any reduction on their meals. Mean Body Mass Index
(BMI) was 21.58 kg/m2 (SD = 7.92).
Mean age for the girls was 12.29 years (SD = 0.92); 37.8% were in 6th grade, 35.5% in 7th
grade, and 26.6% in 8th grade. Regarding race/ethnicity, 58.6% were Caucasian, 23.4% were
Mexican-American, 9.2% were African-American, 2.3% were Asian, and 0.6% were American
Indian. In terms of SES, 24.6% received free lunch, 4.6% received reduced lunch, and 70.8% did
not receive any reduction on their meals. Mean BMI was 21.01 kg/m2 (SD = 4.71).
Measures
Demographic information. The school district provided information on the students’
race/ethnicity, age, grade level, and whether or not they qualified for free or reduced lunch
(which served as our proxy measure of SES).
Academic achievement. The school district provided the students’ scores on the state’s
standardized reading and mathematics examinations (i.e., The Texas Assessment of Knowledge
and Skills [TAKS]). Each school district in the state administers these examinations on the same
dates (for grade and exam) during the month of April each academic school year. Examinations
are scored at the state-level and then reported to each district.
Academic self-concept. Three single items from the Self-Description Questionnaire II
(SDQII; Marsh 1992) measure general and specific aspects of academic self-concept, including
“I am good at mathematics,” “I like reading,” and “I am good at most school subjects.”
Participants rate each item on a 6-point scale that ranges from 1, false, to 6, true. Marsh (1992)
Predictors of Academic Performance 9
has provided extensive information regarding the validity of the items as measures of academic
self-concept.
Cardiorespiratory fitness and body composition. The FITNESSGRAM (The Cooper
Institute, 2007) provides an objective measure of cardiorespiratory fitness through the PACER
(Progressive Aerobic Cardiovascular Endurance Run) test and body composition through the
students’ body mass index (BMI), which is represented in kg/m2. Administered by trained
professionals, the PACER is represented by the number of 20-meter laps students complete
within a specified timeframe and pace. Weight was measured by the researchers (in conjunction
with the physical education teachers at each school) using a Seca digital scale (Model 882) and
recorded to the nearest .1 lb; scales were recalibrated at the beginning of each testing day. Height
and weight was transformed into BMI within the FITNESSGRAM program. The
FITNESSGRAM/ACTIVITY manual (The Cooper Institute, 2007) provides extensive
information about the validity and reliability of the PACER and BMI as measures of
cardiorespiratory fitness and body composition, respectively.
Physical activity. The FITNESSGRAM (The Cooper Institute, 2007) provides three selfreport questions to assess how often individuals participated in activities that were focused on
improving aerobic fitness, strength, and flexibility. Each item is rated in terms of the number of
days, out of the last seven, they engaged in the described physical activities (e.g., for aerobic
fitness, a timeframe of 30 to 60 minutes or more per day is specified). Thus, scores range from 0
to 7; higher numbers indicate more days (during the last week) that the individual engaged in that
type of physical activity at the required level. The FITNESSGRAM/ACTIVITY manual (see The
Cooper Institute, 2007) provides extensive information about these questions as reliable and
valid representations of physical activity.
Predictors of Academic Performance 10
Self-esteem. The 10-item general self-esteem scale from the Self- Description
Questionnaire II (SDQII-GSE; Marsh 1992) measures how proud and satisfied adolescents are
with themselves. On items such as “Overall, I have a lot to be proud of,” participants respond
using a 6-point scale that ranges from 1, false, to 6, true. Total score is the mean; higher scores
represent greater esteem. Marsh, Ellis, Parada, Richards, and Heubeck (2005) reported
Cronbach’s alphas that ranged from .80 to .89 in a sample of male and female adolescents;
Cronbach’s alpha for the current sample was .88. Marsh and colleagues (1992; Marsh et al.,
2005) have provided extensive information regarding the scale’s validity.
Social support. Eight items from the Multidimensional Scale of Perceived Social
Support (MSPSS; Zimet et al., 1988) were used to measure how much help and support
participants believe they receive from friends and family. On items such as “My family helps me
make decisions,” participants respond using a 7-point scale that ranges from 1, very strongly
disagree, to 7, very strongly agree. Total score is the mean; higher scores represent more
perceived social support from friends and family. Zimet et al. (1988) reported a Cronbach’s
alpha of .88 and a two to three month test-retest reliability of .85 in a sample of male and female
undergraduates; Cronbach’s alpha for the current sample was .89. Zimet and colleagues (Zimet
et al., 1988; Zimet, Powell, Farley, & Werkman, 1990) have provided extensive information
regarding the scale’s validity.
