An Examination of Work-Related Correlates of Job Satisfaction In Programmer/Analysts

Transcription

1 An Examination of Work-Related Correlates of Job Satisfaction In Programmer/Analysts By: David K. Goldstein John F, Rockart Center for Information Systems Research Massachusetts Institute of Technology Sloan School of Management E Massachusetts Avenue Cambridge, MA A bstract In the past, research examining the work-related correlates of job satisfaction in programmer~analysts has focused on the relationship between characteristics of the job itself and job satisfaction. It is hypothesized that certain aspects of relationships with co-workers, project leaders, and users will also be significant correlates of job satisfaction. Specifically, the relationship between job satisfaction, role conflict, role ambiguity, and quality of leadership provided by supervisors and peers is examined. A questionnaire measuring job characteristics, role conflict and ambiguity, leaclership characteristics, and job satisfaction was administered to 118 programmer~analysts at four companies. The results indicate that both role and leadership variables correlated at least as highly with job satisfaction as job characteristics and that the addition of role and leadership variables to job characteristics significantly increases the explained variance in job satisfaction. Keywords:MIS management, personnel management, programmers, systems analysts, job satisfaction, job characteristics, role conflict, role ambiguity, leadership style ACM Categories: D.2.9, K.6.1 Introduction The study of job satisfaction is a major research area in industrial and organizational psychology. One estimate is that over 3000 articles have been written examining job satisfaction in general, and the satisfaction of certain types of workers in specific [14], Few MIS researchers, however, have applied the knowledge gathered by industrial and organizational psychologists to the study of job satisfaction in progr.ammer/analysts. There are strong reasons for studying job satisfaction in programmer/analysts. Research has shown that job satisfaction is negatively correlated with absenteeism, turnover, and other outcome variables [14]. Turnover is of special relevance for MIS managers, due to the shortage of experienced programmer/analysts and the high cost of training new hires. Ives and Olson [12] and Rockart [18] have found that the difficulties of hiring and training programmer/analysts are among the most important problems faced by MIS managers. This research focuses on understanding one key aspect of job satisfaction in programmer/analysts. It examines the work-related correlates of satisfaction -- correlates of job satisfaction stemming from the work carried out by programmer/ analysts. Two benefits can be gained from this type of research. First, studying work-related correlates of job satisfaction illuminates ways by which to improve the design of information systems jobs. Second, it provides insights into how changes in the work environment can affect jobs. This is especially important given the many changes that are occurring in the ways in which information systems are developed. Researchers, including Awad [1], Bostrom [3], Couger and Zawacki [5], and Willoughby [20], have studied work-related correlates of job satisfaction in programmer/analysts (see Bartol and Martin [2] for a review of this research). Of these studies, Couger and Zawacki s is the largest and most well-known. There is, however, a major shortcoming to the Couger and Zawacki study. The model that the researchers used, Hackman and Oldham s Job Characteristics Model [9, 10], only examines the relationship between the tasks carried out by workers and the workers feelings about their jobs. It assumes that workers do their jobs more or less independently. MIS Quarterly~June

2 This is not necessarily true for programmer/ analysts who tend to work closely with other programmer/analysts, users, and information systems managers in carrying out their jobs. This research examines two sets of variables that can be added-to the Job Characteristic Model and can measure some work-related correlates of job satisfaction not considered by Couger and Zawacki. One set of variables measures the amount of conflict and ambiguity in a worker s job [13]. The level of conflict and ambiguity perceived by workers has been found to be negatively correlated to job satisfaction [3, 11, 15, 17, 19]. High levels of conflict and ambiguity tend to occur among workers who have a great deal of contact with others, particularly if that contact is with people in other groups within the organization [13]. Since contact with information systems users is often an important part of the programmer/analyst s job, role perceptions should be significant correlates of job satisfaction. The second set of variables measures the quality of leadership provided by the worker s supervisors and peers [4]. These variables have been found to be positively correlated to job satisfaction in several studies (see Bowers and Seashore [4] for a review of this research). This paper describes a study extending Couger and Zawacki s [5] previous research. In order to place the study in perspective, the work of these researchers will first be reviewed. Following this, the two sets of variables introduced above will be discussed in greater