Download as pdf or txt
Download as pdf or txt
You are on page 1of 12

Journal of Occupational Health Psychology © 2013 American Psychological Association

2013, Vol. 18, No. 3, 338 –349 1076-8998/13/$12.00 DOI: 10.1037/a0033114

The Development and Validation of a New Multidimensional Job


Insecurity Measure (JIM): An Inductive Methodology
Patrick O’Neill and Peter Sevastos
Curtin University

This study outlines the development and validation of a new 4-dimensional job insecurity measure (JIM).
Items were generated from interviews with Australian employees facing an objective threat of job loss.
The measure was then validated on a North American sample of 1,004 respondents. Exploratory and
confirmatory factor analyses (EFA and CFA) with tests of sample invariance supported an 18-item
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

measurement model consisting of 4 correlated but distinct subscales of insecurity: Job Loss, Job Changes,
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Marginalization, and Organizational Survival. The results indicate convergent and discriminant validity
as well as high internal consistency for the instrument. Significant associations with psychological
well-being (job related affective well-being, job satisfaction) and organizational attitudes (organizational
commitment, trust in management, intention to resign) established the criterion-related validity of each
subscale. These findings support the use of the instrument in academic and applied settings.

Keywords: job insecurity, validation, well-being, downsizing, restructuring

In the context of widespread downsizing and restructuring employee attitudes, such as organizational commitment (! "
(Datta, Guthrie, Basuil, & Pandey, 2010), the past two decades #.36), trust (! " #.50), and intentions to quit (! " .28).
have seen a growing interest from academics and practitioners in
the effects of job insecurity as “one of the most important stressors Previous Job Insecurity Measures
in contemporary working life” (De Cuyper, Bernard-Oettel, Bern-
tson, De Witte, & Alcaro, 2008, p. 493). Survey findings from 21 Despite the prevalence of job insecurity and its well-
Organisation for Economic Co-Operation and Development documented effects on individual health and organizational vital-
(OECD) member countries (Organisation for Economic Co- ity, few attempts have been made to systematically examine the
Operation and Development, 1997, 1999) indicate that perceptions dimensionality of the construct itself using sophisticated validation
of job insecurity rose steadily in the 1980s and 1990s. More procedures (e.g., Lee, Bobko, Ashford, Chen, & Ren, 2008; Probst,
recently, annual job satisfaction surveys conducted by the Society 2003). In the pursuit of more substantive research questions, many
for Human Resource Management (SHRM) between 2008 and studies have relied on single item “global” measures, most of
2011 have consistently reported job security as the top employee which capture the likelihood or probability of job loss (e.g., “Do
concern and the most important determinant of job satisfaction you expect to be in your current position five years from now?”;
above compensation, benefits, workplace safety, relationship with Dekker & Schaufeli, 1995). With at least three observable vari-
one’s supervisor, the work itself, and senior management– ables necessary to define a latent construct, the reliability of these
employee communication (Society for Human Resource Manage- measures is dubious due to measurement error (Bentler & Chou,
ment, 2011). 1987).
The impact of job insecurity on employee well-being is sup- Beyond single-item measures, job insecurity has been concep-
ported by meta-analytic research showing consistent negative as- tualized as either a unidimensional or multidimensional construct.
sociations between job insecurity and job satisfaction (! " #.41), Those studies adopting a unidimensional approach define job
general mental health (! " #.24), and physical health (! " #.16; insecurity as an overall concern for the future of one’s job (e.g.,
Sverke, Hellgren, & Näswall, 2002), and longitudinal studies Johnson, Messe, & Crano, 1984). Researchers have used scales
report the enduring effects of job insecurity on mental strain (e.g., with multiple items capturing perceived uncertainty or emotions,
Hellgren & Sverke, 2003). Beyond the human costs of job inse- such as fear over future job loss—for instance, Johnson et al.’s
curity, organizations may also pay a high price in terms of em- (1984) seven-item subscale of the Work Opinion Questionnaire
ployee attitudes and behaviors. Sverke and colleagues (2002) (e.g., “The thought of getting fired really scares me”) and Caplan,
found negative associations between job insecurity and several Cobb, French, Van Harrison, and Pinneau’s (1975) four-item Job
Future Ambiguity scale (e.g., “whether your job skills will be of
use and value five years from now”).
Previous studies incorporating unidimensional measures have
Patrick O’Neill and Peter Sevastos, School of Psychology, Curtin Uni- been criticized for failing to display the psychometric properties of
versity, Perth, Western Australia. their scales or for describing, in sufficient detail, the statistical
Correspondence concerning this article should be addressed to Patrick procedures applied (Ashford, Lee, & Bobko, 1989). These psy-
O’Neill, 309-145 Clarence St., Ottawa, Ontario, Canada, K1N-1B7. E-mail: chometric limitations are compounded by variability in the item
patrick13@rogers.com content both in terms of the target of insecurity (e.g., one’s job,
338
JOB INSECURITY MEASURE 339

career, responsibilities) and varying emphasis on cognitive percep- career opportunities in [the organization] are favorable”) and three
tions or affective reactions to job insecurity. The practice of quantitative (e.g., “I am worried about having to leave my job
merging purportedly distinct constructs at the measurement level before I would like to”) items. These two scales were only mod-
prevents researchers from uncovering important differences be- erately correlated (r " .28) and demonstrated adequate internal
tween the constructs and theoretically meaningful relationships consistency reliability. Further longitudinal analyses showed that
with outcomes. quantitative and qualitative job insecurity predicted subsequent
Greenhalgh and Rosenblatt (1984) were the first researchers to mental and physical health, whereas qualitative job insecurity was
argue in favor of a multidimensional operationalization of job the sole predictor of job satisfaction and turnover intention.
insecurity, in which perceptions of job insecurity are said to extend
beyond a concern over job loss to the loss of desirable job features,
Cognitive and Affective Job Insecurity
such as lack of promotion opportunities and current working
conditions. Most theoretical perspectives informing the study of Recent research has argued for a two-dimensional job insecurity
job insecurity highlight the importance of seeing the construct as measurement structure consisting of cognitive (perceived proba-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

