RESEARCH AND PRACTICE
IN HUMAN RESOURCE MANAGEMENT

Highlight, copy & paste to cite:

Han, N. C., Ko, J., Price, J. L. & Muler, C. W., (1995). Organisational Commitment in South Korea, Research and Practice in Human Resource Management, 3(1), 39-68.

Organisational Commitment in South Korea

Nae Chang Han, Jong-Wook Ko, James L. Price & Charles W. Muler

Abstract

A causal model of organisational commitment developed on data collected from Western Societies is estimated for a hospital in South Korea. The estimation is conducted with data collected be questionnaires. Two endogenous variables, job satisfaction and organisational commitment, are assessed with widely used organisational measures; endogenous variables are assessed with single-item measures; exogenous variables are assessed with single-item measures selected from well established indices. Data are analysed with LISREL maximum likelihood menthod. Despite the great differences between Western societies and South Korea, the model works quite well in the latter. Ten variables are important in explaining organisational commitment: job satisfaction, opportunity, met expectations, work involvement, positive affectivity, negative affectivity, autonomy, role conflict, supervisory support, and distributive justice. These ten variables agree with the proposed model. The model explains eighty-five percent of variance in organisational commitment.

Introduction

The research reported in this paper estimates a causal model of organizational commitment for a hospital in South Korea. The site of this hospital is important. Organizational scholars seek to develop models which are applicable to all societies, yet these scholars mostly estimate the models they develop within a narrow range of societies. Almost all research regarding organizational commitment has been conducted in Western societies, especially in the United States and Great Britain. Nonwestern sites need to be studied, and the research reported in this paper examines one such site, South Korea. Generalizability is thus the core concern of this research.

Meyer and Allen (1987, 1990) distinguish affective, continuance, and normative dimensions of commitment. Affective commitment refers to an emotional attachment to the organization. For this dimension, the strongly committed member identifies with, is involved in, and enjoys membership in the organization. Affective commitment is sometimes termed “attitudinal commitment” and “loyalty.” It is the most common conceptualization of commitment in the literature and is best represented by the work of Mowday and his colleagues (Mowday, Porter and Steers, 1982). Continuance commitment is an individuals assessments of the utility of remaining with the organization. This dimension (Becker, 1960) involves assessing the potential loss of valued investments associated with leaving the organization. Finally, normative commitment is defined in terms of duty. The work of Wiener (1982) suggests that individuals will behave in accordance with organizational goals because “they believe it is the right and moral thing to do.”

Affective commitment is this study’s focus. Most studies of organizational commitment have focused on affective commitment and used Mowday and his colleagues’ measure in estimating their models (Porter, Steers, Mowday and Boulian, 1974). It is easiest to build a model of organizational commitment if one focuses on the major data base in the field; this is what this paper does. Commitment henceforth means affective commitment.

The theoretical importance of commitment stems mostly from its negative impact on absenteeism and turnover (Mowday, Porter and Steers, 1982; Porter, Crampton and Smith, 1976; Steers, 1977). There is also literature which argues that increased commitment produces better job performance (Mowday, Porter and Dubin, 1974; Porter, Crampton and Smith, 1976). Absenteeism, turnover, and job performance are important because of their link with effectiveness, a traditional concern in the study of organizations.

The Causal Model

Overview of the Model

The model to be estimated is grounded in expectancy theory. Basic to this theory is the idea that employees enter work organizations with expectations and values, and if these expectations and values are met, the employees will likely be satisfied with their jobs and be committed to the organization. Expectations are beliefs about what will characterize the work organization, whereas values are conceptions of preferred courses of action. Vroom (1964) was the first major scholar to apply expectancy theory to work organizations, and Porter and his colleagues (Porter and Steers, 1973; Mowday, Porter and Steers, 1982) are probably the major exponents of this theoretical perspective. As will soon be indicated, job satisfaction is an intervening, endogenous variable in the model.

To apply expectancy theory, it is critical to specify what the employees expect and value. Empirical research on satisfaction and commitment indicates which work conditions and environmental features employees have expectations and values about. This paper refers to these work conditions as “structural variables” and the external features as “environmental variables.”

Employees bring more than expectations and values into the work organization; they also bring their basic personality dispositions. This paper calls these additional characteristics “individual variables.” Traditional expectancy theory, as illustrated by Vroom’s research, focuses only on expectations and values, and does not examine these individual variables.

Table 1
Definitions of Variables in the Causal Model
Variable Definition
Job Satisfaction an affective orientation toward the overall job situation
Opportunity availibility of alternative jobs in the environment
Kinship Involvementa the existence of obligations to, and support from, relatives residing in the community
Met Expectations degree to which employees preconceived ideas about organizational life are met
Work Involvement belief in the centrality of the work role in one’s life
Positive Affectivity a dispositional tendency to experience pleasant emotional states
Negative Affectivity a dispositional tendency to experience unpleasant emotional states
Autonomy degree to which an employee exercises power in performing his/her job
Job Stress degree to which an employee’s abilities fall below a job’s demands
Role Ambiguity degree to which role expectations are unclear
Role Conflict degree to which role expectations are incompatible
Workload degree to which work role demands are excessive Social Support availability of helpful others
Coworker Support degree to which employees have close friends in their immediate work unit
Supervisory Support degree to which supervisors are helpful in job-related matters
Organizational Support degree to which an organization is responsive to an employee’s well-being
Routinization degree to which jobs are repetitive
Distributive Justice degree to which rewards and punishments are related to performance inputs
Promotional Chances degree to which vertical opportunities exist for an individual within an organization
Pay money and its equivalent received by employees for their services

a Kinship involvement is viewed as a social support variable in this study.

Assumptions which relate to specific elements of the model are described throughout the following subsection. At a more general level, the model assumes an exchange of benefits between the organization and its employees. Organizations typically exchange the rewards at their disposal — most of the structural variables in the model are rewards — in return for the contributions their employees make. Employees are satisfied with their work and are motivated to do their jobs in exchange for the rewards dispensed by the organization.

Figure 1
A Causal Model of Organizational Commitment
A Causal Model of Organizational Commitment

The different elements of the model will now be described. Basic documentation for the model is contained in the work of Han (1992) and Price and Mueller (1986b). The following discussion mostly cites studies to clarify the main elements of the model.

Elements of the Model

This subsection discusses the model’s variables, and the relationships among the variables. Table I defines the variabs and Figure 1 diagrams the relationships.

The model has two environmental variables, opportunity and kinship involvement. Opportunity is a job-market variable emphasized by economists; it captures the “external labor market” (Kalleberg and Sorensen, 1979). The emphasis on the environment is consistent with the focus on environmental determinants studied using the ecological perspective (Carroll, 1988; Hannan and Freeman, 1977). The impact of opportunity on satisfaction and commitment is based on the assumption that an employee is free to move. It is also assumed that employees will become dissatisfied and uncommitted if they know that similar employees elsewhere are getting more rewards. A comparative component — the similar employees — is thus built into the model. The importance of this comparative component is emphasized by equity theory (Adams, 1963, 1965).

The emphasis on kinship involvement comes from the work of Price and Mueller (Blegen, Mueller and Price, 1988; Price and Mueller, 1986b) who term this variable “kinship responsibility.” In the present paper, the variable is relabelled “kinship involvement” to indicate that it refers not just to kinship obligations but also to social support from the kinship system. In their research, Price, Mueller and colleagues have found that kinship involvement decreases turnover by increasing satisfaction and commitment. Based on their research, it is assumed that high kinship involvement increases satisfaction and commitment, because ft provides assistance for job-related problems. Assistance for job-related problems means that kinship involvement is also a social-support variable, albeit one that is outside the organization.

The model includes four individual variables: met expectations, work involvement, positive affectivity, and negative affectivity. Met expectations comes from the work of Porter and his colleagues (Mowday, Porter and Steers, 1982). The study reported in this paper is concerned only with whether expectations are met or not; it was not possible to collect data about employees’ expectations before they entered the organization. Figure 1 thus refers to “met expectations” rather than to “expectations.”

