ࡱ> q` RsbjbjqPqP::M%4)))h^**|Nz+z+"++++++MMMMMMM$OhyQMT?++T?T?M++MLLLT?L++MLT?MLLL+n+ za)EL MM0NLORKORLORL +N2L7:j+++MMLX+++NT?T?T?T?#))  The Impact of Positive and Negative Affectivity on Job Satisfaction* Sang-Wook Kim Department of Sociology Sungkyunkwan University 53 Myungryun 3 Ga-Dong Jongno-Ku, Seoul 110-745 Korea Telephone: 02-760-0412 Fax: 02-744-6169 E-mail: swkim@skku.edu Jong-Wook Ko Department of Urban Administration Anyang University San 102-103 Samseong-Ri, Buleun-Myon, Ganghwa-Kun, Incheon, 417-833 Korea Telephone: 032-930-6011 Fax: 032-930-6211 E-mail: jwko@aycc.anyang.ac.kr May 2005 Abstract The research reported in this paper examines the impact of positive and negative affectivity on job satisfaction. Data was collected from two organizations in South Korea, one which manufactures automobiles and the other which provides airline services. The method of collection was through the use of questionnaires and the resultant data was analyzed by LISREL. The main conclusion is that, while traditional structural determinants used by many organizational scholars to explain job satisfaction remain important, affectivity variables are also important and should be utilized to supplement traditional analysis. Key Word: Positive Affectivity, Negative Affectivity, Personality Traits, Job Satisfaction The Impact of Positive and Negative Affectivity on Job Satisfaction The purpose of this paper is to examine the impact of positive and negative affectivity on job satisfaction. Positive and negative affectivity are, respectively, tendencies to experience pleasant/unpleasant emotional states (Clark & Watson, 1991; Watson & Clark, 1984; Watson & Pennebaker, 1989). The affectivity variables are dispositions that closely correspond to two of the five main personality variables, namely extraversion (positive affectivity) and neuroticism (negative affectivity) (Judge et al., 2002). They are different concepts rather than opposite ends of a continuum (Agho et al., 1993), whose impact is mediated by selective perception and/or interpretation. A pleasant person, for example, is drawn towards positive features in his or her environment as opposed to negative ones and/or interprets ambiguous stimuli in a positive manner. Job satisfaction is the extent to which employees like their jobs (Vroom, 1964). The study of job satisfaction has a long history in the study of organizations. Western Electric Research (Roethlisberger & Dickson, 1939) initiated a sizeable amount of research on job satisfaction, which they labeled morale. Later, the work of Smith and her colleagues (Smith et al., 1969) on measurement helped to change this label from morale to job satisfaction. This paper focuses on a global view of job satisfaction, thus ignoring such facets as pay, co-workers, and so forth. A sizeable amount of literature supports the impact of positive and negative affectivity on job satisfaction (Brief et al., 1988; Brief et al., 1995; Burke et al., 1993; Cropanzano et al., 1988: Levin & Stokes, 1989; Necowitz & Roznowski, 1994; Staw et al., 1986; Staw & Ross, 1985). Two types of impact are emphasized. Firstly, these affectivity variables can either displace or supplement the traditional determinants usually investigated by structurally oriented organizational scholars. Scholars such as these, conducting traditional research, commonly do not investigate dispositional variables. Lincoln and Kalleberg (1990), for example, in their investigation on job satisfaction and commitment, focused mainly on such structural variables as differentiation, centralization, and formalization rather than on dispositional variables. Secondly, and more important, the affectivity variables can contaminate, or bias, the measurements of other determinants of job satisfaction, such as the traditional determinants of structurally oriented organizational scholars. Since these scholars traditional determinants are commonly assessed by perceptual measures (Price, 1997), this potential contamination poses a major threat to the proper explanation of job satisfaction. Of course, both of these impacts pose serious challenges to traditional investigations. However, there is also a sizeable amount of literature that presents a skeptical view on the impact of the affectivity variables on job satisfaction (Chen & Spector, 1991; Davis-Blake & Pfeffer, 1989; Moyle, 1995; Munz et al., 1996; Schanbroeck et al., 1992; Schanbroeck & Ganster, 1991; Spector et al., 1988; Spector & OConnell, 1994; Williams & Anderson, 1994; Williams et al., 1996). This literature is not totally for or against the importance of affectivity variables. Typically, it finds some empirical support for one of the two affectivity variables, often negative affectivity, since this is more commonly investigated. This study, however, examines the impact of both affectivity variables on job satisfaction, and as such is broader than such other typical investigations. As will be shown, this study seeks in diverse ways to improve on previous investigations into the impact of the affectivity variables on job satisfaction. This is the basic rationale behind the research that is presented. A typical investigation of the affectivity variables impact on job satisfaction examines the impact of job-stress variables, commonly referred to as the stress-strain relationship. In this literature, strain is usually indicated by job satisfaction, and job-stress variables are noted as especially vulnerable to contamination by the affectivity variables. Sometimes autonomy and social support are also investigated, for these variables are believed to moderate the job-stress variables impact on job satisfaction (Karasek & Thorell, 1990). Employees who experience high job stress, for instance, may not be dissatisfied if they have a high degree of autonomy and/or social support. However, the investigation for this paper examines the affectivity variables impact with a causal model of job satisfaction (Agho et al., 1993; Kim et al., 1996; Price, 1977; Price, 2001; Price & Mueller, 1981; Price & Mueller, 1986). This model includes job stress, autonomy, and social support plus other determinants which may be contaminated by the affectivity variables and which may influence job satisfaction. This investigation is thus broader than a typical study on the affectivity variables impact on job satisfaction once again, a study that improves on previous investigations on the impact of the affectivity variables. To reiterate: seeking improvements is what has driven this research. The Causal Model The causal model used in this research is based on the work of Price/Mueller and their colleagues at the University of Iowa. Like the Lincoln and Kalleberg model, the Price/Mueller model is basically a