multidimensional in nature (Mauno, Leskinen & Kinnunen, 2001). bility of job loss) and affective (fear and worry over potential job
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Consistent with this view, job insecurity is defined herein as “a loss) job insecurity (Huang, Lee, Ashford, & Ren, 2010). Although
concern about the future existence of one’s job or features asso- few researchers would disagree that job insecurity includes both
ciated with the job” (Lim, 1997, p. 252). From this perspective, job cognitive and affective elements, Huang and colleagues (2010)
insecurity could exist in a restructuring environment that does not note a lack of consistency in the measurement of these two
include layoffs but may involve unfavorable changes to the job. dimensions. Some researchers have combined cognitive and affec-
tive scales into a composite job insecurity index (Reisel & Banai,
2002), whereas others have treated the two dimensions as related
Job Loss and Job Changes Insecurity
but distinct (Konig & Staufenbiel, 2006).
Attempts to operationalize job insecurity as a multidimensional In an attempt to clarify the cognitive and affective dimensions of
construct have distinguished between insecurity due to job changes job insecurity, Probst (2003) developed the Job Security Index and
versus job loss (Hellgren, Sverke, & Isaksson, 1999). For instance, the Job Security Satisfaction scale. The Job Security Index as-
Ashford et al.’s (1989) 51-item Job Insecurity Scale (JIS) mirrors sesses “an individual’s cognitive appraisal of the future of his or
Greenhalgh and Rosenblatt’s (1984) theoretical model, and con- her job with respect to the perceived level of stability and contin-
sists of five subscales measuring seven job features and 10 changes uance of that job” (Probst, 2003, p. 452), whereas the Job Security
to the job as a whole in terms of their importance, the likelihood Satisfaction scale measures satisfaction with job security, which
they would be lost, and their perceived powerlessness to resist the Probst equated to affective job insecurity. Respondents are asked
threat. The multiplicative combination of these components form to rate, on a 3-point scale (Yes/?/No), the extent to which a series
an index of overall job insecurity. of adjectives describe the future of their job and their job security
A common criticism of Ashford et al.’s (1989) 51-item JIS is its (e.g., “questionable,” “nerve wracking”). Although Probst’s study
exhaustive length (Kinnunen, Mauno, Nättie, & Happonen, 1999; is one of few to apply CFA, the author notes that “the resulting fit
Mauno et al., 2001), which makes it of questionable use in an indices are nonetheless lower than would be desired according to
applied setting. Lee and colleagues (2008) responded to this crit- conventional standards” (p. 459). Moreover, separate factors
icism by creating abridged (37 items) and “bare bones” (25 items) emerged for the negatively and positively worded items of the Job
versions of the JIS based on confirmatory factor analysis (CFA). Security Satisfaction scale, leading the author to recommend that
The authors did not, however, establish the relative fit of their “future research apply confirmatory factor analyses . . . to confirm
proposed five-dimensional model with other theoretically plausi- its factor structure.” The use of a categorical three-point rating
ble factor structures (Sobel & Bohrnstedt, 1985). This is an im- scale may have contributed to these tentative results, because such
portant omission, given that researchers have questioned the va- scales cannot be evaluated using the recommended maximum
lidity and reliability of the Probability scale (Mauno et al., 2001), likelihood method in CFA: They can only be assessed using
and others claim that the Importance and Powerlessness scales distribution-free methods.
(Kinnunen et al., 1999; Rosenblatt & Ruvio, 1996) are not neces-
sary components of job insecurity. Researchers have also ques-
Job Insecurity From the Worker’s Perspective
tioned the use of a multiplicative rating scale (e.g., Jacobson,
1991a). Cheng and Chan (2008) excluded studies using the JIS in A common feature of the preceding job insecurity measures is
their meta-analysis based on research by Evans (1991) indicating the deductive approach employed in their development. Specifi-
that the relationship between multiplicative measures and criterion cally, items were developed from the researcher’s perspective or a
variables is subject to a scaling effect that may give rise to spurious preexisting theoretical framework. Implicit in this approach is the
results in bivariate correlational analysis. assumption that researcher’s interpretation of the nature of the
As a more parsimonious alternative to the JIS, Hellgren, Sverke, construct and its key dimensions is a reasonable approximation of
and Isaksson (1999) developed a job insecurity measure (JIM) the worker’s. However, researchers have yet to agree on a common
with subscales measuring job insecurity that is quantitative (per- definition and measure of job insecurity, creating inconsistency
ceived threats to the continuity of the job itself) and qualitative and confusion on both fronts. Moreover, researchers study job
(perceived threat to valued qualities of the employment relation- insecurity from a unique “frame” or perspective, which may be far
ship). Principal components analysis (N " 375) supported a two- removed from that of employees in a downsizing or restructuring
factor solution consisting of four qualitative (e.g., “My future environment.
340 O’NEILL AND SEVASTOS

The discrepancy between the researcher’s and worker’s views of represents the core dimensions of the construct identified by work-
job insecurity was identified by Bartunek and Seo (2002; p. 239) ers facing an objective threat of job loss, (b) captures both the
as “an important threat to the content validity of the construct” in cognitive and affective experience of job insecurity without con-
that “it might miss important dimensions or include trivial dimen- founding the measure with affective outcomes, (c) encompasses
sions in the setting studied.” To overcome these deficiencies, the insecurity over the loss of one’s job and valued features of that job,
authors highlight the advantages of qualitative or inductive meth- and (d) is short enough to be used in an applied setting. To meet
ods to (a) gain a richer understanding of job insecurity from the these objectives, the present study describes the development,
worker’s perspective, (b) confirm or disconfirm academic defini- validation, and cross-validation of a new job insecurity measure.
tions of the construct in a local context, and (c) uncover new Study 1 describes the process used to develop a pilot JIM based on
dimensions that have not been included in previously developed interviews with a sample of Australian workers in a downsizing
scales. By generating items directly from insecure workers, the organization. Study 2 examines the dimensionality and construct
inductive approach to scale development has the capacity to iden- validity of the four-dimensional JIM on a North American sample
tify the various environmental cues and social dynamics that using exploratory and confirmatory factor analysis (EFA and
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

employees draw from to make sense of threatening environments, CFA). The criterion-related validity of the four-dimensional JIM is
This document is copyrighted by the American Psychological Association or one of its allied publishers.

which may not be intuitively apparent to researchers. Once uncov- then tested by exploring its relationship with a variety of outcomes
ered, these dimensions can then be interpreted in light of prevailing related to psychological well-being and organizational attitudes.
theoretical models or may serve to enrich and expand such models
(Klandermans & van Vuuren, 1999). Study 1: Scale Development
To the authors’ knowledge, Lahey’s (1984) Job Security Scale
is the only job insecurity measure to be developed based on this
qualitative approach. Lahey (1984) conducted semistructured in-
Sample
terviews with 22 employees representing five organizations to The research setting for the present study was an Australian
produce a 65-item job security pilot questionnaire, which was private sector company in the forest products industry. At the time
administered to 487 employees from three organizations (service, of data collection, the company employed approximately 710
manufacturing, and civil service). Principal component analysis employees and was divided into six separate operating units: forest
was used to produce the 44-item Job Security Scale, which consists harvesting, saw milling, processing, domestic retail, international
of five dimensions: company concern for the individual, job per- retail, and administration. With an estimated but unannounced 400
formance, company growth and stability, job permanence, and redundancies expected by the end of the year, all branches of the
individual commitment. Unfortunately, at the time of Lahey’s company faced the objective threat of impending redundancies or
study, the use of CFA techniques—which would have enabled her complete closure.
to extend her findings beyond a preliminary exploration of the
instrument’s components—was less widespread. This method-
Item Generation
ological limitation has given rise to a common criticism that the
Job Security Scale suffers from a lack of parsimony (Jacobson, An initial pool of job insecurity items was generated by a series
1991b) and offers “relatively little phenomenological contribution of 12 semistructured group interviews conducted with a random
to the concept of job insecurity because some dimensions could sample of 120 employees (n " 60), supervisors (n " 34), and
pass for correlates of job insecurity (e.g., work performance)” managers (n " 26). The participants in this sample were between
(Rosenblatt & Ruvio, 1996, p. 588). 19 and 62 years of age, and 32% were female. Interviews were
conducted during normal working hours, in groups of 10, each
lasting 45 min. Participants were asked to complete the phrase “I
Summary and Overview
feel insecure in my job when. . .” Each interview was audio re-
In summary, despite the many theoretical advances that have corded and transcribed, yielding a total of 176 specific statements
been made since systematic research into job insecurity began in relating to job insecurity. Item generation was discontinued when
the 1980s, researchers have been unable to converge on a concise no new items were identified after several consecutive interviews.
measurement model with strong psychometric properties (Sverke By excluding items that appeared too ambiguous, or redundant,
et al., 2002; van Wiek & Pienaar, 2008). This lack of consensus the two authors and a postgraduate student selected a set of 46
may be largely attributed to a paucity of validation studies using items before independently assigning each item into a domain
sophisticated CFA techniques; the proliferation of job insecurity based on conceptual similarity. Interrater consistency (kappas) for
scales derived from different theoretical orientations and perspec- the assignment of items to domains (Cohen, 1960) for all three
tives; and a growing preference for more parsimonious measures. pairs of the three raters was .79, .83, and .87. When there was
Also evident in the literature is the absence of a psychometrically disagreement, a third panel member arbitrated resulting in four job
robust job insecurity measure developed from the unique perspec- insecurity dimensions labeled and defined as follows: job loss
tive of the insecure worker—rather, most researchers have “fit” the (perceived uncertainty about the continuity of one’s job and the
worker’s perspective into a preexisting theoretical framework. An likelihood of job loss), organizational survival (the degree to
acceptable job insecurity measure would add rigor to future re- which the organization is perceived to be economically viable and
search and enable the development of evidence-based interven- investing in future growth), marginalization (the perception that
tions aimed at reducing job insecurity. one is being ignored by management and excluded from the
Collectively, these developments support the need for a job broader social activities of the organization), and job changes (the
insecurity measure with sound psychometric properties that (a) anticipated loss of valued features of one’s job).
JOB INSECURITY MEASURE 341

