Revue européenne de psychologie appliquée 59 (2009) 113–122

Original article

Developing a French version of the Occupational Stress Indicator (OSI)

Validation d’une version française de l’Occupational
Stress Indicator (OSI)
D. Steiler a,∗ , B. Paty b
a Département management et comportement, Grenoble École de management, BP 127, 12, rue Pierre-Sémard, 38003 Grenoble cedex 01, France
b Laboratoire de psychologie appliquée EA 3793, université de Reims Champagne-Ardenne, 57, rue Pierre-Taittinger, 51096 Reims cedex, France
Received 6 November 2006; received in revised form 31 October 2008; accepted 3 November 2008

Abstract
This research presents the validation of a French version of the Cooper, Sloan and Williams’ Occupational Stress Indicator – OSI – (1988), which
consists of seven specific scales: Sources of Pressure, Type A Behaviour, Locus of Control, Coping Strategies, Mental Health, Physical Health and
Job Satisfaction. After a translation and back-translation procedure, committee evaluation and a pilot trial on 20 managers, 290 volunteer managers
provided data to test its psychometric qualities. Confronted with the difficulty of reproducing and validating the original factorial structure and the
weakness of some subscales reliability results, a complementary procedure, including exploratory and confirmatory factor analysis, was used to
further improve the psychometric qualities of the French OSI. Even though the main seven-factor structure was maintained, little similarity existed
between the published 25-subscale score keys of the original version and the more parsimonious 12-subscale structure, which the present study
brought to light.
© 2008 Elsevier Masson SAS. All rights reserved.
Résumé
Cette recherche présente les travaux de validation d’une version française de l’Occupational Stress Indicator de Cooper, Sloan et Williams –
OSI – (1988), qui est constitué de sept échelles spécifiques : sources de pression, comportement de Type A, lieu de contrôle, stratégies de coping,
santé mentale, santé physique et satisfaction au travail. À l’issue d’une procédure incluant une traduction et traduction en retour, l’évaluation d’un
comité d’experts ainsi que le test de l’outil auprès d’un groupe pilote de 20 managers, 290 managers volontaires ont permis de recueillir les données
nécessaires à la validation de ses qualités psychométriques. Confronté à la difficulté de reproduction et de validation de la structure factorielle
d’origine ainsi qu’à la faiblesse de certaines sous-échelles en termes de fidélité, une procédure complémentaire a été mise en œuvre. Elle incluait
une analyse factorielle exploratoire et confirmatoire dans le but d’améliorer les qualités psychométriques. Si la constitution finale de cette version
française conserve les échelles initiales, la structure factorielle de chacune d’entre elles a considérablement évolué pour passer de 25 sous-échelles
à une version plus parcimonieuse de 12 sous-échelles.
© 2008 Elsevier Masson SAS. Tous droits réservés.

Keywords: OSI; Validity; Reliability; Work stress; France

Mots clés : OSI ; Validité ; Fidélité ; Stress au travail ; France

1. Introduction
For more than two decades, the results of many studies have
suggested that occupational stress is a major concern for com-
panies on an individual and organisational level (ILO, 1992;
∗ Corresponding author.
E-mail address: dominique.steiler@grenoble-em.com (D. Steiler).
Robinson and Inkson, 1994; Tarquinio, 2008; Darr and Johns,
2008). If, as stated by Verborgh (1992),“workplace assessment is
the first stage of any action being undertaken in order to improve
the work environment” (pv), the measurement of the risks of pro-
fessional stress represents one of the essential conditions of its
success.
Although French companies are beginning to acknowledge
stress as a key strategic issue, few studies actually examine