structural explanation. Since detailed documentation concerning the different components of the Price/Mueller model is contained in a series of other publications, this paper will only present a condensed description of the model. However, enough information will be presented to enable the reader to grasp the models essential elements. These essential elements must be grasped to understand the results of our research. This condensed description is justified, because the major thrust of the paper is not directed towards the estimation of a model of satisfaction, but rather towards the affectivity variables impact on satisfaction. The model contains environmental, structural, and psychological variables. Kinship responsibility and opportunity are the models two environmental variables. Kinship responsibility denotes the existence of obligations to relatives residing in the community. The emphasis is on close kin mothers and fathers, for example rather than on more distant kin, such as uncles and aunts. Opportunity the availability of alternative jobs in an environment is a labor market variable typically emphasized by economists. The amount of employment is a common measure of opportunity. In the model, strong kinship responsibility increases and high opportunity decreases satisfaction, respectively. The core of the model consists of twelve structural variables. Each of the structural variables is either a reward or a punishment, depending on the form of the variable. For example, autonomy and routinization are, respectively, a reward and a punishment. Routinization the extent to which jobs are repetitive is a technology variable since it refers to the process of transforming inputs into outputs. The opposite of routinization is often labeled as Variety. This view of routinization is based on the work of Perrow (1967). Autonomy is a power variable with a narrow focus on the extent of decision-making in the job. Participation in workgroup decisions, for instance, is ignored by autonomy. This view of autonomy is based on the research by Breaugh (1989). This model contains four job-stress variables: role ambiguity, role conflict, workload, and resource inadequacy. Each of these four variables refers to a situation in which it is difficult to conform to job obligations. Role ambiguity means unclear obligations; role conflict indicates inconsistent obligations; workload signifies the amount of work to do; and, resource inadequacy notes the lack of means to do the work. Job stress is viewed from the perspective of the Survey Research Center of the University of Michigan (House, 1981). There are two social support variables in the model, supervisor and peer. These refer the instances where an employee can receive social assistance from his/her supervisor and/or peers. This view of social support also comes from the Survey Research Center (House, 1981). Historically, social scientists have referred to peer support in discussions on workgroup cohesion and primary groups. There are two justice variables: distributive justice is the extent to which rewards and punishments are related to job performance and procedural justice is the degree to which norms are applied universally to all employees. The justice concepts are often investigated in research on the notion of fairness. Promotional chances degree of potential vertical mobility within an organization is closely akin to classic sociological investigations into vertical mobility. Pay is a variable emphasized by economists and refers to money and its equivalents that employees receive for their services to the organization. Cash income is the usual measure of pay, since equivalents, or fringe benefits, are difficult to measure. For the structural variables, satisfaction is increased by high amounts of autonomy, social support, distributive/procedural justice, promotional chances, and pay. High amounts of routinization and job stress decrease satisfaction. There are three psychological variables in the model: job involvement, positive affectivity, and negative affectivity. Literature about involvement (the willingness to exert effort on the job) is often found in discussions on motivation, central life interests, and the Protestant work ethic. The affectivity variables have already been described. Satisfaction is increased by high involvement and positive affectivity and decreased by negative affectivity. Three general assumptions characterize the model. First, it is assumed that there is an exchange of benefits between the organization and its employees. If an organization provides good working conditions, the employees, in exchange, will fulfill their job obligations as set forth by the organization. This view is emphasized by the exchange approach (Gouldner, 1960). Second, it is assumed that employees will act to maintain a balance of rewards over punishments; for example, if an employee receives a very good outside job offer, he/she will leave unless the organization increases the rewards enough to make them stay. The greater the balance of rewards over punishments, the better it is from the employees perspective. March and Simon (1958) adopt this type of motivational approach in their classic work on organizations. Third, it is assumed that employees bring expectations into the workplace and, if these expectations are met, the employees are satisfied. Expectations are cognitions regarding the nature of the workplace. The emphasis on expectations comes from the expectancy approach advanced by Mowday and his colleagues (1982). (Excluded from this paper, to simplify presentation, are some additional assumptions and intervening processes for a number of exogenous determinants. Price (2000) contains a full statement about the Price/Mueller model.) The model has three scope conditions. First, it applies to work organizations, that is, social systems in which employees are paid for their labor. It is not clear how to explain satisfaction for members in voluntary associations, such as churches or trade unions. Second, the model applies to organizations in Western societies. Most estimations for the model and its components have been done in the West and it is not clear how the model will work with regard to Asian societies. Since this study was done in South Korea, it is, in effect, testing the validity of the second scope condition. The South Korean samples and sites are a major asset of this study, because there are few comparative studies of the affectivity variables outside the West. Third, the model applies for reasons which are not apparent to employees who work full-time and who plan long-term relationships with their employers. It is not clear how the model will work for contingent employees. No demographic variables