Professional Documents
Culture Documents
Applied Ergonomics: R.S. Bridger, K. Brasher, A. Dew, S. Kilminster
Applied Ergonomics: R.S. Bridger, K. Brasher, A. Dew, S. Kilminster
Applied Ergonomics
journal homepage: www.elsevier.com/locate/apergo
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.
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 711
the GHQ-12 in these studies enables the affective responses of 1.2. Naval deployment patterns
personnel to different aspects of Service life to be assessed in
relation to their consequences for the well-being of individuals. Naval personnel can serve in a shore-based establishment or
The link between the work environment, psychological strain onboard a ship. When on ship, they are usually appointed for
and the development of physical and medical ailments is well- a period of two years and the ship is their main place of work, be it
established. For example, Salonen et al. (2008) found that pro- at sea or alongside. During the appointment they can be deployed
longed strain predicted musculoskeletal disorders, nervous system, at sea for periods of six months or more without returning home.
eye, ear and metabolic diseases and mental disorders in Finnish When at sea, personnel work long hours, usually with no time off at
factory workers over a twenty year period. The Whitehall II study weekends e essentially, they work an extended form of shift-work,
showed that an individual’s position within the hierarchy of an known as ‘watches’. Shore-based personnel in the NS may work as
organisation was a much better predictor of heart attack than hard or even harder than personnel at sea but they can leave the
obesity, smoking and high blood pressure (Marmot et al., 1991). work environment, live in their own homes and practice favourite
A lack of control at work has also been linked to ill-health (Bosma recreations and hobbies in their spare time. Personnel are guar-
et al., 1998) which supports the predictions of the well-known anteed to work no more than 660 days away from their homeport
‘DemandeControleSupport’ (DCS) model of Karasek. The link in any rolling three-year period, providing at least 465 days ashore
between strain and ill-health may be direct, because of the meta- or alongside. Once home, they may find themselves working or
bolic effects of strain (see Miller et al., 2007, for a review) or indi- training at a shore-base for a while before the next deployment.
rect. Schei (1994), for example, found that strain was associated Appointments at sea are often interspersed with shore postings of
with physical inactivity and the consumption of tobacco, ‘junk food’ similar length. Thus, working life in the NS differs from working life
and alcohol in Norwegian military conscripts. in other occupations because the place and conditions of work and
even job content change regularly over the course of a career. Cross-
1.1. Theories and models sectional studies of NS personnel are of limited value because many
of the stressors identified may be transient and unlikely to be
Karasek’s model (Karasek and Theorell, 1990) predicts job strain linked to long-term health or employee turnover outcomes.
when personnel are exposed to occupational stressors such as high The above suggests that repeated measures of strain are needed
job demands and a lack of control over work. Over-commitment, in if strain and its consequences are to be properly understood.
Siegrist’s (1996) Effort Reward Imbalance (ERI) model of occupa- Further, the use of repeated measures may solve some of the
tional stress regards over-commitment of work as a maladaptive problems of cross-sectional research designs. For example, if
coping strategy e a pattern of excessive striving which leads to ill- occupational stressors measured at a particular time predict future
health. In the short term, over-commitment reflects motivation. In strain in personnel whose role is unchanged, then these might be
the long term, over-commitment can only be maintained if it is considered to be more important than transient mood states or
reinforced by personal development, recognition and goal factors that only influence strain experienced concurrently. State
attainment. variables such as the GHQ-12 might be expected to vary over time
Schaufelli and Baker (2004) discuss the DemandeResources for a wide variety of reasons such as mood swings and events
model, which, although developed as a model of burnout, proposes outside of work. It is clear from the writings of Schaufelli and Baker
a mechanism by which strain may lead to adverse outcomes. High (2004) for example, that demand resource imbalances in the
job demands deplete energy resources, resulting in withdrawal in workplace will only lead to adverse outcomes if they are sustained
order to conserve resources, and job performance drops as a result. and that burnout is the end point of a process rather than an event.