Work involvement is discussed in the work of McClelland and his colleagues, though with different terminology (McClelland, Atkinson, Clark and Lowell, 1953), and is elaborated by Kanungo and his colleagues (Kanungo, 1982; Misra, Kanungo, Rosentiel and Stuhler, 1985). Kanungo (1982) distinguishes between job and work involvement. The former indicates involvement with a specific job, whereas the latter refers to involvement with work in general. The study reported in this paper is concerned only with the latter. Work involvement is a property that employees bring into the organization. It is assumed that high work involvement leads to greater satisfaction and commitment, because highly motivated employees are likely to work harder and receive more rewards for their efforts.

Positive affectivity and negative affectivity have been recently emphasized as important determinants of work orientations by Watson and Clark and their colleagues (Clark and Watson, 1991; Watson and Clark, 1984; Watson, Pennebaker and Folger, 1986; Watson and Tellegen, 1985). These two affectivities are commonly referred to as “dispositional” variables. In addition to their expectations and values, employees bring their dispositions into the organization. Like the previously-mentioned environmental variables, these two dispositional variables constitute a departure from the traditional emphasis on the structural determinants of satisfaction and commitment (Kalleberg, 1977).

Some literature suggests that the two dispositional variables may contaminate the measurements of job stress and social support (Brief, Burke, Atieh, Robinson and Webster, 1988). For example, employees who are predisposed to experience pleasant emotional states may falsely underestimate job stress and overestimate social support. Therefore, it is argued that without controls for these dispositional variables, biased results may be produced. Although this possibility is not emphasized in the literature, the measures of other determinants, particularly opportunity and promotional chances, may also be contaminated by the dispositional variables. Employees who are predisposed to experience pleasant emotional states may overestimate the availability of alternative jobs in the environment and the chances to advance within an organization, whereas employees who are predisposed to experience unpleasant emotional states may underestimate these alternatives and chances. Job stress, social support, opportunity, and promotional chances could all be contaminated by the optimism/pessimism aspects of positive and negative affectivity. The possible confounding influences of the dispositional variables will be examined in this study.

The model has eleven structural variables: autonomy, role ambiguity, role conflict, workload, coworker support, supervisory support, organizational support, routinization, distributive justice, promotional chances, and pay. Role ambiguity, role conflict, and workload are dimensions of job stress, whereas coworker support, supervisory support, and organizational support are dimensions of social support.

Autonomy is the distribution of power from the perspective of an individuaPs job. Excluded from this definition is the power that individuals exercise in their immediate work unit. Material pertinent to autonomy is often found in discussions of “centralization,”“control,”“participation,” and “power.”

Job stress has three dimensions: role ambiguity, role conflict, and workload. These job-stress variables stem from the research (House, 1980, 1981) at the Survey Research Center (SRC) of the University of Michigan. Job-stress variables are often termed “stressors.”

Social support has three internal sources: coworkers, supervisors, and the organization. The social support label also comes from the SRC. Concern for social support, especially the coworker dimension, dates back to the Western Electric Research in the late 1920s and early 1930s (Roethlisberger and Dickson, 1938), studies on primary groups in the I 940s and 1 950s (Shils and Janowirz, 1948; Sbus, 1950), and research on cohesion in the 1950s (Seashore, 1954). Among the three internal social-support variables, organizational support is new. Recent research (Elsenberger, Huntington, Hutchison and Sowa, 1986; Greenberger, Goldberg, Hamill, ONeil and Payne, 1989; Miller, 1984) suggests that organizational policies or practices that support employees’ well-being favorably influence the employees’ work orientations toward the organization. As previously indicated, kinship involvement is also a social-support variable. The difference is that coworker support, supervisory support, and organizational support indicate assistance from within the organization, whereas kinship involvement represents assistance from outside the organization.

Routinization is a technology variable, because it indicates the nature of the transformation process, that is, the means whereby input into the organization is changed into output from the organization. The label of “routinization” in this study comes from the work of Perrow (1967). Material pertinent to routinization is found in the discussions of “variety,” as in the Job Diagnostic Survey (Hackman and Oldham, 1975).

Distributive justice has its roots in the work of Adams (1963, 1965) and Homans (1961). Literature on distributive justice contends that when employees perceive that rewards are distributed in proportion to the contribution to the organization, they will define the situation as fair, and that fairness produces increased satisfaction and commitment. What is assumed in the literature is the importance of comparisons in judging fairness. In other words, it is assumed that when an employee’s input-output (rewards) ratio is.proportional to that of other employees, the employee judges the rewards as fair. These other employees will usually be from within the organization.

Promotional chances has long been a critical variable to sociologists, since it fits well with their traditional concern about vertical mobility. Labor market theorists distinguish between internal and external labor markets. Promotional chances captures a key element of an “internal labor market,” whereas opportunity, an environmental variable mentioned earlier, captures the “external labor market.” Mobley and his colleagues (Mobley, Griffith, Hand and Meglino, 1979; Mobley, 1982) emphasize the importance of future rewards as a determinant of organizational behavior, and these future rewards are captured by promotional chances.

The role of pay as an important incentive for employees has long been investigated in the study of organizations. It includes both the cash income received and the monetary equivalents (commonly termed “fringe benefits”) of such income. The work of Lawier (1973) provides a balanced discussion of the importance of pay as an incentive.

The model has one intervening, endogenous variable, job satisfaction. The causal order of satisfaction and commitment is not well established (Curry, Wakefield, Price and Mueller, 1986; Vandenberg and Lance, 1992). It is also unclear whether the impacts of determinants on commitment are direct, indirect (through satisfaction), or both. Following the most widely adopted sequence (Lincoin and Kalleberg, 1990), it is posited that satisfaction precedes commitment. In addition, all the exogenous variables are hypothesized to impact commitment both directly and indirectly (through satisfaction).

The model does not include demographic variables. Price and Kim (1993) and Price (1995) argue that these variables are proxies for general determinants of commitment, and since the model seeks to include the widely-cited determinants, they exclude them. Five demographic variables — age, tenure, gender, education, and union membership — are, however, used as controls. If the model is properly specified by the variables listed in Table 1, none of these five variables should be significant. The demographic variables are, therefore, used to check the model’s completeness.

Moderating Effects

Four moderators will be examined in this research. First, the model hypothesizes that the determinants’ impact on satisfaction and commitment will be moderated by values, that is, the relationship between the determinants and satisfaction-commitment will differ depending on the valuation of the determinants. (Values are not postulated to moderate the impact of the individual variables on satisfaction and commitment.) Although Price and Mueller (1986b) have not found moderating effects for values in their research, values are included in the model and estimated because of the long tradition of research in expectancy theory.

Second and third, job stress is hypothesized to impact directly on satisfaction and commitment; and indirectly on commitment through satisfaction. However, a considerable amount of literature (Beehr, 1976; Cohen and Wills, 1985; Karasek and Theorell, 1990; Kaufman, 1983) suggests that job stress’s negative impact on satisfaction and commitment is moderated by social support and/or autonomy. When high levels of social support and/or autonomy exist, job stress is believed to have little or no impact on satisfaction.

Fourth, the last moderator to be estimated is being a primary income-earner. Pay is anticipated to impact directly on satisfaction and commitment and indirectly on commitment through satisfaction. In contrast, Muchinsky and Thttle (1979) suggest that pay’s impact on satisfaction and commitment will vary, depending on whether or not an employee is a primary income-earner. They suggest that as a household’s dependence on one earner’s pay increases, pay will be more important to the household and to the employee, and so pay’s impact on satisfaction and commitment will also increase.

In Figure 1, the only moderator shown is values; the other three moderators lack the data base that expectancy theory provides for values.

The Causal Model and South Korea

The model is estimated with data collected in South Korea, a society very different from the West where the model was developed. Four characteristics of South Korea will now be indicated and their implications for the model discussed.

First, the most important characteristic of South Korean society is that its social order is fundamentally based on Confucian doctrine, whose essential core is to maintain the immutable harmony underlying both the universe and human society (Chung, 1989). Because all events are presumed to be prearranged by fate, Confucian doctrine indicates that the individual should accept the world as given and adapt to it.

Second, South Korea is a hierarchical society where human relations are circumscribed by the overwhelming ascendancy of Confucian hierarchy norms. Three cardinal principles, called “samgang” in Korean, constitute the heart of all human relations (Chung, 1989; Kim, 1990). Samgang prescribes the subordination of the subjects to the rulers, the son to the father, and the wife to the husband. Loyalty and obedience are emphasized as virtues central to all human relations.

