THE CONS!DERAT§ON 'AND mmmwe STRUCTURE RELAUONSHIP: e f ‘ - POTENTEAL MODERATOR VAREABLES Thesis for the Begree 0f MA. -.' mum STATE UNWERSITY _ GREGORY EDWARD HUSZCZO- ‘ 1974 .. ""‘.?‘.’ LI B RA RY Michigan $3” Univcfilty ( ' “a?” M. . ‘5' MW“ av 3‘ ~ HUME & SUNS' ‘ BUUK BINDERY INC. LIBRARY BINDE ‘5 ABSTRACT THE CONSIDERATION AND INITIATING STRUCTURE RELATIONSHIP: POTENTIAL MODERATOR VARIABLES By Gregory Edward Huszczo This thesis focused on three areas concerning leadership: l. The relationship between Consideration and Initiating Struc- ture and the allegation that ggfl;_and l;§;_are independent. 2. The possibility that some individual differences variables (particularly, personal background characteristics) moderate the relationship between Cgfl;_andilL§L 3. The possibility that l;§;_is a moderator of the consideration- satisfaction relationship and Cgfl;_is a moderator of the initiating structure-satisfaction relationship. An overview of leadership research focusing on studies involving Consideration and Initiating Structure was provided. House's Path-Goal theory of leadership was reviewed in depth because it combines the Cgfl;: 1.5. leadership research with the need to investigate the effects of moderator variables in leadership research. The subjects of this study were 232 hourly production workers in large automobile plants. Data were collected by questionnaire using the following instruments: the Consideration and Initiating Structure Gregory Edward Huszczo scales developed by Stogdill, l965 (LBDQ Form XII), a Biographical Data Blank developed for the Chrysler Foreman Training Project, and the Job Involvement, Job Identification scales developed by Ruh, Johnson, and Scontrino [1973]. All of these scales are provided in the appendices of this thesis. CON;_and I;§;_were found to be highly correlated. Several explanations are offered and serious questionning of the appropriateness of items on these scales with no concern to situational variables is discussed. None of the variables tested succeeded in moderating the Qgfl;;l;§;_relationship. Also, CON; failed to moderate the l;§;r Satisfaction relationship and I;§;_failed to moderate the £954: Satisfaction relationship. Several explanations are offered, focusing on the measurement "problems" the Qgfl;_and_L£§; scales have in a sample like that used in this study. THE CONSIDERATION AND INITIATING STRUCTURE RELATIONSHIP: POTENTIAL MODERATOR VARIABLES By Gregory Edward Huszczo A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF ARTS Department of Psychology 1974 ACKNOWLEDGMENTS I would like to publicly thank some people whose help and support not only aided the completion of this thesis but also aided my personal development. Certainly, at the top of such a list would be my parents. Their acceptance of my endeavors and the constant support and caring they have given to me all my life has been exactly what I needed and desired. I thank you both, Mom and Dad. My only repayment is a love I give freely to you. I wish to also thank my committee members who contributed professionally as well as personally throughout my efforts. Dr. Jack Wakeley, my chairman, helped make me appreciate clear, concise writing. His patience and encouragement helped me through many frustrations. Dr. Terry Allen helped me believe in my own ideas. Our con- versations of the last couple of years opened up some new avenues of thinking and led me to value "making sense" out of data as more impor- tant than rigidly following rules of interpretation. Dr. Michael Moore contributed greatly in getting this thesis under way. A course I took from him turned me on to some new ideas and literature. His encouragement helped me follow those ideas. I would like to give a special thanks to Bill Brown whose assistance in computer programming was crucial in completing this thesis. I would also like to "thank" another computer consultant who ii spent all my allotted computer money doing the wrong analysis on the wrong set of data during my first attempt at completing this thesis. A word of thanks is in order for Drs. Robert Ruh and Jack Dawson who collected the data for this study and volunteered their time to explain their efforts to me. This thesis is a smaller study within the larger project these two men were administering. Their encouragement was very helpful. Finally, but certainly not least, I would like to thank Mrs. Sue Neesner, who typed the drafts in the development of this thesis. She made me realize the importance of a good secretary. TABLE OF CONTENTS Page LIST OF TABLES ......................... v CHAPTER I. A LOOK AT THE LITERATURE ............ . . . . l The Behavioral Approach and the Ohio State Studies 2 Development of the LBDQ . . . . ............ 4 Consideration and Initiating Structure ........ 5 Major Reviews of CON. and I. S. Research ..... 8 Moderator Variables of the_ CON. and I. S. Relationship. l4 House' 5 Leadership Theory ........... . . . . 15 Conclusions from Literature Review .......... 25 II. METHODS ......................... 27 Subjects . . . .................... 27 Instruments ...................... 28 Analyses ....................... 29 III. RESULTS ......................... 33 IV. DISCUSSION ....................... 39 Worthy of Note ........... . . . . . . . . . 39 The CON. -I. S. Relationship .............. 39 Invest1gation of Potential Moderator Variables . . . . 47 CON. , I. S. and Satisfaction Measures .......... 49 Summary ................ . ....... 5l LIST OF REFERENCES ..................... . . 53 APPENDICES A. CONSIDERATION SUBSCALE OF LBDQ--FORM XII ........ 58 B. INITIATING STRUCTURE SUBSCALE 0F LBDQ--FORM XII ..... 59 C. CHRYSLER FOREMAN TRAINING RESEARCH PROJECT-— BIOGRAPHICAL DATA BLANK ................ 60 D. SATISFACTION SCALES ................... 61 E. ITEM AND SCALE INTERCORRELATION MATRIX WITH COMMUNALITIES IN DIAGONAL AND CORRELATIONS CORRECTED FOR ATTENUATION . . . . . .......... 62 iv LIST OF TABLES Table Page 1. Percent of common variance accounted for by four factors ........................ 5 2. Means, standard deviations, and reliabilities of 3.1. and CON. scores in studies using form XII of the LBDQ— .......................... 29 3. Dichotomization of potential moderator variables to be used .......................... 3l 4. Means, standard deviations, and reliabilities of scales used in this study ........... . ....... 33 5. Correlation matrix for total sample (N = 232) ...... 35 6. Results of investigation of potential moderator variables of CON. -L. S relationship using validation sample (odd- numbered— subjects only,n ll7). . . ....... 36 7. Results of investigation of potential moderator variables of CON.-I.S. relationship using total sample (N = 232) . 37 8. Results of investigation of use of I. S. and CON. as moderator variables in L. S. -satisfaction an EC CON. - satisfaction relationships ............... 38 9. Factor analysis with varimax rotation .......... 43 10. Factor analysis with principle component solution . . . . 