Applied Ergonomics 42 (2011) 710e718

Contents lists available at ScienceDirect

Applied Ergonomics
journal homepage: www.elsevier.com/locate/apergo

Job stressors in naval personnel serving on ships and in personnel serving

ashore over a twelve month period
R.S. Bridger*, K. Brasher, A. Dew, S. Kilminster
Institute of Naval Medicine, Crescent Rd, Alverstoke, PO12 2DL, UK

a r t i c l e i n f o a b s t r a c t

Article history: Sixty one percent of respondents to a questionnaire survey of occupational stress (Phase I) returned
Received 4 August 2009 follow-up questionnaires twelve months later (Phase II). The Phase I questionnaire measured psycho-
Accepted 17 November 2010 logical strain resulting from exposure to occupational stressors, and measured the presence of stress
buffers and demographic, psychological and lifestyle-related confounding factors, including age, rank
Keywords: and gender, mood state and the occurrence of stressful life events outside of work. The prevalence of
Occupational stress
strain was 31% at Phase I and 33% at Phase II. Fifty percent of personnel had no strain on either occasion,
Psychological strain
15% had strain on both occasions and the remainder had strain on one occasion. The main stressor
Work environment
Maritime ergonomics
associated with strain at Phase I was an inability to disengage from work and this stressor accounted for
a greater proportion of the variance in strain in personnel serving on ships than those serving ashore. The
twelve-month follow-up questionnaire (Phase II) re-assessed psychological strain. A multiple linear
regression analysis was conducted to identify factors measured at Phase I that predicted strain at Phase II
in previously strain-free individuals. A lack of autonomy and control and dissatisfaction with living
conditions predicted strain twelve months later in those serving on ships. Of the living conditions
assessed, lack of privacy was the most strongly associated with strain twelve months later in those
serving on ships. These stressors were not associated with strain twelve months later in those serving
ashore. The findings suggest that improvements to the design of the environment on ships may have
benefits for the psychological health of personnel.
Crown Copyright Ó 2010 Published by Elsevier Ltd. All rights reserved.

1. Introduction role in the organisation and career development; relationships at

The prevalence of psychological strain, defined as an adverse
reaction to work demands, in Naval Service (NS) personnel is high
compared to other occupational groups (Slaven, 2002; Kilminster
and Bridger, 2007). At any time, approximately 30% of personnel
are found to be experiencing strain, as measured by the number of
symptoms of anxiety and depression measured using the General
Health Questionnaire (GHQ-12, Goldberg and Williams, 1988). This
prevalence rate of strain is similar to that found in other uniformed
Service occupations, such as the UK Police Force (Collins and Gibbs,
2003) but higher than that found in the general UK population and
in Civil Servants (Dollard et al., 2003; Marmot and Brunner, 2004).
Cooper and Marshall (1976) introduced the person environment
fit concept to relate physical and psychological characteristics of
people and exposure to work environment stressors. They identi-
fied key areas of investigation: the work demands themselves; the
* Corresponding author. Tel.: þ44 2392768220; fax: þ44 2392504823.
E-mail address: Bob.Bridger300@mod.uk (R.S. Bridger).
work; organisational structure and climate. Cross-sectional studies
of occupational stress in Naval personnel have quantified the
affective responses of personnel to different aspects of Service life
(Slaven, 2002; Kilminster and Bridger, 2007; Bridger et al. 2008). In
all of these surveys, involving stratified samples of over 2000
people, a reliable and valid ‘Work and Well-Being Questionnaire’
(WWBQ) has been used, which was developed using standard
psychometric techniques. The 30% prevalence rate of strain has
been replicated in three questionnaire surveys over eight years and
a picture has emerged of how the different aspects of working life
in the Service are perceived. For example, support from leaders has
been rated positively, on average. Workefamily conflict has been
rated negatively on average, particularly by married males. The
absence of the positive (support) and the presence of the negative
(workefamily conflict) have both been found to correlate with high
scores on the GHQ-12 questionnaire, indicating strain. Some
aspects of Service life measured by the WWBQ do not correlate
with strain at all and might best be described as sources
of dissatisfaction that personnel are able to cope with. Lack of
resources (including personnel shortages) is an example. The use of

