1. Universiteit Leiden - Faculteit der Letteren
Sinologisch Instituut
Peter J. Zobel
CULTURAL RELEVANCE OF THE
ORGANIZATIONAL IDENTIFICATION CONSTRUCT
IN CHINA
the case of foreign-invested enterprises
Doctoraalscriptie
2. Table of Contents
Page
Preface 1
The Importance of Organizational Identification 2
1 Literature Review 3
1.1 Review of Theoretical Concepts 3
1.1.1 The Emic/Etic Approach in Cross-Cultural Research 3
1.1.2 China’s Culture 5
1.1.2.1 Hofstede’s Dimensions of Culture 5
1.1.2.2 Collectivism 7
1.1.2.3 Motivation in China 8
1.1.3 Social Identity Theory and the Organization 9
1.1.3.1 Social Identity Theory 9
1.1.3.2 Organizational Identification 10
1.1.4 Related Concepts 13
1.1.4.1 Corporate Identity 13
1.1.4.2 Climate 13
1.1.4.3 Corporate Culture 14
1.1.4.4 Commitment 14
1.2 Models of Organizational Identification 15
1.2.1 Ashforth and Mael’s Model 15
1.2.1.1 Description 15
1.2.1.2 Discussion 16
1.2.2 Rotterdam Organizational Identification Test (ROIT) 17
2 Analysis 18
2.1 Sample and Procedure 18
2.2 Instrument and Variables 19
2.3 Preliminary Analysis and Data Description 21
2.4 Data Analysis 22
2.4.1 Factor Analysis 23
2.4.1.1 Organizational Identification 23
2.4.1.2 Perceived Organizational Prestige 24
2.4.1.3 Job Satisfaction and Commitment 24
2.4.1.4 Goals and Values 25
2.4.1.5 Climate 25
2.4.1.6 Communication Climate 26
2.4.1.7 Received Information 26
2.4.2 Regression Analysis 27
3. 2.4.2.1 Ashforth and Mael’s Model 28
2.4.2.2 ROIT Model 30
3 Discussion 32
3.1 Limitations 32
3.1.1 Sample 32
3.1.2 Construct 33
3.1.2 Method 33
3.2 Assessing the Cultural Validity of Organizational Identification 34
3.2.1 Ashforth and Mael’s Model 34
3.2.2 ROIT Model 35
3.2.3 Cultural Influences on the Organizational Identification Construct 36
3.3 Implication 37
3.3.1 Implications for Organizational Identification in China 37
3.3.2 Implications for Further Research 37
Conclusion 38
Bibliography 39
Appendix 44
1 Organizational Identification Questionnaire in English 45
2 Organizational Identification Questionnaire in Chinese 52
3 Summary Variables 65
4. Preface
Inspired by Hofstede’s cross -cultural research and Professor van Riel’s corporate communication class, I asked myself,
whether the identification of employees with their company is influenced by their nationality. My starting hypothesis was
that some of the factors that influence organizational identification will be more salient in China because of cultural dif-
ferences between Eastern and Western cultures. Thus, the original title of my thesis was “Cultural Differences in Organ-
izational Identification between Chinese and Germans”.
After a promising start in China, where the response rate to my survey was relatively high, problems emerged where I did
not expect them at all. The German company, which belongs to the same multinational as the Chinese enterprises and was
supposed to serve as a benchmark for comparing China with Germany, was no longer willing to participate. In the opin-
ion of management, asking questions about job satisfaction and communication climate would have posed “the risk that
restlessness and instability is unnecessarily brought into the staff” (letter of the personnel manager).
So I landed up with having more than 300 completed Chinese and 150 blank German questionnaires. Because I was un-
able to find another company that was willing to be surveyed, I decided to shift the focus of the thesis to the cultural rele-
vance of the organizational identification construct and the instrument employed. This was only the second best solution
because the items in the questionnaire were not primarily selected on basis of their validity. The questionnaire was based
on a Dutch questionnaire, which was developed by the Corporate Communication Center of Erasmus University Rotter-
dam. Because of the length of the original questionnaire, some parts were omitted while in other parts items were deleted
or altered to shorten the questionnaire to a manageable size. These changes, however, made a comparison with published
data on reliability difficult.
Two models, namely Ashforth and Mael’s (1989) and Smidts et al.’s (1999), were selected to test the validity of the o r-
ganizational identification construct in China because of their comprehensive predictions. After factor analyzing the ques-
tionnaire to verify the factors involved, these predictions were used to validate the models in the Chinese context by em-
ploying multiple regression techniques.
I cannot possibly thank everyone who was involved in this thesis but I want to give special thanks to Professor van Riel,
who made the Rotterdam Organizational Identification Test questionnaire available to me, and to Dr. Arnold and William
Valentino of the holding company in China, who had the courage to give their placet to this survey despite the ‘risk’
involved. I also want to thank Liu Zhengrong who checked my Chinese translation of the questionnaire and hosted me
during my stay in China. Furthermore, I want to thank Mr. Thomas, who tried to help me find a German company, Sherry
Xu, who informed the human resource departments of the upcoming survey, and Wendy Huang, Sherry Lee, Liu Zhen-
grong, Yan Jie, and Zhu Yiqun, who assisted me in carrying out the surveys.
Peter Zobel
Amsterdam, February 2000
Page 1 of 65
5. The Importance of Organizational Identification
The rapid growth of foreign direct investment in China in the last twenty years has impacted not only China’s national
economy but also the Chinese labor market. The serious shortage of skilled labor has led to increasingly higher compen-
sation levels and a high rate in staff turnover (Ma, 1998: 2; Mandel and Labonté-Klajda, 1998: 7). The turnover problem
is one of the greatest challenges for foreign companies because localization is important to make foreign enterprises prof-
itable (Business China, 8 June 98). Because of fierce competition for these skilled employees, pure material rewards seem
no longer sufficient. Therefore, more and more companies compete for these employees by offering non-cash benefits,
like housing loans or even stock options (Mandel and Labonté-Klajda, 1998: 7; SCMP, 13 August 98).
Another measure that might prove a useful retention tool is corporate communication. A strong corporate identity, which
presents the organization in a consistent way to external target groups, can, for example, attract highly qualified appli-
cants. Yet, communication with stakeholders outside the company is just one side of corporate communication. On the
other side, a powerful corporate identity enhances the likelihood of identification with the company (Antonoff, 1986: 21;
Van Riel, 1995: 29). Organizational identification is closely related to a comfortable working environment, friendships at
work, and mutual trust. All these are predictors of low turnover and high productivity (Abrams et al., 1998: 1037;
CEIBS, 1998: 2; The Economist Conferences, 1998: 35; Ma, 1998: 9, Mandel and Labonté-Klajda, 1998: 7; Scott et al.,
1999: 427; WSJE, 18 January 00).
Before rushing to strengthen corporate communication efforts in China, foreign-invested companies should first analyze
the status quo to identify problem areas. One way to do so is to measure the degree to which employees identify with
their company. But there is some danger in simply administering an organizational identification questionnaire in China.
Up to now, it is unclear whether these tests, which were developed in America and Europe, are culturally appropriate
(Yang, 1986: 163). Equating measures of both antecedent and consequent variables in two or more societies may prove
difficult (Mrinal et al., 1994: 32; Smith and Bond, 1993: 47). Problems may also arise when one tries to interpret the
results and does not take the different cultural backgrounds into account (Matsumoto, 1994a: 5).
The purpose of this paper is then to assess the cultural relevance of the organizational identification construct in China
by administering a Western questionnaire in foreign-invested enterprises that belong to a large German multi-national in
the pharmaceutical and chemical industry. With this “imposed etic” approach (Berry, 1989: 726) cross -culturally valid
parts of the questionnaire can be identified and the construct in the Chinese context evaluated.