44 CHAPTER I A LOOK AT THE LITERATURE Leadership is a concept considered important and frequently encountered but seldom understood. Literally thousands of studies, articles, and books have dealt with leadership and its various aspects. It is recognized as a very important factor in the welfare of our country today. The health of our economy, the progress of our people, and the success of all sorts of organizations depends to some extent upon the quality of leadership associated with them. It would seem that with so much value and interest attached to good leadership, many clearcut facts would be known concerning this phenomenon. Unfortunately, for a variety of reasons, such is not the case. Lack of agreement over definitions, lack of consistent use of various concepts across studies, exclusion of factors in some studies, inclusion of factors in others have all led us to the problem of lack of generalizable knowledge con- cerning leadership. As a result, there is a great mass of literature in the field which has little organization, few common assumptions, and many different theoretical and methodological approaches. A possible major reason for this situation of confusion is the apparent "faddish" approach to the study of leadership. A review of the literature reveals the major approaches to be the personality- trait approach (of the "Great Man Theory"), the peer-nomination approach (sociometric studies), the leaderless temporary group approach (leader-emergent studies), the participation-influence approach (Scanlon Plan studies, effects of participation studies), the types of power approach (referent, expert, etc.), the role-function approach (small group studies), the leadership and organizational out- puts approach, the behavioral approach (description studies), and the situational variables approach (Fiedler). Each of these approaches seem to be inspired by (a) successful finding(s) which set off a rash of studies using that approach. This has left us with several problems including subareas within leadership that have no overlap between approaches, apparent "loose-ends" left unresolved within individual approaches, and some difficulty in replicating some results of researchers using individual approaches. The Behavioral Approach and the Ohio State Studies In the last twenty-five years perhaps the most dominant approach (based on amount of literature generated) is the behavioral approach. The behavioral approach is quite simple in its philosophy. It states that the best way to study and to define leadership is in terms of what leaders g9_rather than in terms of what leaders are. Thus, one is concerned with leader "behaviors" rather than leader traits. In outline form, the behavior approach: l. Collects critical incidents of good and bad leadership behavior. 2. Scales each incident to determine, in the eyes of ex- perts, how "good" or how "bad" each behavior is judged to be. 3. Develops a check-list type questionnaire . . . . A leadership "score" can be computed by using the median scale value of the behaviors which have been checked [Blum & Naylor, 1968, p. 421]. The major impetus and best example of this approach is the Ohio State Leadership Studies. Begun in 1945, this interdisciplinary endeavor attempted to describe the behavior of leaders in terms of: l) Aflhag does an individual do while he operates as a leader, and 2) fl2w_does he go about what he does? [Hemphill & Coons, 1957, p. 6]." "While this program was responsible for a variety of signifi- cant findings, it is quite likely that the most important contribution was isolation of 'Consideration' and 'Initiating Structure' as basic dimensions of leadership behavior in formal organization [Korman, 1966, p. 349]." These dimensions emerged, through factor-analytic pro- cedures, in the develOpment of a measuring instrument named the Leader Behavior Description Questionnaire (LBDQ) during these Ohio State studies. Since the purpose of this thesis is to provide further investigation of the two concepts Consideration and Initiating Struc— ture, it is appropriate to take a detailed look at the development of the LBDQ. When the Ohio State Leadership studies were initiated in 1945, "no satisfactory theory or definition of leadership was available.“ [Shartle, l957, p. l]. To provide a starting point, the research) staff defined leadership as "the behavior of an individual when he is directing the activities of a group toward a shared goal." They decided ng§_to define leadership as good leadership to avoid criterion contamination since their goal was the description of leadership first and then its evaluation as a separate research project. The Leader Behavior Description Questionnaire was its major outcome. Development of the LBDQ The LBDQ started with 1800 items generated by the staff members of the Personnel Research Board at Ohio State and two advanced univer- sity classes and was reduced to 150 items by "expert judges" who sorted them into nine "a priori" dimensions of leadership (Initiation, Member- ship, Representation, Integration, Organization, Domination, Communica- tion, Recognition, and Production). This original form was administered to several groups and the results showed high correlations among the dimensions. A factor analysis was performed to reorganize the items into fewer and more independent categories of leader behavior. This factor analysis was performed on data from 300 Air Force crew members describing 52 air crew commanders. The results may be seen in Table l. The two major factors (accounting for more than 80 percent of the com- mon variance) were labeled "Consideration" and "Initiating Structure." The items that loaded on the "Consideration" factor described behavior like friendship, mutual trust and respect, and good "human relations" between the leader and his group. The items that loaded on the "Initiating Structure" factor described behavior such as organizing and defining the relationships between himself and the group, in defining interactions among group members, establishing production methods and scheduling. Fleishman [l953] used these findings to develop a form of this questionnaire to be used in industry. This form is called the Supervisory Behavior Description (530) questionnaire and its final form consisted of twenty-eight items on the Consideration dimension and twenty items on the Initiating Structure dimension. This form Table l. Percent of common variance accounted for by four factors [Factor Analysis of Halpin & Niner, 1954, p. 41]. Factor Factor Percent of Number Designation Common Variance I Consideration 49.6 II Initiating Structure 33.6 III Production Emphasis 9.8 IV Sensitivity (Social Awareness) 7.0 and slight variations of it has been widely used in empirical research, particularly in military organizations, industry, and education. A questionnaire parallel to the $80 that also measures the two dimensions Consideration and Initiating Structure is the Leadership Opinion Ques- tionnaire [Fleishman, 1957]. However, this form deals with the per- ceptions of the leaders themselves of how they should act instead of the perceptions of the subordinate of how their leaders act. The form used in this thesis is called Form XII of the LBDQ. It represents further revisions of the LBDQ most notably by Stogdill [1962]. Details of this particular form will be presented in the methods section of this thesis. Consideration and Initiating Structure Fleishman and Harris [1962, p. 43-44] provided definitions of the two dimensions the studies at Ohio State have isolated. They are: Consideration