Attempts to maintain performance (the ‘strain coping mode’) are Thus, self-reported ratings of work demands that predict GHQ-12
characterised by over-activation of the sympathetic nervous system scores over several time points will be more significant than those
and this is the mechanism by which stressors in the work envi- that only predict GHQ-12 score measured concurrently.
ronment get ‘into the body’. Long-term over-activation of the Consequently, as a first step towards improving the explanatory
hypothalamicepituitaryeadrenocortical axis appears to be the power of previous cross-sectional NS research, the current study
mechanism and it can have long-term health consequences (Miller investigated stressor exposure, concurrent strain, and strain twelve
et al., 2007). In the DemandeResources model, autonomy and months into the future in a cohort of personnel across the NS. The
support from leaders and peers are resources that support active purpose was to determine whether exposure to occupational
coping with work demands, thus lowering strain. Although stressors of the kind described in the models predict future strain in
burnout is rare in the NS, it is thought to be the end point of personnel serving on ships and in personnel serving in shore-based
a process of energy depletion so we would expect job strain to establishments. Identification of occupational stressors causing
precede burnout (Demerouti et al., 2001) and some would argue future strain has clear advantages for the improvement of in-
that ‘strain’ as used in the present paper is a milder equivalent of Service stress management.
the concept of ‘emotional exhaustion’ used in burnout research. In particular, the study enabled the temporal stability of strain to
Thus, although explicit burnout is almost unknown in Naval be better understood and it enabled the predictive validity of
personnel, there is a conceptual link between these models dis- workplace stressors to be assessed. The null hypothesis that
cussed above and the processes currently under investigation in the workplace stressors do not predict future strain in personnel
present research. serving on ship or ashore was tested.
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 2. Method
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- 2.1. Work and Well-Being Questionnaire (WWBQ)
nation of current and future strain in personnel serving ashore and
personnel serving onboard ships. The deployment system used by The WWBQ was used to measure occupational stressors in the
the NS is outlined below, to explain why these two groups of Naval sample at Phase I of this longitudinal survey. The factors
personnel need to be looked at separately. 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 personnel (10% of males and 60% of females in the NS, the latter to
groups of five to eight personnel of equivalent rank in three Naval ensure adequate sub-sample size for analysis) who were randomly
establishments (Slaven, 2002). A pilot questionnaire was then selected and were stratified by branch. Seasonal effects on dispo-
circulated to 600 randomly selected personnel. Factor analysis of sition were controlled for by conducting the survey at the same
the returns was conducted to identify items that loaded discretely time of year as the previous cross-sectional surveys conducted in
onto different factors. The item set was reduced by calculating 1999 and 2004 (Slaven, 2002; Bridger et al., 2007). The question-
alpha coefficients for each of the scales and removing items until naire was sent in three postings, with only non-respondents being
the value of alpha exceeded 0.7. Additional scales were added in targeted in subsequent postings. Demographic information was
2006 (Kilminster and Bridger, 2007) and psychometric reduction collected from the NS personnel database and linked to an anon-
and validation was carried out using a sample of 28 naval ymous respondent ID code. The initial and subsequent question-
personnel. The purpose of the psychometric reduction was to naires were marked with the ID code so that WWBQ responses
confirm the factor structure of the original WWBQ and to reduce could be compared within subjects over time. Participants were
the length of the new WWBQ by extracting 3e5 items from parent informed that all their answers would be treated as ‘medical in
scales of ERI, over-commitment and mood. The new scales confidence’, and that while the project workers could identify them
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. Table 2
WWBQ health and demographic measures.
A five-day testeretest reliability study was conducted and unreli-
able items were removed. Most of the items in all scales of the final Other health and demographic Response format
WWBQ were highly reliable with Spearman correlation coefficients WWBQ measures
over 0.7. Psychometric revalidation of the WWBQ was carried out in Job specialisation Y/N
Job preference area Y/N
the present study and is reported in Table 1.
Physically demanding work Y/N
The new WWBQ measured occupational stressor exposure, Alcohol intake Self-reported units per week
exposure to stressful life events and main confounding factors and BMI Self-reported height and weight
effect modifiers. Psychological strain was measured using the GHQ- Smoking Y/N
12. Health and demographic measures are summarised in Table 2. Exercise participation 4-point Likert scale
Health complaints Tick-box for list of conditions
Self-assessed general health 5-point Likert scale
2.2. Study design Sea-sickness 5-point Likert scale
Absenteeism Self-reported number of absences
A prospective longitudinal design was used. In JanuaryeMarch SLE Based on Holmes and Rahe (1967)
Expectations of future service Self-reported length of service left
2007 (Phase I) the WWBQ was posted to a sample of 4949 RN
R.S. Bridger et al. / Applied Ergonomics 42 (2011) 710e718 713
by their respondent ID code, to anyone else their responses would to the Phase II survey. Using the strain case scoring method, the
be anonymous. The key linking service numbers to respondent IDs percentages of personnel falling into each of the following cate-
was stored securely according to the requirements of the Data gories were calculated:
Protection Act, 1998 and the project received approval form the No strain on either occasion.