0003-6870/$ e see front matter Crown Copyright Ó 2010 Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.apergo.2010.11.005
 R.S. Bridger et al. / Applied Ergonomics 42 (2011) 710e718
the GHQ-12 in these studies enables the affective responses of
personnel to different aspects of Service life to be assessed in
relation to their consequences for the well-being of individuals.
The link between the work environment, psychological strain
and the development of physical and medical ailments is well-
established. For example, Salonen et al. (2008) found that pro-
longed strain predicted musculoskeletal disorders, nervous system,
eye, ear and metabolic diseases and mental disorders in Finnish
factory workers over a twenty year period. The Whitehall II study
showed that an individual’s position within the hierarchy of an
organisation was a much better predictor of heart attack than
obesity, smoking and high blood pressure (Marmot et al., 1991).
A lack of control at work has also been linked to ill-health (Bosma
et al., 1998) which supports the predictions of the well-known
‘DemandeControleSupport’ (DCS) model of Karasek. The link
between strain and ill-health may be direct, because of the meta-
bolic effects of strain (see Miller et al., 2007, for a review) or indi-
rect. Schei (1994), for example, found that strain was associated
with physical inactivity and the consumption of tobacco, ‘junk food’
and alcohol in Norwegian military conscripts.
1.1. Theories and models
Karasek’s model (Karasek and Theorell, 1990) predicts job strain
when personnel are exposed to occupational stressors such as high
job demands and a lack of control over work. Over-commitment, in
Siegrist’s (1996) Effort Reward Imbalance (ERI) model of occupa-
tional stress regards over-commitment of work as a maladaptive
coping strategy e a pattern of excessive striving which leads to ill-
health. In the short term, over-commitment reflects motivation. In
the long term, over-commitment can only be maintained if it is
reinforced by personal development, recognition and goal
attainment.
Schaufelli and Baker (2004) discuss the DemandeResources
model, which, although developed as a model of burnout, proposes
a mechanism by which strain may lead to adverse outcomes. High
job demands deplete energy resources, resulting in withdrawal in
order to conserve resources, and job performance drops as a result.
Attempts to maintain performance (the ‘strain coping mode’) are
characterised by over-activation of the sympathetic nervous system
and this is the mechanism by which stressors in the work envi-
ronment get ‘into the body’. Long-term over-activation of the
hypothalamicepituitaryeadrenocortical axis appears to be the
mechanism and it can have long-term health consequences (Miller
et al., 2007). In the DemandeResources model, autonomy and
support from leaders and peers are resources that support active
coping with work demands, thus lowering strain. Although
burnout is rare in the NS, it is thought to be the end point of
a process of energy depletion so we would expect job strain to
precede burnout (Demerouti et al., 2001) and some would argue
that ‘strain’ as used in the present paper is a milder equivalent of
the concept of ‘emotional exhaustion’ used in burnout research.
Thus, although explicit burnout is almost unknown in Naval
personnel, there is a conceptual link between these models dis-
cussed above and the processes currently under investigation in the
present research.
The present paper is part of a longitudinal study in which
a cohort of approximately 2500 NS personnel is being followed for
five years. The findings of Phase I of this study (2007) have already
been reported by Bridger et al. (2008). The present paper deals with
a twelve-month follow-up of this study and focuses on the expla-
nation of current and future strain in personnel serving ashore and
personnel serving onboard ships. The deployment system used by
the NS is outlined below, to explain why these two groups of
personnel need to be looked at separately.
711
1.2. Naval deployment patterns
Naval personnel can serve in a shore-based establishment or
onboard a ship. When on ship, they are usually appointed for
a period of two years and the ship is their main place of work, be it
at sea or alongside. During the appointment they can be deployed
at sea for periods of six months or more without returning home.
When at sea, personnel work long hours, usually with no time off at
weekends e essentially, they work an extended form of shift-work,
known as ‘watches’. Shore-based personnel in the NS may work as
hard or even harder than personnel at sea but they can leave the
work environment, live in their own homes and practice favourite
recreations and hobbies in their spare time. Personnel are guar-
anteed to work no more than 660 days away from their homeport
in any rolling three-year period, providing at least 465 days ashore
or alongside. Once home, they may find themselves working or
training at a shore-base for a while before the next deployment.
Appointments at sea are often interspersed with shore postings of
similar length. Thus, working life in the NS differs from working life
in other occupations because the place and conditions of work and
even job content change regularly over the course of a career. Cross-
sectional studies of NS personnel are of limited value because many
of the stressors identified may be transient and unlikely to be
linked to long-term health or employee turnover outcomes.
The above suggests that repeated measures of strain are needed
if strain and its consequences are to be properly understood.
Further, the use of repeated measures may solve some of the
problems of cross-sectional research designs. For example, if
occupational stressors measured at a particular time predict future
strain in personnel whose role is unchanged, then these might be
considered to be more important than transient mood states or
factors that only influence strain experienced concurrently. State
variables such as the GHQ-12 might be expected to vary over time
for a wide variety of reasons such as mood swings and events
outside of work. It is clear from the writings of Schaufelli and Baker
(2004) for example, that demand resource imbalances in the
workplace will only lead to adverse outcomes if they are sustained
and that burnout is the end point of a process rather than an event.
Thus, self-reported ratings of work demands that predict GHQ-12
scores over several time points will be more significant than those
that only predict GHQ-12 score measured concurrently.
Consequently, as a first step towards improving the explanatory
power of previous cross-sectional NS research, the current study
investigated stressor exposure, concurrent strain, and strain twelve
months into the future in a cohort of personnel across the NS. The
purpose was to determine whether exposure to occupational
stressors of the kind described in the models predict future strain in
personnel serving on ships and in personnel serving in shore-based
establishments. Identification of occupational stressors causing
future strain has clear advantages for the improvement of in-
Service stress management.
In particular, the study enabled the temporal stability of strain to
be better understood and it enabled the predictive validity of
workplace stressors to be assessed. The null hypothesis that
workplace stressors do not predict future strain in personnel
serving on ship or ashore was tested.
2. Method
2.1. Work and Well-Being Questionnaire (WWBQ)
The WWBQ was used to measure occupational stressors in the
Naval sample at Phase I of this longitudinal survey. The factors
measured by this questionnaire are shown in Table 1.
712 R.S. Bridger et al. / Applied Ergonomics 42 (2011) 710e718

Table 1
Summary of WWBQ subscale content and reliability coefficients.

WWBQ subscale Parent scale Number Example item Response format Cronbach’s
of items alpha
Financial effort reward ERI, Siegrist (1996) 1 How true is this statement: Considering 5-point Likert scale N/A
all my efforts, my salary is adequate
ERI ERI, Siegrist (1996) 4 How true is this statement: I receive 5-point Likert scale 0.44
the respect I deserve from my colleagues
Over-commitment ERI, Siegrist (1996) 3 How true is this statement: Work 4-point Likert scale 0.82
is still on my mind when I go to bed
Role conflict N/A 4 How true is this statement: I work 5-point Likert scale 0.63
on unnecessary tasks
Workefamily conflict N/A 4 How satisfied are you with the amount 5-point Likert scale 0.87
of separation from your spouse/family
Physical work environment N/A 4 Please indicate the extent to which you 5-point Likert scale 0.87
are satisfied with ease of cleaning in
the physical work environment
Physical living environment N/A 9 Please indicate the extent to which you 5-point Likert scale 0.93
are satisfied with privacy in the physical
living environment
Autonomy and control N/A 5 How much can you choose what work 5-point Likert scale 0.91
you will carry out in your job
Adequacy of resources N/A 3 How true is this statement: There is a 5-point Likert scale 0.77
shortage of trained staff in this unit
Leader support N/A 3 How friendly and easy to approach is 5-point Likert scale 0.93
your supervisor
Leader approachability N/A 2 To what extent would you approach your 5-point Likert scale N/A
superior for work-related problems
Peer support N/A 3 To what extent can you count on your 5-point Likert scale 0.88
colleagues to back you up at work
Organisational commitment N/A 6 How true is this statement: I feel I am 5-point Likert scale 0.79
part of the RN
Coping strategy Coping styles 4 To what extent would you talk to others 5-point Likert scale 0.79
inventory about work-related problems
Positive mood PANAS, 3 Indicate how determined you normally feel 5-point Likert scale 0.70
Watson et al., 1988
Negative mood PANAS, 3 Indicate how guilty you normally feel 5-point Likert scale 0.76
Watson et al., 1988