includes behavior indicating mutual trust, respect, and a certain warmth and rapport between the super- visor and his group. This does not mean that this dimension reflects a superficial 'pat-on-the-back,‘ 'first name calling' kind of human relations behavior. This dimension appears to emphasize a deeper concern for group members' needs and includes such behavior as allowing subordinates more partici- pation in decision making and encouraging more two-way com- munication. Initiating Structure includes behavior in which the supervisor organizes and defines group activities and his relation to the group. Thus, he defines the role he expects each member to assume, assigns tasks, plans ahead, establishes ways of getting things done, and pushes for production. This dimension seems to emphasize overt attempts to achieve organization goals. Similar definitions are provided by Fleishman and Peters [1962], Halpin [1954], Hemphill [1955], and Rambo [1958]. The two scales are said to be reliable (Fleishman, Harris and Burtt, 1955 claim test- retest reliabilities over eleven months of .58 and .87 for consideration and .46 and .75 for initiating structure. Also, split—half relia- bilities showing internal consistency have been reported for various samples ranging between .68 and .98). Attempts to establish validity for the two scales have been made by correlating these scales with independent measures of leadership effectiveness and organizational criteria. [See Stogdill and Coons monograph, 1957 for details]. Most notably, Fleishman and Harris [1962] found a relationship between structure and satisfaction measures (i.e., absenteeism, turnover, and grievance rates). They also found that structure moderates the rela- tionship between consideration and these measures. The two scales are claimed to be independent [Fleishman, 1953; Halpin and Miner, 1957] of each other based on the fact that consideration and initiating structure were orthogonal factors in factor analyses of the original items. Conceptually, the leader's behavior with respect to one dimen- sion does not necessarily affect his behavior with respect to the other. Fleishman reports correlations of-.O5 (for a sample of 176 ROTC students who described their superior officers), -.02 (for a sample of 122 foremen who described their own supervisors), and -.33 (for a sample of 394 workers who described the 122 foremen in the above mentioned sample). This last correlation might lead the reader to doubt the true independence of these two scales. Fleishman [1969] defended the claim of independence of the two scales against the reports of significant correlations occurring between them by arguing that there exists positive, negative and non-significant relationships but the median correlation of these studies would be the true value of zero and the significant correlations that have been found repre- sent a simple sampling distribution around that true value. Korman [1966], Weissenberg and Kavanagh [1972], and Lowin, £1L21;.[1959] all present evidence that the independence of the two scales is dubious at best. Since the major purpose of this thesis is to investigate the relationship between consideration and initiating structure, let us take a look at the evidence that the above-mentioned authors present. Major Reviews of CON. and 1.5. Research Korman [1966] provides an extensive review of the research involving consideration and initiating structure. Korman's review deals with the correlations found between these two dimensions and various measures of effectiveness. Korman comes to fairly negative conclusions after reviewing sixteen studies that used the LOQ and sixteen studies that used the LBDQ. The studies that used the LOQ resulted in mostly insignificant correlations for both variables with the effectiveness criteria. "Whatever trend that does exist seems to indicate that consideration might have some relation to a 'pleasantly affective' work situation [Korman, 1966, p. 351]." The results using the LOQ for measurement of Initiating Structure are rather inconsist- ent and the only pattern that shows at all is a prevalence of low correlations. Studies using the LBDQ tended to show "a slightly more consistent pattern of Consideration being related to effective per- formance positively and Initiating Structure negatively, but there is a great degree of inconsistency even among studies using somewhat similar populations [Korman, 1966, p. 354]." Korman also points out that almost all the studies performed on these dimensions are of the concurrent validity variety. Thus, even in the cases where there are significant relationships between leader behavior and effectiveness, there is no evidence whether the variation on these scales is a pre- dictor of different kinds of worker behavior or whether the reverse of this could be the case. Furthermore, Korman points out that the designs of these studies generally just correlate the scale scores with the criterion variables without including situational variables that may be moder- ating these relationships. He concludes that future research of these variables must not only recognize the possible effects of situational variables, but must also systematically conceptualize what the variance due to these variables mean in relationship to leadership behavior. House [1971] attempts to do just that. Attention will be given to his propositions later in this review. Anderson [1966] also reviewed the literature on l;§;_and Qgfl;_ though less exhaustively than Korman. He also comes to the conclusion that studies investigating variables that may moderate the relation- ships between the leader behavior dimensions and effectiveness criteria are sadly lacking. The focus of his paper is his cross-cultural corre- lational study using these dimensions. He points out that differing results occur when the data are analyzed separately within cultural groups. He explains this by speculating that the subordinates had dif- fering role-expectations of the leaders. This is a suggestion that a personal background variable (culture) may moderate relationships involving consideration and initiating structure. Anderson concludes that in almost any situation, a subordinate holds some expectations as to the "proper" level of structuring to be provided by his supervisor. When leader behavior exceeds this expectation, it is likely to be per- ceived by the subordinate as an illegitimate, uncalled for imposition. Weissenberg and Kavanaugh [1972] reviewed seventy-two studies using measures of Consideration and Initiating Structure. Of these 10 they found thirty-seven (51 percent of the studies reviewed) reporting significant positive relationships between the two leadership dimen- sions, seven (10 percent) reporting significant negative correlations, and twenty-eight (39 percent) reporting nonsignificant correlations. This surely throws doubt on the claim that Consideration and Initiating Structure are orthogonal. Furthermore, Weissenberg and Kavanaugh point out that the median for all the studies they reviewed was .36. This finding negates Fleishman's argument for a median of zero and a simple sampling distribution around that true score. They went further and discovered that studies that used the LOQ to measure Consideration and Initiating