MoD Research Ethics Committee. Strain on both occasions (chronic strain).
Between January and March 2008, all Phase I respondents were Strain on one occasion (acute strain)
re-assessed for strain (Phase II). The GHQ-12 was sent to 2596 The sample was then split into four groups e those personnel
personnel who had returned WWBQs in 2007 (the response rate in serving ashore at both Phases; personnel serving on ships at both
2007 was 57%, after adjustment for non-contactable personnel). At Phases; personnel working on a ship at Phase I who moved ashore
Phase II, the serving locations (postal addresses) of all Phase I by Phase II; personnel ashore at Phase I who were on a ship by Phase
respondents were checked on the personnel computerised data- II. GHQ-12 Likert scores for the groups could then be compared.
base. This was done to ensure that personnel who had moved Multiple regression analysis was conducted for concurrent
during the year would still receive the Phase II questionnaire and so strain using the Phase I GHQ-12 Likert score as the dependent
that the researchers would be able identify personnel who had variable and the psychosocial scale scores and demographic
been appointed elsewhere in the interim. predictors in Tables 1 and 2 as predictors. Bivariate correlations
were first conducted to ascertain which variables were most likely
2.3. Questionnaire scoring to be predictive. Only variables found to be significantly correlated
with the outcome variables were entered into the regression
The WWBQ data were scored by summing individual items to equations. Separate analyses were conducted for those serving on
obtain totals for the parent scales used to measure the variables a ship and those serving ashore at Phase I.
listed in Table 1. Stressor exposures were self-reported on five- Multiple linear regression analysis was then conducted for
point Likert scales where high scores indicated greater negativity. strain twelve months later using the Phase II GHQ-12 Likert score as
SLE scores were simple counts of the number of prompted events to the dependent variable and the psychosocial scale scores and
which personnel had been exposed in the previous twelve months. demographic predictors in Tables 1 and 2 as predictors. Four
The GHQ-12 measure of psychological strain was scored in three separate analyses were conducted for those serving on a ship for
different ways: 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
1) Strain prevalence rates were generated by using a binary clas- of 4þ at Phase I) were excluded from all groups to determine
sification of strain ‘caseness’. A strain case is a respondent who whether any of the psychosocial stressor variables measured at
reports four or more strain symptoms on the GHQ-12. The Phase I could predict the emergence of new strain twelve months
strain prevalence rate is given by the percentage of strain cases later. The Phase II questionnaire asked personnel to report whether
in the total sample. Goldberg and Williams (1988) advise that their work role had changed over the year. All personnel in the ship
the choice of the number of symptoms used to describe a strain or shore groups whose work role had changed were also excluded
case depends on the particular population. In the present study, from the Phase II analysis.
a cut-off score of four was used because it has been shown to
have reasonable testeretest reliability (Kilminster and Bridger, 2.4.1. Psychometric revalidation of WWBQ
2007) and to enable comparison with the strain prevalence Returns to Phase I of the study were re-analysed and new values
rates reported in the literature for military and other pop- of Cronbach’s alpha were calculated to confirm the internal reli-
ulations in which a score of four is also used (e.g. Hotopf et al., ability of the scales in the new WWBQ. Table 1 shows that the alpha
2006). values for most WWBQ scales were acceptable, with the exception of
2) The number of strain symptoms an individual reported was the ERI scale, which had a Cronbach’s alpha of 0.44. This was found to
counted. Note, this is an integer score that ranges from ‘0’ be due to one item concerning time pressure, which was removed
(no symptoms) to ‘12’ (strain symptoms on all 12 items on the from the scale used in the regression analyses. The new ERI scale had
GHQ-12). a Cronbach’s alpha value of 0.71 for the remaining three ERI items.