The WWBQ was developed following the conduct of focus
groups of five to eight personnel of equivalent rank in three Naval
establishments (Slaven, 2002). A pilot questionnaire was then
circulated to 600 randomly selected personnel. Factor analysis of
the returns was conducted to identify items that loaded discretely
onto different factors. The item set was reduced by calculating
alpha coefficients for each of the scales and removing items until
the value of alpha exceeded 0.7. Additional scales were added in
2006 (Kilminster and Bridger, 2007) and psychometric reduction
and validation was carried out using a sample of 28 naval
personnel. The purpose of the psychometric reduction was to
confirm the factor structure of the original WWBQ and to reduce
the length of the new WWBQ by extracting 3e5 items from parent
scales of ERI, over-commitment and mood. The new scales
measured: effort reward imbalance; over-commitment; positive
and negative mood (from the PANAS scale of Watson et al., 1988);
Stressful Life Events (SLE, Holmes and Rahe, 1967), and coping style.
A five-day testeretest reliability study was conducted and unreli-
able items were removed. Most of the items in all scales of the final
WWBQ were highly reliable with Spearman correlation coefficients
personnel (10% of males and 60% of females in the NS, the latter to
ensure adequate sub-sample size for analysis) who were randomly
selected and were stratified by branch. Seasonal effects on dispo-
sition were controlled for by conducting the survey at the same
time of year as the previous cross-sectional surveys conducted in
1999 and 2004 (Slaven, 2002; Bridger et al., 2007). The question-
naire was sent in three postings, with only non-respondents being
targeted in subsequent postings. Demographic information was
collected from the NS personnel database and linked to an anon-
ymous respondent ID code. The initial and subsequent question-
naires were marked with the ID code so that WWBQ responses
could be compared within subjects over time. Participants were
informed that all their answers would be treated as ‘medical in
confidence’, and that while the project workers could identify them
Table 2
WWBQ health and demographic measures.
Other health and demographic Response format
WWBQ measures