are used in the model. It is our belief and that of other researchers that demographic variables do not specify exactly what is producing their hypothesized effects (Price, 1995; Price & Kim, 1993). Age is an example of such a variable. If age appears to increase satisfaction, it is not clear what it is about age that produces the increase. Any number of conditions correlated with increased longevity could produce greater satisfaction. Variables without clear content such as this cannot be included in models. However, three demographic variables age, tenure, and education are used as controls in the analysis to assess unknown determinants that may impact on satisfaction. With a perfect model, none of the demographic variables would be significant because its variables would capture all sources of variation. Although we include the demographic variables, they exist as controls, which is not the same thing as fully incorporating them into the causal model. Methodology Two data sets are used in our analysis. The samples and sites provide two estimations of the affectivity variables impact on satisfaction. It is assumed that two estimations are better than one. It is also fortunate that the two samples and sites are different, for reasons which will be explained. It is also assumed that different samples and sites are better than repeated model estimations using identical samples and sites. This paper consists of the re-analysis of data collected for another purpose. Study 1 The first data set consists of a sample of employees in an automobile manufacturing company in South Korea (Kim, 1996). Based on the number of vehicles produced and sold in 1993, the company was South Koreas second largest automobile company at the time. The company used to have its corporate headquarters in the capital city of Seoul, two large plants located in Seouls suburbs, and it maintained numerous sales department/maintenance offices across the country. Self-administered questionnaires were distributed in the Summer of 1993 to 2,468 employees in the company; these employees worked at the corporate headquarters, two plants, and twenty randomly selected sales departments/maintenance offices. In translating the original English questionnaires into Korean, particular attention was given to retain the contextual equivalence between the original and its translation. The translation, for example, was managed by South Koreans who had received Ph.Ds in sociology in the United States. The employees returned 1,773 usable responses, the response rate being 71.8 percent. Females, those whose gender information was missing, part-time workers, and administrative/sales/maintenance workers were eliminated from the analysis, for the most part due to the small numbers represented and/or to make the sample more homogenous. The exclusion of these personnel, as well as the listwise deleted missing cases, brought the final sample analyzed for Study 1 to 1,289. This sample therefore consisted of full-time, male, blue-collar workers. Study 2 The second data set represents a sample of employees in an airline company (Ko, 1996). This company is the largest airline company in South Korea, with over 16,000 employees. It has a corporate headquarters in Seoul and maintains dozens of branches across the country. Judging from sales and production statistics, this company, as well as Study 1s auto company, is one of South Koreas Chaebols, or giant business conglomerates. Self-administered questionnaires were distributed in the Spring of 1995 to a total of 970 employees working at the corporate headquarters. The back-translation method (Brislin 1970) in which the English version was translated first into Korean and then translated again back to English was used to ensure equivalence between the original and its translation. Of the questionnaires distributed, 619 were returned for a response rate of 63.8 percent. Exclusion of females and fifteen cases with too much missing data, plus the listwise deletion of missing cases, brought the final sample analyzed for Study 2 to 461. The sample analyzed in Study 2 consisted of full-time, male, white-collar workers. Measurement The models affectivity variables were operationalized with items adapted from measures received through correspondence with Watson (see Appendix). Several researchers (Agho et al., 1993; Kim et al., 1996) have indicated that the measures have acceptable psychometric properties. Satisfaction, the dependent variable, was measured by six items adapted from the Brayfield and Rothe (1951) scale (see Appendix). Tapping a global assessment of an employees affective response to his/her job, the scale has very acceptable measurement properties (Brooke et al., 1988; Cook et al., 1981; Mueller et al., 1994). The models remaining variables were assessed with widely used organizational measures whose psychometric properties are well documented. As would be expected, perceptual measures were used to assess the variables - a customary practice in organizational research (Price, 1997). Except for the composite measure of kinship responsibility and pay, all variables were operationalized by multiple-item scales that used a five-point Likert-type answering format with verbal anchors. Each scale contained positively and negatively worded items to minimize the effects of subject response sets. The measures discriminant and convergent validity was evaluated by Exploratory Factor Analysis and, if necessary, by Confirmatory Factor Analysis (CFA). After dropping a few items that exhibited inappropriately weak or double loadings, most of the remaining measures demonstrated discriminant and convergent validity in both studies. Reliability was assessed by Cronbach Alpha; the results indicated adequate internal consistencies for almost all constructs. Table 1 presents the descriptive statistics of the concepts, reliability estimates, and sources of the measures. Both studies, of course, used identical measures. ----------------------------- Table 1 About Here ----------------------------- Analysis Data were analyzed by the maximum likelihood (ML) estimation procedures in LISREL8 (Joreskog & Sorbom, 1993). LISREL is most appropriate in this research for three reasons: (1) almost all of the models variables have multiple indicators; (2) use of latent variables in estimating the model corrects for random measurement errors or unreliabilities in manifest variables; and (3) ML estimation procedures provide a statistical evaluation of the overall goodness of the model fit to the data. Among the several model-fit statistics provided by LISREL 8, the study uses the Comparative Fit Index (CFI). The CFI is used because of its robustness for large sample sizes. The multivariate analysis of LISREL