1162-9088/$ – see front matter © 2008 Elsevier Masson SAS. All rights reserved.
doi:10.1016/j.erap.2008.11.001
114 D. Steiler, B. Paty / Revue européenne de psychologie appliquée 59 (2009) 113–122
the whole concept (Godin et al., 2006). In this context, it is
important for French occupational psychologists and physi-
cians, managerial executives, stress management consultants
and even government policy-makers to be able to under-
stand and accurately assess the problem of occupational stress,
which constitutes the first step in the overall occupational
stress management process. Unfortunately, French-speaking
researchers seem to have a much more limited panel of
tools than their English-speaking colleagues. A first review
of the literature on the measurement of professional stress
led to the conclusion that there is, at present, no French
language composite tool that measures occupational stress.
Therefore, in such a situation and as recommended by Vallerand
(1989), the most relevant solution seems to be to work on
the transcultural validation in French of questionnaires that
exist in English and to extend the range of tools available,
while improving the quality and accuracy of intercultural
studies, paying careful attention to bias and equivalence (Van
de Vijver and Tanzer, 2004).
The aim being essentially to provide an overall measure-
ment tool in French, our choice tended towards composite stress
measurement tools, in pursuit of work by Spielberger (1994),
who stated that most of these tools included stressors, strains,
personality characteristics and coping skills.
Although validation of its factorial structure appeared to
be necessary (Hurrell et al., 1998, p. 371), the occupational
stress indicator (OSI) seemed to fulfill the desired criteria of
a wide scope of work stress variables; it had some mediators–
moderators and stress outcome measurements, an excellent nor-
mative database and had already been validated in at least five
foreign languages: Portuguese (Cunha et al., 1992), Chinese (Lu
et al., 1997), Brazilian Portuguese (Swan et al., 1993), Dutch
(Evers et al., 2000) and Bulgarian (Russinova et al., 1997).
Consequently, the goal of this article is to describe the psy-
chometric assessment of the French version of the OSI, which
led to a substantial revision of the scales, very close in their fac-
torial structures to the English version proposed by Lyne et al.
(2000).
2. The OSI
As presented by the authors, the OSI was developed more
as an indicator than as a test. It was built to provide a broad
measurement of work stress in order to help organisations and
collaborators plan organisational change and/or individual adap-
tation. The design goal of the OSI was simple and pragmatic:
“to produce an instrument that could provide a comprehensive,
integrated, relevant and accurate measure of occupational stress”
(Zalaquett and Wood, 1997).
The OSI was used in a wide variety of situations and presently
constitutes a normative database from a survey of more than
20,000 people. As described by Evers et al. (2000), there are
three main objectives to its use:
• to provide managerial boards with information for making
work stress management decisions;
• to evaluate stress management programs;
• finally, to conduct scientific research.
As an operational tool, the OSI has investigated many differ-
ent occupational groups; for example, nurses (Baglioni et al.,
1990), anaesthesists (Cooper, 1999), police officers (Biggam
et al., 1997; Kirkcaldy and Cooper, 1992), IT personnel (Lim
and Teo, 1996), senior civil servants (Bogg and Cooper, 1995;
Renault de Moraes et al., 1993). It has also been useful in evalu-
ating the results of Employee Assistance Programs (Highley and
Cooper, 1994) and in observing stress in managerial situations
like downsizing actions, for example (Flude, 1994).
Based on the Cooper and Marshall stress model (1976) and as
observed by Spielberger (1994), it “incorporates some aspects
of the P-E Fit (French and Caplan, 1972; French et al., 1982)
and Demand–Control models (Karasek, 1979; Meier et al.,
2008), and Lazarus’s Transactional Process Theory” (Lazarus
and Folkman, 1984). The basic conceptual relationship between
the component scales of the OSI highlights the strain-stress
approach of this assessment tool, as adapted from Cooper et
al. (1988). Integrating the three main components of this rela-
tionship, the figure firstly puts forward the model hypothesis of
a predictive relationship between job pressure and stress-related
outcomes, that is, health and job satisfaction. Secondly, it sup-
poses that individual characteristics (Type A behaviour pattern
and Locus of Control) moderate this relationship. Finally, the
model also stresses the moderating impact of coping strategies
on stress-related outcomes.
The main limitation of the model, however, is its lack of
accuracy and indications with respect to relations between the
variables. Furthermore, it bears no association with a specific
psychological model, perhaps due to the authors’ desire to com-
bine individual and collective aspects. Even though the OSI
model is not clearly positioned in either of the two main psycho-
logical stress models, it depicts a very linear model that comes
closer to interactional theory than transactional theory.
Item and factor analysis on a sample of British managers
led the authors to build a seven-scale, 25-subscale and 167-
item indicator, rated on a six-point Likert-type scale. The first
scale, Sources of Pressures (61 items) and the three scales
assessing moderating variables, Type A (14 items), Locus of
Control (12 items) and Coping Strategies (28 items), represent
the independent variables. The last three questionnaires, Mental
health (18 items), Physical Health (12 items) and Job Satisfac-
tion (22 items) represent the dependent variables.
The second limitation of the OSI is that the reliability of 15 out
of the 25 subscales appeared to be really low. More specifically,
three scales presented very problematic results: Type A, Locus
of Control and Coping Strategies, with in some cases, the split-
half as low as 0.20 or 0.10 (Cooper et al., 1988). Further studies
(Cooper and Williams, 1991; Davis, 1996; Kahn and Cooper,
1991; Robertson et al., 1990; Williams and Cooper, 1997) on
the English or even non-English versions (Cunha et al., 1992;
Lu et al., 1995; Swan et al., 1993) yielded better reliability scores
on larger samples, but they all still demonstrated weaknesses for
the same three scales.
Thirdly, even though the OSI is useful in a corporate context,
given the scope of information collected, its factorial structure
 D. Steiler, B. Paty / Revue européenne de psychologie appliquée 59 (2009) 113–122
presents some weaknesses. Albeit the seven principal factors
seem consistent, so much cannot be said for the second level
factors (i.e., subscales). Furthermore, at least two scales are not
built on clear factor analysis: Sources of Pressure and Coping
Strategies.
The Sources of Pressure scale was constructed from the lit-
erature review, because of a low ratio between the number of
respondents and the number of items (Cooper et al., 1988).
Davis (1996) analysed it later to find a four-factor structure,
while Evers et al. (2000) revealed a strong one-factor structure
using principal axis factoring.
The Coping Strategies scale was initially content analysed
because of the redundancy problem of a dominant single fac-
tor solution (Cooper et al., 1988). No other study has since
confirmed its structure. Kirkcaldy and Furnham (1999) found
a two-factor structure, as did Lyne et al. (2000), but Luo Lu et
al. (1995) found a three-factor structure.
The Job Satisfaction scale was also factor analysed and pro-
duced a three-factor solution in Davis’ research and a two-factor
solution in Evers et al.’s study (2000).
Finally, Lyne et al. (2000) built a complex methodology by
extracting potential factorial structure solutions to test the repli-
cability of the OSI subscales. The results were unambiguous: no
solution reproduced the original OSI version, as it stood.
Notwithstanding the OSI authors’ insistence that it was more
of an indicator than a test, the low reliability of scores and the
lack of accuracy in the factorial structure nevertheless begged
for improvement in order to maintain the credibility of the OSI
and propose a valid French version. Additionally, the unidimen-
sionality of the scales was checked with confirmatory factor
analysis.
3. Method
3.1. Procedure
With the permission of the publisher (NFER-Nelson), the OSI
underwent a back-translation procedure and then three bilingual
experts were consulted to give detailed comments on face valid-
ity, content equivalence and conceptual equivalence of the draft
version developed by Jeanrie and Richard (1997). A pilot trial
was then initiated in real conditions with a small group of volun-
teers (20 white-collar workers) to ensure that the items would be
understood by the target sample. Data were collected in the pres-
ence of researchers by distributing a pencil and paper version to
volunteers. Confidentiality was guaranteed through bar-coded
forms.
Potential participants were first informed by a presentation
letter defining the research context and objectives, procedure,
confidentiality and informing them that they could attend a