detail. Finally, the results of a survey designed to examine the relative impact of these three sets of variables on job satisfaction will be presented and analyzed. Overview Hackman and OIdham s [9, 10] Job Characteristic Model (JCM), the model used Couger and Zawacki [5], is the dominant paradigm in the literature on job design [6]. The premise of the model is that personal and workrelated outcome variables, such as job satisfaction and work effectiveness, are related to objective characteristics of the job. The model focuses on five core job characteristics - skill variety, task identity, task significance, a~onomy, and feedback from the job itself -- that are hypothesized to be related to three key psychological states -- experienced meaningfulness of the work, experienced responsibility for the work, and knowledge of results. These psychological states are, in turn, related to personal and work related outcomes, including job satisfaction. These relationships are moderated by an individual s knowledge and skill, growth need (need for personal accomplishment), and satisfaction with his or her job contexts, such as pay and supervision. Couger and Zawacki [5] tested the JCM by surveying over 1000 programmers, systems analysts, and programmer/analysts. They found significant and moderately strong correlations between ratings on each of the five job characteristics and job satisfaction (correlation range was.26 to.40). In addition, they discovered that information systems personnel scored more highly on job satisfaction, job characteristics, and growth need measures than many of the other workers studied by Hackman and Oldham [10]. The JCM, however, assumes that the job itself provides most of the motivation and satisfaction for the worker. Hackman and Oldham state that the model was developed to study "jobs that are done independently by individuals working more or less alone" [10, p. 61 ]. As mentioned earlier, this is not necessarily true in systems development jobs, where programmer/analysts typically work in teams and spend a great deal of time dealing with users, co-workers, and managers. Therefore, extensions to Hackman and Oldham s model are needed to study the impact of workrelated factors, other than job characteristics, on the job satisfaction of programmer/analysts. One extension to the JCM is the inclusion of role perceptions developed by Kahn, Wolfe, Quinn, Stroek, and Rosenthal [13]. A role is defined as a set of expectations about the behavior of a person who is in a particular position within an organization. The study of roles explicitly considers the interactions of the worker with others. It is broader in scope than the JCM, which only examines the relationship between the worker and the tasks he or she performs. Kahn, et al noted that the growth of organizations had led to more complex and specialized jobs, such a~ systems development, where there is a great deal 104 MIS Quarterly~June 1984

3 of interdependency among people. The conflict and ambiguity found in these jobs can be a major source of personal.stress and an important correlate of job satisfaction. In their research, Kahn et al identified two variables -- role conflict and role ambiguity -- as potential sources of stress. Role conflict is "the degree of incongruity or incompatibility in the expectations or requirements communicated to a focal person" [3, p. 92]. These researchers identified several components of role conflict. Person-role conflict -- the extent to which role expectations are ingongruent with the orientations, standards, or values of the focal.person. Intrasender conflict -- the extent to which role requirements are incompatible with the resources or capabilities of the focal person. Intersender conflict -- the extent to which role requirements or expectations from one party oppose those from one or more other parties. Role overload -- the extent to which the various role expectations communicated to the focal person exceed the amount of time available for their accomplishment [3, p. 93]. Role ambiguity is "the degree to which desired expectations are vague, ambiguous, or unclear, thereby making it difficult for the person to fulfill the requirements (of his role)" [3, p. 93]. In their study of various jobs, Kahn, et al. found role conflict and role ambiguity to be high among people whose jobs require being innovative or involved boundary spanning, dealing with people outside of the person s work group. In more recent research, Bostrom [3] examined role conflict and role ambiguity in 75 userdesigner dyads~ each involved in maintaining an information system. He found significant negative correlations between the role conflict and ambiguity felt by the designer and his or her job satisfaction. He also found significant negative correlations between those variables and the user s satisfaction with the information system. Bostrom s results are consistent with the findings of other researchers who have examined the relationship between role variables and job satisfaction in many different types of jobs and organizations [11, 15, 17, 19]. The addition of role conflict and role ambiguity variables allow us to examine work-related correlates of job satisfaction not considered in the JCM. Interactions with users, co-workers, and information systems managers can be an important part of programmer/analysts