494 GALLOWAY, CONNER, AND POPE

opponents of homework, that parents viewed homework positively, rather than as a detractor
from their child’s participation in other activities. As such, questions about the relation between
homework and family and leisure time require further empirical study.

Homework in privileged communities: Advantage or cost?

Last, studies on homework often lack exploration or discussion of social class, yet Kralovec
and Buell (2000) argue that homework does not affect all students equally. Rather, they state,
the practice of homework “appears to further disadvantage the already disadvantaged” (Kralovec
& Buell, 2000, p. 70) and acts as a sorting mechanism to magnify inherent class differences.
Students in middle class and wealthy communities have adequate resources, materials and physical
space, are more likely to receive outside help to complete their work (e.g., a private tutor), and
are less often forced to choose between completing homework and essential family needs or
Downloaded by [UQ Library] at 18:37 03 November 2014

responsibilities (Cooper, 2007; Kralovec & Buell, 2000). These advantaged students are, as
Demerath (2009) described, developing the work habits and the kind of psychological capital to
help them succeed in a credential-focused system. As Bempechat (2004) observed, parents who
advocate against homework actually place their students’ academic preparation in jeopardy due
to the academic demands they will eventually face. Nonetheless, it is not clear to what extent
homework, even in advantaged communities, supports student learning, school engagement, and
well-being. The present study fills this gap.

Present Study and Research Questions

We examine nonacademic effects of homework and seek to understand the role that homework
plays as a stressor in students’ lives in advantaged communities. The following research questions
guided our work: How much homework are students in a set of advantaged, high-performing
high schools doing each night, and how do they view this homework load? How is homework
load associated with (a) students’ stress; (b) students’ physical health (including symptoms like physical symtoms
headaches, stomach problems, problems sleeping); (c) students’ time for outside activities; and
(d) students’ behavioral engagement in school?

METHOD

Although we could have taken a number of different approaches to gathering data on homework,
we decided to go straight to the source and ask the students themselves. In general, student
voice has been lacking from the conversations about educational issues that deeply affect them,
yet recent scholarship has called for attending to students’ perspectives and involving students
in opportunities to effect change (see Cook-Sather, 2006; Thiessen & Cook-Sather, 2007). We
support the premise put forth by Schultz and Cook-Sather (2001):
It is crucial to listen to what students have to say because until we truly understand what students
are experiencing—what and how education means, looks, and feels to them—our efforts at school
reform will not go very far.” (p. 2)

Our methods were designed to gather student voice.

 NONACADEMIC EFFECTS OF HOMEWORK 495

We invited students to respond to both Likert-type and open-ended questions on a survey.

Students’ responses to the Likert-type items allowed us to examine the relation between homework
and both well-being and engagement. The open-ended items provided us with richer student voice,
and enabled us to explore how students view their homework, and if, when, and how homework
impacts their daily lives.

Study Context

This research was part of a larger university-based intervention and research project designed to
understand how to create school and home contexts that support adolescent academic integrity,
engagement, and mental and physical health in communities where achievement pressure is
particularly high (communities self-identify and volunteer to participate). Schools that participate
in the intervention are invited to take part in a survey to gather baseline data on students’
Downloaded by [UQ Library] at 18:37 03 November 2014

perceptions of their home and school contexts, as well as student self-reports on their achievement
goals, academic integrity, engagement, and health. We draw here from a subset of the data in the
survey, focused on the present study’s research questions.

Participants

A sample of 4,317 high school students from 10 high-performing high schools (four public and six
private) participated in this study. The mean age of participants was 15.7 (SD = 1.20), with 1,211
ninth-grade students (28.1% of the sample), 983 tenth-grade students (22.8% of the sample),
1018 eleventh-grade students (23.6% of the sample), and 836 twelfth-grade students (19.4% of
the sample); 269 students (6.2%) did not report their grade level. Of the sample, 54% was female.
Students self-reported their ethnicity: 48% European American; 31% Asian or Asian American;
4% Hispanic; 2% African American; and 0.5% Native American. A small but notable percentage
(10.5%) checked multiple categories or “other” for their race and/or ethnicity, and 4% elected
not to respond.
All of the schools in our study were college preparatory schools in advantaged, upper middle
class communities and had elected to participate in the study as part of the larger research and
intervention project. Median household income in these communities exceeds $90,000.00 per
year. The schools themselves were some of the most high-performing schools in the area. The
four public schools consistently demonstrated outstanding performance on standardized tests,
with Academic Performance Index scores above 800, in a 200 to 1000 range, and rankings in the
9th or 10th decile in comparison to similar schools (the 10th decile is the highest ranking) (Bay
Area News Group, 2011). All schools had high graduation rates, with the majority of the students
attending 4-year colleges and universities, and more than 93% attending a 2- or 4-year institution
(the overall rate in the region is 50%, with 29% representing community college goers; California
Postsecondary Education Commission, 2012).