over 0.7. Psychometric revalidation of the WWBQ was carried out in Job specialisation
Job preference area
the present study and is reported in Table 1.
Physically demanding work
The new WWBQ measured occupational stressor exposure, Alcohol intake
exposure to stressful life events and main confounding factors and BMI
effect modifiers. Psychological strain was measured using the GHQ- Smoking
12. Health and demographic measures are summarised in Table 2. Exercise participation
Health complaints
Self-assessed general health
2.2. Study design Sea-sickness
Absenteeism
A prospective longitudinal design was used. In JanuaryeMarch SLE
Expectations of future service
2007 (Phase I) the WWBQ was posted to a sample of 4949 RN
Y/N
Y/N
Y/N
Self-reported units per week
Self-reported height and weight
Y/N
4-point Likert scale
Tick-box for list of conditions
5-point Likert scale
5-point Likert scale
Self-reported number of absences
Based on Holmes and Rahe (1967)
Self-reported length of service left
 R.S. Bridger et al. / Applied Ergonomics 42 (2011) 710e718
by their respondent ID code, to anyone else their responses would
be anonymous. The key linking service numbers to respondent IDs
was stored securely according to the requirements of the Data
Protection Act, 1998 and the project received approval form the
MoD Research Ethics Committee.
Between January and March 2008, all Phase I respondents were
re-assessed for strain (Phase II). The GHQ-12 was sent to 2596
personnel who had returned WWBQs in 2007 (the response rate in
2007 was 57%, after adjustment for non-contactable personnel). At
Phase II, the serving locations (postal addresses) of all Phase I
respondents were checked on the personnel computerised data-
base. This was done to ensure that personnel who had moved
during the year would still receive the Phase II questionnaire and so
that the researchers would be able identify personnel who had
been appointed elsewhere in the interim.
2.3. Questionnaire scoring
The WWBQ data were scored by summing individual items to
obtain totals for the parent scales used to measure the variables
listed in Table 1. Stressor exposures were self-reported on five-
point Likert scales where high scores indicated greater negativity.
SLE scores were simple counts of the number of prompted events to
which personnel had been exposed in the previous twelve months.
The GHQ-12 measure of psychological strain was scored in three
different ways:
1) Strain prevalence rates were generated by using a binary clas-
sification of strain ‘caseness’. A strain case is a respondent who
reports four or more strain symptoms on the GHQ-12. The
strain prevalence rate is given by the percentage of strain cases
in the total sample. Goldberg and Williams (1988) advise that
the choice of the number of symptoms used to describe a strain
case depends on the particular population. In the present study,
a cut-off score of four was used because it has been shown to
have reasonable testeretest reliability (Kilminster and Bridger,
2007) and to enable comparison with the strain prevalence
rates reported in the literature for military and other pop-
ulations in which a score of four is also used (e.g. Hotopf et al.,
2006).
2) The number of strain symptoms an individual reported was
counted. Note, this is an integer score that ranges from ‘0’
(no symptoms) to ‘12’ (strain symptoms on all 12 items on the
GHQ-12).
3) A ‘Likert’ strain score was obtained. Given that the 12 items on
the GHQ-12 can each be assigned a Likert score from 0 to 3, the
GHQ-Likert score ranges from ‘0’ to ‘36’ (higher numbers
indicating increased strain).
2.4. Data analysis
Returns were double data entered and the database was sub-
jected to range checking.
GHQ-12 scores at Phases I and II were the outcome measures for
analysis of concurrent and future strain. The WWBQ items
measured at Phase I were treated as predictors. Descriptive statis-
tics were generated for the sample so that response rates for
different demographic subgroups might be compared. The internal
consistency of the WWBQ scales used to measure psychosocial
stressors was re-checked by calculating values of Cronbach’s alpha.
Bivariate analyses using t-tests and Chi-Square tests were con-
ducted on the data collected in Phase I to investigate whether
baseline data differed for Phase II respondents and non-respon-
dents. Logistic regression was then used to check for response bias
713
to the Phase II survey. Using the strain case scoring method, the
percentages of personnel falling into each of the following cate-
gories were calculated:
No strain on either occasion.
Strain on both occasions (chronic strain).
Strain on one occasion (acute strain)
The sample was then split into four groups e those personnel
serving ashore at both Phases; personnel serving on ships at both
Phases; personnel working on a ship at Phase I who moved ashore
by Phase II; personnel ashore at Phase I who were on a ship by Phase
II. GHQ-12 Likert scores for the groups could then be compared.
Multiple regression analysis was conducted for concurrent
strain using the Phase I GHQ-12 Likert score as the dependent
variable and the psychosocial scale scores and demographic
predictors in Tables 1 and 2 as predictors. Bivariate correlations
were first conducted to ascertain which variables were most likely
to be predictive. Only variables found to be significantly correlated
with the outcome variables were entered into the regression
equations. Separate analyses were conducted for those serving on
a ship and those serving ashore at Phase I.
Multiple linear regression analysis was then conducted for
strain twelve months later using the Phase II GHQ-12 Likert score as
the dependent variable and the psychosocial scale scores and
demographic predictors in Tables 1 and 2 as predictors. Four
separate analyses were conducted for those serving on a ship for
the year, those ashore for the year, those who moved from ship to
shore and vice versa. Strain sufferers at Phase I (GHQ-12 symptoms
of 4þ at Phase I) were excluded from all groups to determine
whether any of the psychosocial stressor variables measured at
Phase I could predict the emergence of new strain twelve months
later. The Phase II questionnaire asked personnel to report whether
their work role had changed over the year. All personnel in the ship
or shore groups whose work role had changed were also excluded
from the Phase II analysis.
2.4.1. Psychometric revalidation of WWBQ
Returns to Phase I of the study were re-analysed and new values
of Cronbach’s alpha were calculated to confirm the internal reli-
ability of the scales in the new WWBQ. Table 1 shows that the alpha
values for most WWBQ scales were acceptable, with the exception of
the ERI scale, which had a Cronbach’s alpha of 0.44. This was found to
be due to one item concerning time pressure, which was removed
from the scale used in the regression analyses. The new ERI scale had
a Cronbach’s alpha value of 0.71 for the remaining three ERI items.
3. Results
Table 3 presents the response rates for Phase I respondents. Of
the original 4949 in the sample, 407 could not be contacted,
resulting in 4542 questionnaires being sent to personnel. The
response rate to Phase I was 57%. The lowest response rates were
from personnel under 25 years of age. Females tended to have
a higher prevalence of strain than males.
Of the 2596 personnel who returned questionnaires at Phase I,
2468 were still serving in the NS at Phase II. 1305 questionnaires
were returned at Phase II which equates to a 61% response rate
(after correction for 315 non-deliverables). Overall, the strain
prevalence rate at Phase II was 33.5% compared to 31.5% at Phase I
(Bridger et al., 2009). Approximately one half of those with strain at
Phase II were not the same individuals who had strain at Phase I:
1) 15% had psychological strain on both occasions
2) 35% had strain at either Phase I or Phase II but not on both
occasions
3) 50% had no strain on either occasion
714 R.S. Bridger et al. / Applied Ergonomics 42 (2011) 710e718

Table 3
Total sample, number of respondents, response rates and strain prevalence for NS groups at Phase I.

Group Original Number Response Strain prevalence

sample size responding rate (%) rate (%) (and 95% CI)
All 4542 2596 57 32(30e33)

Male RN officer 530 441 83 23(19e27)

RN ratinga 1823 959 53 30(27e33)
RM officer 69 46 67 31(18e45)
RM other ranks 469 153 33 25(18e32)
Age (yrs) <25 619 154 25 34(26-42)
25e34 1011 403 40 32(28e37)
35 1261 1042 83 25(28e37)

Female RN officer 334 287 86 39(33e45)

RN rating 1275 686 54 37(34e41)
RM other ranksb 42 24 57 23(14e32)
Age (yrs) <25 623 238 38 37(31e43)
25e34 682 425 62 38(34e43)
35 346 334 97 37(31e42)
a
‘Rating’ includes all ranks below the level of officer.
b
Females serving in the Royal Marines Band.

There was a small, but statistically significant difference
between the chronic strain rates (strain on both occasions) in males
and females, with females suffering from chronic strain more
frequently (c2 ¼ 4.86, df ¼ 1, p < 0.05). No differences in chronic
strain were found between those serving on ship or ashore or
between officers and ratings (personnel below the rank of officer).
For the sample as a whole, the mean change in GHQ-Likert score
(Phase II GHQ-Likert score e Phase I GHQ-Likert score) was 0.02,
with a standard deviation of 6.69 (Fig. 1).
3.1. Checks for response bias to Phase II
(avoidance) and age (younger age) measured at Phase I accounted
for 6.9% of the variance in non-response to Phase II. Younger
personnel and those who adopted avoidance coping styles were
less likely to respond at Phase II.
3.2. Deployment
Table 4 presents GHQ-12 Likert scores of personnel in the
different deployment groups.
There were no statistically significant differences in strain
between the groups.