jobs. These interactions should not directly affect skill variety, task identity, or any of the other job characteristics. However there are ways in which these interactions can affect the level of role conflict and role ambiguity perceived by programmer/analysts. For example, consider a hypothetical user-designer dyad like the ones studied by Bostrom. If the designer perceives that the user s requests for changes are vague the designer could experience an increase in role ambiguity. Similarly, if the designer perceives that the user s requests are unnecessary he or she could experience an increase in role conflict. The quality of leadership provided by a programmer/analyst s supervisor and peers is another potentially significant correlate of job satisfaction. Since systems development is usually a team effort the quality of the interactions and of the leadership provided by a programmer/ analyst s supervisor and peers could be important correlates of job satisfaction. The relationship between leadership and organizational and personal outcomes has been well documented. In reviewing the leadership research, Bowers and Seashore [4] found that four characteristics of leadership behavior were common to many studies. Support -- behavior that enhances someone else s feeling of personal worth and importance. Interaction facilitation -- behavior that encourages members of the (work) group develop close, mutually satisfying relationships. Goal emphasis -- behavior that stimulates an enthusiasm for meeting the group s goal or achieving excellent performance. Work facilitation -- behavior that helps achieve goal attainment by such activities as scheduling, coordinating, planning, and by providing resources such as tools, materials, and technical knowledge [4, p. 247]. These four characteristics can be used to examine the quality of leadership prov.ided by MIS Quarterly~June

4 Corre/ates of Job Satisfaction both the supervisor and the peers of a programmer/analyst. Bowers and Seashore found significant positive correlations between measures of these characteristics and job satisfaction and performance. As with role perceptions, leadership characteristics measure a work-related correlate of job satisfaction not considered by the JCM. An improvement in the quality of leadership provided by supervisors and peers should increase the job satisfaction of programmer/analysts and should not directly affect any of the job characteristics. The survey described below examines the relative importance of job characteristics, role perceptions, and leadership characteristics as correlates of job satisfaction in programmer/analysts. It is hypothesized that role conflict, role ambiguity, and the characteristics of supervisor and peer leadership discussed above will be significant correlates of job satisfaction in programmer/ analysts. Further, it is hypothesized that the addition of role perceptions and leadership characteristics to job characteristics will significantly increase our understanding of workrelated correlates of job satisfaction. Research Method Subjects and procedure Data were collected from 125 subjects at four large companies, each with over 100 programmer/analysts. Two were manufacturing companies located in the Northeast. The third was a Northeastern insurance company, and the fourth was a Midwestern insurance company. The programmer/analysts who participated in the study were involved in both the development of new systems and the maintenance of existing systems. Seven subjects were excluded from the study because they were not programmer/ analysts, leaving a final sample of 118 programmer/analysts. Participation in the study was voluntary. Subjects were contacted by the information systems management at the companies and were asked to participate. Approximately 75% of those contacted agreed to take part in the study. The questionnaire used to collect the data was administered in a classroom setting to groups of between seven and thirty subjects at a time. Participants completed the full Job Diagnostic Survey -- the instrument developed by Hackman and OIdham to study the JCM [10] -- plus sections added to measure role perceptions and leadership-characteristics. The procedures recommended by Hackman and OIdham [10, Appendix D] were followed in administering the questionnaire. Measures Role Conflict and Role Ambiguity The role conflict measure is based on an eight item scale developed by Rizzo, House, and Lirtzman [1 7]. Miles and Perrault [1 5] found that the scale could be separated into subscales that measure each of the components of role conflict identified by Kahn, eta/. [13]. The fourth subscale, used to measure role overload, was not significantly correlated with job satisfaction in a study of programmer/analysts [3] and is not included in this study. A sample item for each subscale, with the number of items in the subscale in parentheses, include: Person-role conflict (2) -- "1 have to do things that should be done differently." Intrasender conflict (2) -- "1 receive an assignment without adequate resources and materials to execute it." Intersender conflict (4) -- "1 do things that are apt to be accepted by one person and not by others." The six item role ambiguity measure developed by Rizzo, et al. [17] is used. A sample item follows: "There are clear planned goals and objectives for my job." Schuler, Brief, and AIdag [19] conducted a scale analysis of both the role conflict and role ambiguity measures and recommend their continued use. Supervisor and Peer Leadership Measures The scales developed by Bowers and Seashore [4] are used to measure supervisor and peer leadership characteristics. The first set of scales consists of thirteen items that measure the four supervisor leadership characteristics. The second set consists of eleven items that measure 106 MIS Quarterly~June 1984