Procedure

Students with parent consent completed a 40-min survey during the school day. We asked
schools to pay particular attention to schedules, to be sure not to administer the survey near final
examinations, or advanced placement testing, which are particularly stressful times of year for
 496 GALLOWAY, CONNER, AND POPE

students. Nine of the participating schools elected to conduct the survey in the spring term, between
late January and April (one school implemented the survey in early December). These varying
dates were a necessary accommodation to make to our school partners, who best understand the
limitations and affordances their particular schedules allowed. Four of the participating schools
elected to have students complete a paper version of the survey, while the remaining schools
administered an online version. Questions asked on the paper and online surveys were identical.

Measures

The student survey included Likert-type and open-ended questions. Likert-type questions ana-
lyzed for the present study assessed students’ self-reported homework load and perceptions of
homework load, well-being (measured here through questions on stress over schoolwork, per-
formance anxiety, physical health, sleep behavior, and time for outside activities), behavioral
Downloaded by [UQ Library] at 18:37 03 November 2014

engagement, and enjoyment of schoolwork. These items included a mix of scales drawn from
previously validated work and individual items relevant to our research questions. The open-
ended items invited students to write about experiences with stress in their lives more generally
and at school specifically. Although the open-ended items were not designed around the topic of
homework, we draw upon them here to understand students’ experiences with homework, and
specifically how it can act as a stressor in their lives.
The open-ended questions were posed in the middle of the survey, after students had responded
to items about their school context, personal goals, academic integrity, time for activities, per-
formance anxiety, and behavioral engagement. Questions related to homework load, usefulness
of homework, and health (among others not included in this study) followed the open-ended
questions.

Homework load

Homework load was assessed by asking students to report how many hours of homework they
do on a typical week day. Students were asked to exclude time when they were taking breaks,
instant messaging, or using the computer for recreational use.

Homework usefulness

Homework usefulness was assessed by two items: “In general, how useful is your homework
for helping you learn the material?” and “In general, how well does your homework prepare you
for tests, papers, or projects?” Students rated these items from 1 (not at all useful/well) to 5 (very
useful/well). These items were used to create a homework usefulness construct, α = .82.

Stress

We gauged student academic-related stress using a four-item measure of performance anxiety,

which included questions such as “How often do you worry about taking tests?”, and “How often
do you worry about school assignments?” rated on a 5-point scale ranging from 1 (never) to 5
(always). This scale was drawn from Pintrich, Smith, Garcia, and McKeachie (1991). Reliability
for this measure was α = .79. We later asked students to identify what in their lives right
 NONACADEMIC EFFECTS OF HOMEWORK 497

now causes them the most stress and why (an open-ended question). After asking this question,
we measured stress over schoolwork: “How often do you feel stressed by your schoolwork or
academic experience?” rated on a 5-point scale ranging from 1 (never) to 5 (always). We then
asked students to respond to an open-ended question: “If you have felt stressed, what is the most
stressful part of your schoolwork or academic experience?”
To analyze the open-ended data, we reviewed student responses for common themes across
the schools (first looking for what students indicated caused stress and then examining themes
for cases where homework was indicated as a primary stressor). We did not begin the process
with predetermined categories; rather we allowed categories to emerge following the process for
open coding. From here, we moved to axial coding to make connections among our categories
(Strauss & Corbin, 1998). Clear themes began to emerge after examining responses from our first
two schools. We used these themes as a guide for coding the rest of the open-ended responses,
allowing for the addition of new themes along the way. We then formed propositions regarding
Downloaded by [UQ Library] at 18:37 03 November 2014

the data and earmarked representative responses, as well as any negative cases, to use to report
our findings in this paper (Taylor & Bogdan, 1998).