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 3. Results
GHQ-Likert score ranges from ‘0’ to ‘36’ (higher numbers
indicating increased strain). 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
2.4. Data analysis response rate to Phase I was 57%. The lowest response rates were
from personnel under 25 years of age. Females tended to have
Returns were double data entered and the database was sub- a higher prevalence of strain than males.
jected to range checking. Of the 2596 personnel who returned questionnaires at Phase I,
GHQ-12 scores at Phases I and II were the outcome measures for 2468 were still serving in the NS at Phase II. 1305 questionnaires
analysis of concurrent and future strain. The WWBQ items were returned at Phase II which equates to a 61% response rate
measured at Phase I were treated as predictors. Descriptive statis- (after correction for 315 non-deliverables). Overall, the strain
tics were generated for the sample so that response rates for prevalence rate at Phase II was 33.5% compared to 31.5% at Phase I
different demographic subgroups might be compared. The internal (Bridger et al., 2009). Approximately one half of those with strain at
consistency of the WWBQ scales used to measure psychosocial Phase II were not the same individuals who had strain at Phase I:
stressors was re-checked by calculating values of Cronbach’s alpha.
Bivariate analyses using t-tests and Chi-Square tests were con- 1) 15% had psychological strain on both occasions
ducted on the data collected in Phase I to investigate whether 2) 35% had strain at either Phase I or Phase II but not on both
baseline data differed for Phase II respondents and non-respon- occasions
dents. Logistic regression was then used to check for response bias 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.
There was a small, but statistically significant difference (avoidance) and age (younger age) measured at Phase I accounted
between the chronic strain rates (strain on both occasions) in males for 6.9% of the variance in non-response to Phase II. Younger
and females, with females suffering from chronic strain more personnel and those who adopted avoidance coping styles were
frequently (c2 ¼ 4.86, df ¼ 1, p < 0.05). No differences in chronic less likely to respond at Phase II.
strain were found between those serving on ship or ashore or
between officers and ratings (personnel below the rank of officer). 3.2. Deployment
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, Table 4 presents GHQ-12 Likert scores of personnel in the
with a standard deviation of 6.69 (Fig. 1). different deployment groups.
There were no statistically significant differences in strain
between the groups.
3.1. Checks for response bias to Phase II
WWBQ data from Phase I were used to compare respondents 3.3. Modelling concurrent (Phase I) GHQ score
and non-respondents to Phase II. Small but statistically significant
differences were found between respondents and non-respon- Two linear regression models were developed using GHQ-12
dents. Respondents to Phase II reported less workefamily conflict, Likert data from Phase I (concurrent strain) for personnel either on
less autonomy and control and lower commitment to the Service at a ship or serving ashore at Phase I. A forward stepwise approach
Phase I. They also reported greater over-commitment to work role was taken. Table 5 summarises the results of the analyses showing
and reported using more approach-based coping strategies. the main variables associated with GHQ score at Phase I and the
Respondents to Phase II were older, had served longer and had amount of variance explained by the models.
experienced fewer stressful life events. For concurrent strain (i.e. strain at Phase I) the models for the
When response/non-response was analysed as a binary two groups share similar characteristics, accounting for over 40% of
outcome using logistic regression analysis, only coping style 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%).
Change in Strain Jan 07 - Jan 08
300
3.4. Modelling future (Phase II) GHQ score
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
control at work, it is particularly severe when serving at sea 4.3. Limitations of the study
because the work environment and the living environment are one
and the same and the possibilities for any kind of independent One of the limitations of the present study is the low response
action either at work or in the limited leisure time available, are rate of 57% at Phase I, and the fact that only 61% of these Phase I
minimal. For concurrent strain then, there is evidence that some respondents responded at Phase II. Although the checks for
components of both the DCS and ERI models are important response bias at Phase II found little evidence of direct bias (i.e.