The third characteristic is personalism. One interacts with another as a total person, not segmentally (Chang, 1982). The tendency to emphasize close personal ties, which is called as “uiri” in Korean, makes it difficult to establish personal relations based on the narrow, basis of contract. These personalistic relationships engender a sense of mutual obligation among the people involved.

Fourth, South Korea is a collectivistic society. The place of the individual in society is negligible and more emphasis is placed on the group (Chung, 1989; Vogel, 1991). People are urged to be loyal to the groups to which they belong, and to sacrifice for their groups. As with the principles of samgang, the emphasis on the importance of the group over the individual originates in Confucian doctrine, and this phenomenon is evident in all social groups, especially in family and kinship relations. The fulfillment of one’s family obligations and loyalty to one’s family are stressed as fundamental to all social virtues.

These four characteristics of South Korean society have their social counterparts at the organizational level. An attempt will now be made to examine the implications of these characteristics for elements of the model.

The South Korean focus on harmony may increase the importance of social support and role conflict. All the social-support variables — coworkers, supervisors, the organization, and kinship — would appear to be influenced by the South Korean focus on harmony. The idea of help in social support implies cooperation, which constitutes the core meaning of harmony. Among the job-stress variables, role conflict should be more important than role ambiguity and workload, since it appears to be more at variance from harmony.

The hierarchical focus in South Korea may increase the importance of supervisory support and promotional chances. A focus on hierarchy means an upward emphasis in the organization, and supervisors are the next level up in the organization — thus their help would be critical to an employee. An upward focus would also imply the preference to be a supervisor, and the way to do this is to be promoted — thus promotional chances may be especially significant.

The focus on personalism in South Korea may increase the importance of social support, especially its organizational components of help from coworkers and supervisors. Personalism may also decrease the importance of the role ambiguity dimension of job stress. First, consider social support. ‘When social relations are intense among coworkers and between supervisors and subordinates, it is difficult for their relations to remain within the narrow confines of the employment contract. These intense social relations probably increase the importance of social support. Next, consider role ambiguity, Organizational scholars have long viewed intense social relations among employees as examples of departures from the specificity required by contemporary organizations. The specificity required by contractual relations is the opposite of the role ambiguity component of job stress. If an organization does not stress contractual relations, then role obligations need not be clearly specified.

The collectivistic focus of South Korea will possibly increase the importance of kinship involvement, while simultaneously reducing the importance of autonomy and distributive justice. (a) An employee who is deeply concerned with his/her kin is an individual acting in conformity with collectivistic norms — in this case, the obligations of kinship. Thus, the individualistic norms so prized in the West are de-emphasized. Kinship involvement is a dimension of social support. (b) The autonomy so highly sought in the West refers to the individual employee’s ability to make decisions about his/her job. Where collectivistic behaviors are emphasized, it would appear that individual autonomy would be seriously constrained. (c) In the West, when rewards and punishments are related to performance — in short, when distributive justice is high — it is individual performance that is assumed; it is the individual employee who prefers that his/her performance be rewarded or punished. The seniority strongly emphasized in South Korean organizations is consistent with the collectivistic focus of the society, it is thus possible that distributive justice will not be an important variable.

To summarize: the model to be estimated is supported by data collected in Western societies. However, South Korea, where the model is to be estimated, differs from the West and the model may work differently there. An analysis of South Korea suggests that six components of the model are likely to be more important: coworker support, supervisory support, organizational support, role conflict, promotional chances, and kinship involvement. Three components of the model are likely to be less important: role ambiguity, autonomy, and distributive justice. The results in South Korea could thus be quite different from estimations of the model in the West.

Methodology

Site, Sample, and Data Collection

The site of this research is a university hospital in a middle-sized city (population of about 300,000) in South Korea. The hospital consists of seven separate units: headquarters, university hospital, psychiatric ward, dentistry, and three oriental medicine clinics. Four of the seven units are investigated in this research: headquarters, university hospital, dentistry, and one oriental medicine clinic. These four units investigated in this research comprise 85.8 percent of the hospital’s 1,188 employees and 93.7 percent of its 1,098 beds.

The sample is composed of 511 hospital employees and represents all occupational categories in the hospital (physicians, nurses, administrative/clerical workers, technicians, and manual workers). Inclusion of employees from all occupational categories increases the variance of the variables investigated. Three hundred and twenty-two of the employees are females (63.0 percent), and one hundred and sixty-two are union members (31.7 percent). Mean levels of employee age, tenure, and education are 31.0, 4.6, and 14.4 years, respectively.

Data were collectedby questionnaire. The questionnaire was carefully translated into Korean by the first author of this paper. The Korean version was pretested on fifteen employees of the hospital, and based on the pretest results, a final revision was made.

The first author of this paper visited every department of the hospital, explained the purpose of this research to the departments, distributed questionnaires, and collected the questionnaires with the assistance of three hospital employees. Respondents took approximately twenty minutes to complete the questionnaire. In an executive session, the hospital management was briefed about the survey. Questionnaires were distributed and collected between January 6 and January 18, 1992. One thousand and nineteen questionnaires were distributed and 628 questionnaires were collected, for a response rate of 61.6 percent. After 117 questionnaires with missing data are excluded, the final sample consists of 511 respondents, for a net response rate of 50.1 percent.

Measurement

Commitment and satisfaction are assessed by indices widely used in the study of organizations. A four-item index selected from Porter and his colleagues Organizational Commitment Questionnaire (Porter, Steers, Mowday and Boulian, 1974) is used to measure commitment. The four items have a reliability of .74. Satisfaction is measured by a six-item index selected from Brayfield and Rothe (1951). The six items have a reliability of .76.

All of the exogenous variables are measured by single-item indicators selected from well-established indices (Price and Mueller, 1986a). As is customary in most organizational research, these single-item indicators use subjective assessment to gauge objective situations (Price and Mueller, 1986a). Indices are, of course, superior to single indicators, because the degree of validity and reliability is often unknowable for single-item indicators (Mclver and Carmines, 1981). However, some scholars have substituted single indicators for indices, while still retaining adequate validity and reliability (Burnish, 1984; Gardner, Pierce, Dunham and Cummings, 1992; Goldberg, 1972; Russell, Weiss and Mendelsohn, 1989). As can be seen later in the paper, the results of this study provide evidence for the construct validity of the single-item indicators used in this research.

The measures of values and being a primary income-earner (variables used in the tests for moderating effects) require some discussion. The values of twelve determinants (kinship involvement and the eleven structural variables) were measured with questions that asked respondents how much importance they attach to each variable. Each question had five response categories ranging from “very important” to “not important at all.” This is the traditional way to measure values in organizational research (Lincoln and Kalleberg, 1990). Since all thirteen value measures exhibited little variation, each item was converted into a dummy variable after the five response categories of the item were grouped into two approximately equal parts. The two parts are basically “very high importance” and “high importance.” This dichotomy captures the data’s variance. These dummy variables are used to check the moderating effects of values.

Being a primary income-earner is measured using the income statistics collected by the questionnaire. If the employees’ pay constitutes over seventy-five percent of their total household income, they are treated as primary income-earners. ‘(When primary income-earners were defined with different percentages — such as sixty, seventy, eighty, and ninety — the test results for the moderating effect did not change. Seventy-five is used because it is midway between fifty and one hundred.) A single person who works is, of course, a primary income-earner. This is a dummy variable that has a value of one if an employee is a primary income-earner and otherwise has a value of zero.

Data Analysis
Table 2
Reliabilities of the Exogenous Measures
Variables (1)a (2a) (2b) (3) (4) (5) (6) (7) (8) Mean
Environmental
Opportunity .75 .83 .84 .49 .87 .91 .81 .79 .76 .78
Kinship Involvement (.90)
Individual
Met Expectations .89 .82 .86
Work Motivation .72 .86 .71 .74 .58 .72
Positive Affectivity .83 .61 .71 .67 .71
Negative Affectivity .79 .73 .76
Structural
Autonomy .86 .81 .81 .84 .81 .84 .60 .52 .71 .76
Role Ambiguity .90 .93 .93 .77 .85 .61 .62 .80
Role Conflict .84 .65 .77 .75
Workload .59 .85 .63 .72 .71 .70
Coworker Support .84 .61 .73 .86 .84 .85 .78 .49 .74 .75
Supervisory Support .94 .83 .87 .88
Routinization .82 .78 .75 .80 .81 .72 .64 .76 .64 .75
Organizational support (.78)
Distributive Justice .76 .85 .85 .95 .91 .83 .82 .90 .86
Promotional chances .83 .92 .92 .72 .69 .79 .85 .82
Pay (.90)
Demographic
Age (.90)
Tenure (.90)
Gender (.90)
Education (.90)
Union Membership (.90)
Mean .78

a The numbers on this row refer to the studies on which reliability estimates are based. These studies are listed in the references and are as follows: (1) PrIce and Mueller (1981); (2) Price and Mueller (1986b); (3) Mueller and Price (1990); (4) Agho, Mueller and Price (1993); (5) Mueller, Boyer, Price and Iverson (1994); (6) Iverson (1992); (7) Uden-Holman (1992); (8) Wallace (1992).