Structure generally showed independence between those two variables (of the twenty-four studies reviewed, three had significant positive relationships, five had significant negative relationships and sixteen had non-significant relationships). The forty-eight studies reviewed using the LBDQ, however, showed thirty-four positive relation- ships, two negative relationships, and twelve nonsignificant relation- ships between Consideration and Initiating Structure. Remember, the LOQ is a self-report, attitudinal questionnaire which asks the respond- ent to describe his ideal leadership behavior. The LBDQ asks the subordinates to describe the actual behavior of the leader as they perceive it. Thus, just whose perceptions of leader behavior one uses seems to effect what type of relationship will be found between the concepts Consideration and Initiating Structure. Weissenberg and Kavanaugh also found a situational variable that seems to moderate the relationship between CON;_and.;£§L, i.e., organizational level. Of these thirty-eight studies reviewed, sixteen of the twenty-seven studies using first level supervisors reported ll significant relationships while only four of the eleven studies done on managers above the first level reported significant correlations. Lowin, Hrapchak, and Kavanaugh [1969] performed one of the only experimental endeavors studying the relationships of Consideration and Initiating Structure with each other and with measures of productivity and job satisfaction. Their use of an experimental design provides some advantages that are not available in the correlational studies. First, it provides an opportunity to deal with the meaning of each concept as well as the distinction between consideration and initiating structure because the experimental conditions represent conceptual distinctions. Second, an experiment provides the opportunity to inves- tigate directionality of relationships. Third, experimentation allows for closer examination of the perceived relationship between ggfl;_and 14§;_and "thus allows for a clarity (not simplicity!) of interpreta- tion often unattainable in conventional observational research [Lowin gt_gl_._, 1969, p. 241]." Lowin gt_alL, generated three hypotheses: 1) That consideration and initiating structure will be orthogonal when subordinates rate their supervisors, 2) that productivity and quality of work of the subordinates will positively correlate with both consideration and initiating structure of the leader, and 3) that job satisfaction of the subordinate will correlate positively with consideration. They used a 2 x 2 x 2 design where subjects received either a high or low consideration treatment, a high or low initiating structure treatment and with a productivity or quality orientation. The consideration and initiating structure treatments were accomplished by use of scripts 12 formed by using behaviors listed in the LBDQ as high or low consideration behaviors. The productivity or quality orientation was also transmitted verbally to the subjects. The job each subject worked on was a repetitive job of gapping spark plugs. After an interview with the experimenter each subject was hired and then was subjected to one of the treatment combinations. Subjects did not know they were in an experiment, they thought they were applying for a part- time job advertised in the newspaper. All subjects were debriefed after the experiment and paid for their time. The dependent variables studied in this experiment were perceived consideration, perceived initiating structure (using items from the LBDQ), productivity (number produced), quality (errors and deviations), and job satisfaction (using the Brayfield-Roth Index). The results of this experiment give evidence that the manipula- tions of the variables were effective, that the subordinates perception of consideration was effected by his perception of initiating structure but not vice-versa, that consideration had an effect on productivity, quality, and job satisfaction, and that is all. No other main or interaction effects were significant. Lowin §3_al;_reanalyzed their data to use perceived initiating structure and consideration scores to define treatment cells rather than assuming the treatment effect. The new analysis produced very similar findings except the effect con- sideration had on productivity was reduced to nonsignificance but the effect consideration had on job satisfaction increased to significance. Lowin gt_al; conclude that consideration and initiating struc- ture are interdependent under certain circumstances. They speculate 13 that there is reason for anticipating such a relationship. The exercise of authority by supervisor may force some degree of emotional distance between himself and his subordinates. A negative correlation between consideration and initiating structure would then be expected. (This study resulted in a correlation between §9N4_and l;§;_of -.20). However, they cite Anderson's [1966] notion that subordinates have an expectation of what is the "proper" level of structuring for a given situation. When actual structuring exceeds that expectation, the sub— ordinate feels imposed upon and thus will effect his perception of the consideration level of the supervisor. Variables like subordinate competence and task simplicity would surely effect the intercorrelation of CON;_and 1;§;_then. The correlation between EON;_and l;§;_would also be effected by the extent of a halo effect in the situation and any effect due to the method used to measure the variables. The effect that consideration had on job satisfaction was expected and the extent to which job satisfaction was motivational may explain the relationship between consideration and productivity and quality. The lack of correlation between productivity and initiating structure was explained by the authors as the result of the simplicity of the task. The task was simple enough that imposed structure could not improve performance. This is parallel to what House [1971] pre- dicts in his theory that will be discussed later in this review. Lowin gt;31;_re-emphasized the need for further research investigating the effect moderator variables such as job level and other job char- acteristics have on relationships involving consideration and initiating structure. 14 The same article by Lowin gt_gl;_[l969] reports another study they performed concerning the perception of supervisory behavior. In this study they wanted to investigate the above-mentioned finding that perceptions of consideration were affected by level of initiating structure but not vice-versa. A panel of thirty-three male undergrad- uates, unfamiliar with the concepts, were trained to rate behaviors of consideration and initiating structure. The results of rating tests involving leader behavior gave evidence that the judges were able to learn the concepts and reflected a marked negative correlation between perceptions of consideration and initiating structure. To test whether perceptions of consideration were affected more by initiating structure than vice-versa several analyses were performed to discover if judges confused consideration or initiating structure items more often. The results seemed to indicate that the judges perceived consideration as affecting initiating structure as much as initiating structure affected consideration. Moderator Variables of the CON. and 1.8. Relationship A common theme of Korman [1966], Anderson [1966], Lowin, 22.21; [1969], and Weissenberg and Kavanaugh [1972], is that there are moder- ator variables affecting the relationships between Consideration and Initiating Structure with each other, and these dimensions, with measures of effectiveness. Katzell, Barrett, and Parker [1961] suggest specifically that effectiveness of particular kinds of leadership practices might be moderated by such variables as the size of the company and the degree of urbanization where the company is located. 