predictors of strain. personnel with strain at Phase I were no less likely to respond at
Phase II than personnel with no strain) response rates were lower
4.2. Future strain for younger personnel. It is likely that had a greater percentage of
younger personnel responded to the questionnaire, an even greater
The findings summarised in Table 6 suggest that the DCS model proportion of the variance in strain may have been accounted for in
provided a better explanation of future strain than the ERI model for those working on ships. This is because young personnel in the
personnel working on ships. More importantly, perhaps, the find- Navy tend to be of lower rank, have less autonomy and control,
ings suggest that high levels of engagement at one time do not carry out more physically demanding work and have less personal
necessarily predict adverse psychological outcomes later on and space and privacy on ships than more senior personnel. Older, more
this is what would be expected if over-commitment, or high senior officers tend to work ashore more frequently than younger,
engagement, is regarded as a temporary coping strategy. lower ranked personnel. Indeed in the current sample the average
A perceived lack of autonomy and control, and dissatisfaction with age of personnel working aboard ships at both Phases was signifi-
living conditions on ships predicted strain twelve months later in cantly lower than the average age of personnel working ashore
personnel working and living on ships. These stressors would (33.6 yrs compared to 36.9 yrs). The greater proportion of younger,
appear to be chronic stressors in those exposed to them e if present more junior ranked personnel on ships may be a partial explanation
in the work environment of healthy individuals they increase like- as to why autonomy and control only predicted strain in those
lihood of strain in the future. These findings exhibit some agree- working on ships, as these younger personnel have less authority
ment with those of Sandal et al. (2006) in explaining the causes of and less autonomy and control than older more senior ranked
strain in people working in isolated and extreme environments. personnel.
Satisfaction with the work environment has long been known A second limitation is that the models predict only a small
to be a component of job satisfaction (Sundstrom, 1986). The main percentage of the variance in strain at Phase II. However, this might
areas of dissatisfaction with the environment on ships are main- be expected given that the day to day demands on personnel,
tainability and ease of cleaning of equipment, the air quality, noise, particularly those serving on ships, will vary greatly depending on
crowding, lack of privacy and the provision of sanitary facilities whether the ship is at sea or alongside, or whether it is partici-
(Dew et al., 2007). The present findings suggest that the physical pating in exercises or is actively deployed. Also, several of the Phase
environment in warships may, in some individuals, act as II regression models will have had less statistical power than the
a chronic occupational stressor. From Table 1 it can be seen that Phase I models because of the need to excluded Phase I strain cases
the nine-item WWBQ scale used to measure satisfaction with and personnel whose role had changed, thus reducing the sample
living conditions has a high value of alpha (0.93). This suggests size.
that there is no particular aspect of the physical environment on In general, questionnaire studies of military personnel tend to
ships that is the source of dissatisfaction. The scale has items yield low response rates of 30e50% (e.g. the 2009 UK Armed Forces
concerning: vibration exposure; noise; privacy; crowding; provi- Continuous Attitude Survey yielded a response rate of 32%, Deputy
sion of sanitary facilities; ease of cleaning; maintainability; space Chief of Defence Staff, 2009) so there is always a challenge to
for personal effects and indoor climate, yet it has high internal produce instruments that combine brevity (to encourage
consistency, suggesting that it is both the constraints of life aboard a response) with psychometric robustness.
and the intrusiveness of the physical and social surroundings that
some people find stressful. It would seem that those who are 4.4. General discussion
sensitive to the constraints of such environments do not become
habituated to them over time and are more likely to develop strain Previous studies of NS personnel have shown that, for most
when exposed to the environment for a long time (e.g. twelve Naval personnel, serving on ships is no more stressful than serving
months). ashore, as evidenced by GHQ-12 prevalence rates and strain scores.
For those working ashore at both Phases, the best psychosocial The present findings are of relevance to stress management policy
predictor of strain 12 months in the future was workefamily and to ergonomics because they demonstrate that the causes of
conflict. For those at sea, workefamily conflict did not predict strain in a specific occupational group differ depending on the
future strain. Sandal et al. (2006) has pointed out that when people environment they work in and depending on the time scale over
have to work in isolated or extreme environments, they are often which the occupational demands operate. There are many argu-
exposed to unusual or intense stressors that are not often found in ments for incorporating ergonomics into the design processes of
daily life. At the same time, however, they are completely isolated systems and traditionally these arguments focus on outcomes such
from many of the routine demands of daily life. Thus, the total as physical health and safety. The study presents tentative evidence
demands acting on personnel deployed at sea may be no more than to support an ergonomics approach to ship design by demon-
those experienced by personnel working ashore e it is the nature of strating that ship habitability is important for the mental health of
the demands that differs, not the total demand. This explanation is those aboard and not just for comfort and convenience. Habitability
in accord with the data in Table 4 e the GHQ-12 scores in the of ships is likely to be increasingly important in future with the
different deployment groups do not differ significantly. design of more highly automated ships with fewer people onboard.