Procedure
Approval for the study was received from the university’s IRB for Human Subjects
Research as well as from the school district’s associate superintendent and the principals at each
of the five middle schools. Prior to participating in the study, parental consent and child assent
was obtained. At each school, the authors assisted PE instructors in the administration of the
Predictors of Academic Performance 11
FITNESSGRAM protocol to obtain the state-required, annual fitness testing results during the
2009–2010 academic year.
FITNESSGRAM testing occurred at each school during a one week period that was
scheduled between November 2009 and March 2010; the testing dates were determined by the
principal at each school (Time 1). During each week-long session, the children completed the
PACER as well as had their height and weight measured. The students from whom consent and
assent had been obtained also completed the questionnaires during their PE classes.
Questionnaires took approximately 15 minutes to finish. To link questionnaire data to the results
from their FITNESSGRAM tests and with data supplied by the district, participants put their
student ID numbers (but no other identifying information) on their questionnaires. Following
completion of the questionnaires, students were entered into a random drawing for a series of
cash prizes that were given away at each school. Because the TAKS is a state mandated and
administered achievement test, all students across the five middle schools took the math and
reading examinations on the same dates in April, 2010 (Time 2). Depending on when fitness
testing was scheduled at each middle school, FITNESSGRAM and questionnaire data were
collected one to five months prior to the students taking their TAKS examinations.
Data Analysis
First, we addressed the issue of missing data and found that only between 0% and 1.9%
were missing across the questionnaire items. Because items were either missing completely at
random or at random, we replaced the values using the expectation maximization procedure
(Schlomer, Bauman, & Card, 2010). Second, we examined the distributional properties (e.g.,
skewness) of all the measures and found them to be within acceptable levels so no
Predictors of Academic Performance 12
transformations were made to the data. Next, we computed the zero-order correlations among all
the variables separately by sex.
To determine the relative contribution of the different variables in predicting the boys’
and girls’ performance on their math and reading examinations, we used hierarchical multiple
regression and conducted the analyses separately by sex. For each regression model, there were
five steps. At Step 1, we entered the families’ SES level to control for influences of social
affluence, prosperity, and education (Cottrell et al., 2007; Edwards et al., 2011; Roberts et al.,
2010). At Step 2, we entered the reading or math self-concept score (matched to the achievement
test being used as the criterion variable) as well as the overall academic self-concept item. We
entered these variables at this step to control for influences related to students’ perceptions of
their academic abilities (e.g., Green et al., 2006). At Step 3, we entered the number of PACER
laps, which represented the student’s level of cardiorespiratory fitness, and their BMI (i.e., body
composition). At Step 4, we entered the three variables for physical activity – aerobic, strength,
and flexibility. Finally, at Step 5, we entered the students’ general self-esteem and perceived
social support.
Results
Descriptive Statistics
Correlations, means and standard deviations for both the predictor and criterion variables
are presented by sex in Table 1. The highest correlation among the predictor variables was .56,
which suggests that multi-colinearity would not be an issue during the regression analyses.
With the exception of BMI (which was related only to reading test scores), aerobic
physical activity level (which was related only for boys), and flexibility physical activity level
(which was related only for girls), the bivariate relationships between the predictor variables and
Predictors of Academic Performance 13
each measure of academic achievement were significant and in the expected direction. For
example, greater self-esteem, more social support, and higher levels of cardiorespiratory fitness
all were related to better performance on the math and reading tests for boys and girls.
Regression Analyses
Reading achievement. For the boys, the inclusion of SES at Step 1 was significant,
accounting for 9.7% of the variance, Adj. R2 = .095, F (1, 559) = 59.78, p < .0001. Step 2 of the
model, in which we included the academic self-concept measures, was significant and accounted
for an additional 5.2% of the variance, F (2, 557) = 17.17, p < .0001. At Step 3, entry of the
PACER and BMI accounted for an additional 3.9% of the variance, F (2, 555) = 13.19, p <
.0001. At Step 4, the inclusion of the physical activity measures was not significant, F (3, 552) =
1.17, p = .32, ΔR2 = .005. Finally, at Step 5, the addition of self-esteem and social support
accounted for an additional 2.6% of the variance, F (2, 550) = 9.19, p < .0001. The overall
model was significant, accounting for 21.9% (Adj. R2 =.205) of the variance in the boys’ reading
scores, F (10, 550) = 15.41, p < .001. Within the full model, after controlling for household SES
level and the boys’ general and reading-specific academic self-concept, only their
cardiorespiratory fitness levels (i.e., PACER) and their perceived social support from family and
friends predicted better performances on the TAKS reading exam. See Table 2 for the detailed
statistics from Step 5 of the regression analysis.
