Switching Off Mentally: Predictors and Consequences of Psychological Detachment From Work During Off-Job Time

Abstract

Psychological detachment from work refers to the off-job experience of “switching off” mentally. It is hypothesized that a high degree of workload encountered during the work day has a negative impact on subsequent detachment processes and that psychological detachment from work is positively related to well-being. Eighty-seven individuals from various occupations provided questionnaire and daily survey measures over a period of 3 working days. Multilevel analysis showed that workload was negatively related to psychological detachment from work during evening hours. Psychological detachment from work was associated with positive mood and low fatigue. The negative relationship between psychological detachment and fatigue was particularly strong on days with high time pressure.

Full text

Switching Off Mentally: Predictors and Consequences of
Psychological Detachment From Work During Off-Job Time
Sabine Sonnentag and Ute-Vera Bayer
University of Konstanz
Psychological detachment from work refers to the off-job experience of “switching off” mentally.
It is hypothesized that a high degree of workload encountered during the work day has a negative
impact on subsequent detachment processes and that psychological detachment from work is
positively related to well-being. Eighty-seven individuals from various occupations provided
questionnaire and daily survey measures over a period of 3 working days. Multilevel analysis
showed that workload was negatively related to psychological detachment from work during
evening hours. Psychological detachment from work was associated with positive mood and low
fatigue. The negative relationship between psychological detachment and fatigue was particularly
strong on days with high time pressure.
Keywords: job stress, overload, psychological detachment, recovery, well-being
Literally hundreds of studies have shown that
stressful work situations are associated with poor
individual well-being and increased health risks (for
reviews, cf. Ganster & Schaubroeck, 1991; Kahn &
Byosiere, 1992; Sonnentag & Frese, 2003). Particu-
larly, high workload was found to be related to im-
paired health and well-being (Sparks, Cooper, Fried,
& Shirom, 1997; Van der Doef & Maes, 1999).
In addition, researchers have argued that the in-
ability to rest and recover from work might have
severe negative effects on individual health and well-
being (Eden, 2001; Lundberg & Lindfors, 2002;
McEwen, 1998; Meijman & Mulder, 1998). Empiri-
cal research has shown that insufficient recovery is
associated with poor psychological and physical
health, such as psychosomatic complaints and burn-
out (Elfering, Grebner, Semmer, & Gerber, 2002;
Sluiter, Van der Beek, & Frings-Dresen, 1999).
Moreover, research evidence suggests that individu-
als’ well-being and work behavior benefit from re-
spite time during evening hours (Sonnentag, 2001,
2003) as well as from longer respite periods such as
vacations (Westman & Eden, 1997; Westman & Et-
zion, 2001). It seems, however, that it is not merely
the amount of the time available for respite and
recovery that matters; the quality of the respite ex-
perience also plays an important role in the recovery
process (Etzion, Eden, & Lapidot, 1998; Lounsbury
& Hoopes, 1986; Westman & Eden, 1997).
Eden (2001) suggested that psychological detach-
ment from work, that is, not thinking of one’s work
during off-job time, should be a core factor in future
respite research. Psychological detachment from
work has been shown to be one of the major factors
that contribute to the improvement of well-being
during longer respite periods (Etzion et al., 1998).
Until now, researchers have focused on psychologi-
cal detachment during relatively long respite periods,
such as those lasting 2 weeks or longer (Etzion et al.,
1998). To our knowledge, short-term psychological
detachment from work, such as that occurring during
the evenings of normal work weeks, has not yet been
investigated. Moreover, it is largely unclear which
factors support or hinder psychological detachment
from work.
There are at least two reasons why it is important
to extend research on psychological detachment to
shorter time intervals. First, the effects of longer
respites (e.g., vacations) fade out relatively quickly
(Fritz & Sonnentag, in press; Westman & Eden,
1997); therefore, it is useful to investigate which
factors help in maintaining well-being in between
these periods. Second, if individuals do not recover
sufficiently within shorter time intervals, strain reac-
tions may accumulate and result in severe impair-
ments of health and well-being (Meijman & Mulder,
1998). Therefore, it is crucial to examine the ante-
cedents and potential benefits of short-term detach-
ment from work.
Sabine Sonnentag and Ute-Vera Bayer, Department of
Psychology, University of Konstanz, Konstanz, Germany.
Correspondence concerning this article should be ad-
dressed to Sabine Sonnentag, Department of Psychology,
University of Konstanz, Postbox D42, Konstanz D-78457,
Germany. E-mail: sabine.sonnentag@uni-konstanz.de
393
First publ. in: Journal of Occupational Health Psychology 10 (2005), 4, pp. 393-414
Konstanzer Online-Publikations-System (KOPS)
URL: http://www.ub.uni-konstanz.de/kops/volltexte/2008/5644/
URN: http://nbn-resolving.de/urn:nbn:de:bsz:352-opus-56447

In this article, we examine short-term psycholog-
ical detachment from work occurring during the eve-
nings of normal work weeks. We pursued two spe-
cific goals: First, we were interested in the
relationship between work-situation variables, partic-
ularly quantitative workload and psychological de-
tachment. Second, we examined whether psycholog-
ical detachment from work during evening hours
helps in improving well-being after work. Figure 1
displays the conceptual model that we tested in this
study. We propose that the degree of psychological
detachment from work is negatively affected by a
person’s quantitative workload, even when control-
ling for individual differences such as action–state
orientation (Kuhl, 1994b) that might be related to
psychological detachment. In addition, we propose
that psychological detachment from work has an ef-
fect on well-being at bedtime, even after taking into
Figure 1. Conceptual model. Solid lines represent hypothesized relationships and dotted
lines show potential effects of control variables included in the analyses.
394

account workload, time spent on specific off-job ac-
tivities, and well-being when returning home from
work. More specifically, on the basis of well-known
models of affect (Russell, 1980; Watson & Tellegen,
1985), we were interested both in pleasure (positive
mood) and displeasure (negative mood) as aspects of
well-being. With respect to negative mood, we par-
ticularly focused on low arousal (cf. Russell, 1980)
and examined fatigue as an aspect of displeasure that
is highly relevant in the context of job-related distur-
bances of well-being (deCroon, Sluiter, Blonk, Bro-
ersen, & Frings-Dresen, 2004).
We focus on well-being at bedtime as a core out-
come variable for several reasons. First, the question
of how job-related factors affect individuals’ func-
tioning and experiences off the job is an important
topic within occupational health psychology and re-
lated fields (Adams, King, & King, 1996; Hart, 1999;
Rothbard, 2001). Second, impaired well-being does
not disappear from one day to the next but tends to
spill over into subsequent working days (Totterdell,
Spelten, Smith, Barton, & Folkard, 1995). In addi-
tion, an individual’s well-being at bedtime has an
impact on sleep quality and sleep efficiency (Morin,
Rodrigue, & Ivers, 2003), which in turn affects job-
related behavior during the subsequent day (Krueger,
1989).
Because one can assume that the degree of psy-
chological detachment differs substantially from day
to day, we are mainly interested in the predictors and
consequences of day-specific levels of psychological
detachment. Therefore, we examined not only be-
tween-person effects but also within-person varia-
tions of workload, psychological detachment, posi-
tive mood, and fatigue. Other aspects of well-being
that might be also affected by low psychological
detachment, but that develop within a longer time
frame (e.g., health complaints, job satisfaction, life
satisfaction), were beyond the scope of the present
study.
Psychological Detachment Concept
In traditional working arrangements of the past,
most individuals not working at home had the oppor-
tunity to detach themselves from work during off-job
time. They physically left their working place and
went home or to other places where they spent their
evening hours or weekends. However, nowadays,
being physically away from the working place does
not necessarily imply leaving one’s work behind in
psychological terms. In modern work contexts, indi-
viduals have to focus on getting their work done,
regardless of the location at which they complete
their tasks. For example, while at home, individuals
may have to accomplish additional job-related tasks
such as reading reports or preparing material for the
next working day. In addition, when not deliberately
accomplishing job-related tasks, individuals may
continue to think about their jobs, ruminate about
job-related problems, or reflect about future opportu-
nities. When staying psychologically attached to their
jobs during evening hours, individuals may not fully
benefit from their off-job time.
Etzion et al. (1998) introduced the term sense of
detachment for describing “the individual’s sense of
being away from the work situation” (p. 579). De-
tachment implies not being occupied by work-related
duties. For example, when detached from work, an
individual will not receive job-related phone calls at
home and will refrain from job-related activities. In
addition, detachment involves disengaging oneself
psychologically from work. When psychologically
detached from work, one stops thinking of or rumi-
nating about job-related problems or opportunities. In
this article, we use the term psychological detach-
ment to emphasize the psychological component of
disengaging from work during off-job time—as op-
posed to being simply physically absent from the
workplace. Psychological detachment implies a dis-
traction from job-related thoughts. For psychological
detachment to occur, it is not sufficient for one to
change location by leaving the working place; one
must also take a break from thinking about work-
related issues.
Predictors of Psychological Detachment
We propose that job-related factors, particularly
quantitative workload, are related to the inability to
psychologically detach from work during evening
hours. Quantitative workload refers to a high amount
of work and implies that an individual has too much
to do in too little time. Workload is experienced as
time pressure and is often dealt with by working
faster or by working longer hours (Major, Klein, &
Ehrhart, 2002). There is evidence from longitudinal
research that high workload is one of the core job-
related predictors of poor health and well-being (Car-
ayon, 1993; Ganster, Fox, & Dwyer, 2001; Spector,
Chen, & O’Connell, 2000). Moreover, research has
shown that quantitative workload has an impact on
individuals’ functioning within nonwork domains.
For example, after having faced high workload dur-
ing the day, physiological unwinding after work is
prolonged (Frankenhaeuser, 1981; Meijman, Mulder,
395