debriefing meeting. They were also invited, on a voluntary basis,
to attend the research presentation and data-collection meeting
within their respective companies. The volunteers attended the
meeting and completed the questionnaires by way of consent (no
signature was required). All of the data were finally computed
on SPSS 10.1® .
115
3.2. Participants
From four companies, 290 volunteer managers out of
344 participated in the survey, giving a response rate of 84.30%.
The sample comprised 23% women and 77% men, ranging from
25 to 63 years. The mean age was 45.27, with a standard devi-
ation of 6.44. Three quarters of the managers (n = 221, 76.2%)
were between 40 and 53 years old. Sixty-two board managers,
119 managers and 109 team managers represented 21.4, 31 and
37.6% of the population, respectively.
3.3. Analysis
The procedure was set up in two stages.
Since there were ambiguities and limits in the original ver-
sion of the psychometric results, the first stage integrated an
internal consistency measure with Cronbach’ alpha (Cronbach,
1951) and an attempt to reproduce the original factorial struc-
ture through principal axis factoring (PAF-PA2) with oblimin
rotation, as presented in the OSI manual (Cooper et al., 1988,
p. 57).
The difficulty of reproducing and validating an original facto-
rial structure identical to Lyne et al. (2000) and the weakness of
some subscales reliability results, especially within the Type A,
Locus of Control and Coping Strategies subscales (Steiler
and Cooper, 2004), led to a complementary procedure includ-
ing exploratory and confirmatory factor analysis with AMOS
(Arbuckle and Wothke, 1995). Only the results of this second
procedure are presented hereafter.
In this second stage and before beginning the computing
factorisation, a “Kaiser-Meyer-Olkin” sampling adequacy pro-
cedure (KMO) was used to diagnose whether the data contained
enough correlations to warrant factorisation (Kaiser, 1974).
Only thereafter, in order to produce factorial options, were all
of the scales factor-analysed using four-factor extraction tests.
All these tests are proposed by Zwick and Velicer (1986) as the
most common rules to determine the number of components and
are described by Lyne et al. (2000): the Velicer Minimum Aver-
age partial (MAP) test (Velicer, 1976), Armor’s theta (Yaffee,
1997), scree test (Cattel, 1966) and the Kaiser-Guttman eigen-
values greater than one criterion (Kaiser, 1960). As presented by
Velicer (1976, p. 326), the MAP test is one of the most appropri-
ate rules used in the first stage of factor analysis. The MAP test is
a stopping rule that determines the number of common factors in
a correlation matrix. Armors’ theta (Armor, 1974; Yaffee, 1997)
is a factor reliability ratio. A factor cannot explain a significant
proportion of a variance if theta is below 0.50. The Scree test is
a subjective factor assessment based on a graph of eigenvalues;
it was only used if the other tests failed to produce a satisfactory
factorisation structure or were difficult to interpret. Finally, the
Kaiser-Guttman criterion was used to ensure that only factors
with an eigenvalue greater than 1.0 were accepted.
At the end of this part of the procedure, each OSI scale had
up to four factorial options to test.
In the third step, to reduce the optional factor solutions
provided by the exploratory factor analysis and determine the
best options, each option was subjected to minimum residual
116 D. Steiler, B. Paty / Revue européenne de psychologie appliquée 59 (2009) 113–122
(Minres), which is a way of performing factor analysis by the
least square (Jöreskog, 2003). As presented by Lyne et al. (2000,
p. 201), “Minres is a method of factor analysis that does not rely
on iterative item communality estimation procedures and has the
feature of extracting common factors that minimize the residual
correlations between all items, having extracted n factors”. All
options, including original OSI score keys, were examined. As
it was assumed that subscales could be correlated, oblique rota-
tion was applied. Solutions were rotated using oblimin rotation
with a Delta parameter that swept from 0 to − 10.5. Pair-wise
data deletion was used in each analysis. Acceptable loading was
fixed at greater than 0.30.
When one or two factorial solutions appeared correct for each
scale, the confirmatory factor analysis tested their soundness in
terms of factorial structure. Its goal was to assess whether the
proposed models “fit” the data in a deductive approach.
To assess fit, it is generally recommended to use several
indexes per construct (Hu and Bentler, 1999). Five measures
were used to assess fit:
• chi-square/degree of freedom ratio (relative chi-square, chi-
square/df);
• Goodness of Fit Index (GFI);
• Adjusted Goodness of Fit Index (AGFI);
• Root Mean Square of Approximation (RMSEA);
• and Standardized Root Mean Square Residual (SRMR).
A chi-square/df ratio less than 3 seems to be indicative of an
acceptable model (Arbuckle and Wothke, 1995, pp. 399–400).
The closer GFI and AGFI are to 1.0, the better the fit of the
model to the data. Even if no formal rules exist at the present
time, the levels of acceptability of GFI are increasingly strict
in scientific literature. This is particularly true in management
and economics, where a fit index is accepted as good above 0.90
for GFI and AGFI (Croutsche, 2002; Roussel et al., 2002). The
RMSEA should be about 0.06 or less for an acceptable fit and
SRMR should be about 0.08 or smaller for an acceptable fit (Hu
and Bentler, 1998).
4. Results
4.1. Exploratory factor analysis: factor extraction tests
Following Norusis’ scale (Norusis, 1993), all KMO indexes
were good from “medium” (0.69) to “marvellous” (0.90).
Results of the principal component factor extraction tests
were very close to Lyne et al.’s results (2000, p. 203) for the
same work. The number of factors suggested by the four dif-
ferent factor extraction tests will be illustrated hereafter in the
respective sections of each factor.
Because of problems in interpreting Sources of Pressure,
MAP and theta factors, the Scree test was also used for this
scale only, in spite of its subjectivity and tendency to overes-
timate factors. The Kaiser-Guttman criterion was used just to
ensure that only factors with eigenvalues greater than 1.0 were
accepted.
4.2. Exploratory factor analysis: factor options assessment
4.2.1. Sources of Pressure
The Map test indicated a six-factor solution, identical to the
original OSI, while the Armor’s theta indicated five. Both solu-
tions were thereafter rejected by the Minres procedure.
The Scree test was then considered as it showed a marked
breakpoint at three factors. After the Minres analysis, which
supported the solution, items with double loadings were elim-
inated in order to finally achieve a solution with respectively
20, 11 and 16 items for each factor. The first factor presented
pressure-related job characteristics and was therefore called
“Job Pressures”. The second factor was clearly interpreted as
“Managerial Sources”. The third factor, which was slightly
more difficult to interpret, with some items centred on the
home/work relationship and others more centred on personal
achievement, was called “Home/work relationship and personal
achievement”. This factor could be considered as hypothetical
and at an early stage of development.
4.2.2. Type A
Factor extraction tests gave two options for the Type A scale:
a one- and a three-factor solution.
The Minres factorisation showed poor correspondence
between the OSI score keys and the three-factor solution. For
example, seven unique loadings on one factor attracted items
from each of the three OSI subscales. This observation showed
that Minres factor extraction failed to support the three-factor
solution and it was therefore rejected.
The second option was the one-factor solution from the MAP
test and Armor’s theta. This solution was supported by Minres
factor extraction, where only three items failed to load.
4.2.3. Locus of Control
From the two extracted test options (one- and three-factor
solutions), the original three-factor solution was tested first.
After Minres, there appeared to be very little correspondence
between the original OSI subscales and the three-factor solution
identified, which led us to reject the solution.
The one-factor solution from the MAP test and Armor’s theta
was then tested. Even though four items failed to load, the solu-
tion was still supported by the Minres factor extraction test.
4.2.4. Coping Strategies
Factor extraction tests gave three options for the Coping
Strategies scale.
In the six-factor solution, a lot of weaknesses, such as the
first factor that presented six unique loadings distributed among
four of the six OSI subscales, resulted in its rejection.
The three-factor solution was then tested and accepted, at
this stage of the study, with one problem-focused cognitive
methods factor, one problem-focused behavioural and organi-
sational methods factor and one emotion plus avoidance plus
social support-focused factor.
Finally, MAP and Armor’s theta factor extraction tests
yielded a two-factor solution. Examination of this solution led
to a first factor, centred on the problem and a second factor,
 D. Steiler, B. Paty / Revue européenne de psychologie appliquée 59 (2009) 113–122 117

