Meyer and Allen Model of Organizational Commitment
Meyer and Allen Model of Organizational Commitment
Meyer and Allen Model of Organizational Commitment
as a means of assessing the degree to which each scale taps the theorized
structure of the corresponding constructas defined by Allen and Meyer
(1990); Meyer and Herscovitch (2001) and Meyer et al. (2006) that each is
designed to reflect.
Overall Analysis of Commitment Construct Definitions Meyer and Herscovitch
(2001, p. 301) propose that commitment is a force that binds an individual to
a course of action of relevance to one or more targets. Employees are
theorized to experience this force in the form of three bases, or mindsets:
affective, normative, and continuance, which reflect emotional ties, perceived
obligation, and perceived sunk costs in relation to a target, respectively (Allen
and Meyer, 1990). Thus, any scale that purports to measure organizational
commitment should tap one of these mindsets and should reference the target,
what the employee is committed to, be it the organization, a team, a change
initiative, or a goal.
Additionally, Meyer and Herscovitch (2001, p. 311) argue that commitments
include behavioral terms that describe what actions a commitment implies.
Specifically, these terms can take the form of focal and discretionary behavior.
A focal behavior is one believed to be integral to the concept of commitment to
a particular target, such that all three mindsets should predict this behavior. It
is the behavior to which an individual is bound by his or her commitment. For
example, for organizational commitment, the focal behavior is theorized to be
maintaining membership in the organization. In contrast, discretionary
behaviors are optional, in the sense that the employee has some flexibility in
defining the behavioral terms of his commitment. Some mindsets, but not
others, may predict these behaviors. Meyer and Herscovitch argue that
different behavioral terms should be
1 There is no section that addresses measurement issues specific to the ACS,
because there are no such issues. The only problem identified with the ACS is
that one ACS item lacks the proper item-content structure, and this is
addressed in the first section of the paper because it is a model level
problem that characterizes all the three scales.
9Meyer and Allen Model of Organizational Commitment: Measurement Issues
included in item wording, depending on the kind of behavior the researcher is
trying to predict.
Finally, Meyer et al. (2006) note that commitment has both cognitive and
affective elements. The cognitive elements are the behavioral terms and the
basis of the commitment, and the affective element comprises whatever
feelings a specific mindset invokes (e.g., in the case of NC, pride and/or guilt).
omitting staying/leaving wording from the one ACS item (and also deleting
them from the NCS and CCS) is
11Meyer and Allen Model of Organizational Commitment: Measurement Issues
recommended for two reasons: (1) As Bozeman and Perrewe (2001) note, if a
commitment scale contains behavioral-terms wording, it isnt possible to test
theoretical models that specify behavioral intentions as mediators of the
commitment-behavior relationship (e.g., the Mobley, Horner, and
Hollingsworth, 1978; and Hom and Griffith, 1995 models of turnover); and (2)
the Meyer and colleagues recommendation is seemingly impossible to follow if
we are interested in assessing the relationship between commitment and
multiple behaviors in a single study. That is, if the research goal involves
studying organizational commitment and trying to predict a single behavior,
such as job performance, then the ACS, NCS, and CCS items could be modified
to include performance-related (and not turnover-related) terms. But if the goal
is to study the impact of commitment on turnover, job performance,
absenteeism, and citizenship behavior in the same study, it would be unwieldy,
probably impractical, to modify scale items to refer to all of these behaviors
simultaneously. Thus, the best solution is to purge the notion of behavioral
terms from the commitment construct, and also from its measures.
Also, concerning Meyer and Herscovitchs (2001) explanation that AC predicts a
wider range of behaviors more strongly than do NC or CC: it could be that
rather than AC being an inherently stronger/broader binding force, perhaps AC
is typically a better predictor of non-turnover behaviors such as absenteeism,
job performance, and citizenship because its items do not specifically reflect
turnover cognitions while the NCS and CCS items do. By drawing the
respondents attention to turnover cognitions, the NCS and CCS items might
cause respondents to perceive them as relevant only to that outcome, and thus
mute their reflection on how these forms of commitment bind them to nonturnover related outcomes. Thus, an approach that purges behavior-terms
wording from these scales seems warranted.