Van Dormolen, & Cremer, 1992). In addition, high
workload is associated with the experience of work–
family conflict (Geurts & Demerouti, 2003). Thus,
workload is an aspect of the work situation that spills
over into the nonwork domain and continues its in-
fluence on the individual after the end of the working
day. Therefore, it is likely that workload also hinders
psychological detachment from work during evening
hours.
Recently, Geurts, Kompier, Roxburgh, and Hout-
man (2003) argued that high workload negatively
affects health and well-being because it limits the
opportunities for recovery. More specifically, we as-
sume that high workload makes psychological de-
tachment from work less likely. When looking at the
effects of workload on psychological detachment
from a daily perspective, one can differentiate be-
tween chronic and day-specific workload. Chronic
workload refers to the more permanent level of work-
load that is present every day, whereas day-specific
workload refers to the degree of workload present on
the particular day. For example, it might be that in a
given job chronic workload is relatively low but that
from time to time day-specific workload becomes
extremely high.
There are several reasons for the proposed nega-
tive effect of chronic and day-specific workload on
psychological detachment. First, when confronted
with high chronic or day-specific workload, individ-
uals may be more inclined to take some work home
and continue to work on job-related tasks. By defi-
nition, they cannot detach from work while they are
busy with job-related duties. In addition, even when
not actually working they may not detach because
they might anticipate having to do something later in
the evening or because work material is still around
despite the individual having finished the task.
Second, even when not taking work home and not
deliberately accomplishing job-related tasks at home,
it might be difficult to detach psychologically from
work in the evening when facing a high chronic or a
high day-specific workload. Compared with a low-
workload situation, the likelihood of not having fin-
ished all tasks when leaving the workplace in the
evening might be higher when working under a high
workload. Awareness that not all tasks have been
finished might make it difficult to detach from work.
Third, particularly when high workload is not a
single event but continues for longer time, as in the
case of high chronic workload, individuals might also
anticipate a high workload for the next working day.
As a consequence they possibly worry about how to
manage all the tasks that must be accomplished dur-
ing the following day. Therefore, psychological de-
tachment might be difficult. We proposed the
following:
Hypothesis 1: Chronic and day-specific work-
load will be negatively related to psychological
detachment from work during evening hours.
Consequences of Psychological Detachment
We assumed that psychological detachment from
work plays a core role in recovery processes, and
therefore we proposed that it would have a positive
effect on well-being. Recovery can be conceptualized
as a process opposite to the strain process that has
been caused by exposure to stressors. In other words,
recovery eliminates— or at least alleviates—the
mood and performance-related effects of stressors
and restores individual well-being and performance
potential. As a result, an individual’s functioning
returns to its prestressor level (Craig & Cooper, 1992;
Meijman & Mulder, 1998). Recovery from work-
related strain and its associated improvement of
mood requires that work stressors cease to impact on
the individual and that no further demands are put on
the resources that were called upon during the work
process (Craig & Cooper, 1992; Meijman & Mulder,
1998). In simple terms, the probability to recover will
be low if a person continues to pursue job-related
activities during off-job time (Sonnentag, 2001).
Still, the type of activity pursued is not the only factor
relevant for the onset of recovery processes. We
assume that psychological detachment from work is
crucial for recovery to occur. For example, imagine
an individual watching a TV program but at the same
time thinking about an unfinished work task or wor-
rying about how to cope with a high workload antic-
ipated for the next working days to come. In this
situation, the likelihood of recovery will be low be-
cause the resources needed at work are continuously
called upon during off-job time.
Research on mood regulation emphasizes the im-
portance of detachment and distraction for mood
improvement. In everyday situations, individuals re-
gard engagement in pleasant distraction from nega-
tive experiences and from bad mood as a highly
successful strategy for mood regulation (Thayer,
Newman, & McClain, 1994). Totterdell and Parkin-
son (1999) investigated the actual effects of different
mood regulation strategies. They found that partici-
pants’ mood improved after doing something dis-
tracting. Rumination about job-related issues may be
seen as an extreme form of not psychologically de-
396

taching from work in evening hours. Experimental
research has shown that distraction reduces depressed
mood in normally nondepressed persons, whereas
rumination results in an increase of depressed mood
(Morrow & Nolen-Hoeksema, 1990). Moreover, ru-
mination was found to be positively related to other
strain indicators, such as feelings of nervousness and
elevated levels of cortisol (Roger & Najarian, 1998;
Young & Nolen-Hoeksema, 2001). On the basis of
theoretical work about recovery processes and em-
pirical research on mood regulation, we proposed that
psychological detachment from work during evening
hours improves well-being at bedtime:
Hypothesis 2: Psychological detachment from
work during evening hours will be positively
related to well-being at bedtime.
Psychological detachment might not be not equally
important in all situations, though. For example,
when facing a highly stressful work situation during
the day, psychological detachment during the
evening might be particularly needed in order to
recover and restore one’s well-being. When continu-
ing to think about work-related issues after a stressful
working day, the likelihood might be high that one’s
thoughts refer to stressful events. As a consequence,
well-being will suffer. However, after a less stressful
working day, job-related thoughts during the evening
might be less negative. Therefore, the relationship
between psychological detachment and well-being
will be weaker. One might even argue that after a
highly successful and enjoyable working day the
relationship between low psychological detachment
(i.e., continuing to think about the positive events
encountered during the day) and well-being will be
positive. More specifically, with respect to workload
as a core stressor we proposed that the relationship
between lack of psychological detachment and poor
well-being would be strong after a high-workload
day because after such a day strain reactions would
be particularly high (Teuchmann, Totterdell, &
Parker, 1999). To successfully regulate one’s well-
being during the evening of a high-workload day, it is
particularly important to psychologically detach from
work, that is to stop thinking about the very situation
that caused the strain (Totterdell & Parkinson, 1999).
If one continues to think about work during such an
evening, it is likely that thinking about work will be
characterized by highly stressful thoughts that in turn
are negatively related to one’s well-being. In case of
low day-specific workload however, work-related
thoughts during evening hours—if they occur—
might be less stressful and more positive. In such a
situation, refraining from work-related thoughts will
not be so crucial for one’s well-being. We made the
following hypothesis:
Hypothesis 3: Day-specific workload moderates
the relationship between psychological detach-
ment during evening hours and well-being at
bedtime. Psychological detachment will be
more strongly associated with well-being on
days with high day-specific workload than on
days with low day-specific workload.
Control Variables
When analyzing the effects of psychological de-
tachment on well-being one has to take into account
that not only psychological detachment from work
experienced during evening hours but also a range of
other variables might affect well-being at bedtime.
For example, well-being at bedtime might be also
affected by well-being when coming home from
work or from the level of workload experienced
during the work day (Frankenhaeuser, 1981; Meij-
man et al., 1992). Also, off-job time activities pur-
sued during the evening might have an impact on
well-being at bedtime (Sonnentag, 2001). To rule out
these alternative interpretations, we controlled for
workload, time spent on off-job time activities, and
previous well-being when analyzing the effects of
psychological detachment on well-being at bedtime.
With respect to the prediction of psychological
detachment from work, there might be individual
difference variables that account for variance in an
individual’s ability to detach from work. We assume
that action–state orientation (Kuhl, 1994b) is a key
variable when it comes to individual differences in
psychological detachment from work. Action-ori-
ented individuals are able to allocate their attention to
the present situation and the task at hand, whereas
state-oriented individuals tend to ruminate about past
situations and failures. To examine the effects of
quantitative workload above and beyond the effects
of more dispositional factors, we controlled for ac-
tion–state orientation when predicting psychological
detachment from work.
In addition, psychological detachment from work
might depend not only on job-related and individual-
difference factors but also on off-job factors. The
types of activities an individual pursues and the time
he or she spends on them are crucial for psycholog-
ical detachment to occur. For example, spending a
high amount time on job-related activities during
397

evening hours might be negatively related to psycho-
logical detachment. Spending much time on activities
that are not job-related might increase the likelihood
of detaching psychologically from work. Moreover,
individuals facing a high workload might choose
other off-job activities to detach— or to continue
thinking about work—than might individuals facing
less workload. To investigate the effect of workload
on psychological detachment above and beyond the
effects of off-job activities, we controlled for the
types of activities individuals pursued during off-job
time.
Also demographic variables such as age, gender,
and number of children might have an effect on
psychological detachment from work and on well-
being. Therefore, we controlled for age, gender, and
number of children in the analyses.
Method
Sample and Procedure
To recruit participants, we contacted the personnel de-
partments of 10 organizations (7 private organizations and 3
public service organizations) as well as self-employed in-
dividuals. These organizations and self-employed individu-
als produced and delivered a broad range of different goods
and services including computer software, pharmaceutical
products, consultancy services, distribution services, and
public services. We approached these organizations and
self-employed individuals because they were known to offer
or hold demanding jobs. In addition, these organizations
were expected to be interested in employee health and
well-being issues. After the managers of all 10 organiza-
tions and self-employed individuals expressed interest in the
study and agreed that at least one department within the
organization was willing to participate, we directly con-
tacted employees and self-employed individuals through
personal meetings, mostly held within the organizations.
We introduced the study as an investigation on work and
recovery, explained the procedure of the study, and solicited
participation by emphasizing the relevance of the research
topic and by assuring anonymity and confidentiality of all
responses. Through these meetings, we distributed 142 sur-
vey packages to the potential participants. Each survey
package included a letter describing the purpose of the
study, a questionnaire, and daily survey material. All items
were in German.
Of the 142 survey packages distributed, 97 were returned,
representing a response rate of 68%. Ten of these returned
survey packages were not usable because of missing data on
core person-level variables, leaving a total of 87 completed
questionnaires and daily surveys for data analysis (final
response rate: 61%).
The mean age of participants was 41.6 years (SD ⫽ 9.7).
Fifty-three percent of participants in the sample were men,
whereas 47% were women. Participants came from a broad
range of professional backgrounds. Thirty-six percent of the
participants held a supervisory position. Five participants
(5.7%) were self-employed. On average, participants were
highly educated. Fifty-nine percent of the sample had a
university degree, 19% had completed a technical college or
a craftsman’s diploma, and 18% had completed an appren-
ticeship program; only 5% of the sample had no formal
educational degree beyond required school.
The majority of the participants (74.7%) lived with a
partner, whereas 25.3% lived alone. Some participants
(3.4%) lived as single parents with their children or lived
with another person besides a partner or a child. In total,
47% of the participants had no children, 17% had one child,
28% had two children, and 8% had three or more children.
Questionnaire Measures
Our data collection procedure was based on questionnaire
and daily survey measures. Participants had to complete the
questionnaire before they started to fill in the daily survey.
We used the questionnaire to assess chronic workload and
control variables at the person level. An overview of all
questionnaire measures is given in the upper part of Table 1.
Chronic workload. We assessed hours of work and time
pressure as aspects of chronic quantitative workload. Spe-
cifically, participants reported their chronic work hours by
responding to an item in the questionnaire (“How many
hours per week do you actually work for your job?”).
Chronic time pressure was measured with five 5-point Lik-
ert items (1 ⫽ very seldomly/never to 5 ⫽ very often) from
the time pressure scale developed by Semmer (1984) and
Zapf (1993), a measure that is often used in German speak-
ing countries for assessing quantitative workload (Frese,
1985; Garst, Frese, & Molenaar, 2000; Semmer, Zapf, &
Greif, 1996; Cronbach’s
␣
⫽ .79). A sample item was “How
often do you work under time pressure?”
Control variables at the person level. We assessed par-
ticipants’ gender, age, and number of children with single
items. We measured action–state orientation with the 12-
item preoccupation subscale of the Action Control Scale
(ACS–90; Kuhl, 1990, 1994a). A sample item was “If I’ve
worked for four weeks on one project and then everything
goes completely wrong with it: (a) It takes me a long time
to adjust myself to it. (b) It bothers me for a while, but then
I don’t think about it any more,” with (a) indicating a
state-oriented answer and (b) indicating an action-oriented
answer. Cronbach’s alpha was .73. Items were scored such
that a high score represented high action orientation.
Daily Survey Measures
The daily survey had to be completed over a period of 3
consecutive work days. On each day, participants were
asked to respond to items on two measurement occasions:
(a) when returning home from work and (b) before going to
bed. We decided to use self-paced daily survey materi-
als—as opposed to beepers— because participants’ working
and sleeping times differed largely from person to person
and from day to day. To avoid the possibility that the
process of completing the daily survey affected the partic-
ipants’ well-being, we asked participants to complete, on all
days at both measurement occasions, well-being items first
and then the other daily survey items. The lower part of
Table 1 gives an overview over all daily survey measures.
398