To accomplish this, I start with a review of relevant literature and describe two models of organizational identification.
Next, I discuss the results of an organizational identification survey in Chinese subsidiaries of a German multinational.
After factor analyzing and reliability testing the observed data, the relation between antecedent and consequent variables
of organizational identification is mapped by means of multiple regression. Finally, I compare the predictions of the two
models with actually observed results and discuss limitations and further research directions.
Page 2 of 65
6. Cultural Relevance of the Organizational Identification Construct in China
- The Case of Foreign-Invested Enterprises
1. Literature Review
“Grau, teurer Freund, ist alle Theorie, und grün des Lebens goldener Baum.”
Johann Wolfgang von Goethe Faust
Before looking at two specific models of organizational identification, I first give an introduction into the problems of
cross-cultural research and the emic/etic approach. Next follows a description of the Chinese culture to clarify the cultural
issues in this study. After that, I briefly discuss social identity theory on which organizational identification is based.
After looking into the different aspects of organizational identification I provide definitions on related concepts to make
clear the distinctions between the latter and organizational identification.
1.1 Review of Theoretical Concepts
1.1.1 The Emic/Etic Approach in Cross-Cultural Research
“The scientist who adopts the emic approach cannot, by definition, do cross-cultural work. The one who
adopts the etic can easily miss the most important aspects of the phenomena he wishes to study.”
Harry Triandis The Analysis of Subjective Culture
Starting from linguistics, where Kenneth Pike came up with the concept of emics and etics, the emic/etic concept found
its way into many other disciplines, among them anthropology and cross-cultural psychology (Headland, 1988: 15-17).
Matsumoto defines it as follows: “An etic refers to findings that appear to be consistent across different cultures; that is,
an etic refers to a universal truth or principle. An emic, in contrast, refers to findings that appear to be different across
cultures; an emic, therefore, refers to truths that are culture-specific” (1994a: 5, author’s emphasis).
Although both approaches have their weaknesses, “even the specialist, coming from one culture to a sharp ly different
one, has no other way to begin its analysis than by starting with a rough, tentative (and inaccurate) etic description of it”
(Pike, 1967: 40). Berry suggested therefore an ‘imposed etics-emics-derived etics’ approach. “The initial step is usu ally
taken armed with a concept or instrument rooted in the researcher’s own culture (one that is really an emic concept or
instrument for that culture) but which is used as an etic orientation, in two senses: it is assumed by the researcher to be a
valid basis for studying a phenomenon in another culture (the tools for this being brought in from outside for the purpose)
and it is assumed by the researcher to be a valid basis for comparing the phenomenon in the two cultures” (Berry, 1989:
Page 3 of 65
7. 726, author’s emph asis). The results of this approach are then compared with the original concept for the shared features.
These “common a spects for which comparison takes place” are then termed ‘derived etics’ (Berry, 1989: 727).
To assess the cross-cultural validity of a concept or instrument, the assumptions of the imposed etics approach have to be
verified. The similarity in the network of relationships has to be demonstrated to ensure that the same underlying con-
struct is being measured. If the relevance of the construct or the validity of the instrument in the new culture cannot be
demonstrated it is called a “pseudoetic approach”. For Triandis, who coined this term, this is an “emic approach deve l-
oped in a Western culture (usually the United States) which is assumed to work as an etic approach”, where “instruments
based on American theories, with items reflecting American conditions, are simply translated and used in other cultures”
(1972: 39).
For this study an imposed etics approach will be used. The aim is, however, not to show whether the organizational iden-
tification construct is etic but to validate this construct in the setting of foreign-invested enterprises in China. The Chinese
sample, which consists of Chinese employees of enterprises, in which the same German multinational has a majority
stake, is by no means representative for the Chinese workforce let alone the Chinese people. It is in this limited set that
the validity and relevance of the organizational identification construct is tested.
Page 4 of 65
8. 1.1.2 China’s Culture
“Until we leave our community, we often remain oblivious to the dynamics of our shared culture.”
Nancy Adler International Dimensions of Organizational Behavior
1.1.2.1 Hofstede’s Dimensions of Culture
It is important to realize that there are differences between countries that have their roots in culture. These differences can
be characterized in terms of Hofstede’s dimensions of national culture. The dimensions are:
Power distance, which Hofstede defines as “the extent to which the less powerful members of institutions and o r-
ganizations within a country expect and accept that power is distributed unequally” (1994: 28; emphasis omitted).
Individualism, which “pertains to societie s in which the ties between individuals are loose: everyone is expected to
¡
look after himself or herself and his or her immediate family”. The opposite of individualism, collectivism, “pertains
to societies in which people from birth onwards are integrated into strong, cohesive ingroups, which throughout peo-
ple’s lifetime continue to protect them in exchange for unquestioning loyalty” (Hofstede, 1994: 51, emphasis omi t-
ted).
Masculinity, which “pertains to societies in which social gender roles are clearly d istinct (i.e., men are sup-posed to
¡
be assertive, tough, and focused on material success whereas women are supposed to be more modest, tender, and
concerned with the quality of life)”. Femininity, on the opposite side of the scale, “pertains to societies i n which so-
cial gender roles overlap (i.e., both men and women are supposed to be modest, tender, and concerned with the qual-
ity of life)” (Hofstede, 1994: 82 -83). As this terminology is rather misleading and easily misunderstood, Adler’s te r-
minology will be used in this paper. Adler relabeled masculinity ‘career success’ and femininity ‘quality of life’
(Adler, 1997: 47 and note 2).
Uncertainty avoidance is the fourth dimension that Hofstede found in his empirical research. It is defined as “the
¡
extent to which the members of a culture feel threatened by uncertain or unknown situations” (Hofstede, 1994: 113,
emphasis omitted). Later on, the Chinese Culture Connection found that this dimension is not meaningful in all coun-
tries, which means that it is emic. These researchers also found that there is another dimension in Asian countries,
which they labeled ‘long-versus-short-term orientation’ or ‘ Confucian work dynamism’. According to Hofstede
‘uncertainty avoidance’ is related to the ‘search for absolute truth ’ while ‘Confucian dynamism’ related is to the
‘search for virtue’ (1994: 171).
The tables on the next page show the relative position of countries with Chinese culture1 on these dimensions in
Hofstede’s original research in 1968 and 1972 (Table 1.1) and the results of a recent research by Fernandez et al. in
1989/1990 (Table 1.2).
1
Mainland China was not included at that time.
Page 5 of 65
9. Table 1.1 Relative Position of Countries with Chinese Culture (Hofstede)
Power Career Uncertainty Confucian
Individualism
Distance Success Avoidance Dynamism
China (PRC) N/A N/A N/A N/A 118
Hong Kong 68 25 57 29 96
Taiwan 58 17 45 69 87
Singapore 74 20 48 8 48
Germany 35 67 66 65 31
Mean 56.83 43.06 47.89 65.45 46.26
Sources: Hofstede, 1994, Table 2.1 (p. 26), Table 3.1 (p. 53), Table 4.1 (p. 84), Table 5.1 (p. 113), Table
7.1
(p. 166); own calculations.
Table 1.2 Relative Position of Mainland China (Fernandez et al.)
Power Career Uncertainty Confucian
Individualism
Distance Success Avoidance Dynamism
China (PRC) 14.50 10.38 15.27 14.46 N/A
Germany 11.89 11.64 10.46 12.36 N/A
Mean 12.72 11.55 11.46 14.03 N/A
Source: Fernandez et al., 1997: 47-49.
Both surveys show a high score of China or Chinese cultures on power distance and collectivism (i.e. a low score on
individualism) relative to the mean score of the surveyed countries. On Confucian dynamism, a dimension originating in
the Chinese culture, China scores highest. The mixed results for uncertainty avoidance might be due to the cross-cultural
invalidity of this dimension.