Finally, as noted earlier, Meyer et al. (2006) recently proposed that CC has an
affective component, in that an employees experience of CC should be
characterized by feelings such as anxiety or security/insecurity concerning the
sunk costs and side bets that tie them to the organization. Currently, only one
of the original CCS items include wording that taps these feelings (CCS, see
Appendix B, which refers to the respondents being afraid of what might
happen if they lost their jobanother of the CCS items does mention a general
feeling about their commitment). Thus, like the NCS, the CCS may no longer
be up-to-date with respect to recent conceptual modifications to the construct,
in this case there is a need for items that capture this affective dimension. Note
that Powell and Meyers (2004) high-sacrifice, only version of the CCS, omits
this item and thus does not contain any items that directly tap affective
content; nor does the 1993 version of the scales, the implication being that
CCS items should be reworded to capture this affective aspect of CC.
The recommendations for eliminating behavioral terms wording from all of the
scales and adding affective content to the CCS should precede the
recommendations made in the remainder of the paper, since these
recommendations pertain to the fundamental correspondence between the
scales and the constructs they purport to measure.
The Icfai Journal of Organizational Behavior, Vol. VI, No. 4, 200712
NCS Measurement Issues This section addresses some important measurement
issues related specifically to the NCS: the discriminant validity between the
NCS and the ACS; whether the item wording of the NCS has evolved as the
concept of NC has evolved; and the discriminant validity between the NCS and
the CCS.
Discriminant Validity of NCS vs. ACS As noted by Bergman (2006), in the US
and Canadian studies, neither the eight-item nor the six-item NCS has been
shown to have a high degree of discriminant validity with the ACS. Although the
NCS items invariably load on a separate factor from ACS items in Confirmatory
Factor Analysis, both versions of the scale tend to be highly correlated with the
ACS (0.77 for the 1993 six-item and 0.54 for the original 1990 eight-item
versions, Meyer et al., 2002). While these correlations are not indicative of
complete redundancy between AC and NC, in Western countries NC often offers
little additional explanatory power when modeled as a predictor of outcomes in
conjunction with AC.
Somewhat paradoxically, in non-Western cultures, the NCS and the ACS tend to
be even more highly correlated, and yet the NCS arguably has shown greater
discriminant validity in these settings, since it tends to contribute significantly
to outcome prediction. For example, in the analysis of Chang et al. (2007), a
Taiwanese sample reported a correlation between the ACS and NCS of 0.66, but
structural-equation analysis showed that even when controlling the ACS, the
NCS was a strong predictor of turnover intentions. Likewise, Chen and
Francescos (2003) Chinese study found a correlation between the ACS and
NCS of 0.64, but also found that the NCS played a key role in moderating the
relationship between the ACS and three dimensions of citizenship behavior.
Thus, despite high intercorrelations, there is an evidence of construct
distinctiveness, at least in Eastern cultures, perhaps reflecting the collectivist
natures of those cultures, in which commitment based on obligation might
have more resonance (Meyer and Allen, 1997).
Conceptual Changes in NC Allen and Meyer (1990) first introduced the concept
of NC. Since then, researchers have noted that the definition of NC has
changed (Allen, 2003 and Bergman, 2006). The original 1990 NCS was
designed to capture an NC construct that was based largely on Wieners (1982)
work on the internalization of social loyalty norms to organizations. In 1993, NC
was reconceptualized somewhat as an obligation to stay with the organization,
without specific reference to social pressures about loyalty (Meyer et al., 1993).
This conceptual shift was built into the revised 1993 NCS. However, recently,
NC has been altered again, to reflect reciprocity for a benefit (Meyer et al.,
2002), and still more recently, Meyer et al. (2006) refined this reciprocity theme
further, seemingly positing a two-dimensional concept of NC that includes an
indebted obligation aspect reflecting the perceived need to meet others
expectations, which is theorized to be correlated with CC and a moral
imperative aspect that reflects striving to meet valued outcomes, which is
theorized to be correlated with AC.
13Meyer and Allen Model of Organizational Commitment: Measurement Issues
Thus, the NCS, which hasnt been revised since 1993, has not been modified to
keep up with the recent conceptual revisions, and thus probably does not
adequately reflect the theorized construct, which now bears a strong
resemblance to social exchange-based constructs such as the psychological
contract (Rousseau and Wade-Benzoni, 1995). An analysis of the original eightitem scale (Appendix B), shows that only item seven, which references being
taught to remain loyal to ones organization and thus seems to tap the
indebted obligation theme, directly taps either of these dimensions. The sixitem 1993 scale includes two items that mention obligation, but neither
indicates that the source of this obligation is others expectations. Nor do any
of the items reflect the moral imperative theme.