5 the four characteristics of peer leadership. Sample supervisor leadership items include: Support -- "To what extent is your supervisor friendly and easy to approach?" Goal emphasis -- "To what extent does your supervisor encourage people to give their best effort?" Work facilitation -- "To what extent does your supervisor show you how to improve your performance?" Interaction facilitation -- "To what extent does your supervisor encourage the people who work for him to work as a team?" Yunger and Hunt [21 ] found these characteristics similar to the characteristics identified in the Ohio State Leadership Behavior Description (LBDQ) scale. Job Characteristics Hackman and Oldham s Job Diagnostic Survey (JDS) provides measures of five job characteristics. Each characteristic is measured with three questions from two different sections of the survey. Sample items include: Skill variety -- "The job requires me to use a number of complex or high-level skills." Task identity -- "The job is arranged so that I do not have the chance to do an entire piece of work from beginning to end." Task significance -- "This job is one where a lot of other people can be affected by how well the work gets done." Autonomy -- "The job requires a lot of cooperative work with other people." Feedback from the job itself -- "Just doing the work required by the job provides many chances for me to figure out how well I am doing." Job Satisfaction Four scales from the JDS are used to measure job satisfaction. The principle measure is Hackman and Oldham s general satisfaction scale. The other scales measure satisfaction with the opportunity of growth in the job, with co-workers, and with supervision. Sample items for each scale, with the number of items in the scale in parentheses, includes: General satisfaction (5) -- "Generally speaking, I am very satisfied with my job." Growth satisfaction (4) -- "How satisfied are you with the amount of personal growth and development you get in doing your job?" Co-worker satisfaction (3) -- "How satisfied are you with the people you talk to and work with on your job?" Supervisor satisfaction (3) -- "How satisfied are you with the amount of support and guidance you receive from your supervisor?" Results Means, standard deviations, and internal consistency reliabilities for the scales and subscales used in the study are presented in Table 1. All but one of the reliabilities is within the satisfactory range specified by Nunnally [16]. The reliabilities are consistent with results presented in previous studies [10, 19, 21]. The only exception is the person-role conflict subscale. The relative magnitudes of the means and standard deviations for this sample are very similar to the ones found in the samples examined by Couger and Zawacki [5] and Bostrom [3]. Correlations between the independent variables -- job characteristics, role perceptions, and leadership characteristics -- and the dependent variable -- job satisfaction -- are found in Table 2. All of the independent variables are significantly correlated with the satisfaction variables. The correlations between the job characteristics and general satisfaction are slightly higher than those obtained by Couger and Zawacki [5]. Of the job characteristics, autonomy correlated most highly with the satisfaction measures. Among the role perceptions, role ambiguity correlated more highly (negatively) with satisfaction than role conflict. Of the role conflict subscales, person-role,. conflict -- the perception that your job should be done differently -- corpelated most highly (negatively) with the satisfaction measures. These results are consistent with Bostrom ~ [3] findings. Among the leadership characteristics, supervisor work facilitation and supervisor support correlated most highly with job satisfaction. MIS Quarterly~June

6 Table 1. Summary Statistics for Dependent and Independent Variables Internal Consistency Variable Mean Std Dev Reliability Skill Variety Task Identity Task Significance Autonomy Feedback from Job Role Ambiguity Role Conflict Person-Role Intrasender Intersender Peer Support 5.O6 1.O1 Peer Goal Emphasis Peer Work Facilitation Peer Interaction Facilitation Supervisor Support Supervisor Goal Emphasis Supervisor Work Facilitation Supervisor Interaction Facilitation General Satisfaction Growth Satisfaction Co-Worker Satisfaction Supervisor Satisfaction Note: all variables are seven-point scales reliability calculated using Cronbach s alpha 108 MIS Quarter/y/June 1984

