DIMENSIONS AND DIRECTIONS OF PERCEIVED INFLUENCE OVER WORK BEHAVIORS IN RELATION TO HIERARCHICAL LEVEL AND JOB SATISFACTION DISSERTATION FOR THE DEGREE CF'PHD. MICHIGAN STATE UNIVERSITY FRED FALLIK 1977 T“! w‘ iii-:1 r- -Ir--—— 4'- 0; lflllllllllllllfllllllljllljlll“Milli ~ ‘.W§' Y-v-‘l ‘Ix'rr-r. This is to certify that the thesis entitled Dimensions and Directions of Perceived Influence Over Work Behaviors in Relation to Hierarchical Level and Job Satisfaction presented by Fred Fallik has been accepted towards fulfillment of the requirements for 2’ w ‘ r-rr-H 1' ' I ' t “(I4 ’v‘ —' fl . (/\,~/M,/Z)—L- ,/ Maicr professor Date July 29, 1977 0-7639 T ABSTRACT DIMENSIONS AND DIRECTIONS OF PERCEIVED INFLUENCE OVER WORK BEHAVIORS IN RELATION TO HIERARCHICAL LEVEL AND JOB SATISFACTION BY Fred Fallik This study investigated some specific aspects of the traditional as compared with the contemporary approaches to organizational influence over work behaviors. The traditional approach to organizational influence assumes that influence is based on rational and legal authority and is distributed in a hierarchical fashion. More contemporary approaches assume that influence is primarily a psychological phenomenon having little to do with an individual's hierarchical position. Two widely accepted propositions, relevant to both the traditional and contemporary approaches, were investigated: 1. Control or influence over work behaviors tends to increase with successive levels in the organiza- tional hierarchy. Fred Fallik 2. The more an individual influences his own or others' work behaviors, the greater will be his own job satisfaction. This study attempted to measure and evaluate aspects of these two broad propositions relevant to the two sets of assumptions mentioned above. In particular, three independent and two dependent variables were defined. The first independent variable was Hierarchical Level determined by measuring each individual's relative position within the organizational structure. The second independent variable was Direction of Perceived Influence, defined as the degree of perceived influence exerted 23 the respondent compared to the degree of perceived influ- ence exerted by the respondent. The third independent variable was Dimensions of Perceived Influence, defined as the degree of perceived influence exerted through a specific dimension of influence. Ten separate dimen- sions of perceived influence were measured in this study: co-workers, equipment, rules and regulations, boss, sub- ordinates, other departments, customers, physical environ- ment, goals and standards, and the respondent himself. The two dependent variables were overall Degree of Perceived Influence as measured on a five point Likert Scale and Job Satisfaction measured by the Brayfield- Rothe Job Satisfaction Index. The data for the study were collected from two large commercial banking organizations located in the Fred Fallik same geographical area. Volunteer subjects were sampled from three levels of the organizational hierarchy: rank and file, supervisors, and middle management. On con- secutive days, 220 subjects from the first bank and 99 subjects from the second bank were administered the Per- ceived Influence questionnaire developed for this study and the Brayfield-Rothe Job Satisfaction Index. Responses to these questionnaires were compared both across organizational levels within each bank and between levels across banks. The results indicated some similarities and differences. A consistent response across all levels and banks was that individuals tended to consider themselves as the primary determinant of their own work behavior. Another consistent finding was that, as a whole, there was no significant difference between influence 22 respondents compared to influence by respondents on others' work behaviors. In most hierarchical levels respondents tended to judge rules and departmental goals and standards as exerting a significant degree of influ- ence over their work behaviors. Co-workers were gener- ally judged to have little influence over the respond- ent's work behavior. In contrast, respondents generally judged themselves as having a significant degree of influence over their departmental goals and standards but with little influence over their boss' work behaviors. Fred Fallik In terms of job satisfaction, managerial level respondents from both organizations exhibited significant correlations between specific dimensions of influence and job satisfaction as did the rank and file workers in one of the two banks but not in the other. The results of this study were interpreted in terms of both the traditional and contemporary approaches to organizational influence. It was concluded that the study tended to support aspects of both approaches. The implications of the study were discussed in relation to organizational theory, practice, and future research. DIMENSIONS AND DIRECTIONS OF PERCEIVED INFLUENCE OVER WORK BEHAVIORS IN RELATION TO HIERARCHICAL LEVEL AND JOB SATISFACTION BY Fred Fallik A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Psychology 1977 List of List of Chapter I. II. III. IV. Table of Contents Introduction and Literature Review . . Relationship Between Influence Hierarchical Level . . . . . . Relationship Between Influence Job Satisfaction . . . . . . . Statement . . . . . . . . . . Methodology . . . . . . . . . . . . . Measurement of the Variables . . . Hierarchical Level . . . . . . Job Satisfaction . . . . . . . Influence Over Work Behaviors a. Influence Questionnaire - Validity Assessment . . . b. Influence Questionnaire - Reliability Assessment . . Sampling Procedures and Data , Collection............ Tables Figures A. The and B. The and Problem A. 1. 2. 3. B. Results A. Analysis Methods and Results Pertain- ing 1. to Issues One to Four . . . . Perceived Influence, Hierarchical Results (Alpha) . . . . . . . ii iv vii 14 21 24 24 24 25 26 29 45 47 S7 64 66 iii B. Analysis Methods and Results Pertain- ing to Issues Five and Six . . . . . . 88 V. Conclusions and Implications . . . . . . . 98 A. Implications for Organizational Theory 0 O O O O O O O O O O O O O O O 100 B. Implications for Organizational praCti-Ce O O O O O O O O O O O O O O O 104 C. Implications for Future Research . . . 105 Appendix - Research Questionnaire . . . . . . . . 108 References . . . . . . . . . . . . . . . . . . . . 117 Table 9. 10. 11. 12. 13. 14. 15. 16. 17. List of Tables Item Correlation Matrix for Part A (Alpha) Item Correlation Matrix for Part A (Beta) Item Correlation Matrix for Part 8 (Alpha) Item Correlation Matrix for Part B (Beta) Factor Loadings, Eigenvalues, and Percent- ages for Part A (Alpha) . . . . . . . . . Factor Loadings, Eigenvalues, and Percent- ages for Part A (Beta) . . . . . . . . . . Factor Loadings, Eigenvalues, and Percent- ages for Part 8 (Alpha) . . . . . . . . . Factor Loadings, Eigenvalues, and Percent- ages for Part B (Beta) . . . . . . . . . . Item Pair Correlations (Alpha) . . . . . . Item Pair Correlations (Beta) . . . . . . Factor Loadings, Eigenvalues, and Percent- ages for Item Pairs (Alpha) . . . . . . . Factor Loadings, Eigenvalues, and Percent- ages for Item Pairs (Beta) . . . . . . . . Coefficient Alpha for Parts A & B and Total (Alpha and Beta) . . . . . . . . . . Item Correlations by Pairs (Alpha) . . . . Item Correlations by Pairs (Beta) . . . . Comparison of Responses: Alpha and Beta . Response Means by Level and Organization . iv 31 31 32 32 33 34 38 39 42 42 43 44 48 48 48 54 S9 Table 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. Number of Cases by Item and Organization . Analysis of Variance - Bank Alpha . . . . Analysis of Variance - Bank Beta . . . . . Analysis of Variance for Directions and Items (Rank and File Workers - Alpha) . . Analysis of Variance for Directions and Items (Supervisors - Alpha) . . . . . . . Analysis of Variance for Directions and Items (Management - Alpha) . . . . . . . . Pairwise Comparison of Item Means based on Rank and File Workers' Responses (Part A - Alpha) 0 C O O O C O O C O O O O C C O C . Pairwise Comparison of Item Means based on Supervisors' Responses (Part A - Alpha) . Pairwise Comparison of Item Means based on Rank and File Workers' Responses (Part B - Alpha) 0 O O O O O O O O O O O O O O O O O Pairwise Comparison of Item Means based on Supervisors' Responses (Part B - Alpha) . Rank Order Correlations Comparing Organiza- tional Levels on Part A and Part B of the Questionnaire (Alpha) . . . . . . . . . . Analysis of Variance for Directions and Items (Rank and File Level - Beta) . . . . Analysis of Variance for Directions and Items (Supervisory Level - Beta) . . . . . Analysis of Variance for Directions and Items (Supervisory Level - Beta) . . . . . Pairwise Comparison of Item Means based on Rank and File Responses to Part A (Beta) . Pairwise Comparison of Item Means based on Supervisors' Responses to Part A (Beta) . Pairwise Comparison of Item Means based on Managers' Responses to Part A (Beta) . . . 6O 67 68 7O 7O 71 73 73 74 74 75 78 78 79 81 81 82 Table 35. 36. 37. 38. 39. 40. 41. 42. 43. vi Pairwise Comparison of Item Means based on Rank and File Responses to Part B (Beta) . . Pairwise Comparison of Item Means based on Supervisors' Responses to Part B (Beta) . . Pairwise Comparison of Item Means based on Managers' Responses to Part B (Beta) . . . . Rank Order Correlations Comparing Organiza- tional Levels on Part A and Part B (Beta) . Rank Order Correlations for all Levels in AlphaandBeta............... Multiple Regression of Items with Job Satis- faction (Managerial Level - Alpha) . . . . . Multiple Regression of Items with Job Satis- faction (Rank and File Level - Beta) . . . . Multiple Regression of Items with Job Satis- faction (Management Level - Beta) . . . . . Multiple Regression of Item Pairs with Job Satisfaction (Management Level - Beta) . . . 82 83 83 84 87 9O 92 94 94 List of Figures Figure 1. Average Response to Influence Questionnaire Items by Level (Alpha) . . . . . . . . . . . 62 Average Response to Influence Questionnaire Items by Level (Beta) . . . . . . . . . . . 63 vii Chapter One Introduction and Literature Review I. Introduction Almost by definition, the organization of human beings into formal collectives implies control over their behaviors. Tannenbaum (1968) states that to speak of organizations is to speak of control. The universal occurrence of organizational control has provided a topic of common interest to a variety of disciplines. Tannenbaum (1957) cites three reasons for this widespread concern and interest in the topic of control and influence over human behavior: 1. Control over human behavior has broad social and political implications. 2. Control over human behavior has implications for the satisfaction of human needs. 3. Control over human behavior has implications for organizational functioning. Many authors appear to agree with Tannenbaum with respect to the importance of control or influence in relation to organizational structure, job satisfaction and productivity. Despite agreement on its importance, however, there remain many issues concerning control and influence over work behaviors. By what methods 1 2 should influence be exerted? How much should be exerted? Who should exert it? What are the behavioral and atti- tudinal effects of control? Different generations of authors have, of course, attempted to answer these questions in a manner consist- ent with their discipline and culture. In general, two distinct approaches to organizational influence have been evidenced. The first approach to organizational control can be traced back to the writings of Max Weber. Influence over human behavior, according to this traditional approach, is distributed in a hierarchical fashion-- increasing with level or rank in the hierarchy. The distribution of influence is based on legal and rational values prescribing the influence relationship among organizational members. The second conceptualization of organizational influence can be traced back to the writings of Chester Barnard. In contrast to the traditional approach, neither legitimacy nor hierarchy plays a particularly central role in the influence process. Influence, according to this contemporary approach, is primarily a result of a complex of interactions among individuals, each in terms of his own self-interest. Of concern to both conceptualizations of organ- izational influence are two interrelated propositions: 3 1. That control or influence over work behaviors tends to increase with successive levels in the organiza- tional hierarchy. 2. That the more an individual influences his own or others' work behaviors, the greater will be his job satisfaction. Although these propositions have attracted much interest, little empirical evidence exists as to their validity as assessed through the perceptions of organi— zation members. The following review of literature will examine these two propositions in some detail and suggest alternative conceptualizations. II. Literature Review This section will be concerned with the relation- ships among three variables implicit in the two proposi- tions above: influence, hierarchical level, and job satisfaction. Taken together, these propositions sug- gest that influence over work behaviors is related to one's hierarchical level, and influence, in turn, relates to one's job satisfaction. A. The Relationship between Influence and Hierarchical Level Probably one of the most widely noted aspects of organizational behavior is that some individuals have more control or influence over their own and others' work behaviors than do other individuals. Differences in 4 the ability to control one's own work behaviors ("work autonomy" Dubin, st 21., 1976) and the work behaviors of others have traditionally been attributed to differences in formal authority. This traditional conception of influence asserts that, since formal authority tends to increase with successive levels in the organizational hierarchy, so must influence. Those at the top of the organizational hierarchy have more control or influence over their own and others' work behaviors than do those individuals at lower levels. Tannenbaum, st 31., (1974) maintains this position clearly: "First, authority is distributed hierarchically in organizations. Individuals at upper levels have more power and exercise more control than those at successively lower levels" (p. 7). This traditional conception of the relationship of authority and influence is common to many classical approaches to organizational behavior (Cartwright, 1965; Gibson, 33.21., 1973; Rosen and Weaver, 1972). As typi- cally defined, power is the latent or potential ability to guide, direct, or influence work behaviors (Dalton, ‘gt‘gl., 1968). Influence is any successful direction or alteration in behavior. The logic underlying the traditional approach to influence appears to be that power is distributed hierarchically in terms of formal authority and formal authority, in turn, results in influence over work behaviors. Formal authority, the legitimate right to direct others, is thus generally equated with influence over the work behaviors of others. 5 While few would deny that formal authority tends to increase with the hierarchical level of organization members, several have questioned whether formal authority and influence over others' work behaviors increase pro- portionally. In general, critics of the traditional approach to influence, those who advocate a contemporary approach to organizational influence, have stressed the role of individual perception and multiple determinants. Hackman (1969), for instance, claims that the "objective" properties of a task affect behavior and attitude inasmuch as they are perceived by individuals. It is this perception and subjective redefinition which primarily determine work behavior and attitude rather than the "objective" aspects of the task. Many contemporary authors appear to agree with this position (Argyris, 1957; Korman, 1971; Porter, st 31., 1975; Stagner, 1956). Melcher (1976) cautions that the objective structure of organizational authority is often at variance with the day to day process of decision making. To assess influ- ence, he notes, researchers must measure individual per- ceptions and not rely on organizational charts. One of the few empirical attempts to measure perceived influence has been conducted by Tannenbaum (1968). In a wide variety of organizational settings, Tannenbaum attempted to investigate the perceived dis- tribution of organizational influence in relation to hierarchical level. Using survey research techniques, 6 individuals at successively higher levels in the organi- zation were asked to indicate the extent to which those above them determined what "went on" in the respondent's work group. By averaging responses within each organi- zational level and comparing across levels, Tannenbaum claims to have developed an accurate and realistic description of organizational influence. By means of this technique, "control" graphs of organizational influence were developed depicting the average amount of perceived influence in relation to hierarchical level. This control graph permitted both intra- and interorganizational comparisons. According to Tannenbaum, organizations which showed little difference in average influence among levels were termed "demo- cratic." Those organizations which showed large differ— ences in average perceived influence among levels were classified as being relatively "autocratic." Tannenbaum's approach to measuring influence in terms of individual perceptions is probably more realistic than the common traditional approach of measuring influ- ence in terms of formal authority ascribed to individ- uals' hierarchical level. In contrast, however, some methodological issues may be raised which cast some doubt on his conclusions. In particular, the methods employed by Tannenbaum appear questionable for the following reasons: 1. The referent work behaviors judged by respondents were those of the respondent's work group, not those 7 of the respondent. Tannenbaum required subjects to indicate how much influence successive levels in the organizational hierarchy had over the respondent's work group. Judgments made by respondents as to perceived