Discriminant Validity of NCS vs. CCS Powell and Meyers (2004) analysis of CC
found that contrary to expectations, social-cost side bets predicted NC stronger
than CC, and based on this speculation predicted that NC might be a special
form of CC. This idea has face validity, since a perceived obligation to remain
with the organization, particularly one based on reciprocity, could be
experienced by the employee as a psychic cost, in the form of guilty
conscience, that would have to be incurred if the employee were to break the
obligation and leave. Similarly, Wasti (2002, p. 529), noting that Beckers
original commitment theory contemplated generalized cultural expectations
about remaining as an antecedent, says that Becker has in effect proposed
that CC does not develop from calculative costs only, but has normative bases
(my emphasis) as well. This further confounds the issue of discriminant
validity of the NCS as measuring a distinct form of commitment, since it implies
that NC might be an antecedent of CC.
Research Recommendations for the Refinement of NCS Concerning the
discriminant validity between the ACS and NCS, researchers who feel
comfortable using the 1990 and 1993 definitions of NC would get benefit from
comprehensive item-level confirmatory factor analysis or item response
analysis to determine the redundancy between ACS and NCS at the item, level
(not just scale) as a means of pinpointing redundant items and thus clarifying
the distinctiveness of the NC construct. Ideally, this would be done on samples
drawn from both western and eastern cultures, so that the cross-cultural
validity of the item-level analysis can be established. It might be that a twocomponent model of commitment (CC and AC) is a better descriptor of an
employees commitment in some cultural settings.
Concerning the issue of the bi-dimensional nature of NC, the NCS should be
rewritten to include items that specifically tap the moral imperative and
indebted obligation dimensions of Meyer et al. (2006). Doing so will also help
resolve the discriminant validity issues regarding AC and CC, since whether NC
is an antecedent or dimension of CC, or has discriminant validity with AC, can
be validly tested only after we are confident that NC is being measured with
appropriate items. Rewriting the NCS to reflect these dimensions should be
done after or concurrently with the removal of behavioral terms wording as
recommended in the preceding section of the paper.
The Icfai Journal of Organizational Behavior, Vol. VI, No. 4, 200714
CCS Measurement Issues Since McGee and Fords (1987) seminal study,
researchers have been concerned with the item wording and factor structure of
the CCS. Recently, researchers studying CC have formulated differing views
about how specific CCS items should be with regard to the kinds of costs and
side bets that it taps. Also, advances in factor model structure research open
the door to analyze whether or not CC is a higher-order construct with two subdimensions.
Specificity of CCS Items Allen (2003, p. 242) noted that CC reflects the degree
to which the employee recognizes, or is aware that she or he is bound to stay
because of the costs associated with leaving, not the mere existence of the
costs themselves. If, objectively, the employee has incurred costs, but
subjectively is not aware of these costs and thus does not experience them as
binding him/her to the organization, then CC is not present. Further, the level
of awareness can stem from various events or perceptions, the nature or
substance, which can be quite idiosyncratic to the individual. For this reason,
Allen argues that the best CC items are those that capture the recognition of
perceived costs but do so without reference to their specific source.
The emphasis on perceived costs is also important because it pertains to the
accumulated evidence that the CCS is bi-dimensional, i.e., it includes items that
reflect both High Sacrifices (HS) that would be incurred in order to leave the
organization, and Low Alternatives (LA) items that reflect the degree to which
since failure to do so would mean that we are omitting a key component of the
CC construct (Mackenzie et al., 2005).
Research Recommendations for CCS Concerning the issue of the specificity of
CCS items, to an extent, Wastis (2002) position has merits. There is a rich
research tradition on calculative commitment, which is based on economic
exchange, which traces its theoretical lineage back to March and Simon (Mayer
and Schoorman, 1996), and that has been shunted aside over the past 15 or so
years as the Meyer/Allen model has gained ascendancy in the literature. Future
research could compare measures of calculative commitment against Meyer
and Allens CCS to determine which kind of cost-based commitment has the
most explanatory power. But within the context of the Meyer/Allen model,
Allens (2003) perspective is more compelling. CC is based on Beckers (1960)
side-bet theory, and as Wasti (2002) acknowledges, Becker contemplated
commitment based on a broad array of costs, ranging from economic to
cultural and social to psychological. Thus, the CCS should include items that
encompass this full range of possible costs, even if this means some
conceptual clarity between CC and AC/NC is lost.
Concerning the other possible CCS models issue, the reflective model
possibility could be empirically investigated using second order CFA with CC
posited as a second order factor that underlies the HS and LA sub-dimensions.
The critical tests to be performed is to determine if this second order factor
structure provides a good fit to the data, and if the second order CC construct
directly impacts on outcomes such as turnover behavior and job performance,
while direct paths from HS and LA to these outcomes are non-significant.