For the girls, the inclusion of SES level was significant, accounting for 14.5% of the
variance, Adj. R2= .143, F (1, 648) = 109.61, p < .0001. Step 2, which included entry of the
academic self-concept measures, accounted for an additional 7.3% of the variance, F (2, 646) =
30.24, p < .0001. At Step 3, the girls’ cardiorespiratory fitness levels and body composition
accounted for an additional 4.6% of the variance, F (2, 644) = 20.19, p < .0001. At Step 4, the
Predictors of Academic Performance 14
physical activity measures were not significant, accounting for less than 1% of the variance, F (3,
641) = 1.78, p = .15. Finally, at Step 5, the addition of the girls’ self-esteem and perceived social
support was not significant, explaining only 0.2% of the variance, F (2, 639) = 0.930, p = .40.
The overall model was significant, accounting for 27.2% of the variance (Adj. R2 =.261) of the
girls’ reading scores, F (10, 639) = 23.91, p < .0001. Within the full model, after controlling for
the girls’ families’ SES and their academic self-concept, only their cardiorespiratory fitness and
their body composition predicted their reading scores (see Table 2).
Math achievement. For the boys, the entry of SES level was significant, accounting for
8.0% of the variance, Adj. R2 = .080, F (1, 623) = 54.46, p < .0001. Step 2, in which we entered
the academic self-concept measures, was also significant, F (2, 621) = 62.41, p < .0001, ΔR2 =
.15. At Step 3, the boys’ cardiorespiratory fitness levels and their body composition accounted
for an additional 4.7% of the variance, F (2, 619) = 20.11, p < .0001. At Step 4, the inclusion of
the physical activity measures was not significant, explaining less than 1% additional variance, F
(3, 616) = 2.40, p = .07. Finally, at Step 5, the inclusion of the boys’ self-esteem and perceived
social support also was not significant, F (2, 614) = 1.26, p = .28, ΔR2 = .003. The overall model
was significant and accounted for 29.2% (Adj. R2 =.281) of the variance of the boy’s
mathematics scores, F (10, 614) = 25.35, p < .0001. After controlling for the boys’ families’
SES levels and their academic self-concept, only their level of cardiorespiratory fitness predicted
how they performed on the math TAKS test (see Table 3).
For the girls, the inclusion of SES at Step 1 of the model was significant, accounting for
9.9% of the variance, Adj. R2 = .098, F (1, 648) = 71.50, p < .0001. At Step 2, inclusion of the
academic self-concept measures was also significant, F (2, 646) = 73.42, p < .0001, ΔR2 = .17.
At Step 3, the girls’ cardiorespiratory fitness level and body composition were significant,
Predictors of Academic Performance 15
explaining an additional 2.3% of the variance, F (2,644) = 10.36, p < .0001. At Step 4, the
inclusion of the physical activity measures was not significant, F (3, 641) = 2.65, p = .05, ΔR2 =
.01. Finally, at Step 5, self-esteem and perceived social support were not significant, F (2, 639) =
0.42, p = .66, accounting for less than 0.1% of the variance. The overall model was significant,
accounting for 29.9% (Adj. R2 =.288) of the variance of the girls’ mathematics scores, F (10,
639) = 27.22, p < .0001. After controlling for their families’ SES level and their academic selfconcept, only the girls’ cardiorespiratory fitness explained their performance on the math TAKS
test (see Table 3).
Discussion
Except for BMI, aerobic activity, and flexibility, the physical and psychological variables
were related in the expected directions with the math and reading achievement scores. These
variables shared between 2% and 20% of the variance across the two measures of academic
achievement, supporting previous research that has demonstrated that physical activity levels,
physical fitness, body composition, and various measures of psychosocial functioning (e.g.,
social support) are associated with better academic performances among children and
adolescents (e.g., Cottrell et al., 2007; Edwards et al., 2011; Grissom, 2005; Hillman et al., 2009;
Kristjansson et al., 2010; Roberts et al., 2010; Rosenfeld et al., 2000). However, when all the
variables were considered within the regression analyses, only certain effects remained
significant as expected. Because of our study’s longitudinal design, these variables can now be
considered “risk factors” (Stice, 2002) in relation to middle school students’ performances on
state-mandated math and reading examinations. The identification of specific risk factors is
essential for the development of data-based, effective prevention programs.