7 Table 2. Correlation of Independent Variables With Satisfaction Variables (N = 118) General Growth Co-Worker Supervisor Variable Satisfaction Satisfaction Satisfaction Satisfaction Skill Variety Task Identity Task Significance Autonomy Feedback From Job Role Ambiguity Role Conflict Person-Role Intrasender Intersender Peer Support Peer Goal Emphasis Peer Work Facilitation Peer Interaction Facilitation Supervisor Facilitation Supervisor Goal Emphasis Supervisor Work Facilitation Supervisor Interaction Facilitation In general, the role perceptions correlated more highly with the satisfaction variables than either the leadership characteristics or the job characteristics. For example, the median correlation of role perceptions with general satisfaction was.48, while the median correlations of the job characteristics and the leadership characteristics were.42 and.37, respectively. Of the two types of leadership characteristics, supervisor leadership tended to correlate more highly with job satisfaction than peer leadership. Correlations among the independent variables were also computed. There are significant, but relatively weak correlations between leadership characteristics and job characteristics (median =.21). The correlations between role perceptions and both leadership characteristics and job characteristics tended to be much stronger (median r =.29 and.31, respectively). See [7] for tables of correlations among the independent variables. Multiple regression analyses were performed to examine the amount of variance in job satisfaction explained by each of the three sets of independent variables. The job characteristics, role perceptions, and leadership characteristics were MIS Quarterly~June

8 Set of Variables Table 3. Variance in Job Satisfaction Explained by Each Set of Independent Variables (N = 118) Number of Variables Adjusted F-Statistic in Set R ~ R 2 for Regression Job Characteristics 5 Role Perceptions 2 Leadership Characteristics 8 ** -- result significant at the.01 level "* ** ** each regresse S against the general satisfaction variable. The results, presented in Table 3, indicate that all three sets of variables explain a significant amount of the variance in job satisfaction. Among the three sets, job characteristics accounted for the most variance in satisfaction, with role perceptions and leadership characteristics accounting for the second and third most variance, respectively. Further regression analyses were performed to test the hypotheses that the addition of role perceptions and leadership characteristics to job characteristics significantly increases the amount of variance in job satisfaction explained by the independent variables. The results shown in Table 4 indicate that the addition of role perceptions and leadership characteristics, either individually or together, significantly increase the explained variance in job satisfaction. The adjusted total variance in job satisfaction explained by the independent variables increases from 39% to 50%. Table 4 also provides data that allow us to analyze whether both additional sets of variables are needed or whether one additional set of variables is sufficient. If we start with a base regression containing job and leadership characteristics as. independent variables and then add role perceptions, we significantly increase the explained variance in job satisfaction. If, however, we start with job characteristics and role perceptions and then add leadership characteristics, the increase in variance explained is not significant. Table 4. Increase in Variance Due to the Addition of Role and Leadership Variables (N = 118) Base Set of Variables Added Base Total R 2 Variables R 2 R 2 Change F-Stat. for Change Adjusted Total Rz Job Characteristics role perceptions "*.390 Job Characteristics leadership characteristics *.444 Job Characteristics role perception and leadership characteristics 3.7**.496 Job Characteristics and Role Perceptions leadership characteristics Job Characteristics and role perceptions Leadership Characteristics * -- result significant at the.05 level * * -- result significant at the.01 level MIS Quarterly~June 1984

9 - role + conflict leadership characteristics job satisfaction - role + -ambiguity The correlation and regression analyses presented in Tables 2 and 4 indicate that there is a strong relationship between leadership characteristics and role perceptions, and between role perceptions and job satisfaction. The analyses prompted a further exploratory investigation into the relationship among these variables. House and Rizzo [11] observed that the relationship between leadership variables and job satisfaction is moderated by role variables. That is, positive leadership characteristics tend to reduce role conflict and ambiguity and lead to increased satisfaction. The relationships can be diagrammed as above: This finding is confirmed in the present study by the partial correlation analysis presented in Table 5. In all cases but one, controlling for either role conflict or role ambiguity reduced the relationship between leadership characteristics and general satisfaction. First order partial correlations controlling for role ambiguity were, in general, smaller than first order partials controlling for role conflict, indicating that role ambiguity is a more important intervening variable than role conflict. Controlling for both role conflict and role ambiguity did not reduce the partial correlations any more than controlling for just role ambiguity. Table 5. Partial Correlations of Leadership Characteristics and General Satisfaction Controlling for Role Conflict and Ambiguity (N = 118) GENERAL SATISFACTION Leadership Characteristics 2nd Order 1st Order 1st Order Partial Partial Partial Control for Zero Order Control for Control for Role Conflict Correlation Role Ambiguity Role Conflict and Ambiguity Peer Support.35 Peer Goal Emphasis.35 Peer Work Facilitation.32 Peer Interaction Facilitation.36 Supervisor Support Supervisor Goal Emphasis Supervisor Work Facilitation Supervisor Interaction Facilitation Note: r.15 are significant at the.05 level.47,38, o MIS Quarterly~June