influence might be more easily defined if the focus of judgment was on the respondent's own work behaviors rather than the comparatively amor- phous and ill-defined entity of the "work group's" behaviors. Subjects were required to judge the influence exerted by successively more distant levels in the organiza- tion hierarchy. Respondents at the lowest levels of the organizational hierarchy were asked to judge the amount of influence exerted by individuals at each level above them. Successive levels were required to judge the influence of those above them. A study by Rothaus, gt 21. (1965) demonstrated that role familiarity affected rating validity. Those at the lower levels in the hierarchy were probably less familiar with the role of higher levels than were those at these higher levels. Litterer (1973), in addition, points out that the level one occupies in the organizational hierarchy frequently biases his perception of others' role in the organization. It would appear likely that requiring respondents to judge the influence of successively more distant and abstract levels is a tenuous technique at best. 8 Ratings errors could possibly be minimized and held constant by limiting respondent's judgment to those hierarchical levels immediately adjacent to his. Perceived influence was measured in one direction only, that of superiors on subordinates. Tannenbaum made no attempt to measure the perceived influence of the respondents on others. While this important distinction in direction of influence has been made repeatedly (McCormick and Tiffin, 1975; McGregor, 1960; Pelz, 1952), no attempt has been made, so far as can be determined, to measure the relationship between perceived influence from others and perceived influence on others. Mechanic (1962) suggests that influence can be exerted by and on members at all levels of the organizational hierarchy. Measuring the relationship between these two directions of perceived influence might add an important distinc- tion to the concept of control ignored in Tannenbaum's studies. Perceived influence was measured in terms of one dimension only, that of formal authority. Although Tannenbaum (1960) defines control or influence as "any process in which a person (or group of persons, or organization of persons) determines or intention- ally affects what another person (or group or organ- ization) will do," his measurement methodology may not reflect this definition. By asking respondents 9 to judge influence exerted by successive levels in the organization hierarchy, influence was probably judged in terms of formal authority. Although formal authority may represent an important dimension or mode of potential influence, a wealth of evidence indicates that it is but one of many dimensions through which work behavior may be influenced. That control or influence over work behaviors may be exerted through many dimensions has been noted by a variety of authorities. Cartwright (1965), for instance, points out that "behavior down the line" may be influenced by many factors in addition to formal directives. Several authors have attempted to list the dimen- sions that they believe are most salient in the deter— mination of work behaviors. Dalton and Lawrence (1972), Argyris (1964), and Seiler (1957) all identify three primary dimensions of influence: 1) formal organiza- tional policies and rules. 2) informal group norms and social influences. 3) individual self-control, values and attitudes. Other authors have cited additional dimensions (Davis, 1972; Leavitt, 1976). As a whole, it would appear that many authorities consider influence to be a multidimensional process in contrast to the uni- dimensional approach taken by Tannenbaum (1968). Based on a review of the literature and dis- cussions with a wide variety of organizational members 10 several dimensions are frequently noted as important factors in the influence process: 1. 2. 3. 4. 5. 6. 7. 8. 10. 11. 12. the co-workers or peer group. the nature of the technology, equipment or machinery. the formal rules, regulations, or procedures. the immediate boss or supervisor. the customers or clients (if any). other departments or sections. subordinates (if any). the physical work environment (lighting, noise, etc.). the personnel practices and performance appraisal methods. one or more unions in which the focal person is a member. one or more unions in which the focal person is not a member. the individual himself. In various combinations, these dimensions have received considerable attention in the organizational literature. So far as can be determined, no systematic attempt has been made to measure the relative importance of each of these dimensions in influencing work behaviors and attitudes. If investigated at all, each dimension has generally been treated separately and in isolation without consideration to relative influence or inter- action. 11 Unlike the distribution of formal authority, it probably cannot be assumed that the influence of each of these dimensions will tend to increase with hierarchical level. Argyris (1972), in fact, suggests that the "potency" of various dimensions will differ in relation to specific hierarchical levels. With reference to four of the cited dimensions, Argyris maintains that struc- tural dimensions (roles), technology, and administrative controls may be more important at lower levels of the organization whereas leadership may be more salient at upper levels. Ritchie (1974) suggests that it is the perception and attitude of organization members toward influence exerted from a multitude of dimensions that determines work behavior rather than the "objective" degree of influence. The discussion up to this point has attempted to examine the first of the two propositions noted at the beginning of this chapter, namely, that influence over work behaviors tends to increase with the hierarchical level of organization members. In general, two distinct approaches can be identified which are addressed to this proposition. The traditional approach suggests that successive levels in the organizational hierarchy exert more influence than do those below them because they have more formal authority. In contrast, the contemporary approach suggests that influence may have little to do with hierarchical level or formal authority and is 12 primarily a function of personal perceptions and self- interest. Tannenbaum, although identified with the con- temporary approach through his emphasis on perceived rather than objective definition of influence, measures a rather limited and methodologically weak "influence." In particular, four aspects of Tannenbaum's research were questioned: 1. The emphasis on the group behavior as the referent rather than the respondent's work behavior. 2. Requiring respondents to generalize over several hierarchical levels rather than those immediately adjacent to them. 3. Identifying influence only in terms of one direction, namely, that of others on the respondent's work group. 4. Identifying influence only in terms of one dimension, that of formal authority. One major purpose of this study was to take a contemporary approach to the relationship between influ- ence and hierarchical level by expanding on the defini- tion of influence suggested by Tannenbaum. In particular, three concepts related to the influence process will be measured in this study with regard to the contemporary approach to influence: 1. Directional influence - the degree of perceived influence exerted from others in the organization compared to the degree of perceived influence exerted 13 by the respondent on others. Directional influence will be measured without regard to respondent's hierarchical level or dimension of influence. Dimensional influence - the degree of perceived influence exerted by a specific mode or method com- pared to all other modes or methods. Dimensional influence will be considered without regard to respondent's hierarchical level or direction of influence. Hierarchical influence - the average degree of per- ceived influence reported for individuals at a specific hierarchical level in the organizational structure compared to all other levels. Hierarchical influence will be considered without regard to specific directions or dimensions of perceived influence. The identification of these three aspects may lead to a broader conceptual and methodological approach to the measurement of perceived influence. Since, it has been argued, previous approaches to organizational influ- ence may have been limited, a number of issues regarding the relationship between influence and hierarchical level may now be approached: 1. Is directional influence a relevant concept? Is there a measureable difference between perceived influence exerted on respondents and degree of influ- ence exerted by respondents on others? B. 14 Is dimensional influence a relevant concept? Is there a measureable difference among different dimensions of influence in terms of perceived influ— ence? Is hierarchical influence a relevant concept? Is there a measureable difference between respondents at different hierarchical levels in terms of per- ceived influence? Is perceived influence contingent upon specific directions of influence, dimensions of influence, and/or hierarchical levels? The Relationship between Influence and Job Satis- faction. The second proposition noted at the beginning of this chapter dealt with the relationship between influence over work behaviors and job satisfaction: job satisfac- tion tends to increase with influence. This notion has found widespread support from a number of prominent authorities (McGregor, 1960; Likert, 1967; O'Connell and Cummings, 1973; Tannenbaum, 1968). Blauner (1964) per- haps best states the logic underlying this proposition: "When work provides an opportunity for control, meaning and self-expression, it becomes an end in itself rather than a means to live." As was the case in the previous section, the relationship between job satisfaction and influence has often been defined in terms of formal authority. The 15 consistent finding that job satisfaction tends to increase with hierarchical level (Herzberg, st 21., 1957) has frequently been cited as empirical support for the effects of formal authority on job satisfaction. It is generally reasoned that increases in formal authority with hierarchical level promote increased job satisfac- tion. Probably the clearest statement of this position is by Argyris (1964). Argyris maintains that the hierarchical distribu- tion of authority results in an inherently unequal dis- tribution of influence. Individuals at the lower ends of the organizational hierarchy have relatively little "say" over their work behaviors while those at the higher levels have a great deal of influence. For the mature individual, this "powerlessness" may have negative con- sequences in terms of job satisfaction. In general, Argyris maintains, the greater the degree of control exerted by management, the less likely it is that sub- ordinates would be satisfied with their jobs. This conceptualization of job satisfaction and influence, while currently popular, has not gone unchallenged. Vroom (1964), commenting on a number of studies which reported increasing job satisfaction with hier- archical level, questions the cause. Many things, he notes, increase with hierarchical level: status, pay, type of tasks, etc. The relationship between each of these variables and job satisfaction will be unclear, he 16 claims, until further research is conducted. Vroom's cautious approach may be justified since a great deal of indirect evidence suggests that formal authority may have little relationship to job satisfaction. Indirect evidence bearing on the relationship between formal authority and job satisfaction has, for the most part, come from "field" studies. The basic premise underlying these studies appears to be that job satisfaction should tend to increase or decrease with respect to changes in degree of formal authority. Although many studies do indicate increased job satis- faction with increased authority, literature reviews by Vroom and Yetton (1974), Hulin (1971), and Hulin and Blood (1968) indicate that this increasein job satisfac- tion does not always occur. Increases in formal authority are not necessarily accompanied by increased job satis- faction. In a recent study by Tannenbaum, st 31. (1974) an attempt was made to measure the relationship between job satisfaction and perceived influence. Using a variety of organizations from both socialist and capitalist countries, Tannenbaum reasoned that socialist managed organizations should show a more egalitarian distribution of influence than would capitalist managed organizations. The results supported his contention; on the average, socialist organizations did display a characteristically "flat" distribution of influence compared to capitalist 17 organizations. Despite the differences in control pat- terns between the two types of management systems, how- ever, no differences in job satisfaction were found. Two interpretations of this last result might be made. First, if Tannenbaum did measure perceived con- trol as he claims, then his results might be interpreted as indicating that there exists little relationship be- tween how much "say" an individual has and his/her job satisfaction. Second, if Tannenbaum measured only a limited aspect of control (formal authority), then it is possible that only this dimension has little to do with job satisfaction. A related study by Dalton, Barnes, and Zaleznik (1968) provides some support for this second interpretation. Dalton, st 31. attempted to measure change in perceived work autonomy and job satisfaction as a result of change in the formal authority structure. Nampa, a large engineering company, was restructured to eliminate one level of top management. The authority and responsi- bilities of this eliminated level were assigned to first and second level managers. This group of managers who had had their formal authority increased were termed the CAPS (Coordinated Action Plan) group. A control group of first and second level managers remained unaffected throughout the organizational restructuring program. In a retrospective appraisal, roughly 33% of the experimental (CAPS) group reported an increase in work l8 autonomy compared to a 9% increase in the control (non- CAPS) group. In terms of job satisfaction, roughly 44% of the CAPS group reported increased job satisfaction compared to 26% of the non-CAPS group. Since these per- centages were based on a retrospective judgment by organization members, some doubt may be cast on their interpretation. Certainly, a large percent of the exper- imental group reported little changein work autonomy and job satisfaction although the tendency toward increased autonomy and satisfaction was evident. No indication was made in the study as to the relationship between increased autonomy and job satisfaction. It is possible that those individuals in the CAPS group who reported increased autonomy were not those who showed an increase in job satisfaction. Clearly, the relationship between formal authority and job satisfaction was not unquestion- ably demonstrated in this study. One possible explanation of these results found by Tannenbaum (1974) and Dalton, gt _1. (1968) concerns the previous discussion on influence and hierarchical level. If formal authority were to be considered but one dimension of the influence process, it might be that other dimensions could be more salient in affecting job satisfaction. In examining the relationship between dis- tribution of formal authority and job satisfaction, neither Tannenbaum nor Dalton considered the multidirec- tional or multidimensional possibility of perceived l9 influence over work behaviors. With consideration given primarily to the role of formal authority in job satis- faction, both Tannenbaum and Dalton appear to adhere to the traditional approach to organizational influence previously mentioned. So far as can be determined, no attempts have been made to measure the multidimensional and multidirectional approach to organizational influence with job satisfaction. To do so would appear to be more consistent with the contemporary approach to influence. This section of the literature review has attempt- ed to examine the relevant evidence pertaining to the second proposition noted at the beginning of the chapter; job satisfaction tends to increase with increasing influ— ence over work behaviors. Aside from the methodological problems of measuring perceived influence, it was noted that by defining perceived influence in terms of formal authority little evidence exists to directly support this proposition. The contemporary conception of the relationship between multidirectional and multidimensional influence and job satisfaction is, as yet, unmeasured. Taking this contemporary conception of influence into account, two issues regarding the relationship between influence over work behaviors and job satisfaction call for empirical investigation: 1. Is job satisfaction significantly related to the degree of perceived influence resulting from any one or combination of dimensions or directions of per- ceived influence? 20 2. Does the relationship, if any, between dimensions and/or directions of perceived influence vary with hierarchical level? The purpose of this review was to examine the current state of knowledge regarding the relationship between hierarchical level and influence on the one hand and influence and job satisfaction on the other. It was found that, from both a conceptual and methodological point of view, limitations and ambiguities abound. For the most part, the traditional conceptualization of influence in relationship to hierarchical level and job satisfaction has received little empirical support. So far as can be determined, little evidence exists as to the validity of the contemporary approach and definition of influence. It was the purpose of this study to investigate some aspects of this second approach. Chapter Two Problem Statement It was the purpose of this study to measure cer- tain of the relationships discussed in the literature review. More specifically, two related sets of proposi- tions were investigated. First, an attempt was made to measure the relationship between the hierarchical level of respondents and degree of perceived influence over work behaviors. As cited in Chapter One, the definition of perceived influence was expanded to include the con- cepts of hierarchical influence, directional influence, and dimensional influence. Four issues were posed in the previous chapter based on this expanded definition: 1. Is hierarchical influence a significant factor? Do individuals at different hierarchical levels differ significantly from each other in terms of perceived influence over work behaviors? 