The Icfai Journal of Organizational Behavior, Vol. VI, No. 4, 200716
The formative model possibility could be assessed by modeling CC as a latent
construct that is formed by the combined influence of latent LA and HS
constructs. This is not the same as simply adding the scores on all of the LA
and HS items to create a single CC scale (as has been the usual practice in the
literature). The CC construct would include a construct-level error term, which
implies that the CC construct has surplus meaning (Mackenzie et al., 2005),
including the possibility that perhaps some third factor, in addition to HS and
LA, contributes to the formation of CC as well. Thus, the aggregate CC
construct cannot be reduced to a simple mathematical sum of its indicators, in
this case the HS and LA constructs. Tests of these models may reveal that the
LA items may yet to be found to play a role in the formation or reflection of CC.
However, even if empirical tests of the formative/reflective model indicate a
very good fit, these results might still be trumped by the conceptual argument
that low alternatives items do not fall within the proper domain of CC, as
conceived by Becker (1960). The tests for the usefulness of formative and
reflective models should not be conducted until the CCS items have been
restructured so as to remove behavioral terms wording, and have been
rewritten to include affective content, as recommended in the analysis of
Meyer and Allens framework section.
Recommendations for Further Research Two remaining issues that are
important for further refinement of the Meyer/Allen measures are the
generalizability of the model in a micro sense, and the relationship between
the commitment constructs and recently developed work attitudes that appear
to tap at least a part of the same conceptual space. Note that these issues
should not be addressed until the recommendations outlined in the preceding
sections of the paper have been implemented, since tests of generalizability
and nomological net placement should not be conducted until valid and
structurally sound commitment scales have been constructed.
Micro Generalizability Macro generalizability refers to the generalizability of the
Meyer/Allen commitment measures to populations beyond the Western
population in which it was originally developed. Currently, a robust research
program studying the validity of the model in East Asian cultures (e.g., Chen
and Francesco, 2003 and Chang et al., 2007), Mid-Eastern cultures (e.g., Wasti
2002 and Cetin 2006), and Eastern Europe (e.g., Vandenberghe et al., 2001) is
being implemented.
In contrast, micro generalizability refers to the validity of the model within subpopulations in the broader Western culture in which it was developed.
Historically, research on organizational commitment within the Meyer and Allen
paradigm has focused on full-time paid organizational members. But, not all
organizational members have these characteristics. A growing number of
organizations are employing part-time, temporary,
17Meyer and Allen Model of Organizational Commitment: Measurement Issues
and contract workers (Meyer et al., 2006), and some organizations include
volunteers. Concerning the commitment of members of these sub-population,
research is largely lacking, though some initial steps have been taken. For
example, Dawley et al. (2005) extended the analysis of the Meyer/Allen
constructs to one such sub-population, volunteers, specifically chamber of
commerce board members, and found that the high sacrifice CCS subscale
did not measure a meaningful construct.
Dawley et al. (2005) explained this finding by noting that volunteer chamber
board members often have paid employment with other organizations, and
would not suffer monetary losses to the same degree as regular paid
employees. They argue that their findings should not be interpreted as HS is
not an important basis of commitment for volunteer chamber of commerce
board members, but instead that the CCS contains items that are too narrowly
(Contd...)
The Icfai Journal of Organizational Behavior, Vol. VI, No. 4, 200724
Appendix A (...Contd)
Moral Imperative Dimension
I am loyal to this organization because my values are largely its values.
This organization has a mission that I believe in and am committed to.
I feel it is morally correct to dedicate myself to this organization.
Original Commitment Scale Items (Allen and Meyer, 1990)
Affective Commitment Scale Items
1. I would be very happy to spend the rest of my career with this organization.*
#
2. I enjoy discussing about my organization with people outside it.#
3. I really feel as if this organizations problems are my own.#
4. I think that I could easily become as attached to another organization as I
am to this one.(R) #
5. I do not feel like part of the family at my organization.(R) #
6. I do not feel emotionally attached to this organization.(R) #
7. This organization has a great deal of personal meaning for me.#
8. I do not feel a strong sense of belonging to my organization.(R) #
Continuance Commitment Scale Items
1. I am not afraid of what might happen if I quit my job without having another
one lined up.(R) * #
2. It would be very hard for me to leave my organization right now, even if I
wanted to.*
3. Too much in my life would be disrupted if I decided to leave my organization
now.*
4. It wouldnt be too costly for me to leave my organization now.(R) *
5. Right now, staying with my organization is a matter of necessity as much as
desire.*
6. I feel that I have very few options to consider leaving this organization.+ *
Appendix B
(Contd...)
25Meyer and Allen Model of Organizational Commitment: Measurement Issues