10 Discussion The results presented above support our hypotheses that both leadership characteristics and role perceptions are important work-related correlates of job satisfaction in programmer/ analysts. These results lead to conclusions that have significant implications for both MIS researchers and practitioners. In this section, conclusions and implications are discussed in detail. Recommendations are then made for future research. The most important conclusion of this research is that extensions are needed to the Job Characteristic Model to allow a better understanding of the work-related correlates of job satisfaction in programmer/analysts. There is strong evidence that role perceptions should be added to the JCM. The correlations and regression analyses indicate that both role conflict and role ambiguity are significantly correlated to satisfaction and that, taken as a pair, the variables account for over 30% of the variance in job satisfaction. Adding role perceptions to a base regression containing job characteristics or to one containing job characteristics and leadership characteristics increases the explained variance in job satisfaction by 9% and 5%, respectively. There is moderately strong evidence that leadership characteristics should also be added to the JCM. The correlation analysis indicates that each of the peer and supervisor leadership characteristics are significantly correlated with job satisfaction. The regression analysis in Table 3 indicates that taken as a whole the leadership characteristics explain about 25% of the variance in satisfaction. The regression analysis in Table 4 indicates that adding leadership characteristics to a base regression containing job characteristics increases the explained variance in satisfaction by 9%. The addition of leadership characteristics to a base regression containing job characteristics and role perceptions, however, does not significantly increase the explained variance in satisfaction. Taken as a whole, the results indicate that leadership characteristics should be included in future studies of programmer/analysts. The cost of adding the variables is rather low (they require only two pages of the questionnaire) and they could illuminate leadership problems that might not be apparent if only job characteristics and role perceptions were examined. We cannot however, rule out the possibility that their addition would not significantly increase understanding of job satisfaction. The conclusion discussed above has significant implications for researchers and managers. If the JCM were to be used as the sole basis for work redesign, the redesign would only concentrate on understanding and modifying the characteristics of the tasks that the programmer/analysts perform. This approach to job changes would lead to proposals for work redesigns with serious shortcomings. Two potential problems with JCMbased work redesigns can be foreseen. First, a JCM-based work redesign that focuses only on job characteristics will ignore factors that significantly affect job satisfaction. For example, consider a hypothetical information systems departmenthat typically works on large development projects involving many diverse user groups. In this department, one group of programmer/analysts performs mainly programming tasks and a second group performs mainly analysis tasks. If it were determined that job satisfaction, skill variety, and task identity were all low, an advocate of job characteristics theory might recommend combining the tasks of the analysts and programmers to improve skill variety, and task identity [10]. This change, however, could lead to increases in role ambiguity and role conflict, especially among the programmers, who would be performing the less welldefined analysis tasks and who would be dealing more with users. Second, strategies for work redesign based on changing the jobs of users and information systems managers would not be considered if the JCM were the sole basis for the redesign. For example, in some companies the training of user liaisons might be an effective method for improving the job satisfaction of programmer/ analysts. By facilitating communication between programmer/analysts and users, the liaisons could reduce the amount of intersender role cor~flict perceived by the programmer/analysts. They could also reduce role ambiguity by eliminating some of the less well-defined tasks. Improving the leadership training of information systems managers, and of programmer/analysts 112 MIS Quarterly~June 1984