15 Vroom [1964] feels that the effectiveness of particular kinds of supervisory practices would be moderated by the wishes and expectancies of the subordinates. House [1971] takes directly off from this specu- lation of Vroom's. He uses expectancy theory to explain how leader behavior affects the subordinates and thus the measures of effective- ness commonly studied in the literature on Consideration and Initiating Structure. His theory asserts there are three kinds of moderator vari- ables on which the effects of leader behavior are contingent: task variables, environmental variables and individual difference variables. The details of House's theory lay the conceptual framework for an explanation of the effects of subordinate's expectations on their perceptions of leadership behavior. House's Leadership Theory House's Path-Goal Theory of leadership combines the research efforts using the expectancy theory of motivation and the Path-Goal hypothesis of Georgopolous, Mohoney and Jones [1957]. House states: The central concept of Expectancy Theory is that the force on an individual to engage in a specific behavior is a function of (1) his expectations that the behavior will result in specific outcomes, and (2) the sum of the valences, that is, personal utilities or satisfactions, both intrinsic and extrinsic, that he expects to derive from the outcomes. Research findings indi- cate that the function is a non-linear monotonically increasing product of expectations and valences [House, 1971b, p. 4]. House uses the Galbraith and Cummings [1967] expectancy NOdEI- This model is an extension of Vroom's [1964] model with the addition that some of the valences associated with a behavior are intrinsic to the behavior itself and some are extrinsic consequences of that l6 behavior. Notice that when the behavior itself is largely intrinsically valent, then the instrumentality (which here is Evan's [1968] Path instrumentality defined as "the cognition of the degree to which fol- lowing a particular path [behavior] will lead to a particular outcome"-- similar to Vroom's "expectancy") must be one or close to that. After all, behavior that is intrinsically valent can be assumed to be also intrinsically motivational because valences associated with it are completely certain. This is the motivation theory House uses: n M = Ivb + E] [Iva + E=1 (E2 Evi)] i = l,..., n M = Motivation to work. E1 = The individual estimates the probability that his behavior will accomplish some work goal . . . So he must consider his abilities to do what is necessary and also the barriers that stand in his way of accomplishing the task, and also the support he can count on from others to complete the task. E2 = The individual also estimates the probability that accomplishing the task will get him the rewards that he wants (i.e., that are valent to him). IVb = The individual gives some subjective value to the intrinsic valence that the behavior itself (the behavior necessary for the goal accomplishment) would bring to him. IV = The individual gives some subjective value to the intrinsic valence that accomplishing the work goal will bring him. l7 EVi = The individual gives some value to the extrinsic valences associated with the outcomes he will personally receive for accomplishing the task. Using this as a stepping stone, House points out that a leader's behavior can affect each of those variables that will lead to motivation. EVi The leader (at least in part) helps determine the extrinsic rewards associated with accomplishment of the work goal. He may assist in raises in wages, in promotions, certainly in [arecognition of the accomplishment, and possible assignments to the more interesting or better parts of jobs. The leader can increase the expectancy of the subordinate of the reward(s) occurring upon completion of the work goal. He can clarify the connection between rewards and work-goal accomplish- ment by being consistent in recognizing and rewarding the accom- plishments. When the subordinates see this behavior then their estimate of E2 will increase. E1 = The leader can affect the three areas that an individual con- siders in his estimate of E1. That is, he may provide support for the subordinates efforts; he may help train the individual and thus increase the subordinates abilities to perform the task; and he may reduce the barriers that stand in the subordinate's way of accomplishing the task. These last two leader behaviors (increasing abilities and reducing barriers) are sometimes not possible in certain operations. However, providing support for the subordinates should always be possible. l8 IVb = The leader behavior associated with increasing this variable is very similar to those associated with increasing E]. In order to make the work behavior itself more intrinsically valent, the leader may reduce frustrating barriers, be supportive especially in times of stress, and be considerate of the different needs of the subordinates. This may mean increasing the social inter- action on the job, encouraging more participation of the sub- ordinates in planning the task, praising the worker personally for his work related behavior, or any other method that the leader has at his disposal to make the work itself more satisfy- ing for the subordinates. IVa = The leader may increase the intrinsic valence associated with the subordinate's accomplishing the task by making the rewards more in line with the subordinates higher level needs and by assigning subordinates to positions that upon accomplishment produce the rewards the subordinate wants. For example, by giving more influence and goal control to subordinates with high needs for independence [Vroom, 1959; Campion, 1968] or high needs for achievement [Litwin & Stringer, 1968], the leader will increase the intrinsic valence associated with work goal accomplishment for these people. House further points out that when a leader has influence with his superiors and peers, he may be able to influence the goals and design of the whole (or at least the parts significant to his subordi- nates) organization task system. Any efforts on his part that help bring the goals and procedures more in line with his subordinates 19 needs and desires will bring about increases in one or more of the variables that influence motivation. House goes on to clarify and modify these statements concerning the behaviors of the leader and their effects on motivation by stating eight propositions. Basically, these deal with situational factors and make predictions concerning the effect of particular leader behaviors possible. These propositions may be stated as: 1. "The motivational functions of the leader consists of increasing personal payoffs to subordinates for work-goal attainment, and making the path to these payoffs easier to travel by clarifying expectancies, reducing road blocks and pitfalls, and increasing the opportunities for personal satisfaction en route [House, 1972, p. 6]." 