Having identified an initial classification of job insecurity items, The four items developed for the Organizational Survival Inse-
a set of 26 pilot items was written by the authors to measure each curity (OSI) scale are consistent with Lahey’s (1984) “company
of the four job insecurity dimensions. We reviewed the literature growth and stability” scale, such as “This organization seems to
and preexisting measures relevant to each dimension and, to the have clear goals and a definite strategy for achieving them” and
extent possible, the language for the items was taken from the “Management appears to be preparing in advance and planning for
original statements made by interviewees. The items also needed the future.” By signaling a clear strategic direction and investments
to be consistent with the construct definitions and to be behavioral in future growth, these behaviors are likely to foster employee
in nature, and not confound with affective or other attitudinal confidence in a company’s economic stability and its ability to
variables (Hinkin, 1995). By not directly measuring affective meet future demands. In the context of environmental uncertainty,
reactions to perceived job insecurity such as fear or worry, we Jacobson (1991b) and others (Klandermans & van Vuuren, 1999)
sought to avoid confounding the measure with affective outcomes. have discussed the importance of such behaviors in enhancing or
A peer review of the measure was conducted by three university reducing the perceived probability and/or severity of job loss.
professors from psychology and business administration depart- Additionally, longitudinal evidence supports the direct causal im-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

ments with experience consulting in organizations and developing pact of organizational performance (financial and market) on em-
This document is copyrighted by the American Psychological Association or one of its allied publishers.

psychometric instruments. This was followed by an independent ployee satisfaction with job security (Schneider, Hayes, Smith, &
review by 12 members of the sample organization representing a Salvaggio, 2003).
range of occupational categories and across employment levels. The four-item Marginalization Insecurity (MI) scale reflects
Upon completing the instrument, each reviewer was provided with perceived social exclusion on the part of one’s manager (e.g., “I
definitions of the four constructs and asked to review the instru- feel as though management is avoiding me”) and from the activ-
ment in terms of clarity, wording, face validity, content validity, ities of the organization as a whole (e.g., “I feel like I am being
and ease of usage. The peer review resulted in the elimination of given the “silent treatment” in this organization”). Although such
two items and the rewording of others creating a 24-item pilot behaviors have yet to be examined in the literature as elements of
measure. job insecurity, items related to social exclusion, such as the “silent
treatment,” have appeared in the literature as indicators of inter-
personal deviance (Bennett & Robinson, 2000), workplace inci-
Pilot (JIM) vility (Blau & Andersson, 2005), and workplace ostracism (Ferris,
Brown, Berry, & Lian, 2008). Ferris et al. (2008) found that higher
The seven-item Job Loss Insecurity (JLI) scale captures uncer- levels of workplace ostracism (the perception that one is being
tainty over future job loss (e.g., “I’m not sure of how long my job ignored or excluded at work) were associated with a decrease in
will last”), as well as a lack of control or powerlessness (e.g., “No well-being, job satisfaction, affective commitment, and productiv-
matter how hard I work, there is no guarantee that I am going to ity.
keep my job”) and the importance of job loss to the individual Respondents were asked to indicate their level of agreement
(e.g., “The possibility of losing my job occupies my thoughts with statements on a 7-point Likert scale ranging from 1 (very
constantly”). Uncertainty has been identified as one of the most inaccurate) to 7 (very accurate). Scale anchors were modeled after
common psychological states arising from organizational change Hackman and Oldham’s (1975) Job Diagnostic Survey. All items
and a central aspect of job insecurity (De Witte, 1999). Employee were worded in the direction of insecurity with the exception of the
perceptions of employment uncertainty are also featured as an reverse-scored OSI scale.
environmental determinant of well-being within Warr’s (1987)
vitamin model, under the “environmental clarity” group of fea-
tures. Although Ashford and colleagues (1989) measure power- Discussion
lessness and importance as separate dimensions of job insecurity, The purpose of Study 1 was to develop a preliminary set of job
we sought to test Jacobson’s (1991a) theoretical proposition that insecurity items based on interviews with workers facing an ob-
both elements are likely to be embedded in the experience of jective threat of job loss and provide an initial factor structure for
employment uncertainty. further testing. Whereas two of the constructs identified in the
The seven items developed for the Job Changes Insecurity (JCI) study (JLI and JCI) have surfaced in theoretical models of job
scale builds on Hellgren et al.’s (1999) four-item qualitative Job insecurity and were similar to preexisting measures, two new
Insecurity Scale by referring to the anticipated loss of valued job dimensions (OSI and MI) were uncovered. In order to determine
features in general (e.g., “unfavorable changes to my job,” “many whether these four job insecurity dimensions are an accurate
features of my job that I value the most”), as well as specific facets reflection of the underlying constructs and demonstrate sound
of the job, such as “opportunities for promotion and advancement” psychometric properties, the pilot JIM was validated and cross-
and “physical working conditions.” These changes have high ex- validated in Study 2.
ternal validity in that they represent some of the most well-
documented changes stemming from corporate restructuring and
Study 2: Scale Validation and Cross-Validation
downsizing (Mishra, Spreitzer, & Mishra, 1998), such as salary
freezes (“the rewards of my job are likely to diminish”), cutbacks We examined the dimensionality and construct validity of the
in plant and equipment technology (“overall, my physical working pilot JIM on a North American sample using EFA and CFA with
conditions are likely to deteriorate”), and a decreased investment tests of invariance across samples. Based on the four job insecurity
in training (“I expect to have fewer resources to meet the perfor- dimensions identified in Study1, we hypothesized that a four-
mance requirements of my job”). dimensional structure would prove to be a more appropriate model
342 O’NEILL AND SEVASTOS

for employee’s perceptions of job insecurity than a unidimensional Sample 2 was then utilized for conducting CFA to test the extent
model, and that the four-dimensional model would generalize to which the factorial structure derived from EFA fit the data. The
across different employee samples. The two testable hypotheses EQS version 6.1 structural equations modeling (SEM) program
therefore are as follows: (Bentler, 2005) was used to conduct the CFA. Analyses of model
fit involved testing a sequence of three measurement models of
Hypothesis 1: A four-dimensional measurement model of job increasing complexity. The first model, which was used as a
insecurity (job loss, job changes, marginalization, and orga- baseline, was a null model, which argues for complete indepen-
nizational survival) will demonstrate a fit to the data that is dence (zero covariance) between items (Bentler & Bonett, 1980).
superior to a unidimensional model. Model 2 was a unidimensional solution with all of the items
loading onto a single factor. Model 3 consisted of the hypothesized
Hypothesis 2: The four-dimensional job insecurity measure- four-dimensional structure of job insecurity. The maximum like-
ment model derived from the validation sample will general- lihood method of estimation was used to estimate all models, and
ize (i.e., be invariant) to data drawn from a cross-validation the input for all analyses was the covariance matrix. To test these
sample.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

models, only random error was taken into consideration.


This document is copyrighted by the American Psychological Association or one of its allied publishers.