11 themselves, might be another effective way to improve job satisfaction which would not be considered if the JCM were used as the sole basis for work redesign. Many programmer/analysts and managers have a great deal of technical training in programming and systems design, have technical degrees, and have followed a career path that leads from program maintenance to program development, systems design, systems analysis, and project management. These programmer/ analysts and information systems managers lack training and experience in management -- planning, organizing, and controlling work -- and in interpersonal skills. Training in these areas could lead directly to reduced role conflict and ambiguity and, indirectly, to improved job satisfaction. Since these skills will make programmer/analysts and information systems managers more effective in carrying out their jobs, this training could also lead to improving productivity and quality of systems. At this point it is importanto interject a word of caution concerning the implications of the conclusion discussed above. Since the correlation and regression analyses used in this study were based only on cross-sectional data, we cannot directly conclude that changes in job characteristics, role perceptions, or leadership characteristics will cause corresponding changes in job satisfaction. The only way to show that changes in the independent variables will lead to changes in satisfaction is with a longitudinal study. Cross-sectional research can only suggest areas were changes might be warranted. A second conclusion of this study concerns the causal link between leadership characteristics, role perceptions, and job satisfaction. The partial correlation analysis in Table 5 supports the findings of House and Rizzo [1 1 ] that role perceptions, and particularly role ambiguity, act as intervening variables between leadership characteristics and job satisfaction. This indicates that an increase in the amount of aid provided by supervisors and peers could cause a decrease in the amount of role ambiguity and role conflict, which, in turn, could cause an increase in job satisfaction. A word of caution concerning the use of partial correlations is also needed. Even though it is logical to assume that role perceptions act as intervening variables between leadership characteristics and job satisfaction, the direction of this causal link cannot be determined with certainty by a partial correlation analysis. The data in Table 5 could also be used to support the hypothesis that role perceptions are determinants of both leadership characteristics and job satisfaction. That is, high role conflict and role ambiguity lead to the perception of poor leadership and poor job satisfaction. Implications for Future Research One area where research is warranted is the examination of the effects of certain task-related factors on job satisfaction and its correlates. For example, a survey could be used to examine the differences in job satisfaction between programmer/analysts involved in systems development and those involved in systems maintenance. It is likely that maintenance programmer/analysts would be less satisfied than development programmer/analysts because their jobs are more routine. We might also find that maintenance programmer/analysts derive more satisfaction from job characteristics because they deal with fewer people outside their work group. If this were the case, the strategies needed for improving systems maintenance jobs would be different than the strategies recommended in the previous section for improving systems development jobs. A second area where further cross-sectional research could be carried out is the study of individual differences. In the present study, we examined the correlates of job satisfaction for the average programmer/analyst, ignoring differences among different types of programmer/analysts. The study of these differences could provide important insights. For example, there might be some programmer/ analysts who prefer a less structured work environment and who could be very satisfied in a job with high levels of role ambiguity, while others might prefer a more structured and less ambiguous environment. Future research could provide ways to identify these individual differences and to determine the proper work environment for these different types of programmer/analysts. MIS Quarterly~June

12 There are several areas where longitudinal research can be used to examine job satisfaction. One area of particular importance is the study of the effects of change in the information systems environment on job satisfaction. Goldstein [8] presents a research plan for studying one such change, the introduction of a productivity aid. Longitudina~ research can also be used to examine the effects of other job or organizational changes on job satisfaction. For example, one might want to study the change in job satisfact(on, role perceptions, and leadership characteristics caused by giving programmer/analysts and project leaders a leadership training course as recommended in the previous section. One might also want to examine the effects of introducing user liaisons on role perceptions and job satisfaction. The main conclusion of this research is that extensions are needed to the JCM to better understand the work-related correlates of job satisfaction. This conclusion has important implications for researchers and practioners interested in redesigning programmer/analysts jobs. Work redesigners must not focus solely on modifying the tasks performed, but in addition, must also examine the interactions between information systems users and programmer/analysts that could affect role conflict and role ambiguity. They must consider the quality of leadership provided by