The differences between China and Germany are quite obvious for all dimensions. While China is characterized by large
power distance and low individualism, Germany scores lower than average on power distance and is relatively more indi-
vidualistic. For the other dimensions, this pattern is repeated: Germany and China are always on opposite sides of the
scale.
Page 6 of 65
10. 1.1.2.2 Collectivism
The individualism-collectivism dimension is “a key determinant of how cultural influence might affect workplace dyna m-
ics” (Earley, 1993: 319) and therefore especially interesting for this paper. In a collectivist culture, “the group takes
precedence over the individual members’ interests, and is the centre of their loyalty” (Tayeb, 1996: 57). Collectivism
denotes “a balan ce between self-seeking and the maintenance of harmony with the community, as relative to the individu-
alism of western cultures” (Cheng, 1996: 239). In China’s collectivist culture the group is very important: relationships
among the people who generate results are emphasized, rather than the results themselves. This corresponds with Child’s
findings that managers in the PRC “attached significantly more importance than the other groups to (¼) having co -
workers who co-operate well with each other. They attached significantly less importance to having the opportunity for
promotion to higher level jobs” (Child, 1994: 180). Cooperation among members of a group is thus emphasized in China
(Beamer, 1998: 56).
Collective cultures “stress the needs of a group; memb ers identify themselves as individuals through their groups. Hierar-
chical differences and vertical relationships are emphasized; one’s role, status, and appropriate behaviors are more clearly
defined by position” (Matsumoto, 1994b: 119). Therefore, “Chines e may be characterized as placing a high value on
identification with their various in-groups” (Bond, 1996: 225). Important for this study involving the concept of identity
is that the conceptualization of self in China is always in a relational context as a ‘we’ identity (Gao et al., 1996: 282).
Schwartz found that his mainland China samples were high on the importance attributed to ‘hierarchy’ (emphasizing
power and ranking in social affairs and distribution of resources and thus related to power distance) and ‘mastery’values
(emphasizing energetic self-assertion to control the social and physical environments). At the same time, however, these
samples scored low on the importance of ‘egalitarian commitment’ values (emphasizing a transcendence of egocent ric
concerns by embracing an interpersonal morality and social principled-ness) (Bond, 1996: 216-217). Schwartz therefore
concluded that China is “not a prototypical ‘collectivist’ society” (cited in Bond, 1996: 217). To sum it up, Chinese co l-
lectivism pertains to small in-group circles but not to a larger community of people.
Page 7 of 65
11. 1.1.2.3 Motivation in China
Another important difference between Chinese and Western culture, related to collectivism, is the ranking of needs,
which impacts possible incentive schemes. Strongly associated with the individualist pole were the working goals ‘per-
sonal time’, ‘freedom to adopt an own approach’ and ‘challenging work’. For the collectivist pole, however, ‘training
opportunities’, ‘phys ical working conditions’ and ‘use of skills and abilities on the job’ were the most important working
goals (Hofstede, 1994: 51-52).
Achievement in the Chinese context is almost always in terms of group achievement rather than in terms of personal goals
achievement (Ho, 1986: 27; Yu, 1994: 50). Yang reports that in comparison with their American counterparts Chinese
students score high in nurturance, social oriented achievement, and endurance (1986: 110, Table 4.1). In Taiwan, em-
ployees value high salary, good working conditions, and public recognition for their contribution (Hui and Tan, 1996:
365). For Hong Kong, researchers found “higher scores for social needs than in other countries, as well as lower auton-
omy and self-actualization needs” (Lockett, 198 8: 488, author’s emphasis). In mainland China, a general survey of a
large sample of working people in Beijing found that ‘pay’ ranks first followed by ‘good interpersonal relations’, ‘good
working conditions’ and the ‘opportunity to learn’ (Westwood and L eung, 1996: 405, Table 5; see also Fisher and Yuan,
1998: 516).
There are however shifts in the motivational pattern. Hui and Tan, for example, find that “economic reform and modern i-
zation have resulted in Chinese employees’ focusing their attention on pe rsonal, extrinsic rewards” (1996: 367). For
young university graduates recruited by a joint venture (JV), Westwood and Leung report that they “rated aspect of their
career development and intrinsic work factors more highly than ‘opportunity for contributing to society’. They also rated
‘the general public’ ninth in importance out of eleven organizational constituents, and ‘myself’ second” (Westwood and
Leung, 1996: 389). For these graduates the most salient motivating factor for wanting to join a JV was greater develop-
ment opportunities (Westwood and Leung, 1996: 395). They felt “that JVs would be more characterized by trust and
respect of the individual and would better engender a sense of belonging and pride in the company” (West -wood and
Leung, 1996: 396).
A survey in Taiwan (Hui and Tan, 1996: 368) found that older people did more strongly endorse collective interests than
younger ones. In another survey by Wang reported in Hui and Tan younger employees “valued personal growth and
achievement, as well as extrinsic rewards such as status and income” (Hui and Tan, 1996: 368). This is consistent with
Fisher and Yuan’s study where promotion and growth were more important to younger employees (1998: 526). Thus, a
distinction has to be made between employees of state-owned enterprises (SOEs) and employees of private or foreign-
invested enterprises, as well as between older and younger employees.
Page 8 of 65
12. 1.1.3 Social Identity Theory and the Organization
“However large or small the membership, the need for cohesion between the members exists in all
organizations, not just companies.”
Arie de Geus The Living Company
1.1.3.1 Social Identity Theory
Social Identity Theory was developed and formalized by Henri Tajfel in 1978. He postulated that, at least in Western
society, “an individual strives to achieve a satisfactory concept or image of himself” (Tajfel, 1978b: 61). Social identity is
then “that part of an individual’s self -concept which derives from his knowledge of his membership of a social group (or
groups) together with the value and emotional significance attached to that membership” (Tajfel, 1978b: 63, author’s
emphasis). A description of the essence of a group includes one to three of the following components: “a cognitive com-
ponent, in the sense of the knowledge that one belongs to a group; an evaluative one, in the sense that the notion of the
group and/or of one’s membership of it may have a positive or a negative value connotation; and an emotional component
in the sense that the cognitive and evaluative aspects of the group and one’s membership of it may be accompanied by
emotions (such as love or hatred, like or dislike) directed towards one’s own group and towards others which stand in
certain relations to it” (Tajfel , 1978a: 28-29). In a recent research note, Bettencourt and Hume were able to confirm this
notion (1999: 119).
The social identity of an individual “can be made up of identifications with many different groups” (Abrams, 1992: 59).
But for membership in a group it is essential “that the individuals concerned define themselves and are defined by others
as members of a group” (Tajfel and Turner, 1979: 40). This has four consequences: First, “an individual will tend to
remain a member of a group and seek membership of new groups if these groups have some contribution to make to posi-
tive aspects of his social identity” (Tajfel, 1978b: 64). Second, an individual will tend to leave the group if it does not
make a positive contribution, and leaving the group psychologically and/or objectively is a viable alternative. Third, if it
is impossible to leave the group, the individual will either try a more positive reinterpretation or actively attempt to
change the situation from within the group. Fourth, “the ‘positive as pects of social identity’ and the reinterpretation of
attributes and engagement in social action only acquire meaning in relation to, or in comparison with, other groups” (T a-
jfel, 1978b: 64). This last point is important for models of organizational identification because characteristics of one’s
group as a whole “achieve most of their si gnificance in relation to perceived differences from other groups and the value
connotation of these differences” (Tajfel, 1978b: 66).
