You are on page 1of 17

Journal of Childrens Health: Vol. 2, No. 1, pp.

87102 (2004)
Copyright C 2004 ASP
ISSN: 0888-0018 print / 1521-0669 online
DOI: 10.1080/15417060490447387

School-Year Employment Among South Texas Middle


School Students: Effects on Academic, Social, Mental
Health, and Physical Functioning
Nancy F. Weller, DrPH, BSN,1 Sharon P. Cooper, PhD,2 Susan R. Tortolero, PhD,3
Steven H. Kelder, PhD, MPH,3 and Sohela Hassan, DrPH4
1
Department of Family and Community Medicine, Baylor College of Medicine, Houston,
Texas; 2 School of Rural Public Health, Texas A & M University, Bryan, Texas; 3 Center for
Health Promotion and Prevention Research, School of Public Health, University of Texas
Health Science Center at Houston, Houston, Texas; 4 Mental Sciences Institute, University of
Texas Health Science Center at Houston, Houston, Texas

Corresponding author: Nancy F. Weller, DrPH, BSN, Department of Family and Community Medicine, Baylor College of Medicine, 3701 Kirby, Suite 600, Houston, Texas 77098.
Telephone: 713-798-3633. Fax: 713-798-3644. E-mail: nweller@bcm.tmc.edu

ABSTRACT
This study describes the effect of different weekly work intensity levels on child
functioning in a sample of middle school students (N = 5,367) in rural South Texas
where economically disadvantaged and Hispanic students are heavily represented.
Anonymous surveys were conducted in sixth through eighth grade students regular classrooms. The following effects were associated with longer hours of weekly
school-year employment: 1) decreased engagement in school, hours of weeknight
and weekend sleep, and satisfaction with amounts of leisure time, and 2) increased
psychological stress, frequency of headaches, and substance use. It was concluded
that parents and professionals should continue to closely monitor the number of
weekly hours that students work during the school year.
Key Words: employment, workload, child, minority groups, schools, alcohol drinking, mental health, tobacco, marijuana smoking, cocaine, sleep.

INTRODUCTION
State, national, and international child labor laws prohibit the employment of
children 14 years of age and younger (Texas Employment Commission 1992; U.S.
Department of Labor 1990; United Nations Childrens Fund 1997). Despite these

87

Weller et al.

proscriptions against work in young children, emerging evidence suggests that some
unknown proportion of young people work during their middle school years, possibly during the school year and in formal sector positions (Institute of Medicine
1998; Steinberg and Cauffman 1995).
Only a few scientific studies have documented that paid work occurs among children and early adolescents (Dunn 1998; Steinberg and Cauffman 1995; Miller 1995;
Rothstein 2001; Huang 2001; Miller and Kaufman 1998). Weekly work intensity, a
critical variable in the study of youth employment, was not reported in these findings,
and research among older youth indicates that, while positive benefits to working
have been observed, school-year work may have negative effects on adolescent development, especially as hours of weekly work increase (Steinberg and Cauffman 1995;
Ruggiero 1984; Steinberg and Dornbusch 1991; Steinberg et al. 1993; Weller 2003a;
2003b; 2003d).
For example, several studies among employed high school students have reported
negative effects on school performance when older teens work past 20 hours a week,
past 15 hours for younger adolescent students (Schoenhals 1997; Steinberg and
Cauffman 1995; Chaplin and Hannaway 1996; Institute of Medicine 1998; Weller
2003d). Other studies found similar thresholds for health risk behaviors, future
educational attainment, and parent-adolescent relations (Carr 1996; Greenberger
and Steinberg 1986; Mortimer and Shanahan 1994; Steinberg and Dornbusch 1991;
Mihalic and Elliot 1997; Manning 1990; Wright 1997; Wofford 1988; Bachman and
Schulenberg 1993; Resnick 1997; Mortimer 1996; Weller 2003d). Still other studies
reported that working fewer than 10 hours weekly posed little risk to most high school
students and might be beneficial (Steinberg and Cauffman 1995; Bachman and
Schulenberg 1993). Although these findings among older adolescents are somewhat
useful in examining employment among middle schoolers, the results of such studies
cannot be directly generalized to younger workers, since substantial differences may
exist between the two populations in job tasks, frequency and intensity of work,
developmental abilities, and work-related experience.
Finally, no information, to the authors knowledge, is available on the effects
of different weekly work intensity levels on the development of mostly minority and
low-income populations. These youngsters may work more hours weekly due to their
socioeconomic circumstances, possibly sustaining more adverse employment effects.
This article reports on a cross-sectional study of the health and social effects of
school-year employment among middle school students, a large proportion of whom
are poor, Hispanic youth who live in rural south Texas.
We examined the effect of different weekly work intensity levels on four categories of outcome variables: 1) school performance and engagement, 2) health risk
behaviors, 3) physical and mental health issues, and 4) social life. This study sought
to document the relation between these factors and weekly work intensity in order to assist in prioritizing prevention efforts to reduce negative outcomes in these
youngest and most vulnerable workers.