2. The leader's clarifying of the path-goal relationship and thus increasing expectancies will have positive motivational effects to the extent that: a. it reduces role ambiguity. Role ambiguity is negatively valent to subordinates [Rizzo e;_al;, 1970] and it "is usually associated with low subjective estimates of the probabilities that effort will be rewarded.“ b. it makes possible the exercise of externally imposed con- trols. Externally imposed controls are motivational because they make possible the allocation of valences contingent on desirable behavior. This works only when the leader con- trols the positively valent rewards the subordinate desires, when the leader controls the punishment associated 20 with the negatively valent aspects, when the rewards and punishments are contingent on performance, and when the contingency is clearly perceived by the subordinates. (Note that here House referred to motivation for per- formance.) “Whether performance motivated by external controls is satisfying to the subordinate depends on his needs, values, and perceptions of equity in the exchange of effort for rewards [House, 1972, p. 7]." When the system controls the objectives and routines of the task, a leader's attempts to clarify path-goal relationships may be seen as redundant with existing conditions. If the use of these externally imposed controls by the leader are seen as redundant, such controls may increase performance but will also decrease satisfaction. Leader behavior directed at need Satisfaction of subordinates will result in increased performance to the extent that such satisfaction increases the net positive valence associated with goal-directed behavior. Satisfaction desired by subordinates from the behavior of his leader equals the frequency of that behavior, multiplied by the subordinates preference for it. Thus, the leader behavior must be congruent with the subordinates behavioral preferences to be satisfying. The extent that it is or is not congruent is the extent that it is or is not satisfying. Subordinate preferences are determined by the personality and cultural training of subordinates. 21 7. The extent that particular leader behaviors are seen as legitimately required to facilitate the subordinate's task performance determines the extent that such behavior will increase the expectancies of subordinates that their effort will lead to work-goal attainment (E1). Thus, E] is a func- tion of the frequency of the leader behavior multiplied by the degree to which the subordinates perceive that behavior as facilitating their task performance. This perception is determined by the objective performance requirements of the task and the subordinates' familiarity with those task require- ments. 8. The more the subordinate is dependent on the leader, the greater will be the effect of the leader's behavior on the subordinate's expectancy that work goal accomplishment will result in desired outcomes. These propositions assert then, that there are three kinds of moderator variables on which the effects of leader behavior is con- tingent: task variables, environmental variables, and individual dif- ference variables. It is clear that behavior that clarifies path-goal relation— ships, reduces role ambiguity, or imposes external controls would be considered initiating structure behaviors. Leader behavior providing support and providing need satisfaction would be examples of considera- tion behaviors. In a presentation of his theory [House, 1971], House reviewed the literature involving initiating structure as measured by 22 perceptions of the subordinates (LBDQ). He came to basically the same conclusions the previous reviewers came to. He found the relationships between initiating structure, performance measures, and satisfaction measures were inconsistent. He found that these relationships were affected by job level, group size, and level of consideration. He then proceeds to give a post hoc interpretation of these results using his Path-Goal Theory constructs. Basically, he explains positive correlations of l;§;_with measures of performance and satisfaction by citing evidence concerning situational variables (e.g., evidence of ambiguity in the job situation where the measures were taken--thus increased consideration and satisfaction ratings). He also uses situational variables to explain negative and zero correlations too (e.g., in routine jobs l;§;_is seen as redundant external control and thus lowers ratings of consideration, and satisfaction or in autonomous jobs (sales) initiating structure would not clarify expec- tations that performance will result in rewards, since the performance- reward relationship is determined by the compensation system. Thus increased, l;§;_would be of no help and may actually hinder). House admits he made some assumptions concerning the moderator variables he used in this post hoc analysis. So, he included an empirical test of his theory also. A report of his study follows. Dessler and House administered the LBDQ [form XII Stogdill, 1965], a modified form of Evans' [1958] expectancy scales (Expectancy I and II), a modified form of Porter's scales designed to measure opportunities for valent intrinsic or extrinsic outcomes, a role ambiguity scale [Rizzo, House and Lirtzman, 1970] and a scale developed 23 to measure certainty or predictability of task demands. All scales had (reliabilities in excess of .70. The following hypotheses were tested: 1. The correlations between leader initiating structure and subordinate role ambiguity, satisfaction and path-goal expectancies will be moderated by the degree of respondent task certainty. The higher the certainty, the smaller will be: (a) the negative correlation with role ambi- guity; and (b) the positive correlation with satisfaction and expectancies. 2. The correlations between leader consideration and sub- ordinate satisfaction and expectancies will be positively moderated by task certainty. The consideration-role ambiguity will be negatively moderated. 