It is not surprising that, in the groups who changed location Not only is it difficult and time consuming to extract personnel who
between Phases (moving from ship to shore or shore to ship) only cannot cope with such environments, but with ‘leaner-manned’
personal factors predicted Phase II GHQ-12 score, as having moved vessels, personnel are more specialised and therefore more difficult
to a new job, personnel were no longer exposed to the occupational to replace. However, these points should be treated with caution,
stressors measured at Phase I. given that the main focus of the questionnaire measure used in this
R.S. Bridger et al. / Applied Ergonomics 42 (2011) 710e718 717
study was not ship environments. The measure of ship habitability 12 months. For those who changed from working on a ship to
used in the present study was crude, with only 9-items measured working ashore, and vice versa, only personal factors predicted
on five-point Likert scales and more detailed studies are called for future strain.
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- 5. Conclusions
tion. This is almost certainly because many personnel exhibited
change in GHQ-12 score over time (Fig. 1) possibly due to fluctua- This study has identified some important psychological and
tions in work demands. Amongst military personnel, there is some physical variables in the prediction of current strain and future
evidence that personnel recover quickly from high levels of strain if strain in Naval personnel. In particular, the role of satisfaction with
moved to a less stressful role. Rona et al. (2008) presented evidence living conditions aboard ships in the prediction of future strain is
that mental health problems were more likely in personnel highlighted. Several occupational stressors from general models of
deployed for 13 months or more in a three-year period than those occupational stress were found to predict future strain in
deployed for shorter periods. The number of deployments in the personnel on ships (lack of autonomy and control and dissatis-
same period was not associated with mental health problems faction with living conditions, particularly the lack of privacy),
because the individual deployments were shorter and provided suggesting that these are chronic stressors that can have adverse
more opportunity for recovery. Thus, it would seem that the effects one year later. This is an important finding in applied
varying nature of NS working life might protect individuals from ergonomics because it demonstrates that design decisions taken
chronic strain. Chronic exposure to isolated and extreme environ- early, that may be difficult to reverse later, can lead to long-term
ments is often avoided by the way deployments are organised. problems to which personnel do not adapt. Personal factors and
Indeed, Hotopf et al. (2006) found few differences in the health of problems outside work predicted strain in personnel working
regular UK military personnel who had deployed to the 2003 Iraq ashore.
war and those who had not. According to Rona et al., when chronic
exposure does occur (due to personnel shortages, for example)
References
psychological strain results. It is noteworthy that the present data
indicate that it is not just the strain levels that change over time, but Bosma, H., Peter, R., Siegrist, J., Marmot, M., 1998. Two alternative job stress models
the chronicity and nature of the stressors to which personnel are and the risk of coronary heart disease. American Journal of Public Health 88,
68e75.
exposed.
Bridger, R.S., Kilminster, S., Slaven, G., 2007. Occupational stress and strain in the
The study provides some evidence for the distinction between Royal Navy 1999 and 2004. Occupational Medicine 57, 92e97.
acute and chronic stressors. Chronic stressors may provoke no Bridger, R.S., Brasher, K., Dew, A., Kilminster, S., 2008. Occupational stress and strain
reaction initially but cause strain symptoms to develop in people in the Royal Navy 2007. Occupational Medicine 58, 534e539.
Bridger, R.S., Brasher, K., Dew, A., Munnoch, K., Kilminster, S., 2009. Chronics and
exposed to them. In the present study, the living conditions acute psychological strain in Naval personnel. Occupational Medicine 59,
aboard ships, particularly the lack of privacy, are a chronic 454e458.
stressor that operates over a 12-month time period, indicating 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
that, some personnel at least, do not habituate to these force. Occupational Medicine 4, 256e264.
deprivations over 12 months. The link with Phase II strain Cooper, C.L., Marshall, J., 1976. Occupational sources of stress: a review of the
demonstrates that a poor environment with little privacy is literature relating to coronary heart disease and mental ill-health. Journal of
Occupational Psychology 49, 11e28.
anxiety-provoking in some people. This is a novel finding in DCDS(Pers), 2009. Armed Forces Continuous Attitude Survey. http://www.mod.uk/NR/
maritime research but is in accord with studies of office workers. rdonlyres/54D61DA9-C1D0-4B9D-8B89.84C05193F098/0/afcas08_main_rpt.pdf.