In 1982, Tajfel’s colleague John Tur ner came up with the cognitive redefinition of social identity. He proposed that “a
social group can be defined as two or more individuals who share a common social identification of themselves or, which
is nearly the same thing, perceive themselves to be members of the same social category” (Turner, 1982: 15). Turner
based this minimalist definition of a social group on his observation that “members of a social group seem often to share
no more than a collective perception of their own social unity and yet this seems to be sufficient for them to act as a
group” (1982: 15). He distinguishes between the “Social Cohesion model”, where group -belongingness tends to have an
affective basis, and his “Social Identification model”, where group -belongingness has primarily a perceptual or cognitive
Page 9 of 65
13. basis (Turner, 1982: 16). According to him, the former “has been productive in investigating the dynamics of small, face -
to-face groups” while the latter’s subject matter is “large -scale social category membership such as nationality, class, sex,
race or religion” (Turner, 1982: 22). In Turner’s opinion, “social cohesion may be neither necessary nor sufficient for
group formation” (1982: 22) while it “may arise as a direct product of social identification” (1982: 25).
1.1.3.2 Organizational Identification
It seems that the concept of organizational identification has many different definitions because most articles lack a clear
statement whether organizational identification is used as a noun (to describe a state of being) or as a process.
As a noun, organizational identification refers to the self-definition of a person in terms of organizational membership
(see Turner’s definition of social identification, 1982: 17 -18). This state is the outcome of an identification process and
thus remains transitory.
As a process, organizational identification is both the process of identifying and of being identified. While Hall et al.
define organizational identification as “the process by which the goals of the organization and those of the individual
become increasingly integrated or congruent (1970: 176-177, emphasis added), Mael and Ashforth define organizational
identification as “perceived oneness”, i.e. as an outcome of the identifi cation process (1992: 103).
The nature of identification is dynamic, or as Russo puts it: “Neither the process nor the product of identification is u n-
changing” (1998: 77). Authors that use the static concept of organizational identification acknowledge th is dynamic na-
ture by using a feedback loop from the consequences to the antecedents of organizational identification (Ashforth and
Mael, 1989: 26; Dutton et al., 1994: 253).
Organizational identification not only varies in degree but also in the reasons for identification (O’Reilly and Chatman,
1986: 493). Kagan believes that the ultimate motive for identification is “a desire for the positive goal states commanded
by the model” (1958: 304). Three motives may contribute to this, namely (1) cohesion and affi liation, (2) achievement
and self-actualization and (3) prestige or self-esteem enhancement. Hall and Schneider, for example, distinguish two
paths to satisfy different needs: the single-organization career, where individuals are able to satisfy their ‘needs for secu-
rity and affiliation’, and the multi -organization career, where individuals are “forced to move to find growth opportun i-
ties” (1972: 349). Brown names ‘providing opportunities for personal achievement’, ‘power within the organization’and
the absence of competing sources of identification as criteria for the selection of an object for identification (1969: 346).
Mael and Ashforth state, based on social identity theory, “that individuals identify partly to enhance self -esteem” (1992:
105), while Dutton et al. maintain that “members will find organizations attractive when their social identities there pr o-
vide them with a sense of distinctiveness” (1994: 246). These motives for identification are similar to those described in
various motivational theories, although labels sometimes differ.2
2
Schein summarizes and compares these theories (1980: 85-87). The labels for the different motives are ‘affiliation’, ‘achiev e-
ment’, and ‘power’ in McClelland’s categories, ‘existence needs’, ‘growth needs’, and ‘relatedness needs’ in Alderfer’s categ o-
ries, and ‘affiliation’, ‘self -actualization’, and ‘self -esteem needs’ in Maslo w’s needs hierarchy.
Page 10 of 65
14. Identification can also be a passive process of being identified. In my opinion, it is this passive identification, which is
studied in minimal group experiments, where persons are randomly assigned to groups to study group behavior. This ad
hoc team identification can be termed ‘situated identification’. It is “created by situational cues signalling shared interests
and maintains as long as the cues persist” (Rousseau, 1998: 218). Organizational identification, on the other hand, is in
my opinion ‘deep structure identification’, i.e. “the cognitive schema formed in work settings across role, over time, and
across situations that leads to congruence between self-at-work and one’s broader self concept” (Rousseau, 199 8: 218).
The outcomes and conclusions from group experiments can, in my opinion, not be generalized to organizational identifi-
cation because they depend on impermeable group boundaries (see Ellemers, 1991; Ellemers et al., 1988). But bounda-
ries of organizations are more or less permeable. It will generally be easier for organizational members to leave an or-
ganization in case it does not satisfy their needs than for group members, where group is defined in terms of a social cate-
gory. Furthermore, organizational identification has an intermediate position between “small, face -to-face groups” and
“large -scale social category memberships” described by Turner (1982: 22). Therefore, one would expect that components
of both the ‘Social Cohesion model’ and the ‘Social Identification model’ (i.e. affective and cognitive components) are
important for organizational identification.
Recent research supports the view that organizational identification is not a unidimensional but a multidimensional con-
struct (Brown et al., 1986: 284; Hinkle et al., 1989: 306; Scott, 1997: 494-497; Smidts et al., 1999: 13). Two components
appear typically in the reviewed literature: ingroup similarity and distinctiveness towards outgroups (see Brown, 1969;
Lee, 1971; Rotondi, 1975; O’Reilly and Chatman, 1986; Dutton et al., 1994). If we reconsider the different definitions,
and especially the motivation to identify, we can think of an organizational identification model where the dimensions are
based on different needs and divided into an internal and an external oriented identification (see Figure 1.1).
Figure 1.1 Theoretical Components of Organizational Identification
individualistic, collectivist,
internal component internal component
(achievement) (affiliation)
OID
relational,
external component
(awareness)
This model views organizational identification as a dynamic, multidimensional concept, based on Rotondi’s de finition of
organizational identification as the “process whereby an individual identifies with the organization” (1975: 95). Hinkle et
al. label the three components mentioned in Tajfel’s definitions of social identity (1978a: 28 -29; 1978b: 63) “knowledge
or cognitive aspects of group membership”, “emotional -affective aspects of belonging to the group”, and “evaluative
aspects of group membership” (1989: 306 -307). This tripartite structure reflects different underlying needs. The need for
harmony and affiliation is fulfilled by group cohesion and congruity, the need for achievement and self-actualization is
Page 11 of 65
15. fulfilled by a rational-evaluative component, and the need for prestige and self-esteem enhancement is fulfilled by a com-
ponent which relates identification with the outside.
Ashforth and Mael’s (1989) mono -dimensional model of organizational identification would thus represent the external
aspect, while “Acknowledgment & Perceived Opportunities” and “Likemindedness/Congruence” of the ROIT model
(Smidts et al., 1999: 13) represent the internal aspects of identification.3 The two internal components are in line with
Hall and Schneider’s notation of different dynamics of identification in two types of career. Some people choose a career
that satisfies needs for security and affiliation, while others go for self-fulfillment and personal growth opportunities (Hall
and Schneider, 1972: 349).
This three components model has so far not been tested in a field study. As the relative importance of these components
may, depending on organizational and societal characteristics, vary considerably, this model might prove useful to com-
pare company across sectors and countries.
3
“Pride & Involvement”, the third component extracted by Smidts et al. (1999, 13) is in my opinion a consequence of the identi-
fication process and should thus be separated from organizational identification.
Page 12 of 65
16. 1.1.4 Related Concepts
“Far from reducing national differences, organizational culture maintains and enhances them.”
Nancy Adler International Dimensions of Organizational Behavior
After the analysis of organizational identification in the previous section, it will come as no surprise that boundaries to
related concepts are sometimes blurry. In this section, I review four related concepts on the organizational level to make
the distinctions between them clearer before moving on to models of organizational identification.