centred on emotion. The Minres factor extraction test supported
this solution.
After this stage, confirmatory factor analysis was expected to
assess which of the two retained models best fit the data.
After this stage, the confirmatory factor analysis was expected
to assess which of the three proposed models best fit the
data.
4.3. Confirmatory factor analysis

4.2.5. Mental Health
For the Mental Health scale, the factor extraction tests gave
a single-factor solution similar to that found in the OSI manual.
This scale was supported by Minres factor extraction, which
showed that 17 of the 18 Mental Health items load on a single
factor. As supposed in the pilot trial phase, some of the Mental
Health scale items seemed to be complex, subsequently leaving
room for improvement.
4.2.6. Physical Health
For the Physical Health scale, the one-factor solution result-
ing from the extraction tests was also consistent with the unique
OSI option. All 12 items loaded on a single factor and the scale
was supported by Minres factor extraction.
4.2.7. Job Satisfaction
Factor extraction tests gave four options for this scale (1, 2,
3 and 5-factor solutions). As for most of the other scales, the
attempt to replicate the original five-factor version failed.
Even though the one-factor solution was accepted, it was
considered to have a limited operational purpose, since in an
operational context, it would only give an overall vision of
satisfaction.
Given that the two-factor solution, chosen from the theta
results to compare with Lyne’s solution (2000, p. 220), presented
almost exactly the same structure with a first extrinsic satisfac-
tion factor as with a second intrinsic satisfaction factor, it was
supported by Minres analysis.
We then proceeded to analyse the three-factor solution pro-
posed by MAP. In a first attempt to qualify the three factors,
observation led us to define the first factor as “Organisational
Extrinsic Satisfaction”, the second as “Person/Job Fit Satisfac-
tion” and the third as “Personal Rewards Satisfaction”. While the
first factor measured the external conditions of satisfaction, the
second measured motivational conditions, and the third assessed
the results or “rewards” obtained from the expectations, this
three-factor solution was also supported by Minres analysis.
To observe the results of this step of the study, the table
below presents the confirmatory analysis results of the trans-
lated French version, based on the original UK version’s factorial
structure (left side of the table, OT-OSI) and the results of the new
French version’s factorial solution from the exploratory analysis
results (right side of the table, OSI-F) (Table 1).
As shown, the OSI-F structure almost systematically presents
better goodness-of-fit indexes than the original version, even if
fit indexes are not strictly those recommended by the literature.
Concerning the seven scales of the OSI-F, results appear better
and all items are significant at 0.01.
The OSI-F Sources of Pressure model fit indexes show that
only Chi2 /df and SRMR can be interpreted as acceptable. GFI,
AGFI and RMSEA are not within the accepted level of goodness-
of-fit, but are better than the original translated OSI’s (OT-OSI)
results. Even if the OSI-F approaches an acceptable level and
although some discrepancy can be expected between different fit
indexes (i.e., relative versus absolute fit), the magnitude of this
difference suggests enough misspecification to require a review
of the model. Observation of estimated parameters shows that
17 items are below 0.50. In an attempt to improve the Sources
of Pressure scale’s goodness-of-fit indexes, these items were
deleted. Although the results presented slightly better indexes,
they still did not reach the accepted level. The decision was taken
to reintegrate these items, firstly because they finally contributed
to other good psychometric assessment criteria and secondly, to
avoid drifting too far away from the original version. In conclu-
sion, this scale could not be confirmed. Further research will be
required to pursue its development.
Type A scale fit indexes of the OSI-F version could be con-
sidered as acceptable, even if AGFI and RMSEA are slightly
below the usually accepted level; thus, confirming the model
as a one-factor model. For this scale, three items had a smaller
than 0.50 estimate parameter. The deletion of these items gave a
result of greater than 0.90 AGFI, which is an improvement. But
nevertheless, like Sources of Pressure, the number of items and
the similarity to the original version were preferred to a more
precise confirmatory model.

Table 1
Confirmatory factor analysis: fit indexes for OSI scales in the original UK structure and in the new French version structure.
Original translated OSI (OT-OSI) French OSI (OSI-F)

Chi2 /df GFI AGFI RMSEA SRMR Chi2 /df GFI AGFI RMSEA SRMR

Sources of Pressure 2.54 0.59 0.56 0.08 0.10 2.34 0.71 0.68 0.07 0.08
Type A 3.13 0.88 0.83 0.09 0.09 2.78 0.91 0.87 0.08 0.07
Locus of Control 2.17 0.93 0.89 0.07 0.07 0.89 0.98 0.97 0.00 0.03
Coping Strategies 2.74 0.76 0.58 0.08 0.08 2.41 0.86 0.83 0.07 0.07
Mental Health 2.37 0.87 0.84 0.07 0.06 2.37 0.87 0.84 0.07 0.06
Physical Health 2.46 0.92 0.88 0.08 0.05 2.46 0.92 0.88 0.08 0.05
Job Satisfaction 3.49 0.77 0.71 0.10 0.08 2.84 0.85 0.81 0.09 0.07

In italics: the best-fit indexes for the presented solutions – see Appendix A for standardised solutions.
118 D. Steiler, B. Paty / Revue européenne de psychologie appliquée 59 (2009) 113–122

Table 2
Reliability of the new OSI-F score keys.
Scale domain Subscale label Number of items Cronbach’s Alpha

Sources of Pressure Job pressures 20 0.90

Managerial sources 11 0.83
H/W relationships and personal achievement 16 0.85
Total 47 0.93
Type A Behaviour – 11 0.70
Locus of Control – 8 0.70
Coping Strategies Problem-focus 10 0.77
Emotion-focus 9 0.72
Total 19 0.76
Mental Health – 17 0.86
Physical Health – 12 0.85
Job satisfaction Organisational extrinsic satisfaction 9 0.83
Person/job fit satisfaction 6 0.86
Personal rewards satisfaction 7 0.82
Total 22 0.92