2. Is dimensional influence a significant factor? Do the ten dimensions of influence differ significantly from each other in terms of perceived influence over work behaviors? 3. Is directional influence a significant factor? Is influence from others significantly different from influence on the work behaviors of others? 21 22 4. Is the degree of perceived influence contingent upon specific hierarchical levels, dimensions of influence, and/or directions of influence? The second set of issues this study attempted to answer concerned the relationship between the expanded definition of influence over work behaviors and job satis- faction. Two questions, in particular, were posed regarding this relationship: 1. Is job satisfaction significantly related to the de- gree of perceived influence resulting from any one or combinations of dimensions or directions of per- ceived influence? 2. Does the relationship, if any, between dimensions and/or directions of perceived influence vary with hierarchical level of respondents? As presented, these six questions attempted to take into consideration the contemporary approach to organizational influence. Both in terms of conceptual orientation and methodological considerations, this study attempted to circumvent limitations evidenced in previous studies of this topic. Conceptually, an attempt was made to define influence in terms of a variety of aspects previously omitted. Ten separate dimensions were defined through which influence might be exercised. In addition, two directions (influence from others and influence on others) were defined. 23 In terms of methodology, an attempt was made in this study to avoid some of the limitations noted in previous studies. First, an attempt was made to obtain a cross-section of organization members at most levels in the hierarchy. Outside of Tannenbaum's work, few studies have attempted to measure several levels in the hierarchy but have confined their attention to specific levels. Second, two similar but independent work organ- izations in the same industry were used in this study to provide some evidence on reliability and generalizability of results. Third, subjects were required to judge influence only in terms of their own current work behav- iors and those of the immediately adjacent levels in the organizational hierarchy. This methodology contrasts to the techniques employed by Tannenbaum and Dalton. In sum, this study investigated the possibility of an expanded definition of organizational influence. In doing so, an attempt was made to circumvent some of the methodological limitations of previous studies. Chapter Three Methodology A. Measurement of the Variables l. Hierarchical Level Both of the independent organizations in this study were primarily concerned with banking and financial operations. Both banks (denoted Alpha and Beta) are located in the same geographical area and engaged in similar business operations. For both organizations, only the "headquarters" personnel were involved in the study (N for Alpha = 220, N for Beta = 99). The hierarchical level of organization members was determined by asking each respondent to indicate his/her job title and work section or department. After data collection was accomplished, this researcher then consulted with the directors of personnel at each of the two organizations used in this study. Using the hier- archical definitions and characteristics proposed by Porter, Lawler, and Hackman (1975, p. 91) the personnel directors classified job titles and sections in terms of hierarchical level. As defined by Porter, 35 31. the definition and characteristics of each hierarchical level are: 24 25 Level Four: Top management; concerned with overall goal formation and policy decisions regarding allocation of resources. Level Three: Middle management; concerned with sub-goal formation and plans for implementing deci- sions from above and coordinating activities from below. Level Two: Lower management; concerned with implement- ing decisions made at higher levels and coordinating and directing the work of employees at the lowest level of the organ- ization. Level One: Rank and file; concerned with carrying out specific task activities. Since these definitions were used in both organ- izations, a common criterion for measurement of the hier- archy variable was established. This procedure permitted comparisons across organizations despite their initial differences in number of hierarchical levels. 2. Job Satisfaction A second variable measured in this study was job satisfaction, the affective attitude of organization members toward their work (Bass and Barrett, 1972). Measurement of job satisfaction was accomplished through the use of the Brayfield-Rothe Job Satisfaction Index. A review of this instrument by Robinson, Athanasiou, and Head (1974) indicated that it had a high degree of 26 internal consistency (r = .87) and good discrimination ability between known job satisfied and job dissatisfied groups. An additional advantage was that it was self- administered and could be administered in a brief period of time (10 minutes). 3. Influence over Work Behaviors Measurement of perceived influence over work behaviors required the development of a new instrument for the purpose of this study. To measure the twelve dimensions outlined in the review of literature, two groups of ten questions were developed. Omitted from this study were the dimensions concerned with unions. The deletion of the two union oriented questions was at the direct request of the participating organizations. Although neither bank had, at the time of this study, an active union the personnel directors believed that by asking union oriented questions, a degree of anxiety on the part of organization members might be created. Degree of perceived influence, as a result, was measured through ten of the original twelve dimensions. The first group of ten questions attempted to measure the degree to which organization members per- ceived various independent dimensions as influencing their work behaviors. The second group of ten questions attempted to measure perceived influence of the respondent over the work behavior and activity of others. The referent dimensions in both the first and second set of 27 influence questions were identical; the direction of influence varied. The influence questionnaire was thus divided into two parts: the degree of perceived influ— ence from others (Part A) and the degree of perceived influence on others (Part B). As previously mentioned, this dichotomy was suggested by McCormick and Tiffin (1975). Each of the ten dimensions of influence used in this study was measured by a single questionnaire item. This was done following the conventions set by previous authors who attempted to measure perceived influence (Tannenbaum, 1968; Lawrence and Lorsch, 1967; Dalton, Barnes, and Zaleznik, 1968). Patchen (1962) compared two different approaches to measuring influence-~one approach used a specific decisional situation and the second approach used a global index. Both methods yielded sim- ilar results. The question format used in this study was based on that used by Tannenbaum (1968). More specifically, respondents were asked to indicate, in terms of a Likert format, how much influence a given dimension had over their work behaviors (part A) or how much influence the respondent had over the work behavior or activity of others (part B). Several trial drafts of the instrument were devised and pretested twice with subsequent revisions to dimensions, item stem, response format, and directions. During the summer of 1975 a preliminary version of the 28 questionnaire was administered to a group of 100 organi— zation members from a variety of companies as part of a class project. Two weeks after the initial administra- tion, a second administration of the same group of respondents was accomplished. Test-retest reliability proved to be fairly high (rtt = +.82). Several problems in directions and item stems required a second adminis- tration of the questionnaire to a group of 32 air force officers. All subjects in this second pretest were stu- dents in a seminar conducted by this researcher. After completing this second pretest, all subjects were asked to indicate ambiguities in wording, concepts, directions, response format, and to suggest additional dimensions of influence which influenced their work behaviors but were not cited in the questionnaire. The suggestions made by this group were incorporated in the version of the ques- tionnaire as used in this study. The Appendix displays the influence questionnaire used in this study and the version of the Job Satisfaction Index which was used in this study. In addition to the content and format considera- tions required of a newly developed instrument, attention must also be given to its psychometric properties. In the following two sections attempts were made to measure the validity and reliability of the influence question— naire. 29 a. Influence Questionnaire - Validity Assessment An important issue in the development of this questionnaire was the extent to which it measured, for the samples used in this study, a relatively homogeneous construct "degree of perceived influence" or a hetero- geneous collection of independent items. Although there are several methods for estimating construct validity, one of the most common and useful techniques is to meas- ure the extent to which factor structure is similar under two independent administrations (Nunnally, 1967). To accomplish this four sets of data were derived. Responses to the twenty items comprising the influence questionnaire were subdivided into two parts. Part A was Comprised of the responses to the first ten items ("influence from others") and Part B was comprised of the last ten items ("influence on others"). Parts A and B were first analyzed separately for each of the two organizations involved in this study thus yielding four independent analyses. Subsequent analyses were conducted based on item pairs, combining parallel items (questions 1 and 11, questions 2 and 12, etc.) for both organiza- tions. ' All four data sets-~0rganization Alpha Part A, Organization Beta Part A, Organization Alpha Part B, Organization Beta Part B--were factor analyzed using a variety of rotational techniques. More specifically, varimax, quartimax, and oblique rotations were conducted. 30 For the most part, these different rotational solutions yielded similar results in terms of factor loadings and eigenvalues. Displayed in Tablesl, 2, 3, and 4 are the correlation matrices upon which rotational procedures were accomplished. Tables 1 and 2, representing item correlations for Alpha and Beta on the first ten items, appear to be, in general "flatter" than responses to Part B for both organizations. Similarly, in comparing Tables 1 and 3 (Alpha) with Tables 2 and 4 (Beta), corre- lations for Alpha appear to be somewhat lower than those for Beta. Due to similarity in results for different rota- tional procedures, only the factor analyses using varimax rotation are displayed for each part of the questionnaire and organization. Analysis of Part A of the question- naire for Alpha and Beta will be attempted first followed by interpretations of Part B analysis for Alpha and Beta. These factor analyses and their pertinent statistics are presented in Tables 5 through 8. Analysis of responses to Part A, Tables 5 and 6, show that rotation to simple structure produced four factors. For both Alpha and Beta, factor one accounted for almost 50% of item variance. Factors two, three, and four account for parallel and decreasing percentages of variance in both organizations. The item to factor loadings in both banks appear to be largely parallel. In both organizations, items 1, 2, 3, 7, and 8 have their 31 Table 1 Item Correlation Matrix for Part A (Alpha). 11 12 I3 I4 I5 I6 I7 I8 I9 I10 Total 11 1.00 .093 .284 .175 .340 .193 .160 .059 .178 .103 .492 12 1.00 .185 .021 .260 .135 .036 .213 .098 .016 .357 I3 1.00 .202 .111 .328 .110 .191 .305 .081 .578 I4 1.00 -.04 -.03 .129 .173 .165 .035 .344 IS 1.00 .019 .102 .128 .257 .035 .508 I6 1.00 .244 .047 .093 .021 .460 I7 1.00 .138 .173 .127 .498 I8 1.00 .248 .033 .426 I9 1.00 .180 .543 110 1.00 .351 Table 2 Item Correlation Matrix for Part A (Beta). I1 I2 I3 I4 I5 I6 I7 I8 I9 I10 Total I1 1.00 .258 .360 .413 .322 .240 .325 .157 .186 .039 .599 I2 1.00 .355 .317 .026 .255 .190 -.11 -.01 -.00 .496 I3 1.00 .434 .127 .464 .383 .055 .121 —.26 .555 I4 1.00 .418 .404 .184 .140 .103 -.09 .594 I5 1.00 .368 .148 .381 .250 .133 .536 I6 1.00 .365 .008 .170 .003 .610 I7 1.00 .405 .115 .110 .566 I8 1.00 .196 .038 .384 I9 1.00 .358 .405 I10 1.00 .125 32 Table 3 Item Correlation Matrix for Part B (Alpha). 111 112 113 114 115 116 117 118 119 120 Total 111 1.00 .072 .421 .455 .704 .410 .265 .217 .449 .656 .725 112 1.00 .315 .141 .106 .198 .017 .204 .125 .083 .331 113 1.00 .399 .377 .302 .082 .282 .528 .359 .654 114 1.00 .266 .453 .392 .293 .520 .530 .699 115 1.00 .293 -.01 .206 .433 .608 .618 116 1.00 .301 .224 .325 .424 .598 117 1.00 .195 .261 .303 .467 118 1.00 .385 .410 .508 119 1.00 .565 .716 120 1.00 .773 Table 4 Item Correlation Matrix for Part B (Beta). 111 112 113 114 115 116 117 118 119 120 Total 111 1.00 .271 .348 .389 .497 .366 .269 .304 .355 .592 .708 112 1.00 .405 .309 .275 .207 .175 .395 .271 .262 .508 113 1.00 .501 .327 .376 .396 .294 .442 .301 .606 114 1.00 .244 .394 .291 .298 .502 .498 .636 115 1.00 .478 .199 .209 .336 .518 .629 116 1.00 .489 .412 .257 .383 .633 117 1.00 .110 .244 .393 .486 118 1.00 .359 .376 .587 119 1.00 .457 .633 120 1.00 .751 33 Table 5 Factor Loadings, Eigenvalues, and Percentages for Part A (Alpha). Factors One Two Three Four h2 item 1 .506 —.066 -.072 -.319 .367 item 2 .317 -.074 -.013 .269 .179 item 3 .519 .206 .135 -.049 .332 item 4 .221 .074 .393 -.153 .232 item 5 .590 -.633 -.325 .038 .857 item 6 .479 .627 —.404 .098 .795 item 7 .314 .125 .042 -.066 .120 item 8 .356 -.025 .360 .331 .367 item 9 .473 -.053 .236 -.025 .283 item 10 -.l69 -.012 -.097 .114 .051 Factor Eigenvalue % of Var. Cum. % One 1.731 48.3 48.3 Two .871 24.3 72.6 Three .643 18.0 90.6 Four .339 9.5 100.0 34 Table 6 Factor Loadings, Eigenvalues, and Percentages for Part A (Beta). Factors One Two Three Four h2 item 1 .548 .058 -.162 .029 .331 item 2 .364 .279 -.054 -.214 .259 item 3 .577 .343 .195 -.l62 .515 item 4 .671 .198 -.291 .224 .624 item's .519 -.278 -.243 .317 .506 item 6 .604 .144 -.173 -.128 .431 item 7 .563 -.129 .472 -.354 .682 item 8 .425 -.469 .534 .325 .802 item 9 .307 -.359 -.207 -.019 .267 item 10 -.036 .671 .319 .412 .723 Factor Eigenvalue % of Var. Cum. % One 2.443 47.5 47.5 Two 1.152 22.4 69.9 Three .904 17.6 87.5 Four .641 12.5 100.0 35 highest loadings on factor one. Although items 5, 6, and 9 have their highest loadings on different factors, they all display significant correlations (r item-factor greater than .30) on parallel factors. Of the ten items, only items 4 and 10 display a lack of similarity. The significance of each factor was assessed by means of the Kaiser criterion (Child, 1970). Only those factors whose eigenvalues exceed unity, according to this criterion, are considered significant. According to this criterion, factor one in both Alpha and Beta is significant. Factor two, barely sig- nificant in Beta, is not in Alpha. The remaining factors do not approach a level of significance. It might be concluded that at least one group factor results from data rotation in each organization. While the specific item loadings are generally parallel, Tables 5 and 6 indi- cate the existence of unique factors comprised of items which are organization specific. To determine the degree of similarity between organizations in terms of factor structure, a coefficient of factorial similarity (Harmon, 1970) was computed com- paring Part A of the questionnaire for Alpha and Beta. The coefficients for each of the four factors were .93161, .01088, .27404, and .18219 for factors one through four respectively. Of these coefficients, only factor one approaches a significant level of correspond- ence between organizations. 