1.1.4.1 Corporate Identity
Albert and Whetten consider the “criteria of central character, distinctiveness, and temporal continuity as each necessary,
and as a set sufficient” to define identity (1985: 265). Identity “serves the function of identification and is in part acquired
by identification (Albert and Whetten, 1985: 267). It is thus important for the company to distinguish itself “on the basis
of something important and essential” (Albert and Whetten, 1985: 266). This organizational identity, however, mu st not
be confused with corporate identity.
Corporate identity is a measure on the part of the company and describes “the way in which a company presents itself to
its target groups” (Van Riel, 1995: 28). The objectives of corporate identity are (1) to enh ance the identification of em-
ployees with their company and (2) to enhance the identification of external stakeholders (Keller, 1987: 50).
Organizational identification depends on corporate identity as the self-presentation of a company’s identity, but al so on
identification needs of employees, and actual behavior of the company (Gutjahr and Keller, 1995: 91).
1.1.4.2 Climate
Reichers and Schneider define climate as “shared perceptions of organizational policies, practices, and pr ocedures, both
formal and informal” (1990: 22). In the same organization multiple climates can coexist (Reichers and Schneider, 1990:
Table 1.1), because collective climates are based upon the “perception of individuals who share common multi -
dimensional descriptions of their work environment” (Joyce and Slocum, 1990: 133). These perceptions can vary as a
function of the position and are related to sub-unit performance and job satisfaction (Reichers and Schneider, 1990: Table
1.1). According to Grunig, climate consists of “employee perceptions of what the organization is like” while “satisfaction
consists of the evaluations or affective responses of employees of the organization” (1992: 551). Bartels, on the other
hand, defines climate as the degree of job satisfaction (1995: 131). In Bartels view, a good climate increases the willing-
ness to perform, which in turn increases the degree of identification with the goals of the company (1995: 131).
For Poole, the whole concept of climate is “bound up with commu nication” (1985: 81). This leads us to a special kind of
climate, namely the communication climate, which is defined in terms of trust, openness, credibility, accuracy, and fre-
quent interaction (Grunig, 1992: 540). These dimensions are positively correlated to satisfaction (Poole, 1985: 95; Fal-
Page 13 of 65
17. cione et al., 1987: 201-202). Falcione et al.’s research suggests furthermore, “that communication climate is necessary
for member satisfaction but not sufficient” (1987: 222, author’s emphasis).
1.1.4.3 Corporate Culture
Culture and climate are sometimes confused because these concepts are closely related. They can, however, be distin-
guished in that culture is “a deeper, less consciously held set of meanings”, while climate is a “ma nifestation of culture”
(Reichers and Schneider, 1990: 24). Hofstede et al. describe climate as shorter-term and culture as longer-term character-
istics of an organization and claim that climate “should be easier to change than culture” (1993: 489). Schein also empha-
sizes the longer-term characteristic of culture. He defines culture as a “learned product of group experience” that is “to be
found only where there is a definable group with a significant history” (Schein: 1989: 7).
According to Sriramesh et al. “organizational culture consists of the sum total of shared values, symbols, meanings, b e-
liefs, assumptions, and expectations that organize and integrate a group of people who work together” (1992: 591). Pe r-
ceptions of organizational practices can be described along the following six dimensions of corporate culture: (1) process
versus result orientation, (2) employee versus job orientation, (3) parochial versus professional, (4) open versus closed
systems, (5) loose versus tight control, and (6) normative versus pragmatic (Hofstede et al., 1993: 483).
1.1.4.4 Commitment
Commitment is another usually vaguely defined concept that overlaps with satisfaction and identification. Mowday et al.,
for example, define organizational commitment as the “relative strengths of an individual’s identification with and in-
volvement in a particular organization” (1979: 226, emphasis added). Reichers goes even further and defines commi t-
ment as “a process of identification with the goals of an organization’s multiple constituencies” (1985: 465, emphasis
added; see also Barge and Schlueter, 1988: 131).
Mowday et al. distinguish between attitudinal and behavioral commitment (1979: 225). Considering Salancik’s definition
where the “degree of commitment derives from the extent to which a person’s behaviors are binding” (Salancik, 1982: 4,
emphasis added), attitudinal commitment seems identical with organizational identification, at least for an organizational
identification concept that includes affective aspects. To avoid confusion, in this paper commitment will only refer to the
behavioral commitment component and is considered a consequence of identification.
Page 14 of 65
18. 1.2 Models of Organizational Identification
1.2.1 Ashforth and Mael’s Model
1.2.1.1 Description
Mael and Ashforth define organizational identification as the “perception of oneness with or belongingness to an organ i-
zation, where the individual defines him or herself in terms of the organization(s) in which he or she is a member” (1992:
104, author’s emphasis). Consequently, an individual experiences the organization’s successes and failures as his own
(Mael and Ashforth, 1992: 103).
According to Ashforth and Mael’s adaptation of social identity theory, an individual need not expend any efforts towards
the group’s goals to actually identify, “rather an individual need only perceive him - or herself as psychologically inter-
twined with the fate of the group” (1989: 21). This means that identification with a group “can arise even in the absence
of interpersonal cohesion, similarity, or interaction and yet have a powerful impact on affect and behavior” (Ashforth and
Mael, 1989: 26). This follows Turner’s (1982) cognitive redefinition of social identity and contrasts with Tajfel’s conce p-
tualization that includes affective and evaluative aspects (Tajfel, 1978b: 63).
In Ashforth and Mael’s view, the tendency to identify will depend o n the positive or negative distinctiveness of the
group’s values and practices (1989: 24). Second, it will depend on the prestige of the group because “through inter -group
comparison, social identification affects self-esteem” (Ashforth and Mael, 1989: 25 ). Furthermore, the salience of the
out-groups will enhance the degree of identification as awareness of out-groups “reinforces awareness of one’s in -group”
(Ashforth and Mael, 1989: 25). Finally, traditional group formation factors, like interpersonal interaction, similarity,
liking, proximity, shared goals or threat, and common history, may also affect the degree of identification but are in Ash-
forth and Mael’s opi nion not necessary for identification to occur (1989: 25).
Consequences of identification include “support for and commitment to” the organization as “individuals tend to choose
activities congruent with salient aspects of their identities” (Ashforth and Mael, 1989: 25). In addition, identification may
result in “loyalty to, and pride in, the gr oup and its activities” and may also engender “internalization of, and adherence
to, group values and norms and homogeneity in attitudes and behavior” (Ashforth and Mael, 1989: 26). Furthermore, it
affects “outcomes conventionally associated with group for mation, including intragroup cohesion, cooperation, and altru-
ism, and positive evaluations of the group” (Ashforth and Mael, 1989: 26). Finally, “it is likely that social identification
will reinforce the very antecedents of identification, including the distinctiveness of the group’s values and practices,
group prestige, salience of and competition with out-groups, and the traditional causes of group formation” (Ashforth and
Mael, 1989: 26).
Ashforth and Mael’s model of organizational identification with t heir predicted antecedents and consequences (but not
the feedback loop) are shown in Figure 1.2 on the next page.
Page 15 of 65
19. Figure 1.2 Ashforth and Mael’s Model of Organizational Identification
Salience of Traditional Group
Distinctiveness Prestige Out-groups Formation Factors
Organizational Identification
Support, Commitment,
and Internalization Loyalty and Pride Group Formation Outcomes
1.2.1.2 Discussion
Ashforth and Mael define social identification as “the perception of oneness with or belongingness to some human aggr e-
gate” (1989: 21) and argue that “organizational identification is a specific form of social identification” (1989: 22). Ho w-
ever, in my opinion Turner’s cognitive redefinition of social identity is not valid on the organizational level. Social iden-
tity is usually a result of passive categorization, i.e. of being categorized. Although an individual may emphasize some
categorical membership more than others, this membership cannot easily be changed. Organizational identification, on
the other hand, is the result of an active choice with more permeable boundaries. The difference becomes clearer if we
compare a social identity, like nationality or being a soldier, with organizational membership, like being an employee of
XYZ or being a member of a political party.