Predictors of Academic Performance 16
Although there were two differences across sex (which we discuss in more detail below),
findings were consistent in terms of which variables were risk factors and how much variance
they explained in the achievement test scores. As expected, both SES, and the measures of
academic achievement were significant predictors of the girls’ and boys’ performances. Those
students who came from higher SES homes earned higher scores on the math and reading
examinations, which is consistent with past research (Cottrell et al., 2007; Edwards et al., 2011).
In addition, general and specific measures of academic self-concept were related to the students’
test scores, such that the boys and girls who either believed they were good in most of their
school subjects, liked reading, or thought they were good at math actually performed better.
Similarly, Stringer and Heath (2008) found that 4th and 5th grade boys’ and girls’ self-perceptions
of their academic competence predicted their performance on standardized achievement tests
taken one year later. Students who perform well on academic tasks, and attribute their success to
internal factors, such as effort, are likely to see increases in their academic self-concept in
general, but also specific to the area where they have experienced the success. Thus, providing
students with the tools to be academically successful and helping them develop a belief in their
abilities can have long-term positive effects on their academic performances.
The only variable to be related consistently to the students’ performance was their level
of cardiorespiratory fitness. Consistent with past studies (e.g., Aberg et al., 2009; Cottrell et al.,
2007; Edwards et al., 2011; Grissom, 2005; Roberts et al., 2010), the more fit the students’ were
in terms of their cardiorespiratory functioning, the better they performed on both the math and
reading examinations. This finding is noteworthy for three reasons. First, because of our
longitudinal design, we can conclude that middle school boys’ and girls’ level of
cardiorespiratory fitness influences how they perform on subsequent academic tests. Higher
Predictors of Academic Performance 17
levels of fitness also are associated with improvements in cognitive control, including inhibition
(ability to selectively pay attention to relevant information) and working memory, as well as
increases in response accuracy and more cognitive flexibility to handle task demands (e.g.,
Chaddock et al., 2011; Pontifex et al., 2011), which are suggested to explain the improvements in
academic performances. Second, the relationship between fitness and academic performance
existed even after we had controlled for the influences of the students’ family SES and their
academic self-concept, variables that have been strongly related to achievement test scores (e.g.,
Cottrell et al., 2007). Third, although higher levels of physical activity (i.e., aerobic, strength,
and flexibility) were associated with higher levels of cardiorespiratory fitness, particularly
among the boys in our study, it was only the students’ fitness levels that predicted their academic
performances. As the results from longitudinal studies, such as ours, continue to establish a
temporal association between fitness and subsequent improvements in academic measures,
researchers will want to continue to study the neurocognitive and physiological mechanisms that
may underlie this relationship (Chaddock et al., 2011).
The relations between the physical and psychosocial variables and the two measures of
academic achievement differed slightly for the boys and girls, which had been suggested by past
research (e.g., Eveland-Sayers et al., 2009). For the reading exam, the variables accounted for
more variance in the girls than the boys and were predictive for one but not the other.
Specifically, for boys (but not for girls), higher levels of social support had a direct and positive
influence on how they performed on the exam, which is consistent with Ahmed et al. (2010) who
found parental and peer support were related directly to academic achievement. For girls, but
not boys, having a larger BMI was related to better performances on the reading exam, which is
both counterintuitive as well as inconsistent with the past research (e.g., Eveland-Sayers et al.,
Predictors of Academic Performance 18
2009; Grissom, 2005; Kristjansson et al., 2010). However, like past studies (e.g., Cottrell et al.,
2007) being overweight was not as important for understanding boys’ and girls’ performances on
state-mandated achievement tests as was their level of physical fitness.
Limitations and Directions for Future Research
There are a few limitations to the present study. First, most of the psychosocial and
physical variables used were self-report, though we did incorporate objective measures of
cardiorespiratory fitness and body composition. Future studies might consider a more objective
measure of physical activity, such as having students wear accelerometers for a one to two week
period of time, to complement the manner in which physical fitness was assessed. It will be
important for future studies to further examine the relationship between physical activity and
fitness and their relative contributions (when both are measured objectively) to academic
performance. Second, the sample was obtained from only one school district in the southern
United States, thus limiting its generalizability to similar suburban areas. The sample, however,
was diverse in terms of racial/ethnic and socio-economic status, and accurately represented the
overall demographics of the students in the district. Future studies might examine the influence
of such physical and psychosocial variables in urban or rural districts and among high school
students. Third, not all relevant physical and psychosocial factors were included in the study. We
did use variables that had been shown to relate to academic achievement, but time constraints
placed on us by the district in terms of access to the students prevented us from including other
potential predictors, such as anxiety, depression, nutritional status, to name a few. Future
research may want to examine the relative predictive utility of these other factors, as well as
variables such as student delinquency to determine their long-term effect on academic
achievement. Finally, although our study used a longitudinal design, the timeframe between
Predictors of Academic Performance 19
when the physical and psychosocial variables were measured to when the students took the math
and reading examinations was limited to one to five months. Future research should examine the
long-term effects of fitness and other psychological variables, determining whether the effects
uncovered in this study last from year to year across middle and high school.