88

Journal of Childrens Health. Vol. 2, No. 1, 2004

Effects of School-Year Employment Among Middle School Students

METHODS
Sample
During May 1995, data in this descriptive, cross-sectional study were collected as
part of the Safe and Drug Free Schools (SDFS) Programs regular assessment of the
prevalence of substance use among Texas Education Agency Region Two students
who represented 27 middle schools in 11 contiguous counties in south Texas. This
data collection provided an opportunity to examine issues of employment in relation
to health and other behaviors.
The SDFS program coordinator recruited all 42 school districts in these counties.
Fifteen districts declined participation due primarily to time constraints. Participating and nonparticipating districts were located in small towns and rural areas except
for a single, small urban area in the nonparticipating group. The distribution of
the educational status of parents and student ethnic characteristics was quite similar
among participating and nonparticipating districts. About 60% of students in participating districts were eligible to receive free or reduced cost lunches; the median
percentage of Hispanic students in participating districts was 72% (Texas Education
Agency 1993).
The total middle school population in the participating schools was 11,523. At the
larger schools, classes were randomly selected by grade from a master list of second
period classes using a random number table. In schools with fewer than 200 students,
all students were surveyed. The number of students eligible to complete the survey
was 8,757. A total of 7,420 sixth through eighth graders, representing 85% of the
eligible, was surveyed. The number of students who responded was 7,302 (2,365
sixth graders, 2,487 seventh graders, and 2,450 eighth graders). Excluded from the
analysis were cases with missing data on one or more variables (2,035), leaving a
sample size of 5,367. Of these, 3,008 reported current or recent employment within
the last six months.
The University of Texas Health Science Center Institutional Review Board approved the study protocol and survey instrument (HSC-SPH-95-018). During the
week prior to survey administration, parents received a passive informed consent
form explaining the study and providing an opportunity to decline participation.
Following the Centers for Disease Control and Preventions (CDC) Youth Risk Behavior Study (YRBS) protocol, teachers received instruction packets and administered
the survey during students regular classes (Kolbe at al. 1993). The survey was anonymous; students recorded their responses directly on a computer-scannable answer
sheet.
Analysis and Measures
Survey questions were selected from instruments previously used in other major
studies of adolescent development. Table 1 provides a description of the outcome
variables and their characteristics.

Journal of Childrens Health. Vol. 2, No. 1, 2004

89

Weller et al.

Table 1. Description of outcome variables and characteristics of survey measures


for school performance/engagement, health risk behaviors, social life,
and physical and mental health
Outcome variable

Coding

School performance/engagement
Weekly homework hours
Actual hours ranging from 07 +
Grades usually made
Range from 1 = Fs to 7 = As
Weekly unexcused absences
Actual times ranging 1 = 0 to 8 = 12+
Weekly cheating episodes
Actual times range 1 = 0 to 8 = 12+
Frequency class sleeping
Range from 1 = never to 5 = everyday
Frequency school tardies
Range from 1 = never to 5 = everyday
Confidence re: graduating
Range from 1 = none to 5 = graduate
college
Risk behaviors
Days smoked past month
Range from 1 = 0 to 7 = all 30 days
Cigarettes/day past month
Range from 1 = 0 to 7 = 20+ per day
Days used snuff past month
Range from 1 = 0 to 7 = all 30 days
Days used alcohol past month Range from 1 = 0 to 7 = all 30 days
Days drunk past month
Range from 1 = 0 to 7 = 20+ per day
Frequency marijuana use in
Range from 1 = 0 to 7 = 100 times
life
Frequency cocaine use in life Range from 1 = 0 to 6 = 40+ times
Frequency of inhalant use in Range from 1 = 0 to 6 = 40+ times
life
Frequency steroid use in life
Range from 1 = 0 to 6 = 40+ times
Frequency of injecting drugs Range from 1 = 0 to 6 = 40+ times
in life
Sum frequency fruits and
Range from 1 = 0 to 4 = 3+ times
veggies yesterday (3 items)
Sum frequency fat yesterday
Range from 1 = 0 to 4 = 3+ times
(3 items)
Days aerobic exercise last week Range from 1 = 0 to 8 = 7 days
Physical/mental health concerns
Hours week night sleep
Actual hours 1 = 10+ to 8 = 3 or less
Hours weekend sleep
Actual hours 1 = 10+ to 8 = 3 or less
Frequency head/
Range from 1 = never to 5 = everyday
stomachaches
Frequency colds, minor
Range from 1 = never to 4 = often
illnesses
Frequency feeling stressed
Range from 1 = never to 5 = always
Social life
Satisfaction with amount
Range from 1 = not at all to 5 = very
leisure time
Weekday afternoons with
Actual days 1 = 0 to 6 = 5 days
friends
Independent variable
Weekly work hours
Actual hrs ranging from 1 = 0 to
8 = 31+

Reliability Source
NA**
NA
NA
=.68
NA
NA
NA

Temple
MTF
Temple
Temple
Temple
Temple
MTF

YRBS
YRBS
YRBS
YRBS
YRBS
YRBS

= 80%
= 76%
= 72%
= 68%
= 64%
= 88%

= 73% YRBS
NA
YRBS
= 42% YRBS
= 15% YRBS
NA

YRBS

NA

YRBS

NA

YRBS

NA
NA
= .67

Bradley
Bradley
CES-D

= .67

CES-D

= .88

CES-D

NA

MTF

NA

MTF

NA

Temple

*Temple University Psychology Department, University of Michigan Monitoring the Future (MTF) Survey,
Centers for Disease Control and Preventions Youth Risk Behavior Survey (YRBS), EP Bradley Hospital
Sleep Research Laboratory, Center for Epidemiologic Studies Depression Scale (CES-D).
**NA = not available.