managers and co-workers and how improvements in this area can lead to higher job satisfaction. Acknowledgement The authors acknowledge the helpful comments of Lotte Bailyn, Judith Quillard, Michael Treacy and the anonymous referees on earlier versions of this article. References [1] Awad, E.M, "Prediction of Satisfaction of Systems Analysts, Programmers," Data Management, Volume 15, Number 1, January 1977, pp [2] Bartol, K.M. and Martin, D.C. "Managing Information Systems Personnel: A Review of the Literature and Managerial Implications," MIS Quarterly, Special Issue December 1982, pp [3] Bostrom, R.P. "Role Conflict and Ambiguity: Critical Variables in the User- Designer Relationship," Proceedings of the Seventeenth Annual Computer Personnel Research Conference, 1981, pp [4] Bowers, D.G. and Seashore, S.E. "Predicting Organizational Effectiveness with a Four-Factor Theory of Leadership," Administrative Science Quarterly, Volume 11, Number 2, September 1966, pp [5] Couger, D. and Zawacki, R. Motivating and Managing Computer Personnel, John Wiley and Sons, New York, New York, [6] Evans, M.G., Kiggundu, M.N., and House, R.J. "A Partial Test and Extension of the Job Characteristics Model of Motivation," Organizational Behavior and Human Performance, Volume 24, Number 3, December 1979, pp [7J Goldstein, D.K. "A Further Examination of the Determinants of Job Satisfaction in Programmer/Analysts," Center for Information Systems Research, Working Paper 96, Cambridge, Massachusetts, [8] Goldstein, D.K. "The Effects of Structured Development Methods on the Job Satisfaction of Programmer/Analysts: A Theoretical Model," Center for Information Systems Research, Working Paper 90, Cambridge, Massachusetts, [9] Hackman, J.R. and Oldham, G.R. "Motivation Through the Design of Work: A Test of a Theory," Organizational Behavior and Human Performance, Volume 16, Number 2, August 1976, pp [10] Hackman, J.R. and OIdham, G.R. Work Redesign, Addison-Wesley, Reading, Massachusetts, [11] House, R.J. and Rizzo, J.R. "Role Conflict and Ambiguity as Critical Variables in a 114 MIS Quarterly~June 1984

13 Model of Organizational Behavior," Organizational Behavior and Human Performance, Volume 7, Number 3, June 1972, pp [12] Ives, B. and Olson, M.H. "Manager or Technician? The Nature of the Information Systems Managers Job," MIS Quarterly, Volume 5, Number 4, December 1981, pp [13] Kahn, R.L., Wolfe D., Quinn R., Snoek J.D., and Rosenthal R. Organizational Stress: Studies in Role Conflict and Role Ambiguity, John Wiley and Sons, New York, New York, [14] Locke, E.A. "The Nature and Causes of Job Satisfaction," in Dunnette, M.D. (ed.) Handbook of Industrial and Organizational Psychology, Rand McNally, Chicago, Illinois, [15] Miles, R.H. and Perrault, W.D. "Organizational Role Conflict: Its Antecedents and Consequences," Organizational Behavior and Human Performance, Volume 17, Number 1, October 1976, pp [16] Nunnaliy, J.C. Psychometric Theory, Second Edition, McGraw-Hill, New York, New York, [17] Rizzo, J.R., House, R.J. and Lirtzman, S.J. "Role Conflict and Ambiguity in Complex Organizations," Administrative Science Quarterly, Volume 15, Number 2, June 1970, pp [18] Rockart, J.F. "The Changing Role of the Information Systems Executive: A Critical Success Factors Perspective," Sloan Management Review, Volume 24, Number 1, Fall 1982, pp [19] Schuler, R.S., AIdag, R.J. and Brief, A.P. "Role Conflict and Ambiguity: A Scale Analysis," Organizational Behavior and Human Performance, Volume 20, Number 1, October 1977, pp [20] Willoughby, T.C. "Staffing the MIS Function," ACM Computing Surveys, Volume 4, Number 4, December 1972, pp [21] Yunger, G.W. and Hunt, J.G. "An Empirical Comparison of the Michigan Four-Factor and the Ohio State LBDQ Leadership Scales," Organizational Behavior and Human Performance, Volume 17, Number 1, October 1976, pp About the Authors Bavi~f K. (~oldsteit~ is a doctoral candidate in the Information Systems area at the Sloan School of Management at M.L T. He holds a B.Sc. degree in mathematics from McGill University, a S.M. degree in appfied mathematics (computer science) from Harvard University, and a M.E. degree in information science also from Harvard. Mr. Goldstein s research interests are in the systems development process and in information systems personnel. He is currently completing his Ph.D thesis entitled Programmer/Analysts Perceptions of Their Jobs: The Role of Task Differences.,lohn! =. R~ katt, Director, ClSR, and Senior Lecturer of Management Science, Sloan School of Management, M.I.T., has taught and conducted research within the areas of management planning and control systems and the use of computer-based information systems. He presently supervises several ClSR research projects. Prior to joining the Sloan School in 1966, Dr. Rockart spent four years with IBM. His most recent research interests are the "critical success factor" concept, top managerial information use, and the management of end user computing. He is co-author of the book Computers and the Learning Process (with Michael Scott Morton). Among his articles are "Chief Executives Define Their Own Data Needs" and "The CEO Goes On- Line " both of which have appeared in the Harvard Business Review. Dr. Rockart serves on the Board of Directors of three organizations and consults and teaches for several major companies. MIS Quarterly~June