36 With respect to Part A of the influence question- naire, several conclusions may be made. First, at least one significant group factor emerged in both factor analyses. Second, non-significant unique factors resulted from both rotations. Three, the pattern of item to factor loadings approached significance only in regards to factor one. While most factor analyses can be interpreted in a variety of ways depending on the rotational procedure used and criterion of significance, this study will interpret the results of Part A analyses as indicating the existence of one group factor relating to perceived influence from others. Although this construct displays a degree of heterogeneity both within each factor analysis and between organizations, the degree of congruency can- not be dismissed. The occurrence of unique factors which show little similarity across organizations may well be due to differences in sampling procedures or organiza- tional structure evidenced in this study. The pervasive- ness of these contaminating variables will be discussed under section II, Sampling Procedure and Data Collec- tion. Part B of the influence questionnaire, containing items 11 to 20, is a parallel version of Part A with the exception that the direction of influence is reversed. While Part A attempts to measure various aspects and degrees of perceived influence exerted on the respondent, 37 Part B attempts to measure the perceived influence exerted by the respondent on others through various dimensions. As was the case in analyzing the results of Part A of the questionnaire, several factor analytic solutions were employed. Since similar results were obtained using these different solutions, only the varimax rotation is displayed in Tables 7 and 8 for Alpha and Beta respective- ly. Inspection of the correlation matrix corresponding to each of these factor analyses, Tables 5 and 6, sug- gests a moderate degree of item similarity. Beta, com- pared with Alpha, appears to evidence a slightly higher inter-item correlation but both present many correlations in the moderate range (.30 to .50). Factor analysis results for both Alpha and Beta indicate both a strong degree of item homogeneity and correspondence. Using the Kaiser criterion, both Tables 7 and 8 indicate the existence of a single significant factor relating to perceived influence over others. Factor one in both analyses accounts for almost 73% of item variance. Comparison of item-factor loadings be- tween organizations also indicates a degree of similarity. With the exception of item 12, every item has its high- est loading on the same factor--the first. Harmon's coefficient of factor similarity was computed comparing the pattern of factor loadings by item for both Alpha and Beta. As computed, these coefficients were .98273, 38 Table 7 Factor Loadings, Eigenvalues, and Percentages for Part B (Alpha). Factors One Two Three h2 item 11 .765 -.218 —.l79 .665 item 12 .221 .092 .343 .175 item 13 .627 .040 .523 .669 item 14 .668 .330 -.076 .560 item 15 .717 -.648 -.042 .935 item 16 .544 .173 -.050 .328 item 17 .360 .420 -.289 .389 item 18 .439 .162 .099 .229 item 19 .703 .115 .107 .519 item 20 .809 -.038 -.214 .702 Factor Eigenvalue % of Var. Cum. % One 3.754 72.6 72.6 Two .833 16.1 88.7 Three .584 11.3 100.0 39 Table 8 Factor Loadings, Eigenvalues, and Percentages for Part B (Beta). Factors One Two Three h2 item 11 .657 -.114 -.263 .514 item 12 .455 -.113 .293 .306 item 13 .627 .061 .334 .508 item 14 .635 -.079 .205 .452 item 15 .596 -.l80 —.355 .513 item 16 .603 .053 -.032 .368 item 17 .492 .806 -.026 .893 item 18 .512 -.234 .191 .354 item 19 .580 —.076 .045 .344 item 20 .736 -.026 -.246 .603 Factor Eigenvalue % of Var. Cum. % One 3.537 72.9 72.9 Two .782 16.1 89.0 Three .535 11.0 100.0 40 .53469, and .72846 for factors one, two, and three respectively. Only factor one indicated a significant degree of congruency between organizations. These results lend themselves to the interpreta- tion that, so far as perceived influence over others is concerned, a general factor may be said to exist. Al- though significant loadings are noted for specific items on other factors, these factors do not represent signif- icant percentages of variance. Questions 12 and 17 in particular may represent unique factors which vary according to specific administrations of the instrument. To summarize the analyses and interpretations related to both Part A and Part B of the influence ques- tionnaire, several conclusions may be drawn. Both meas- ures indicated a degree of homogeneity among items. Part B appears to represent, comparatively, a more homo- geneous trait since it may be composed of a general in addition to specific factors. Part A, on the other hand, may be primarily composed of group and specific factors. In comparing item-factor loading similarity between organizations, only factor one in each analysis exhibited a strong degree of consistency. Each of the other fac- tors which emerged from factor rotation did not "cross validate." It was maintained that these results could be interpreted to indicate that both Parts A and B measure, to a degree, a relatively homogeneous trait pertaining to perceived influence over work behaviors. 1.... ills”: .. ....|, IA..h.MnRflnbr.to1l|iQ f, 41 Those variations which were evident in results might be attributed to differences in organizations and possibly sampling procedures. To further assess the nature of perceived influ- ence and to determine the plausibility of an alternative conceptualization, an additional series of analyses was conducted. In contrast to the initial procedure of dividing the questionnaire into two parts ("influence from" and "influence on"), item pairs were computed by obtaining the difference between responses to parallel items. Each of the resulting ten pairs thus represented the difference between influence from others and influence on others with respect to each of the ten dimensions of influence. These results were then factor analyzed (varimax rotation) to determine the structure, if any, with respect to the entire questionnaire. The item-pair correlation matrix for Alpha is depicted in Table 9 and in Table 10 for Beta. Factor analysis results for Alpha and Beta are found in Tables 11 and 12 respectively. Inspection of Tables 11 and 12 indicates that item pair characteristics represent several different constructs. The item communalities, for instance, sug- gest that, for many of the item pairs, a great deal of variance remains unexplained through factor solutions. Although factor one represents a significant amount of item Variance in both Alpha and Beta, four factors were derived from Alpha while only three resulted from Beta responses. 42 Table 9 Item Pair Correlations (Alpha). 1 2 3 4 5 6 7 8 9 10 1 1.00 .233 .227 .394 .309 .157 .225 .038 .048 .229 2 1.00 .357 .263 .050 .183 .149 .098 .226 -.07 3 1.00 .418 -.03 .369 -.02 .181 .304 .094 4 1.00 -.01 .289 .249 .357 .357 .214 5 1.00 -.08 -.21 .062 .040 .349 6 1.00 .235 .190 .296 .034 7 1.00 .117 .330 .027 8 1.00 .333 .219 9 1.00 .137 10 1.00 Table 10 Item Pair Correlations (Beta). 1 2 3 4 5 6 7 8 9 10 1 1.00 .040 .405 .438 .420 .111 .480 .209 .239 -.06 2 1.00 .468 .227 .359 .209 .124 .094 .076 -.13 3 1.00 .629 .338 .476 .613 .307 .646 -.12 4 1.00 .414 .364 .450 .235 .703 -.12 5 1.00 .205 .148 .378 .444 -.10 6 1.00 .712 .406 .370 .010 7 1.00 .456 .309 .062 8 1.00 .534 .133 9 1.00 .183 10 1.00 43 Table 11 Factor Loadings, Eigenvalues, and Percentages for Item Pairs (Alpha). Factors One Two Three Four h2 pair 1 .251 .694 -.109 .243 .616 pair 2 .449 .058 .039 .070 .212 pair 3 .673 .143 .180 -.106 .51? pair 4 .273 .266 .325 .216 .297 pair 5 .035 .310 .079 -.098 .113 pair 6 .323 -.039 .237 .147 .184 pair 7 .037 -.026 .137 .515 .286 pair 8 .114 .049 .486 .097 .261 pair 9 .318 .080 .490 .128 .364 pair 10 -.125 .508 .453 -.062 .483 Factor Eigenvalue % of Var. Cum. % One 1.78888 53.7 53.7 Two .65521 19.7 73.3 Three .52391 15.7 89.0 Four .36514 11.0 100.0 44 Table 12 Factor Loadings, Eigenvalues, and Percentages for Item Pairs (Beta). Factors One Two Three h2 pair 1 .433 .023 .157 .213 pair 2 .215 .303 -.055 .141 pair 3 .595 .445 .158 .578 pair 4 .679 .237 .173 .548 pair 5 .300 -.003 .310 .187 pair 6 -.026 .768 .161 .616 pair 7 .141 .375 .167 .188 pair 8 .089 .169 .633 .437 pair 9 .438 .199 .648 .651 pair 10 .054 .027 .290 .088 Factor Eigenvalue % of Var. Cum. % One 2.49901 68.6 68.6 Two .64195 17.6 86.2 Three .50440 13.8 100.0 45 Item loadings for specific pairs also appear to vary to a greater degree than were noted in previous solutions. Of the ten item pairs in this analysis, four (pair 3, pair 5, pair 8, and pair 9) had their highest loadings on the same factors. For the most part, the factor rotations may be said to indicate the existence of unique or, at best, small group factors comprised of two to three item pairs. To a large degree, the compo- sition of the factors varies from Alpha to Beta. The efficacy of defining responses to the influence question- naire in terms of clusters of item pairs was thus con- sidered minimal. Although item pairs were used in rela- tion to job satisfaction, each pair was treated as a separate dimension of influence. b. Influence Questionnaire — Reliability Assessment Consistency of responses to the items com- prising the influence questionnaire was assessed in several ways. To a large degree, a primary concern was the possibility of response styles as a factor affecting results independently of the variables under consideration in this study. Tables 1 through 4, as previously noted, cite the inter-item correlation matrices for Alpha and Beta in relation to items 1 to 10 (Part A) and 11 to 20 (Part B). In general, these matrices indicate a relatively "flat" relationship among items. The last column in each of 46 these four tables is the item—total correlation for each question. Total score was considered as the sum of responses to each part of the questionnaire. With the exception of item 10 for Beta, every item-total correla- tion was significantly different from zero. The exist- ence of low to moderate intercorrelations with moderate to strong item-total correlations may be interpreted to indicate the lack of response styles. Consistent pat- terns of response style would be indicated by relatively high intercorrelations among items. Since this, in gen— eral, was not the case here, it would appear that subjects tended to respond to each question independently of every other question. High item-total correlations evident in the analyses may be taken as indications of independent contributions to total score. The internal consistency of Part A, Part B, and total responses to all twenty items was assessed by use of Coefficient Alpha. Table 13 lists those values obtained through each analysis for both Alpha and Beta. Both Part A and Part B evidence a good deal of internal consistency as did the consistency of total scores. Part A for Alpha displays the lowest level of internal consistency, .62, all other coefficients being in the low to high .80's. The relationship between Part A and Part B was not totally independent of each other. For Alpha, Part A total score correlated with Part B total score .59. Beta correlation between total score on A 47 and total score on B was somewhat lower, .42. The consistency between Part A and Part B as measured through item pairs for similar dimensions suggests a weak to moderate relationship. Tables 14 and 15 present the item pair correlations for both Alpha and Beta respectively. For Alpha, the median correlation between pairs is .30. Beta, in contrast, shows a median correlation of .23 for all item pairs. So far as reliability is concerned, both Parts A and B appear to indicate a substantial degree of con- sistency both within each part and for the questionnaire as a whole. Item pair correlations, on the other hand, indicate little relationship, for the most part, between different directions of influence based on the same dimension of influence. As was previously mentioned, the test-retest reliability of an earlier draft of the influ- ence questionnaire was +.82. In sum, it was concluded that the questionnaire, as a whole, indicated a strong degree of reliability. B. Sampling Procedures and Data Collection As previously indicated, two organizations were used in this study--Alpha and Beta. Both organizations are banks engaged in commercial enterprises within the same geographic region, the middle Atlantic seaboard. Alpha, the larger of the two banks, employed at the time of this study nearly 4300 individuals in all branches of its operation. Of these individuals, about 2000 or 48% 48 Table 13 Coefficient Alpha for Parts A & B and Total (Alpha and Beta). Part A Part B Total Alpha .6246 .8279 .8395 Beta .8245 .8624 .8834 Table 14 Item Correlations by Pairs (Alpha). Item Pair Correlation l. 1, 11 .320 2. 2, 12 .458 3. 3, 13 .265 4. 4, 14 .044 5. 5, 15 .154 6. 6, 16 .431 7. 7, 17 .479 8. 8, 18 .295 9. 9, 19 .389 10. 10, 20 .209 Table 15 Item Correlations by Pairs (Beta). Item Pair Correlation '1. l, 11 .292 2. 2, 12 .279 3. 3, 13 .084 4. 4, 14 .251 5. 5, 15 .410 6. 6, 16 .240 7. 7, 17 .555 8. 8, 18 .148 9. 9, 19 .222 10. 10, 20 .040 49 were stationed at their main headquarters building. Beta, on the other hand, employed slightly under 3600 individ- uals of whom 1100 or 34% were located at the main head- quarters complex. Only those individuals employed at the main headquarters building for each bank were in- cluded in this study. Both from a logistical point of view and in terms of sampling, this decision to involve only headquarters personnel was a necessary one. To involve organization members employed at branch locations would have required mailing questionnaires through bank channels. This loss of control would have most likely resulted in a number of unreturned question- naires. In addition, being mailed through bank channels would have possibly created the impression that the organizations themselves were using the questionnaire results. Branch personnel also represented a limited sample of organizational functions and 1evels--being con- cerned primarily with small individual checking and sav- ings accounts. Headquarters staff, on the other hand, covered all levels of the organizational hierarchy and represented a wide diversity of functions. Carrying out data collection in a centralized location both sim- plified the process and permitted the researcher to be personally identified with a local university rather than the organization itself. The decision to carry out this investigation in banks was done for several reasons. Primary, of course, 50 was the obtaining of permission to do the study from appropriate authorities within each bank. After explain- ing the nature and purpose of the study, this permission was secured. Another reason for using banking organiza- tions was that they employed a considerable number of individuals at all levels of the hierarchy. Larger sam- ple size, all other things being equal, tends to increase the power of statistical tests employed in this study. A final reason for using banks in this study concerned the degree of differentiation among hierarchical levels. Because of their traditional involvement with money, most banks rely heavily on formal authority structure and clearly defined roles. Since hierarchical level was treated as an independent variable in this study well defined hierarchical distinctions were required. The participation of these banks was motivated by a promise on the part of this researcher to provide feedback of the study's results to the directors of per- sonnel. It was also agreed that any publication result- ing from this study would not identify the banks by name. All subjects in this study were volunteer par- ticipants. In all cases, care was taken to insure their anonymity so far as was possible. Respondents were asked to indicate only their job title and department or section. Any additional information, it was thought, would have increased apprehension regarding 51 confidentiality. Feedback to participants was provided through the in-house company newsletter. Both the sample selection and data collection procedures differed from Alpha to Beta. Despite the researcher's efforts to minimize any procedural varia— tion, the requirements imposed by the personnel depart- ments and situational circumstances took precedence. The sample selection procedure used at Alpha was more systematic than the procedure used at Beta. From the bank's computer listing, the personnel director at Alpha did a simple random selection of 400 organization members employed at all levels in the headquarters building. With a cover letter from this researcher on his university stationery, each of those selected for the study was sent a memo from the personnel office requesting that they volunteer to take part in the study and assuring them that the result of the study would not be used for any administrative purpose. Respondents were asked to "drop by" when they had a few minutes to spare anytime during the appointed day. A large training room was set aside for the purpose of data collection. During the data collection procedures no bank officials were present--only this researcher and his associate. This was done to minimize apprehension on the part of respond— ents. Both the researcher and his associate wore uni- versity name tags to further divorce the study from bank affiliation. 