90

Journal of Childrens Health. Vol. 2, No. 1, 2004

Effects of School-Year Employment Among Middle School Students

Independent Variable
Items pertaining to employment were obtained from the North Carolina Teens at
Work Questionnaire, the Massachusetts Teens at Work Questionnaire, and the Temple University Psychology Departments School-Year Work Questionnaire (Personal
Communications, K. Dunn 1994; L. Davis 1995; L. Steinberg 1995). The independent variable, weekly work hours spent at a paying job, was expressed as a four-level
variable with the following categories: not employed, employed 110 hours per week
(low intensity), employed 1120 hours per week (moderate intensity), and employed
21+ hours weekly (high intensity). Since the employment of very young workers occurs mostly in informal and unstable settings, current/recent weekly work hours
usually spent at a paying job(s) was considered as an indicator of how many weekly
hours a student usually worked or has workeda typical school-year work pattern.
Outcome Variables
Four categories of outcome variables were included on the survey: school performance and engagement, health risk behaviors, physical/mental health concerns,
and social life. Seven items measured school performance and engagement. Grades
usually made was used as a measure of school performance. In an analysis of several thousand students, Dornbusch reported that the correlation between self- and
school-reported grade point average was almost 0.80 (Dornbusch 1987). To measure
school engagement, students reported weekly homework hours and the frequency of
unexcused absences, tardies, class sleeping, and cheating (Steinberg and Dornbusch
1991; Bachman and Schulenberg 1993).
Health risk behavior items were taken from the Centers for Disease Control and
Preventions (CDC) 1995 Youth Risk Behavior Survey (YRBS) (Kolbe et al. 1993). The
test-retest reliability of the 1991 instrument was tested in 1992. Approximately 75%
of items were rated as having substantial or higher reliability (kappa = 61100%)
(Brener et al. 1995). Items pertaining to somatic complaints, social life, and sleep
were taken from the Center for Epidemiologic Studies Depression Scale (Radloff
1977), the Monitoring the Future survey (Personal Communication, Survey Research
Center 1995) and the Chronobiology/Sleep Research Laboratory of E.P.Bradley
Hospitals School Sleep Habits Survey (Personal Communication, M. Carskadon
1995), respectively.
Covariates
Students provided information pertaining to gender, grade, race/ethnicity,
and maternal/ paternal educational status, a measure of socioeconomic level.
Race/ethnicity was collapsed into three categories: non-Hispanic white, Hispanic,
and all other race/ethnic group combined (black, Asian/Pacific Islander, American Indian/Alaskan Native, mixed race). Maternal and paternal educational status
was indexed into three categories: one/both did not graduate high school, both
graduated high school, one/both graduated college or attended graduate school.

Journal of Childrens Health. Vol. 2, No. 1, 2004

91

Weller et al.

Methods of Analysis
The relationships between the outcome and independent variables were investigated using SPSS 8.0 (SPSS 1995).
Data Analysis
The analysis consisted of several steps: First, four multivariate analyses of covariance (MANCOVAS) were performed to examine the relation between weekly work
hours and the four categories of outcomes (academic factors, risk behaviors, physical/mental health concerns, and social life), holding constant the effect of gender, grade, race/ethnicity, and parent education. MANCOVA was chosen because
it provides an omnibus test of significance across all variables within a conceptual
grouping, and it allows an independent assessment of the effect of the independent
variable on each dependent variable within the conceptual grouping.
Second, for each significant univariate variable detected in the first step, we performed one-way analysis of covariance and post-hoc Scheffes. The latter test produces
adjusted means, compares the means of the work intensity levels to each other, and
reports all significant factor contrasts.
Third, to address our specific concerns regarding low- versus high-intensity work
and for comparison to other studies results, an effect size (the d-index) was calculated. The d statistic is a scale-free measure of the separation between two group
means and is calculated as follows: mean for 110 hours minus mean for 21+ hours
divided by their average standard deviation (Cohen 1988). For example, we examined the difference in weekly unexcused absences between low- and high-intensity
workers. A d-index of .21 favoring low-intensity workers indicates that the average
weekly absences of these workers was .21 standard deviations below the average score
of high-intensity workers.
RESULTS
Characteristics of the Overall Sample
The overall sample was comprised of almost equal proportions of males and
females. Sixth graders made up 29% of the sample; seventh and eighth graders
comprised 35% and 36%, respectively. Fifty-five percent of the sample reported
Hispanic ethnicity, while 32% stated that they were white, and 13% claimed other
races and ethnicities. Sixty percent had parents whose educational backgrounds did
not extend past high school.
Table 2 presents the prevalence of work among middle schoolers by weekly work
intensity level. Fifty-six percent of the overall sample reported current or recent
employment within the last six months. Among workers, about 80% of middle school
students were employed for 110 hours weekly; 12% reported working 11 or more
hours weekly. A greater proportion of workers were male (54%). Thirty-eight percent
of white, 50% of Hispanic, and 13% of other races/ethnicities reported working for
pay.