Day-specific workload. In addition to chronic workload
assessed with the questionnaire, we measured day-specific
workload with the daily survey on each day at the first
measurement occasion, that is, when returning home from
work. Specifically, participants indicated their day-specific
work hours by responding to the question “How many hours
did you work today?” For measuring day-specific time
pressure, we adapted three items from Semmer’s (1984) and
Zapf’s (1993) time pressure scale. Specifically, we formu-
lated the items in a way that they applied to the situation on
the specific day (e.g., “Today I had to work under time
pressure”). Participants responded to these items on a
5-point Likert scale (1 ⫽ not true at all to 5 ⫽ very true).
We used this subset of three items instead of the full version
for economical reasons. Cronbach’s alphas computed sep-
arately for the 3 days ranged between .89 and .91.
Well-being when returning home from work. We as-
sessed well-being when returning home from work in the
daily survey at the first measurement occasion, immediately
after participants returned home from work. More specifi-
cally, we measured positive mood and fatigue with mea-
sures developed by Nitsch (1976). These measures assess an
individual’s present state and are widely used in German-
speaking countries for measuring mood states in various life
domains. Research has demonstrated that these measures
show good construct validity and sensitivity to mood
changes (Apenburg, 1986; Martin & Weber, 1976; Udris &
Barth, 1976).
Specifically, on each of the 3 days, participants responded
to seven items assessing positive mood (“cheerful,” “good
humored,” “content,” “joyous,” “hilarious,” “carefree,”
“harmonic”) and to five items assessing fatigue (“depleted,”
“tired out,” “in need of recovery,” “rested” (recoded), “re-
created” (recoded)). Participants were instructed to report
with respect to every adjective how they felt “now, at this
moment, after work.” Participants provided their responses
on 6-point Likert scales ranging from 1 ⫽ barely to 6 ⫽
completely. Cronbach’s alphas computed separately for the
3 days ranged between .91 and .94 for positive mood and
between .86 and .89 for fatigue. To examine the distinctive-
ness of our positive mood and fatigue measures, we per-
formed a set of confirmatory factor analyses. For each of the
3 days, we performed separate analyses in which we com-
pared a one-factor model (with all items loading on one
latent factor) with a two-factor model (with all positive
mood items loading on a positive mood factor and with all
fatigue items loading on a fatigue factor). Although model
fit for the two-factor models was not perfect, for all 3 days,
the two-factor model fit the data better than did the one-
factor model, ⌬
␹
2
(1, N ⫽ 81) ⬎ 128.54, p ⬍ .01.
Psychological detachment from work. We assessed
psychological detachment from work on a daily basis with
the daily survey at the second measurement occasion (i.e.,
each day at bedtime). We asked participants to indicate their
level of psychological detachment experienced during five
different types of activities performed on the respective day.
We used this activity-specific psychological detachment
measures to gather more detailed information and because
we assessed that the degree of psychological detachment
might differ from one activity to another. More specifically,
we first gave short descriptions of five activity categories
and a list of prototypical activities within each category.
These activity categories and related descriptions had been
developed in an earlier study (Sonnentag, 2001). The five
activity categories were composed of the following: (a)
work-related activities (e.g., finishing or preparing for work
duties and doing one’s private administration such as com-
pleting one’s tax declaration or paying bills), (b) household
and child-care activities (e.g., cooking, doing the dishes,
shopping, taking care of the children), (c) low-effort activ-
ities (e.g., watching TV, taking a bath), (d) social activities
(e.g., meeting with others, making a phone call in order to
chat), (e) physical activities (e.g., sports, cycling). For each
of these activity categories, participants had to indicate
whether they had performed this type of activity on that day.
If they answered “yes,” they were asked to report the
Table 1
Overview Over Measures
Construct Instrument
Frequency of
assessment Time of assessment
Chronic time pressure Questionnaire Once
Chronic work hours Questionnaire Once
Action-state orientation Questionnaire Once
Demographic variables: gender, age, no. of children Questionnaire Once
Day-specific time pressure Daily survey 3 days Upon return from work
Day-specific work hours Daily survey 3 days Upon return from work
Positive mood when returning home Daily survey 3 days Upon return from work
Fatigue when returning home Daily survey 3 days Upon return from work
Psychological detachment Daily survey 3 days At bedtime
Positive mood at bedtime Daily survey 3 days At bedtime
Fatigue at bedtime Daily survey 3 days At bedtime
Time spent on work-related activities Daily survey 3 days At bedtime
Time spent on household and child-care activities Daily survey 3 days At bedtime
Time spent on low-effort activities Daily survey 3 days At bedtime
Time spent on social activities Daily survey 3 days At bedtime
Time spent on physical activities Daily survey 3 days At bedtime
399

amount of time they had spent on the respective category
and to respond to three psychological detachment items on
a 5-point Likert scale (1 ⫽ not true at all to 5 ⫽ very true):
(a) “While performing this activity, I forgot completely
about my working day”; (b) “While performing this activ-
ity, I could ‘switch off’ completely”; and (c) “While per-
forming this activity, I had to think about my work again
and again” (recoded). If they answered “no” (i.e., if they had
not performed the specific type of activity on the respective
day), they were asked to proceed to the items referring to the
next activity category.
We computed Cronbach’s alphas of this psychological
detachment measure separately for each of the five activity
categories and for each of the 3 days. For work-related
activities, Cronbach’s alphas computed separately for the 3
days ranged between .74 and .88. For household and child-
care activities, Cronbach’s alphas ranged between .72 and
.82. For low-effort activities, Cronbach’s alphas ranged
between .77 and .86. For social activities, Cronbach’s al-
phas ranged between .74 and .91. For physical activities,
Cronbach’s alphas ranged between .83 and .90.
To examine the validity of our psychological detachment
measure, we compared the degree of psychological detach-
ment experienced during the pursuit of work-related activ-
ities with the degree of psychological detachment experi-
enced during the pursuit of other activities. Specifically, for
each of the five activity categories, we averaged the psy-
chological detachment scores across the 3 days. If a partic-
ipant executed an activity on 1 or 2 days only, we based this
average psychological detachment score on the data as-
sessed on these single days. Analysis showed that partici-
pants experienced less psychological detachment during
work-related activities than they did during the other type of
activities: household and child-care activities, t(44) ⫽
⫺4.58, p ⬍ .01; low-effort activities, t(51) ⫽⫺8.01, p ⬍
.01; social activities, t(44) ⫽⫺5.72 p ⬍ .01; and physical
activities, t(20) ⫽⫺5.21, p ⬍ .01 (all t tests with Bonfer-
roni adjustment). It is interesting to note that psychological
detachment was low during work-related activities, al-
though work-related activities comprised also nonjob activ-
ities such as completing one’s tax declaration or paying
bills. One could expect that the psychological detachment
score was even lower when referring only to job-related
activities in this category.
Subsequently, we used the day-specific and activity-spe-
cific psychological detachment scores to compute overall
day-specific psychological detachment measures for each of
the 3 days. Specifically, we averaged the psychological
detachment score provided for the five activity categories
across all activities the person performed on a specific day.
This procedure resulted in one psychological detachment
score for each participant per day. Very few participants
performed all five activities in a single day. Therefore, we
could not compute the usual Cronbach’s alpha coefficient
for these overall day-specific psychological detachment
measures. For an indication of the internal consistency of
the overall day-specific psychological detachment mea-
sures, we correlated the activity-specific psychological de-
tachment measures provided separately for the five activi-
ties with the overall day-specific psychological detachment
measures. For work-related activities, the correlations be-
tween the activity-specific psychological detachment mea-
sure and the overall day-specific psychological detachment
measure ranged between .63 and .86. For household and
child-care activities, the correlations ranged between .64
and .84. For low-effort activities, the correlations ranged
between .78 and .83. For social activities, the correlations
ranged between .85 and .88. For physical activities, they
ranged between .60 and .82. Overall, these correlations
correspond to good item-total correlations.
Well-being at bedtime. We assessed well-being at bed-
time in the daily survey at the second measurement occasion
immediately before the participants went to bed. On each of
the 3 days, participants were asked to respond to seven
6-point Likert items measuring positive mood and to five
6-point Likert items measuring fatigue. Specifically, partic-
ipants were asked to indicate how they felt “now, at this
moment, before going to bed.” Items were identical to those
administered for measuring well-being when returning
home from work. Cronbach’s alphas computed separately
for the 3 days ranged between .93 and .95 for positive mood
and between .76 and .83 for fatigue. Again, we conducted a
set of confirmatory factor analyses in which we compared a
one-factor model with a two-factor model. These analyses
showed that for all 3 days, model fit was better for the
two-factor model than for the one-factor model, ⌬
␹
2
(1, N ⫽
81) ⬎ 65.96, p ⬍ .01.
One might argue that well-being at bedtime is a rather
irrelevant outcome measure because individuals go to sleep
anyway. To investigate whether well-being at bedtime mat-
ters beyond the specific moment we examined its effect on
well-being during subsequent work days. We assessed well-
being on Day 3 with seven items from the irritation–strain
measure developed by Mohr (1986; cf. Frese, 1999; Cron-
bach’s
␣
⫽ .88). Sample item are “Upon returning home
from work, I was rather irritated” and “I got angry easily.”
Participants were instructed to answer the irritation–strain
items with respect to how they felt during the past 2 days
(i.e., Day 2 and Day 3). Multiple regression analysis showed
that positive mood and fatigue at bedtime of Day 1 ex-
plained 17% of the variance in irritation–strain, F(2, 84) ⫽
8.661, p ⬍ .01. This result suggests that well-being at
bedtime is a relevant predictor of well-being experienced
during the subsequent work days.
Control variables at the day level. For control variables
at the day level, we assessed the amount of time participants
spent on the five types of off-job activities. In the daily
survey at the second measurement occasion (i.e., immedi-
ately before going to bed), participants reported how much
time (in minutes) they had spent on work-related activities,
household and child-care activities, low-effort activities,
social activities, and physical activities.
Data Analysis Procedure
For each participant, we gathered data at the person level
(e.g., chronic workload, demographic variables) and at the
day level (e.g., day-specific workload, psychological de-
tachment, well-being). Day-level data were nested within
the person-level data. To analyze our data we used multi-
level analysis, also known as hierarchical linear modeling
(Bryk & Raudenbush, 1992). Such a multilevel approach is
most appropriate for analyzing hierarchically structured
data sets such as the one gathered in the present study. It is
superior to ordinary least square regression analysis because
it does not require independence of observations but allows
for dependent observations within the higher level data
400