Physical health

We assessed physical health by whether students reported experiencing any of seven stress- related to page
related physical symptoms in the month before taking the survey (sweating, headache, exhaustion, 494
weight loss, weight gain, stomach problems, and/or sleeping difficulties). Each was a dichotomous
variable where students answered that they either had or had not experienced the reaction because
of stress. We also asked students to report how many hours they slept on school nights.

Time for other activities/endeavors

We were interested in how homework was associated with students’ perceptions of their ability
to have time for sleep, relationships, and activities. Three questions addressed perceptions of time:
“How often does the amount of schoolwork you have keep you from getting enough sleep?” rated
on a 5-point scale ranging from 1 (never) to 5 (always); “How often does schoolwork or studying
keep you from having time for your family, friends, or other activities?” also rated on a 5-point
scale ranging from 1 (never) to 5 (always); and “Have you ever been forced to drop an activity or
hobby which you enjoyed because schoolwork took too much of your time?” rated on a yes/no
scale. This final question is from West and Wood’s (1970) Academic Pressure Scale.

Behavioral engagement

To assess students’ behavioral engagement, we used a previously established measure (Marks,

2000), which consisted of three items:
1. How often do you try as hard as you can in school?
2. How often do you pay attention in your classes?
3. How often do you complete your school assignments?
 498 GALLOWAY, CONNER, AND POPE

Each of these items was rated from 1 (never) to 5 (always). Reliability (α) for this measure was
.68. As done in the study by Marks (2000), we also asked students how often they enjoyed their
schoolwork, from 1 (never) to 5 (always).

Demographic and achievement information

The end of the survey asked students to provide demographic information and self-reported
grade point average. For demographics, students reported on gender, race and/or ethnicity, grade
in school, and age.

RESULTS
Downloaded by [UQ Library] at 18:37 03 November 2014

Hours of Homework, Stress, and Physical Health

On average, students reported spending 3.11 hr (SD = 1.46) per night on homework. In general,
they found their homework as “somewhat useful” in helping them to learn the material (M =
3.08, SD = 0.98) and “somewhat useful” in helping them to prepare for tests, papers, or projects
(M = 3.13, SD = 0.98). Frequencies indicated that only 6% of students found their homework
“very useful” for either learning or preparation.
Although our sample averaged more than 3 hr of homework per night, homework hours differed
by grade, F(3, 3912) = 20.10, p < .001, ηp 2 = .02, and gender, F(1, 3912) = 166.0, p < .001, ηp 2 =
.04. In addition, 9th- and 12th-grade students reported the fewest hours per night, less than did the
10th-grade students at p < .05, whereas 11th-grade students reported significantly more than did
any other grade. Female students indicated doing more hours per night than did male students.
This finding was also statistically significant (see Table 1 for means and standard deviations). A
one-way analysis of variance indicated that the schools in our sample also differed in average
amount of homework, F(9, 4151) = 17.90, ηp 2 = .04. Four (one private, three public) schools had
means below 3.00 hr per night, with the lowest school mean at 2.38 (SD = 1.07). The remaining
schools (five private schools and one public school) had means above 3.00, with highest mean at
3.59 (SD = 1.25).
As expected, many students in our study reported experiencing stress, compromised health, or
lack of balance. Most experienced distress and/or lacked time to engage in important life tasks
outside of school. The majority (72%) reported being often or always stressed over schoolwork
(M = 3.92, SD = 0.90), and many reported that they experienced physical symptoms due to

TABLE 1
Nightly Homework Hours, by Grade and Gender

Boys Girls

Grade M SD M SD
9th 2.75 1.26 3.11 1.25
10th 2.91 1.42 3.43 1.37
11th 3.03 1.48 3.66 1.44
12th 2.49 1.47 3.35 1.67
 NONACADEMIC EFFECTS OF HOMEWORK 499

stress (82% reported experiencing at least one physical symptom in the past month, with 44% of
the sample experiencing three or more symptoms). Overall, students reported getting less sleep
than the National Sleep Foundation’s (2000) recommended 8.5 to 9.25 hr per night for healthy
adolescent development. On average, students in our sample reported 6.80 hr of sleep on school
nights (SD = 1.24), and 68% stated that schoolwork often or always kept them from getting enough
sleep each night. Many (63%) reported that the amount of work they received often or always
made it challenging to spend time with family and friends, and a similar percent (61%) indicated
that they had been forced to drop an activity they enjoyed because of their school workload.