The Locus of Control structure was confirmed given that the
Locus of Control fit indexes of the French version were above
the more restrictive level of fit and, indeed, showed the best fit
of all of the scales.
For the Coping Strategies scale, two solutions were accepted
after the Minres analyses. The two-factor solution presented
better-fit indexes than the three-factor solution. If some of the
fit indexes were not totally satisfactory (GFI, AGFI, RMSEA),
they were nevertheless better than those of the original version.
In conclusion, even if this scale would have to be improved in
the future, it was accepted in the meantime and the three-factor
model rejected.
Mental Health, with only 17 items in the OSI-F versus 18 in
the OT-OSI, and Physical Health presented the same results in
the original translated version as in the French version. As for
the previous scale, even if they left room for improvement, the
structures were accepted at that level.
For the Job Satisfaction scale, three solutions were accepted
after Minres analysis. Estimated parameters were all above 0.50
except for three items. In an attempt to improve this scale’s
goodness-of-fit indexes, these items were deleted and honed
down to the French structure to 133 items from its initial
136-item shape (Steiler, 2005). The one-factor model was not
confirmed with only SRMR at an acceptable index of fit. The
two-factor solution came closer, but was still too low to confirm
the structure (e.g., AGFI at 0.78). Neither did the three-factor
solution reach the stricter level of goodness of fit, but it did
present fit indexes that were way above those of the original
version.
To conclude this subsection, for each of the analysed sub-
scales, even if the stricter level of goodness of fit were not
reached, they were better for the OSI-F than for the original
translated version. They can be considered as correct and in
accordance with some statistical scientific references (Arbuckle
and Wothke, 1995, pp. 399–400; Browne and Cudeck, 1993; Hu
and Bentler, 1998) but also with criteria used in studies that
focus on questions closely related to the present research topic
(Ingledew et al., 1992; Leiter and Robichaud, 1997).
4.4. Reliability of the French version
Cronbach’s alpha was used to measure the internal consis-
tency of the OSI. A Cronbach’s alpha above 0.70 and below 0.90
can be considered as high, without the side effects of redundancy
or restricted assessment.
Only alphas of scales that were accepted in the confirma-
tory analysis procedure are presented in the following table. As
shown, all the seven total scales are good. Unlike the original
version and other non-English studies, the Type A and Locus
of Control scales have a good reliability coefficient of 0.70 and
all subscales have an internal consistency coefficient over 0.70
(Table 2).
5. Discussion
The previous analyses concluded in a unique final struc-
ture for all scales, suggesting that a composite tool, adapted
to the French context, could be proposed to measure the
main components of the stress process: work stress variables,
mediators–moderators and stress outcomes. Beyond this overall
strength, it is interesting to support the above results by linking
them to the literature.
Concerning the Coping Strategies scale, the two accepted fac-
torial solutions that emerged from Minres analysis were a two-
factor solution and a three-factor solution. The existence of these
two solutions seems relevant in view of Latack and Havlovic’s
work (1992). Indeed, although the two-factor solution, as shown
by these authors, defines a structure based on the notion of “focus
of coping” (problem focus and emotion focus), the three-factor
solution, on the other hand, shows promise of what could become
a scale integrating different coping methods, that they described
as cognitive or behavioural centred, avoidance centred or con-
trol centred and solitary versus collective. However, the results
of confirmatory analysis and reliability analysis of these two
solutions prompted us to choose the two-factor solution, even
though it would certainly benefit from further development in
order to allow a broader assessment of coping methods.
 D. Steiler, B. Paty / Revue européenne de psychologie appliquée 59 (2009) 113–122
In terms of the Job Satisfaction scale, the literature review
seems to favour a three-factor approach, already considered
in a previous study of the OSI (Davis, 1996). In accordance
with Ndoye (2000, p. 442), a job satisfaction analysis facilitates
the understanding of links that exist between the three main
variables of the concept: organisational, professional and per-
sonal. Additionally, for Herzberg (1966) and later for Frances
and Mogenet (1988), there is an important link between motiva-
tion and satisfaction; therefore, it is not possible to be satisfied
without a certain amount of motivation for any one aspect of
the job. Furthermore, the “development of motivations depends
on obtained satisfactions and their circumstances of occur-
rence” (Frances and Mogenet, 1988, pp. 23–24). Campbell and
Pritchard (1976) present pro- and retroactive effects between
motivation and satisfaction. Thus, proactive effects imply that
only people with expectations can be satisfied. Retroactive
effects imply that future motivation (or expectation) depends
partly on personal satisfaction obtained previously in satis-
factory external conditions. Finally, to close the loop, overall
satisfaction depends on obtaining personal satisfaction, which,
in turn, depends on professional expectations in the context of
an organisational environment.
Thus, the selected three-factor model seems to conform to the
authors’ vision. It proposes a more useful and accurate manner of
observing the levels of organisational, professional and personal
aspects of satisfaction.
As presented in the OSI manual, the Sources of Pressure scale
was initially derived from previous studies about work stress,
because their initial sample was too small for factorial analy-
sis. The first step of the present study rectified this omission,
but failed to confirm the structure. While confirmatory analy-
sis did not confirm the retained three-factor solution, even if fit
indexes were better than for the original version, exploratory
methodology led to a more stable structure, albeit not ideal, that
could advantageously stand in for the original scale until further
research provides a new confirmed Sources of Pressure scale.
Instead of the original OSI version three-factor model for
Type A scale with unsatisfactory psychometric results, the
present study concluded with a one-factor reliable and valid
model. Such a one-factor model has previously been accepted
by Luo Lu et al., 1995. Of course it could be argued that a Type A
behaviour pattern is a multidimensional concept. Nevertheless,
from the original pool of items, it was not possible to confirm this
multidimensionality. A future research project might endeavour
to test a new pool of items in order to assess not only an overall
Type A behaviour pattern, but also subcomponents as suggested
by Spence et al. (1987): Achievement Striving versus Impatience
Irritability.
The OSI Locus of Control scale has been questioned in sev-
eral previous studies (Cunha et al., 1992; Davis, 1996; Ingledew
et al., 1992; Robertson et al., 1990) either with regard to its
reliability or even its construct validity. Results of the French
translated Locus of Control scale in its original factor struc-
ture led to similar weak results. Conformity of these results
with the literature was not only reassuring but also a basis
for development. The results obtained in the present study on
the new one-factor structure are partially different. Concerning
119
reliability, no other study reached a good Cronbach’s alpha. In
terms of validity, although Robertson et al. (1990) concluded
with “less impressive” evidence of validity because of poor dis-
criminant validity, our results are slightly better, although not
entirely satisfactory for demonstrating strong validity. Further-
more, contrary to Ingledew et al.’s findings (1992), a one-factor
model was supported by the Minres extraction test and strongly
confirmed through the confirmatory factor analysis (CFA). Of
course, it is a shorter scale, assessing only external Locus of
Control, but as developed by Dubois (1985), previous stud-
ies proposing multidimensionality were not entirely convincing.
Although these results are more satisfactory than those of other
studies in terms of psychometrics, they do not exclude at least
two issues. First, as underlined by Davis (1996), the Locus of
Control scale does not determine whether results assess a per-
sonality trait that mediates the individual’s response to sources of
pressure or if they assess perception of a job environment. A pre-
vious study, focusing on the original OSI scale, concluded with a
“state measure” rather than a “trait measure” and thus, questions
the role of Locus of Control in the theoretical model (Rees and
Cooper, 1992). Similar work is necessary on the current French
structure to determine exactly what this scale assesses or at least
to understand if the weak convergent validity of the scale is not
just an output of an attempt to validate a state (French OSI Locus
of Control scale) using a trait alternative measure (with Rotter’s
test). Second, within the remaining items, interpretation can be
very different for people who are managers and for those who are
managed. Adequately argued by Ingledew et al. (1992), a given
item focusing on board actions could lead to an internal Locus
of Control-oriented response from a manager and an external-
oriented response from people who are managed. Additionally,
people may answer while positioning themselves as either man-
agers or “managed” and thus, give opposite answers. Perhaps, a
first evolution could simply be to improve the Locus of Control
scale’s instructions. As the OSI was developed for managers,
it might be relevant to ask respondents “Considering your job
as a manager, you are required to record the extent to which
you agree. . .”. Pursuant to these remarks and before the pro-
posed developments are undertaken, French-OSI users should
bear in mind that Locus of Control scores must be interpreted
with precaution.
For the Mental Health scale of the original version, the
authors adopted a one factor-model, even if their factor analysis
concluded in a three-factor model. They took this option to avoid
putting future users “in a position of performing a diagnosis”
(Cooper et al., 1988, p. 51) and with the aim of providing insight
into the level of well-being. In the present study, exploratory and
confirmatory analyses also lead to the same one-factor model.
They take the same kind of precautions without conceding the
importance of a psychometric analysis, which would finally be
consistent with Robertson et al.’s results (1990).
The Physical Health scale was simply not subjected to sta-
tistical analysis in the original OSI manual. The authors argued
that the items represent typical physical manifestations of stress
included in psychosomatic measures. As for the Mental Health
scale, they proposed a scale that avoids diagnosis and allows
assessment of a physical index of stress-related complaints.
120 D. Steiler, B. Paty / Revue européenne de psychologie appliquée 59 (2009) 113–122