92

Journal of Childrens Health. Vol. 2, No. 1, 2004

Effects of School-Year Employment Among Middle School Students

Table 2. Demographic characteristics of south Texas middle school students by


weekly work intensity level

Overall
Characteristics gender
Male
Female
Grade
Sixth
Seventh
Eighth
Race
White
Hispanic
Other
Parent education
College graduate
High school grad
<High school grad

Percent
who do not
work
(n = 2,359)
%

Percent
who work
110 hours
(n = 2,382)
%

Percent
who work
1120 hours
(n = 365)
%

Percent
who work
>21 hours
(n = 261)
%

Total
%

44.0

44.4

6.8

4.9

100.0

38.0
49.7

46.9
41.9

8.5
5.2

6.6
3.2

100.0
100.0

45.1
43.5
43.5

45.2
45.0
43.1

6.1
6.9
7.3

3.6
4.6
6.1

100.0
100.0
100.0

34.2
49.9
42.9

53.0
39.5
43.5

7.4
6.5
6.5

5.4
4.1
7.0

100.0
100.0
100.0

41.6
45.2
45.7

46.5
43.9
42.4

6.3
7.1
7.1

5.5
3.8
4.8

100.0
100.0
100.0

Weekly Work Hours and the Outcome Variables


Table 3 presents the results for the relationship between weekly work intensity
and the outcome variables (the means for the outcome variables appear in the no
job, 110 hours, 1120 hours, and 21+ hours columns).
School Performance/Engagement
We found that weekly work intensity was associated with declines in school performance and engagement, Wilks Lambda = 12.6, p < .01. High-intensity workers had
more unexcused absences, cheated more frequently per week, slept in class, and
were tardy to school more frequently than nonworkers and low-intensity workers
(p < .001 for all). Moderate-intensity workers made poorer grades than nonworkers (p < .01). However, low- and moderate-intensity workers spent more homework
hours weekly than nonworkers (p < .001). In general, students working 21+ hours
weekly exhibited the poorest school engagement profile.
Health Risk Behaviors
Weekly work intensity was positively associated with increased risk behaviors,
Wilks Lambda = 4.9, p < .001. High- and moderate-intensity workers reported using cigarettes and smokeless tobacco on more days in the past month, smoking
more cigarettes per day on the days smoking occurred out of the past month, getting drunk on more days in the past month, and drinking at least one alcoholic

Journal of Childrens Health. Vol. 2, No. 1, 2004

93

Weller et al.

Table 3. Means of dependent variables by work intensity, south Texas middle


school students
Variable

No job
mean (a)

110 hrs 1120 hrs 21+ hrs


Effect
mean (b) mean (c) mean (d) size (D)* p Value

School performance/engagement (n = 2,972)


Homework hours/week
2.14(b,c)
2.44
2.44
2.34
Grades usually made
5.24(c)
5.22
4.93
4.99
Weekly absences
.62(c,d)
.67(c,d) 1.00
1.29
Weekly cheating
1.03(c,d)
1.07(c,d) 1.51(d)
1.93
Frequency class sleeping
1.99(b,c,d) 2.12(c,d) 2.59
2.61
Frequency school tardies
1.97(c,d)
2.00(c,d) 2.31
2.40

.13
.15
.40
.42
.47
.34

<.001
<.01
<.001
<.001
<001
<.001

.60
.61
.63
.54
.57
.60
.38
.54
.00
.41
.006
.03
.29

<.001
<.001
<.001
<.001
<.001
<.001
<.001
<.001
Ns
<.001
Ns
<.001
<.001

Physical/mental health concerns (n = 5,059)


Hours week night sleep
7.41(c,d)
7.41(c,d) 6.98
6.84
Hours weekend sleep
7.33(c,d)
7.14(c,d) 6.65
6.51
Frequency of headaches
2.87(c,d)
2.92(d)
3.07
3.13
Frequency of minor illness 2.39
2.35
2.42
2.49
Frequency of stress
2.97(c,d)
3.03
3.20
3.28

.29
.35
.24
.10
.26

<.001
<.001
<.001
Ns
<.001

Social activities (n = 5,229)


Satisfaction w/leisure time 3.13(c,d)
3.05
2.92
Days with friends
2.69(b,d) 3.03
3.05

.16
.23

<.01
<.001

Days smoked past month


Cigarettes/day past month
Days with snuff past month
Days with drink past month
Days drunk past month
Lifetime marijuana use
Lifetime cocaine use
Lifetime inhalant use
Lifetime steroid use
Lifetime needle use
Fruits/vegetables yesterday
Fat yesterday
Exercise past week

Health risk behaviors (n = 1904)


1.67(c,d)
1.75(c,d) 2.15(d)
1.59(c,d)
1.68(c,d) 2.05
1.10(c,d)
1.15(c,d) 1.38(d)
1.66(c,d)
1.70(c,d) 2.19
1.44(c,d)
1.50(c,d) 1.78
1.48(c,d)
1.53(c,d) 1.95
.08(c,d)
.09(c,d)
.16
.14(c,d)
.16(c,d)
.25
.55
.50
.54
.07(c,d)
.07(c,d)
.13
4.73(b,c,d) 4.81
4.72
5.43(b,c,d) 5.76
5.64
4.22 (b,c,d) 4.82
4.96

2.79
2.50
1.83
2.54
2.33
2.58
.19
.30
.61
.15
4.73
5.49
4.90

2.90
3.12

Notes: Sample size varies due to cases containing missing information.