structure (Snijders & Bosker, 1999). With respect to the
present study, multilevel analysis takes the dependence of
day-level data within each person into account.
We used the MLn program for data analysis (Rasbash &
Woodhouse, 1996). In our analyses, predictor variables at
the person level (chronic workload, demographic variables)
were Level 2 data, and predictor variables at the day level
(day-specific workload, off-job time activities, psychologi-
cal detachment, positive mood and fatigue when returning
home) were Level 1 data. We centered person-level predic-
tor variables around the grand mean and day-level predictor
variables around the respective person mean. For testing our
hypotheses, we followed a hierarchical test procedure and
compared a set of nested models. In all sets of analyses we
started with a null model that included only the intercept. In
the subsequent steps we consecutively added the predictor
variables of interest. We tested the improvement of each
model with a likelihood ratio statistic. For more information
about a similar approach to data analysis, see Sonnentag’s
(2001) study.
Results
Table 2 shows means, standard deviations, and
zero-order correlations between the study variables.
Person-level and day-level correlations are displayed.
To correlate variables measured at the day level with
variables assessed at the person level, we averaged
day-level measures across the 3 days. Day-level vari-
ables were not centered before correlating them with
other day-level and person-level variables. With re-
spect to person-level correlations, the person was the
level of analysis (N ⫽ 87); with respect to day-level
correlations, the day was the level of analysis (k ⫽
221).
Because employees from various industry types
participated in the study, we examined whether in-
dustry type was related to the core outcome variables.
Specifically, we used the International Standard In-
dustrial Classification of All Economic Activities
(ISIC Revision 3.1; United Nations, 2002) for cate-
gorizing the participating organizations into seven
industry types (manufacturing; retail; real estate,
renting, and business activities; public administration
and defense; education; health; and other community,
social, and personal service activities) and included
these categories as dummy variables in the analysis.
Analysis revealed no significant main effects of in-
dustry types on well-being measures or psychological
detachment. There were also no significant interac-
tion effects between industry type and psychological
detachment on well-being and no significant interac-
tion effects between industry type and workload on
psychological detachment. Therefore, we did not dif-
ferentiate between various industry types in further
analyses.
Similarly, because our sample included five self-
employed individuals and because employment sta-
tus may be related to the core outcome variables, we
included employment status as a dummy variable in
our analysis (0 ⫽employed,1⫽self-employed).
There were no significant main effects of employ-
ment status on well-being measures or psychological
detachment. Neither the interaction effects between
employment status and workload measures on psy-
chological detachment nor the interaction effects be-
tween employment status and psychological detach-
ment on well-being were significant. There was also
no evidence for any three-way interaction effect in-
cluding employment status. Therefore, in further
analyses we did not differentiate between employed
and self-employed participants.
Test of Hypotheses
In Hypothesis 1, we predicted a negative effect of
quantitative workload on psychological detachment
from work. We used multilevel modeling for testing
this hypothesis and compared nested models. Specif-
ically, we compared three nested multilevel models:
a null model, Model 1, and Model 2. In the null
model, the intercept was the only predictor. In Model
1, we entered demographic variables (gender, age,
number of children; Level 2), action–state orientation
(Level 2), and time spent on off-job time activities
(Level 1) as control variables. In Model 2, we added
chronic workload (Level 2) and day-specific work-
load (Level 1) as the core predictor variables of
interest. We tested the improvement of each model
above the previous one by computing the difference
between the respective likelihood ratios. This differ-
ence follows a chi-square distribution (with degrees
of freedom equal to the number of new parameters
added to the model). Results are displayed in Table 3.
Analysis showed that Model 1, in which we en-
tered the control variables, did not show an improve-
ment over the null model (difference of ⫺2 ⫻ log ⫽
13.105, df ⫽ 9, ns). Model 2, which also included
workload measures, fit the data better than did Model
1 (difference of ⫺2 ⫻ log ⫽ 26.569, df ⫽ 4, p ⬍
.01). High chronic time pressure and long day-spe-
cific work hours had negative effects on psycholog-
ical detachment from work. Also, action–state orien-
tation was significantly related to psychological
detachment, with action-oriented individuals show-
ing higher psychological detachment scores. Taken
together, Hypothesis 1 was supported for chronic
time pressure and day-specific work hours as predic-
tor variables.
401

Table 2
Means, Standard Deviations, and Correlations
Variable MSD123456789101112131415161718
1. Gender
a
1.53 0.50 —
2. Age 41.60 9.70 ⫺.10 —
3. Number of children 0.99 1.11 .20 .39 —
4. Action-state orientation 6.43 2.90 .23 ⫺.03 ⫺.08 —
5. Chronic work hours 43.64 11.75 .40 ⫺.10 .00 .25 —
6. Chronic time pressure 3.16 0.78 .01 .05 .15 .18 .44 —
7. Day-specific work
hours 8.62 1.90 .35 ⫺.11 ⫺.02 .25 .84 .47 — .21 .01 ⫺.33 ⫺.14 ⫺.09 ⫺.10 ⫺.22 ⫺.05 ⫺.12 .21 .15
8. Day-specific time
pressure 2.61 0.90 ⫺.16 .00 ⫺.10 .09 .07 .39 .13 — .07 ⫺.11 ⫺.09 ⫺.02 ⫺.10 ⫺.19 ⫺.28 ⫺.07 .47 .12
9. Time spent on work-
related activities 30.98 45.58 ⫺.16 ⫺.03 ⫺.06 ⫺.04 ⫺.08 .09 .03 ⫺.04 — .01 ⫺.03 ⫺.14 ⫺.11 ⫺.27 ⫺.05 ⫺.12 .10 .09
10. Time spent on
household and child-
care activities 44.49 54.01 ⫺.26 .24 .19 ⫺.09 ⫺.38 ⫺.12 ⫺.41 ⫺.10 .05 — .02 ⫺.11 ⫺.04 ⫺.03 ⫺.05 ⫺.03 ⫺.01 .05
11. Time spent on low-
effort activities 78.84 61.83 .10 ⫺.12 ⫺.20 .05 ⫺.13 ⫺.20 ⫺.14 ⫺.13 .01 .03 — ⫺.19 ⫺.21 .07 .22 .19 ⫺.13 ⫺.15
12. Time spent on social
activities 73.82 80.18 ⫺.00 ⫺.07 .09 .08 ⫺.21 ⫺.10 ⫺.08 .01 ⫺.19 ⫺.06 ⫺.11 — ⫺.02 .12 .03 .15 ⫺.02 .00
13. Time spent on physical
activities 12.40 19.56 .04 ⫺.01 ⫺.04 .09 ⫺.08 ⫺.05 ⫺.14 ⫺.15 ⫺.08 ⫺.01 ⫺.14 .06 — .15 .05 .13 ⫺.10 ⫺.13
14. Psychological
detachment 3.72 0.69 .16 ⫺.10 .06 .21 ⫺.16 ⫺.31 ⫺.25 ⫺.19 ⫺.33 .02 .10 .14 .23 — .40 .50 ⫺.43 ⫺.38
15. Positive mood when
returning home 3.81 0.86 .23 ⫺.23 .04 .25 .05 ⫺.06 ⫺.00 ⫺.17 ⫺.12 ⫺.04 .26 .07 .21 .50 — .74 ⫺.58 ⫺.41
16. Positive mood at
bedtime 3.86 0.91 .19 ⫺.27 .00 .29 ⫺.03 ⫺.01 ⫺.08 .02 ⫺.09 ⫺.02 .23 .23 .20 .54 .86 — ⫺.36 ⫺.41
17. Fatigue when returning
home 3.42 0.92 ⫺.15 ⫺.09 ⫺.04 ⫺.14 .05 .28 .15 .43 .20 ⫺.00 ⫺.17 .02 ⫺.27 ⫺.53 ⫺.58 ⫺.41 — .66
18. Fatigue at bedtime 3.48 0.87 ⫺.15 ⫺.04 .08 ⫺.11 .10 .12 .15 .11 .13 .01 ⫺.20 .03 ⫺.20 ⫺.41 ⫺.46 ⫺.45 .74 —
Note. Below the diagonal: correlations at the person level (N ⫽ 87). All correlations ⱖ .21 are significant at p ⬍ .05; all correlations ⱖ .27 are significant at p ⬍ .01. Above
the diagonal: correlations at the day level (k ⫽ 221). All correlations ⱖ .13 are significant at p ⬍ .05; all correlations ⱖ .17 are significant at p ⬍ .01.
a
1 ⫽ female, 2 ⫽ male.
402

In Hypothesis 2, we proposed that psychological
detachment has a positive effect on well-being at
bedtime. Hypothesis 3 stated a moderator effect of
day-specific workload on the relationship between
psychological detachment and well-being at bedtime.
Positive mood at bedtime and fatigue at bedtime were
the dependent variables in these analyses. For each of
these two dependent variables, we compared six
nested multilevel models: a null model, Model 1,
Model 2, Model 3, Model 4, and Model 5. In the null
model the intercept was the only predictor. In Model
1, we entered demographic variables (Level 2) as
control variables. In Model 2, we entered chronic
(Level 2) and day-specific (Level 1) workload mea-
sures and well-being when coming home from work
(Level 1). More specifically, to predict positive mood
at bedtime, we entered positive mood when returning
home from work as a control variable; to predict
fatigue at bedtime, we entered fatigue when coming
home from work as a control variable. In Model 3,
we entered time spent on the five off-job activities as
predictors (Level 1). In Model 4, we entered psycho-
logical detachment from work (Level 1) as the core
predictor variable of interest in testing Hypothesis 2.
A significant improvement of Model 4 over Model 3
would indicate support for Hypothesis 2. In Model 5,
we entered the interaction terms between day-specific
workload (Level 1) and psychological detachment
(Level 1). A significant improvement of Model 5 over
Model 4 would indicate support for Hypothesis 3.
Table 4 summarizes the findings for positive mood
at bedtime as the dependent variable. Model 1, which
included demographic variables as control variables,
showed a significant improvement over the null
model (difference of ⫺2 ⫻ log ⫽ 8.732, df ⫽ 3, p ⬍
.05). Specifically, age was a significant predictor with
older persons showing less positive mood at bedtime.
In Model 2, we entered workload measures and pos-
itive mood when returning home from work. This
model showed improvement over Model 1 (differ-
Table 3
Multilevel Estimates for Models Predicting Psychological Detachment From Quantitative Workload
Variable
Null model
a
Model 1
b
Model 2
c
Estimate SE t Estimate SE t Estimate SE t
Fixed effects
Intercept 3.7340 0.07363 50.713 3.7280 0.0701 53.212 3.3650 0.2385 14.109
Gender 0.0682 0.1514 0.451 0.1104 0.1561 0.707
Age ⫺0.0086 0.0081 ⫺1.062 ⫺0.0097 0.0075 ⫺1.259
No. of children 0.0893 0.0729 1.225 0.1179 0.0691 1.705
Action-state orientation 0.0591 0.0253 2.337* 0.0802 0.0240 3.341**
Time spent on work-related
activities ⫺0.0009 0.0010 ⫺0.979 ⫺0.0010 0.0009 ⫺1.027
Time spent on household
and child-care activities ⫺0.0002 0.0008 ⫺0.208 ⫺0.0004 0.0007 ⫺0.616
Time spent on low-effort
activities 0.0007 0.0008 0.793 ⫺0.0001 0.0008 ⫺0.147
Time spent on social
activities 0.0009 0.0006 1.489 0.0007 0.0006 1.169
Time spent on physical
activities 0.0020 0.0016 1.303 0.0015 0.0015 1.002
Chronic work hours ⫺0.0081 0.0071 ⫺1.146
Chronic time pressure ⫺0.2643 0.0979 ⫺2.699**
Day-specific work hours ⫺0.0920 0.0350 ⫺2.631**
Day-specific time pressure ⫺0.0854 0.0580 ⫺1.472
Intercept
variance SE
Intercept
variance SE
Intercept
variance SE
Random effects
Level 1 0.2813 0.0344 0.2723 0.0334 0.2474 0.0302
Level 2 0.3569 0.0727 0.3154 0.0656 0.2634 0.0564
Note. N ⫽ 87. k ⫽ 221 days.
a
⫺2 ⫻ log(lh) ⫽ 471.043.
b
⫺2 ⫻ log(lh) ⫽ 457.938, Diff ⫺2 ⫻ log ⫽ 13.105, df ⫽ 9.
c
⫺2 ⫻ log(lh) ⫽ 431.369,
Diff ⫺2 ⫻ log ⫽ 26.569***, df ⫽ 4.
* p ⬍ .05. ** p ⬍ .01. *** p ⬍ .001.
403