Despite the fact that this choice seems meaningful from an oper-
ational point of view, as previously, we chose to extract and
confirm a one-factor model that would prove reliable and valid.
For all scales except Type A, confirmed factorial structures
generally account for more than one third of explained vari-
ance. Although there is room for improvement, these results
are in an acceptable range, as presented by other psychologi-
cal assessment tools (i.e., Spielberger’s STAI or Smith’s CISS).
Moreover, it is important to remember that the retained factors
were based on a relatively strict procedure: several extraction
tests, eigenvalue greater than 1.0, validation of the structure
by Minres analysis, acceptance by confirmatory analysis and
good reliability. Furthermore, while explained variance results
of the original OSI scales are higher, they do not go so far as
to represent reality, since the original OSI structure was not
confirmed.
While the exploratory analysis procedure of this study aims to
reduce potential factor structure and to determine which models
best fit the data, it is paramount to remember, when interpreting
confirmatory analysis result, that several models can be ade-
quately confirmed. Accepted models are best within a defined
procedure and because this work is the first attempt to confirm
OSI structure, further study and development will certainly help
in discovering a better-fit solution.
6. Strengths and limitations
trial lowered the bias resulting from the translation of certain
idiomatic expressions that could have led to misunderstanding
in the French context. Thus, despite the fact that many precau-
tions were indeed taken to limit eventual sources of difference of
structure resulting from methodological (for example, a sample
size that should, strictly speaking, have been larger for this type
of analysis) or cultural problems, at this stage we cannot entirely
exclude their impact.
The present study aimed to validate a French version of the
OSI and results conclude that a difference of structure exists,
which can partially be explained by problems of a psychometric
order, but which is probably underpinned by cultural factors for
which we can only envisage hypotheses.
Considering these aspects and although the factorial structure
quality of one scale has to be strengthened, major psychometric
improvements have made this 133-item version available for
further studies in the French context1 .
Finally, given the limits of this work, there is a good
case for reporting the present results, if only to stimulate
other French-speaking researchers to attempt to replicate our
findings.
Appendix A. French OSI score key using item number
from the original UK version

Despite the questionable size of the sample, which would Scales Items Number of Items
have benefited from being bigger, the very high response rate of
Sources of pressure (SP) Total scale 47
the study was nonetheless reassuring.
Job pressures (SP1) 2 3 4 6 8 9 10 11 12 20
One should also remember that the difference in factorial 15 16 18 21 22 24 27
structure could be attributed to a number of methodological or 30 35 53 59
cultural causes. In terms of methodology, the main causes that Managerial sources (SP2) 5 14 20 31 32 44 50 11
could lead to a difference in structure are set forth by Schaffer 51 55 56 57
H/W relationships and 19 23 29 34 37 38 40 16
and Riordan (2003). One of them refers to the equivalence of
personal achievement 41 43 46 47 48 49 54
sample. In the present study, sample variation could, of course, (SP3) 58 60
have been a source of variation in the factorial structure. Indeed, Type A (TA) 1 2 3 4 6 7 9 10 12 13 11
the overall population, on which the OSI is based, counts more 14
than 14,000 participants from a wide range of organisations. In Locus of control (LOC) 1 2 4 6 8 10 11 12 8
Coping strategies (CS) Total scale 19
our study, although the diversity of the participating companies
Problem-focused (C1) 4 5 8 9 10 15 16 19 21 10
enabled us to come close to the characteristics of the original 22
population, their substantially lower number could have led to a Emotion-focused (C2) 2 7 11 12 17 18 20 25 9
source of differentiation, for example, due to the heavier impor- 28
tance of factors, such as the sector of industry, the number of Mental health (MH) 1 2 3 4 5 6 7 8 10 11 17
12 13 14 15 16 17 18
years of experience of participants or even the relative proportion
Physical health (PH) 1 2 3 4 5 6 7 8 9 10 11 12
of sexes (only 22% were female). 12
As argued by Lu et al. (1995, p. 154), difference in fac- Job satisfaction (JS) Total scale 19
tor structure may also indicate “different dimensionalities in Organisational extrinsic 1 8 9 10 13 18 20 21 8
occupational stress-related perception and experiences among (JS1)
Person/job fit (JS2) 4 5 11 12 16 22 6
different cultures”. Certain methodological precautions were
Personal rewards (JS3) 3 6 14 15 17 5
also taken in an attempt to minimize some of the cultural causes
of difference in factorial structure. Thus, to ensure semantic Total 133
equivalence, this study used three precautions: a back-translation
procedure, a pilot trial and the participation of what Schaffer
and Riordan (2003, p. 190) called insiders and outsiders, that is,
“bilingual and bicultural researchers from each culture working 1 Please contact Professor Dominique Steiler at dominique.steiler@grenoble-