D statistics = magnitude of difference between 110 hr/wk and 21+ hr wk workers.
Letters (a,b,c,d) = significantly different post-hoc Scheffe means, e.g., for homework, nonworkers are
significantly different from 110 and 1120 hr/wk workers at p < .001.
Wilks lambda, multivariate tests: school performance/engagement 12.6, p < 001; risk behaviors 4.9,
p < .001; physical and mental health 5.6, p < .001; and social activities 9.3, p < .001. Ns = nonsignificant.
p Value indicates significant univariate variables in overall MANCOVA even though corresponding
Scheffe may be nonsignificant.

beverage on more days in the past month than nonworkers or low-intensity workers.
High- and moderate-intensity workers also reported more frequent lifetime use of
marijuana, cocaine, inhalants, and injectable drugs than low-intensity and nonworkers. Employees in all three work categories reported more frequently exercising

94

Journal of Childrens Health. Vol. 2, No. 1, 2004

Effects of School-Year Employment Among Middle School Students

aerobically than nonworking students (p < .001 for all health risk behavior variables). Overall, students working 21+ hours weekly exhibited a poorer health risk
behavior profile.
Physical and Mental Health Concerns
We found that weekly work intensity was associated with increased physical and
mental health concerns, Wilks Lambda = 5.6, p < .001. For example, high- and
moderate-intensity workers reported reductions in hours of weeknight and weekend
sleep, more frequent headaches, and increased feelings of stress compared to lowintensity workers and nonworkers (p < .001 for all these variables). Again, highintensity workers exhibited the poorest physical and mental health profile.
Social Life
Weekly work intensity was both positively and negatively associated with social life
factors measured in this study, Wilks Lambda = 9.3, p < .001. Moderate- and highintensity workers were less satisfied with amounts of leisure time than nonworkers
and low-intensity workers (p < .01), and low- and high-intensity workers spent more
time with friends after school than nonworkers (p < .001).
Effect Sizes
The magnitude of differences among students working at various weekly work
intensity levels is also provided in Table 3. This table provides the means for each
of the outcome variables, as well as the effect size (d statistic), which represents the
contrast between students working 110 hours weekly and those working 21+ hours
weekly. We chose this contrast because of the controversy surrounding high-intensity
work among student workers; on the other hand, low-intensity work has been shown
to provide certain benefits to employed students.
Table 3 indicates that the magnitude of the difference between workers with
high versus low involvement in work is highly dependent on the outcome variable.
The greatest differences in effect sizes are found in the substance use variables, the
academic variables (absences, cheating, sleeping in class, and tardies), and physical
and mental health concerns (hours of nightly sleep, frequency of headaches and
feeling stressed, and time spent with friends after school). The effect sizes for other
variables were more moderate.
DISCUSSION
This study investigated the relation between different weekly work intensity levels and certain academic, social, and health outcomes in a sample of south Texas
middle school students heavily represented with low-income, Hispanic youth. Our
findings suggest that working longer hours weekly may have untoward effects on student functioning: most notably, employment for greater than 20 hours weekly was

Journal of Childrens Health. Vol. 2, No. 1, 2004

95

Weller et al.

positively associated with using several substances, psychological stress, and


headaches. Dissatisfaction with amounts of leisure time and reduced hours of
weeknight and weekend sleep were also associated with 20+ hours of work, as were
several academic factors, including weekly absences and cheating, tardies to school,
and sleeping in class. Weekday afternoon time spent with friends after school and
aerobic exercise during the past week were also positively related to high intensity
work.
Weekly Work Intensity
A critical and controversial issue in the study of youth employment among older
adolescents is weekly work intensity or the number of hours that a youth works
weekly (Steinberg and Cauffman 1995; Ruggiero 1984; Steinberg and Dornbusch
1991; Steinberg et al. 1993; Weller 2003a, b, d; Institute of Medicine 1998). For
the past several decades, researchers and policy makers have sought to determine
a clear weekly work hours threshold (usually defined as more than 20 hours per
week among older teens, possibly 15 hours per week among younger youth) before
which youth employment appeared to benefit youngsters and past which the negative
effects of youth employment were more apparent. Although study results have been
mixed and a clear weekly work hours threshold remains indefinite, considerable
evidence has accumulated supporting the benefits of low- to moderate-intensity work
among older adolescents and the possible ill effects of higher-intensity employment
during the school year (Steinberg and Cauffman 1995; Institute of Medicine 1998).
Unfortunately, weekly work intensity among children and early adolescents was not
investigated in these studies of older youth, and the findings from these studies may
not directly generalize to considerably younger youth. One purpose of the current
study, therefore, was to investigate the number of hours that appear to constitute
too much weekly work in children and early adolescent youth.
Similar to the results of studies among older youth, our findings among middle schoolers did not clearly identify a cut-off point beyond which negative
work effects for most variables were observed (Greenberger and Steinberg 1986;
Institute of Medicine 1998; Bachman and Schulenberg 1993; Steinberg and
Cauffman 1995; Steinberg and Dornbusch 1991; Weller 2003c). However, as in
our high school sample, negative results for many academic, risk behaviors, physical/mental health, and social variables were more pronounced in the higher two
work intensity categories (1120 hours and 21+ hours weekly) (Weller 2003d). For
many variables (e.g., cigarettes per day in the past month and weekly absences from
school), these results appear to suggest that as weekly work hours increase, corresponding complications may also increase. In the case of many variables, no significant differences in the means were observed between nonworkers and low-intensity
workers (weekly cheating, lifetime needle use, number of days in the past month during which the student drank five or more drinks in a row within a couple of hours).
Moreover, in some cases, low intensity workers fared slightly better than nonworkers
(homework hours per week), possibly indicating the benefits of working a few hours