Assessment of model fit. The first objective of CFA was to


Having identified a final measurement model, we tested the
test the extent to which the factorial structure derived from EFA fit
instrument’s criterion-related validity—the extent to which a con-
the data for the calibration sample. Analyses of model fit involved
struct is related to variables derived from theory (Hinkin,
testing a sequence of three measurement models of increasing
1995)— by exploring its relationship with a variety of outcomes
complexity. The first model, which was used as a baseline, was a
related to psychological well-being and organizational attitudes.
null model, which argues for complete independence (zero cova-
riance) between items (Bentler & Bonett, 1980). Model 2 was a
Sample single first-order factor under which all job insecurity items are
Data were collected from a large North American media com- subsumed. Model 3 consisted of the four-dimensional structure of
pany (N " 2,900). A total of 1,185 surveys were completed, job insecurity derived from EFA.
representing 47% of the employees who received the invitation The appropriateness of each of the three competing models was
e-mail. This data set contained four central divisions (TV, radio, examined using several indices of fit. Global assessments of fit
corporate, content), consisting of 21 departments and several geo- were based on the Satorra-Bentler scaled statistic (S-B$2) and the
graphically dispersed locations. The sample was comprised of 38% difference in chi square and degrees of freedom. Additional fit
professionals, 10% senior managerial or executive personnel, 19% indices included the robust comparative fit index (RCFI) and the
middle managers, 6% supervisors, 16% human resources and standardized root mean square residual (SRMR). The RCFI ranges
administrative support, 10% technical staff and tradespersons, and from zero to 1.00, with a value !0.95 indicating an acceptable fit
1% laborers. The sample was roughly split in terms of gender (Hu & Bentler, 1999). Values below .08 for the SRMR are deemed
(53% male; 47% female), and the average age and tenure of acceptable (Hu & Bentler, 1999), whereas those above this cutoff
respondents was 42 years and 10 years, respectively. The majority value would indicate a misspecification of the model. A supple-
of respondents were employed full-time (93%), were nonunion mentary third index was used, the root mean square of approxi-
members (94%), and held either a postsecondary diploma (37%) or mation (RMSEA), which provides information on the degree of
a university degree (33%). imprecision in the point estimates and is relatively independent of
From the main data set, 93 of the cases were eliminated due to sample size. Point estimates of .05 or less indicate a good fit,
missing information, followed by another 17 cases identified as whereas a value approaching .08 would represent reasonable errors
univariate outliers, with scores exceeding three standard deviations of approximation (Steiger, 1999), although Hu and Bentler (1999)
from the mean. This resulted in 1,004 cases available for subse- recommend a cutoff value of .06 for this index as an indication of
quent analyses. Using a listwise selection procedure, this main a reasonable fit. RMSEA values above .10 indicate poor fit. All
sample was randomly split into two equal samples of 502 each. model fit evaluations for the present study were based on MacCa-
Anderson and Gerbing (1988) recommend the random assignment llum, Browne, and Sugawara’s (1996) framework, after demon-
of two subsamples, where one sample is used to develop the strating adequate power and sample size requirements.
measurement model and the other to validate the solution obtained. Convergent and discriminant validity. Convergent validity
In line with this recommendation, one of these samples (n " 502) was assessed by examining the t-ratios for the factor loadings of
was used in its entirety to perform the EFA. Sample 2 was then the job insecurity items on their specified dimensions. Statistically
randomly split into two equal subsamples of 251 each and utilized significant t-ratios indicate that the items in the model are signif-
for CFA. One of these two samples was designated the calibration icantly related to their target constructs.
sample and the other the cross-validation sample. Discriminant validity for each pair of constructs was assessed by
setting estimated covariance parameters free and then performing
a separate analysis in which the relationship between them was
Statistical Analyses
constrained to 1.00 (Anderson & Gerbing, 1988). A chi-square
EFA was first performed on Sample 1 (n " 502) to obtain a difference test (with one degree of freedom) on the values obtained
factor solution suitable for CFA. We used the maximum likelihood for the constrained and unconstrained models was performed
oblique estimation procedure for the EFA and chose the direct (Jöreskog, 1971). A significant difference in chi square suggests
oblimin method of rotation, as it generated the most easily inter- nonequivalence and indicates that the constructs are distinguish-
pretable and psychologically meaningful solution. able (Bagozzi & Phillips, 1982).
JOB INSECURITY MEASURE 343

Testing for invariance. Having identified the measurement Measurement is on a 7-point scale (1 " very inaccurate to 7 "
model that most adequately describes the data, this best-fitting very accurate).
model was then compared for its invariance across the calibration Organizational commitment. A five-item scale was used to
and validation samples. A series of “chi-square difference tests” measure organizational commitment (from the nine-item scale by
(Bollen, 1989, p. 292) was then performed between two nested Cook & Wall, 1980). Respondents were asked to indicate their
models. After establishing the best baseline model for each group level of agreement with items on a 7-point Likert scale (1 " no, I
separately, tests of invariance between two or more samples pro- strongly disagree to 7 " yes, I strongly agree). A sample item
ceed in a hierarchical fashion (Byrne, Shavelson, & Muthén, includes, “I am quite proud to be able to tell people whom it is I
1989). Invariance was first sought by comparing the baseline work for” (% " .82).
model with a model that constrains item-factor loadings in the Trust in management. We used Cook and Wall’s (1980)
replication sample. If the results of the comparison are not statis- five-item Trust in Management scale. Sample items include “Man-
tically significant (i.e., constrained parameters are equal across agement at (organization) is sincere in its attempts to meet the
groups), a further comparison of this model and one with addi- workers’ point of view” and “Management can be trusted to make
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

tional constraints on factor covariances is performed (Bentler, sensible decisions for (organization)’s future” (% " .87). Re-
This document is copyrighted by the American Psychological Association or one of its allied publishers.

2005). An even more restrictive test is to constrain, in addition to sponses are on a 7-point scale (1 " no, I strongly disagree to 7 "
the loadings and covariances among factors, the error variances of yes, I strongly agree).
the observed variables. If the results of the comparison between the Intention to resign. We used a two-item scale derived from
two nested models are statistically nonsignificant, then we may the Michigan Organizational Assessment Questionnaire (Seashore,
conclude that the four-factor measurement model is invariant or Lawler, Mirvis, & Cammann, 1982) to assess the extent to which
generalizable across samples. a respondent intended to resign from the organization: “I will
Criterion-related validity. We sought to test the criterion- probably look for a new job within the next year” and “I often
related validity of the four job insecurity constructs by examining think about quitting” (% " .92). Responses are on a 7-point scale
their relationship with a variety of constructs with meta-analytic (1 " no, I strongly disagree to 7 " yes, I strongly agree).
links to job insecurity (Cheng & Chan, 2008; Sverke et al., 2002).
Correlational analyses were conducted on the main data set (N " Results
1,004), which included the JIM and variables related psychological
well-being (job-related affective well-being, job satisfaction) and EFA. An EFA was conducted on the 24-item JIM, identifying
organizational attitudes (organizational commitment, trust in man- five factors with eigenvalues greater than 1. Because the fifth
agement, intention to resign). Prominent theories of occupational factor contained only two items (“Management at this company
stress and well-being have identified job insecurity as a type of pays attention to me when I’m speaking”; “Management at this
work-related stressor that is potentially detrimental to the mental company notices me”), these were deleted and a second EFA was
conducted on the 22-item covariance matrix. Out of the 22 pilot
health and attitudes of employees (Katz & Kahn, 1978;Warr,
items, four of these had loadings below the conventional cutoff of
1987). Accordingly, we expected all four job insecurity subscales
.40 (Hinkin, 1995) and were therefore eliminated. This analysis
to be positively associated with intention to resign and negatively
identified four job insecurity factors with an eigenvalue greater
associated with all other criteria.
than 1, comprising 18 items and accounting for 58.02% of the
variance.
Measures The first factor capturing JCI contained six items accounting for
39.98% of the variance. The content of this scale dealt with
JIM. A preliminary 24-item pilot JIM was used for EFA and changes to job features in general (e.g., “many features of my job
CFA. The final 18-item JIM validated in the present study was that I value the most”) that would apply across organizations and
used to test for criterion-related validity. JLI (% " .90) and JCI jobs and that would threaten an employee’s ability to perform the
(% " .90) were both measured with six-item scales, whereas MI job (e.g., “I expect to have fewer resources to meet the perfor-
(% " .88) and OSI (% " .87) were tapped using three items each. mance requirements of my job”). The second JLI factor consisted
Scoring for the four subscales was based on a simple, unweighted of six items accounting for 6.48% of the variance. The third factor
average of the scale items. consisting of four items and accounting for 6.85% of the variance
Job-related affective well-being. Two scales developed by assessed MI. Finally, the three OSI items represented 4.70% of the
Warr (1990) were used to measure the two dimensions of affective variance.
well-being in the work context: Anxiety–Contentment (% " .82) CFA. The job insecurity measurement model identified
and Depression–Enthusiasm (% " .88). The six-item, 6-point through the EFA was the input for CFA. This 18-item model,
scales ask respondents to think of the past few weeks and rate a illustrated in Figure 1, was tested using Sample 2. Table 1 displays
series of emotions related to the job (e.g., tense, optimistic). Scores fit indices for this model (RCFI " .973; RMSEA " .048; CI "
range from 1 (never) to 6 (all of the time). .040 –.056; SRMR " .041) in comparison with a null model.
Intrinsic and extrinsic job satisfaction. The Job Satisfaction Table 2 and Table 3 show the relevant fit indices for the null,
Scale developed by Warr, Cook, and Wall (1979) was used to single, and four-factor models across the calibration and cross-
measure intrinsic job satisfaction (e.g., satisfaction with variety in validation samples. Results for the calibration sample reveal that
the job, responsibility given; % " .86) and extrinsic job satisfaction the four-factor model provided the best fit of the data. The single-
(e.g., satisfaction with pay, promotion, and career path; % " .78). factor solution, when compared with the four-factor model, gen-
These scales consisted of eight and seven items, respectively. erated lower scores on the robust CFI (.637) as well as the point
344 O’NEILL AND SEVASTOS
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Figure 1. The job insecurity measurement model.