ML estimation requires the fulfillment of a few statistical assumptions. Two of the most important assumptions linearity and low multicollinearity were tested. Linearity was tested by standard F-tests that decompose linear and nonlinear components; the results indicated no significant deviations from linear association between any independent and dependent variables in either of the studies. Multicollinearity was tested by the Eigenvalue decomposition method (Gunst, 1983). The smallest Eigenvalue was larger than .05 in both studies, indicating that problematic symptoms of multicollinearity did not arise among the independent variables. Additional information pertinent to multicollinearity is provided in Table 2, which presents the LISREL-corrected correlation coefficients for all variables in the analysis. --------------------------- Table 2 About Here --------------------------- Results The following results for Study 1 and Study 2 are presented separately. For each study, the causal model is first estimated without positive and negative affectivity. This is the way structurally oriented organizational scholars have traditionally estimated their models, that is, without psychological or dispositional variables. In Table 3, results for both studies are contained in Model 1. The next three estimations for each study involve, successively, three additions to Model 1: negative affectivity (Model 2), positive affectivity (Model 3), and negative/positive affectivity (Model 4). Three estimations are necessary because the results change for each estimation. Table 3s results are reported as standardized coefficients (betas). Degrees of significance whether .05 or .001, for example - are ignored. What is important is whether or not a determinant is significant. ------------------------- Table 3 About Here ------------------------- Study 1 Model 1. All of satisfactions determinants are significant except resource inadequacy and peer support. A large number of significant determinants is to be expected, since the sample is quite large (N=1,289). Age, as anticipated, has a positive and significant impact on satisfaction. Although the models quality is not the major focus in this research what is critical is the affectivity variables importance, it is interesting to note that all the signs are as expected: the explained variance is forty-two percent (which is comparatively large), and the Comparative Fit Index is .964 (which is acceptable). The quality of the model is important, because of the comparative nature of the research; the South Korean samples and sites would not be appropriate if the quality of the model were poor. Comparison of Models 1 and 2. When the estimation includes negative affectivity, three determinants that were significant in Model 1 become insignificant: kinship responsibility, role conflict, and pay. This means that these three determinants are contaminated, or biased, by negative affectivity. Only one of these determinants, role conflict, is a job-stress variable. As previously noted, job-stress variables typically occupy an important place in the study of affectivity variables. When negative affectivity is introduced into the analysis, ten of the determinants that were significant in Model 1 remain significant and determinants that were not significant in Model 1 (resource inadequacy and peer support) remain insignificant. The introduction of negative affectivity would not be expected to produce significance where none previously existed. However, as anticipated, negative affectivity does decrease satisfaction in a negative manner. The positive sign and significance for age continues in this comparison. The explained variance now stands at forty-seven percent a five percent increase from Model 1 and the Comparative Fit Index remains basically unchanged. Comparison of Models 1 and 3. When the estimation includes positive affectivity, three determinants that were significant in Model 1 become insignificant: kinship responsibility, promotional chances, and pay. These three determinants are thus contaminated, or biased, by positive affectivity. None of these determinants is a job-stress variable. Ten determinants that were significant in Model 1 remain significant after the introduction of positive affectivity. Resource inadequacy and peer support, which were not significant in Model 1, remain insignificant. As expected, positive affectivity has a positive impact on satisfaction. The positive sign and significance for age continues. The explained variance now stands at forty-eight percent a six percent increase from Model 1 and the Comparative Fit Index remains almost unchanged. In short, kinship responsibility and pay are contaminated, or biased, by both affectivity variables. However, role conflict is contaminated only by negative affectivity and promotional chances is contaminated only by positive affectivity. Comparison of Models 1 and 4. When the estimation includes both affectivity variables, four determinants that were significant in Model 1 become insignificant: kinship responsibility, role ambiguity, promotional chances, and pay. Nine determinants are significant in both models. No insignificant determinants in Model 1 become significant in Model 4. Both affectivity variables are significant in the predicted directions. Age continues to have a positive and significant impact on satisfaction. The explained variance increases to fifty-one percent an eight percent increase from Model 1 and the Comparative Fit Index remains almost the same. Study 2 Study 2 will be discussed with the same order as Study 1. The results in this study are less complicated than those in Study 1 because fewer determinants are significant. This is to be expected since the sample size is considerably smaller (N=461). Model 1. Six and nine determinants are, respectively, significant and insignificant. The signs for all significant determinants agree with the model, the R2 is fifty percent, and the Comparative Fit Index (.920) is acceptable. None of the demographic variables is significant. As in Study 1, the quality of the model is quite acceptable, an important outcome in a comparative study. Comparison of Models 1 and 2. No changes occur in significance when negative affectivity is added to the analysis. Negative affectivity is significant and has the anticipated negative impact on satisfaction. There are no changes in the explained variance, the Comparative Fit Index, or the demographic variables. In short, there appears to be no contamination. Since the explained variance did not change, negative affectivity does not add anything of substance to this analysis. Comparison of Models 1 and 3. The addition of positive affectivity to the analysis produces only one change