WWBQ data from Phase I were used to compare respondents
and non-respondents to Phase II. Small but statistically significant
differences were found between respondents and non-respon-
dents. Respondents to Phase II reported less workefamily conflict,
less autonomy and control and lower commitment to the Service at
Phase I. They also reported greater over-commitment to work role
and reported using more approach-based coping strategies.
Respondents to Phase II were older, had served longer and had
experienced fewer stressful life events.
When response/non-response was analysed as a binary
outcome using logistic regression analysis, only coping style
Change in Strain Jan 07 - Jan 08
300
3.3. Modelling concurrent (Phase I) GHQ score
Two linear regression models were developed using GHQ-12
Likert data from Phase I (concurrent strain) for personnel either on
a ship or serving ashore at Phase I. A forward stepwise approach
was taken. Table 5 summarises the results of the analyses showing
the main variables associated with GHQ score at Phase I and the
amount of variance explained by the models.
For concurrent strain (i.e. strain at Phase I) the models for the
two groups share similar characteristics, accounting for over 40% of
the variance in GHQ-12 Likert score. Over-commitment accounted
for more of the variance in ship-serving personnel than those
ashore (23% versus 9%).
3.4. Modelling future (Phase II) GHQ score

Table 6 shows the regression models for Phase II strain in the

different deployment groups. In all cases, the models explained far
200
less (approximately one third to one quarter of the strain) of Phase
Frequency

II strain than was explained at Phase I. Further, the models for ship-

100 Table 4
Mean (and standard deviation) GHQ-12 Likert scores for personnel serving in four
deployment groups over twelve months (min ¼ 0, max ¼ 36).
Mean =0.17 GHQ-12 Phase 1 GHQ-12 Phase II N
Std. Dev. =3.775
0 N =1,259 (2007)a (2008)a
-15 -10 -5 0 5 10 15
ShipeShip. 12.9 (5.2) 12.6 (6.0) 176
GHQPIIIminusGHQPI
ShipeShore 12.2 (5.2) 11.7 (6.4) 89
Change in GHQ12 Symptoms ShoreeShip 12.1 (4.8) 12.9 (5.9) 97
ShoreeShore 12.5 (5.6) 12.5 (5.8) 897
Fig. 1. Phase II GHQ-12 symptom score minus Phase I GHQ-12 symptom score a
Higher Scores Indicate Greater Strain.
(max þ12, min 12).
 R.S. Bridger et al. / Applied Ergonomics 42 (2011) 710e718 715