52 The sample selection and data collection pro- cedures were markedly different at Beta. At the request of the personnel director of this bank, no systematic selection procedure was used. He indicated that he felt this would reduce apprehension on the part of the employ- ees. Instead, a notice was placed next to the time clocks or bulletin board in each department stating the date of the study, its purpose, and requesting employee participation. Bank officers were notified by telephone. Data collection was carried out by this research— er and his associate accompanied by the personnel director. Bank policy required all "visitors" to be escorted by bank personnel for security reasons. Each department was, in turn, visited. The employees in each department were assembled by the personnel director. He then described the nature of the study and asked them to participate. A sufficient number of questionnaires were left and collected several hours later. The differences in sampling and data collection procedures between the two banks had two discernible effects. First, the number and percentages of respondents differed from bank to bank. Of the 400 sampled cases in Alpha, 227 actually completed the questionnaire. A much smaller percentage of questionnaires was completed by Beta members. Although the actual number of organization members contacted at Beta was not identified, those com- pleting the form appeared to represent far less than 50 53 percent. Secondly, the number of items marked incorrect- ly or left unmarked was slightly higher in frequency for Beta than for Alpha. At Alpha, it was possible to inspect the questionnaires for completeness and accuracy before the majority of respondents left the research situation. Thus, fewer than one percent of all questions were omitted in data analysis. In Beta, since data were collected some hours after completion of the question- naire, it proved futile to attempt to identify individual respondents who had only partially or incorrectly com- pleted the questionnaire. In cases where hierarchical level could not be identified the entire questionnaire was discarded. Where hierarchical level could be identi- fied but some responses were incorrectly scored or miss- ing, only those responses which were usable were incor— porated in data analysis. The sampling and data collection procedures resulted in an unequal distribution of frequencies across banks. A comparison was made in terms of number of responses for each organizational level compared to the total number of organization members at each level of the hierarchy. The results are presented in Table 16. A 2 x 4 contingency table, banks by four levels of hierarchy (rank and file, supervisors, management, and top management) yielded a Chi Square value of 43.786. This value was significant at the .01 level. Departures from expected frequency were most evident for Beta's 54 Table 16 Comparison of Responses: Alpha and Beta Alpha Beta Number of % of Number of % of Level Respondents N Sample Respondents N Sample rank & file 144 1715 8.4 35 903 3.9 supervisor 64 212 30.2 35 126 27.8 management 12 75 16.0 29 50 58.0 top mgmt. 7 7 35.0 1 7 14.3 227 2009 11.3 100 1086 9.2 management level; this cell alone contributed a value of 21.6. Table 16 shows the relationships among hierarch- ical levels and banks. Less than 4% of the rank and file workers at Beta participated in the study compared to 8.4% in Alpha. In contrast, 58% of the managers at Beta participated com- pared to 16% for Alpha. For both banks, the number of cases at top management level was very low (n = 7 for Alpha and n = l for Beta). Consequently, top management responses were discarded and all subsequent data analyses incorporated only three levels of hierarchy: rank and file, supervisors, and management. In all data analyses to follow, Alpha and Beta responses were treated separately. A greater degree of confidence could be placed in any results of this study if both organizations evidenced the same relationships 55 among variables than would be the case if found in one bank alone. For all practical purposes, both banks appeared to be similar in most aspects of operation and personnel. While obvious differences existed in terms of number of employees and number of hierarchical levels, so far as possible these differences were taken into account both methodologically and statistically. Since no demographic or biographic data were collected on the respondents, it was impossible to tell whether or not those who volun- teered for the study differed in any significant way from those who did not. This chapter was comprised of two sections: section I examined the nature of the influence question- naire developed in thisriudy whereas section II discussed the sampling and collection procedures. The validity and reliability estimates of the influence questionnaire, while far from ideal, did indi- cate the existence of some common elements and relation- ships among questionnaire items. It was concluded that the questionnaire did measure perceived influence from others and perceived influence on others to a large de— gree. In each organization, the relationship among items did vary somewhat resulting in unique aspects of perceived influence. Differences in item relationships were ascribed to specific situational factors possibly 56 resulting from different sample characteristics and data collection methods. Part II of this chapter described the nature of the organizations involved and techniques used to measure the variables under consideration. Although sampling procedures and data collection varied from bank to bank, the effect of these differences represented an unknown factor in the study. Any differences in sample character- istics were assumed to be randomly distributed in both organizations. Chapter Four Results To answer the questions posed in the problem sec- tion, several measurements were obtained for each subject in this study. First, each subject was assigned a score indicating his or her hierarchical level in the following manner: 1 = rank and file level 2 = supervisory level 3 = management level 4 = top management level It will be recalled that level 4, tOp management level, responses were discarded due to the small number of respondents in this category (N = 8). A second score for each subject was obtained by requiring subjects to complete the Brayfield—Rothe Job Satisfaction Index. A simple sum of item responses expressed job satisfaction of respondents. Each item was scored on a l to 5 scale with appropriate item reversals. High scores were considered indicative of a greater degree of job satisfaction compared to low scores a 57 58 Thirdly, each respondent was measured in terms of perceived influence. In all, 20 items were presented. These 20 items represented the 10 dimensions of perceived influence over work behaviors. The first set of ten items pertained to influence from others while the second set of parallel items pertained to the degree of per- ceived influence on the work of others. Responses to each of these 20 items were scored: 1 = very little or no influence 2 = a little influence 3 = some influence 4 = a great deal of influence 5 = a very great deal of influence In all cases where subjects omitted responses or the item did not apply to the respondent's job, the response was scored "0" and deleted from analysis on an item by item basis. On an overall basis, the average response to each of the twenty items comprising the influence questionnaire are presented in Table 17. Columns represent a given hierarchical level within a bank while rows are specific questionnaire items. Each mean represented in this table is based upon different number of cases depending on the organization, hierarchical level, and item number. It should be noted that row and column means within each bank, although unequal in number of cases, are proportion- al as is evidenced in Table 18. Table 18 presents the 59 Table 17 Response Means by Level and Organization Alpha 1 Alpha 2 Alpha 3 Beta 1 Beta 2 Beta 3 11 2.92 2.98 3.00 2.57 3.17 2.90 12 3.48 3.36 3.58 3.23 3.43 3.17 I3 3.84 3.95 4.25 3.89 3.80 3.65 I4 3.49 3.38 3.33 3.34 3.83 3.38 IS 2.65 3.16 3.22 2.55 3.32 3.30 I6 2.79 3.48 3.17 2.91 3.09 3.07 17 3.34 3.70 3.45 3.95 2.33 2.91 18 3.43 3.19 3.17 3.32 3.44 3.21 19 3.81 4.00 4.17 3.45 3.71 3.66 110 3.89 4.28 3.75 4.09 4.24 4.35 111 2.75 3.41 3.25 2.77 2.91 3.10 112 2.76 2.77 2.36 2.83 2.79 2.34 113 2.87 3.22 2.83 3.03 2.85 2.97 114 2.07 2.73 2.75 2.34 2.29 2.38 115 3.17 3.84 3.78 2.10 3.40 3.74 116 1.99 2.75 2.58 2.09 2.06 2.79 117 2.68 3.40 3.00 3.28 1.92 2.33 118 2.69 2.72 2.50 2.82 2.37 2.38 119 2.96 3.54 3.25 2.97 3.06 3.07 120 2.54 3.38 2.83 2.57 2.85 2.93 60 Table 18 Number of Cases by Item and Organization Alpha 1 Alpha 2 Alpha 3 Beta 1 Beta 2 Beta 3 11 144 64 12 35 35 29 12 144 64 12 35 35 29 13 144 64 12 35 35 29 14 144 64 12 35 35 29 15 51 42 9 11 19 20 16 143 64 12 34 34 29 17 104 47 11 21 18 22 18 144 64 12 34 34 29 I9 144 64 12 33 34 29 110 144 64 12 35 33 29 111 144 64 12 35 33 29 112 144 64 12 35 34 29 113 144 64 12 35 34 29 114 144 64 12 35 34 29 115 46 43 9 10 20 19 116 143 64 12 34 35 29 117 93 47 ll 18 13 18 118 144 64 12 33 34 29 119 144 63 12 33 35 29 120 143 64 12 35 35 29 61 number of cases per question and hierarchical level. With the exception of items five and seven, fifteen and seventeen, non-response percentages were minimal. Loss of data was more evident in the case of Beta than Alpha due, possibly, to data collection differences. Those items in which a large percentage of sub- jects did not indicate degree of perceived influence are similar with respect to their content. Items five and fifteen refer to degree of subordinate influence while items seven and seventeen refer to customer or client influence. Questionnaire items in each case included a "does not apply" response category. Thus, low number of cases for these items indicates the reality of respond- ents' jobs. Figures 1 and 2 depict the values expressedin Table 17. Visual inspection of these figures indicates a degree of parallelism among hierarchical levels within each organization. Inspection also reveals the possi- bility of variance among item means. Comparing organiza- tions as a whole, visual inspection suggests that a somewhat similar pattern of responses emerges in both banks. Subsequent statistical analyses were conducted to assess these relationships. Since the unequal number of cases per level and item were proportional and reflected population parame— ters rather than experimental error, Winer (1968) recom- mends the use of least squares solutions in computing nt1afv‘ F unlike-Dr: , I... ._ in! 51% 62 mEmuH ON ma ma 5H 0H ma «H ma NH HH OH 0 m b m m C O O C v C h C O C C . O 0 C 0 s. . a x g i I a . c . . A \ ’ ~ a on. O a . . .. Q 4 ’ l. O 0.. I... ~ . .2 “H .. ‘ a o. 0 \Co I \.o a. .. . a u y’o‘ I‘loJ \ a .. x . . . .. : . . . x . xx. . . 4. . I a o t to 00 o y o. loo 0 O N .. o. ‘ a o f o a \ . .. . 3 x 1... ‘ o O o. u v \ co 0 O. .0 ‘ C‘ .0. on o. o a _ co. co : .u x m u x .T Ls r. .. C owns» Hm>mH o3» Hm>oH mco Hm>mH n Inuit m N H .1. 4 .x .\ .xmnaflmq >n memuH whamccofiummso mucOSHMCH on Oncoammm mmmuo>< . mucODHMCH . om>wmuumm H wusmwm "'| 63 memuH ON mH mH 5H wH mH ¢H MH NH HH OH m w b .1; H I a (a 09". s o . .. i .2 . .. Us a H ’ o... .fl a A . H 3 . m z .. T. . . H .. ..\\o . 1.. H . . _ \\J 3 x . .. I i . x . a c i. . / ~ O a O 0 ’~ . o s. e .. _ . I . H . to a \\‘d . . o m a H . a c .. . . s .s H H .n L ‘1. 3 . 0 ’ \u LL .H mourn Hm>oH u ..... N 03» Hm>mH u mco Hm>wH u IIIII Hm>mq >3 memuH muHmccoHummso mucwsHMCH Cu m d m N H \J a x 3 H \ H H , c. , ca. .i d» ”I 43 m\ I.» a < .xmummv oncoommm ommu0>< mucmsHMCH pm>Hmuumm N OMDOHE Eur, lad... 64 analysis of variance effects. In essence, least squares solutions weight or adjust each main and interaction effect by the number of cases entering into the summa- tion. All analysis of variance procedures, where appro- priate, were based on this procedure. A. Analysis Methods and Results Pertaining to Issues One to Four As stated in the problem section (Chapter Two), the purpose of this study was to measure and evaluate some specific relationships among the variables of hier- archical level, dimensions of perceived influence, direc- tion of perceived influence, and job satisfaction. In particular, six issues were to be assessed. The first four of these dealt with the relationships among overall perceived influence, dimensions of influence, directions of perceived influence, and contingent relationships among these variables: 1. On the average, do hierarchical levels differ sig- nificantly from each other in terms of overall per- ceived influence? 2. On the average, do dimensions of influence differ significantly from each other in terms of perceived influence? 3. On the average, do directions of influence differ significantly from each other in terms of perceived influence? .. irri- 65 4. On the average, is the degree of perceived influ- ence contingent upon specific hierarchical levels, dimensions of influence, and/or directions of per- ceived influence? In analysis of variance terminology these four questions may be phrased more explicitly. Issue one above refers to the main effects for hierarchy (H). A significant main effect for this variable would indicate that the three levels of hierarchy (h1 = rank and file workers; h2 = supervisors; h3 = managers), when summed over the ten items in both directions, will differ sig— nificantly from each other. Issue two refers to the main effect for items (I). A significant main effect for this variable would indicate that the ten levels of questions differ significantly from each other whem summed over directions of influence and hierarchical level. The ten levels of the items variable are: 11 a co-worker dimension 12 = machinery dimension I3 = rules dimension I4 a supervisor dimension 15 = subordinate dimension I6 = other sections dimension I7 = customers dimension I8 environment dimension 19 standards dimension 110 = self dimension 66 The third issue refers to the main effect for direction of influence (D). A significant effect for this variable would indicate that, on an overall basis, influence from others (d1) differs significantly from influence on others (d2). The fourth issue indicates the possibility of interaction effects. To evaluate these issues an omnibus analysis of variance test was conducted. In this unequal N design, two of the factors were considered within subjects fac- tors (I and D) and one was considered a between subjects' factor (H). Data for Alpha and Beta were analyzed sepa- rately. The overall results of these analyses are pre- sented in Tables 19 and 20. As is evident from inspec- tion of these tables, significant effects resulted from both analyses. These effects will be analyzed first for Alpha and then for Beta. Finally, results from both organizations will be compared. 1. Perceived Influence, Hierarchical Results (Alpha) Table 19 indicates that the main effects for hierarchy and items were statistically significant. In addition, the first order interaction effects of hier- archy by items, hierarchy by direction and items by direction were also statistically significant. Finally, the second order interaction of hierarchy by items by 67 Table 19 Analysis of Variance - Bank Alpha. Source df SS MS F p<. Hierarchy 2 102.55 51.27 14.664 .01 Error H 217 758.74 3.49 Items 9 274.33 30.48 21.75 .01 Error 1 1953 2736.67 1.40 Direction 1 9.08 9.08 1.93 .166 Error D 217 1020.08 4.70 H x I 18 121.80 6.77 4.83 .01 Error H x 1953 2736.67 1.40 H x D 2 80.88 40.44 8.60 .01 Error H x 217 1020.08 4.70 I x D 9 165.36 18.37 13.77 .01 Error 1 x 1953 2606.31 1.33 H x I x D 18 56.04 3.11 2.33 .01 Error H x x D 1953 2606.31 1.33 Table 20 Analysis of Variance - Bank Beta. 68 Source df SS MS F _p<. Hierarchy 2 29.68 14.84 3.70 .03 Error H 96 384.53 4.00 Items 9 592.85 65.87 44.96 .01 Error I 864 1265.77 1.47 Direction 1 17.81 17.81 3.49 .065 Error D 96 489.41 5.10 H x I 18 67.89 3.77 2.57 .01 Error H 864 1265.77 1.47 H x D 2 1.65 .83 .16 .85 Error H 96 489.41 5.10 I x D 9 165.62 18.40 15.56 .01 Error I 864 1021.99 1.18 H x I x 18 52.91 2.94 2.48 .01 Error H x D 864 1021.99 1.18 69 direction was significant. In fact, only the overall effects of directions of influence were not significant. Interpretation of these results required several analyses whose purpose was to assess the nature of the interaction effects. With note to the fact that the triple interaction was significant, it was decided to con- duct a simple effects analysis holding hierarchy constant. Thus, for each level in the organizational hierarchy, the interaction between direction and items was assessed. For those hierarchical levels in which main (directions or items) or interaction effects (directions by items) were noted, Scheffé or appropriate pairwise comparisons were conducted. For Alpha, the above procedure first resulted in three two-way analyses of variance--one for each hier- archical level. The outcomes of these analyses are pre- sented in Tables 21, 22, and 23. Inspection of these three tables reveals that both items and the interaction between items and direc- tions are significant for rank and file workers; direc- ticms, items and the interaction between the two are significant for supervisors; no effects were present for the management level in Alpha. Subsequent analyses were conducted on Alpha level one and level two members to determine the nature of the interaction effects. All subsequent analyses were comprised of Scheffé post-hoc multiple comparisons of items. Critical 70 Table 21 Analysis of Variance for Directions and Items (Rank and File Workers - Alpha). Source SS df MS F Ap A. Direction 118.01 1 118.01 30.86 .01 B. Items 1723.42 9 191.49 83.184 .01 D x 1 424.64 9 47.183 21.206 .01 errorA 546.84 143 3.824 errorB 2967.83 1287 2.302 errorA.B 2864.01 1287 2.225 Table 22 Analysis of Variance for Directions and Items (Supervisors - Alpha). Source SS df MS F p A. Direction 6.23 l 6.23 1.859 n.s. B. Items 477.32 9 53.04 22.76 .01 D x I 166.35 9 18.48 8.812 .01 errorA 211.17 63 3.35 errorB 132.08 567 2.33 error . 1189.25 567 2.10 71 Table 23 Analysis of Variance for Directions and Items (Management - Alpha). Source SS df MS F p A. Direction 6.19 l 6.19 1.54 n.s. B. Items 24.70 9 2.74 1.36 n.s. I". D x I 33.16 9 3.68 1.90 n.s. ‘] errorA 44.26 11 4.02 errorB 199.95 99 2.02 g errorA.B 191.89 99 1.94 values for pairwise of items comparisons were set at the alpha .01 level to maintain a degree of power. For both level one and level two members in Alpha every cell mean for each of the ten items and two directions were com- pared. This process resulted in 210 pairwise comparisons for each level. The critical difference values necessary for significance were computed separately for items 5, 7, 15, and 17 due to gross differences in sample size. The critical difference values necessary for each comparison were: level one: (all items but 5, 7, 15, 17) = .249 (items involving 5, 7, 15, 17) .443 level two: (all items but 5, 7, 15, 17) = .513 (items involving 5, 7, 15, 17) = .613 Although all pairwise comparisons between means were conducted, only those involving questions within 72 the same direction of influence are presented. It was thought that little additional insight was to be gained by making comparisons between directions of influence. Tables 24 through 27 present the results of these com- parisons. Tables 24 and 26 are based on the results from level one of Alpha. Table 24 presents the pairwise com- parisons between each of the ten items comprising Part A of the influence questionnaire ("influence from others"). In this and Tables 25, 26, and 27, item means are arranged in descending order. Significant differences between means are indicated by an asterisk (‘). Mean differences which fall below the necessary level of significance are denoted by n.s. The mean value for a given item is indi— cated next to that item number. For both levels one and two, the perceived degree of influence exerted by a specific dimension varies some- what. For the most part, both rank and file workers agree that self-influence over workers behaviors (item 10) is significantly greater than most other dimensions. The influence of rules (item 3) and goals or standards (item 9) appears also to be a significant factor in determining work behaviors for both levels. Despite differences in hierarchical position, both levels agree that neither co-workers (item 1) nor subordinates (item 5) influence them to any great extent. The greatest amount of disagreement regarding degree of perceived Table 24 10(3.89) mqummhxow (3.84) (3.81) (3.49) (3.48) (3.43) (3.34) (2.92) (2.79) (2.65) Table 25 10(4.28) HUlmNimeWKO (4.00) (3.95) (3.70) (3.48) (3.38) (3.36) (3.19) (3.16) (2.98) 73 Pairwise Comparison of Item Means based Rank and File Workers'Responses (Part A Alpha). Items 10 3 9 4 2 8 7 l 6 5 ns - - - - - - - - - ns ns - - - - - - - - O 0 0 .. _ .. ._ ... .. .. o o a n3 _ _ - .. _ .. ‘ ' ' ns ns - - - - - ‘ ‘ ‘ ns ns ns - - - - O O 0 t I O . ... ... .