ence of ⫺2 ⫻ log ⫽ 22.871, df ⫽ 5, p ⬍ .01).
Positive mood when returning home was a strong
predictor of positive mood at bedtime. In Model 3,
we entered time spent on off-job activities as addi-
tional predictors. The improvement of Model 3 over
Model 2 was nonsignificant. However, inspection of
the estimates of the several off-job activities revealed
time spent on physical activities as a significant pre-
dictor of positive mood at bedtime. In Model 4, we
added psychological detachment to the model. Model
Table 4
Multilevel Estimates for Models Predicting Positive Mood From Leisure Time Activities and
Psychological Detachment
Variable
Null model
a
Model 1
b
Model 2
c
Estimate SE t Estimate SE t Estimate SE t
Fixed effects
Intercept 3.8590 0.1002 38.513 3.5040 0.3183 11.008 3.3300 0.3404 9.783
Gender 0.2333 0.1986 1.175 0.3798 0.2202 1.725
Age ⫺0.0275 0.0110 ⫺2.510* ⫺0.0277 0.0108 ⫺2.561*
Number of children 0.0785 0.0978 0.802 0.0568 0.0984 0.577
Positive mood when
returning home 0.2644 0.0775 3.412***
Chronic work hours ⫺0.0152 0.0102 ⫺1.487
Chronic time pressure 0.0583 0.1400 0.417
Day-specific work
hours ⫺0.0381 0.0315 1.212
Day-specific time
pressure ⫺0.0737 0.0558 1.312
Time spent on work-
related activities
Time spent on
household and child-
care activities
Time spent on low-
effort activities
Time spent on social
activities
Time spent on physical
activities
Psychological
detachment
Day-Specific Work
Hours ⫻
Psychological
Detachment
Day-Specific Time
Pressure ⫻
Psychological
Detachment
Intercept
variance SE
Intercept
variance SE
Intercept
variance SE
Random effects
Level 1 0.2434 0.0298 0.2438 0.0299 0.2096 0.0257
Level 2 0.7741 0.1327 0.6887 0.1197 0.6795 0.1158
Note. N ⫽ 87. k ⫽ 221 days.
a
⫺2 ⫻ log(lh) ⫽ 505.237.
b
⫺2 ⫻ log(lh) ⫽ 496.506, Diff ⫺2 ⫻ log ⫽ 8.731*, df ⫽ 3.
c
⫺2 ⫻ log(lh) ⫽ 473.635,
Diff ⫺2 ⫻ log ⫽ 22.871***, df ⫽ 5.
d
⫺2 ⫻ log(lh) ⫽ 464.262, Diff ⫺2 ⫻ log ⫽ 9.373, df ⫽ 5.
e
⫺2 ⫻ log(lh) ⫽
448.667, Diff ⫺2 ⫻ log ⫽ 15.595***, df ⫽ 1.
f
⫺2 ⫻ log(lh) ⫽ 445.329, Diff ⫺2 ⫻ log ⫽ 3.338, df ⫽ 2.
* p ⬍ .05. *** p ⬍ .001.
404

4 showed a significant improvement over Model 3
(difference of ⫺2 ⫻ log ⫽ 15.595, df ⫽ 1, p ⬍ .01).
The estimate of psychological detachment was highly
significant, indicating that psychological detachment
contributed significantly to the prediction of positive
mood at bedtime— beyond the effects of positive
mood when returning home from work and beyond
the effect of time spent on specific off-job activities.
In Model 5, we entered the interaction terms between
day-specific workload and psychological detach-
Model 3
d
Model 4
e
Model 5
f
Estimate SE t Estimate SE t Estimate SE t
3.3050 0.3425 9.647 3.2630 0.3405 9.583
3.3310 0.3430 9.711 0.3895 0.2215 1.758 0.4094 0.2198 1.861
0.3789 0.2218 1.708 ⫺0.0279 0.0109 ⫺2.561* ⫺0.0275 0.0108 ⫺2.546*
⫺0.0273 0.0109 ⫺2.500* 0.0550 0.0990 0.555 0.0502 0.0981 0.512
0.0555 0.0992 0.560 0.2652 0.0766 3.463*** 0.2660 0.0768 3.465***
0.3365 0.0789 4.267*** ⫺0.0160 0.0103 ⫺1.556 ⫺0.0164 0.0102 ⫺1.607
⫺0.0155 0.0102 ⫺1.512 0.0524 0.1408 0.372 0.0517 0.1395 0.371
0.0570 0.1410 0.404 ⫺0.0060 0.0303 ⫺0.199 ⫺0.0029 0.0306 ⫺0.094
⫺0.0341 0.0313 ⫺1.090 ⫺0.0345 0.0532 ⫺0.649 ⫺0.0229 0.0533 ⫺0.430
⫺0.0454 0.0562 ⫺0.808 ⫺0.0010 0.0008 ⫺1.204 ⫺0.0011 0.0008 1.287
⫺0.0014 0.0009 ⫺1.592 0.0004 0.0006 0.583 0.0003 0.0006 0.524
0.0002 0.0006 0.323 0.0003 0.0007 0.449 0.0004 0.0007 0.551
0.0004 0.0007 0.514 0.0005 0.0005 0.982 0.0005 0.0005 0.947
0.0008 0.0005 1.474 0.0026 0.0013 2.010* 0.0026 0.0013 2.043*
0.0032 0.0014 2.364* 0.3028 0.0746 4.062*** 0.3318 0.0760 4.372***
⫺0.0014 0.0963 ⫺0.014
⫺0.2662 0.1637 ⫺1.626
Intercept
variance SE
Intercept
variance SE
Intercept
variance SE
0.1936 0.0237 0.1726 0.0211 0.1704 0.0208
0.6973 0.1181 0.7040 0.1177 0.6905 0.1156
405

ment. This model did not result in any improvement
of model fit (difference of ⫺2 ⫻ log ⫽ 3.338, df ⫽
2, ns). Taken together, for positive mood as depen-
dent variable, our data show support for Hypothesis 2
but not for Hypothesis 3.
Table 5 shows the results for fatigue at bedtime as
the dependent variable. Model 1, which included
demographic variables as predictors, showed no sig-
nificant improvement over the null model. In Model
2, we added chronic and day-specific workload and
Table 5
Multilevel Estimates for Models Predicting Fatigue From Leisure Time Activities and
Psychological Detachment
Variable
Null Model
a
Model 1
b
Model 2
c
Estimate SE t Estimate SE t Estimate SE t
Fixed effects
Intercept 3.4800 0.09387 37.0725 3.9790 0.3074 12.944 4.1750 0.3294 12.675
Gender ⫺0.3267 0.1916 ⫺1.705 ⫺0.4282 0.2131 ⫺2.009*
Age ⫺0.0096 0.0106 ⫺0.907 ⫺0.0103 0.0105 ⫺0.982
Number of children 0.1275 0.0947 1.346 0.1415 0.0953 1.484
Fatigue when returning
home 0.4939 0.0815 6.059***
Chronic work hours 0.0125 0.0099 1.261
Chronic time pressure 0.0587 0.1353 0.434
Day-specific work
hours 0.0060 0.0433 0.138
Day-specific time
pressure ⫺0.1120 0.0743 ⫺1.508
Time spent on work-
related activities
Time spent on
household and child-
care activities
Time spent on low-
effort activities
Time spent on social
activities
Time spent on physical
activities
Psychological
detachment
Day-Specific Work
Hours ⫻
Psychological
Detachment
Day-Specific Time
Pressure ⫻
Psychological
Detachment
Intercept
variance SE
Intercept
variance SE
Intercept
variance SE
Random effects
Level 1 0.4888 0.0597 0.4881 0.0596 0.3700 0.0452
Level 2 0.5673 0.1184 0.5368 0.1137 0.5639 0.1096
Note. N ⫽ 87. k ⫽ 221 days.
a
⫺2 ⫻ log(lh) ⫽ 587.368.
b
⫺2 ⫻ log(lh) ⫽ 583.602, Diff ⫺2 ⫻ log ⫽ 3.766, df ⫽ 3.
c
⫺2 ⫻ log(lh) ⫽ 543.970, Diff
⫺2 ⫻ log ⫽ 39.632***, df ⫽ 5.
d
⫺2 ⫻ log(lh) ⫽ 539.081, Diff ⫺2 ⫻ log ⫽ 4.889, df ⫽ 5.
e
⫺2 ⫻ log(lh) ⫽ 534.560,
Diff ⫺2 ⫻ log ⫽ 4.521*, df ⫽ 1.
f
⫺2 ⫻ log(lh) ⫽ 528.478, Diff ⫺2 ⫻ log ⫽ 6.082*, df ⫽ 2.
* p ⬍ .05. *** p ⬍ .001.
406