96

Journal of Childrens Health. Vol. 2, No. 1, 2004

Effects of School-Year Employment Among Middle School Students

weekly (Steinberg and Cauffman 1995; Institute of Medicine 1998). Finally, for a few
variables (exercise during the past week), working middle schoolers employed at all
three intensity levels performed significantly better than nonworkers.
School Performance and Engagement
For several decades since youth employment has become the norm in our society, adolescent researchers have expressed considerable concern about the impact
of long hours of student employment on the academic performance of older adolescents, youth who may legally work under federal and state child labor laws. The
current study centers around the effect of such employment among children and
early adolescents, youth who are still growing intellectually, socially, physically, and
emotionally, and whose informal employment is not protected by state and federal
regulations and agencies.
The results of our investigation regarding the effect of school-year employment
on the academic performance and school engagement of our middle school sample
confirmed the findings of studies among older youth, including our high school
sample. Studies investigating the relation between longer work hours and academic
performance (GPA) commonly report a weak association, a reciprocal relationship,
or mixed results (Marsh 1991; Mortimer and Finch 1986; Mihalic and Elliot 1997;
Mortimer and Johnson 1998; Mortimer et al. 1996; Schoenhals et al. 1997; Steinberg
1993; Weller 2003d). Methodologic characteristics, such as design and selection
effects, contribute to these ambiguities.
We found statistically significant, but relatively modest, differences in school performance between nonworkers and moderate-intensity workers, the latter reporting
about one-quarter letter grade lower GPA. Other researchers, who report very similar findings among older youth, suggest that students may maintain their grades
despite their jobs by cheating, copying, cutting classes, and other compensatory
strategies (McNeil 1984; Bills 1995). Our middle schoolers who worked 1120 hours
and 21+ hours weekly during the school year appear to have cheated, slept in class,
and been tardy to school significantly more frequently than nonworkers and students working 110 hours weekly. The effect sizes for these variables ranging from
.34 to .47 also indicate substantial differences in these practices between low- and
high-intensity workers, suggesting the latter may have to compensate more for their
job commitments than students employed fewer hours weekly.
Health Risk Behaviors
Our findings regarding substance use reveal an important association between
working longer hours and higher rates of monthly tobacco and alcohol use and
lifetime marijuana, inhalant, cocaine, and injectable drug use. An analysis of the
effect sizes for these substances suggests an important distinction in the behaviors of
students working 10 or fewer hours weekly and those working more than 20 hours
weekly. These effect size differences favoring low- over high-intensity middle school

Journal of Childrens Health. Vol. 2, No. 1, 2004

97

Weller et al.

workers in the range of .38 to .63 standard deviations are of adequate size to justify
concern and are especially pronounced for substance use variables compared to
other variables measures in this study.
Our results are similar to the findings of several other studies among older
workers, including our high school sample (Bachman and Schulenberg 1993;
Greenberger and Steinberg 1986; Mihalic and Elliott 1997; Mortimer et al. 1996;
Resnick et al. 1997; Steinberg and Dornbusch 1991; Weller 2003d). A 1991 crosssectional study found an effect size of .30 for drug use variables favoring low-intensity
workers in a large, primarily middle-class and white sample from California and Wisconsin (Steinberg and Dornbusch 1991). Moreover, the associations in our sample
held after applying statistical controls for gender, grade, race/ethnicity, and parent
education, indicating that the negative substance use correlates of student employment cross age, racial/ethnic, and socioeconomic lines.
One interpretation of these findings would suggest that part-time work is related
to drug use, and the more hours worked, the greater the likelihood of substance use.
However, due to the cross-sectional nature of this study, it is impossible to determine
whether the observed differences between low- and high-intensity workers reflect the
genuine effects of employment, differential selection into the workplace influenced
by real differences in workers (those with preexisting drug/behavioral problems or
drug susceptibility or biology that influences amount of work due to addictions or
academic weaknesses that encourage high-intensity work over school), or differential
reporting by high- and low-intensity workers.
Social Life and Mental/Physical Health Concerns
One criticism of student employment during the school year, especially when it
involves long work hours, is that it may steal time from other important developmental activities (Resnick 1997; Bachman and Schulenberg 1993; Carskadon 1990;
Mortimer and Johnson 1998). One might suspect that satisfaction with leisure time,
for example, would be negatively related to increasing weekly work intensity. Such
was the case; the relationship among our middle school sample was strong and
nearly linear and paralleled the results of our south Texas high school study (Weller,
2003a). Similar to the findings of other studies, we also observed significant differences in the frequency of feeling stressed and having headaches favoring lowintensity/nonworkers over high-intensity workers (Steinberg and Dornbush 1991;
Mortimer 1994; Mortimer and Johnson 1998; Resnick 1997; Weller 2003d). Interestingly, high-intensity workers reported spending more time with friends after school
than nonworkers and low-intensity workers.
Limitations
Several caveats should be discussed in interpreting study results: 1) information
was self-reported, subject to accuracy of recall and not possible to confirm due to
study design; 2) thresholds for weekly work hours were limited to the categories
used in this analysis; 3) the cross-sectional design does not allow a determination