— o o o o o o 0 n5 .. .. O 0 t t O O 0 n5 n5 _ significant difference non-significant difference ns Pairwise Comparison of Item Means based Supervisors' Responses (Part A - Alpha) Items 10 9 3 7 6 4 2 8 5 1 ns - - - - - - - - - ns ns - - - - - - - - ns ns ns - - - — - - - c o 0 ns _ _ _ _, _, _ ‘ ‘ ‘ ns ns - - - - - ‘ ‘ ‘ ns ns ns - - - - ’ ' ‘ ' ns ns ns - — - ‘ ‘ ‘ ‘ ns ns ns ns — - ‘ ‘ ‘ ‘ ’ ns ns ns ns - significant difference non-significant difference ns on on 74 Table 26 Pairwise Comparison of Item Means based on Rank and File Workers' Responses (Part B - Alpha). Its—‘32 15 19 13 12 11 18 17 20 14 16 15(3.l7) — - - - - — - - _ _ 19(2.96) ns - - - - — - - - - 13(2.87) ns ns - - - - - — _ - F7 12(2.76) ns ns ns - - - - - — - - 11(2.75) ns ns ns ns - — - - - - 18(2.69) ‘ ns ns ns ns - - - — — 17(2.68) ° ns ns ns ns ns - - - - A 20(2.54) ‘ ‘ ’ ns ns ns ns - - - y 14(2.o7) . . o . . . . . _ _ 16(1.99) ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ns - significant difference non-significant difference ns Table 27 Pairwise Comparison of Item Means based on Supervisors' Responses (Part B - Alpha). Items 15 19 11 17 20 13 12 16 14 18 15(3.84) - - - - - - — _ - - 19(3.54) ns - - - - - - - - - ll(3.41) ns ns - - — — - - _ _ 17(3.40) ns ns ns - - - - - _ - 20(3.38) ns ns ns ns - - — - - - 13(3.22) ‘ ns ns ns ns - - - — — 12(2.77) ' * a - . n5 _ - - _ l6(2.75) ‘ ‘ ‘ ‘ ‘ ns ns - - - 14(2.73) ‘ ‘ ‘ ’ ‘ ‘ ns ns - - 18(2.72) ' ' ‘ ‘ ' ' ns ns ns - significant difference non-significant difference ns 75 influence between levels appears to concern those dimen- sions exerting a moderate amount of influence over work behaviors. Comparing the item mean differences on Part B of the questionnaire for both level one and level two of Alpha, a good deal of similarity is evidenced. Both levels indicated that they, as individuals, exerted a great deal of influence on subordinates (item 15) and goals and standards (item 19). In general, both levels indicated little influence over their boss or supervisor (item 14) or other departments (item 16). Aswas the case with Part A of the questionnaire, disagreement between levels is most evident for those middle or moderate dimensions of influence. To further compare the degree of similarity in ranking of questions in terms of perceived influence, Spearman Rank Order correlations were computed. In these comparisons, the rank order of items for each level in Alpha was compared with the rank order at all other levels for both Part A and Part B of the questionnaire. Table 28 summarizes these correlations. Table 28 Rank Order Correlations Comparing Organizational Levels on Part A and Part B of the Questionnaire (Alpha). Level Comparisons 1 vs. 2 1 vs. 3 2 vs. 3 Part A .736 .790 .790 Part B .636 .483 .885 76 Table 28 indicates that a moderate to strong relationship exists between levels in terms of ranking of items on both parts of the questionnaire. It is evi- dent that, despite differences in hierarchical level, organization members perceive dimensions of influence in a fairly similar manner. To summarize the results of these analyses deal— ing with the relationships among hierarchical level, dimensions of influence, and direction of influence for Alpha, several conclusions may be reached. Although both hierarchy and dimensions of influence were significant in an overall sense, the relationships between these two variables and direction of influence varied with hier- archical level. Subsequent analyses showed that, for level one and two organization members, the interaction between items and direction of influence was significant although this was not the case for level three organiza- tion members. Pairwise comparisons of items within a given direction showed that self, goals and standards, and rules dimensions were believed by organization mem- bers at both levels to exert a great deal of influence over their work behaviors with co-workers and subordinates exerting little influence. Similarly, both organization levels generally agreed that they exerted a great deal of influence over subordinates and goals and standards but little influence over their boss or other departments. 77 Correlational analysis tended to support the similarity in ranking between organizational levels. The analysis of Beta responses to Part A and B of the influence questionnaire will be conducted in a similar manner to Alpha. Table 20 presents the omnibus analysis of vari- ance for the variables of hierarchy, direction of influ— ence, and items with respect to bank Beta. Inspection of this table reveals that the main effects for hierarchy and items were statistically significant but such was not the case for direction of influence. The first order interaction effects for hierarchy by items and items by direction were significant as was the triple interaction of hierarchy by items by direction. As was the case with Alpha, the third order interaction effects were analyzed by simple effects analysis, holding hierarchical level constant and assessing the relationship between direction of influence and items at each level of the organizational hierarchy. The results of these analyses for each hier— archical level are presented in Tables 29, 30, and 31 for rank and file workers, supervisors, and management levels respectively. Table 29 indicates that, while directions of influence was not a significant variable, there was a significant interaction between items and direction of influence. Overall, items varied in terms of their per- ceived influence. 78 Table 29 Analysis of Variance for Directions and Items (Rank and File Level - Beta). Source SS df MS F p A. Direction 15.61 1 15.61 2.90 n.s. B. Items 580.59 9 64.51 27.69 .01 D x I 48.00 9 5.33 2.49 .01 errorA 182.89 34 5.38 errorB 713.11 306 2.33 errorA.B 654.49 306 2.14 Table 30 Analysis of Variance for Directions and Items (Supervisory Level - Beta). Source SS df MS F p A. Direction 37.36 1 37.36 9.67 .01 B. Items 497.88 9 55.32 25.12 .01 D x 1 166.28 9 18.48 9.8 .01 errorA 131.44 34 3.87 errorB 673.92 306 N 2.20 error . 576.92 306 1.89 ”Nil a. (urinary. T 79 Table 31 Analysis of Variance for Directions and Items (Management Level - Beta). Source SS df MS F p A. Direction 13.20 1 13.20 3.80 n.s. B. Items 216.72 9 24.08 11.15 .01 D x I 127.72 9 14.19 7.89 .01 errorA 97.25 28 3.47 errorB 544.34 252 2.16 errorA.B 453.34 252 1.80 Table 30 indicates that, for supervisors, both the direction and items variables showed significant main effects. As was the case with rank and file workers, a significant interaction between question and direction occurred. Management level responses, Table 31, shows that items as a whole varied in terms of perceived influence while directions of influence did not. Once again, the interaction between items and direction of influence was statistically significant. Significant interactions, as was the case with Alpha, were compared using the Scheffe procedure, alpha = .01 level. Items 5, 7, 15, and 17 were compared using a different critical difference level due to large dif- ferences in sample size for these items. The critical 80 difference values necessary for each pairwise comparison were: level one: (all items but 5, 7, 15, 17) = .525 (items involving 5, 7, 15, 17) = .829 level two: (all items but 5, 7, 15, 17) = .4739 (items involving 5, 7, 15, 17) = .5994 #3 level three: (all items but 5, 7, 15, 17) = .5362 (items involving 5, 7, 15, 17) = .581 Comparison of items within each hierarchical level, although computed for all pairs, are presented in ; Tables 32 to 37 only in terms of the same direction of influence. Thus items 1 through 10 are compared with each other and items 11 through 20 are compared with each other for each level in the organizational hierarchy. Inspection of Tables 32, 33, and 34 show a high degree of consistency among levels in terms of perceived influence on their work behaviors. All three levels in the organizational hierarchy see themselves, as individ- uals, as the most influential dimension on their work behaviors. In addition, goals and standards (item 9), rules and regulations (item 3), and boss or supervisor (item 4) are seen with some regularity as being relatively influential dimensions over the respondent's work behav- iors. In contrast, other departments (item 6) and co- workers (item 1) are generally judged to be relatively weak influences over the respondent's work behavior. Table 32 10(4.09) mwmmmbxowq (3.95) (3.89) (3.45) (3.34) (3.32) (3.23) (2.91) (2.57) (2.55) Table 33 10(4.24) \lmHmmooxown (3.83) (3.80) (3.71) (3.44) (3.43) (3.32) (3.17) (3.09) (2.33) Pairwise Comparison of Item Means based 81 and File Responses to Part A (Beta). Items 10 7 3 9 4 8 2 6 1 ns - - - - - - - - ns ns - - - - - - - O 0 ns .. - .. - _ — ‘ ‘ ns ns - - - - - 0 ‘ ' ns ns — - - - ‘ ' ‘ ns ns ns - - - ' ’ ° ‘ ns ns ns - - O t 0 O O O 0 n5 _. O C 0 C C ‘ 0 ns ns ° a significant difference = non-significant difference ns Pairwise Comparison of Item Means based Supervisors' Responses to Part A (Beta) on Rank on Items 10 4 3 9 8 2 S l 6 ns - - - - - - - - ns ns - - - - - - - ‘ ns ns - - - - - - ‘ ns ns ns - - - - - ° ns ns ns ns - - - - ' ‘ ' ns ns ns - - - ' ‘ ‘ ‘ ns ns ns - - ° ' ' ‘ ns ns ns ns - O 0 O 0 O 0 I O 0 ‘ a significant difference ns = non—significant difference Table 34 10(4.35) (3.66) (3.65) (3.38) (3.30) (3.21) (3.17) (3.07) (2.91) (2.90) quwoompwxo Table 35 17(3.28) 13(3.03) 19(2.97) 12(2.83) 18(2.82) 11(2.77) 20(2.57) 14(2.34) 15(2.10) 16(2.09) 82 Pairwise Comparison of Item Means based on Managers' Responses to Part A (Beta). Items 10 9 3 4 5 8 2 6 7 1 O _ .. .. ... _ .. ... - — 0 ns ... .. .. - .. - — ... ' ns ns - - - - - - - ‘ ns ns ns - - - - - - ‘ ns ns ns ns - - - - - ' ns ns ns ns ns - - - - ’ ‘ ’ ns ns ns ns - - — ‘ ° ‘ ns ns ns ns ns - - ' ‘ ‘ ns ns ns ns ns ns - ‘ a significant difference non-significant difference Pairwise Comparison of Item Means based on Rank and File Responses to Part B (Beta). Items 17 13 19 12 18 11 20 14 15 16 ns - - - - - - - - - ns ns - - - - - - - - ns ns ns - - - - - - - ns ns ns ns - — - - - — ‘ ns ns ns ns - - - — - ' ns ns ns ns ns - - - - ' ‘ ‘ ns ns ns ns - - - o o o o o 0 n3 n5 _ _ ' ’ ‘ ‘ ‘ ‘ ns ns ns - ° = significant difference ns = non-significant difference Table 36 15(3.40) l9(3.06) ll(2.91) 13(2.85) 20(2.85) 12(2.79) 18(2.37) 14(2.29) 16(2.06) l7(1.92) Table 37 15(3.74) 11(3.10) l9(3.07) 13(2.97) 20(2.93) l6(2.73) 18(2.38) 14(2.38) 12(2.34) 17(2.33) 83 Pairwise Comparison of Item Means based on Supervisors' Responses to Part B (Beta). Items 15 19 ll 13 20 12 18 14 16 17 ns - - - - - - - - - ns ns - - - — - - - - ' ns ns - - - - - - - ' ns ns ns — - - - - - ‘ ns ns ns ns - - - - - o o t o 0 n5 _. - .. .. o o a o o 0 n5 - .. - o . . c a . ns ns _ - ‘ ' ‘ ‘ ‘ ‘ ns ns ns - ‘ = significant difference ns = non-significant difference Pairwise Comparison of Item Means based Managers' Responses to Part 8 (Beta). Items 15 ll 19 13 20 16 18 14 12 17 O ... .. .. .... .. _ .. ... .- 0 n5 _ .. .. .. .. - .. - ‘ ns ns - - - - - - - ‘ ns ns ns - - - - ~ ~ * ns ns ns ns - - — - - O O O t 3 ns _. _. _ .— o c o o a n5 n5 .. .. . ' ‘ ‘ ‘ ‘ ns ns ns - - ‘ ‘ ‘ ‘ ' ns ns ns ns — ‘ 2 significant difference ns = non-significant difference on ””7 84 Tables 35, 36, and 37 demonstrate a lower degree of consistency among hierarchical levels in response to Part B of the questionnaire. There appears to be some agreement between supervisors and management that they influence subordinates more than any other dimension while both agree that they influence customers and cli- {E ents least of all (item 17). Rank and file workers and 1‘ supervisors both rank influence over goals and standards : as relatively high as do management personnel. In gen- f 4 eral, it would appear that the agreement between super- 5 visors and management is higher than the agreement be- tween rank and file and supervisors. To quantify the extent of agreement among levels in terms of rank order of influence dimensions, a Spearman rank order correla- tion was computed comparing all levels. These results are found in Table 38 below. {Table 38 Rank Order Correlations Comparing Organizational Levels on Part A and Part B (Beta). Level Comparisons 1 vs. 2 1 vs. 3 2 vs. 3 Part A .510 .515 .873 Part B -.106 -.269 .903 The correlations found in Table 38 (and to a degree in Table 28) were, to a great extent, dependent upmn1 the agreement between ranking of subordinate 85 influence and customer influence. Computation of rank order correlations without items 5 and 7 or 15 and 17 in most cases substantially increased the degree of corres- pondence among levels. For example, the comparison be- tween levels one and two on Part B increased from -.106 to +.554 with the elimination of items 15 and 17. F To summarize these analyses made with regard to Beta, several tentative conclusions may be drawn. First, although main effects for hierarchy and dimensions of influence were present, their interpretation should be ?1 contingent upon the interaction between hierarchical level, direction of influence, and dimension of influ- ence. Analysis of the interaction of dimension by items on a level by level basis indicated that perceived influ- ence was significantly different from item to item. In all levels, self-influence over work behaviors was judged to be higher than most other dimensions. In general, there appeared to be a great deal of similarity among dimensions in terms of ranking of different dimensions. This was particularly evident when comparing supervisors and management level members in Beta. In an overall sense, the results found in Alpha were in general agreement with those resulting from Beta responses. The omnibus analyses of variance were strikingly similar. In both organizations, significant main effects were noted for hierarchy and items varia- bles but in neither bank was direction of influence a 86 significant main effect. Similarly, both organizations displayed significant first order interactions of hier- archy by items and items by direction. Results differed in that Alpha results indicated an interaction between hierarchy by direction while Beta did not. Both organiza- tions showed a second order interaction among hierarchy, r2 items, and direction of influence. Subsequent post-hoc analyses indicated that the relationship between items and direction of influence was similar in both banks. With the exception of Alpha's management level, all other hierarchical levels showed an interaction effect between direction and item. Pairwise item mean comparisons indicated that, within the same organization, a degree of similarity in ranking occurred across levels of the organizational hierarchy. Rankings appeared to be more similar across levels for Alpha than for Beta and for Part A compared to Part B of the questionnaire. To quantify the relationship between organizations more clearly, a series of comparisons were made to assess the extent to which the same level in different organi- zations agreed on the rank order of items for both Part A and B of the questionnaire. A Spearman rho was com- puted based on the rankings for each of the ten ques- tions on Part A and Part B. The results are indicated in Table 39. As was the case with the other rank order com- parisons, items concerning subordinate and customer 87 Table 39 Rank Order Correlations for all Levels in Alpha and Beta. Levels Compared Alpha l-Beta 1 Alpha 2-Beta 2 Alpha 3-Beta 3 Part A .794 .455 .712 Part B .358 .694 .700 influence were the source of most disagreement. Compu- tation of the above correlations subtracting the effects of subordinate and customer questions substantially in- creased most of the above correlations. Even with the inclusion of these items there appears to be substantial agreement between organizational levels in the different banks as to the relative importance of influence dimen— sions. This substantial amount of agreement both among levels within the same organization and between organi- zations indicates a degree of generalization regarding the results of this study. The results described so far in this chapter have been addressed to four issues: 1. On the average, do hierarchical levels differ signif- icantly from each other in terms of overall perceived influence? 2. On the average, do dimensions of influence differ significantly from each other in terms of perceived influence? 88 3. On the average, do directions of influence differ significantly from each other in terms of perceived influence? 4. On the average, is the degree of perceived influence contingent upon specific hierarchical levels, dimen- sions of influence, and/or directions of perceived V” influence? 