estimate of .159 for the RMSEA, with an unacceptable 90% for the invariance of loadings across groups, the $2 increment from
confidence interval that ranges from .149 to .168, and a SRMR of the baseline was not statistically significant, '$2 " 15.807, 18 df,
.122. This result indicates that the single-factor model does not fit p ( .05. Holding the factor covariances across groups invariant
the data well. By contrast, the point estimates on the robust CFI after constraining the factor loadings to be equal across groups did
(.954), and the RMSEA (.058) with a 90% confidence interval of not result in a statistically significant difference between nested
.046 to .069, indicate that the four-factor model provides a very models, '$2 " 8.187, 6 df, p ( .05. When the most restrictive
good fit to the data (MacCallum et al., 1996). Similar results are model was tested, which included invariant error variances in
reported for the cross-validation sample, as can be seen in Table 3. addition to loadings and covariances, the results were not statisti-
Hypothesis 1 is therefore supported. cally significant, '$2 " 479.486, 18 df, p ( .05. For this final
Convergent and discriminant validity. Displayed in Table 4 model, the point estimates for the RMSEA (.035; CI [.029, .040]),
are the convergent validities for both samples of the 18 items CFI (.961), and SRMR (.060) all indicate that this model fits the
across the four factors for both subsamples. All of the t-ratios for data well. Hypothesis 2 is therefore supported.
the loadings in both subsamples were statistically significant (p & Intercorrelations and reliability statistics. Displayed in Ta-
.001), indicating that each item in the model was significantly ble 6 are the summary statistics for the main sample (N " 1,004),
related to its intended construct. including means, standard deviations, reliability estimates, and
When the discriminant validities of the four job insecurity intercorrelations for the four job insecurity dimensions. Alpha
constructs were examined for Subsample 1, in every case, the reliabilities for each dimension far exceed the minimum require-
difference in $2 between the baseline model with the covariance ment of .70 (Nunnally, 1978), confirming the reliability of the
parameters set free and the model in which the same parameters instrument. Intercorrelations among the four job insecurity factors
were fixed at 1 was statistically significant. That is, all of the chi are not excessively high, suggesting the subscales are related but
squares for the unconstrained model were substantially lower, independent. In line with previous research (Hellgren et al., 1999),
indicating discriminability among the four subscales. JCI showed a strong correlation with JLI (.57), which is to be
Sample invariance. Table 5 shows the chi-square difference expected in a restructuring environment where layoffs are often
tests with the associated degrees of freedom for nested models accompanied by changes in work roles and responsibilities. Of the
based on the multisample analysis. After imposing all constraints

Table 2
Table 1 Confirmatory Factor Analysis of Calibration Sample
Confirmatory Factor Analysis of Sample 1 Data
Model $2 S-B$2 df RCFI RMSEA (CI) SRMR
Model $2 S-B$2 df RCFI RMSEA (CI) SRMR
Null 3019.718 2495.107 153 — — —
Null 5551.426 — 153 — — — One factor 1223.560 984.723 135 .637 .159 (.149, .168) .122
Four factor 362.267 276.987 129 .973 .048 (.040, .056) .041 Four factor 279.926 236.191 129 .954 .058 (.046, .069) .058
Note. n " 502. Power " 1.00 for RMSEA. CFI " comparative fit index; Note. Power " 1.00 for RMSEA. CFI " comparative fit index; RCFI "
RCFI " robust comparative fit index; RMSEA " robust root mean error robust comparative fit index; RMSEA " robust root mean error of ap-
of approximation; S-B$2 " Satorra-Bentler scaled statistic; SRMR " proximation; S-B$2 " Satorra-Bentler scaled statistic; SRMR " standard-
standardized root mean square residual. ized root mean square residual.
JOB INSECURITY MEASURE 345

Table 3 develop scale items from qualitative interviews with workers fac-
Confirmatory Factor Analysis of Cross-Validation Sample ing an objective threat of job loss. The results of EFA and CFA support
the measurement of job insecurity through the four correlated but
Model $2 S-B$2 df CFI RCFI RMSEA (CI) SRMR distinct subscales of job loss, Job Changes, Marginalization, and
Null 3154.242 2354.741 153 — — — — Organizational Survival Insecurity. Tests of invariance across
One factor 1322.152 1020.154 135 .598 .162 (.152, .171) .126 loadings, factor covariances, and error covariances held across two
Four factor 252.877 218.060 129 .960 .053 (.040, .064) .046 samples, thus confirming the statistical equivalence of the four-
Note. Power " .997 for RMSEA. CFI " comparative fit index; RCFI " factor structure between the two sets of data. In addition to its
robust comparative fit index; RMSEA " robust root mean error of ap- strong construct validity, the results indicate high internal consis-
proximation; S-B$2 " Satorra-Bentler scaled statistic; SRMR " standard- tency reliability for the measure. That the four job insecurity
ized root mean square residual. dimensions generated from qualitative interviews with an Austra-
lian sample were validated and cross-validated on a North American
two new job insecurity subscales, MI showed the strongest asso- sample further corroborates the generalizability of the instrument.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

ciations with JCI (r " .56) and JLI (r " .44). Although OSI These findings, along with the demonstrated criterion-related va-
This document is copyrighted by the American Psychological Association or one of its allied publishers.

showed moderate correlations with JLI (r " .33) and MI (r " .39), lidity of the four job insecurity subscales, support the use of the
stronger associations were found for JCI (r " .51). instrument in academic and applied settings.
Criterion-related validity. Consistent with previous re-
search, all correlations between job insecurity subscales and out- Theoretical and Practical Implications
come measures were statistically significant (p & .001) and in the
expected direction. With respect to both psychological well-being The strong fit of the four-dimensional measurement model
and organizational attitudes, the lowest correlations were generally suggests that job insecurity arises largely as a function of cues in
found for JLI, which demonstrated moderate to strong associations the work environment and social dynamics, indicating the likeli-
with depression#enthusiasm (r " #.38), intrinsic job satisfaction hood of job loss, changes to the job itself, the experience of being
(r " #.41), extrinsic job satisfaction (r " #.41), organizational marginalized, and a lack of information on the strategic direction
commitment (r " #.40), trust in management (r " #.45), and and viability of the organization. This multidimensional measure is
intention to resign (r " .39). These results are fairly consistent expected to pave the way for richer theoretical frameworks to
with previous meta-analyses. Only OSI had a lower association advance our understanding of the causes and consequences of job
with anxiety#contentment (r " #.23) relative to the JLI (r " insecurity.
#.31). By contrast, the strongest associations were reported for The theoretical rational behind the emergence of OSI and MI as
JCI for all of the outcomes including depression#enthusiasm (r " dimensions of job insecurity warrants further discussion. During
#.59), anxiety#contentment (r " #.41), intrinsic job satisfaction periods of organizational change and uncertainty, employees are
(r " #.59), extrinsic job satisfaction (r " #.63), organizational inclined to seek ways of reducing their uncertainty by gaining
commitment (r " #.60), and intention to resign (r " .54). Similar information on the strategic direction and economic viability of the
correlations were found for MI and OSI, with the strongest rela- organization (Casey, Miller, & Johnson, 1997). Research into
tionships found for intrinsic job satisfaction (MI, r " #.57; OSI, organizational change suggests that even when top managers know
r " #.51) and extrinsic job satisfaction (MI, r " #.55; OSI, r " what changes will occur, they are often unable or unwilling to
#.53), followed by organizational commitment (MI, r " #.50; report information on the company’s status or discuss changes
OSI, r " #.57), depression#enthusiasm (MI, r " #.48; OSI, r " with employees (Mirvis & Marks, 1986). Left uncertain about their
#.44), and anxiety#contentment (MI r " #.31; OSI r " #.23). futures, employees are often forced to rely on unintended clues and
rumors, which are typically an ineffective way of reducing anxiety
(Napier, Simmons, & Stratton, 1989). It is therefore reasonable
General Discussion
that a primary contributor to job insecurity (OSI) would be the
The purpose of this study was to develop and validate a new job absence of signs from senior management indicating the financial
insecurity measure that represents, in a parsimonious way, the viability of the company and plans for future growth. Thus, one
salient dimensions of job insecurity from the perspective of the strategy for reducing job insecurity is for management to openly
insecure worker. The study was the first since Lahey (1984) to and honestly communicate this information to employees.