Relation Between Hours of Homework and Student Well-being and Engagement

Bivariate correlations indicated that more hours of homework was associated with increased
school stress and physical distress, and decreased ability to cultivate skills outside of school
Downloaded by [UQ Library] at 18:37 03 November 2014

(see Table 2 for correlations among all variables). Students who spent more hours on homework
tended to be more behaviorally engaged in school, but were simultaneously more stressed about
their school work and tended to report more physical symptoms due to stress, fewer hours of sleep
on school nights, less ability to get enough sleep, and less ability to make time for friends and
family. These correlations also indicated that homework hours had a stronger linear relation to
student well-being variables (with rs between |.22| and |.41|) than to student grade point average
(r = .14).
To look more closely at the relation between homework load and student outcome variables, we
conducted a series of hierarchical linear regressions, controlling for demographic variables and
school effects to understand how much variance homework predicted in our outcome variables.
Though effects were small, time spent on homework had the single biggest effect for our stress
and time variables (see Table 3): for performance anxiety, homework accounted for 11% of the
variance; for stress over schoolwork, 12%; for time spent with family and friends, 14%; and for
time spent sleeping, 13%. We also found positive significant effects for homework load on hours
of sleep and physical symptoms due to stress, though effect sizes were less than .10.

TABLE 2
Bivariate Correlations Between Hours of Homework, Well-Being, and Engagement

1 2 3 4 5 6 7 8 9 10 11

1. Homework hours — 0.36∗∗ .34∗∗ 0.22∗∗ −0.31∗∗ 0.41∗∗ 0.39∗∗ .27∗∗ .05∗∗ .01 .14∗∗
2. Stress over schoolwork — .63∗∗ 0.41∗∗ −0.27∗∗ 0.51∗∗ 0.48 ∗∗ .12∗∗ −.16∗∗ .01 −.05∗
3. Performance anxiety — .39∗∗ −.23∗∗ .48∗∗ .44∗∗ .19∗∗ −.10∗∗ −.02 −.11∗∗
4. Physical problems — −0.31∗∗ 0.34∗∗ 0.25∗∗ −.05∗∗ −.12∗∗ .08∗∗ −.13∗∗
5. Hours of sleep — −0.44∗∗ −0.21∗∗ .14∗∗ .13∗∗ −.19∗∗ .04∗
6. Lack of time for sleep — 0.56 ∗∗∗ .09∗∗ −.09∗∗ .05∗∗ .04∗
7. Lack of friend/family time — .18∗∗ −.08∗∗ −.04∗ .09∗∗
8. Behavioral engagement — .38∗∗ −.17∗∗ .36∗∗
9. Enjoyment — −.01 .22∗
10. Age — −.02
11. Grade point average —

∗p < .05. ∗∗ p < .01. ∗∗∗ p < .001.

 500 GALLOWAY, CONNER, AND POPE

TABLE 3
Hierarchical Multiple Regression Analyses Examining Relationship Between Homework Load and Well-Being

Performance School Physical Time from Time from

anxiety stress symptoms family and friends sleep Sleep

Predictor β β β β β β
Step 1: Demographic variables
Age −.04∗ .01 .08∗∗∗ −.05∗∗ .05∗∗ −.24∗∗∗
White students −.04 .05 .06∗∗∗ .06∗∗ −.04 .13∗∗∗
Non-Asian students of color −.04 .02 .07∗∗ .04∗∗ −.02 .04
Gender .16∗∗∗ .12∗∗∗ .25∗∗∗ .02 .08∗∗∗ −.10∗∗∗
Grade point average −.17∗∗∗ −.11∗∗∗ −.16∗∗∗ .03 −.02 .09∗∗∗
R2 .06 .04 .10 .02 .03 .10
Step 2: Schools
Downloaded by [UQ Library] at 18:37 03 November 2014