among the determinants: peer support is no longer significant. This, of course, indicates contamination by positive affectivity. Peer support is not a job-stress variable. Positive affectivity is thus significant and has the expected positive impact on satisfaction. The explained variance increases five points to fifty-five percent, whereas the Comparative Fit Index decreases slightly (from .920 to .909). None of the demographic variables is significant. In brief, only peer support appears to have been contaminated by positive affectivity. Since negative affectivity did not contaminate any variables, positive affectivity appears to be more important. Comparison of Models 1 and 4. When the estimation includes both affectivity variables, peer support is no longer significant, the explained variance increases to fifty-six percent (a six percent increase), the Comparative Fit Index declines slightly (from .920 to .905), and the demographic variables still remain insignificant. What is most interesting is that, among the affectivity variables, only positive affectivity is significant and its sign is as anticipated. In short, only one determinant (peer support) is contaminated, and positive affectivity is more important than negative affectivity in terms of contamination and net effects. Discussion As indicated earlier, the emphasis on positive and negative affectivity poses two challenges to the traditional explanation of satisfaction espoused by structurally oriented organizational scholars. First, rather than explaining satisfaction exclusively by structural variables, the affective variables offer a psychological explanation. This psychological explanation can either supplement or displace the traditional explanation. Second, because measures of traditional structural determinants may be contaminated, or biased, by the affectivity variables, these dispositional variables need to be controlled. These challenges must now be explicitly assessed given the findings of this study. It is clear that the affectivity determinants do not displace the traditional structural determinants. With both affectivity variables included in the estimations, seven and five structural determinants, respectively, in Study 1 and Study 2 continued to be significant. In short, an extreme position regarding the affectivity determinants importance vis--vis the structural determinants is clearly inappropriate. The traditional structural determinants continue to be important. To say that the affectivity variables fail to displace the structural variables, however, does not imply that the affectivity variables are unimportant. Consider the following evidence, indicating that the affectivity variables have important impacts on satisfaction. First, in Study 1, when both affectivity variables were included in the analysis (Model 4), three structural determinants (role ambiguity, promotional chances, and pay) became insignificant, thus indicating contamination. Kinship responsibility an environmental determinant also changed from significant in Model 1 to insignificant in Model 4, again indicating contamination. In Study 2, one structural determinant (peer support) lost significance when the affectivity variables were included (Model 4). These examples of contamination clearly indicate the affectivity variables importance. Second, the explained variance increased by eight and six percent, respectively, when the affectivity variables were added to Study 1 and 2. These increases in explained variance in both studies are critical evidence of the affectivity variables importance in explaining satisfaction. Third, in Study 1, both affectivity variables were significant in the predicted directions. The beta for positive affectivity (.220) was higher than any other apart from routinization (-.254), a determinant of long-standing importance in organizational analysis. In Study 2, of the affectivity variables, only positive affectivity was significant, but its beta (.286) was the highest of all betas. Thus, scholars who argue for the importance of the affectivity variables are correct to do so. Among organizational scholars, this point is most relevant to the research of Brief and his colleagues and Staw; among psychologists, it is most relevant to the research of Clark/Watson and their colleagues. That is, organizational scholars should incorporate the affectivity variables into their explanations of satisfaction as these variables can supplement traditional investigations. This incorporation will, of course, necessitate some major changes in traditional investigations, given that these studies ignore dispositional variables. Past investigations of the affectivity variables importance have focused heavily on the job-stress determinants. It is believed that these determinants are especially likely to be contaminated by the affectivity variables. The model estimated in this research contained four job-stress determinants: role ambiguity, role conflict, workload, and resource inadequacy. Of the five determinants which appeared to be contaminated, only one of them was a job-stress variable, role ambiguity. The remaining four were traditional structural variables. In short, contemporary research overemphasizes the importance of job-stress variables. It is thus important to examine the role of the affectivity variables and their impact on satisfaction with the kind of comprehensive model that was used in our research. Other, simpler models used in much of the research to date have tended to generate misleading results. In addition to job stress as an exogenous determinant, some investigations of the affectivity variables have also examined autonomy and support. This study includes autonomy and two types of support, supervisor and peer, as exogenous determinants. As indicated in the results section, only peer support appeared to be contaminated. Had this research only focused on job stress, autonomy, and support the exogenous determinants commonly examined in this field it would have neglected to account for three important contaminated determinants: kinship responsibility, pay, and promotional chances. This again emphasizes the importance of investigating the affectivity variables impact on satisfaction with a comprehensive model. Research on the importance of the affectivity variables has typically examined negative affectivity to the relative neglect of positive affectivity. In this research, we included both affective variables as determinants of satisfaction in Study 1. However, in Study 2, only positive affectivity was significant. Our studies indicate, therefore, that positive affectivity is more important than negative affectivity. This suggests that researchers should examine the impact of both affectivity