Two statistical techniques were used to analyze the data, LISREL and multiple regression. The causal model presented in Figure 1 was estimated with LISREL 8 (Joreskog and Sorbom, 1993). The sample covariance matrix for the observed indicators was used as input data. To correct for measurement errors, the reliabilities of the single-item measures were set to the values presented in the last column of Table 2. Most of those reliability estimates were obtained by averaging the reliabilities of the measures reported in previous studies conducted by Price and Mueller and their colleagues. The reliability of organizational support was set to .78, the average of the average reliabilities; the reliabilities of kinship involvement, pay, and the five demographic variables were set to .90. The theta-delta matrix for the x-variables was obtained by multiplying 1 minus the reliability estimate by the observed variance for each x-variable. The latent variables corresponding to the x-variables were set to the same scale as the x-variables (Bollen, 1989; Joreskog and Sorbom, 1989). Parameter estimates were obtained with the maximum likelihood method.

There is a lack of consensus on how to assess a model’s goodness-of-fit. Therefore, according to the recommendation by Bollen (1989) and Bollen and Long (1993), multiple measures of fit — such as the Normed Fit Index (NFl; Bentler and Bonnet, 1980), the Comparative Fit Index (CFI; Bentler, 1990), and the Incremental Fit Index (IFI; Bollen, 1989)—will be used to evaluate a model’s fit to the data.

Checks for linearity and multicollinearity were conducted using multiple regression. Significance tests for deviations from linearity and a graphical examination of the relationships indicate that significant or meaningful nonlinear relationships do not exist. To check multicolliriearity, the R2s for the equations regressing each independent variable on the other independent variables were compared with the R2s for the equation regressing the dependent variable on all of the independent variables (Berk, 1983). If the former is greater than the latter, there is multicollinearity. The dependent variables used in this analysis are satisfaction and commitment. None of the R2s for each independent variable are found to be greater than or equal to the R2 for the dependent variable, so there is no problem of multicollinearity among the exogenous variables. The correlation matrix in Table 3 provides additional information pertinent to multicollinearity. (Complete information on the results for these tests is available.)

Results

Moderating Effects

(1) Consider first the interactions between values and the twelve determinants (kinship involvement and the eleven structural variables). When each dependent variable is regressed on the exogenous variables and then the first set of twelve interaction terms is entered as a block, the R2 increment due to the addition of the twelve interaction terms is not significant for either satisfaction or commitment. (2) When the twelve interaction terms between the four social-support vanables and the three job-stress variables are added as a block, the R2 increment is again not significant for either satisfaction or commitment. (3) When the three interaction terms between autonomy and the three job-stress variables are entered as a block, the R2 increment is significant for satisfaction, but not for commitment. Individual t-tests for the three interaction terms show that only one interaction term, that between autonomy and workload, has a significant impact on satisfaction. ‘Mien the relationship between workload and satisfaction is examined for high and low levels of autonomy, the pattern is opposite to the one predicted in the literature. (4) Finally, when the interaction term between being a primary income- earner and pay is added, the R2 increment is significant for satisfaction, but not for commitment. When the relationship between pay and satisfaction is examined for primary and non-primary income-earners, the pattern is again opposite to the one predicted in the literature.

Table 3
Zero-order Correlations among the Exogenous Variables (n=511)a
Variables (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18) (19) (20) (21)
Variables (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18) (19) (20) (21)
Opportunity (1) 1.000
Kinship Involvement (2) -.031 1.000
Met Expectations (3) .16l**.0461.000
Work Motivation (4) .15O**.026.0271.000
Positive Affectivity (5) -.028.105*.155.478***1.000
Negative Affectivity (6) .023.107*-.312***.052-.166**1.000
Autonomy (7) -.005.024.258***.105*.286***..143**1.000
Role Ambiguity (8) .016-.018-.078.196***331***.092.276***1.000
Role Conflict (9) -.054.196***-.170**.252*177**.274***.070-.1611.000
Workload (10) .155**-.040.109*.100-.021.205***.041.13l*.299***1.000
Coworker Support (11) .054-.072.112*.230***.202***-.041.132*-.051-.072-.0711.000
Supervisory Support (12) -.090*-.005.262***.128*.320***-.128.352***110*.015-.027.278***1.000
Organizational Support (13) .041.022.278***.056-.033-.113*.170**-.025-.219***-.128*.114*.117*1.000
Routinization (14) .059-.108*-.115*.154**-.181**.071-.223***.248***.033.135*-.055-.245***.0141.000
Distributive Justice (15) .134**-.017-.382***.075.102*-.237***.270***.013-.070.108*.082.238*.141**-.0381.000
Promotional Chances (16) .104*-.055.314***.018.079-.232***.368*-.153**-.143**.044.100*.250***.230***.127*.233** 1.000
Pay (17) -.032.109*.066.006-.004-.098*-.003-.020.082-.006.021-.039.032.071.027 .173*** 1.000
Age (18) .017.461***.091*.070.140**.201***.128**-.157**-.120*.054.004.005.027-.285***.065 .110* .022 1.000
Tenure (19) .129**.221***.084*.047.114*.181***.128**-.067-.002-.007-.050-.043-.060.104*.057 .049 .255*** .604*** 1.000
Genderb (20) .212***.430***.098*-.071.017.247***-.025-.016.282***-.038.058-.057.094*-.156.035 .253*** .032 .601*** .185*** 1.000
Education (21) .164**-.052.119**-.065.001.131**.241***-.069..220***.056.079.016.196***-.033.134** .571*** .096* .067 -.033 .423* 1.000
Union Membershipc (22) -.183*** -.031 -.136** .031 .020 .037 -.186*** .119** .122* -.105* -.029 .103* -.210*** .179*** -.087* .376*** .140** .192*** .255*** .340*** -.565***

*P<.05, **P<.01, ***P<.001 (one-tailed)
a These are corrected for unreliability because they are LISREL estimates.
b 1= male, 0 = female.
c 1 = union member, 0 = non-union member.

In sum, no substantial evidence is found for the moderating effects of values, social support, autonomy, or being a primary income-earner. (To conserve space, the results for these tests for moderating effects are not presented in the paper.)

Estimates of the Causal Model

Results for the estimation of the causal model are shown in Model 1, Table 4. Consider first the results for the goodness-of-fit of Model 1. The results show that CFI and NFl are above .90, whereas NFl is just below .90. Therefore, the fit of Model 1 to the data is acceptable. Consider next the effects of the exogenous variables on satisfaction (the third column in Table 4). The results show that eight variables have significant effects on satisfaction. Met expectations, work involvement, positive affectivity, autonomy, and supervisory support have positive effects on satisfaction, whereas negative affectivity, role ambiguity, and role conflict have negative effects. These significant effects are all in the direction predicted by the model presented in Figure 1. Among the demographic variables, only age is significant and it increases satisfaction, as anticipated. The explained variance for satisfaction is sixty-three percent.

When the results for commitment are considered (the fourth column in Table 4), five variables have significant effects on commitment. Satisfaction, met expectations, autonomy, and distributive justice have positive effects on commitment, whereas opportunity has a negative effect, All of these results are consistent with the model. Among the five demographic variables, two —gender and education — are significant. Being a male increases commitment, whereas higher education decreases it. The results for gender are not anticipated; however, the findings for education are anticipated. The explained variance for commitment is eighty-five percent.