together” as a team. The back-translation procedure and the pilot em.com for any French OSI request.
 D. Steiler, B. Paty / Revue européenne de psychologie appliquée 59 (2009) 113–122
References
Arbuckle, J.L., Wothke, W., 1995. Amos 4.0 user’s guide. SmallWaters Corpo-
ration, Chicago.
Armor, D.J., 1974. Theta reliability and factor scaling. In: Costner, H.L. (Ed.),
Sociological methodology 1973–1974. Jossey-Bass, San Francisco, p. 28.
Baglioni, A.J.J., Cooper, C.L., Hingley, P., 1990. Job stress, mental health and
job satisfaction among UK senior nurses. Stress Medicine 6, 9–20.
Biggam, F.H., Power, K.G., MacDonald, R.R., 1997. Coping with the occupa-
tional stressors of police work: a study of Scottish officers. Stress Medicine
13.
Bogg, J., Cooper, C.L., 1995. Job satisfaction, mental health and occupational
stress among senior civil servants. Human Relations 48, 327–341.
Browne, M.W., Cudeck, R., 1993. Alternative ways of assessing model fit. In:
Bollen, K.A., Long, J.S. (Eds.), Testing structural equation models. Sage,
Newburry Park, California, pp. 136–162.
Campbell, J.P., Pritchard, R.D., 1976. Motivation theory in industrial and orga-
nizational psychology. In: Dunnette (Ed.), Industrial and organizational
psychology. Rand Mc Nally, Chicago, pp. 63–130.
Cattel, R.B., 1966. The scree test for the number of factors. Multivariate Behav-
ioral Research 1, 245–276.
Cooper, C.L., 1999. Occupational stress, job satisfaction and well-being in
anaesthetists. Stress Medicine 15, 111–126.
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, 11–28.
Cooper, C.L., Sloan, S.J., Williams, S., 1988. Occupational stress indicator
management guide. NFER-Nelson, Windsor.
Cooper, C.L., Williams, J., 1991. A validation study of the OSI on a blue collar
sample. Stress Medicine 7, 109–112.
Cronbach, L., 1951. Coefficient alpha and the internal structure of tests. Psy-
chometrika 16, 297–334.
Croutsche, J.-J., 2002. Étude des relations de causalité : utilisation des mod-
èles d’équations structurelles. La Revue des sciences de gestion 198,
81–97.
Cunha, R.C., Cooper, C.L., Moura, M.I., Reis, M.E., Fernandes, P., 1992. Por-
tuguese version of the OSI: a study of reliability and validity. Stress Medicine
8, 247–251.
Darr, W., Johns, G., 2008. Work, stress, health and absenteism. Journal of
Occupational Health Psychology 13, 293–318.
Davis, A.J., 1996. A re-analysis of the Occupational Stress Indicator. Work
Stress 10, 174–182.
Dubois, N., 1985. Contribution à l’étude de la dimensionalité du concept de
« locus of control ». L’Année psychologique 85, 27–40.
Evers, A., Frese, M., Cooper, C.L., 2000. Revisions and further develop-
ments of the Occupational Stress Indicator: LISREL results from four
Dutch studies. Journal of Occupational and Organisational Psychology 73,
221–240.
Flude, R., 1994. Downsizing: selecting who should stay and who should
go. Employee Counselling Today. Retrieved: 6/6, from http://emerald-
library.com/.
Frances, R., Mogenet, J.R., 1988. Motivation et satisfaction au travail. EAP,
Issy-Les-Moulineaux.
French, J.R.P., Caplan, R.D., 1972. Organisational stress and individual strain.
In: Murow, A.J. (Ed.), The failure of success. Amacon, New York, pp. 30–66.
French, J.R.P., Caplan, R.D., van Harrison, R., 1982. The mechanisms of job
stress and strain. John Wiley and Sons, Chichester.
Godin, I., Desmarez, P., Kittel, F., 2006. Short communication: work stress
assessment and instability of employment. Stress and Health 22, 51–58.
Herzberg, F., 1966. Work and the nature of man. Staples, London.
Highley, J.C., Cooper, C.L., 1994. Evaluating EAP’s. Personnel Review 234,
46–59.
Hu, L., Bentler, P.M., 1998. Fit indices in covariance structure modelling:
sensitivity to underparameterized model misspecifications. Psychological
Methods 3, 424–453.
Hu, L., Bentler, P.M., 1999. Cutoff criteria for fit indexes in covariance structure
analysis: conventional criteria versus new alternatives. Structural Equation
Modeling 6, 1–55.
121
Hurrell, J.J., Nelson, D.L., Simmons, B.L., 1998. Measuring job stressors and
strains: where we have been, where we are, and where need to go. Journal
of Occupational Health Psychology 3, 368–389.
ILO, 1992. Preventing stress at work, vol. 11. Bureau international du travail,
Genève.
Ingledew, D.K., Hardy, L., Cooper, C.L., 1992. On the reliability and the validity
of the locus of control scale of the occupational stress indicator. Personality
and Individual Differences 13, 1183–1191.
Jeanrie, C., Richard, B., 1997. Translating tests with the international test
commission’s guidelines: keeping validity in mind. Unpublished manuscript.
Jöreskog, K.G., 2003. Factor analysis by Minres. Retrieved: June 6th 2003, from
http://www.ssicentral.com/lisrel/minres.pdf.
Kahn, H., Cooper, C.L., 1991. A note on the validity of the mental health
and coping scales of the occupational stress indicator. Stress Medicine 7,
185–187.
Kaiser, H.F., 1960. Application of electronic computers to factor analysis. Edu-
cational and Psychological Measures 20, 141–151.
Kaiser, H.F., 1974. An index of factorial simplicity. Psychometrika 39, 31–36.
Karasek, R.A., 1979. Job demmands, job decision latitude and mental strain:
implications for job design. Administrative Science Quaterly 24, 285–308.
Kirkcaldy, B.D., Cooper, C.L., 1992. Managing stress of change: occupational
stress among senior police officers in Berlin. Stress Medicine 8, 219–231.