3 The foregoing series of analyses have indicated that issues one, two, and four should be answered in the affirmative. Issue three should, of course, be answered 2 in the negative; neither bank showed overall effects for directions of influence. Issue four acts as a modifier to the affirmative answers to questions one and two. B. Analysis Methods and Results Pertaining to Issues Five and Six The previous section of this chapter attempted to measure the relationships among different aspects of perceived influence and hierarchical level. The remain- ing issues posed in the Problem Statement (Chapter Two) addressed the relationships among dimensions of influence, hierarchical level, and job satisfaction. More specific- ally, two issues were posed: 1. Is job satisfaction significantly related to the degree of perceived influence associated with any one or combination of dimensions of perceived influ- ence? 89 2. Does the relationship, if any, between job satis- faction and dimensions of perceived influence vary significantly with different levels in the organiza- tional hierarchy? As was the case in answering the first four issues, the results for each organization will be analyzed separately and then results compared. To answer these issues concerning the relation- ships among hierarchical level, dimensions of perceived influence and job satisfaction, a series of multiple regression analyses were conducted. For each organiza- tion a patterned sequence of regressions was followed. To answer issue one above, a multiple regression involv- ing all twenty items with job satisfaction was conducted. Next, the difference between item pairs (ll-Ill, 12-112, etc.) were regressed on job satisfaction. This pair analysis was conducted to determine whether the difference between influence from others and influence on others was significantly related to job satisfaction. Both of these analyses ignored differences among organizational mem- bers in terms of hierarchical position. To answer the second issue posed in this section, rmJltiple regressions for all items (11 to I20) and item pairs.were conducted separately for each hierarchical level on job satisfaction. In terms of all responses from Alpha, both the regression of single items and item pairs indicated no 90 significant degree of association between any one or combination of items. So far as the entire organization was concerned, it appears that perceived influence over work behaviors bears little relationship to job satis— faction. Such was not the case when item and item pair scores were regressed on job satisfaction at each hier- fin archical level in Alpha. For item scores, neither level one nor two in Alpha displayed a significant regression on job satisfac- tion. The third level in Alpha, managerial level, did result in significant regressions. Table 40 summarizes the relationship between specific items and job satis- faction. Table 40 Multiple Regression of Items with Job Satis- faction (Managerial Level - Alpha). Item R R2 R2change r Beta 17 .628 .394 .394 -.628 -.907 3 .860 .739 .345 .346 1.158 12 .920 .846 .107 .012 -.907 5 .985 .969 .123 .168 .653 10 .996 .992 .022 -.433 .220 11 .998 .996 .004 -.201 -.111 Taken together, the combination of items 17, 3, 12, 5, 10, and 11 account for over 99% of the variance in job satisfaction scores for Alpha's management level. Items 10 and 11 account for little change in the multiple 91 regression. It would appear then that, in general, the less managers influence customers, the more they influ- ence rules and regulations; the less influence they have on technology and the more they influence subordinates, the greater will be their job satisfaction. Item pair regressions on levels one and two re- sulted in negligible regressions. Again, level three showed a significant regression on job satisfaction but only for item pair seven (item 7 - item 17). The differ- ence between how much customers and clients influenced the respondent and how much the respondent influenced customers and clients correlated .601 with job satisfac- tion and accounted for over 36% of the variance in job satisfaction scores for management level respondents. Thus, for those management level members who had cus- tomers or clients, the amount of perceived influence resulting from this dimension significantly related to job satisfaction. It will be recalled from Tables 34 and 36 that customers, as an influence dimension, was ranked low in terms of degree of perceived influence. Whatever little influence exists as a result of these dimensions appears to singularly affect job satisfaction. Responses of organization members at bank Beta were analyzed in the same manner as Alpha. First, with- out regard to hierarchical level, both item and item pair scores were regressed on job satisfaction. Like the results with Alpha, no significant relationships 92 were evidenced indicating that, as an organization, per- ceived influence makes little difference in terms of job satisfaction of respondents. This relationship, when measured level by level, did indicate some significant relationships among items and item pairs. So far as the rank and file workers at Beta are concerned, job satisfaction is significantly related to specific questionnaire items. Table 41 below indicates the degree of relationship among items and job satisfac- tion. Table 41 Multiple Regression of Items with Job Satis- faction (Rank and File Level - Beta). Item R R2 chhange r Beta 5 .581 .338 .338 .581 .924 17 .924 .854 .516 -.428 -.917 4 .983 .966 .113 .285 .421 16 .999 .998 .032 .217 -.236 In all, the degree of perceived influence exerted by or on these dimensions accounts for over 90% of the variance in job satisfaction score for rank and file workers at Beta. The relationship between subordinate influence and influence on customers in terms of job satisfaction is, of course, significant only for those rank and file workers who do have subordinates or inter- act with customers. Item 16, influence on other *4; 93 departments or sections, accounts for a minimal increase in the multiple regression and can probably be disre- garded. Item pair regression was also significant for rank and file workers in Beta. Item pair four (item 4 - item 14) correlated .57 with job satisfaction and accounted for 32% of the variance in job satisfaction scores for these respondents. Other item pairs added little additional correlation increments. Reference to Tables 32 and 35 indicates that, like Alpha, items which account for a significant proportion of variance in job satisfaction are rated rather low in degree of perceived influence compared to other dimensions. The results here indicate that the greater the difference between influ- ence on and from one's boss or supervisor, the greater the job satisfaction of rank and file workers in Beta. As was the case with Alpha, supervisory level members' responses showed little relationship between job satisfaction and specific items. Item pair regres- sions, in addition, were not significantly related to respondents' job satisfaction at the supervisory level. Level three, management personnel, did indicate a strong relationship between both items and item pairs and job satisfaction. Table 42 summarizes the regres- sion among items and job satisfaction for this level of Beta. 94 Table 42 Multiple Regression of Items with Job Satis- faction (Management Level - Beta). Item R R2 R2change r Beta 5 .356 .127 .127 .356 1.11 8 .776 .602 .475 -.343 -l.03 10 .968 .937 .335 .271 .639 For managers at Beta, the greater the influence of subordinates, the less one is influenced by the environment, and the more respondents influence their own work behaviors, the greater the job satisfaction in general. These three items, in total, account for almost 94% of the variance in job satisfaction scores among man- agers. Item pairs, in addition to single items, corre- lated significantly with job satisfaction for managers. More specifically, pair two and pair eight regressed significantly on job satisfaction. Table 43 displays the summary of the relationship between item pairs and job satisfaction for managers at Beta. Table 43 Multiple Regression of Item Pairs with Job Sat- isfaction (Management Level - Beta). 2 2 Pair R R R change r Beta 2, 12 .366 .134 .134 .366 .406 8, 18 .573 .328 .194 -.345 -.742 Not included in Table 43 are other item pairs which increase the multiple regression less than 10%. 95 The results of this table would appear to indicate that the greater the difference between influence from tech— nology and influence on technology and the smaller the difference between influence from the environment and influence on the environment, the greater will be the job satisfaction of managers in general. Tables 34 and 37 indicate that both machinery and environment are both rated fairly low in terms of degree of perceived influ- ence by managers. Two issues were posed at the beginning of this section pertaining to the relationships between perceived influence, hierarchical level and job satisfaction of organization members. The multiple regressions conducted in this section have suggested some possible responses to these issues. 1. So far as the total organization is concerned, there is little evidence in this study that perceived influ- ence, either as related to single items or in com- binations, significantly correlates with job satis- faction. 2. The hierarchical level of respondents does appear to make a difference in the relationship between per- ceived influence over work behaviors and job satis- faction. For both Alpha and Beta, managerial level respondents did indicate that specific items and item pairs related to job satisfaction. Rank and file workers at Beta also indicated a significant 96 relationship between specific items and item pairs with job satisfaction. Several rather broad conclusions can be reached as a result of the analyses conducted in this chapter. Measurement of the relationships among hierarchical level, dimensions of perceived influence, and directions of perceived influence indicate that the specific rela- tionships among these variables do not remain constant across all levels of the hierarchy. There do exist some regularities. For instance, almost all respondents indi- cated that they considered themselves as the primary determinant of their work behaviors, believed they were not influenced by co-workers, and believed that they influenced other departments very little. The relation- ships among dimensions also appear to be fairly consist- ent across organizations but the strength of the rela- tionship varies from level to level. Influence, in general, appears to bear little relationship to job satisfaction when the organization as a whole is considered. Given a specific hierarchical level, there is some evidence that the amount and direc— tion of influence significantly relates to job satisfac- tion but the nature of this relationship appears to be specific to a given level, organization, and dimension. Only one common dimension appeared in all significant regressions-~degree of perceived influence from subord- inates (item 5). Significantly, those item pairs which 97 did show a significant regression tended to be rated as exerting a relatively low degree of perceived influence. In general, those dimensions rated high in degree of perceived influence were not those which significantly related to job satisfaction. The relationships among the variables of this study, in sum, present a very contingent picture of their distribution and effects. wxi Chapter Five Conclusions and Implications The overall goal of this study was to measure and assess some specific relationships among perceived influ- ence, hierarchical level and job satisfaction. A review of the literature had indicated that two different approaches, the traditional and the contemporary, had developed. Central to the traditional approach to organ- izational influence was the identification of influence with formal authority resulting from an individual's hierarchical level. The contemporary approach, on the other hand, proposed that influence was a perceptual phenomenon, multivariate in nature. For the most part, the identification of formal authority with influence over work behaviors appears to be commonly accepted among organizational theorists and practitioners. The chain of logical assumptions--hier- archical level determines formal authority which then determines influence over work behaviors and, in turn, affects job satisfaction--is common in the organizational literature. The results of this study tend to indicate that this causal chain may not be without fault. In general, 98 99 it was found that the distribution of influence and its relationship to job satisfaction may be more contingent and situation specific than the traditional approach would suggest. Certain specific aspects of formal authority were perceived as being statistically more significant than other dimensions (e.g., rules, boss, goals and standards), but their importance varied with direction of influence and hierarchical level. In addition, some consistent relationships did exist both within and between organizations. Contrary to the assumptions of the traditional approach but con- sistent with the American image of self-reliance, re- spondents at all levels tended to judge themselves as the primary determinant of their own work behaviors. The next most consistent finding, more in keeping with the traditional approach, was that respondents in both banks indicated that they influenced subordinates and departmental goals and standards to a greater degree than other dimensions. The relationships among other dimensions and degree of perceived influence varied some- what with hierarchical level and bank, but a moderate degree of correlation between hierarchical levels and banks with respect to the dimensions of influence was noted. The relationships among dimension and direction of influence and job satisfaction, however, varied from the traditional conception. In general, little in the 100 way of a direct relationship existed between perceived influence over work behaviors and job satisfaction. Where specific relationships did exist, they appeared to be limited to a given level in the hierarchy, dimension of influence and organization. Only item five, influ- ence over subordinates, correlated significantly with job satisfaction for both Alpha's and Beta's managerial level. Other dimensions, while significant for a specific organizational level, showed little generality. In general, the results of this study tended to support aspects of both the traditional and contemporary approaches to organizational influence. The value of these results was enhanced somewhat by comparing results across banks and examining interorganizational consisten- cies and relationships. As organizations, banks may be more structured in terms of authority and role relation- ships than is the "typical" customer service organization. As such, they probably provided a conservative test of the traditional conception and thus allow a degree of generalization to the results of this study. The implications of these results will next be examined with respect to organizational theory, prac- tice, and future research applications. A. Implications for Organizational Theory The review of literature, Chapter One, pointed out that traditional conceptualizations of influence tended to identify formal authority with influence over work 101 behaviors. The results of the factor analyses conducted as part of this study tended to indicate that various aspects of formal authority such as rules and regulations, boss or supervisor, departmental goals and standards, etc. were judged as being only part of the influence construct. In general, it was indicated that influence over work behaviors was comprised of a single group factor which was significant in both organizations and one or two specific factors which did not replicate from bank to bank. Influence, as a construct, appears to incorporate several additional aspects, only some of which represent formal authority. It might be fruitful to take into con- sideration these various aspects of influence as proposed by the contemporary approach to influence. In addition to the classical formal authority approach to influence, aspects of more psychologically oriented theories (e.g., Tannenbaum, Argyris) concerning influence over work behaviors and attitudes might be re- examined. According to Tannenbaum's conception of organi- zational influence, the banks used in this study should have indicated a pattern of successively increasing control from lower levels to higher levels. While most classical theorists in addition to Tannenbaum would con- cur with this position, the results of this study did not bear this out. Inspection of Figures 1 and 2 indicatesa degree of response parallelism among levels 102 in both banks. If the results of this study are taken at face value, it might be that organizations which are objectively authoritarian in structure are perceived as being "democratic" in process by their members. This finding that individuals at all levels in each bank judge themselves as the major determinant of their work behaviors tends to support this conclusion. It would appear that the formal structure of organizational authority and judgments as to perceived influence are at variance. Perhaps formal theories of organizations should become more sensitive to the role of perceptual realities expressed by organization members. The important role of social co-worker influ- ences, much cited in the social science and behavioral literature, was not evident in this study. In most cases, respondents indicated that their co-workers had little influence on their work behaviors. On the other hand, individuals judged themselves as having a moderate degree of influence over their co-workers' behaviors. From a perceptual point of view, it might be that individuals are more prone to view themselves as a source of influence rather than a referent when co- workers were considered. It would appear that, in total, the results of this study do have some implication for organizational theory. Most of the findings of this study indicate that degree of perceived influence varies with 103 organizational level and dimension of influence. Tradi- tional theory appears to paint a rather simplistic picture of what, evidently, is a complex relationship. In terms of the relationship between influence and job satisfaction, traditional and contemporary con- ceptualizations also appear to be too general. Argyris (1964), for instance, maintains that, as control over work behaviors increases on the part of organization members, so should job satisfaction. A significant rela- tionship was found in this study between influence and job satisfaction but only for specific levels and specific dimensions of influence. As was the case in Tannenbaum's (1974) study, on an organization-wide basis no significant relationships existed between influence and job satisfaction. It was only when specific hier- archical levels were isolated that certain dimensions emerged as being statistically significant. So far as the organizations in this study were concerned, the relationship between perceived influence and job satis— faction was not as direct as Argyris' conceptualization would indicate. In spite of its democratic appeal, it would appear that influence over work behaviors, for the majority of the respondents, bears little relationship to their job satisfaction. The results of this study would indicate that additional research may be needed to identify more salient and general determinants of job satisfaction. 