Affectivity Variables

To investigate the confounding effects of the two affectivity variables previously discussed, the LISREL analysis was performed again with these two variables removed. The results for the removal of the two affectivity variables are shown in Model 2, Table 4. The analysis will focus on the nine variables — the three job-stress variables, the four social-support variables, opportunity, and promotional chances —which the affectivity variables may contaminate.

Table 4
Zero-order Correlations and Standardized LISREL Estimates for the Model (n = 511)
Determinants & Demographic Variables Zero-order Correlations Model 1 Model 2
Satisfaction Commitment Satisfaction Commitment Satisfaction Commitment
R-square .631 .853 .584 .853
NFIc .880 .879
CFId .917 .916
IFIe .923 .921*
Intervening Variable
Job Satisfaction .763*** .494*** .507*
Environmental Variables
Opportunity -.131* -.376*** -.064 -.232*** -.036 -.229***
Kinship Involvement .111* .117* -.041 -.004 -.023 -.004
Individual Variables
Met Expectations .491*** .672* .268*** .295*** .315*** .297***
Work Involvement .300*** .261*** .114* .044 .201*** .046
Positive Affectivity .486*** -.373*** .226*** .013
Negative Affectivity -.355*** -.338*** -.121** -.029
Structural Variables
Autonomy .466*** .436*** .162** .146** .192*** .147
Role Ambiguity -.376*** -.223*** -.141** .011 -.199*** .008
Role Conflict .102* -.118* -.123* -.073 -.125* -.076
Workload -.001 -.083 .043 .053 -.002 .047
Coworker Support .211*** .171** .040 -.011 .041 -.013
Supervisory Support .366*** .436*** .085* .073 .142** .076*
Organizational Support .208*** .249*** .042 .056 .010 .054
Routinization -.323*** -.265*** -.058 .001 -.057 .002
Distributive Justice .266*** .437*** -.008 .134*** .002 .137***
Promotional Chances .316*** .190*** -.026 -.020 -.035 -.018
Pay .072 .016 .060 -.058 .068 -.058
Demographic Variables
Age .368*** .205*** .252** -.068 .268** -.070
Tenure .256*** .130** .024 -.054 .032 -.052
Gendera .163** .031 -.098 .125* -.080 .132*
Education .174*** -.070 .060 -.282*** .084 .282***
Union Membershipb -.2O4*** -.052 -.064 .002 -.032 .006

*P<.05, **P<.01, ***P<.001 (one-tailed).
a 1 = male, 0 = female.
b 1 = union member, 0 = non-union member.
c Normed Fit Index = (f0 - f1)/ f0, where 10 is the value of a discrepancy (unction for some baseline model and (1 is the value of the discrepancy function for some model of interest.
d Comparative Fit Index = 1- max ((T1 - df1), 0)/ max ((T0 - df0), (T1 - df1), 0), where T0 is the value of the test statistic for the baseline model, T1 is the value of the test Statistic for the model of interest, and df0 and df1 are the degreees of freedom associated with f0 and f1 defined as above respectively.
e Incremental Fit Index = (f0 - f1)/(f0 - df1/(N-1)). where f0, f1, and df1 are defined as above.

First, consider the situation for satisfaction. When the two affectivity variables are controlled (the third column in Table 4), eight variables are significant. Two of these eight variables are, of course, the affectivity variables. ‘When the affectivity variables are removed (the fifth column in Table 4), the same six variables are significant. In short, no changes, in terms of significance, occur when the affectivity variables are not controlled. However, it is also important to examine the ehanges in the magnitudes of the significant beta coefficients after the removal of the affectivity variables- The following increases occur for the six variables after the removal: met expectations, 17.5 percent; work involvement, 39.6 percent; autonomy, 18.5 percent; role ambiguity, 41.1 percent; role conflict, 1.6 percent; and supervisory support, 67.1 percent. The explained variance decreases about six percent (.05 7) when the affectivity variables are removed.

Next, consider the situation for commitment. ‘When the affectivity variables are controlled (the fourth column in Table 4), five variables are significant. The two affectivity variables are not among the five significant variables. When the affectivity variables are removed (the sixth column in Table 4), six variables are significant. Five of the six variables were significant when the affectivity variables were controlled; however, a new variable, supervisory support, becomes significant. In terms of magnitudes after removal, four of the significant beta coefficients increase (satisfaction, 2.6 percent; met expectations, 0.7 percent; autonomy, 0.7 percent; and distributive justice, 2.2 percent), whereas one decreases (opportunity, 1.3 percent). These changes in magnitudes are small, however. There is no change in explained variance after the removal of the affectivity variables.

The results are thus mixed for satisfaction and commitment when the affectivity variables are removed. No changes occur among the significant variables for satisfaction, whereas one change occurs for commitment. The changes in magnitudes of the beta coefficients and the explained variance are greater for satisfaction than for commitment. On balance, the removal of the affectivity variables influences satisfaction more than it influences commitment.

Decomposition of Effects

The determinants may influence commitment directly or indirectly through the intervening variable, satisfaction, and thus the effects of the determinants are decomposed into direct and indirect effects. The decomposed effects of the determinants on commitment are presented in Table 5. The investigation of these effects focuses on the theoretical variables.

Table 5
Direct, Indirect, and Total Causal Effects on Commitment
Determinants Direct Effects Indirect Effect Total Effects
Intervening Variable
Job Satisfaction .494*** .494***
Environmental Variables
Opportunity -.232*** -.031 -.263***
Kinship Involvement -.004 -.020 -.024
Psychological Variables
Met Expectations .295*** .132*** .427***
Work Involvement .044 .056* .100*
Positive Affectivity .013 .112** .125*
Negative Affectivity -.029 -.060* .089*
Structural Variables
Autonomy .146** .080** .226***
Role Ambiguity .011 .069** -.058
Role Conflict -.073 .061* .134**
Workload .053 .021 .074
Coworker Support -.011 .020 .009
Supervisory Support .073 .042 .115*
Organizational Support .056 .021 .077
Routinization .001 -.029 -.028
Distributive Justice .134*** -.004 .130**
Promotional Chances -.020 -.013 -.033
Pay -.058 .030 -.028
Demographic Variables
Age -.068 .124* .056
Tenure -.054 .012 -.042
Gender .125* -.049 .078
Education -.282 .030 .252**
Union Membership .002 -.032 -.030

*P < .05, **P < .01, ***P < .001 (one-tailed)

Of the eighteen theoretical variables, ten variables have significant total effects on commitment. Satisfaction (.494) has largest total effect on commitment, followed by met expectations (.427), opportunity (-.263), autonomy (.226), role conflict (-.134), distributive justice (.130), positive affectivity (.125), supervisory support (.115), work involvement (.100), and negative affectivity (-.089). The ten total effects are as predicted.

Most of these ten variables have greater direct than indirect effects. The proportion of the direct effect to the total effect for each of the ten variables is as follows: satisfaction, 100.0 percent; distributive justice, 97.1 percent; opportunity, 88.2 percent; met expectations, 69.1 percent; autonomy, 64.6 percent; supervisory support, 63.5 percent; role conflict, 54.5 percent; work involvement, 44.0 percent; negative affectivity, 32.6 percent; and positive affectivity, 10.4 percent. The analysis indicates that satisfaction is not a strong mediator of the exogenous variables, since most of these variables have their strongest impact directly on commitment.

Discussion

The Causal Model and Soth Korea

There are two issues to be addressed in this subsection: one, the extent to which estimation of the model in South Korea emphasized or deemphasized different components of the model in an anticipated manner; and two, the extent to which the results of the estimation agree with predictions based on the model. It is possible, for example, that specific variables anticipated to be important empirically might not be important, yet these unanticipated findings might be consistent with predictions based on the model.

It was previously anticipated that six components of the model were likely to be especially important: in South Korea: coworker support, supervisory support, organizational support, role conflict, promotional chances, and kinship involvement. It was also anticipated that three components of the model were likely to be less important in South Korea: autonomy, the role ambiguity dimension of job stress, and distributive justice.

Of the six variables anticipated to be especially important, only two, supervisory support and role conflict, were significant (see Total Effects in Table 5). Four anticipated results were thus incorrect. Of the three variables anticipated to be less important, two (autonomy and distributive justice) were significant (see Total Effects in Table 5). IWo of the anticipated results were thus incorrect. In short, based on their prior analysis of South Korean society, the investigators were not very successful in anticipating results.