Kirkcaldy, B.D., Furnham, A., 1999. Stress coping styles among German man-
agers. Journal of Workplace Learning 11, 22–26.
Latack, J.A., Havlovic, S.J., 1992. Coping with job stress: a conceptual evalu-
ation framework for coping measures. Journal of Organizational Behavior
13, 479–508.
Lazarus, R.S., Folkman, S., 1984. Stress, appraisal and coping. Springer, New
York.
Leiter, M.P., Robichaud, L., 1997. Relationships of occupational hazards with
burnout: an assessement of measures and models. Journal of Occupational
Health Psychology 2, 35–44.
Lim, V.K.G., Teo, T.S.H., 1996. Gender differences in occupational stress and
coping strategies among IT personnel. Women in Management Review 11,
20–28.
Lu, L., Cooper, C.L., Chen, Y.C., Hsu, C.H., Li, C.H., Wu, H.L., et al., 1997.
Chinese version of the OSI: a validation study. Work Stress 11, 79–86.
Lu, L., Cooper, C.L., Yen Ching Chen, Chia Ho Hsu, Chi Hua Li, Hung Luan
Wu, et al., 1995. Chinese version of the OSI: a study of reliability and factors
structures. Stress Medicine 11, 149–155.
Lyne, K.D., Barrett, P.T., Williams, C., Coaley, K., 2000. A psychometric eval-
uation of the Occupational Stress Indicator. Journal of Occupational and
Organisational Psychology 73, 195–220.
Meier, L.L., Semmer, N.K., Elfering, A., Jacobshagen, M., 2008. The double
meaning of control: three-way interactions between internal resources, job
control and stressors at work. Journal of Occupational Health Psychology
13, 244–258.
Ndoye, A.K., 2000. L’insatisfaction au travail des professeurs du second degré
du Sénégal. Revue des sciences de l’éducation 26, 439–462.
Norusis, M.J., 1993. SPSS professional statistics. SPSS, Chicago.
Rees, D.W., Cooper, C.L., 1992. The Occupational Stress Indicator locus of
control scale: should this be regarded as a state rather than a trait measure?
Work and Stress 6, 45–48.
Renault de Moraes, L.F., Swan, J.A., Cooper, C.L., 1993. A study of occupa-
tional stress among government white collar workers in Brazil using the
occuptional stress indicator. Stress Medicine 9, 91–104.
Robertson, I.T., Cooper, C.L., Williams, J., 1990. The validity of the Occupa-
tional Stress Indicator (OSI). Work Stress 4, 29–39.
Robinson, P., Inkson, K., 1994. Stress effects on the health of chief executives
of business organizations. Stress Medicine 10, 27–34.
Roussel, P., Durieu, F., Campoy, E., El Akremi, A., 2002. Méthodes d’équations
structurelles : recherche et application en gestion. Economica, Paris.
Russinova, V., Vassileva, L., Randev, P., Jiliova, S., Cooper, C.L., 1997. Psy-
chometric analysis of the first Bulgarian version of the Occupational Stress
Indicator. International Journal of Stress Management 4, 111–119.
Schaffer, B.S., Riordan, C.M., 2003. A review of cross-cultural methodolo-
gies for organizational research: a best-practices approach. Organizational
Research Methods 6, 169–215.
122 D. Steiler, B. Paty / Revue européenne de psychologie appliquée 59 (2009) 113–122
Spence, P.E., Helmreich, R.L., Pred, R.S., 1987. Impatience vs. achieve-
ment striving in the Type A pattern: differential effects on students’ health
and academic achievement. Journal of Applied Psychology 72, 522–
528.
Spielberger, C.D., 1994. Professional manual for the job stress survey (JSS).
Psychological Assessment Resources Inc, Odessa, FL.
Steiler, D., 2005. Une mesure du stress professionnel : fidélité et validité
factorielle de la version française de l’Occupational Stress Indicator. In:
Chasseigne, G., Lassare, D. (Eds.), Stress et société, vol. 2. Presses Univer-
sitaires de Reims, Reims, pp. 89–122.
Steiler, D., Cooper, C.L., 2004. French version of the Occupational Stress Indi-
cator (OSI): preliminary assessment of reliability and validity. Stress and
Health 20, 231–237.
Swan, J.A., Renault de Moraes, L.F., Cooper, C.L., 1993. Developing the
Occupational Stress Indicator (OSI) for the use in Brazil: a report on
the reliability and validity of the translated OSI. Stress Medicine 9, 247–
253.
Tarquinio, C., 2008. Work is good for you! Psychological approaches to the prob-
lematic of occupational health. European Review of Applied Psychology 58,
199–200.
Vallerand, R.J., 1989. Vers une méthodologie traditionnelle de validation tran-
sculturelle de questionnaires psychologiques : implication pour la recherche
en langue française. Canadian Psychology 30, 662–680.
Van de Vijver, F., 2004. Bias and equivalence in cross-cultural assessment: an
overview. European Review of Applied Psychology 54, 135–199.
Velicer, W.F., 1976. Determining the number of components from the matrix of
partial correlation. Psychometrika 41, 321–327.
Verborgh, E., 1992. Foreword. In: Simmons, S., Stampe Oland, J. (Eds.), Work-
place Assessment. European Foundation for the Improvement of Living and
Working Conditions, Luxembourg.
Williams, S., Cooper, C.L., 1997. Occupational Stress Indicator. In: Zalaquett,
C.P., Wood, R.J. (Eds.), Evaluating stress, vol. 1. The Scarecrow Press Inc,
Lanham & London, pp. 251–276.
Yaffee, R.A., 1997. Common correlation and reliability analysis with SPSS
for Windows. Retrieved: May 23 2003, from http://www.nyu.edu/its/socsci/
Docs/correlate.html.
Zalaquett, C.P., Wood, R.J., 1997. Evaluating stress: a book of resources, vol. 1.
The Scarecrow Press, Inc, London.
Zwick, W.R., Velicer, W.F., 1986. Comparison of fives rules for determining the
number of components to retain. Psychological Bulletin 99, 432–442.