3. Leader consideration will have a positive covarying influ- ence on the relationship between leader initiating structure and subordinate satisfaction and expectancies, and a negative covarying influence with respect to role ambiguity. Specifically, when leader consideration is held constant, the moderating effect of task certainty will be more pronounced [House, 1972, p. 21]. After adjusting the moderator groups for homogeneity of occupa- tional level, Dressler and House report these results. Support for the first hypothesis was found with respect to Expectancy II and Role Ambiguity but not for Expectancy I, intrinsic job valence or extrinsic valence opportunities. The second hypothesis was strongly supported with respect to Expectancy I and opportunities for both intrinsic and extrinsic valence. The third hypothesis was also strongly supported. Nine of the 11 r's between initiating structure and the dependent variables decrease in the predicted direction as task certainty increases. This study provides rather strong support for House's theory. In the discussion section of his paper [House, 1972], House proposes a modification of the theory to include the effect due to covariance 24 of consideration on any relationships between structure and dependent variables. House proposes that such covariance would have its greatest effect under conditions where the task is less satisfying. The present findings indicate that covariance of consideration suppresses the r between structure and the dependent variables under conditions of high task certainty. It is likely that tasks with more certain demands are also more dissatisf ing for at least a majority of the respondents [House, 1972, p. 24 . The findings of the study referred to in his paper clearly show that under conditions of low task certainty, with consideration held con- stant, the relationships between leader structure and the dependent variables are in line with the predictions of House's theory. House mentions that the findings of this study indicate that as task cer- tainty increases, leader consideration becomes an increasingly important function. Under high certainty, consideration facilitates performance, thus raising the expectancy that effort will lead to goal attainment and goal attainment will lead to the rewards that are desired. Caution should be taken in considering the results reported by House. No cross-validation on the correlations was attempted and all of his results are correlational in nature--though, despite this, House infers some causality. House suggests further research be per- formed on the relationships he describes. He states the need for more moderator variable approaches to understand the complex relationships between Initiating Structure, Consideration and effectiveness criteria. 25 Conclusions from Literature Review After reviewing all of the previous literature, the present author has come to several conclusions: --that the literature about QQN;_and l;§:_presents a complex picture of results. --that there is sufficient evidence to state that a relation- ship exists between perceptions of Consideration and Initiating Struc- ture [Anderson, (1966), Lowin §§Lg1;_(1969), House (1972), and Weissenberg and Kavanaugh (1972)]. --that there exist moderator variables that influence the relationships of l;§;_to EON; and of each to measures of leader effec- tiveness (performance and satisfaction measures). --that research in this area that includes moderator variables is insufficient. --that the most promising moderator variables to investigate include task characteristic variables, situational variables, and individual difference variables of both the leaders and the sub- ordinates. --that the speculations of Lowin ££_Ql; [1969] and especially the Path-Goal Theory of Leadership of House [1972] provide a reasonable framework to explain the apparent inconsistent findings of studies involving Consideration and Initiating Structure. The purpose of this thesis is to expand the literature in three areas: 1) Further investigate the claim of independence in the CON;;1;§;_ relationship; 2) Investigate the possibility that some individual 26 difference variables (in particular personal background characteristics) moderate the relationship between perceived consideration and perceived initiating structure; 3) Investigate the possibility of initiating structure as a moderator of the consideration-satisfaction relationship and consideration as a moderator of the initiating structure-satisfaction relationship. CHAPTER II METHODS This study was part of a larger program investigating the effectiveness of an experimental foreman training program and uses data collected from the hourly employees with instruments embedded in a larger questionnaire. Subjects The subjects of this study were 232 hourly production workers in the Detroit area Chrysler Corporation manufacturing plants. An attempt was made to collect data on 592 subjects by means of a ques- tionnaire using three procedures: 1) personal administration of the questionnaire; 2) mail and returned through the mail; and 3) distribu~ tion at the plant and returned through the mail. The usable data - includes the data of subjects collected in the plant (24) the data of subjects returned via the mail (258). The return rate (calculated as the proportion of data collected from the available sample) was 48 per- cent. Of these 282 subjects some (50) did not complete those sections of the questionnaire used for this study. The data for those subjects were also dropped thus leaving 232 usable subjects. 27 28 Instruments Three instruments were employed in this study. 1) Measures of Consideration and Initiating Structure on which the subjects rated their foreman's behavior were achieved through the use of the Leader Behavior Description Questionnaire [Form XII, Stogdill, 1965]. The items of these scales are listed in Appendices A and B. Marder [1960] reported the first use of this form of the scales in a study of an army division and a state highway patrol organization. Day [1961, 1963], Stogdill [1962] and Goode [1963] used revised forms of this questionnaire in studies on an industrial organization, a group of ministers, leaders in a community development, United States Senators, and presidents of corporations. The subscales, means, standard deviations, and reliabilities from previous studies (deter- mined by a modified Kuder-Richardson formula--modified in that each item was correlated with the remainder of the items in its subscale rather than with the subscale score including the item, thus "Conserva- tive" estimate of subscale reliability) are all shown in Table 2. 2) The second instrument used in this study was the Biographi- cal Data Blank developed for the Chrysler Foreman Training Research Project. A copy of this instrument appears in Appendix C. 3) The third instrument used consisted of two scales: Job Involvement and Job Identification (see Appendix 0). These scales, developed by Ruh, Johnson, and Scontrino [1973], are used as a means of measuring a worker's satisfaction with his job. Responses to items of these scales were presented in Lickert-type format. 29 Table 2. Means, standard deviations, and reliabilities of I.S. and CON. scores in studies using form XII of the LBDQ. (I) 801 U) U'l G) W >5 'P-H +4 +9 : +J> s. +4 4»: c c m Subscales .53 3'— “.3'4'3 33 'E ‘s’.’ 2% {'3’ 83-93 8 U) 30 L: m :0) O-g— LOP m.,— p >~,ov- .25.. 00 'l— EU Qt!) Om r—U) f6 5: .91; :22 .5 cs :32 e2 '32: s (D ID. (LL! 2 0...! 