Page 1
IJOMEH 2010;23(3) 239
O R I G I N A L P A P E R S
International Journal of Occupational Medicine and Environmental Health 2010;23(3):239 – 253
DOI 10.2478/v10001-010-0029-0
PSYCHOSOCIAL WORKING CONDITIONS
AND ACTIVE LEISURE-TIME PHYSICAL ACTIVITY
IN MIDDLE-AGED US WORKERS
BONGKYOO CHOI1, PETER L. SCHNALL1, HAIOU YANG1, MARNIE DOBSON1, PAUL LANDSBERGIS2,
LESLIE ISRAEL1, ROBERT KARASEK3,4, and DEAN BAKER1
1 University of California Irvine, Irvine, USA
Center for Occupational and Environmental Health
2 The State University of New York Downstate School of Public Health, Brooklyn, USA
Department of Environmental and Occupational Health Sciences
3 University of Massachusetts Lowell, Lowell, USA
Department of Work Environment
4 Copenhagen University, Copenhagen, Denmark
Department of Psychology
Abstract
Objectives: This study was to examine whether psychosocial work characteristics such as job control, psychological job
demands, and their combinations are associated with leisure-time physical activity (LTPA) in US workers. Materials and
Methods: 2019 workers (age range: 32 to 69) from the National Survey of Midlife Development in the United States
(MIDUS) II study (2004–2006) were chosen for this cross-sectional study. Job control and job demands were measured by
standard questionnaire items. Active LTPA was defined as “moderate or vigorous” level of physical activity. Results: After
controlling for covariates (e.g., age, race, education, income, physical effort at work, obesity, and alcohol consumption),
high job control was associated with active LTPA. Active jobs (high control and low demands) and low-strain jobs (high
control and high demands), compared to passive jobs (low control and low demands), increased the odds for active LTPA.
The associations varied by sex and education level. Job demands alone were not associated with active LTPA. Conclusions:
Having on-the-job learning opportunities and decision authority on their tasks may be conducive to active LTPA in middle-
aged US workers.
Key words:
Job control, Job demands, Education, Obesity, MIDUS
Grant Sponsor: Center for Social Epidemiology, Marina Del Rey, California. Grant Number: Proposal # 48004, Office of Research Administration, University of
California Irvine.
Address reprint requests to B. Choi, Center for Occupational and Environmental Health, University of California Irvine, 5201 California Avenue, Suite 100, Irvine,
CA, 92617, USA (e-mail: b.choi@uci.edu).
INTRODUCTION
Regular physical activity reduces the risk of all-cause
mortality, coronary heart disease, high blood pressure,
stroke, type 2 diabetes, metabolic syndrome, colon can-
cer, breast cancer, and depression [1,2]. Thus it has been
widely recommended as a key public health policy [1,3,4]
in the United States (US): for instance, adults should en-
gage in at least 30 minutes of moderate-intensity physical
activity (e.g., brisk walking) on five days of the week.
However, according to the recent statistics from the Be-
havioral Risk Factor Surveillance System (BRFSS) [5],
only half of US adults meet the recommended level
of physical activity and about 25% of adults do not en-
gage in any physical activity during their leisure time.
More importantly, the proportions have not changed
Page 2
O R I G I N A L P A P E R S B. CHOI Et AL.
IJOMEH 2010;23(3)240
(e.g., reading books, sports activities, social participation,
and political activities), but in contrast, those who have
a “passive job” (a combination of low control and low
demands), would undergo a process of “skill atrophy and
unlearning” [10, p. 94]. This group would be least active
in the leisure activities. In addition, those who had “low
strain” jobs (a combination of high control and low de-
mands) or “high strain” jobs (a combination of low con-
trol and high demands) would have intermediate levels
of the leisure activities. In a national Swedish study [6],
sports activities during leisure time were more prevalent
in those who had high control, high demands, and an ac-
tive job (vs. passive job).
Other theories have focused on the internal process of how
people change behavior [11–13]. These theories posit that
for successful adoption of healthy behaviors, a person
needs to be highly motivated and also to have skill and
ability to actualize a well-conceived plan for the behavior
against barriers in reality.
In sum, the above theories imply that job control, psycho-
logical job demands, and an active job (vs. passive job)
would be associated with LTPA in a working population.