Table 4
Invariance Analysis of the Four-Factor Job Insecurity Measurement Model

Models MFF $2 df S-B $2 '$2 ' df S-B '$2 CFI SRMR RMSEA (CI)

Null 6173.961 306 4842.732 — — — — — —


Baseline (same form and pattern) 532.803 258 454.449 — — — .957 .053 .039 (.033, .045)
Loadings Invariant 548.610 276 469.836 15.807a 18 15.387a .957 .057 .037 (.032, .043)
Loadings ) Factor Covariances Invariant 556.797 282 474.670 8.187a 6 4.834a .958 .060 .037 (.031, .043)
Loadings ) Covariances ) Error Variances Invariant 580.616 300 479.486 4.816a 18 13.20a .961 .060 .035 (.029, .040)
Note. Power " 1.00 for RMSEA. CFI " comparative fit index; CI " confidence interval; MFF $2 " minimum fit function chi square; RMSEA " root
mean square error of approximation; S-B $2 " Satorra-Bentler scaled chi square; SRMR " standardized root mean square residual.
a
Nested model chi-square differences are not statistically significant.
346 O’NEILL AND SEVASTOS

Table 5
Standardized Coefficients and T-Ratios for Subsamples 1 and 2

Subsample 1 Subsample 2
Standardized Standardized
Job insecurity items Factor coefficient z coefficient z

1. The possibility of losing my job occupies my


thoughts constantly. F2 (JLI) .695 14.441 .774 15.372
2. No matter how hard I work there is no guarantee that
I am going to keep my job. — .719 15.931 .725 16.717
3. I am certain of losing my job. — .778 15.688 .779 12.779
4. I’m not sure of how long my job will last. — .804 18.850 .849 22.905
5. I am uncertain about my future with this organization. — .765 16.394 .750 15.343
6. The probability of being laid-off is high. — .842 17.199 .824 15.620
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

7. Senior management is really trying to build this


This document is copyrighted by the American Psychological Association or one of its allied publishers.

organization and make it successful. F3 (OSI) .785 17.417 .841 18.417


8. Management appears to be preparing in advance and
planning for the future. — .850 17.475 .795 14.312
9. This organization seems to have clear goals and a
definite strategy for achieving them. — .676 12.916 .723 12.134
10. Overall, my physical working conditions are likely to
deteriorate. F1 (JCI) .850 12.710 .820 10.761
11. I am expecting unfavorable changes to my job. — .934 14.823 .911 15.289
12. I expect to have fewer resources to meet the
performance requirements of my job. — .698 10.494 .772 12.119
13. The rewards of my job are likely to diminish. — .771 13.582 .724 12.151
14. I will probably lose many features of my job that I
value the most. — .855 17.287 .879 18.890
15. I wish my job could go back to the way it used to be. — .772 16.765 .785 15.916
16. I feel like I am being given the “silent treatment” in
this organization. F4 (MI) .888 13.873 .878 12.854
17. I am often excluded from discussions or meetings that
affect me. — .707 12.514 .810 15.983
18. I feel as though management is avoiding me. — .925 14.931 .925 13.350
Note. The critical values for two tailed tests are z " 1.96, p " .05; z " 2.58, p " .01; z " 3.29, p " .001. JCI " Job Changes Insecurity; JLI " Job
Loss Insecurity; OSI " Organizational Survival Insecurity; MI " Marginalization Insecurity; z " z score or t-ratio (unstandardized coefficients/standard
error).

Given the prevalence of social exclusion in the workplace redundant employees is that their social identity has been tainted,
(Ferris et al., 2008), it may be that incumbents form an association causing them to be excluded from the normal social activities of
between marginalization and job loss based on either their own the work group. This theory is rooted in the phenomenon of social
experience of exclusion or witnessing the exclusion of others, prior stigma proposed by sociologists (Falk, 2001). According to Goff-
to being made redundant. Through this experience, marginaliza- man (1970), a stigma occurs when the actual social identity of
tion would come to be viewed by the employee as a precursor to individuals does not meet society’s normative expectations of the
job loss. One plausible explanation for the marginalization of attributes the individuals should possess. In this way, an individ-

Table 6
Means, Standard Deviations, Alpha Reliabilities, and Pearson Intercorrelations Among Study Variables

Variables Mean SD 1 2 3 4 5 6 7 8 9 10 11

1. Job loss insecurity 3.23 1.52 .901


2. Job changes insecurity 2.93 1.40 .568 .904
3. Marginalization insecurity 2.40 2.40 .438 .564 .882
4. Organizational survival
insecurity 2.62 1.23 .330 .510 .391 .868
5. Job depression–enthusiasm 4.55 .91 #.384 #.590 #.479 #.437 .879
6. Job anxiety–comfort 3.81 .90 #.314 #.414 #.315 #.229 .579 .818
7. Intrinsic job satisfaction 5.02 1.03 #.413 #.592 #.575 #.506 .654 .353 .864
8. Extrinsic job satisfaction 5.01 .97 #.411 #.628 #.546 #.531 .622 .450 .774 .783
9. Organizational commitment 5.43 1.19 #.398 #.596 #.504 #.575 .662 .332 .618 .614 .822
10. Trust in management 5.04 1.25 #.447 #.627 #.572 #.665 .599 .352 .614 .665 .792 .868
11. Intention to resign 2.78 1.73 .391 .539 .451 .405 #.598 #.344 #.587 #.576 #.738 #.601 .918
Note. N " 1,004. Alpha reliabilities appear in the diagonal. All correlations are significant at the .001 level (two-tailed).
JOB INSECURITY MEASURE 347

ual’s social identity is tainted, and they are assumed to be incapa- longitudinal invariance. The ability to meaningfully interpret lon-
ble of fulfilling the role requirements of social interaction. Stig- gitudinal analyses becomes compromised to the extent that unsta-
mata attached to the unemployed are now well documented by ble measurement exists (Vandenberg & Lance, 2000). By employ-
sociologists (Heatherton, Kleck, Hebl, & Hull, 2000). ing multiwave, multivariable modeling, researchers can determine
Interestingly, the MI item identifying management as the source the structural stability of the measurement model over time and it’s
of social exclusion showed the highest standardized coefficient suitability for longitudinal research.
across samples. Given that stigma are highly dependent on social, Third, although significant correlations between job insecurity
economic, and political power (Haslam, Jetton, Postmes, & Has- factors and outcome variables were detected, the strength and
lam, 2009), it makes sense that managers would be especially direction of causation is not clear, given the cross-sectional nature
influential in shaping perceptions of marginalization among em- of the research design. Previous longitudinal research using uni-
ployees. Accordingly, managers can play a pivotal role in mini- dimensional job insecurity measures suggests that job loss insecu-
mizing MI by maintaining group cohesion and inclusiveness dur- rity acts as an antecedent of job well-being (e.g., Hellgren &
ing periods of restructuring. Sverke, 2003). However, given the multidimensional nature of job
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Statistically significant correlations between the four job inse- insecurity, it is likely that the causal relationships are more com-
This document is copyrighted by the American Psychological Association or one of its allied publishers.