According to Ashforth and Mael, affects and behaviors “serve as antecedents or consequences of the cognition” (1989:
35). Although this distinction between affects and organizational identification as a cognitive construct is theoretically
possible, it is not measurable in practice because of the dynamic nature of organizational identification. Furthermore,
research indicates that “affective aspects of identification may warrant more importance” (Hinkle et al., 1989: 314-315)
Page 16 of 65
20. 1.2.2 Rotterdam Organizational Identification Test (ROIT)
The authors of the Rotterdam Organizational Identification Test (ROIT) define organizational identification with Mael
and Ashforth’s words of “perceived oneness with an organization” (1992: 103). Unlike Ashforth and Mael (1989), whose
theory is rooted in the cognitive redefined social identity, Smidts et al. include affective states (Smidts et al., 1999: 6).
This means that Smidts et al. follow Tajfel’s definition of social identity as “that part of an individual’s self -concept
which derives from his knowledge of his membership of a social group (or groups) together with the value and emotional
significance attached to that membership” (Tajfel, 1978b: 63, italics omitted).
Like Ashforth and Mael (1989), Smidts et al. believe that strongly identifying employees are more likely to show a sup-
portive attitude towards their organization (Smidts et al., 1999: 3). They view organizational identification as an underly-
ing concept of commitment, and thus expect that strong identification will result in better performance of the organization
(Smidts et al., 1999: 3).
Smidts et al. emphasize the role of communication on organizational identification. According to the results of their re-
search, identification appears “to be influenced mainly by the communication climate” (Van Riel, 1995: 63). Because this
study uses the ROIT questionnaire, a discussion of its content and sub-scales will be reported in the following section. A
discussion of the preliminary model of the ROIT can be found in Van Riel’s Principles of Corporate Communication
(1995: 60-65) and is reproduced as Figure 1.3 below.
Figure 1.3 Preliminary Model ROIT Scale (Van Riel, 1995: 61)
Personal Organizational
Characteristics Characteristics
Salience of Traditional Group
Distinctiveness Prestige Out-groups Formation Factors
Organizational Identification
Employee Communication
Page 17 of 65
21. 2 Analysis
2.1 Sample and Procedure
Ninety-one items of the original Dutch Rotterdam Organizational Identification Test (ROIT) questionnaire were selected
for this research and translated into English. The English version was supplemented with additional questions concerning
personal background (see Appendix 1) and translated into Chinese by the author. This translation was checked by two
native speakers, one of them who works as a human resource manager.
The final Chinese version (see Appendix 2) was distributed to 337 Chinese employees working for four different enter-
prises of a German multinational company in Beijing, Shanghai and Wuxi in March 1999 via the human resource de-
partments. Prior to distribution, corporate communications informed human resource personnel of the upcoming survey
with a telephone call. The questionnaires, carrying a short instruction on the cover, were distributed to respondents in the
afternoon and supposed to be returned the next morning to make individual completion more likely (see Appendix 2,
cover page).
Because of the small size of some of the companies all employees present on the day of the survey were included to guar-
antee anonymity. Participation in this survey was voluntary; participants were assured of individual anonymity. The com-
pleted questionnaires were returned to the researcher either individually or in provided envelopes via the human resource
department. In total, 291 employees participated in this survey, for a response rate of 86 percent.
Page 18 of 65
22. 2.2 Instrument and Variables
The adapted ROIT questionnaire attempts to measure organizational identification and its hypothesized antecedents. All
items were answered on a 5-point, Likert-type scale, with 1 indicating strongly disagree and 5 meaning strongly agree,
unless otherwise specified. A questionnaire consisted of ten parts.
The first part referred to organizational identification, which was measured for both department-level and company-level
identification, each using a ten-item scale. It consisted of five items taken from the original ROIT questionnaire and the
first five items of Mael and Ashforth’s identification instrument (1992: 122). These statements were formulated parallel
for department-level and company-level identification. The respondents were asked to rate the statements, such as “Pe o-
ple in my company really back me” on the disagree/agree scale.
The second part consisted of a five-item scale referring to external prestige as perceived by the employees. Slight modifi-
cations were made to the original scale (ROIT) by deleting an item related to financial soundness of the company. As
foreign-invested enterprises, all companies surveyed compare favorably to most state-owned enterprises. The respondents
were asked to indicate their perception of the reputation of their company with regard to different stakeholders, such as
the general public or customers.
The third part of the questionnaire referred to satisfaction with the job and the company, and included statements assess-
ing commitment and intent to leave. The original ROIT instrument for this scale was reduced from thirteen to eight items
with the last three being items originally used in Mowday et al.’s Organizational Commitment Questionnaire (OCQ)
(1979: 228, Questions 1, 2, and 9). Examples of this section are “I enjoy doing my job” and “I am willing to put in extra
effort in order to help the company be successful”.
In the fourth part, respondents were asked to assess the extent to which the company strives for the goals and values
stated in various guidelines of the German parent company. Nine items referred to goals and values like “social respons i-
bility” or “leadership in research and development”. Respondents were asked to rate their company on a 5 -point, Likert-
type scale, with 1 indicating We do next to nothing about that at XYZ and 5 We do a lot about that at XYZ. Another nine
items referred to the way these goals are pursued, for example “creativity” or “flexibility”. The scale ranged from You
will never encounter something like this at XYZ to You will definitely encounter something like this at XYZ. Furthermore,
respondents were asked to indicate to what extent they personally agreed with the goals and the way of attaining these
goals, a question found in both ROIT and OCQ (Mowday et al., 1979: 228, Question 5). These additional two items
again used the 5-point strongly disagree to strongly agree scale.
In the fifth part, 16 items referred to organizational culture. Except for the last item, which was taken from Keller’s Co r-
porate Identity questionnaire (Keller, 1987, Question 38), all items came from the ROIT questionnaire, however, reduc-
ing the number to fifteen items. The respondents were asked to indicate on the 5-point disagree/ agree scale to what ex-
tent they agreed to the statements. This part not only measured corporate culture characteristics with statements like “At
this company the job I am doing gets more attention than my person”, but also organizational (communication) climate,
e.g. “I can talk about ever ything with my boss”.
The sixth part of the questionnaire addressed received information. While the ROIT distinguishes between sufficiency
and usefulness of received information, here, only the sufficiency part, consisting of 17 statements, was used for assess-
ment. On a 5-point, Likert-type scale with 1 indicating definitely not enough and 5 meaning extensive, respondents were
Page 19 of 65
23. asked to complete statements such as “Information about customer satisfaction with our products is¼” or “Information
about the quality of my achievements is¼”.
The seventh part referred to internal communication and communication climate within the company. All eight items
were taken from the ROIT questionnaire without any changes and shortening. Respondents were asked to rate their agree-
ment/disagreement with statements like “When my colleagues tell me something, I trust that they are telling the truth” or
“My superior is open to my suggestions” on the 5 -point disagree/agree scale with 1 indicating strongly disagree and 5
meaning strongly agree.
In the eighth part respondents were asked to rank from 1 (most important) to 5 (least important) the sources where they
get their information from according to (a) perceived current importance, as well as (b) desired importance.
The ninth part referred to the communication policy of the company. Of the original eight items of the ROIT only five
were used. The respondents were asked to indicate to what extent they agreed or disagreed with statements like “Emplo y-
ees should be involved to a greater extent in the decision making process” and “Official sources of information most
often give information that is no longer up to date”.