Empirical evidence:
psychosocial working conditions and active LTPA
Empirical evidence for these associations has been in-
conclusive. Some studies have supported the associations
for high job control [6,14–18], high psychological job
demands [6] and an active job [6,15,19]. But other stud-
ies have not supported the associations for high job con-
trol [12,20], high psychological job demands [12,14–18,20]
and an active job [14,21,22]. In addition, only a few stud-
ies [15,16,19,20,22] examined the associations after con-
trolling for possible confounders such as socioeconomic
status, physical effort at work, and working hours. Further-
more, to our knowledge, few studies [15] have examined
the associations in US workers from diverse occupations
and industries.
The aim of this study is to examine whether psychosocial
work characteristics are associated with active (moderate
or vigorous level) LTPA in middle-aged US workers, using
a recent US national dataset.
much during the last fifteen years. The reasons for this
situation are not well understood and there may well be
important unidentified social and environmental deter-
minants of leisure-time physical activity (LTPA). This
study explores the role of psychosocial work character-
istics such as job control, psychological job demands,
and their combinations as occupational determinants
of LTPA in US workers.
Theories: the relationship between work and LTPA
A number of sociologists [6–9] have postulated that work
characteristics could affect leisure-time activities of work-
ers and, generally, there would be a similarity rather
than an antithesis between work and non-work activities
(called “spill-over hypothesis” vs. “compensation hy-
pothesis”). Meissner [7] suggested, in a study on manual
workers at a Canadian wood-product manufacturing fac-
tory, that “the design of industrial work creates or pre-
vents opportunities for the development or maintenance
of discretionary and social skills. When choice of action is
suppressed by the spatial, temporal, and functional con-
straints of the work process, worker capacity for meeting
the demands of spare-time activities which require discre-
tion is reduced. They engage less in those activities which
necessitated planning, coordination, and purposeful ac-
tion” (p. 260). In the study, Meissner [7] reported that
those who had high job discretion engaged more frequent-
ly in sports activities during their leisure time, in contrast,
those who had low job discretion engaged more frequently
in sedentary activities at home (e.g., going for a drive, TV
watching) during their leisure time.
Karasek [6] also proposed a similar, but more sophisti-
cated hypothesis based on the two concepts — job con-
trol (opportunities at work for learning and decision
making) and psychological job demands (degree of men-
tal work demands). According to the active-passive hy-
pothesis of the demand-control (DC) model [6,10], those
who have an “active job” (a combination of high control
and high demands) would develop “more skills through
the trial/failure/success learning process” [10, p. 93] and
an active personality (“feeling of mastery or confidence”,
p. 98). Thus, they would be most active in leisure activities
Page 3
WORK AND LEISURE tIME PHYSICAL ACtIVItY O R I G I N A L P A P E R S
IJOMEH 2010;23(3) 241
gender. However, in the follow-up, less-educated per-
sons and non-whites were relatively more likely to have
dropped out of the study. For the MIDUS II study, 4963
persons (males, 47% and females, 53%) completed a tele-
phone interview only (N = 931) and both the interview
and mailed questionnaires (N = 4032). For this analysis,
we first restricted study subjects to those (N = 2292) who
completed both the interview and mailed questionnaires
(N = 4032), were not pregnant (N = 4954), were work-
ing a paid-job (at least one hour per week at a main job)
in the MIDUS II survey (N = 2469), and were aged less
than 70 years (N = 4177). Finally 2019 workers (1001 males
and 1018 females) who had valid information on the expo-
sure and outcome variables (see below) were chosen for
this analysis.
Main exposures — job control, psychological job
demands, and their combinations
Both job control and psychological job demands were as-
sessed by self-administered questionnaire items, similar
to the ones of the Job Content Questionnaire (JCQ) [24].
Job control was measured with five items about skill dis-
cretion (2 items; e.g., “How often do you learn new things
at work?”) and decision authority (3 items; e.g., “How of-
ten do you have a choice in deciding how you do your tasks
at work?”). Psychological job demands were measured
with three items about work intensity, workload, and time
pressure (e.g., “How often do you have enough time to
get everything done?” — reverse scored). More detailed
information about the items is available elsewhere [25].
The items had a five-point Likert type of response set: all
of the time (1) to never (5) and were summed up for scal-
ing-scoring. The Cronbach alphas of job control and psy-
chological job demands were 0.81 and 0.68, respectively.