fatigue when returning home as additional predictors.
Model 2 showed a highly significant improvement
over Model 1 (difference of ⫺2 ⫻ log ⫽ 39.632,
df ⫽ 5, p ⬍ .01). The estimate of fatigue when
returning home from work was highly significant. In
this model, the estimate of gender was also signifi-
cant, with women showing more fatigue at bedtime.
In Model 3, we entered variables referring to time
spent on off-job activities. Model fit did not improve
(difference of ⫺2 ⫻ log ⫽ 4.889, df ⫽ 5, ns),
Model 3
d
Model 4
e
Model 5
f
Estimate SE t Estimate SE t Estimate SE t
4.2100 0.3305 12.738 4.1500 0.3369 12.318
4.1870 0.3292 12.719 ⫺0.4441 0.2137 ⫺2.078* ⫺0.4164 ⫺0.2172 ⫺1.917
⫺0.4333 0.2129 ⫺2.035* ⫺0.0099 0.0105 ⫺0.941 ⫺0.0092 ⫺0.0107 ⫺0.861
⫺0.0104 0.0105 ⫺0.990 0.1456 0.0956 1.523 0.1371 0.0970 1.414
0.1448 0.0953 1.520 0.4699 0.0815 5.769*** 0.4305 0.0807 5.335***
0.5028 0.0814 6.174*** 0.0128 0.0099 1.287 0.0128 0.0101 1.262
0.0122 0.0099 1.243 0.0654 0.1357 0.482 0.0624 0.1377 0.437
0.0628 0.1352 0.465 ⫺0.0153 0.0436 ⫺0.352 ⫺0.0087 ⫺0.0428 ⫺0.203
⫺0.0012 0.0439 ⫺0.028 ⫺0.1314 0.0740 ⫺1.776 ⫺0.0996 ⫺0.0730 ⫺1.365
⫺0.1207 0.0752 ⫺1.604 ⫺0.0002 0.0011 ⫺0.201 ⫺0.0003 ⫺0.0011 ⫺0.280
⫺0.0000 0.0012 ⫺0.003 0.0008 0.0009 0.927 0.0008 0.0008 0.919
0.0009 0.0009 1.045 ⫺0.0015 0.0010 ⫺1.532 ⫺0.0012 ⫺0.0010 ⫺1.254
⫺0.0016 0.0010 ⫺1.591 0.0001 0.0007 0.206 0.0002 0.0007 0.281
⫺0.0000 0.0007 ⫺0.009 ⫺0.0019 0.0018 ⫺1.026 ⫺0.0016 ⫺0.0018 ⫺0.929
⫺0.0023 0.0018 ⫺1.244 ⫺0.2223 0.1034 ⫺2.150* ⫺0.1755 ⫺0.1023 ⫺1.716
0.0717 0.1266 0.566
⫺0.5349 ⫺0.2183 ⫺2.450*
Intercept
variance SE
Intercept
variance SE
Intercept
variance SE
0.3573 0.0437 0.3438 0.0420 0.3227 0.0394
0.5675 0.1093 0.5781 0.1101 0.6071 0.1131
407

indicating that time spent on the several off-job ac-
tivities had no effect on fatigue at bedtime. In Model
4, we added psychological detachment from work
into the model. Model 4 fit the data better than did
Model 3 (difference of ⫺2 ⫻ log ⫽ 4.521, df ⫽ 1,
p ⬍ .05). The estimate of psychological detachment
from work was significant and showed a negative
sign. Thus, the data supported Hypothesis 2 for fa-
tigue as the dependent variable as well. Finally, in
Model 5 we entered the interaction terms between
day-specific workload and psychological detach-
ment. Analysis showed an improved model fit (dif-
ference of ⫺2 ⫻ log ⫽ 6.082, df ⫽ 2, p ⬍ .05), with
a significant interaction term between day-specific
time pressure and psychological detachment.
To explore the pattern of this interaction effect
further, we followed the procedure proposed by
Aiken and West (1991) and divided our sample
into two subgroups with low versus high time
pressure. Because our moderator variable was a
day-level variable, the units in our subgroups were
days and not persons. Specifically, the low time
pressure subgroup comprised all days with time
pressure below the person-specific mean; the high
time pressure subgroups comprised all days with
time pressure above the person-specific mean. We
performed separate multilevel analyses with sim-
ple slope tests for these two subgroups, with fa-
tigue as the dependent variable and psychological
detachment as the predictor variable. For days with
low time pressure, the model that included psycho-
logical detachment did not show a better model fit
than the model without psychological detachment
(difference of ⫺2 ⫻ log ⫽ 0.292, df ⫽ 1, ns). The
slope of psychological detachment was not signif-
icant (
␥
⫽⫺0.1053, SE ⫽ 0.1933, t ⫽ 0.54, ns).
For days with high time pressure, the model that
included psychological detachment fit the data bet-
ter than the model without psychological detach-
ment (difference of ⫺2 ⫻ log ⫽ 6.058, df ⫽ 1, p ⬍
.05). The slope of psychological detachment
yielded a negative sign, indicating that low psy-
chological detachment was associated with a high
level of fatigue (
␥
⫽⫺0.4927, SE ⫽ 0.1700, t ⫽
⫺2.89, p ⬍ .01). Figure 2 shows that on days with
low time pressure, psychological detachment was
not related to fatigue. However, on days with high
time pressure, psychological detachment was neg-
atively related to fatigue. Taken together, Hypoth-
esis 3 was supported for fatigue as the dependent
variable.
Discussion
The results of this study point to a paradoxical
situation. Previous research has shown that individ-
uals who work long hours report a higher need for
recovery after work (Sluiter, Van der Beek, & Frings-
Dresen, 1999). The results of the present study show
that when confronted with high workload, individu-
als are less successful at detaching themselves psy-
chologically from work. Thus, when workload is
high, need for recovery increases— but at the same
time this high need for recovery is less likely to be
satisfied. In addition, our study shows that psycho-
logical detachment from work is positively associ-
ated with positive mood and low fatigue at bedtime.
Moreover, after working days characterized by high
time pressure, the effect of psychological detachment
on fatigue was particularly strong.
Predictors of Detachment
Two of our four workload measures were found to
be negatively related to psychological detachment.
Thus, with increasing chronic time pressure and long
day-specific work hours it is less likely that an indi-
vidual detaches psychologically from work during
Figure 2. Interaction effect of time pressure and psycho-
logical detachment on fatigue.
408

evening hours. The finding that chronic but not day-
specific time pressure was associated with low psy-
chological detachment suggests that it is not primar-
ily the amount of time pressure that one has faced
during the past working day that makes psychologi-
cal detachment difficult but rather the anticipation
that time pressure will continue during the working
days to come. With respect to work hours, day-
specific work hours seem to be more relevant for
psychological detachment than do chronic work
hours. One may speculate that— up to a certain ex-
tent—individuals might adjust to long chronic work
hours and organize their spare free time in a way such
that they can benefit from it also when returning
home from work late. However, individuals might be
less prepared for long day-specific work hours and
therefore might be less likely to psychologically de-
tach from work during evening hours.
Our results parallel findings from studies that ex-
amined how individuals unwind from work. Franken-
haeuser (1981) and Meijman et al. (1992) reported
that physiological stress indicators stayed elevated
for a longer time and that it took employees longer to
unwind physiologically after stressful working days
than after less stressful days. In a recent study on
overtime, a similar pattern of findings emerged, par-
ticularly for women holding poorly designed jobs
(Rau & Triemer, 2004). It might be that a high level
of physiological arousal impedes psychological de-
tachment from work and that, in turn, negatively
affects well-being. An alternative interpretation
would be that the inability to detach oneself from
work hinders the unwinding process; that is, physio-
logical arousal remains high because one continues to
think about the past working day or anticipates the
next day’s workload. Here, more research is clearly
needed.
Although the degree of psychological detachment
differed across off-job time activities, time spent on
specific off-job activities had no effect on overall
psychological detachment. It might be that the cate-
gories of off-job activities used in this study were too
broad. For example, the question of how much time
one spends on household or low-effort activities as a
whole might not be as relevant as some more specific
types of household or low-effort activities that foster
psychological detachment and others that hinder it. In
addition, some types of off-job activities might in-
clude deliberate preoccupation with job-related
thoughts (e.g., when meeting friends and talking with
them about work). In addition, it might be that there
is not a set of off-job activities that increases psycho-
logical detachment in all persons but that different
activities are useful for different persons. For exam-
ple, one person might best detach by doing sports,
whereas another detaches when concentrating on
household chores.
Our findings imply that just adding “free hours” in
the late evening does not help in detaching psycho-
logically from work. It might be that after having
worked long hours, particularly under time pressure,
one is perhaps already too fatigued to benefit from
free hours and off-job activities. One might speculate
that psychological detachment requires self-regula-
tion, for example, a deliberate effort to stop job-
related thoughts. Research has shown that self-regu-
lation suffers under fatigue (Baumeister, Bratslavsky,
Muraven, & Tice, 1998). Thus, psychological detach-
ment would be impaired in a situation in which
self-regulation is required but not fully available be-
cause of preceding workload.
Psychological Detachment as a Predictor of
Positive Mood and Low Fatigue
Analyses showed that psychological detachment
from work during evening hours was positively re-
lated to positive mood and negatively related to fa-
tigue at bedtime. These effects existed beyond the
strong effects of positive mood and fatigue when
returning home from work. Our findings suggest that
it is not the mere physical distance to one’s work-
place that is important for recovery after work; rather,
psychological detachment seems to be crucial. Our
study adds to Etzion et al.’s (1998) research on de-
tachment during longer respites by showing that
short-term psychological detachment during normal
work weeks plays a role in experiencing positive
mood and low fatigue.
Analysis further revealed an interaction effect
between day-specific time pressure and psycholog-
ical detachment on fatigue. After working days
characterized with high time pressure, psycholog-
ical detachment from work during evening hours
showed a strong negative relationship with fatigue
at bedtime. After working days with low time
pressure, psychological detachment was not related
to fatigue. This finding implies that psychological
detachment is particularly important after high
time pressure days. If one is not able to psycho-
logically detach from work after such working
days, fatigue increases substantially.
The finding that day-specific time pressure mod-
erated the relationship between psychological detach-
ment and fatigue but not the relationship between
409