Table 5 ship-based personnel (the ‘ShipeShip’ group in Table 6) for whom

Psychosocial factors associated with concurrent GHQ score (Phase I) in personnel dissatisfaction with onboard living conditions at Phase I predicted
serving on ships and ashore.
strain at Phase 2. Five item scores were significantly related to strain
Predictor Beta Multiple Multiple P at Phase II in one-tailed comparisons: dissatisfaction with privacy
R2 R2 change
(r ¼ 0.32, P < 0.01); dissatisfaction with noise (r ¼ 0.26, P < 0.05);
Ships dissatisfaction with vibration on the ship (r ¼ 0.22, P < 0.05);
Over-commitment 0.871 0.228 0.228 <0.001
dissatisfaction with sanitary arrangements (r ¼ 0.22, P < 0.05); and
ERI 0.322 0.314 0.086 <0.001
Organisational commitment 0.161 0.348 0.034 <0.001 crowding onboard (r ¼ 0.20, P < 0.05). As might be expected, given
Negative mood 0.499 0.373 0.025 <0.001 the high value of Alpha for this scale, high correlations between the
Autonomy and control 0.153 0.388 0.015 <0.005 item scores were found and so a multiple linear regression analysis
SLE 0.271 0.404 0.016 <0.005 was conducted to determine the relative value of each item’s
Workefamily conflict 0.196 0.414 0.010 <0.005
Total R2 0.414
contribution to Phase II strain using a stepwise approach. A statis-
tically significant model was obtained. Only the privacy score
Shore emerged as a statistically significant predictor of Phase II strain
Negative mood 0.833 0.207 0.207 <0.001 (r ¼ 0.30, r2 ¼ 0.89, F ¼ 10.72, df ¼ 1 and 110, P < 0.001).
Over-commitment 0.660 0.297 0.090 <0.001
Organisational commitment 0.151 0.372 0.075 <0.001
4. Discussion
Positive mood 0.430 0.396 0.024 <0.001
ERI 0.266 0.419 0.023 <0.001
SLE 0.221 0.435 0.016 <0.05 The present study compared subsets of the same workforce
Workefamily conflict 0.179 0.445 0.010 <0.05 working in different physical environments e ships or shore-based
Total R2 0.445
Naval establishments. Although the levels of strain in personnel did
not differ according to the particular environment they were
working in, the variables associated with and predictive of strain
did differ. The differences were most pronounced in the predictive
based and shore-based personnel are quite different. For shore-
models of strain. Mood and family-related factors predicted GHQ-
based personnel, personal factors (mood) and workefamily issues
12 score 12 months later in personnel working ashore, whereas
predicted Phase II GHQ-12 score, which was higher in those with
dissatisfaction with living conditions on ships and a lack of
low mood who had experienced stressful life events and work-
autonomy and control predicted future GHQ-12 score in those
efamily conflict at Phase I. For ship-based personnel, Phase II GHQ-
serving on ships. In most other ways, though, the groups are the
12 score was predicted by dissatisfaction with living conditions
same e all NS personnel of different ranks and genders working
aboard ship, a lack of autonomy and control at work and avoidance
under contractually identical conditions of service.
coping. For personnel whose place of work had changed over the
year, only personal factors measured at Phase I predicted GHQ-12
4.1. Concurrent strain
score at Phase II (as might be expected because they were no longer
exposed to the work stressors they had reported 12 months before).
The regression model shown in Table 5 suggests that the ERI
model predicted significant variance in concurrent strain (strain at
3.5. Supplementary analysis of living conditions and Phase II strain the time the stressors were measured) in NS personnel irrespective
of whether the personnel were serving onboard ship or ashore. Of
The scale used to measure satisfaction with living conditions had the different components of the ERI model, over-commitment to
nine items dealing with: vibration; noise; privacy; crowding; sani- work role appeared to be particularly important in explaining
tary facilities; ease of cleaning; maintainability, personal space and concurrent strain in ship-based personnel. Over-commitment to
climate. Pearson correlation coefficients were calculated for the work role is indicated in the WWBQ by positive responses to items
scale item scores and strain at Phase II. The analysis was restricted to such as: ‘as soon as I get up in the morning, I think about work
problems’; ‘work is still on my mind when I go to bed’; and ‘if I
postpone something that I was supposed to do today, I’ll have
Table 6 problems sleeping tonight’. It is not surprising, given the limited
Psychosocial factors associated with Phase II GHQ-12 score in four deployment possibilities for disengagement from work on ships, that many
groups (excluding Phase I strain cases and personnel whose role had changed). personnel are ‘over-committed’ in the sense used in the ERI model.
Predictor Beta Multiple Multiple P Indeed, over-commitment may be appropriate in certain circum-
R2 R2 change stances and it is very likely that it is linked to engagement as dis-
1. ShoreeShore N ¼ 828 cussed in the DemandeResources Model. However, the present
Negative mood 0.220 0.038 0.038 <0.001 findings linking over-commitment to psychological strain,
Workefamily conflict 0.090 0.063 0.027 <0.001
accounted for 23% of the variance in strain in those serving on ships
Stressful life events 0.180 0.078 0.016 <0.001
Positive mood 0.111 0.089 0.012 <0.001 compared to 9% in those serving ashore. This is an important
finding because over-commitment has been found to be linked to
2. ShipeShip N ¼ 111 the development of health problems, such as coronary heart
Satisfaction with living conditions 0.219 0.055 0.064 <0.05 disease in studies of civilians (e.g. Bosma et al., 1998). The ERI model
Autonomy and control 0.338 0.089 0.042 <0.05 did not provide a complete explanation of concurrent strain in NS
Stress coping 0.462 0.115 0.033 <0.05
personnel. Commitment to the NS also played a role e personnel
3. ShipeShore N ¼ 77 with higher levels of commitment to the NS were less likely to
Age 0.222 0.063 0.063 <0.05 experience concurrent strain, and this applied to both personnel
Mood 0.990 0.012 0.057 <0.05 working ashore and on ships. Further, for ship-based personnel
only, a lack of autonomy and control was also associated with
4. ShoreeShip N ¼ 84 elevated GHQ-12 score. This finding is not unexpected. Although
Organisational commitment 0.280 0.06 0.061 <0.05
military personnel, almost by definition, do lack autonomy and
716 R.S. Bridger et al. / Applied Ergonomics 42 (2011) 710e718
control at work, it is particularly severe when serving at sea
because the work environment and the living environment are one
and the same and the possibilities for any kind of independent
action either at work or in the limited leisure time available, are
minimal. For concurrent strain then, there is evidence that some
components of both the DCS and ERI models are important
predictors of strain.
4.2. Future strain
The findings summarised in Table 6 suggest that the DCS model
provided a better explanation of future strain than the ERI model for
personnel working on ships. More importantly, perhaps, the find-
ings suggest that high levels of engagement at one time do not
necessarily predict adverse psychological outcomes later on and
this is what would be expected if over-commitment, or high
engagement, is regarded as a temporary coping strategy.
A perceived lack of autonomy and control, and dissatisfaction with
living conditions on ships predicted strain twelve months later in
personnel working and living on ships. These stressors would
appear to be chronic stressors in those exposed to them e if present
in the work environment of healthy individuals they increase like-
lihood of strain in the future. These findings exhibit some agree-
ment with those of Sandal et al. (2006) in explaining the causes of
strain in people working in isolated and extreme environments.