98

Journal of Childrens Health. Vol. 2, No. 1, 2004

Effects of School-Year Employment Among Middle School Students

regarding temporality, i.e., whether work or behavior comes first, which suggests
the need for future prospective cohort studies; 4) data from a non-participating
urban school district serving a population of one-quarter million was not collected,
possibly influencing our results; and 5) generalizability of this study is limited to
similar populations of public middle school students from primarily rural geographic
areas, a substantial portion of whom are economically compromised and ethnically
similar to the present study composite.
CONCLUSIONS
In summary, we found evidence suggesting that school-year employment may have
negative effects on south Texas middle school students. Most pronounced were our
findings suggesting that students were more likely to engage in substance use as hours
of work increased. High-intensity work was also associated with psychological distress
and school disengagement. However, for several variables, low-intensity workers had
better outcomes than nonworkers, suggesting that while high-intensity work may
have negative effects, low-intensity work may be beneficial in some cases. A unique
contribution of this study is its description of the effects of academic school-year employment in a previously undescribed, largely low-income and Hispanic population
from a rural geographic area that also has not been studied.
In light of our findings and those of other studies, we believe that parents and professionals involved with youth should supervise the number of weekly hours that adolescents work while attending school. Since substantial numbers of youth combine
school and work, school-based intervention projects could educate youth about the
risks and benefits associated with youth employment. Several of these preventative
occupational health education efforts, many of them school-based, have been developed over the past decade (Institute of Medicine 1998; Runyan and Zakocs 2000).
These programs include occupational health and safety training for adolescents
and education regarding child labor laws and worker rights and responsibilities for
youth, their parents, professionals, and employers. Further investigation is needed,
however, to target such interventions to subpopulations of student employees, to
evolve methods and strategies to market these programs to youth, and to rigorously
evaluate the effectiveness of these health education efforts to these youngest and
most vulnerable workers.
REFERENCES
Bachman JG and Schulenberg J. 1993. How part-time work intensity relates to drug use, problem behavior, time use, and satisfaction among high school seniors: Are these consequences
or merely correlates? Devel Psych 29(2):220235.
Bills DB, Helms LB, and Ozcan M. 1995. The impact of student employment on teachers
attitudes and behaviors toward working students. Youth & Society 27(2):169193.
Brener ND, Collins JL, Kann L, Warren CW, and Williams BI. 1995. Reliability of the youth
risk behavior survey questionnaire. Am J Epidemiol 141(6):575580.

Journal of Childrens Health. Vol. 2, No. 1, 2004

99

Weller et al.

Carr RV, Wright JD, and Brody CJ. 1996. Effects of high school work experience a decade
later: Evidence from the National Longitudinal Survey. Sociol Educ 69:6681.
Carskadon MA. 1990. Patterns of sleep and sleepiness in adolescents. Pediatrician 17:512.
Carskadon M. 1995. Personal Communication. School Sleep Habits Survey. E.P. Bradley Hospital. Chronobiology/Sleep Research Laboratory, Riverside, RI.
Chaplin D and Hannaway J. 1996. High school enrollment: Meaningful connections for atrisk youth. Paper presented at the Annual Meeting of the American Educational Research
Association, New York. Urban Institute, Washington, DC.
Cohen J. 1988. Statistical Power Analysis in the Behavioral Sciences. Lawrence Erlbaum,
Hillsdale, NJ.
Davis L. 1995. Personal Communication. Massachusetts SENSOR Project: Sentinel Notification System for Occupational Risks: Surveillance and Prevention of Work-Related Injury
to Teens. Teens at Work: Follow-Up Questionnaire. Massachusetts Department of Public
Health, Boston, MA.
Dornbusch SM, Ritter PL, Liederman P, Roberts D, and Fraleigh M. 1987. The relation of
parenting style to adolescent school performance. Child Devel 58:12441257.
Dunn K. 1994. Personal Communication. North Carolina Teens at Work Questionnaire. Department of Emergency Medicine, East Carolina University, Greenville, NC.
Dunn KA, Rungan CW, Cohen LR, and Schulman MD. 1998. Teens at work: A statewide study
of jobs, hazards, and injuries. J Adolesc Health 22:1925.
Greenberger E and Steinberg L. 1986. When Teenagers Work: The Psychological and Social
Costs of Adolescent Employment. Basic Books, New York, NY.
Huang L, Pergamit, M, and Shkolnik J. 2001. Youth initiation into the labor market. Mon
Labor Rev August, 1824.
Institute of Medicine. 1998. Protecting youth at work: Health, safety, and development of working children and adolescent in the United States. National Academy Press, Washington,
DC.
Kolbe LJ, Kann L, and Collins J. 1993. Overview of the Youth Risk Behavior Surveillance
System. Public Health Rep 108(suppl):210.
Manning WD. (1990). Parenting employed teenagers. Youth & Society 22:184200.
Marsh HW. 1991. Employment during high school: Character building or subversion of academic goals? Sociol Educ 64:172189.
McNeil L. 1984. Lowering Expectations: The Impact of Student Employment on Classroom
Knowledge. Wisconsin Center for Educational Research, Madison, WI.
Mihalic SW and Elliot D. 1997. Short- and long-term consequences of adolescent work. Youth
& Society. 28(4):464498.
Miller, M (1995). Occupational injuries among adolescents in Washington State, 1988-1991:
A review of workers compensation data (Technical Report Number 35-1-1995). Olympia,
WA: Washington State Department of Labor and Industries.
Miller ME and Kaufman JD. 1998. Occupational injuries among adolescents in Washington
State, 19881991. Am J Ind Med 34:121132.
Mortimer JT and Finch MD. 1986. The effects of part-time work on self-concept and achievement. In Borman K, and Reisman J (eds), Becoming a Worker, Ablex, Norwood, NJ.
Mortimer JT, Finch MD, Ryu S, Shanahan MJ, and Call KT. 1996. The effects of work intensity
on adolescent mental health, achievement, and behavioral adjustment: New evidence from
a prospective study. Child Devel 67:12431261.