Clinical Implications
Given the study’s design, our results support several potential interventions that may be
implemented to boost middle school students’ achievement on state-based reading and math
tests. First, as expected, lower SES level had a negative effect on the children’s test scores. Low
SES students may experience difficulties, such as a lack of resources, little support or academic
help from parents/guardians, and familial discord/stress, that can interfere with needed
motivation, focus, and attention, and, overall, create an environment that is not conducive to or
supportive of learning. Therefore, it becomes necessary to identify these children early in their
school careers and provide as many extra services as possible, such as after school tutoring,
individual and/or family counseling, after school care, summer meal programs, etc. By
ameliorating the potential negative effects associated with low SES environments, children may
see significant improvements in their academic performance. Second, students’ confidence in
their academic abilities in each subject area were strong predictors of performance, so working to
improve their self-concept and self-efficacy would be beneficial. Helping children achieve early
academic successes upon which they can build may lead to improvements in self-concept and a
willingness to engage in more complex and challenging learning in the future. The more children
believe in their own academic abilities, and understand that their efforts lead to success, the more
likely they are put in the time and focus needed to be successful academically. Helping children
see the connection between their efforts and their achievement is a necessary first step (Dweck,
Predictors of Academic Performance 20
2006). Finally, given that cardiorespiratory fitness predicted achievement across both tests, even
after controlling for SES and academic self-concept, it makes sense to dedicate time in which
children can be physically active, such as during PE classes, to have opportunities to become
more fit. Although districts often view PE classes as secondary to their academic mission (e.g.,
Grissom, 2005), our data suggest that students’ are likely to receive tangible academic benefits
from their involvement in PE classes and by improving their cardiorespiratory fitness. If PE
classes are not possible, schools might consider supporting before school walking programs for
kids who arrive early or after school programs that allow students’ access to the gyms and other
areas to be physically active. Our results indicate that students’ academic achievement is
influenced, at least in part, by how physically fit they are.
We examined the relative predictive utility of physical and psychosocial variables in
predicting middle school students’ academic performance on state-mandated math and reading
examinations. After controlling for the students’ families’ SES level as well as the students own
academic self-concept, we found that level of cardiorespiratory fitness was the one consistent
predictor of how they performed. This finding has policy implications when it comes to whether,
and for how long, students will take PE classes, and what communities are doing to provide
children and adolescents with easy and safe access to environments, such as parks, where they
can be physically active. It also highlights the importance of students having opportunities to
develop their cardiorespiratory fitness, whether in school-sponsored PE classes or through
community or private sport and activity opportunities.
Predictors of Academic Performance 21
References
Aberg, M. A., Pedersen, N.L., Toren, K., Svartengren, M., Backstra, B., Johnsson, T., CooperKuhn, C.M., Åberg, N.D., Nilsson, M., & Kuhn, H.G. (2009). Cardiovascular fitness is
associated with cognition in young adulthood. PNAS, 106, 20906-20911. doi:
10.1073/pnas.0905307106
Ahmed, W., Minnaert, A., van der Werf, G., & Kuyper. H. (2010). Perceived social support and
early adolescents’ achievement: The mediational roles of motivational beliefs and
emotions. Journal of Youth Adolescence, 39, 36-46. doi: 10.1007/s10964-00809367-7
Castelli, D. M., Hillman, C.H., Buck, S.M., & Erwin, H. E. (2007). Physical fitness and
academic achievement in third- and fifth-grade students. Journal of Sport & Exercise
Psychology, 29, 239-252.
Chaddock, L., Pontifex, M., Hillman, C., & Kramer, A. (2011). A review of the relation of
aerobic fitness and physical activity to brain structure and function in children. Journal of
the International Neuropsychological Society, 17, 1-11. doi:
10.1017/S1355617711000567
Chomitz, V. R., Slining, M. M., McGowan, R. J., Mitchell, S. E., Dawson, G. F., & Hacker,
K.A., (2009). Is there a relationship between physical fitness and academic achievement?
Positive results from public school children in the northeastern United States. Journal of
School Health, 79, 30-37.