The final section of the questionnaire requested background information about the respondents, referring to gender; age;
education (elementary school, high school, college, university); and position (middle management and above, lower man-
agement, administrative personnel and secretary, worker, or driver).
In total, 108 variables were measured in the survey. Ninety-one of the surveyed items were based on questions of the
ROIT, which in its original form consists of 249 variables.
Page 20 of 65
24. 2.3 Preliminary Analysis and Data Description
A total of 291 of 337 distributed questionnaires were returned, generating a response rate of 86 percent. The data from
these questionnaires were entered into the SPSS 8.0 for Windows statistical package for analysis.
Prior to analysis, all variables were examined through various SPSS programs for accuracy of data entry, missing values,
and fit between their distributions and the assumptions of multivariate analysis following the procedure suggested by
Tabachnik and Fidell (1996: 57-104). The variable for attachment to company (ATTACH), with missing values on al-
most 20 percent of the cases, was deleted.
Fourteen questionnaires had more than four variables with missing values and were excluded from further analysis (N =
277). Twenty cases with extremely low z-scores (z < - 3.29) were found to be univariate outliers. These outliers were
excluded and separately analyzed4. Thus the work set for further analysis consisted of 257 cases, representing 76 percent
of all distributed questionnaires.
The final sample was 57 percent male5, with almost two-third aged between 26 and 35 years. Sixty-five percent of the
respondents had university degrees; 18 percent were college graduates. Sixty-six percent had worked for their company
less than two years, probably due to the short history of these enterprises. Respondents included 22 percent in managerial
position, 36 percent administrative personnel, and 35 percent workers. The Beijing joint venture clearly dominated the
sample with 55 percent of all respondents, while the wholly foreign owned holding company contributed with only 8
percent. This is not due to the differential response quote but to the small size of the holding (Table 2.1).
Table 2.1 Descriptive Statistics
Total JV JV JV
Holding
Sample Beijing Shanghai Wuxi
Distributed questionnaires 337 26 191 48 72
Useable response 76.3% 80.8% 74.3% 58.3% 91.9%
Questionnaires in work set 100% 7.7% 56.7% 10.9% 25.7%
Male : Female 57:41 52:38 47:51 75:25 71:24
University degree 65% 76% 72% 39% 58%
Managerial level 22% 22% 26% 11% 18%
Administrative personnel 36% 57% 36% 36% 20%
Production workers 35% 0 29% 50% 55%
4
The excluded cases were not significantly different from the working sample.
5
Percentages may not add up to 100 percent due to missing values.
Page 21 of 65
25. 2.4 Data Analysis
The data set was analyzed in two ways. First, principal components factor analyses were performed for each of the sub-
scales previously described. Factors with eigenvalues exceeding 1.0 were rotated to simple structure using the varimax
procedure unless stated otherwise. In addition scree plots were used to locate transition points so that sometimes a solu-
tion with fewer factors than suggested by the Kaiser-Guttman rule6 was preferred (Nunnally and Bernstein, 1994: 482).
Factor loadings with an absolute value greater than or equal to .40 were used to define the factors. The extracted factors
were then compared with the literature to check whether the same items define the anticipated scales. In addition, the
alpha coefficient was calculated to test for internal reliability in these multi-item scales (Cramer, 1999: 393). Second, a
series of regressions was conducted to test whether the antecedents, which are described by the two model of organiza-
tional identification, can actually explain organizational identification.
Table 2.2 Research Measures
Measure Items Mean S.D. Alpha
Organizational identification
ROIT model 5 3.69 .69 .80
Ashforth and Mael’s model 5 4.17 .61 .76
Independent variables
Perceived external prestige 5 4.09 .64 .82
Job satisfaction 3 3.86 .74 .70
Goals and values 18 3.99 .64 .93
Climate 4 2.72 .70 .68
Communication climate 7 3.18 .76 .88
Information concerning job 7 3.36 .70 .83
Information concerning company 10 3.14 .83 .93
Source 3 2.53 .72 .62
Dependent variables
Commitment 4 4.30 .59 .75
6
i.e. that the eigenvalue of factor should exceed 1.0.
Page 22 of 65
26. 2.4.1 Factor Analysis
2.4.1.1 Organizational Identification
The ten questions regarding organizational identification on company level were reduced to two organizational identifica-
tion indices, by means of factor analysis. The first factor of the principal components analysis accounts for 28.7 per-cent
of the overall variance, the second for 26.6 percent. For 9 out of 10 variables the factor loadings are clearly on one com-
ponent (see Table 2.3). The first factor can be identified as Ashforth and Mael’s organizational identification concept, the
second as items taken from the ROIT measure. CPRIDE (see Appendix 1, Question 12) was expected to load on the sec-
ond factor. This item seems to be more closely related to Ashforth and Mael’s conceptualization of identification than to
the other items of the ROIT scale. This supports the view that pride is a consequence of the organizational identification
process (see footnote 3 on page 12).
For the means of this analysis, an index was calculated for Ashforth and Mael’s instrument by using the mean over the
last five items (AMID, see Appendix 3). The coefficient alpha for these five items was .76 (see Table 2.2) and therefore
lower than the reported range of .83 to .84 (Mael and Ashforth, 1992: 110). The remaining five items, which all loaded
greater than .4 on the second factor, were used to calculate a mean identification score index for the ROIT model (ROIT).
This second factor yielded a coefficient alpha of .80.
Table 2.3 Rotated Components: Organizational Identification
Component
1 2
CAGREE .665
CPRIDE .512 .430
CBACK .796
CPSGOALS .799
CRESPECT .755
CCRITIC .647
COPINION .550
CPRAISE .811
CWE .723
CSUCCESS .695
Page 23 of 65
30. 2.4.2 Regression Analysis
Standard multiple regression was used for regression analysis, although stepwise multiple regression might have been
preferable because then the order of entry of variables is solely based on statistical criteria (Tabachnik and Fidell, 1996:
150). However, the data was not sufficient to meet the criterion of a cases-to-independent-variable ratio of 40 to 1 for
stepwise regression (Tabachnik and Fidell, 1996: 133). In addition, with stepwise multiple regression an over fitting of
data (problem of ‘data mining’) might have resulted (Tabachnik and Fidell, 1996: 153).
Analysis was performed using SPSS REGRESSION and SPSS FREQUENCIES for evaluating the antecedent and con-
sequent variables in the two models. With 257 respondents and 7 independent variables, the number of cases was well
above the minimum requirement of 111 for testing individual predictors in standard multiple regression (Tabachnik and
Fidell, 1996: 132). Variables with missing values were list wise excluded from analysis.
The correlations between the regressors and the two dependent variables, that is, the two organizational identification
measures ROIT (based on the Rotterdam Organizational Identification Test) and AMID (based on Ashforth and Mael’s
model), are shown in Table 2.5. Note that communication climate, information about the job, and information about the
company are highly correlated with each other.
The only consequence of organizational identification that could be tested was commitment (COMMIT). This was done
using standard linear regression for each of the models.
Table 2.5 Correlations
1 2 3 4 5 6 7 8
1. ROIT
2. AMID .54
3. PEP (perceived external prestige) .54 .47
4. SAT (job satisfaction) .59 .40 .39
5. CLIMATE (climate) .34 .12* .18 .17
6. GV (goals and values) .57 .43 .57 .46 .33
7. COMMCLIM (communication climate) .57 .35 .41 .50 .31 .61
8. INFOJOB (information about the job) .54 .32 .38 .46 .25 .57 .65
9. INFOCOMP (information about the com-
.48 .35 .42 .45 .23 .62 .68 .72
pany)
N = 257
All correlations: p .001, except for (*) p = .057
Page 27 of 65
32. Table 2.7 Regression of Consequence Variables on AMID
Regression of (AMID) F value Significance R2
COMMIT 0.486 F1, 198 = 61.27 .000 .24
CPRIDE 0.497 F1, 252 = 82.68 .000 .25
The linear regression models, which regressed commitment and pride in the company separately on AMID were both
highly significant (p .001; see Table 2.7). The regression of AMID on COMMIT had an explanatory power of 24 per-
cent. Thus, organizational identification as defined by Ashforth and Mael explained 24 percent of the variance in com-
mitment.