However, when predictions from the Western-based model are examined, the results are totally in agreement with the model. All of the ten significant total effects (Table 5) agree with the model. In addition, the explained variance of eighty- five percent (Table 4) also is very high. It is higher than the explained variance for the following four studies: Price and Mueller (1986b), forty-two percent; Mueller and Price (1990), forty-one percent; Mueller, Boyer, Price, and Iverson (1994), seventy-six percent; and Iverson and Roy (1994), seventy percent. It should be noted that all of the results are not perfectly comparable, because some of the studies (Mueller, Boyer, Price and Iverson, 1994; Iverson and Roy, 1994) used LISREL, whereas the remainder used OLS regression. Finally, as indicated by the results for the fit-indices (Table 4), the models fit to the data also is acceptable. It therefore appears that the model works quite well in South Korea. These findings are the most important ones of this study.

These findings also constitute evidence for the construct validity of the single- item measures used in this study. The findings indicate that all the significant relationships among the variables agree with predictions based on the model.

It was previously noted that ten total effects were significant. Since the model has eighteen variables, this means that slightly over half (fifty-five percent) of them are significant. Thus, the question of the model’s adequacy arises. Those variables that are not significant may be genuine determinants of commitment, but may not be significant in the site investigated. This may be because of the large number of variables controlled and the lack of variance among the variables investigated. Multivariate analysis is common in the study of commitment, but few studies use the extensive range of controls used inthis investigation. The single hospital studied here may also lack variationfor the nonsignificant variables. An investigator is likely to obtain greater variance when different organizations are studied, especially if the organizations are quite different. All of the nonsignificant variables have empirical support in the literature and may turn out to be significant in other settings. The investigators, therefore, argue that the model estimated is a good one, even if some of its determinants are not significant in this study.

Affectivity Variables

There is evidence that the affectivity variables contaminate the measurement of some of the exogenous variables. Removal of the affectivity variables increased, sometimes quite substantially, the beta coefficients of the variables that significantly influence satisfaction. The removal of the affectivity variables reduced the explained variance for satisfaction from sixty-three percent to fifty-eight percent. Finally, when affectivity variables were taken out of the analysis of commitment, supervisory support, which was not significant before removal, became significant. The affectivity variables were thus of some importance in the explanation of commitment. Brief and his colleagues are correct in pointing to the affectivity variables’ contaminating potential.

However, the affectivity variables were not major contaminants. No change in significance occurred for the results of satisfaction before and after the removal of the affectivity variables. Supervisory support, which was contaminated, was but one of the eight variables which might have been anticipated to be contaminated. The beta coefficients for the variables that significantly influence commitment were not substantially increased when the affectivity variables were removed. Finally, the explained variance for commitment was unchanged after the removal of the affectivity variables. The affectivity variables’ contaminating potential is, therefore, somewhat exaggerated. Research on satisfaction and commitment which does not control for these variables is not seriously compromised.

However, it should be emphasized that the affectivity variables have substantive and anticipated impacts on satisfaction. Positive affectivity increased satisfaction and its beta coefficient (.226) was second in magnitude after met expectations (.268). Negative affectivity decreased satisfaction, though its beta coefficient (-.121) was but half of that of positive affectivity. As anticipated, the affectivity variables also had significant effects on commitment in terms of total effects. The effects of these two personality dispositions on satisfaction and commitment are important, because traditional research has almost exclusively emphasized structural variables as determinants. Although the contaminating role of the affectivity variables is somewhat exaggerated, their role as determinants of satisfaction and commitment is enhanced. Watson and Clark and their colleagues have made an important contribution to the study of work orientations by emphasizing the affectivity variables.

Moderating Effects

No moderating effects of values, social support/autonomy, or being a primary-income earner were found for satisfaction and commitment. Consider first the matter of values. The argunent for the moderating effects of values is based on the assumption that the determinants are differently evaluated among the employees. However, the results of this study cast some doubt on the truth of the assumption. The results show that there is little variation among the respondents for the valuations of the twelve determinants. When the respondents were asked how much importance they attached to each of the twelve variables, almost all the respondents answered that those variables were “very important” or “important.” It is difficult to find interaction effects when there is little variation for the variables examined. The valuations of the twelve determinants may vary widely among other respondents, but the respondents of this study highly evaluate the twelve variables and thus show little difference in their valuations.

As discussed earlier, a sizable literature suggests that the negative impact of job stress on satisfaction and commitment is moderated by the existence of social support and/or autonomy. This study found that the impact of job stress on satisfaction and commitment does not vary across employees who have different amounts of social support and autonomy.

Muchinsky and Tuttle (1979) suggest that the effect of pay on satisfaction and commitment will vary depending on whether or not the recipient of pay is a primary income-earner. This study found that the effect of pay on satisfaction and commitment does not differ between primary income-earners and other employees.

All of these findings show no convincing evidence for the moderating effects that appear in the literature. These findings are consistent with the empirical research by Price and Mueller (1986b). All of the findings in this study and in previous research validate an additive model. However, these findings do not necessarily reject the contingency approach (Lawrence and Lorsch, 1967; Thompson, 1967; Woodward, 1965) which is prevalent in the study of organizations. The moderating effects may exist in other settings, but they are not found in the sample investigated in this research. The findings of this study and previous research suggest that the validity of the contingency approach is somewhat overemphasized.

Direct and Indirect Effects

The model indicates that the exogenous variables have both direct and indirect effects on commitment. Satisfaction is thus viewed as a mediator for all the exogenous variables impacts on commitment. It is not clear from the literature which exogenous variables impact only directly on commitment, which impact only indirectly through satisfaction, and which impact both directly and indirectly. The model posits a plausible interpretation of direct and indirect effects.

The findings presented earlier indicated that most of the effects of the exogenous variables on commitment were primarily direct. Six exogenous variables — opportunity, met expectations, autonomy, role conflict, supervisory support, and distributive justice — had effects which were mostly direct. However, indirect effects also were important. Four exogenous variables — work involvement, positive affectivity, negative affectivity, and role ambiguity — had mostly indirect effects on commitment. Finally, four of the six variables which had mostly direct effects — met expectations, autonomy, role conflict, and supervisory support — also had substantial indirect effects. It will take more research to determine the direct and indirect effects of the exogenous variables on commitment. In the meantime, the direct and indirect effects posited by the model remain plausible.

Satisfaction is the most important variable influencing commitment. Only met expectations (.427) has total effects approaching satisfaction (.494). Since it is cross-sectional, this research can provide no evidence about the causal ordering of satisfaction and commitment. What can be noted is that the data are consistent with the causal ordering postulated by the model.

In addition to satisfaction, this study found that nine exogenous variables have significant total effects on commitment. These nine exogenous variables represent three types of determinants: environmental (opportunity), individual (met expectations, work involvement, positive affectivity, and negative affectivity), and structural (autonomy, role conflict, supervisory support, and distributive justice). Price and Mueller (1 986b) have consistently found similar results in their research on absenteeism and turnover, in which commitment is an important mediating variable. The findings of this study, and the previous research by Price and Mueller, indicate that future research on commitment should take into consideration all three types of variables as determinants of commitment. The findings, in short, emphasize that an interdisciplinary approach — which integrates the different research traditions of sociology, psychology, and economics — is required to explain commitment.

Three dimensions of social support require brief discussion: coworker support, supervisory support, and organizational support. This study found that coworker support does not have a significant total effect on commitment, whereas supervisory support does. Similar results was reported in recent research on South Korean employee commitment (Yoon, Baker and Ko, 1994), which found that vertical interpersonal attachment (supervisory support) within an immediate workgroup is much more important than horizontal interpersonal attachment (coworker support) in explaining employee commitment in South Korea. The findings of both studies seem to indicate the importance of the hierarchical characteristics in explaining commitment in South Korea.

Little research has been conducted on organizational support as a determinant of commitment. It was not significant in this study, but development of more valid and reliable measures might indicate greater importance for organizational support as a determinant of commitment. Thus this variable should not be excluded from future research on commitment, because it does have some support in the literature.

Demographic Variables

Five demographic variables were used as controls in this research. It was found that age had a significant direct effect on satisfaction, whereas gender and education had significant direct effects on commitment. Finally, the total effect for education was larger than the total effects for all variables except satisfaction and met expectations. What this means is that these demographic variables are serving as proxies for theoretical variables not included in the model.