curity subscales and a range of outcome variables indicate that the plex, suggesting a need to examine the causal ordering of the four
JIM has demonstrated utility in relation to important outcomes for dimensions in their relationship with worker outcomes over time.
the individual and the organization. Although JLI showed low to It is indeed plausible that the well-documented relationship be-
moderate correlations consistent with previous research, stronger tween global job insecurity measures and attitudinal outcomes may
correlations were found for JCI, MI, and OSI in relation to all but be mediated by JCI, MI, or OSI. Understanding the sequential
one criterion (job-related anxiety#comfort). That JCI showed the pattern of relationships between job insecurity dimensions would
highest correlations compared with the three other subscales ex- shed light on the psychological dynamics set in motion by the
tends Hellgren et al.’s (1999) longitudinal research demonstrating process of downsizing.
the relationship between qualitative job insecurity and work atti- Finally, it must be noted that the measure developed in the
tudes. present study constitutes only a preliminary step in an ongoing
For practitioners interested in monitoring job insecurity and its process of iterative psychometric development. Given that the
consequences, the findings from this study have important impli- validation sample consisted of highly educated and largely full-
cations. A parsimonious and validated job insecurity instrument time workers, researchers are encouraged to adapt the instrument
ensures that the most salient facets of the construct are measured and apply the measurement model to a variety of organizational
without becoming unnecessarily taxing for the respondent. With settings. Through this process of iterative refinement, a more
18 items representing four constructs, the JIM represents a parsi- thorough understanding of the job insecurity measurement model
monious alternative to other multidimensional job insecurity can be achieved.
scales, such as Lee et al.’s (2008) abridged (37 items) and “bare
bones” (25 items) versions of the JIS. This in itself should increase References
the response rate of the questionnaire. In addition, by accurately
Anderson, J., & Gerbing, D. (1988). Structural equation modeling in
measuring the most salient job insecurity dimensions identified by practice: A review and recommended two-step approach. Psychological
insecure workers, the JIM may account for additional variance in Bulletin, 103, 411– 423. doi:10.1037/0033-2909.103.3.411
worker outcome variables over and above that explained by uni- Ashford, S. J., Lee, C., & Bobko, P. (1989). Content, causes, ad conse-
dimensional measures. The instrument further enables human re- quences of job insecurity: A theory-based model and substantive test.
source practitioners to assess the prevalence of the four dimensions Academy of Management Journal, 32, 803– 829. doi:10.2307/256569
and perhaps target interventions to reduce job insecurity more Bagozzi, R. P., & Phillips, L. W. (1982). Representing and testing orga-
effectively. nizational theories: A holistic construal. Administrative Science Quar-
terly, 27, 459 – 489. doi:10.2307/2392322
Bartunek, J. M., & Seo, M.-G. (2002). Qualitative research can add new
Limitations and Future Directions meanings to quantitative research. Journal of Organizational Behavior,
23, 237–242. doi:10.1002/job.132
Although the present study has employed robust statistical pro- Bennett, R. J., & Robinson, S. L. (2000). Development of a measure of
cedures, some limitations are noteworthy. First, the collection of workplace deviance. Journal of Applied Psychology, 85, 349 –360. doi:
self-report data may result in an inflation of effect size due to 10.1037/0021-9010.85.3.349
common method variance. Although the results of the CFA using Bentler, P. M. (2005). EQS 6 structural equations program manual.
Harmon’s single-factor test (Podsakoff, Mackenzie, Lee, & Pod- Encino, CA: Multivariate Software (www.mvsoft.com).
sakoff, 2003) suggests that method variance is not a likely source Bentler, P. M., & Bonett, D. G. (1980). Significance test and goodness-
of bias in the data, we did not control for NA in testing for criterion of-fit in the analysis of covariance structures. Psychological Bulletin, 88,
related validity, and therefore method variance may have influ- 588 – 606.
enced the correlational results. Future research should test this Bentler, P. M., & Chou, C. P. (1987). Practical issues in structural mod-
elling. Sociological Methods & Research, 16, 78 –117. doi:10.1177/
possibility.
0049124187016001004
Second, although the validation of the JIM has focused on North Blau, G., & Andersson, L. (2005). Testing a measure of Instigated Work-
American and Australian samples, future research should examine place Incivility in the workplace. Journal of Occupational and Organi-
the invariance of the instrument on non-English speaking samples zational Psychology, 78, 595– 614. doi:10.1348/096317905X26822
and across demographic categories such as gender, age, education, Bollen, K. A. (1989). Structural equations with latent variables. New
occupational group and level. Equally important is the question of York, NY: Wiley.
348 O’NEILL AND SEVASTOS

Byrne, B. M., Shavelson, R. J., & Muthén, B. (1989). Testing for equiv- Hinkin, T. R. (1995). A review of scale development practices in the study
alence of factor covariance and mean structures: The issue of partial of organizations. Journal of Management, 21, 967–988. doi:10.1177/
measurement invariance. Psychological Bulletin, 105, 456 – 466. doi: 014920639502100509
10.1037/0033-2909.105.3.456 Hu, L., & Bentler, P. M. (1999). Cut-off criteria for fit indexes in covari-
Caplan, R. D., Cobb, S., French, J. R. P., Van Harrison, R., & Pinneau, ance structure analysis: Conventional criteria versus new alternatives.
S. R. (1975). Job demands and worker health: Main effects and occu- Structural Equation Modeling, 6, 1–55. doi:10.1080/
pational differences. Washington, DC: U.S. Department of Health Ed- 10705519909540118
ucation and Welfare. Huang, G. H., Lee, C., Ashford, S. J., & Ren, X. P. (2010). Affective job
Casey, M. K., Miller, V. D., & Johnson, J. R. (1997). Survivors’ informa- insecurity: A moderation of the cognitive job insecurity – employee
tion seeking following a reduction in workforce. Communication Re- outcomes relationships. International Studies of Management and Or-
search, 24, 755–781. doi:10.1177/0093650297024006007 ganization, 40, 20 –39. doi:10.2753/IMO0020-8825400102
Cheng, G. H.-L., & Chan, D. K.-S. (2008). Who suffers more from job Jacobson, D. (1991a). The conceptual approach to job insecurity. In J.
insecurity? A meta-analytic review. Applied Psychology: An Interna- Hartley, D. Jacobson, B. Klandermans, & T. van Vuuren (Eds.), Job
tional Review, 57, 272–303. insecurity: Coping with jobs at risk (pp. 23–39). London, UK: Sage.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Cohen, J. (1960). A coefficient of agreement for nominal scales. Educa- Jacobson, D. (1991b). Toward a theoretical distinction between the stress
This document is copyrighted by the American Psychological Association or one of its allied publishers.