Cottrell, L. A., Northrup, K., & Wittberg, R. (2007). The extended relationship between child
cardiovascular risks and academic performance measures. Obesity, 15, 3170-3177.
Dweck, C. S., (2006). Mindset: The New Psychology of Success. Ballantine: New York.
Predictors of Academic Performance 22
Dwyer, T., Sallis, J. F., Blizzard, L., Lazarus, R., & Dean, K. (2001). Relation of academic
performance to physical activity and fitness in children. Pediatric Exercise Science, 13,
225-237.
Edwards, J. U., Mauch, L., & Winkelman, M. R. (2011). Relationship of nutrition and physical
activity behaviors and fitness measures to academic performance for sixth graders in a
midwest city school district. Journal of School Health, 81, 65-73.
Eveland-Sayers, B. M., Farley, R. S., Fuller, D. K., Morgan, D. W., & Caputo, J. L. (2009).
Physical fitness and academic achievement in elementary school children. Journal of
Physical Activity and Health, 6, 99-104.
Fox, C. K., Barr-Anderson, D., Neumark-Sztainer, D., & Wall, M. (2010). Physical activity and
sports team participation: Associations with academic outcomes in middle school and
high school students. Journal of School Health, 80, 31-37.
Green, J., Nelson, G., Martin, A. J., & Marsh, H. (2006). The causal ordering of self-concept and
academic motivation and its effect on academic achievement. International Education
Journal, 7, 534-546.
Grissom, JB (2005). Physical fitness and academic achievement. Journal of Exercise Physiology,
8, 11-25.
Hillman, C. H., Pontifex, M. B., Raine, L.B., Castelli, D.M., Hall, E.E., & Kramer, A. F. (2009).
The effect of treadmill walking on cognitive control and academic achievement in
preadolescent children. Neuroscience, 159, 1044-1054. doi:
10.1016/j.neuroscience.2008.01.057
Kristjansson, A. L., Sigfusdottir, I. D., & Allegrante, J. P. (2008). Health behavior and academic
achievement among adolescents: The relative contribution of dietary habits, physical
Predictors of Academic Performance 23
activity, body mass index, and self-esteem. Health Education Behavior, 37, 51-64. doi:
10.1177/1090198107313481
Lundy, S. M., Silva, G.E., Kaemingk, K.L., Goodwin, J. L, & Quan, S.F. (2010). Cognitive
functioning and academic performance in elementary school children with
anxious/depressed and withdrawn symptoms. Open Pediatrician Medical Journal, 4, 19. doi: 10.2174/1874309901004010001
Marsh, H. W. (1992). Self-Description Questionnaire II: A theoretical and empirical basis for
the measurement of multiple dimensions of adolescent self-concept: A test manual and
research monograph. Macarthur, New South Wales, Australia: University of Western
Sydney, Faculty of Education.
Marsh, H. W., Parada, R. H. & Ayotte, V. (2004). A multidimensional perspective of relations
between self-concept (Self Description Questionnaire II) and adolescent mental health.
Psychological Assessment, 16, 27-41.
Osborne, J. W., & Jones, B. D. (2011). Identification with academics and motivation to achieve
in school: How the structure of the Self influences academic outcomes. Educational
Psychology Review, 23, 131-158. doi: 10.1007/s10648-011-9151-1
Petrie, T. A., Greenleaf, C., & Martin, S. (2010). Biopsychosocial and physical correlates of
middle school boys’ and girls’ body satisfaction. Sex Roles, 63, 631-644. doi:
10.1007/s11199-010-9872-5
Pintrich, P. R., & De Groot, E.V. (1990). Motivaitonal and self-regulated learning components of
classroom academic performance. Journal of Educational Psychology, 82, 33-40.
Predictors of Academic Performance 24
Pontifex, M., Raine, L., Johnson, C., Chaddock, L., Voss, M., Cohen, N., Kramer, A., &
Hillman, C. (2011). Cardiorespiratory fitness and the flexible modulation of cognitive
control in preadolescent children. Journal of Cognitive Neuroscience, 23, 1332-1345
Rapport, M. D., Denney, C.B., Chung, K., & Hustace, K. (2001). Internalizing behavior
problems and scholastic achievement in children: Cognitive and behavioral pathways as
mediators of outcome. Journal of Clinical Child Psychology, 30. 536-551. doi:
10.1207/S15374424JCCP3004_10
Roberts, C., Freed, B., & McCarthy, W. (2010). Low aerobic fitness and obesity are associated
with lower standardized test scores in children. The Journal of Pediatrics, 156, 711-718.
doi: 10.1016/j.peds.2009.11.039
Rosenfeld, L. B., Richman, J. M., & Bowen, G.L. (2000). Social support networks and school
outcomes: The centrality of the teacher. Child and Adolescent Social Work Journal, 17,
205-226.