Satisfaction with the work environment has long been known
to be a component of job satisfaction (Sundstrom, 1986). The main
areas of dissatisfaction with the environment on ships are main-
tainability and ease of cleaning of equipment, the air quality, noise,
crowding, lack of privacy and the provision of sanitary facilities
(Dew et al., 2007). The present findings suggest that the physical
environment in warships may, in some individuals, act as
a chronic occupational stressor. From Table 1 it can be seen that
the nine-item WWBQ scale used to measure satisfaction with
living conditions has a high value of alpha (0.93). This suggests
that there is no particular aspect of the physical environment on
ships that is the source of dissatisfaction. The scale has items
concerning: vibration exposure; noise; privacy; crowding; provi-
sion of sanitary facilities; ease of cleaning; maintainability; space
for personal effects and indoor climate, yet it has high internal
consistency, suggesting that it is both the constraints of life aboard
and the intrusiveness of the physical and social surroundings that
some people find stressful. It would seem that those who are
sensitive to the constraints of such environments do not become
habituated to them over time and are more likely to develop strain
when exposed to the environment for a long time (e.g. twelve
months).
For those working ashore at both Phases, the best psychosocial
predictor of strain 12 months in the future was workefamily
conflict. For those at sea, workefamily conflict did not predict
future strain. Sandal et al. (2006) has pointed out that when people
have to work in isolated or extreme environments, they are often
exposed to unusual or intense stressors that are not often found in
daily life. At the same time, however, they are completely isolated
from many of the routine demands of daily life. Thus, the total
demands acting on personnel deployed at sea may be no more than
those experienced by personnel working ashore e it is the nature of
the demands that differs, not the total demand. This explanation is
in accord with the data in Table 4 e the GHQ-12 scores in the
different deployment groups do not differ significantly.
It is not surprising that, in the groups who changed location
between Phases (moving from ship to shore or shore to ship) only
personal factors predicted Phase II GHQ-12 score, as having moved
to a new job, personnel were no longer exposed to the occupational
stressors measured at Phase I.
4.3. Limitations of the study
One of the limitations of the present study is the low response
rate of 57% at Phase I, and the fact that only 61% of these Phase I
respondents responded at Phase II. Although the checks for
response bias at Phase II found little evidence of direct bias (i.e.
personnel with strain at Phase I were no less likely to respond at
Phase II than personnel with no strain) response rates were lower
for younger personnel. It is likely that had a greater percentage of
younger personnel responded to the questionnaire, an even greater
proportion of the variance in strain may have been accounted for in
those working on ships. This is because young personnel in the
Navy tend to be of lower rank, have less autonomy and control,
carry out more physically demanding work and have less personal
space and privacy on ships than more senior personnel. Older, more
senior officers tend to work ashore more frequently than younger,
lower ranked personnel. Indeed in the current sample the average
age of personnel working aboard ships at both Phases was signifi-
cantly lower than the average age of personnel working ashore
(33.6 yrs compared to 36.9 yrs). The greater proportion of younger,
more junior ranked personnel on ships may be a partial explanation
as to why autonomy and control only predicted strain in those
working on ships, as these younger personnel have less authority
and less autonomy and control than older more senior ranked
personnel.
A second limitation is that the models predict only a small
percentage of the variance in strain at Phase II. However, this might
be expected given that the day to day demands on personnel,
particularly those serving on ships, will vary greatly depending on
whether the ship is at sea or alongside, or whether it is partici-
pating in exercises or is actively deployed. Also, several of the Phase
II regression models will have had less statistical power than the
Phase I models because of the need to excluded Phase I strain cases
and personnel whose role had changed, thus reducing the sample
size.
In general, questionnaire studies of military personnel tend to
yield low response rates of 30e50% (e.g. the 2009 UK Armed Forces
Continuous Attitude Survey yielded a response rate of 32%, Deputy
Chief of Defence Staff, 2009) so there is always a challenge to
produce instruments that combine brevity (to encourage
a response) with psychometric robustness.
4.4. General discussion
Previous studies of NS personnel have shown that, for most
Naval personnel, serving on ships is no more stressful than serving
ashore, as evidenced by GHQ-12 prevalence rates and strain scores.
The present findings are of relevance to stress management policy
and to ergonomics because they demonstrate that the causes of
strain in a specific occupational group differ depending on the
environment they work in and depending on the time scale over
which the occupational demands operate. There are many argu-
ments for incorporating ergonomics into the design processes of
systems and traditionally these arguments focus on outcomes such
as physical health and safety. The study presents tentative evidence
to support an ergonomics approach to ship design by demon-
strating that ship habitability is important for the mental health of
those aboard and not just for comfort and convenience. Habitability
of ships is likely to be increasingly important in future with the
design of more highly automated ships with fewer people onboard.
Not only is it difficult and time consuming to extract personnel who
cannot cope with such environments, but with ‘leaner-manned’
vessels, personnel are more specialised and therefore more difficult
to replace. However, these points should be treated with caution,
given that the main focus of the questionnaire measure used in this
 R.S. Bridger et al. / Applied Ergonomics 42 (2011) 710e718
study was not ship environments. The measure of ship habitability
used in the present study was crude, with only 9-items measured
on five-point Likert scales and more detailed studies are called for
to identify the key aspects of ship design.
Psychosocial factors associated with concurrent strain did not all
predict future strain in the sample of personnel under investiga-
tion. This is almost certainly because many personnel exhibited
change in GHQ-12 score over time (Fig. 1) possibly due to fluctua-
tions in work demands. Amongst military personnel, there is some
evidence that personnel recover quickly from high levels of strain if
moved to a less stressful role. Rona et al. (2008) presented evidence
that mental health problems were more likely in personnel
deployed for 13 months or more in a three-year period than those
deployed for shorter periods. The number of deployments in the
same period was not associated with mental health problems
because the individual deployments were shorter and provided
more opportunity for recovery. Thus, it would seem that the
varying nature of NS working life might protect individuals from
chronic strain. Chronic exposure to isolated and extreme environ-
ments is often avoided by the way deployments are organised.
Indeed, Hotopf et al. (2006) found few differences in the health of
regular UK military personnel who had deployed to the 2003 Iraq
war and those who had not. According to Rona et al., when chronic
exposure does occur (due to personnel shortages, for example)