variables on satisfaction. This is in line with Watsons review of related literature (2000), which consistently tests both the positive and negative components of the affectivity variable. Although model estimation was not the purpose of this research, it is noteworthy that the model estimated works quite well in South Korea, as judged by explained variance, confirmed predictions, and the model fit statistic. The models impressive quality supports the comparative focus of this research. A final note about age, a demographic variable, is required. Age was consistently significant in Study 1 but not Study 2. This means that unknown determinants, not captured by the model, were having an impact on satisfaction in Study 1. This state of affairs is common in the research tradition in which this study is involved; it takes time to improve models and their measures to eliminate the significance of such demographic variables. Conclusions The traditional structural determinants many organizational scholars use to explain job satisfaction remain important; however, the affectivity variables are also important and should be incorporated to supplement traditional structural analysis. This is especially significant because the traditional explanation of satisfaction as a rule does not use dispositional variables, such as positive and negative affectivity. These conclusions are solidly based on improvements this research makes on contemporary work that assesses the affectivity variables impact on satisfaction. Six improvements are noteworthy. First, rather than focusing only on negative affectivity as much contemporary research does this study examines the impact of both affectivity variables. This dual emphasis follows Watsons lead (2000). Second, rather than examining the impact of a limited number of exogenous determinants especially stress, autonomy, and social support this study uses a comprehensive model of satisfaction. Third, rather than using partial statistical techniques to estimate the model as some studies have done (Brief et al., 1988) this research uses LISREL, which provides a better estimation because it corrects for measurement error. Fourth, LISRELs measurement component also provides a Confirmatory Factor Analysis, a powerful measurement technique which assesses the validity of the models variables, especially the affectivities. Fifth, rather than using an inadequate measure of satisfaction as one of the major studies has (Staw and Ross, 1985) this research uses a condensed version of the Brayfield-Rothe Index (1951) that has excellent psychometric properties. Sixth, and finally, this research uses South Korean samples and sites, thereby expanding the scope of research beyond the typically Western-centric focus and bias of other studies. These six improvements are of major importance, providing a strong empirical foundation for the conclusions reported here and existing as the fundamental rationale behind the research undertaken. Four issues require investigation. First, it is not clear why the five variables (kinship responsibility, pay, role ambiguity, promotional chances, and peer support) were contaminated. There are no apparent differences between these five variables and the others in the model. Second, it is not clear why positive affectivity was more important than negative affectivity. Positive affectivitys importance in this research has added significance, since so much previous work on the affectivity variables has claimed that negative affectivity is the more important of the two. Many studies, for instance, focus only on negative affectivity. Third, this research has only investigated satisfaction as a strain variable. The reason for this emphasis is that most research on the affectivity variables examines satisfaction. However, there is some research on the affectivity variables and organizational commitment (Cropanzano et al., 1993), and future research should examine commitment as an illustration of strain. The results for commitment could turn out to be quite different than those for satisfaction. Fourth, and finally, this research has only investigated two of the big five personality variables. All of the five should be investigated. Conscientiousness, for example, seems to be especially promising as a determinant of satisfaction (Judge et al., 2002). To wrap up, this research, in examining the impact of the affectivity variables on satisfaction, supports the traditional structural explanation of satisfaction used by organizational scholars. However, the results gathered here also suggest that the traditional structural analysis be expanded to include the two dispositional determinants, positive and negative affectivity. Although our purpose was not to research model estimation, it is noteworthy that a model devised and estimated in the West works quite well in South Korea. As we have stated, important issues remain, yet their resolution will surely promote a better understanding of the determinants of satisfaction. * The authors would like to thank Drs. James L. Price and Charles W. Mueller (Univ. of Iowa) for their helpful comments on this paper. REFERENCES Agho, A., Mueller, C., & Price, J. (1993). Determinants of employee job satisfaction: An empirical test of a causal model. Human Relations, 46, 1007-1027. Agho, A., Price, J., & Mueller, C. (1993). Discriminant validity of measures of job satisfaction, positive affectivity, and negative affectivity. Journal of Occupational and Organizational Psychology, 65, 185-196. Anderson, J., & Gerbing, D. (1988). Structural equation modeling in practice: A review and recommended two-step approach. Psychological Bulleti,. 103, 411-423. Blegen, M., Mueller, C., & Price, J. (1988). The measurement of kinship responsibility for organizational research. Journal of Applied Psychology, 73, 402-409. Brayfield, A., & Rothe, H. (1951). An index of job satisfaction. Journal of Applied Psychology, 35, 307-311. Breaugh, J. (1985). The measurement of work autonomy. Human Relations, 38, 551-570. Breaugh, J. (1989). The work autonomy scale: Additional validity evidence. Human Relations, 42, 1003-1056. Brief, A., Burke, M., George, J., Robinson, B., & Webster, J. (1988). Should negative affectivity remain an unmeasured variable in the study of job stress? Journal of Applied Psychology, 73, 193-198. Brief, A., Butcher, H., & Roberson, L. (1995). Cookies, dispositional, and job attitudes: The effects of positive mood-inducing events and negative affectivity on job satisfaction in a field experiment. Organizational Behavior and Human Decision Processes, 62, 55-62. Brislin, R. (1970). Back-translation for cross-cultural research. Journal of Cross-Cultural Psychology, 1, 185-216. Brooke, P., Russell, D., & Price, J. (1988). Discriminant