Schlomer, G. L., Bauman, S., & Card, N.A. (2010). Best practices for missing data management
in counseling psychology. Journal of Counseling Psychology, 57, 1-10. doi:
10.1037/a0018082
Stice, E. (2002). Risk and maintenance factors for eating pathology: A meta-analytic review.
Psychological Bulletin, 128, 825-848. doi: 10.1037//0033-2909.128.5.825
Stringer, R. W., & Heath, N. (2008). Academic self-perception and its relationship to academic
performance. Canadian Journal of Education, 31, 327-345.
The Cooper Institute (2007). FITNESSGRAM/ACTIVITYGRAM: Test Administration Manual.
Champaign, IL: Human Kinetics.
Predictors of Academic Performance 25
Wittberg, R., Cottrell, L.A., Davis, C.L., & Northrup, K.L. (2010). Aerobic fitness thresholds
associated with fifth grade academic achievement. American Journal of Health
Education. 41, 284-291.
Zimet, G. D., Dahlem, N.W., Zimet, S.G., & Farley, G.K. (1988). The Multidimensional Scale of
Perceived Social Support. Journal of Personality Assessment, 52, 30-41.
Zimet, G. D., Powell, S.S., Farley, G.K., & Werkman. S. (1990). Psychometric characteristics of
the Multidimensional Scale of Perceived Social Support. Journal of Personality
Assessment, 55, 610-617
Predictors of Academic Performance 26
Table 1
Means, Standard Deviations, and Pearson Product-moment Correlations for the Criterion and Predictor Variables (Girls above the diagonal, n = 650; Boys
below the diagonal, n = 561)
Variable
1. Reading
1
-
2
.65**
-
3
4
5
6
7
8
9
10
11
12
13
.38**
.21**
.29**
.29**
.25**
.00
.17*
.14*
.07
.18*
.18*
.32**
.45**
.25**
.36**
.26**
-.08*
.16*
.15*
.06
.24**
.14*
2. Math
.63**
3. SES
.31**
.27**
-
.19*
.13*
.26**
.22**
-.15*
.16*
.21**
.12*
.24**
.18*
4. SDQ-Math
.20**
.41**
.10*
-
.22**
.58**
.14*
-.04
.19*
.22**
.16*
.39**
.21**
5. SDQreading
.16*
.12*
.06
.27**
.01
.08
.04
-.02
.14*
.16*
6. SDQAcademic
.27**
.30**
.16*
.57**
-.02
.12*
.19*
.16*
.53**
.31**
7. PACER
.23**
.27**
.15*
.06
-.10*
.05
-
-.34**
.25**
.34**
.28**
.24**
.14*
-.02
-.27**
-
-.03
-.11*
.02
-
.53**
.53**
.22**
.16*
.56**
.23**
.15*
.39**
.42**
-
-.07
.11*
8. BMI
-.05
-.12*
-.17*
.00
-.01
9. Aerobic
.07
.08
.11*
.16*
.02
.13*
.26**
-.02
10. Strength
.14*
.12*
.12*
.17*
-.01
.17*
.36**
-.04
.50**
11. Flexibility
.12*
.11*
.04
.10*
.01
.04
.34**
.04
.55**
.48**
-
.17*
.08*
12. SDQ
.25**
.20**
.18*
.33**
.15*
.42**
.22**
-.08
.19*
.25**
.18*
-
.47**
13. MSPSS
.26**
.17*
.14*
.24**
.26**
.34**
.01
.03
.13*
.09*
.06
Mean Girls
(SD)
807.64
(94.43)
764.10
(89.99)
---
4.18
(1.58)
4.35
(1.73)
4.97
(1.28)
30.08
(14.25)
21.01
(4.71)
4.08
(2.20)
4.75
(1.90)
3.11
(2.02)
4.95
(0.90)
5.52
(1.33)
Mean Boys
(SD)
785.30
(96.26)
760.05
(84.63)
---
4.56
(1.53)
3.91
(1.82)
4.93
(1.26)
37.50
(18.50)
21.58
(7.92)
3.71
(2.33)
5.04
(1.92)
3.60
(2.06)
5.05
(0.83)
4.97
(1.53)
-
-.14*
.35**
-.07
-
Note: Reading- TAKS Reading scores (range = 249 to 1002); Math-TAKS Math scores (range = 477to 1025), SES- Socio Economic Status based on students
Qualifying for federal assistance in school meal programs; SDQ-Math - Math Self-Concept (range = 1, low to 6, high); SDQ-Reading - Reading Self-Concept
(range = 1, low to 6, high); SDQ – Academic- General Academic Self-Concept (range = 1, low, to 6, high); PACER- Cardiorespiratory fitness test
Predictors of Academic Performance 27
(range = 0 to 100 laps); BMI - Body Mass Index (kg/m2); Aerobic - Days of aerobic activity (range = 0, low to 7, high); Strength - Days of muscular exercises
(range = 0 to 7); Flexibility - Days of flexibility exercises (range = 0 to 7); SDQ – Self-Description Questionnaire General Self-Esteem (range = 1, low to 6, high);
MSPSS - Perceived Social Support from Family and Friends (range = 1, low to 7, high).