psychological detachment and positive mood war-
rants attention. This pattern of findings suggests that
thinking about work after high time pressure days is
fatiguing because job-related thoughts might refer to
a stress-relevant content, which further draws on
one’s resources. After days with low time pressure,
job-related thoughts might not necessarily imply a
stressful content and, therefore, they do not foster the
fatigue process. Nevertheless, job-related thoughts
during the evening after low time pressure days are
also negatively related to positive mood. It might be
that thinking about the job during evening hours does
not meet individuals’ expectations about a free
evening and therefore individuals may regard low
detachment as a negative experience that affects their
mood. In addition, it might be that low detachment
also negatively impacts the quality of social interac-
tions during the evening and, as a consequence, pos-
itive mood suffers.
Unlike day-specific time pressure, day-specific
long work hours did not moderate the relationship
between psychological detachment and fatigue. Thus,
although day-specific long work hours were nega-
tively associated with psychological detachment, low
detachment after long work days was not more det-
rimental to one’s well-being than it was after shorter
work days. It might be that although thinking about
work in the evening is less frequent after shorter
work days, it is not necessarily less stressful—as
opposed to thinking about work after relatively re-
laxed, low time pressure days. Therefore, short work-
ing hours do not attenuate the relationship between
psychological detachment and fatigue.
Our study showed that low psychological detach-
ment from work during evening hours was related to
poor well-being at bedtime. However, it must not be
taken for granted that low detachment is always
negatively related to well-being. In addition to the
interaction effect found in the present study, one
might argue that there are even days when low de-
tachment is positively related to well-being. For ex-
ample, after having successfully accomplished a ma-
jor project or having received highly positive
feedback from one’s supervisor or an important cli-
ent, continuous thinking about work may boost one’s
positive mood. Thus, the content of work-related
thought might be crucial. A recent study on recovery
during weekends showed that employees who re-
flected about their work in a positive way during the
weekend experienced less exhaustion and less disen-
gagement after returning to work (Fritz & Sonnentag,
2005). In addition, when one is faced with a highly
stressful situation at home, high psychological de-
tachment from work may impede one’s well-being,
because work-related thoughts might help to stabilize
well-being.
Furthermore, one might speculate that there are
also individual differences in the degree to which
persons experience job-related thoughts during off-
job time as stressful. It might be that individuals high
in job involvement even enjoy thinking about their
work during off-job time. For these individuals, the
relationship between psychological detachment dur-
ing evening hours and well-being will be weaker than
for individuals low in job involvement. Thus, future
research on psychological detachment should pay
more attention to situational factors and individual
difference variables.
One might argue that positive mood and fatigue at
bedtime are influenced by an individual’s level of
dispositional affect and not by day-specific psycho-
logical detachment. Although we did not control for
dispositional affect in our analyses, we believe that it
is unlikely that our findings can be primarily ac-
counted for by dispositional affect. When predicting
positive mood and fatigue at bedtime we included the
respective affective measures assessed after work in
the analyses. Dispositional affect should be already
reflected in these after-work measures and therefore
should not be a core explanatory variable for the
relationship between psychological detachment and
positive mood and fatigue at bedtime.
Other Predictors of Positive Mood
and Fatigue
This study revealed a number of additional inter-
esting findings. Analyses showed that men were less
fatigued at bedtime than were women. This result
corresponds to the observation that women react
more negatively to stressful situations (Kivima¨ki et
al., 1997). This finding is also in line with a total-
workload interpretation claiming that women face a
higher total workload, including household and
child-care duties, which makes it more difficult to
recover during evening hours (Mardberg, Lundberg,
& Frankenhaeuser, 1991). In the present study
women spent more time on household and child-care
activities, t(85) ⫽ 2.43, p ⬍ .05; however, time spent
on household activities did not predict fatigue. Thus,
it is not the mere amount of time women spent on
household and child-care activities that contributes to
an increase in fatigue. However, it could be that
women spend more time on the more demanding
types of household tasks or experience a higher re-
410

sponsibility for household and child-care activities
that increase fatigue (cf. Barnett & Gareis, 2002).
Time spent on physical activities was positively
associated with positive mood at bedtime. This find-
ing is in line with a large body of research on the
positive effect of sports on well-being (Byrne &
Byrne, 1993). It is interesting to note that this effect
refers not only to differences between persons but
also to variations within persons; that is, individuals
feel better on days in which they spent more time on
sports than they do on average. Despite the beneficial
effect of physical activities on positive mood, phys-
ical activities did not decrease fatigue. This finding
might be explained by the fact that physical activities
are physically fatiguing. Maybe potentially positive
effects on mental fatigue were counteracted by an
increase in physical fatigue. Future studies should
differentiate between mental and physical fatigue
when examining the effects of physical activities.
Limitations
Some limitations of this study should be noted.
First, the study relied on self-report data, and there-
fore a common method bias cannot be ruled out
completely. However, we gathered data with two
different instruments (questionnaire and daily sur-
vey) and assessed our study variables at different
levels of specificity and detail. For example, we
measured positive mood and fatigue with a set of
rather global items, whereas we assessed psycholog-
ical detachment at the level of each specific activity
and later averaged these specific scores into an over-
all psychological detachment measure. In addition,
work hours are rather objective measures of work-
load and are not easily influenced by response bias
(Frese & Zapf, 1988). Nevertheless, it is desirable
that future research uses measures other than pure
self-report data when examining the relationships
between work overload, psychological detachment,
and well-being. Asking participants’ spouses about
their partners’ psychological detachment would be
one option.
Second, although we provided detailed instruc-
tions about when to complete the daily survey, we
cannot be completely sure that our participants al-
ways complied with the instructions and answered all
items at the requested points in time. However, be-
cause there were data missing at points when some
participants skipped some measurement points, we
are rather confident that participants did not fake their
survey completion times. Nevertheless, to overcome
this problem in future studies, researchers could use
handheld computers for data collection that allow for
the recording of time stamps when specific items are
completed.
Third, our participants were highly educated and
on average held intellectually demanding jobs. Thus,
it remains an open question for future research
whether the relationship between high workload and
low detachment is specific for individuals in intellec-
tually demanding jobs or whether it generalizes be-
yond these work settings. Nevertheless, participants
came from different organizations and had diverse
professional backgrounds. Therefore, our findings are
not limited to one specific organization or profession
but probably generalize to a broader range of orga-
nizations and professions.
Implications for Research and Practice
Our study has some implications for future re-
search and practice. In this study, we examined short-
term effects of psychological detachment that be-
came evident within a few hours. Future research
should address longer term outcomes of psychologi-
cal detachment during evening hours. For example,
one might hypothesize that psychological detachment
from work during evening not only improves well-
being on the same day but helps to protect individual
health and well-being in the long run. Longitudinal
studies over longer time periods are needed. In addi-
tion, it would be interesting to investigate whether
psychological detachment from work during evening
hours has an effect on job-related behavior, particu-
larly job performance, during the next day.
This study indicates that high workload is detri-
mental for psychological detachment. Although de-
sirable, workload cannot always be reduced. There-
fore, it is an important question for future research to
examine how individuals can detach psychologically
from their work, particularly during periods of high
workload. Eden (2001) pointed out that self-efficacy
might play an important role when predicting suc-
cessful recovery. One can speculate that recovery
self-efficacy, that is, one’s belief in the ability to
recover from work during evening hours, is a crucial
factor in order for one to fully benefit from off-job
experiences.
With respect to practical implications, our findings
underscore the importance of psychological detach-
ment for improving positive mood and reducing fa-
tigue during evening hours. Therefore, individuals
should be encouraged to switch off mentally from
work when leaving their working place and arriving
at home. Potentially helpful strategies might include
411

deliberate relaxation exercises (Carlson & Holye,
1993) or the initiation of flow experiences during
leisure time activities (Csikszentmihalyi & LeFevre,
1989). The most important starting point for any
intervention, however, is probably the reduction of
workload. It would be particularly relevant to avoid
chronic time pressure and to refrain from long day-
specific working days. Such a reduction of workload
will not only decrease an individual’s need for recov-
ery (Sluiter et al., 1999) but also have a positive
impact on the recovery process by making psycho-
logical detachment from work more likely. Overall,
psychological detachment from work during evening
hours may be one important factor that contributes to
individuals’ work–life balance.
References
Adams, G. A., King, L. A., & King, D. W. (1996). Rela-
tionships of job and family involvement, family social
support, and work–family conflict with job and life sat-
isfaction. Journal of Applied Psychology, 81, 411– 420.
Aiken, L. S., & West, S. G. (1991). Multiple regression:
Testing and interpreting interactions. Newbury Park,
CA: Sage.
Apenburg, E. (1986). Befindlichkeitsbeschreibung als
Methode der Beanspruchungsmessung [State description
as a method for measuring stress]. Psychologie und
Praxis, 30, 3–14.
Barnett, R. C., & Gareis, K. C. (2002). Full-time and re-
duced-hours work schedules and marital quality: A study
of female physicians with young children. Work & Oc-
cupations, 29, 364 –379.
Baumeister, R. F., Bratslavsky, E., Muraven, M., & Tice,
D. M. (1998). Ego depletion: Is the active self a limited
resource? Journal of Personality and Social Psychology,
74, 1252–1265.
Bryk, A. S., & Raudenbush, S. W. (1992). Hierarchical
linear models: Application and data analysis methods.
Newbury Park, CA: Sage.
Byrne, A., & Byrne, D. G. (1993). The effect of exercise on
depression, anxiety and other mood states: A review.
Journal of Psychosomatic Research, 37, 565–574.
Carayon, P. (1993). A longitudinal test of Karasek’s job
strain model among office workers. Work and Stress, 7,
299 –314.
Carlson, C. R., & Holye, R. H. (1993). Efficacy of abbre-
viated progressive muscle relaxation training: A quanti-
tative review of behavioral medicine research. Journal of
Consulting and Clinical Psychology, 61, 1059 –1067.
Craig, A., & Cooper, R. E. (1992). Symptoms of acute and
chronic fatigue. In A. P. Smith & D. M. Jones (Eds.),
Handbook of human performance (Vol. 3, pp. 289 –339).
London: Academic Press.
Csikszentmihalyi, M., & LeFevre, J. (1989). Optimal expe-
rience in work and leisure. Journal of Personality and
Social Psychology, 56, 815– 822.
deCroon, E. M., Sluiter, J. K., Blonk, R. W. B., Broersen,
J. P. J., & Frings-Dresen, M. H. W. (2004). Stressful
work, psychological job strain, and turnover: A 2-year
prospective cohort study of truck drivers. Journal of
Applied Psychology, 89, 442– 454.
Eden, D. (2001). Vacations and other respites: Studying
stress on and off the job. In C. L. Cooper & I. T.
Robertson (Eds.), International review of industrial and
organizational psychology (pp. 121–146). Chichester,
England: Wiley.
Elfering, A., Grebner, S., Semmer, N. K., & Gerber, H.
(2002). Time control, catecholamines and back pain
among young nurses. Scandinavian Journal of Work
Environment and Health, 28, 386 –393.
Etzion, D., Eden, D., & Lapidot, Y. (1998). Relief from job
stressors and burnout: Reserve service as a respite. Jour-
nal of Applied Psychology, 83, 577–585.
Frankenhaeuser, M. (1981). Coping with stress at work.
International Journal of Health Services, 11, 491–510.
Frese, M. (1985). Stress at work and psychosomatic com-
plaints: A causal interpretation. Journal of Applied Psy-
chology, 70, 314 –328.
Frese, M. (1999). Social support as a moderator of the
relationship between work stressors and psychological
dysfunctioning: A longitudinal study with objective mea-
sures. Journal of Occupational Health Psychology, 4,
179 –192.
Frese, M., & Zapf, D. (1988). Methodological issues in the
study of work stress: Objective vs. subjective measure-
ment and the question of longitudinal studies. In C. L.
Cooper & R. Payne (Eds.), Causes, coping, and conse-
quences of stress at work (pp. 375– 411). New York:
Wiley.
Fritz, C., & Sonnentag, S. (in press). Recovery, well-being,
and performance-related outcomes: The role of work
load and specific vacation experiences. Journal of Ap-
plied Psychology.
Fritz, C., & Sonnentag, S. (2005). Recovery, health, and job
performance: Effects of weekend experiences. Journal of
Occupational Health Psychology, 3, 187–199.
Ganster, D. C., Fox, M. L., & Dwyer, D. J. (2001). Explain-
ing employees’ health care costs: A prospective exami-
nation of stressful job demands, personal control, and
physiological reactivity. Journal of Applied Psychology,
86, 954 –964.
Ganster, D. C., & Schaubroeck, J. (1991). Work stress and
employee health. Journal of Management, 17, 235–271.
Garst, H., Frese, M., & Molenaar, P. C. M. (2000). The
temporal factor of change in stressor–strain relationships:
A growth curve model on a longitudinal study in East
Germany. Journal of Applied Psychology, 85, 417– 438.
Geurts, S. A. E., & Demerouti, E. (2003). Work/non-work
interface: A review of theories and findings. In M. J.
Schabraqc, J. A. M. Winnubst, & C. L. Cooper (Eds.),
The handbook of work and health psychology (pp. 279–
312). Chichester, United Kingdom: Wiley.
Geurts, S. A. E., Kompier, M. A. J., Roxburgh, S., &
Houtman, I. L. D. (2003). Does work– home interference
mediate the relationship between workload and well-
being? A multi-sample analysis. Journal of Vocational
Behavior, 63, 532–559.
Hart, P. M. (1999). Predicting employee life satisfaction: A
coherent model of personality, work, and nonwork expe-
riences, and domain satisfactions. Journal of Applied
Psychology, 84, 564 –584.
Kahn, R. L., & Byosiere, P. (1992). Stress in organizations.
In M. D. Dunnette & L. M. Hough (Eds.), Handbook of
412