Public school 1 −.07∗∗∗ −.06∗∗ −.08∗∗∗ −.02 −.07 .01

Public school 2 −.01 −.02 .00 .00 .00 .03
Public school 3 −.01 −.01 −.06∗∗ .00 .01 −.06∗∗
Private school 1 .02 −.02 .00 .04∗ .03 −.08∗∗∗
Private school 2 .02 −.04 .00 .01 .00 −.02
Private school 3 .02 −.06∗∗ −.05∗ .02 .07∗∗∗ −.05∗
Private school 4 −.05∗∗ −.03 .00 −.01 −.03 .03
Private school 5 −.01 .01 −.03 .08∗∗ .09∗∗∗ .01
Private school 6 .00 −.01 −.03 .02 .05∗ −.07∗∗∗
R2 .01 .01 .01 .02 .03 .02
Step 3: Homework hours
Homework load .362 .21∗∗∗ .40∗∗∗ .37∗∗∗ −.25
R2 .11 .12 .04 .14 .13 .06
Total R2 .18 .17 .15 .18 .19 .18

Note. The Asian sample and the largest school (a public school) were the comparison groups for race and school,
respectively, in all regressions.
ˆp < .10. ∗ p < .05. ∗∗ p < .01. ∗∗∗ p < .001.

For engagement, we found a somewhat different pattern (see Table 4). The demographic
variables (particularly grade point average) explained more variance in the behavioral engagement
model than any other variable, though again, homework hours also showed a small positive
relation. These results suggest (in concert with the literature) that those with higher grade point
averages also tend to be more likely to try hard, complete their work on time, and pay attention in
class. Similarly, those with higher grade point averages also tended to report greater enjoyment
of school. Not surprising, although disheartening, was that homework hours predicted none of
the variance in enjoyment; in other words, those who did more hours of homework did not enjoy
their work any more than those who did less.
We also examined students’ likelihood of dropping out of an activity due to homework load.
Using a point-biserial correlation, we found those doing more homework per night tended to be
more likely to drop out of activities due to workload (r = .25, p < .001). Nonetheless, we did not
find a statistically significant relation between the amount of hours students reported spending on
extracurricular activities and the hours they spent on homework (r = .02).
 NONACADEMIC EFFECTS OF HOMEWORK 501

TABLE 4
Hierarchical Multiple Regression Analyses Examining Relation Between
Homework Load and Engagement

Behavioral engagement Enjoyment

Predictor β β
Step 1: Demographic variables
Age −.19∗∗∗ −.04∗
White students .02 .01
Non-Asian students of color .00 .03
Gender .14∗∗∗ .10∗∗∗
Grade point average .31∗∗∗ .21∗∗∗
R2 .20 .07
Step 2: Schools
Downloaded by [UQ Library] at 18:37 03 November 2014

Public school 1 .09∗∗∗ .07∗∗

Public school 2 .02 −.01
Public school 3 −.03 .00
Private school 1 −.03 .01
Private school 2 .06∗∗∗ .00
Private school 3 .07∗∗∗ .09∗∗∗
Private school 4 .04∗ .08∗∗∗
Private school 5 .08∗∗∗ .08∗∗∗
Private school 6 −.13 −.01
R2 .01 .04
Step 3: Homework hours
Homework load .22∗∗∗ −.02
R2 .05 .00
Total R2 .26 .11

Note. The Asian sample and the largest school (a public school) were the
comparison groups for race and school, respectively in all regressions.
ˆp < .10. ∗ p < .05. ∗∗ p < .01. ∗∗∗ p < .001.

Student Views of Homework as a Stressor

The aforementioned quantitative results presented indicate that a significant number of high
school students in our sample are doing several hours of homework per night, and the more
homework students report doing, the more they report school stress, physical distress (including
lack of sleep), inability to find time for friends and family, and likelihood of dropping one or more
activities because of workload. When we explored students’ responses to the two open-ended
questions—“Right now in your life, what would you say causes you the most stress and why?” and
“If you have felt stressed by your schoolwork, what is the most stressful part of your schoolwork
or academic experience?”—we wondered how they would talk about homework as a stressor.
stress -> physical
symtoms The majority (56%) of students indicated homework as a primary stressor. Homework was listed
mentioned above most often as a primary stressor, although 43% of students listed tests as a primary stressor,
in page 494 and and 33% listed grades and/or getting good grades as a primary stressor. Parent expectations and
497
the college process were also reported by more than 15% of the sample. A small percentage of
students (1%, a total of 59 students) indicated that they were not stressed, either in general and/or