For analysis, the two scale scores were divided into four
groups based on their quartiles. The lowest groups of job
control and psychological job demands scale scores were
chosen as the reference for analysis (see below).
The four quadrants of the DC model — active, low strain,
high strain, and passive jobs — were created by two differ-
ent methods: a) based on the medians of job control and
psychological job demands (hereafter called “four-group
METHODS
Study population
Data from the National Survey of Midlife Development
in the United States (MIDUS) II study [23] were used
for this study. Information on physical effort at work
and leisure-time physical activity was not available in
the MIDUS I study, so it was not possible to perform
a longitudinal analysis. From 1995 to 1996, the MacAr-
thur Midlife Research Network carried out a nation-
al survey (i.e., MIDUS I study) to investigate the role
of behavioral, psychological, and social factors in un-
derstanding age-related differences in physical and
mental health [23]. In the MIDUS I study, 7108 persons
(males, 48% and females, 52%) completed a telephone
interview only (N = 783) or both the interview and
mailed questionnaires (N = 6325). All of the participants
were non-institutionalized, English-speaking adults,
aged 25–74, in the US. They were drawn from four sub-
samples: (a) a national random-digit-dial (RDD) sample
(N = 3487); (b) oversamples from five metropolitan ar-
eas (N = 757); (c) siblings of individuals from the RDD
sample (N = 950); and (d) a national RDD sample
of twin pairs (N = 1914). The response rates of the four
subsamples ranged from 60% to 70%. The four subsam-
ples were very similar to one another in terms of the dis-
tributions of age, education, and gender [23]. The socio-
demographic characteristics of the RDD subsample were
comparable to those of a US population representative
sample, the October 1995 Current Population Survey
(http://www.census.gov/cps). However, the RDD sub-
sample relatively underrepresented those who were
black, young (e.g., aged 25 to 34), or had less education
(i.e., 12 or less than 12 years of formal education) [23].
A follow-up survey of the participants of the
MIDUS I study respondents was conducted from 2004
to 2006. The average follow-up interval was approximate-
ly 9 years later and ranged from 7.8 to 10.4 years. The lon-
gitudinal retention rates among the four subsamples
ranged from 65% to 78% (on average, 70%). There were
no significant (p < 0.01) differences between the follow-
up participants and non-participants in terms of age and
Page 4
O R I G I N A L P A P E R S B. CHOI Et AL.
IJOMEH 2010;23(3)242
Covariates
Various potential covariates were considered in the analy-
sis: data sources, socio-demographic measures [4,27–31],
other working conditions [6,16,17,22,27,32,33], health
conditions [28,34], and health behaviors [28–30]. Spe-
cifically, four data sources: city; siblings; and twin sub-
samples (vs. the national random subsample), age (< 40;
40 to 49; 50 to 59; and ≥ 60 years old), sex, marital status
(married and non-married), race (whites vs. others), an-
nual household income (< $ 60 000; $ 60 000 to $ 99 999;
and ≥ $100 000), and education (high — university/gradu-
ate school graduate; middle — some college education,
but unfinished; and low — high school graduate and lower
education). Working conditions were measured by ques-
tionnaire items: physical effort at work (1 item), coworker
(2 items) and immediate supervisor (2 items) support,
and no coworkers (2 items) and immediate supervisors
(2 items). More detailed information about the items is
available elsewhere [25]. In addition, hours of work per
week at a main job (≤ 40 hrs and > 40 hrs per week) and
other paid jobs (yes vs. no) were also self-reported.
The following health conditions and health behaviors were
measured: major depression assessed by the telephone in-
terview, based on the Diagnostic and Statistical Manual
of Mental Disorders III-R [35]; chronic diseases (those
who have experienced or been treated for any of the fol-
lowing during the past 12 months: arthritis, sciatica, recur-
ring stomach trouble or diarrhea, persistent foot troubles,
trouble with varicose veins, multiple sclerosis, stroke, and
hernia; or those who have ever had heart problems or ever
had cancer); obesity (body mass indexes based on self-re-
ported height and weight information, ≥ 30 kg/m2) [36];
smoking (current smokers vs. non-smokers); alcohol con-
sumption (moderate drinking — up to two drinks per day
for men and one drink per day for women [37] during
the past month and heavy drinking — more than mod-
erate drinking vs. no drinking); and stress-induced over-
eating (those who endorsed either of the following two
items about “how you respond when you are confronted
with difficult or stressful events in your life”: “I eat more
than I usually do” and “I eat more of my favorite foods to
make myself feel better”).
definition”); and b) based on the quartiles of job control
and psychological job demands (resulting in 16 possible
cells) [26, see Figure 2-e, p. 195] for avoiding potential
misclassification of the four quadrants of the DC model
around the medians of job control and psychological job
demands. The 4 cells in the middle were labeled as “mid-
dle group” and the other 12 cells (3 corner cells for each)
were labeled as active, low strain, high strain, and passive
jobs (hereafter called “five-group definition”).