Limitations

Three suggestions are advanced for future research on commitment. (1)Since findings of this study are based on the investigation of only one study, it is difficult to generalize. Therefore, additional research needs to be conducted on different organizations in South Korea. In addition, the model needs to be estimated in other Asian countries, such as China and Indonesia, which may be quite different from South Korea.

(2) Many organizational studies suffer from the problems associated with common method variance. One way to deal partially with these problems in the study of satisfaction and commitment is to ask the respondents’ immediate supervisor also to judge his/her satisfaction and commitment. The immediate supervisor’s judgment would thus supplement the respondent’s self report of satisfaction and commitment. An analysisould thus be done of these two measures of satisfaction and commitment to see how the results compare. Procedurally, it would be difficult to collect this information from the supervisors, but it would be one way to deal partially with the problems associated with common method variance.

(3) A true longitudinal design is required to establish firmly the causal paths among the variables. As indicated earlier, the causal order of satisfaction and commitment is not clear. It is also possible that satisfaction may promote work involvement, rather than the other way around, as hypothesized in the model. This study could not check the causal ordering because it employed a cross-sectional design. A longitudinal study would make it possible to address the problem of causal ordering.

This study has attempted to develop a comprehensive model of commitment and to extend its scope conditions to a nonwestem setting. Implementation of the above suggestions should further improve the models generalizability and utility.

References

Adams, J. S. (1963), Toward an understanding of inequality, Journal of Abnormal and Social Psychology, 67, 422-436.

Adams, J. S. (1965), Inequality in social change, in L. Berkowitz, ed: Advances in experimental and social psychology, 266-279, New ‘York: Academic Press.

Agho, A. O., Mueller, C. W. & Price, J. L. (1993), Determinants of employee job satisfaction: An empirical test of a causal model, Human Relations, 46, 1007- 1027.

Allen, N. I & Meyer, J. P. (1990), The measurement and antecedents of affective, continuance and normative commitment to the organization, Journal of Occupational Psychology, 63, 1-18.

Beehr, T. A. (1976), Perceived situational moderators of the relationship between subjective role ambiguity and role strain, Journal of Applied Psychology, 61, 35-40.

Becker, H. S. (1960), Note on the concept of commitment, American Journal of Sociology, 66, 32-42.

Bentler, P. M. (1990), Comparative fit indexes in structural models, Psychological Bulletin, 107, 238-146.

Bentler, P. M. & Bonnet, D. G. (1980), Significance tests and goodness of fit in the analysis of covariance structures, Psychological Bulletin, 88, 588-606,

Berk, R. A. (1983), Applications of general linear model to survey data, In P. H. Rossi, I. D. Wright, & A. B. Anderson Ed: Handbook of survey research, New York: Academic Press.

Blegen, M. A., Mueller, C. W. & Price, J. L. (1988), Measurement of kinship responsibility for organizational research, Journal of Applied Psychology, 73(3), 402-409.

Bollen, K. (1989), Structural equations with latent variables, New York: Wiley.

Bollen, K. & Long, J. S. (1993), Testing structural equation models, Newbury Park, CA: Sage.

Brayfield, A. H. & Rothe, H. F. (1951), An index of job satisfaction, Journal of Applied Psychology, 35, 307-311.

Brief , A. P., Burke, M. I., Atieh, J. M., Robinson, B. S. & Webster, J. (1988), Should negative affectivity remain an unmeasured variable in the study of job stress, Journal of Applied Psychology, 73(2), 193-198.

Burnish, M. (1984), You dont always get what you pay for: Measuring depression with short and simple versus long and sophisticated scales, Journal of Research in Personality, 18, 81-98.

Caplan, R. D., Cobb, S. & French, I. R. P. (1975), Relationship of cessation of smoking with job stress, personality, and social support, Journal of Applied Psychology, 60, 211-219.

Carroll, G. R. (1988), Ecological models of organizations, Cambridge, MA: Ballinger Publishing Co.

Chang, Y. (1982), Personalism and social change in Korea, in Y. Chang, T. Kwon & P. J. Donaldson, ed: Society in transition: with special reference to Korea, 29-43, Seoul, Korea: Seoul National University Press.

Cohen, S. & Wills, T. A. (1985), Stress, social support, and the buffering hypothesis, Psychological Bulletin, 98, 310-357.

Chung, Y. (1989) The Impact of Chinese culture on Koreas economic development, In Tai, H., ed: Confucianism and economic development: An oriental alternative?, 149-165, Washington, D. C.: The Washington Institute Press.

Clark, L. A. & Watson, D. (1991), General affective dispositions in physical and personality health, In C. R. Cynder & D. R., ed: Handbook of social and clinical psychology, 221-245, New York: Pergamon Press.

Curry, J. P., Wakefield, D. S., Price, J. L. & Mueller, C. W. (1986), On the causal ordering of job satisfaction and organizational commitment, Academy of Management Journal, 29, 847-858.

Eisenberger, R., Huntington, R., Hutchison, S. & Sowa, D. (1986), Perceived organizational support, Journal of Applied Psychology, 69, 615-622.

Gardner, D. G., Pierce, J. L., Dunham, R. B. & Cummings, L. L. (1992), Single- item versus multi-item measurement scales: An empirical comparison, A paper presented at the annual meeting of the Academy of Management, Las Vegas.

Goldberg, D. P. (1972), The detection of psychiatric illness by questionnaire, Oxford: Oxford University Press.

Greenberger, E., Goldberg, S. H., Hamill, S., ONeil, R. & Payne, C. K. (1989), Contributions of a supportive work environment to parents well-being and orientation to work, American Journal of Community Psychology, 17, 755-783.

Hackman, J. R. & Oldham, G. R. (1975), Development of the Job Diagnostic Survey, Journal of the Applied Psychology, 60, 151-170.

Han, N. C. (1992) An empirical study of an organizational commitment model in South Korea, Unpublished doctoral dissertation, University of Iowa.

Hannan, M. T. & Freeman, J. (1977), The population ecology of organizations, American Journal of Sociology, 82(5), 929-966.

Homans, G. C. (1961), Social behavior: Its elementary forms, New York: Harcourt, Bruce.

House, J. S. (1980), Occupational stress and the mental and physical health of factory workers, Ann Arbor, MI: Survey Research Center, University of Michigan.

House J. S. (1981), Work stress and social support, Reading, MA: Addison-Wesley.

Iverson, R. D. (1992), Employee intent to stay: An empirical test of a revision of the Price and Mueller model, Unpublished doctoral dissertation, The University of Iowa.

Iverson, R. D. & Roy, P. (1994), A causal model of behavioral commitment: Evidence from a study of Australian blue-collar employees, Journal of Management, 20, 15-26.

Joreskog, K. G. & Sorbom, D. (1989), LISREL 7: A guide to the program and applications, Chicago: SPSS publications.

Joreskog, K. G. & Sorbom, D. (1993), LISREL 8: Users reference guide, Chicago: Scientific International Software, Inc.

Kalleberg, A. L. (1977), Work values and job rewards: A theory of job satisfaction, American Sociological Review, 5, 124-143.

Kal]eberg, A. L. & Sorensen, A. B. (1979), The sociology of labor markets, Annual Review of Sociology, 5, 351-379.

Kanungo, R. N. (1982), Work alienation, New York: Praeger.

Karasek, R. & Theorell, T. (1990), Healthy work, New York: Basic Books.

Kaufman, G. M. (1983), The determinants of strikes over time and across industries, Journal of Labor Research, 4(2), 159-175.

Kim, K. (1990), The distinctive features of South Koreas development, In P. L. Berger & H. M. Hsiao, ed: In search of an East Asian development model, 197-219, New Brunswick, New Jersey: Transaction Publishers.

Lawler, E. E. III. (1973), Motivation in work organizations, CA: Brooks/Cole Monterey.

Lawrence, P. R. & Lorsch, J. W. (1967), Organization and environment, Boston: Division of Research, Harvard Business School.

Lincoln, J. R. & Kalleberg, A. L. (1990), Culture, control, and commitment: A study of work organization and work attitudes in the United States and Japan, New York: Cambridge University Press.

McClelland, D. C., Atkinson, J. W, Clark, R. A. & Lowell, E. L. (1953), The achievement motives, New York: Appleton-Century-Crofts.