104 The value of such power—equalization programs as manage- ment by objectives, workers' councils, and job enrichment in improving job satisfaction may be fairly situational and limited. B. Implications for Organizational Practice The implications of this study for the practi- tioner in organizational behavior are several. Many organizational development programs (e.g., management by objectives, workers' councils, job enrichment) attempt to redistribute formal authority in a more egalitarian form within the organizational hierarchy. The results of this study indicate that this redistribution of formal author- ity may not directly increase organization members' perception of control over their own or others' work behaviors. For the most part, individuals at all organi- zational levels judged themselves as the primary deter- minant of their own work behaviors. Management's increas- ing the degree of control over work behaviors on the part of those members at the lowest levels of the hierarchy may have little benefit since they tend to perceive them- selves as already controlling their work to a large degree. Proponents of job enrichment and similar programs may be advocating approaches which result in limited benefits to the organization and its membership so far as control over work behaviors is concerned. Attempts to increase job satisfaction by means of increasing workers' control over work behaviors may also 105 be more contingent than most approaches to organizational development suppose. In both banks, only the management level respondents showed a significant relationship be- tween certain specific dimensions of influence and job satisfaction. The finding that Beta's rank and file workers also showed some significant relationship between certain dimensions of influence and job satisfaction may have limited implications. Beta, in general, appeared to be more paternalistic in management philosophy than did Alpha. A general conclusion, then, might be that, for management level members job satisfaction is tied in to perceived influence, but the specific dimensions may be organization contingent. C. Implications for Future Research As is the case with any research study, more questions were raised than were answered. Many factors may have affected the results other than differences in perceived influence over work behaviors: differences in sampling procedure, the possibility of biased responses, construct validity of the questionnaire, trait stability, etc. Continued research on the topics of influence, hierarchical differences, and job satisfaction may lead to more generalizable results and conclusive implications. Several specific recommendations for future research in this area of organizational behavior may be made: 106 Considerable attention should be given to the possi- bility of developing alternative methods for assessing perceived influence. While indications existed in this study that the questionnaire measured perceived influence, the likelihood exists that response bias and perceptual distortion could have consistently affected the results. Some method of assessing con- vergent validity through using two or more measures might resolve some measurement issues raised by this study. Conceivably, a behaviorally anchored rating scale, observational technique, or projective measure could be used to supplement or replace the question- naire used in this study. Aside from the methodological issues, several poss- ible areas of application might increase knowledge in this area. For example, differences in perceived influence might be examined in two or more contrast- ing organizations with different authority structures (e.g., voluntary vs. coercive, production vs. cus- tomer service). Future studies might incorporate union influence as a dimension. In certain types of organizations, the union may prove to be an important or dominant influ- ence over work behaviors. Future studies should also incorporate different measures of job satisfaction. Perhaps the Brayfield- Rothe Index limited the implications of this study 107 since it measured overall job satisfaction. Measure- ment of several factors involved in job satisfaction, such as the intrinsic-extrinsic distinction, might broaden the relationship between influence and job satisfaction. 5. An additional area for future investigation might center on changes in perceived influence patterns in relationship to different types of organizational development programs. If such programs as manage- ment by objectives, job enrichment, and workers' councils do increase the influence of organization members on their work, then it would be logical that perceived influence over work behaviors would increase as a result of such programs. Tannenbaum (1974) has advocated this approach in his treatment of organi- zational influence. In sum, the possibility of much additional research exists in this area of organizational influence. APPENDIX ,Appendix Research Questionnaire This study is concerned with your opinion of your job. The time and effort you take in filling out this questionnaire will help all of us interested in organizations to understand how organizations and people affect each other. There are three parts to this questionnaire. In the first part you are asked to indicate how much influ- ence others have over the way you do your work. In the second part, you are asked to indicate how much influ- ence you have over the work and activities of others. In the third part, you are asked to indicate what it is about your job that you like or don't like. All replies to this questionnaire are confiden— tial and will be used for research purposes 2211. Your responses will be tallied in with those of others. There will be no way of identifying any one individual's response in any report of this project. As soon as you have had a chance to complete this form, please hand it in at the place indicated. Thank you for your cooperation. 108 109 Job Code or Job Title Dept. or Section Part A How much do your co-workers, the people you work with, influence what you do at work and how you do it? “AAA/N vvvvv very little or no influence a little influence some influence a great deal of influence a very great deal of influence How much does machinery, tools, equipment, or phys- ical materials influence what you do at work and how you do it? “AAA“ VVVVV How very little or no influence a little influence some influence a great deal of influence a very great deal of influence much does "paperwork", official rules, regula- tions, and procedures influence what you do at work and how you do it? AAAAA VVVVV How w very little or no influence a little influence some influence a great deal of influence a very great deal of influence much does your boss or supervisor influence hat you do at work and how you do it? “AAA“ ) ) ) ) ) very little or no influence a little influence some influence a great deal of influence a very great deal of influence 9. 110 How much do your subordinates-—the people who report to you—-inf1uence what you do at work and how you do it? very little or no influence a little influence some influence a great deal of influence a very great deal of influence How much do other departments or sections influence what you do at work and how you do it? ( ) very little or no influence ( ) a little influence ( ) some influence ( ) a great deal of influence ( ) a very great deal of influence ow much do customers or clients influence what you H do at work and how you do it? no customers or clients very little or no influence a little influence some influence a great deal of influence a very great deal of influence AAAAAA vvvvvv How much does your physical work environment--the lighting, noise, cheerfulness, temperature, etc.-- influence what you do at work and how you do it? ( ) very little or no influence ( ) a little influence ( ) some influence ( ) a great deal of influence ( ) a very great deal of influence How much do the goals, standards, expectations, and performance criteria of your department or section influence what you do at work and how you do it? very little or no influence a little influence some influence a great deal of influence a very great deal of influence AAA/K" vvvvv 10. How you VVVVV How 111 much do you, as an individual, influence what do at work and how you do it? very little or no influence a little influence some influence a great deal of influence a very great deal of influence Part B much do you influence your co-workers, the people you work with, in determining what they do at work and how they do it? “AAA“ VUVVV How very little or no influence a little influence some influence a great deal of influence a very great deal of influence much do you influence the operation of the machinery, tools, equipment, or physical materials you “AAA“ vvvvv How work with? very little or no influence a little influence some influence a great deal of influence a very great deal of influence much do you influence "paperwork," official rules, regulations, and procedures? “AAA“ vvvvv How very little or no influence a little influence some influence a great deal of influence a very great deal of influence much do you influence your boss or supervisor in determining what he or she does at work and how it is done? AAAAA vvvvv very little or no influence a little influence some influence a great deal of influence a very great deal of influence 9. 112 How much do you influence your subordinates, the people who report to you, in determining what they do at work and how they do it? no subordinates very little or no influence a little influence some influence a great deal of influence a very great deal of influence AAAAAA VVVVVV How much do you influence what other departments or sections do and how they do it? ( ) very little or no influence ( ) a little influence ( ) some influence ( ) a great deal of influence ( ) a very great deal of influence ow much do you influence what the customers or lients do and how they do it? ()3: no customers or clients very little or no influence a little influence some influence a great deal of influence a very great deal of influence VVVVVV How much do you influence your physical work environ- ment--the lighting, noise, cheerfulness, temperature, very little or no influence a little influence some influence a great deal of influence a very great deal of influence How much do you influence the goals, standards, expectations, and performance criteria of your department or section? ) very little or no influence ) a little influence ) some influence ) a great deal of influence ) a very great deal of influence “AAA“ 113 10. How much do ou, as an individual, influence the work behavior and activities of others? ) very little or no influence ) a little influence ) some influence ) a great deal of influence ) a very great deal of influence Part C Some jobs are more interesting and satisfying than others. We want to know how people feel about different jobs. This blank contains eighteen statements about jobs. You are to cross out the phrase below each state- ment which best describes how you feel about your present job. There are no right or wrong answers. We should like your honest opinion on each one of the statements. Work out the sample item numbered (0). 0. There are some conditions concerning my job that could be improved. STRONGLY AGREE AGREE UNDECIDED DISAGREE STRONGLY DISAGREE 1. My job is like a hobby to me. STRONGLY AGREE AGREE UNDECIDED DISAGREE STRONGLY DISAGREE “AAA“ VVVVV 2. My job is usually interesting enough to keep me from getting bored. STRONGLY AGREE AGREE UNDECIDED DISAGREE STRONGLY DISAGREE AAAAA vvvvv 3. It seems that my friends are more interested in their jobs. STRONGLY AGREE AGREE UNDECIDED DISAGREE STRONGLY DISAGREE “AAA“ VVVVV 114 I consider my job rather unpleasant. STRONGLY AGREE AGREE UNDECIDED DISAGREE STRONGLY DISAGREE I enjoy my work more than my leisure time. ( ) STRONGLY AGREE ( ) AGREE ( ) UNDECIDED ( ) DISAGREE < ) STRONGLY DISAGREE I am often bored with my job. ( ) STRONGLY AGREE ( ) AGREE ( ) UNDECIDED ( ) DISAGREE ( ) STRONGLY DISAGREE I feel fairly well satisfied with my present job. ( ) STRONGLY AGREE ( ) AGREE ( ) UNDECIDED ( ) DISAGREE ( ) STRONGLY DISAGREE Most of the time I have to force myself to go to work. STRONGLY AGREE AGREE UNDECIDED DISAGREE STRONGLY DISAGREE AAAAA VVVVV H am satisfiedudth my job for the time being. STRONGLY AGREE AGREE UNDECIDED DISAGREE STRONGLY DISAGREE “AAA“ VVVVV 10. 11. 12. 13. 14. 15. 115 I feel that my job is no more interesting than others I could get. STRONGLY AGREE AGREE UNDECIDED DISAGREE STRONGLY DISAGREE VVVVV H definitely dislike my work. STRONGLY AGREE AGREE UNDECIDED DISAGREE STRONGLY DISAGREE “AAA/N VVVVV I feel that I am happier in my work than most other people. STRONGLY AGREE AGREE UNDECIDED DISAGREE STRONGLY DISAGREE VVVVV Most days I am enthusiastic about my work. STRONGLY AGREE AGREE UNDECIDED DISAGREE STRONGLY DISAGREE “AAA/K VVVVV Each day of work seems like it will never end. STRONGLY AGREE AGREE UNDECIDED DISAGREE STRONGLY DISAGREE “AAA“ VVVVV l—l like my job better than the average worker does. STRONGLY AGREE AGREE UNDECIDED DISAGREE STRONGLY DISAGREE “AAA“ VVVVV 116 16. My job is pretty interesting. STRONGLY AGREE AGREE UNDECIDED DISAGREE STRONGLY DISAGREE “AAA“ VVVVV 17. H find real enjoyment in my work. STRONGLY AGREE AGREE UNDECIDED DISAGREE STRONGLY DISAGREE “AAA“ 18. H am disappointed that I ever took this job. STRONGLY AGREE AGREE UNDECIDED DISAGREE STRONGLY DISAGREE “AAA“ VVVVV REFERENCES References Argyris, C. The Applicability of Organizational Soci— F] ology. London: Cambridge University Press, 1972. Argyris, C. Personality and Organization. New York: Harper and Row, 1957. 1 ‘ . ”fi“r. Argyris, C. Integrating the Individual and the Organi- zation. New York: Wiley and Sons, 1964. ; Blauner, R. Alienation and Freedom. Chicago: Univer- sity of Chicago Press, 1964. Bass, B. M., and Barrett, G. V. Man, Work, and Organ- izations. Boston: Allyn and Bacon, 1972. Bennis, w. Beyond Bureaucracy. New York: McGraw- Cartwright, D. Influence, leadership, and control. Handbook of Organizations, J. G. March (ed.). Chicago: Rand McNally, 1965, pp. 1-47. Child, D. The Essentials of Factor Analysis. New York: Holt, Rinehart, and Winston, 1970. Dalton, G. w., Barnes, L. B., and Zaleznik, A. The Distribution of Authority in Formal Organiza- tions. Cambridge: M.I.T. Press, 1968. Dalton, G. w., and Lawrence, P. R. Motivation and Controlin Organizations. Homewood: Irwin- Dorsey, 1971. Filley, A. C., House, R. J., and Kerr, S. Managerial Process and Organizational Behavior. Glenview, 1976. Gibson, J. L., Ivancevich, J., and Donnelly, J. H. Organizations: Structure, Process, and Behavior. Dallas: Business Publications, Inc., 1973. 117 118 Grusky, O. and Miller, G. The Sociology of Organiza- tions. New York: Free Press, 1970. Hackman, J. R. Toward understanding the role of tasks in behavioral research. Acta Psychologica, Harmon, H. H. Modern Factor Analysis. Chicago: Uni- versity of Chicago Press, 1970. Hulin, C. L. Individual differences and job enrichment: the case against general treatments. In New Perspectives in Job Enrichment, J. R. Maher (ed.). New York: Van Nostrand Rinehold Press, 1971, pp. 159-171. Hulin, C. L. and Blood, M. R. Job enrichment, individual differences and worker responses. Psychological Bulletin, 1968, 62, pp. 44-51. Kelly, J. Organizational Behavior. Homewood: Richard Korman, A. 5. Industrial and Organizational Psychology. New York: Prentice-Hall, 1971. Lawrence, P. R. and Lorsch, J. w. Differentiation and integration in complex organizations. Adminis- trative Science Quarterly, 1967, 12, pp. 1-47. Leavitt, H. J. Managerial Psychology. Chicago: Univer- sity of Chicago Press, 1976. Likert, R. The Human Organization. New York: McGraw- Hill, 1967. Litterer, J. The Analygis of Organizations. New York: Wiley and Sons, 1973. Linton, M. and Gallo, P. S. The Practical Statistician. Monterey: Brooks-Cole, 1975. McCormick, E. J. and Tiffin, J. Industrial Psychology. Englewood Cliffs: Prentice-Hall, 1974. McGregor, D. The Human Side of Enterprise. New York: McGraw-Hill, 1960. Mechanic, D. Sources of power of lower participants in complex organizations. Administrative Science Quarterly, 1962, 1, pp. 349-364. 119 Melcher, A. J. Structure and Process of Organizations: A Systems Approach. Englewood Cliffs: Prentice- Hall, 1976. Miner, J. B. The Management Process: Theory, Research, and Practice. New York: Macmillan, 1973. Nunnally, J. C. Psychometric Theory. New York: McGraw- Pelz, D. C. Influence: a key to effective leadership in the first line supervisor. Personnel, 1952, 22, pp. 209-217. Porter, L. w., Lawler, E. E., and Hackman, J. R. Behavior in Organizations. New York: McGraw-Hill, 1975. Ritchie, J. B. Supervision. In Organizational Behavior: Research and Issues by G. Strauss, R. E. Miles, C. C. Snow, and A. S. Tannenbaum (eds.). Madi— son: Industrial Relations Research Association, 1974, pp. 51-76. Robinson, J. P., Athanasiou, R., and Head, K. B. Meas- ures of Occupational Attitudes and Occupational Characteristics. Ann Arbor: Institute for Social Research, 1969. Rosen, H. and Weaver, C. G. Motivation in management: a study of four management levels. In W. K. Graham and K. H. Roberts (eds.) Comparative Studies in Organizational Behavior. New York: Holt, Rinehart, and Winston, 1972, pp. 15-25. Stagner, R. Perception: applied aspects. In The Psych- ology of Industrial Conflict. New York: John Wiley and Sons, 1956. Sutermeister, R. A. People and Productivity. New York: McGraw-Hill, 1976. Tannenbaum, A. S. Control in Organizations. New York: McGraw-Hill, 1968. Tannenbaum, A. S. and Kahn, R. L. Organizational control structure. Human Relations, 1957, 10, pp. 127- 140. ‘- Tannenbaum, A. S., Kavcic, B., Rosner, M., Vianello, M., and Wiesener, G. Hierarchy in Organiza- tions. San Francisco: Jossey-Bass, 1974. 120 Vroom, V. H. Work and Motivation. New York: John Wiley and Sons, 1964. Winer, J. B. Statistical Principles in Experimental Design. New York: McGraw-Hill, 1962.