Page 29 of 65
33. 2.4.2.2 ROIT Model
Antecedents
Preliminary screening through residuals showed normal distribution of residuals. No multivariate outliers were found that
exceeded the critical value of Mahalanobis distance. Table 2.8 displays the standardized regression coefficient ( ), the
significance of the regressors, as well as significance levels for two random sub-samples including approximately two-
third of the cases, squared correlation (r2), squared semi partial correlations (sr2), R2 and adjusted R2. R for regression
was significantly different from zero, F5, 249 = 44.00, p .001.
Table 2.8 Coefficients ROIT Model
Regressors β Sig. Sig. Sub 1 Sig. Sub 2 r2 sr2
PEP .243 .000 .000 .000 .291 .038
SAT .299 .000 .000 .000 .343 .061
CLIMATE .138 .003 .022 .004 .114 .017
GV .105 .105 .271 .305 .323 .005
COMMCLIM .155 .018 .047 .122 .319 .010
INFOJOB .176 .008 .006 .069 .291 .013
INFOCOMP -.086 .215 .248 .647 .228 .003
2
R = .55 (unique variability = .15; shared variability = .40)
Adjusted R2 = .54
R = .74 (p .001)
The results show that perceived external prestige and job satisfaction (p .001), climate and information about the job (p
.01), and communication climate (p .05) contributed significantly to prediction of organizational identification. In
both random sub samples, external prestige and job satisfaction remained highly significant and climate significant, while
communication climate and information about the job were no longer significant.
Judging from the unique contribution of the regressors, job satisfaction was the most important regressor followed by
perceived external prestige and climate. These three independent variables contributed 11.6 percent uniquely to total
variance in organizational identification. In combination with the other regressors 54 percent of the variability (adjusted)
in organizational identification were explained by the ROIT model.
Because of multicollinearity in the model, I tested on mediating effects of communication climate in respect to the effects
of information about the job and about the company and climate, following Baron and Kenny’s (1986) testing method.
The results showed that information about the company was clearly mediated by communication climate. Without com-
munication climate information about the company was significant at the 5-percent level, including communication cli-
mate it had no longer any significant effect on organizational identification. Climate and information about the job were
only partially mediated, so that they were still significant in a regression model including communication climate.
Thus adding INFOJOB and CLIMATE to a regression model that includes PEP, SAT and COMMCLIM results in a
significant R2 change (see Table 2.9).
Page 30 of 65
34. Table 2.9 Restricted and Unrestricted Regression Model for ROIT
Regressors Sig. r2 sr2 Adjusted R2
Restricted Model .50
PEP .296 .000 .291 .069
SAT .328 .000 .343 .076
COMMCLIM .281 .000 .319 .055
Unrestricted Model .54
PEP .269 .000 .291 .056
SAT .304 .000 .343 .063
COMMCLIM .156 .010 .319 .012
INFOJOB .157 .007 .291 .013
CLIMATE .152 .001 .114 .021
R2restricted = .51 (unique variability = .20; shared variability = .31)
R2unrestricted = .55 (unique variability = .17; shared variability = .38)
Consequences
Although not explicitly stated in Figure 1.3, the ROIT model assumes, too, that organizational identification will result in
higher commitment (Smidts et al., 1999: 3). No multivariate outliers were found that exceeded the critical value of Maha-
lanobis distance. The regression of COMMIT on ROIT model was highly significant. More than a quarter of the vari-
ance in COMMIT was explained by ROIT (see Table 2.10 below).
Table 2.10 Regression of COMMIT on ROIT
Regression of (ROIT) F value Significance R2
COMMIT .526 F1, 198 = 75.84 .000 .28
Page 31 of 65
35. 3 Discussion
3.1 Limitations
Before starting to discuss the results obtained in respect to the cultural relevance of organizational identification using
the models of Ashforth and Mael and Smidts et al., I first want to draw some attention to the limitations of the current
study, which are important in generalizing the results.
3.1.1 Sample
The sample consisted of Chinese employees in foreign-invested enterprises. These employees have characteristics that
distinguish them from the rest of the national workforce.
First, two-thirds of the employees in the current sample were between 20 and 35 years old, while in another survey 62
percent of the employees of a state-owned enterprise were more than 40 years old (Li, 1997: 197).
Second, foreign-invested enterprises, which employ 17.2 percent of the working population (National Bureau of Statis-
tics, 1999: 148), have higher qualified employees: While 83 percent of the respondents in this sample had a college or
university education, only 3.5 percent of China’s total employed population have a college or higher level education (in
Beijing 19.4 percent; National Bureau of Statistics, 1999: 170).9
Third, people applying for jobs at foreign-invested enterprises have a different perception of work. A survey of recent
graduates that applied at a JV in Guangzhou, found that they have a more “individualistic orientation toward their work”
(Westwood and Leung, 1996: 389). Actually, an important reason for their application was “to avoid some of the neg a-
tive features they saw as prevalent in domestic enterprises” (Westwood and Leung, 1996: 385). Their expectations for
working in a JV were connected with having more opportunities for learning and self development and with being occu-
pied in more interesting and challenging work (Westwood and Leung, 1996: 397, Table 1). In the opinion of these gradu-
ates JVs would be “more characterized by trust and respect of the individual and would better engender a sense of be-
longing and pride in the company” (Westwood and Leung, 1996: 396, emphasis added).
To summarize, employees of foreign-invested enterprises differ from employees of state-owned enterprises in that they
are generally young and highly educated and indicate appreciation of intrinsic work factors. The current sample is thus
not an adequate representation of the population. No conclusions about cultural differences can be drawn (Matsumoto,
1994a: 10). The results of the current study apply only to employees of foreign-invested companies.
9
For an example of differences in the educational composition in a state-owned enterprise, a joint venture, and a wholly foreign-
owned enterprise see Li, 1997: 197.
Page 32 of 65
36. 3.1.2 Construct
For a proper test of the organizational identification a longitudinal approach is far more adequate. This will enable the
researcher to distinguish empirically between antecedents and consequences of organizational identification. A further
limitation in this sample was that the construct was not completely covered, i.e. not all relevant domains were sampled.
For the Rotterdam Organizational Identification Test corporate culture was omitted (see Schein’s definition of culture as
a “group with a significant history”, 1989: 7, emphasis added), as well as approximately half of the questions concerning
communication. Ashforth and Mael’s model has prestige of the group, negative and posit ive distinctiveness, and salience
of out-groups as antecedents. However, only the first of these and satisfaction were measured.
3.1.3 Method
Two sources of potential bias are the lack of comparability of samples discussed earlier, and a differential response style.
The latter refers “to a cultural tendency to respond a certain way on test or response scales that is reflective more of cu l-
tural tendency than of the meaning of the actual scale” (Matsumoto, 1994a: 12). There might be, for example, a cultural
tendency for on average higher scoring or extremity scoring. While this does not affect the test of relationship between
the variables, it influences the level of the mean score on a particular item. Therefore, the score on a single item might be
difficult to compare across cultures. Results of an unpublished survey of subsidiaries of a German company in various
countries indicate, for example, that Chinese responses were on average 0.5 points higher on a 5-point, Likert-type scale.
However, this bias does not influence the results of factor analysis and multiple regression in this survey.
Page 33 of 65