psychological strain results. It is noteworthy that the present data
indicate that it is not just the strain levels that change over time, but
the chronicity and nature of the stressors to which personnel are
exposed.
The study provides some evidence for the distinction between
acute and chronic stressors. Chronic stressors may provoke no
reaction initially but cause strain symptoms to develop in people
exposed to them. In the present study, the living conditions
aboard ships, particularly the lack of privacy, are a chronic
stressor that operates over a 12-month time period, indicating
that, some personnel at least, do not habituate to these
deprivations over 12 months. The link with Phase II strain
demonstrates that a poor environment with little privacy is
anxiety-provoking in some people. This is a novel finding in
maritime research but is in accord with studies of office workers.
Oldham and Fried (1987), for example, found that adverse reac-
tions were more common in employees working in offices that
they rated as dark, with few enclosures surrounding employees’
work areas and when employees were seated close to one
another, with many employees occupied an office. All of these
workspace characteristics of offices share some similarities with
the deprivations of the living environment on ships. Thus, lack of
privacy on ships is an anxiety-provoking stressor in some people
and not just a source of dissatisfaction. In the language of
Demerouti et al. (2001), coping with the demands of life onboard
ship is a process of energy depletion that places personnel at risk
of strain. On ships, the resources to meet these demands,
particularly the possibility of providing additional resources in
the form of support or relief, are limited, so continued exposure
leads to strain. Ultimately, it would be expected that continued
exposure would lead to disengagement and lowered perfor-
mance, placing the operational capability of the vessel at risk. A
question for further research in missions where personnel are
required to work in isolated and extreme environments is to
determine the time course of energy depletion, the optimum
periods for deployment and recovery and the scope for amelio-
ration through ergonomic design.
Finally, the paper provides evidence for the validity of the
WWBQ and the constructs it purports to measure. Several
psychosocial stressor variables were associated with strain in those
whose work environment and work role remained unchanged for
717
12 months. For those who changed from working on a ship to
working ashore, and vice versa, only personal factors predicted
future strain.
5. Conclusions
This study has identified some important psychological and
physical variables in the prediction of current strain and future
strain in Naval personnel. In particular, the role of satisfaction with
living conditions aboard ships in the prediction of future strain is
highlighted. Several occupational stressors from general models of
occupational stress were found to predict future strain in
personnel on ships (lack of autonomy and control and dissatis-
faction with living conditions, particularly the lack of privacy),
suggesting that these are chronic stressors that can have adverse
effects one year later. This is an important finding in applied
ergonomics because it demonstrates that design decisions taken
early, that may be difficult to reverse later, can lead to long-term
problems to which personnel do not adapt. Personal factors and
problems outside work predicted strain in personnel working
ashore.
References
Bosma, H., Peter, R., Siegrist, J., Marmot, M., 1998. Two alternative job stress models
and the risk of coronary heart disease. American Journal of Public Health 88,
68e75.
Bridger, R.S., Kilminster, S., Slaven, G., 2007. Occupational stress and strain in the
Royal Navy 1999 and 2004. Occupational Medicine 57, 92e97.
Bridger, R.S., Brasher, K., Dew, A., Kilminster, S., 2008. Occupational stress and strain
in the Royal Navy 2007. Occupational Medicine 58, 534e539.
Bridger, R.S., Brasher, K., Dew, A., Munnoch, K., Kilminster, S., 2009. Chronics and
acute psychological strain in Naval personnel. Occupational Medicine 59,
454e458.
Collins, P.A., Gibbs, A.C.C., 2003. Stress in police officers: a study of the origins,
prevalence and severity of stress-related symptoms within a county police
force. Occupational Medicine 4, 256e264.
Cooper, C.L., Marshall, J., 1976. Occupational sources of stress: a review of the
literature relating to coronary heart disease and mental ill-health. Journal of
Occupational Psychology 49, 11e28.
DCDS(Pers), 2009. Armed Forces Continuous Attitude Survey. http://www.mod.uk/NR/
rdonlyres/54D61DA9-C1D0-4B9D-8B89.84C05193F098/0/afcas08_main_rpt.pdf.
Demerouti, E., Bakker, A.B., Nachreiner, F., Schaufeli, W., 2001. The job
demandseresources model of burnout. Journal of Applied Psychology 86,
499e512.
Dew, A., Brasher, K., Bridger, R.S. 2007. Naval Service Stress Survey 2007: Supple-
mentary Report on Ship Habitability, shore-based Accommodation and Strain.
Unpublished MoD Report.
Dollard, M.f., Winefield, A.H., Winefield, H.R., 2003. Occupational Stress in the
Service Professions. Taylor and Francis, London.
Goldberg, D., Williams, P., 1988. A User’s Guide to the General Health Questionnaire.
NFER Nelson.
Holmes, D., Rahe, J., 1967. The social readjustment scale. Journal of Psychosomatic
Research 11, 213e218.
Hotopf, M., et al., 2006. The health of UK military personnel who deployed to the
2003 Iraq way: a cohort study. The Lancet 367, 1731e1741.
Karasek, R.A., Theorell, T., 1990. Healthy Work: Stress, Productivity and the
Reconstruction of Working Life. Basic Books, New York.
Kilminster, S. Bridger, R.S. 2007. Modified work and Well-Being Questionnaire for
Cohort Study of Stress in the Naval Service: Psychometric Validation. Unpub-
lished MoD Report.
Marmot, M., Brunner, E., 2004. Cohort profile: the Whitehall II Study. International
Journal of Epidemiology 34, 251e256.
Marmot, M.G., Davey-Smith, G., Stansfield, S., Patel, C., North, F., Head, J., 1991.
Health inequalities among British civil servants. Lancet 337, 1387e1393.
Miller, G.E., Chen, E., Zhou, E.S., 2007. If it goes up, must it come down? Chronic
stress and the hypothalamic-pituitary-adrenocortical axis in humans. Psycho-
logical Bulletin 133, 25e45.
Oldham, G.R., Fried, Y., 1987. Employee reactions to workspace characteristics.
Applied Psychology 72, 75e80.
Rona, R.J., Fear, N.T., Hull, L., Greenberg, N., Earnshaw, M., Hotopf, M., Wessely, S.,
2008. Mental health consequences of overstretch in the UK armed forces:
first phase of a cohort study. British Medical Journal. doi:101136/bmj/
3927458572BE.
Salonen, P.H., Arola, H., Nygard, C.H., Huhtula, H., 2008. Long term associations of
stress and chronic diseases. Psychology, Health & Medicine 13, 55e62.
718 R.S. Bridger et al. / Applied Ergonomics 42 (2011) 710e718

Sandal, G.M., Leon, G.R., Palinkas, L., 2006. Human challenges in polar and space
environments. Reviews in Environmental Science and Biotechnology 5,
281e296.
Schaufelli, W., Baker, A.B., 2004. Job demands, job resources and their relationship
with burnout and engagement: a multi-sample study. Journal of Organisational
Behaviour 25, 293e315.
Schei, E., 1994. A strengthening experience:Mental distress during military service.
Social Psychiatry and Psychiatric Epidemiology 29, 40e45.
Siegrist, J., 1996. Adverse health effects of high effort/low reward conditions. Journal
of Occupational Health Psychology 1, 27e41.
Slaven, G., 2002. Work-related Strain in the Naval Service. University of East Lon-
don, Dissertation, London.
Sundstrom, E.D., 1986. Workplaces. Cambridge University Press, England.
Watson, D., Clark, L.A., Tellegen, A., 1988. Development and validation of brief
measures of positive and negative affect: the PANAS scale. Journal of Personality
and Social Psychology 54, 1063e1070.