validation of measures of job satisfaction, job involvement, and organizational commitment. Journal of Applied Psychology, 73, 139-145. Burke, M., Brief, A., & George, J. (1993). The role of negative affectivity in understanding relations between self-reports of stressors and strains: A comment on the applied psychology literature. Journal of Applied Psychology, 78, 402-412. Caplan, R., Cobb, S., French, J., Van Harrison, R., & Pinneau, S. (1975). Demands and worker health. Washington, DC: U.S. Government Printing Office. Chen, P., & Spector, P. (1991). Negative affectivity as the underlying cause of correlations between stressors and strains. Journal of Applied Psychology, 76, 398-407. Clark, L., & Watson, D. (1991).General affective dispositions in physical and psychological health. In C. Snyder & D. Forsyth. (Eds.), Handbook of social and clinical psychology (pp. 221-245). New York: Perganon Press. Cook, J., Hepworth, S., Wall, T., & Barr, P. (1981). The experience of work. London: Academic Press. Cropanzano, R., James, K., & Konovsky, M. (1993). Dispositional affectivity as a predictor of work attitudes and job performance. Journal of Organizational Behavior, 14, 595-606. Davis-Blake, A., & Pfeffer, J. (1989). Just a mirage: The search for dispositional effects in organizational research. Academy of Management Review, 14, 385-400. Gouldner, A. (1960). The norm of reciprocity: A preliminary statement. American Sociological Review, 2, 161-178. Gunst, R. (1983). Regression analysis with multicollinear predictor variables: Definition, detection, and effects. Communications in Statistics: Theory and Methods, 12, 2217-2260. House, J. (1981). Work stress and social support. Reading: Addison-Wesley. Joreskog, K., & Sorbom, D. (1989). LISREL 8: Analysis of linear structural relationships by the method of maximum likelihood. Chicago: SPSS Publications. Judge, T., Heller, D., & Mount, M. (2002). Five-factor model of personality and job satisfaction: A meta-analysis. Journal of Applied Psychology, 87, 530-541. Kanungo, R. (1982). Work alienation. New York: Preager. Karasek, R., & Thorell, T. (1990). Healthy work. New York: Basic Books. Kim, S. W. (1996). Employee intent to stay: The case of automobile workers in South Korea. Unpublished Ph.D. dissertation. University of Iowa. Kim, S. W., Price, J., Mueller, C., & Watson, T. (1996). The determinants of career intent among physicians at a U.S. Air Force hospital. Human Relations, 49, 947-976. Ko, J. W. (1996). Assessments of Meyer and Allens three-component model of organizational commitment in South Korea. Unpublished Ph.D. dissertation. University of Iowa. Levin, I., & Stokes, J. (1989). Dispositional approach to job satisfaction: Role of negative affectivity. Journal of Applied Psychology, 74, 752-758. Lincoln, J., & Kalleberg, A. (1990). Culture, control, and commitment. Cambridge: Cambridge University Press. March, J., & Simon, H. (1958). Organizations. New York: Wiley. McFarlin, D., & Sweeney, P. (1992). Distributive and procedural justice as predictors of satisfaction with personal and organizational outcomes. Academy of Management Journal, 35, 626-637. Mowday, R., Porter, L., & Steers, R. (1982). Employee-organization linkages. Academic Press. New York: New York. Moyle, P. (1995). The role of negative affectivity in the stress process: Tests of alternative models. Journal of Organizational Behavior, 16, 647-668. Mueller, C., Boyer, E., Price, J., & Iverson, R. (1994). Employee attachment and noncoercive conditions of work: The case of dental hygienists. Work and Occupational, 20, 179-212. Munz, D., Huelsman, T., Konold, T., & McKinney, J. (1996). Are there methodological and substantive roles for affectivity in job diagnostic survey relationships? Journal of Applied Psychology, 81, 795-805. Necowitz, L., & Roznowski, M. (1994). Negative affectivity and job satisfaction: Cognitive processes underlying the relationship and effects on employee behaviors. Journal of Vocational Behavior, 45, 270-294. Nunnally, J. (1967). Psychometric theory. New York: McGraw Hill. Perrow, C. (1967). A framework for the comparative analysis of organizations. American Sociological Review, 32, 194-208. Price, J. (1977). The study of turnover. Ames: Iowa State University Press. Price, J. (1995). A role for demographic variables in the study of absenteeism and turnover. The International Journal of Career Management, 7, 26-32. Price, J. (1997). Handbook of organizational measurement. International Journal of Manpower, 18, 303-558. Price, J. (2000). Reflections on the determinants of voluntary turnover. International Journal of Manpower, 22, 600-624. Price, J., & Mueller, C. (1981). Professional turnover: The case of nurses. Bridgeport: Luce. Price, J., & Mueller, C. (1986). Absenteeism and turnover of hospital employees. Greenwich: JAP Press. Price, J., & Kim, S. W. (1993). The relationship between demographic variables and intent to stay in the military: Medical personnel in a U.S. Air Force hospital. Armed Forces and Society, 20, 125-144. Rizzo, J., House, R., & Lirtzman, S. (1970). Role conflict and ambiguity in complex organizations. Administrative Science Quarterly, 15, 150-163. Roethlisberger, F., & Dickson, M. (1939). Management and the worker. Cambridge: Harvard University Press. Schanbroeck, J., & Ganster, D. (1991). The role of negative affectivity in work-related stress. Journal of Social Behavior and Personality, 6, 319-330. Schanbroeck J., Ganster, D., & Fox, M.L. (1992). Dispositional affect and work-related stress. Journal of Applied Psychology, 77, 323-335. Sims, H., Szilagy, A., & Keller, R. (1976). The measurement of job characteristics. Academy of Management Journal, 19, 195-212. Smith, P., Kendall, L., & Hulin, C. (1969). The measurement of satisfaction in work and retirement. Chicago: Rand McNally. Spector, P., Dwyer, D., & Jex, S. (1988). Relation of job stressors to affective, health, and performance outcomes: A comparison of multiple data sources. Journal of Applied Psychology, 73, 11-19. Spector, P., & OConnell, B. (1994). The contribution of personality traits, negative affectivity, locus of control, and type A to the subsequent reports of job stressor and job strains. Journal of Occupational and Organizational Psychology, 67, 1-11. Staw, B., & Ross, J. (1985). Stability in the midst of change: A dispositional approach to job attitudes. Journal of Applied Psychology, 70, 469-480. Staw, B., Bell, N., & Clauson, J. (1986). The dispositional approach to job attitudes: A lifetime longitudinal test. Administrative Science Quarterly, 31, 56-77. Vroom, V. (1964). Work and motivation New York: John Wiley. Watson, D. (1990). Mood and temperment New York: The Guildford Press. Watson, D., & Clark, L. (1984). Negative