industrial and organizational psychology (2nd ed., Vol.
3, pp. 571– 650). Palo Alto, CA: Consulting Psycholo-
gists Press.
Kivima¨ki, M., Vahtera, J., Thomson, L., Griffiths, A., Cox,
T., & Pentti, J. (1997). Psychosocial factors predicting
employee sickness absence during economic decline.
Journal of Applied Psychology, 82, 858 – 872.
Krueger, G. P. (1989). Sustained work, fatigue, sleep loss
and performance: A review of the issues. Work & Stress,
3, 129 –141.
Kuhl, J. (1990). Der Fragebogen zur Erfassung von Han-
dlungskontrolle nach Erfolg, Mi
␤
erfolg und prospektiv
HAKEMP 90 [Questionnaire for assessing action control
after success, failure, and prospectively HAKEMP 90].
Osnabru¨ck, Germany: University of Osnabru¨ck.
Kuhl, J. (1994a). Action versus state orientation: Psycho-
metric properties of the action control scale (ACS–90). In
J. Kuhl & J. Beckmann (Eds.), Volition and personality:
Action versus state orientation (pp. 47–59). Seattle, WA:
Hogrefe & Huber.
Kuhl, J. (1994b). A theory of action and state orientations.
In J. Kuhl & J. Beckmann (Eds.), Volition and person-
ality: Action versus state orientation (pp. 9 – 46). Seattle,
WA: Hogrefe & Huber.
Lounsbury, J. W., & Hoopes, L. L. (1986). A vacation from
work: Changes in work and nonwork outcomes. Journal
of Applied Psychology, 71, 392– 401.
Lundberg, U., & Lindfors, P. (2002). Psychophysiological
reactions to telework in female and male white-collar
workers. Journal of Occupational Health Psychology, 7,
354 –364.
Major, V. S., Klein, K. J., & Ehrhart, M. G. (2002). Work
time, work interference with family, and psychological
distress. Journal of Applied Psychology, 87, 427– 436.
Mardberg, B., Lundberg, U., & Frankenhaeuser, M. (1991).
The total workload of parents employed in white-collar
jobs: Construction of a questionnaire and a scoring sys-
tem. Scandinavian Journal of Psychology, 32, 233–239.
Martin, E., & Weber, A. (1976). Wirkungen einto¨nig-repeti-
tiver Ta¨tigkeiten auf das subjektive Befinden und die
Flimmerverschmelzungs-Frequenz [The effects of mo-
notonous repetitive activities on subjective condition and
on the critical fusion frequency]. Zeitschrift fu¨r Arbe-
itswissenschaft, 30, 183–187.
McEwen, B. S. (1998). Protective and damaging effects of
stress mediators. New England Journal of Medicine, 338,
171–179.
Meijman, T. F., & Mulder, G. (1998). Psychological aspects
of workload. In P. J. D. Drenth & H. Thierry (Eds.),
Handbook of work and organizational psychology: Vol.
2. Work psychology (pp. 5–33). Hove, England: Psychol-
ogy Press.
Meijman, T. F., Mulder, G., Van Dormolen, M., & Cremer,
R. (1992). Workload of driving examiners: A psycho-
physiological field study. In H. Kragt (Ed.), Enhancing
industrial performances (pp. 245–260). London: Taylor
& Francis.
Mohr, G. (1986). Die Erfassung psychischer Befindensbee-
intra¨chtigungen bei Arbeitern [Assessment of psycholog-
ical stress effects in industrial workers]. Frankfurt/M.,
Germany: Lang.
Morin, C. M., Rodrigue, S., & Ivers, H. (2003). Role of
stress, arousal, and coping skills in primary insomnia.
Psychosomatic Medicine, 65, 259 –267.
Morrow, J., & Nolen-Hoeksema, S. (1990). Effects of re-
sponse to depression on the remediation of depressive
affect. Journal of Personality and Social Psychology, 58,
519 –527.
Nitsch, J. R. (1976). Die Eigenzustandsskala (EZ-Skala)–
Ein Verfahren zur hierarchisch-mehrdimensionalen Be-
findlichkeitsskalierung [Own state scale—An instrument
for hierarchical-multidimensional state scaling]. In J. R.
Nitsch & I. Udris (Eds.), Beanspruchung im Sport. Bad
Homburg, Germany: Limpert.
Rasbash, J., & Woodhouse, G. (1996). MLn. London: Uni-
versity of London, Institute of Education.
Rau, R., & Triemer, A. (2004). Overtime work as related to
blood pressure and mood during work, leisure and night
time. Social Indicators Research, 67, 51–73.
Roger, D., & Najarian, B. (1998). The relationship between
emotional rumination and cortisol secretion under stress.
Personality and Individual Differences, 24, 531–538.
Rothbard, N. P. (2001). Enriching or depleting? The dy-
namics of engagement in work and family roles. Admin-
istrative Science Quarterly, 46, 655– 684.
Russell, J. A. (1980). A circumplex model of affect. Journal
of Personality and Social Psychology, 39, 1161–1178.
Semmer, N. (1984). Stre
␤
bezogene Ta¨tigkeitsanalyse
[Stress-related job analysis]. Weinheim, Germany: Beltz.
Semmer, N., Zapf, D., & Greif, S. (1996). “Shared job
strain”: A new approach for assessing the validity of job
stress measurements. Journal of Occupational and Or-
ganizational Psychology, 69, 293–310.
Sluiter, J. K., Van der Beek, A. J., & Frings-Dresen,
M. H. W. (1999). The influence of work characteristics
on the need for recovery and experienced health: A study
on coach drivers. Ergonomics, 42, 573–583.
Snijders, T. A. B., & Bosker, R. J. (1999). Multilevel
analysis. An introduction to basic and advanced multi-
level modeling. London: Sage.
Sonnentag, S. (2001). Work, recovery activities, and indi-
vidual well-being: A diary study. Journal of Occupa-
tional Health Psychology, 6, 196 –210.
Sonnentag, S. (2003). Recovery, work engagement, and
proactive behavior: A new look at the interface between
nonwork and work. Journal of Applied Psychology, 88,
518 –528.
Sonnentag, S., & Frese, M. (2003). Stress in organizations.
In W. C. Borman, D. R. Ilgen, & R. J. Klimoski (Eds.),
Comprehensive handbook of psychology: Vol. 12. Indus-
trial and organizational psychology (pp. 453– 491). New
York: Wiley.
Sparks, K., Cooper, C., Fried, Y., & Shirom, A. (1997). The
effects of hours of work on health: A meta-analytic
review. Journal of Occupational and Organizational
Psychology, 70, 391– 408.
Spector, P. E., Chen, P. Y., & O’Connell, B. J. (2000). A
longitudinal study of relations between job stressors and
job strains while controlling for prior negative affectivity
and strains. Journal of Applied Psychology, 85, 211–218.
Teuchmann, K., Totterdell, P., & Parker, S. K. (1999).
Rushed, unhappy, and drained: An experience sampling
study of relations between time pressure, perceived con-
trol, mood, and emotional exhaustion in a group of ac-
countants. Journal of Occupational Health Psychology,
4, 37–54.
Thayer, R. E., Newman, J. R., & McClain, T. M. (1994).
Self-regulation of mood: Strategies for changing a bad
413

mood, raising energy, and reducing tension. Journal of
Personality and Social Psychology, 67, 910–925.
Totterdell, P., & Parkinson, B. (1999). Use and effective-
ness of self-regulation strategies for improving mood in
a group of trainee teachers. Journal of Occupational
Health Psychology, 4, 219 –232.
Totterdell, P., Spelten, E., Smith, L., Barton, J., & Folkard,
S. (1995). Recovery from work shifts: How long does it
take? Journal of Applied Psychology, 80, 43–57.
Udris, I., & Barth, H. (1976). Beanspruchungsaspekte der
Arbeitsorganisation im Bu¨robereich [Workload aspects
of work organization in offices]. Zeitschrift fu¨r Arbe-
itswissenschaft, 30, 87–91.
United Nations. (2002). International Standard Industrial
Classification of All Economic Activities. (ISIC Rev. 3.1).
Retrieved July 30, 2003, from http://unstats.un.org/unsd/
cr/registry/regcst.asp?Cl⫽17
Van der Doef, M., & Maes, S. (1999). The job demand-
control (-support) model and psychological well-being:
A review of 20 years of empirical research. Work &
Stress, 13, 87–114.
Watson, D., & Tellegen, A. (1985). Toward a consensual
structure of mood. Psychological Bulletin, 98,
219 –235.
Westman, M., & Eden, D. (1997). Effects of a respite from
work on burnout: Vacation relief and fade-out. Journal of
Applied Psychology, 82, 516 –527.
Westman, M., & Etzion, D. (2001). The impact of vacation
and job stress on burnout and absenteeism. Psychology
and Health, 16, 595– 606.
Young, E. A., & Nolen-Hoeksema, S. (2001). Effect of
rumination on the salvia cortisol response to a social
stressor. Psychoneuroenocrinology, 26, 319 –329.
Zapf, D. (1993). Stress-oriented analysis of computerized
office work. European Work and Organizational Psy-
chologist, 3, 85–100.
414