Mclver, J. P. & Carmines, E. G. (1981), Unidimensional scaling, Beverly Hills: Sage Publications.

Meyer, J. P. & Allen, N. J. (1987), Organizational commitment: Toward a three- component model, Research Bulletin No. 660, The University of Ontario, Department of Psychology, London.

Miller, T. I. (1984), The effects of employer-sponsored child care on employee absenteeism, turnover, productivity, recruitment or job satisfaction: What Is claimed and what is known, Personnel Psychology, 37, 277-289.

Misra, S., Kanungo, R. N., Rosentiel, L. V. & Stuhier, E. A. (1985), The motivational formulation of job and work involvement: a cross-national study, Human Relations, 38(6), 501-518.

Mobley, W. H. (1982), Employee turnover, Reading: Addison-Wesley.

Mobley, W. H., Griffith, R. W, Hand, H. H. & Meglino, B. M. (1979), Review and conceptual analysis of the employee turnover process, Psychological Bulletin, 86, 493-522.

Mottaz, C. J. (1988) Determinants of organizational commitment, Human Relations, 41, 467-482.

Mowday, R. T., Porter, L. W. & Dubin, R. (1974), Unit performance, situational factors, and employee attitudes in spatially separated work units, Organizational Behavior and Human Performance, 12, 231-248.

Mowday, R. T., Porter, L. W. & Steers, R. M. (1982), Employee-organization linkages: The psychology of commitment, absenteeism, and turnover, New York: Academic Press.

Muchinsky, P. M. & Thttle, M. L. (1979), Employee turnover: An empirical and methodological assessment, Journal of Vocational Behavior, 14, 43-77.

Mueller, C. W. & Price, J. L. (1990), Economic, psychological and sociological determinants of voluntary turnover, The Journal of Behavioral Economics, 19, 321-335.

Mueller, C. W, Boyer, E. M., Iverson, R. D. & Price, 1. L. (1994), Employee attachment and noncoercive conditions of work: The case of dental hygienists, Work and Occupations, 21, 179-212.

Oldham, G. R. & Brass, D. J. (1979), Employee reactions to an open-plan office: A naturally occurring quasi-experiment, Administrative Science Quarterly, 24, 267-284.

Perrow, C. (1967), A framework for the comparative study of organizations, American Sociological Review, 32, 194-208.

Porter, L. W. & Steers, R. M. (1973), Organizational, work, and personal factors in employee turnover and absenteeism, Psychological Bulletin, 80, 151-176.

Porter, L. W, Steers, R. M., Mowday, R. T. & Boulian, P. V. (1974), Organizational commitment, job satisfaction, and turnover among psychiatric technicians, Journal of Applied Psychology, 59, 603-609.

Price, J. L. (1995), A role for demographic variables in the study of absenteeism and turnover, The International Journal of Career management 7(5), 26-32.

Price, J. L. & Kim, S. (1993), The relationship between demographic variables and intent to stay in the military: Medical personnel in U. S. Air Force hospital, Armed Forces & Society, 20 (1), 125-144.

Price, J. L. & Mueller, C. W (1981), Professional turnover: The case of nurses, Bridgeport: Luce.

Price, J. L. & Mueller, C. W (1986a), Handbook of organizational measurement, Scranton: Harper Collins.

Price, J. L. & Mueller, C. W. (1986b), Absenteeism and turnover among hospital employees, Greenwich, CT: JAI Press.

Rizzo, I. R., House, J. R. & Lirtzmari, S. I. (1970), Role conflict and ambiguity in complex organizations, Administrative Science Quarterly, 15, 150-163.

Roethlisberger F, J., & Dickson, W J. (1938), Management and the worker, Cambridge, MA: Harvard University Press.

Russell, J. A., Weiss, A. & Mendelsohn, G. A. (1989), Affect grid: A single-item scale of pleasure and arousal, Journal of Personality and Social Psychology, 57(3), 493-502.

Seashore, S. E. (1954), Group cohesiveness in the Industrial workgroup, Ann Arbor, MI: Survey Research Center, University of Michigan.

Shils, E. A. (1950), Primary groups in the American army, In R. K. Merton & P. F. Lazarsfeld, ed: Studies in the scope and method of The American Soldier, 16-39, Glencoe, IL: The Free Press.

Shils, E. A. & Janowitz, M. (1948), Cohesion and disintegration in the Wehrmact in World War II, Public Opinion Quarterly, 12, 280-315.

Steers, R. M. (1977), Antecedents and outcomes of organizational commitment, Administrative Science Quarterly, 22, 46-56.

Thompson, J. D. (1967), Organizations in action, New York: McGraw-Hill.

Uden-Holman, T. M. (1992), Commitment and intentions to stay among U. S. Force enlisted personnel, Unpublished doctoral dissertation, The University of Iowa.

Vandenberg, R. I. & Lance, D. E. (1992), Examining the causal order of job satisfaction and organizational commitment, Journal of Management, 18(1), 153-167.

Viscusi, W K. (1979), Employment hazards, Cambridge, MA: Harvard University Press.

Vogel, E. (1991), The four little dragons, Cambridge, MA: Harvard University Press.

Wallace, J. E. (1992), Employee attachment among law firm lawyers: A study of organizational commitment and intent to stay, Unpublished doctoral dissertation, The University of Iowa.

Watson, D. & Clark, L.A. (1984), Negative affectivity: The disposition to experience adversive emotional states, Psychological Bulletin, 96, 465-490.

Watson, D., Pennebaker, J. W & Folger, R. (1986), Beyond negative affectivity: Measuring stress and satisfaction in the work place, Journal of Organizational Behavior management, 8, 141-157.

Watson, D. & Tellegen, A. (1985), Toward a consensual structure of mood, Psychological Bulletin, 98, 219-235.

Wiener, Y. (1982), Commitment in organizations: A normative view, Academy of Management Review, 7, 4 18-428.

Woodward, J. (1965), Industrial organization: Theory and practice, London: Oxford University Press.

Yoon, I., Baker, M. R. & Ko, J. (1994), Interpersonal attachment and organizational commitment: Subgroup hypothesis revisited, Human Relations, 47(3), 329-351.

Appendix

Selected Measures of the Variables
Variable Measure Source
Organizational Commitment (1) 1 am proud to tell others I am part of this hospital. Porter et al. 1974
(2) The hospital inspires the very best job in the way of job performance.
(3) I am glad that I chose this hospital to work for over others I was considering at the time I joined.
(4) This hospital is the best of all possible places to work for me.
Job Satisfaction (1) 1 feel fairly well satisfied with my job. Brayfield and Rothe, 1951
(2) 1 find enjoyment in my job.
(3) I like working here better than most other people I know in this hospital.
(4) I am seldom bored with my job.
(5) I would consider taking another kind of job.
(6) Most days, I am enthusiastic about my job.
Opportunity It would not be easy for me to find a job with another employer that is better than the one I now have. Price and Mueller, 1986a
Kinship Involvementa Blegen et al., 1988
Met Expectations Generally, this hospital has been what I thought it would be. New Measure
Work Involvement Work should be central to life. Kanungo, 1982
Positive Affectivity I usually find ways to liven up my day. Watson, Correspondence
Negative Affectivity There are days when I am “on edge” all of the time. Watson, Correspondence
Autonomy I have much freedom in deciding what tasks I will do. Price and Mueller, 1986a
Role Ambiguity I know exactly what is expected of me in doing my job. Rizzo et al., 1970
Role Conflict I often receive incompatible job requests from different supervisors. Rizzo et al., 1970
Workload I do not have enough time to get everything done in my job. Rizzo et al., 1970
Coworker Support My peers are willing to listen to my job-related problems. Caplan et al., 1975
Supervisory Support My supervisor shows a lot of concern for me. Caplan et al., 1975
Organizational Support This hospital does not value my contribution to its well-being. Eisenberger et al. 1986,
Routinization My job has variety. Price and Mueller, 1986a
Distributive Justice As compared to other employees, I am fairly rewarded for the amount of effort that I put forth. Price and Mueller, 1986a
Promotional Chances I have an opportunity for advancement in this hospital. Price and Mueller, l986a
Payb Price and Mueller, 1986a

aMeasured by the number of dependents.
b Measured by asking respondents to indicate their total monthly income before taxation.