* p < .01, ** p <.001
Predictors of Academic Performance 28
Table 2
Hierarchical Multiple Regression Analyses Predicting Reading Achievement for Boys (N = 561) and Girls (N = 650) at Step 5 of the Model
Step/Predictor
B
SE B
β
t
Boys
Socio-Economic Status (SES)
I like reading
I am good at most school subjects
PACER
Body Mass Index (BMI)
Days of Aerobic activity
Days of Strength activity
Days of Flexibility activity
SDQ-General Self-Esteem
MSPSS-Perceived Social Support
27.18
4.43
0.24
6.14**
3.73
2.22
0.07
1.68
9.45
3.50
0.12
2.70*
1.00
0.23
0.19
4.36**
0.48
0.49
0.04
1.00
-4.07
1.98
-0.10
-2.05
0.66
2.37
0.01
0.28
3.00
2.26
0.06
1.33
6.82
5.19
0.06
1.31
9.68
2.66
0.15
3.65*
Full Model R² = .22, Overall F (10, 550) = 15.41**
Girls
Socio-Economic Status (SES)
I like reading
I am good at most school subjects
PACER
Body Mass Index (BMI)
Days of Aerobic activity
Days of Strength activity
Days of Flexibility activity
SDQ-General Self-Esteem
MSPSS-Perceived Social Support
31.90
3.97
0.29
8.03**
8.19
3.26
0.11
2.51*
11.73
2.07
0.22
5.67**
1.57
0.26
0.24
6.02**
2.47
0.73
0.12
3.38*
3.60
1.84
0.08
1.96
-1.05
2.22
-0.02
-0.47
-3.27
2.05
-0.07
-1.60
-4.42
4.68
-0.04
-0.94
3.47
2.75
0.05
1.26
Full Model R² = .27, Overall F (10, 639) = 23.91**
Note: Step 5 represents the final step of the model when all predictors have been entered.
* p < .01, ** p < .001
Predictors of Academic Performance 29
Table 3
Hierarchical Multiple Regression Analyses Predicting Mathematics Achievement for Boys (N = 625) and Girls (N = 755) at Step 5 of the Model
Step/Predictor
B
SE B
β
t
Boys
Socio-Economic Status (SES)
I like math
I am good at most school subjects
The PACER
Body Mass Index (BMI)
Days of Aerobic activity
Days of Strength activity
Days of Flexibility activity
SDQ-General Self-Esteem
MSPSS-Perceived Social Support
24.84
4.09
0.23
6.08**
4.99
2.05
0.10
2.43**
9.51
3.24
0.13
2.93
1.09
0.22
0.21
5.06**
0.61
0.47
0.05
1.30
-2.71
1.83
-0.07
-1.48
-0.71
2.22
-0.01
-0.32
2.80
2.12
0.06
1.32
7.17
4.91
0.06
1.46
8.41
2.47
0.14
3.40
Full Model R² = .292, Overall F (10, 614) = 25.35**
Girls
Socio-Economic Status (SES)
I like math
I am good at most school subjects
The PACER
Body Mass Index (BMI)
Days of Aerobic activity
Days of Strength activity
Days of Flexibility activity
SDQ-General Self-Esteem
MSPSS-Perceived Social Support
18.85
3.35
0.19
5.62**
7.59
2.30
0.11
2.53**
19.82
2.21
0.35
8.99
1.13
0.22
0.18
5.13**
0.48
0.64
0.03
0.74
2.19
1.60
0.05
1.37
-0.49
1.93
-0.01
-0.26
-3.80
1.76
-0.09
-2.15
-0.66
4.15
-0.01
-0.16
-0.75
2.34
-0.01
-0.32
Full Model R² = .295, Overall F (10, 639) = 27.22**
Note: Step 5 represents the final step of the model when all predictors have been entered.
* p < .01, ** p < .0001