Oldham and Fried (1987), for example, found that adverse reac- Demerouti, E., Bakker, A.B., Nachreiner, F., Schaufeli, W., 2001. The job
demandseresources model of burnout. Journal of Applied Psychology 86,
tions were more common in employees working in offices that
499e512.
they rated as dark, with few enclosures surrounding employees’ Dew, A., Brasher, K., Bridger, R.S. 2007. Naval Service Stress Survey 2007: Supple-
work areas and when employees were seated close to one mentary Report on Ship Habitability, shore-based Accommodation and Strain.
Unpublished MoD Report.
another, with many employees occupied an office. All of these
Dollard, M.f., Winefield, A.H., Winefield, H.R., 2003. Occupational Stress in the
workspace characteristics of offices share some similarities with Service Professions. Taylor and Francis, London.
the deprivations of the living environment on ships. Thus, lack of Goldberg, D., Williams, P., 1988. A User’s Guide to the General Health Questionnaire.
privacy on ships is an anxiety-provoking stressor in some people NFER Nelson.
Holmes, D., Rahe, J., 1967. The social readjustment scale. Journal of Psychosomatic
and not just a source of dissatisfaction. In the language of Research 11, 213e218.
Demerouti et al. (2001), coping with the demands of life onboard Hotopf, M., et al., 2006. The health of UK military personnel who deployed to the
ship is a process of energy depletion that places personnel at risk 2003 Iraq way: a cohort study. The Lancet 367, 1731e1741.
Karasek, R.A., Theorell, T., 1990. Healthy Work: Stress, Productivity and the
of strain. On ships, the resources to meet these demands, Reconstruction of Working Life. Basic Books, New York.
particularly the possibility of providing additional resources in Kilminster, S. Bridger, R.S. 2007. Modified work and Well-Being Questionnaire for
the form of support or relief, are limited, so continued exposure Cohort Study of Stress in the Naval Service: Psychometric Validation. Unpub-
lished MoD Report.
leads to strain. Ultimately, it would be expected that continued Marmot, M., Brunner, E., 2004. Cohort profile: the Whitehall II Study. International
exposure would lead to disengagement and lowered perfor- Journal of Epidemiology 34, 251e256.
mance, placing the operational capability of the vessel at risk. A Marmot, M.G., Davey-Smith, G., Stansfield, S., Patel, C., North, F., Head, J., 1991.
Health inequalities among British civil servants. Lancet 337, 1387e1393.
question for further research in missions where personnel are Miller, G.E., Chen, E., Zhou, E.S., 2007. If it goes up, must it come down? Chronic
required to work in isolated and extreme environments is to stress and the hypothalamic-pituitary-adrenocortical axis in humans. Psycho-
determine the time course of energy depletion, the optimum logical Bulletin 133, 25e45.
Oldham, G.R., Fried, Y., 1987. Employee reactions to workspace characteristics.
periods for deployment and recovery and the scope for amelio-
Applied Psychology 72, 75e80.
ration through ergonomic design. Rona, R.J., Fear, N.T., Hull, L., Greenberg, N., Earnshaw, M., Hotopf, M., Wessely, S.,
Finally, the paper provides evidence for the validity of the 2008. Mental health consequences of overstretch in the UK armed forces:
WWBQ and the constructs it purports to measure. Several first phase of a cohort study. British Medical Journal. doi:101136/bmj/
3927458572BE.
psychosocial stressor variables were associated with strain in those Salonen, P.H., Arola, H., Nygard, C.H., Huhtula, H., 2008. Long term associations of
whose work environment and work role remained unchanged for 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 Siegrist, J., 1996. Adverse health effects of high effort/low reward conditions. Journal
environments. Reviews in Environmental Science and Biotechnology 5, of Occupational Health Psychology 1, 27e41.
281e296. Slaven, G., 2002. Work-related Strain in the Naval Service. University of East Lon-
Schaufelli, W., Baker, A.B., 2004. Job demands, job resources and their relationship don, Dissertation, London.
with burnout and engagement: a multi-sample study. Journal of Organisational Sundstrom, E.D., 1986. Workplaces. Cambridge University Press, England.
Behaviour 25, 293e315. Watson, D., Clark, L.A., Tellegen, A., 1988. Development and validation of brief
Schei, E., 1994. A strengthening experience:Mental distress during military service. measures of positive and negative affect: the PANAS scale. Journal of Personality
Social Psychiatry and Psychiatric Epidemiology 29, 40e45. and Social Psychology 54, 1063e1070.