affectivity: The disposition to experience aversive emotional status. Psychological Bulletin, 96, 465-490. Watson, D. & Tellegren, A. (1985). Toward a consensual structure of mood. Psychological Bulletin, 98, 219-235. Watson, D., Pennebaker, J., & Folger, R. (1987). Beyond negative affectivity: Measuring stress and satisfaction in the work place. Journal of Organizational Behavior Management, 8, 141-157. Watson, D., & Pennebaker, J. (1989). Health complaints, stress, and distress: Exploring the control role of negative affectivity. Psychological Bulletin, 96, 235-254. Williams, L., & Anderson, S. (1994). An alternative approach to method effects by using latent-variable models: Applications in organizational behavior research. Journal of Applied Psychology, 79, 323-331. Williams, L., Gavin, M., & Williams, M. (1996). Measurement and nonmeasurement processes with negative affectivity and employee attitudes. Journal of Applied Psychology, 81, 88-101. APPENDIX VariablesItemsPositive AffectivityI live a very interesting life. I usually find ways to liven up my day. Most days, I have moments of real fun.Negative AffectivityOften I get irritated at little annoyances. My mood often goes up and down. Minor setbacks sometimes irritate me too much.Job SatisfactionI feel fairly well satisfied with my job. Most days, I am enthusiastic about my job. I like working here better than most other people I know in this company. I do not find enjoyment in my job. I am often bored with my job. I would consider taking another kind of job. Reverse-coded.     PAGE  PAGE 6 MQ|^ _ b i k w x r s x { / C Z m n s w @ H l =>iu4@ &h6CJnHtHh6CJmH nHo(sH tHh6CJmH sH  h6CJh6CJnHo(tHPLMNOPQR`x   7 I b  $1$7$8$H$a$Mrb o m  d1$7$8$H$ $1$7$8$H$a$m n   o x!y!!%%'(*,I./#1 231$7$8$H$ $1$7$8$H$a$ d1$7$8$H$MX !"&'89:;W_ )BPSt `a $,6Ach6B*CJnHo(phtHh6CJnHtH h6CJh6CJnHo(tHT!%_c +0^ay}!fg"#U V ~  !! !!2!3!6!?!A!H!V!Y!]!h!w!y!h6CJnHtHh6CJnHo(tH h6CJXy!!" "n"o"""=#>#####$$W$g$$$$$&%<%A%L%z%{%%%%%%%.&1&]&^&&&&&!'"'''''?(P(c(d(~((((0)b)m)n)))))****\*]*****++}+~+++++++++++h66CJ]h6CJnHo(tH h6CJh65CJ\W++,, , ,,&,*,H,I,,,,,,,-- -+-z-{--- .!.J.T.k....+/,/g/u///// 00$0%0&0)0y0~000112222224353k3l3 44v4w4 55Q5R5555576O6Q6R666"7#7777777`8a8h66CJ] h6CJh6CJnHo(tHZa8s8u888!9&9h9i9p9r999M:N::::::: ;;);*;;;;;8<9<<<<<<< ====0=F=G=H=L=R=Z=w===== >!>G>J>>>>>??9?:?????0@1@9@<@G@M@o@|@@@@@@@@@@@@@AA-A0A;ANANNNFOIOTO`OPPPPPP4Q5QQQSSSSSS;T*CJh6B*CJnHo(phtHh66CJ] h6CJh6CJnHo(tHQ8[T[g[[[[ _$_7_Q_Y_b(gzjkkmnnpfrmuwwd d1$7$8$H$$a$1$7$8$H$ $d1$7$8$H$a$ $1$7$8$H$a$[[[\\5\6\\\\]]]]]6^7^{^|^^^^^Z_a___-`.`z`{```9a:aaawb{bbbbbccNcOcccccndodddZe\eieoeeeff@fAfffff)gEgggggggNhOhhhhh6B*CJnHo(phtH h6>*CJh6B*CJphh6CJnHo(tH h6CJQhhhhPiRi_i`iiiiiijyjkklkkkkkkQlRlllllllllllmm`mamnn*CJ h6CJh6CJnHo(tHTVsssssssptqtttttu8unuuvvvvvvvvvDwEwwwwwwwwwxxxxxTyUyyyPzWzgzhzzzzzzz#{p{z{{{||!|&|^|||||h66CJ]nHo(tHh6CJnHtHh65CJ\h6B*CJphh6B*CJnHo(phtH h6>*CJh6CJnHo(tH h6CJDwxz|8ʈ[IJVtٞ *C@d *C@1$7$8$H$ d1$7$8$H$||}:};}D}G}R}Z}]}c}f}}}_~a~~~~~+,À%&ǁс./AIQdÃ]`҄ބ #$]^#/h66CJ]nHo(tHh66CJ]h6CJnHtHh6CJnHo(tH h6CJQ/BKTXbipq\]ĉʼnÊ%&@AScxz !#NO͍Z[vy':<=s|ɏˏ̏ }~ِڐ -2HJVNO?Ih65CJ\h6CJnHtHh6CJnHo(tH h6CJW]^RSѕҕ<= ͗җ!HIR~.:TZhr/0 #FOq|IJtuҝӝ67`aڞ9;BCh6CJnHtHh6CJnHo(tH h6CJZ ST^bq}#3>šءL !"%¤äŤ/02sܥħŧǧ789ڨ̩B[ ()+F]vh66CJ] h66CJ h6CJh6CJnHo(tHX0Ф=ĦЧD6v)˭<;կt `@^` $ *@@da$v./01߮ ^Hegiϰ۰  14?]^`ƴȴʴ еѵӵn4QRSU#$&FYݸ޸h6CJnHo(tH h6CJ]h66CJ] h66CJ h6CJh6CJmH sH Q+ղkٳմF޵`1r7ι8v@Ҽ<Խ_߾ @^` `@^`ùĹŹ*+-Һ0^235ļżǼǽȽɽ3RTѾҾԾ A&GHJexD[\^[/02 h6CJ]h6CJnHo(tHh66CJ] h6CJ h66CJU߾ZSi= @^`<=MNPQSTVWYZ`abdeklmnorsпh0JmHnHu h60Jjh60JUh6jh6Uh6CJnHo(tHh66CJ] h6CJ `@d$If$ `@*d^*`a$ @^`3\ `@d$Ifrkd$$IfTl0^%.0%64 laT `@d$Ifrkd$$IfTl0^%.0%64 laT*U : `@d$Ifrkd>$$IfTl0^%.0%64 laT:;LMOPRSUVXY|qoooooooo d1$7$8$H$ `@Ed^E`rkd$$IfTl0^%.0%64 laT Ybcdopqrs d1$7$8$H$h]h&`#$/0P/ =!"#$%pp 6 00&P P. A!"#O$O% 200P/ =!"#$% $$If!vh55.#v#v.:V l0%6,55.4T$$If!vh55.#v#v.:V l0%6,55.4T$$If!vh55.#v#v.:V l0%6,55.4T$$If!vh55.#v#v.:V l0%6,55.4T8@8 \CJ_HaJmH sH tH J@J ȩ 1 $$d1$7$8$@&H$a$>*CJP@P ȩ 2 $$d1$7$8$@&H$a$ 5>*CJ\F@F ȩ 3$d1$7$8$@&H$ 5CJ\D@D ȩ 4$d1$7$8$@&H$>*CJF@F ȩ 5$$1$7$8$@&H$a$ 5CJ\F@F ȩ 6$d1$7$8$@&H$ 5CJ\(A@( 0 } 4Fi@F \ \ :V 44 la $k@$] L 6B@6 8d1$7$8$H$CJ. @.   !$)@$ t 8.@". 8  !s1iLMNOPQR`x7Ibomno xy% "$I&'#) *+A15999;; ?.A6AVDEEHlJLLM/M0M9M'OO8STSgSSSS W$W7WQWYWZ(_zbkceffhfjmmooprt8}ʀ[IJVtٖ0М=ĞПD6v)˥<;էt+ժk٫լFޭ`1r7α8v@Ҵ<Ե_߶ZSպiػ= ¾þľžƾǾȾɾʾ˾̾;ξϾоѾҾӾԾվ־׾ؾپھ۾ܾݾ޾߾3\ǿ*U :;LMOPRSUVXYbcdopqtI0I0I0I0I0I0I0I0I0I0I0I0I0I0I0I0I0I0I0I0I0I0I0I0I0I0I0I0I0I0I0I0I0I0I0I0I0I0I0I0I0I0I0I0I0I0I0I0I0I00H00000000I0H0000000000000000000000000(000(000(00000000(00000000(00000(000000(000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000 0 0 0 000 0 0 000 0 0 000000 0 00I00I00I00I00I00I00I00I00I00I00I00@0@0I00I00 $$$$$$$$$$$$$$$'y!+a8^z$$$$$$%%%%G3G3|;|;;;j<<<>>AA B BBBچچٙٙ]imm^^""ݨݨBBoo}Ϫܪǫ2<<[[%``""*CCyyt      !"#$%&'()*+,-./0213456789:;=<>?@ABCDEFGHJIKLMONPQRTSUVWYXZ[\]^_`abecdfghijkmlnpoqrstuvwxy=HHi$$$$$$%%%%R3R3;;;;o<<<??AABBBBޙgkqqdd))MMy˪ӪӪުƫѫѫ:DDcc$//ii )44ȶȶJJt  !"#$%&'()*+,-./0213456789:;=<>?@ABCDEFGHJIKLMONPQRTSUVWYXZ[\]^_`abdecfghijkmlnpoqrstuvwxy9E*urn:schemas-microsoft-com:office:smarttagsStateBt*urn:schemas-microsoft-com:office:smarttagscountry-region8v*urn:schemas-microsoft-com:office:smarttagsCity=x*urn:schemas-microsoft-com:office:smarttags PlaceType=y*urn:schemas-microsoft-com:office:smarttags PlaceName9z*urn:schemas-microsoft-com:office:smarttagsplace @zyxzvztzyxztztzxyzyyxzxyzyyxztztztvzvztztzvztztztzxyvEztzEzvzxyzvzvzEzEztzxyztzxyzvzyxzEzxyEzEzvzEvzyxxzvzvvzyxzxyzvzvzEzEW[x7=U`b*4_c. < [ e  # #+2AH04'\ i @!M!{!!!!""))@-H-?0E0>>@@BBCCDEcFgF$G-G2G7GII{JJKKKKMMMMNNPPQQQQQQQQHRYRRRyy{{FNko"06Мٜ=DПןq}.5 #v<C˥ӥ<Agu;CKQt{¨z լ۬/:`hpy19rxX`Ҵ<GOVԵߵipZatw&ǹιvKYMMOOPPRSUVXYqtbnoj!l!&&''22,=2=AA#C)CM M\SaS!V4V}VV,W1WZWbWZZ)_F_)c?clccfghhgjjnmm6tEtHK!0\]МQrAĞDpq٠0B\ 6F]v)O;]ΨϨܨOno +-=Ԫժk٫;<լ!"1ޭ̮`FZIJѲ/0_vmҴ<bcԵ2_߶ BCDٸ7ڹպiػ :02Y[ĿƿRT~;KMMOOPPRSUVXYqt33333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333330!n!LM}VVQWYWddef^8JDiMMOOPPRSUVXYqtMMOOPPRSUVXYqt63 *:;t1J@@[-s@@UnknownGz Times New Roman5Symbol3& z Arial;|i0Batang"heF̕f^3$vtb^tb^!xxd2AHP?2DThe Impact of Positive and Negative Affectivity on Job Satisfaction*Joyce E. CraigjeunOh+'0 ,< LX x   HThe Impact of Positive and Negative Affectivity on Job Satisfaction*Joyce E. CraigNormaljeun36Microsoft Office Word@D|@CQ@>Z@dwat՜.+,D՜.+,< hp  University of Iowa^b EThe Impact of Positive and Negative Affectivity on Job Satisfaction* 0@$_AdHocReviewCycleID_EmailSubject _AuthorEmail_AuthorEmailDisplayName_ReviewingToolsShownOnceX)second paper reviewstephenb@snu.ac.krStephen  !"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~Root Entry F@ڌaData 1TableORWordDocumentSummaryInformation(DocumentSummaryInformation8CompObjm  FMicrosoft Office Word MSWordDocWord.Document.89q