Outcome — active LTPA
Active LTPA was defined by moderate or vigorous level
of LTPA which is long enough to work up a sweat, several
times a week, during the summer or the winter. It is quite
consistent with the contemporary minimum recommenda-
tion of physical activity for US adults [1,3,4]: at least 5 days
of week for moderate physical activity and at least 3 days
per week for vigorous physical activity. In detail, vigorous
and moderate LTPA were each assessed with one item [23]:
“during your leisure time or free time, how often do you
engage in vigorous physical activity that causes your heart
to beat so rapidly that you can feel it in your chest and you
perform the activity long enough to work up a good sweat
and are breathing heavily?”; and “during your leisure time
or free time, how often do you engage in moderate physi-
cal activity, that is not physically exhausting, but it causes
your heart rate to increase slightly and you typically work
up a sweat?”, respectively.
The items were specified further for the summer and
the winter, considering a possible seasonable variation
of LTPA. They had a 6-frequency based response set (sev-
eral times a week, once a week, several times a month,
once a month, less than once a month, and never). A pre-
liminary analysis revealed little seasonal (summer vs. win-
ter) variation in responses to the vigorous and moderate
physical activity items: Spearman correlations were 0.90
and 0.87, respectively. Nonetheless, in this study, we
still retained seasonal specificity in the definition of ac-
tive LTPA (i.e., during the summer or the winter) to mini-
mize potential misclassification of the LTPA outcome
variable (essentially equivalent to controlling for potential
confounding by season).
Page 5
WORK AND LEISURE tIME PHYSICAL ACtIVItY O R I G I N A L P A P E R S
IJOMEH 2010;23(3) 243
conditions; Model 3 — additionally with other working
conditions, data sources, and socio-demographic variables;
and Model 4 — additionally with other working conditions,
data sources, socio-demographic variables, health condi-
tions, and health behavior variables. In addition, the above
multivariate analyses were replicated with stratification
of the data by sex and education (Table 5) in consider-
ation of potential interactions between sex, education, and
the psychosocial work characteristics on active LTPA.
Statistical analyses
Descriptive statistics of vigorous and moderate levels
of LTPA are presented in Table 1. The bivariate associations
of the study variables with active LTPA were examined by
chi square tests (Table 2). The associations were then inves-
tigated through a series of multivariate logistic regression
models (Tables 3 and 4): Model 1 — only two variables (job
control and job demands) or the four quadrants of the DC
model; Model 2 — additionally with other working
Table 1. Prevalence of vigorous and moderate leisure-time physical activity (LTPA) in 2019 US male and female workers
Level of LTPA
Men
N = 1001
(%)
Women
N = 1018
(%)
Total
N = 2019
(%)
Vigorous 30.8 28.3 29.5
Moderate 38.2 39.8 39.0
Vigorous or moderate 44.8 44.4 44.6
Table 2. Active leisure-time physical activity prevalence in relation to study variables in 2019 US workers
Major variable
category
Minor variable
category
Subcategory
Frequency
(%)
Active LTPA
(%)
Data source Subsamples National random 41.9 45.8
City 9.2 47.8
Siblings 16.3 43.2
Twin 32.6 42.8
Socio-demographic Sex Men 49.6 44.8
Women 50.4 44.4
Age (years) < 40 11.7 49.8b
40–49 34.8 45.9b
50–59 36.5 45.4b
≥ 60 17.0 36.6b
Marital status Married 73.8 44.6
Non-married 26.2 44.5
Race White 92.7 45.8c
Others 7.3 29.3c
Education High school or less 25.4 31.4c
Some college 28.8 40.2c
University or more 45.8 54.6c
Annual household income ($) < 60 000 32.7 37.0c
60 000–99 999 33.1 43.3c
≥ 100 000 34.2 53.1c
End of preview.