tional and Psychological Measurement, 20, 37– 46. doi:10.1177/ components of the job insecurity and job loss experiences. In S. B.
001316446002000104 Bacharach (Ed.), Research in the sociology of organizations (Vol. 9, pp.
Cook, J., & Wall, T. D. (1980). New work attitude measures of trust, 1–19). Greenwich, CT: JAI Press.
organizational commitment and personal need fulfillment. Journal of Johnson, M., Messe, L., & Crano, W. D. (1984). Predicting job perfor-
Occupational Psychology, 53, 39 –52. doi:10.1111/j.2044-8325.1980 mance of low income workers: The Work Opinion Questionnaire. Per-
.tb00005.x sonnel Psychology, 37, 291–299. doi:10.1111/j.1744-6570.1984
Datta, D. K., Guthrie, J. P., Basuil, D., & Pandey, A. (2010). Causes and .tb01451.x
effects of employee downsizing: A review and synthesis. Journal of Jöreskog, K. G. (1971). Statistical analysis of congeneric tests. Psy-
Management, 36, 281–348. doi:10.1177/0149206309346735 chometrika, 36, 109 –133. doi:10.1007/BF02291393
De Cuyper, N., Bernhard-Oettel, C., Berntson, E., De Witte, H., & Alarco, Katz, D., & Kahn, R. (1978). The social psychology of organizations. New
B. (2008). Employability and employee’s well-being: Mediation by job York, NY: Wiley.
insecurity. Applied Psychology: An International Review, 57, 488 –509. Kinnunen, U., Mauno, S., Nätti, J., & Happonen, M. (1999). Perceived job
doi:10.1111/j.1464-0597.2008.00332.x insecurity: A longitudinal study among Finnish employees. European
De Witte, H. (1999). Job insecurity and psychological well-being: Review Journal of Work and Organizational Psychology, 8, 243–260. doi:
of the literature and exploration of some unresolved issues. European 10.1080/135943299398348
Journal of Work and Organizational Psychology, 8, 155–177. doi: Klandermans, B., & van Vuuren, T. (1999). Job insecurity: Introduction.
10.1080/135943299398302 European Journal of Work & Organizational Psychology, 8, 145–153.
Dekker, S. W., & Schaufeli, W. B. (1995). The effects of job insecurity on doi:10.1080/135943299398294
psychological health and withdrawal: A longitudinal study. The Austra- Konig, C. J., & Staufenbiel, T. (2006). The difference between cognitive
lian Psychologist, 30, 57– 63. and affective job insecurity. Paper presented at the 21st conference of the
Evans, M. G. (1991). The problem of analyzing multiplicative composites: Society for Industrial and Organizational Psychology, Dallas, TX.
Interactions revisited. American Psychologist, 46, 6 –15. doi:10.1037/ Lahey, M. A. (1984). Job security: Its meaning and measure. (Unpublished
0003-066X.46.1.6 doctoral dissertation). Kansas State University, Manhattan, KS.
Falk, G. (2001). Stigma: How we treat outsiders. Amherst, NY: Pro- Lee, C., Bobko, P., Ashford, S., Chen, Z. X., & Ren, X. (2008). Cross-
metheus Books. cultural development of an abridged job insecurity measure. Journal of
Ferris, D. L., Brown, D. J., Berry, J. W., & Lian, H. (2008). The devel- Organizational Behavior, 29, 373–390.
opment and validation of the Workplace Ostracism Scale. Journal of Lim, V. K. (1997). Moderating effects of work-based support on the
Applied Psychology, 93, 1348 –1366. doi:10.1037/a0012743 relationship between job insecurity and its consequences. Work & Stress,
Goffman, E. (1970). Stigma: Notes on the management of spoiled identity. 11, 251–266. doi:10.1080/02678379708256839
Harmondsworth, UK: Penguin Books. MacCallum, R. C., Browne, M. W., & Sugawara, H. M. (1996). Power
Greenhalgh, L., & Rosenblatt, Z. (1984). Job insecurity: Toward concep- analysis and determination of sample size for covariance structure mod-
tual clarity. The Academy of Management Review, 9, 438 – 448. eling. Psychological Methods, 1, 130 –149. doi:10.1037/1082-989X.1.2
Hackman, J. R., & Oldham, G. R. (1975). Development of the Job Diag- .130
nostic Survey. Journal of Applied Psychology, 60, 159 –170. doi: Mauno, S., Leskinen, E., & Kinnunen, U. (2001). Multi-wave, multi-
10.1037/h0076546 variable models of job insecurity: Applying different scales in studying
Haslam, S. A., Jetton, J., Postmes, T., & Haslam, C. (2009). Social identity, the stability of job insecurity. Journal of Organizational Behavior, 22,
health and well-being: An emerging agenda for applied psychology. 919 –937. doi:10.1002/job.122
Applied Psychology: An International Review, 58, 1–23. doi:10.1111/j Mirvis, P. H., & Marks, M. L. (1986). Merger syndrome: Management by
.1464-0597.2008.00379.x crisis. Mergers and Acquisitions, 21, 50 –55.
Heatherton, T. F., Kleck, R. E., Hebl, M. R., & Hull, J. G. (2000). The Mishra, K., Spreitzer, G. M., & Mishra, A. (1998). Preserving employee
social psychology of stigma. New York: Guilford Press. morale during downsizing. Sloan Management Review, 39, 83–95.
Hellgren, J., & Sverke, M. (2003). Does job insecurity lead to impaired Napier, N. K., Simmons, G., & Stratton, K. (1989). Communication during
well-being or vice versa? Estimation of cross-lagged effects using latent a merger: Experience of two banks. Human Resource Planning, 12,
variable modelling. Journal of Organizational Behavior, 24, 215–236. 105–122.
doi:10.1002/job.184 Nunnally, J. (1978). Psychometric theory. New York, NY: McGraw-Hill.
Hellgren, J., Sverke, M., & Isaksson, K. (1999). A two-dimensional ap- OECD. (1997). Is job insecurity on the increase in OECD countries? In
proach to job insecurity: Consequences for employee attitudes and OECD employment outlook (pp. 129 –160). Paris: OECD.
well-being. European Journal of Work and Organizational Psychology, OECD. (1999). Implementing OECD job strategy: Assessing performance
8, 179 –195. doi:10.1080/135943299398311 and policy. Paris: OECD.
JOB INSECURITY MEASURE 349

Podsakoff, P. M., Mackenzie, S. B., Lee, J. Y., & Podsakoff, N. P. (2003). Society for Human Resource Management. (2011). 2011 employee job
Common method biases in behavioural research: A critical review of the satisfaction and engagement: Gratification and commitment at work in
literature and recommended remedies. Journal of Applied Psychology, a sluggish economy. Alexandria, VA: Author.
88, 879 –903. Steiger, J. H. (1999). STATISTICA power analysis. Tulsa, OK: StatSoft.
Probst, T. M. (2003). Development and validation of the Job Security Sverke, M., Hellgren, J., & Näswall, K. (2002). No security: A meta-
Index and the Job Security Satisfaction scale: A classical test theory and analysis and review of job insecurity and its consequences. Journal of
IRT approach. Journal of Occupational and Organizational Psychology, Occupational Health Psychology, 7, 242–264. doi:10.1037/1076-8998.7
76, 451– 467. doi:10.1348/096317903322591587 .3.242
Reisel, W. D., & Banai, M. (2002). Comparison of a multidimensional and Vandenberg, R. J., & Lance, C. E. (2000). A review and synthesis of the
a global measure of job insecurity: Predicting job attitudes and work measurement invariance literature: Suggestions, practices, and recom-
mendations for organizational research. Organizational Research Meth-
behaviors. Psychological Reports, 90, 913–922.
ods, 3, 4 – 69. doi:10.1177/109442810031002
Rosenblatt, Z., & Ruvio, A. (1996). A test of a multidimensional model of
van Wiek, M., & Pienaar, J. (2008). Towards a research agenda for job
job insecurity: The case of Israeli teachers. Journal of Organizational
insecurity in South Africa. South African Business Review, 12, 49 – 86.
Behavior, 17, 587– 605. doi:10.1002/(SICI)1099-1379(199612)17:
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Warr, P. B. (1987). Work, unemployment, and mental health. Oxford, UK:


1)&587::AID-JOB825(3.0.CO;2-S
Oxford University Press.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Schneider, B., Hanges, P. J., Smith, D. B., & Salvaggio, A. N. (2003). Warr, P. B. (1990). The measurement of well-being and other aspects of
Which comes first: Employee attitudes or organizational financial and mental health. Journal of Occupational Psychology, 63, 193–210. doi:
market performance? Journal of Applied Psychology, 88, 836 – 851. 10.1111/j.2044-8325.1990.tb00521.x
doi:10.1037/0021-9010.88.5.836 Warr, P. B., Cook, J., & Wall, T. (1979). Scales for the measurement of
Seashore, S. E., Lawler, E. E., Mirvis, P., & Cammann, C. (1982). Ob- some work attitudes and aspects of psychological well-being. Journal of
serving and measuring organizational change: A guide to field practice. Occupational Psychology, 52, 129 –148. doi:10.1111/j.2044-8325.1979
New York, NY: Wiley. .tb00448.x
Sobel, M., & Bohrnstedt, G. (1985). Use of null models in evaluating the
fit of covariance structure models. In N. Tuma (Ed.), Sociological Received February 21, 2013
methodology (pp. 152–178). San Francisco, CA: Jossey-Bass. doi: Revision received April 22, 2013
10.2307/270849 Accepted April 24, 2013 !

You might also like