100

Journal of Childrens Health. Vol. 2, No. 1, 2004

Effects of School-Year Employment Among Middle School Students

Mortimer JT and Johnson MK. 1998. New perspectives on adolescent work & the transition to
adulthood. In Jessor R (ed.), New Perspectives on Adolescent Risk Behavior. Cambridge
University Press, New York.
Mortimer JT and Shanahan MJ. 1994. Adolescent work experience and family relationships.
Work and Occupations 21:369384.
Radloff LS. 1977. The CES-D scale: A self-report depression scale for research in the general
population. Appl Psychol Meas 1:385401.
Resnick MD, Bearman PS, Blum RW, Bauman KE, Harris KM, Jones J, Tabor J, Beuhring T,
Sieving RE, Shew M, Ireland M, Bearinger LH, and Udry JR. 1997. Protecting adolescents
from harm. Findings from the National Longitudinal Study on Adolescent Health. JAMA
178(10):823832.
Rothstein DS. 2001. Youth employment in the United States. Mon Labor Rev 124:6
17.
Ruggiero M. 1984. Work as an impetus to delinquency: An examination of theoretical and
empirical connections. Unpublished doctoral dissertation. University of California, Irvine,
CA.
Runyan CW and Zakocs RC. 2000. Epidemiology and prevention of injuries among adolescent
workers in the United States. Ann Rev Pub Health 21:247269.
Schoenhals M, Tienda M, and Schneider B. 1997. The educational and personal consequences
of adolescent employment. Unpublished paper. Department of Sociology, University of
Chicago, Chicago, IL.

R
SPSS Inc. 1995. SPSS for Windows-TM: Advanced Statistics, Release 6.1.3., SPPSS, Inc.,
Chicago, IL.
Steinberg L. 1995. Personal Communication. Temple University Department of Psychology,
School-Year Work Questionnaire, Philadelphia, PA.
Steinberg L and Cauffman E. 1995. The impact of employment on adolescent development.
In Vasta R (ed), Ann Child Development 11:131166.
Steinberg L and Dornbusch S. 1991. Negative correlates of part-time employment during
adolescence: Replication and elaboration. Devel Psych 27(2):304313.
Steinberg LD, Fegley J, and Dornbusch JM. 1993. Negative impact of part-time work
on adolescent adjustment: Evidence from a longitudinal study. Devel Psych 29:171
180.
Survey Research Center. 1995. Personal Communication. Institute for Social Research. Monitoring the Future Survey. The University of Michigan, Ann Arbor, MI.
Texas Education Agency. 1993. Snapshot 93: 1992-93 School District Profiles. DOC#
GE417001. TEA, Austin, TX.
Texas Employment Commission. 1992. Texas employment commission rules. Texas Employment Commission, Austin, TX.
United Nations Childrens Fund. 1997. State of the Worlds Children 1997. Oxford University
Press, Oxford, UK.
U.S. Department of Labor, Employment Standards Administration, Wage and Hour Division.
1990. Child labor requirements in nonagricultural occupations under the Fair Labor Standards Act. U.S. Department of Labor, Washington, DC.
Weller NF, Basen-Engquist K, Cooper SP, Kelder SH, and Tortolero SR. 2003a. School-year
employment among Hispanic high school students in rural south Texas: Prevalence and
patterns. Texas J Rural Health 21:5771.

Journal of Childrens Health. Vol. 2, No. 1, 2004

101

Weller et al.

Weller NF, Cooper SP, Basen-Enguist K, Kelder SH, and Tortolero SR. 2003b. The prevalence
and patterns of occupational injury among south Texas high school students. Texas Med
99:5257.
Weller NF, Cooper SP, Tortolero SR, Kelder SH, and Hassan S. 2003c. Work-related injury
among south Texas middle school students: Prevalence and patterns. South Med J 96:1213
1220.
Weller NF, Kelder SH, Cooper SP, Basen-Engquist K, and Tortolero SR. 2003d. School-year
employment among high school students: Effects on academic, social, and physical functioning. Adolescence 38:441458.
Wright JP, Cullen FT, and Williams N. 1997. Working while in school and delinquent involvement: Implications for social policy. Crime and Delinquen 43:203221.
Wofford S. 1988. A Preliminary Analysis of the Relationship Between Employment and Delinquency/Crime for Adolescents and Young Adults. National Youth Survey No. 50. Institute
of Behavioral Science, University of Colorado, Boulder, CO.

102

Journal of Childrens Health. Vol. 2, No. 1, 2004

You might also like