00. .40. DO. U) Initiating Mean 3 86 3.97 3 66 3.87 3 72 3.85 3.83 3.77 3.88 Structure SD .57 .45 54 49 57 .50 56 42 .55 Rel. .79 .75 .78 .70 72 77 .78 80 72 Considera- Mean 3.71 3.69 3.71 4.25 4.11 4.15 4.23 4.13 4.11 tion SD .56 .65 .58 .58 .47 .40 .55 .41 .59 Rel. .76 .87 .84 .85 .77 .78 .83 .76 .85 Analyses In order to "further investigate the claim of independence in the CON;:1;§L_relationship," an 11 x 11 correlation matrix was gen- erated. The eleven variables included each subjects': mean Considera- tion scale score, mean Initiating Structure scale score, mean Job Involvement scale score, mean Job Identification scale score, age, company seniority, job seniority, own education level, education level of father, size of community where raised, and father's occupation. Reliabilities for the Consideration, Initiating Structure, Job Involve- ment, and Job Identification scales were determined using Cronbach's Alpha Reliability Coefficient [Cronbach, 1951]. This measures 3O internal-consistency test reliability and was selected because unlike most other reliability formulas, "Cronbach's is not restricted to dichotomous scoring, thus it is especially useful for Likert-type scales [Gilmore, 1968, p. 1]." In order to "investigate the possibility that some individual difference variables may moderate the relationship between perceived consideration and perceived initiating structure," the seven personal background characteristic variables and two satisfaction measures were each dichotomized (see Table 3) and the correlation between CON;_ and l;§;_for each of these subgroups was calculated. A test for the difference between independent correlations was performed using Fisher 2 scores [see Guilford, 1965, p. 163-164]. This analysis was performed on the data of all odd-numbered subjects, n = 117, (to be referred to as the Validation sample). Any significant findings were to be cross-validated on the data of all even-numbered subjects, n = 115. The division of the total sample into validation and cross- validation samples was accomplished to minimize the possibility of results due only to chance in this exploratory-type study. Finally, in order to "investigate the possibility of initiating structure as a moderator of the consideration-satisfaction relationship and consideration as a moderator of the initiating structure- satisfaction relationship," the variable initiating structure was dichotomized around the sample mean (see Table 3) of 2.0. Correlations for the CON;:Job Involvement and the CON;7Job Identification relation- ships were calculated for both the High and Low l;§;_groupings. Then the variable consideration was dichotomized around its sample mean 31 amp om.N v meoum zoo 3o; mm om.~ m meoum .zou cmwz. cowumemcwmcoo . . . . . . meauuaepm pep oo N v weoum m H 304 pm oo N A meoum m H gap: mcwpc_w_=m mm. oo.e v aeoum emaz eavewpemeH mam; em oo.e A macaw cum: umwewpcaeH aem> =o_pmawcwpcmww om oo.m w meoom cam: um>_o>cH poz amp oo.m A «Loom :mmz cw>Po>cH pcmem>Fommw xcmpm memo co x:w_m mums co co_pma=uuo omp N Lo F mmcoammm nm—wamcs NFP m1m mmcoammm umppwxm m.em;ame xcwpm mums co x=m_m mama ummwmm mews: om m Lo .N .F Foeam N¢F co m Lo w cane: zuwcseeou mmcoamwm mmcoammm Lo mNWm xcmpm memo xcmFm memo Locum; m_F co mmm_ Lo mmmb Lo .m.u wp_ co emzmw; Lo .m.w cusp meoz o . =m= mmcoammm =m= mmcoammm e um xcmpm mpmo m mxcmpm ms» co =xeoz m mm H o :p .m.: can» mmmA mm— umem Lo mmm__oo meoE co .m.: mm m co mmmp go .mmm__oo @50m $0 um =FF= mmcoammm .N_ mmcoammm . now pcmmmea xnweo_cmm NFF mgpcoz om eowcsw om? co mgucoz mm A eowcmm non Auweowcmm mop mgpcoz omp eovcaq cu mcucoz omp A eowcmm acmgsou FFF upo .me> om mcao> FNP u_o .me> cm A upo mm< Ampasmm #:Pom Leopzu FQOF AmFQEmm p=_o¢ mmouzu _mno_ m_nmPLm> Papow Lev N gzoemaam quop Lev P aaoemnzm F 2 z .ummz mg on mmpnmwem> Loymemuoe FMFOcmpoa we cowpm~_eouo56wo .m m—amh 32 of 2.5 (see Table 3). Correlations for the I.S.-Job Involvement and the I.S.-Job Identification relationships were calculated for both the High and Low CON. groupings. The test of difference between independent correlations of each pair was again applied. CHAPTER III RESULTS First, the data on the four scale scores used in these studies were investigated. Table 4 summarizes the results found in this sample. Table 2 (in previous chapter) showed the same type of data found in nine previous studies. A comparison of the two tables shows that the reliability coefficients are very similar but the CON;_and l;§;_scores of this sample are considerably lower than any of the nine studies referred to in Table 2. The data on the Job Involvement and Job Identification scales for this sample show scale reliabilities of a respectable nature (Alpha = °83 and .78 respectively). Table 4. Means, standard deviations, and reliabilities of scales used in this study. Subscale Mean SD Rel. (Cronbach's Alpha) Consideration 2.49 0.83 .86 Initiating Structure 2.02 0.76 .76 Job Involvement 3.18 0.81 .83 Job Identification 3.92 0.59 .78 33 34 Table 5 shows the intercorrelation matrix for all eleven variables used in this study. The correlations are all Pearson- Product-Moment correlation coefficients not corrected for attenuation. One should take special note of the correlation between CON; and.l;§; (.593). When corrected for attenuation this increases to r = .76. Table 6 displays the results of the investigation of potential moderator variables of the CQNLzlLSL relationship. AZ 2 1.96 is neces- sary for a significant (.05) difference between correlations. Since no groups were found significantly different, the cross-validation analysis was abandoned. In order to satisfy this researcher's curiosity, and in order to explore the possibility that a larger sample might establish a result approaching significance in the validation sample; the same analysis was performed on the total sample (Valida- tion and Cross Validation samples combined). Table 7 presents those results. Note that even in this larger sample, no significant dif- ferences were found. Finally, Table 8 displays the results of the use of 1;§;_as a moderator of the Consideration—Satisfaction relationship and the use of EON; as a moderator of the l;§;7Satisfaction relationship. Again, no significant differences were found. 35 Fmp. Fo. pm mm”. u mo. pm muumnnzm oom Lo; L we mucmu_ewcmwm ooo.— moo. mam. moo. FF”. moo. ooo.~ omp. noF. Fwo. omo. ooo._ amp. mw_.1 New. ooo.~ N¢—.u mm¢.1 ooo._ mmv. ooo._ Npo. epo. mop.u pmo. —oo. F— wmo.1 wvo.1 Fmp.u Nvo.1 mmo.a op mmm.n mnP.1 mm_.1 moo. ,emo. m m¢¢.1 o¢N.1 o¢N.1 Now. va. m NFm. opm. cop. wwo.1 —mo.1 N wwn. mew. woo. ovp.1 woF.1 m ooo._ mmm. me. omN.1 mm~.1 m ooo._ mvm. mmp.1 oom.1 v ooo.~ o¢_.1 mmm.1 m ooo.F mom. N ooo.~ P .ouo cmmwmm Logan; Lo Lme mo .cwm .cmm eageee .eou .um .um new .00 Ap_v Aoev Amv Amy ARV Amv mm< meoum .m>< meoom .m>< meoum meoum .pcaeH new .>=H non .a>< .a>< .m.H .zoo Amv ASS Amv ANV A_v .ANMN u zv mFQEmm Papou com xwepme :owpmpmeeou .m mFQMH 36 Table 6. Results of investigation of potential moderator variables of CON.-I.S. relationship using validation sample (odd numbered subjects only, n = 117). Potential Moderating Dichotomized r Variable Label CON.,I.S. n 2 Old .646 56 Age .121 Young .633 61 1 (Co.) Senior .739 31 Co. Seniority .326 (Co.) Junior .578 86 (Job) Senior .674 62 Job Seniority .753 (Job) Junior .588 55 H.S. or More .675 73 Ed. of Self .127 Less than H.S. .536 44 More than G.S. .645 62 Ed. of Father .090 G.S. or Less .635 55 Size of Com- Urban .604 70 munity where .928 raised Rural .706 47 Father's Skilled .672 56 753 occupat'0" Unskilled .586 61 Involved .682 71 Job Involvement .118 Not Involved .548 46 Very Identified .681 39 Job Identification .468 Less Identified .627 78 37 Table 7. Results of investigation of potential moderator variables of CON.-I.S. relationship using total sample (N = 232) Potential Moderating Dichotomized r Variables Label CON.,I.S. N Z Old .578 121 Age .353 Young .608 111 (Co.) Senior .637 70 Co. Seniority .794 (Co.) Junior .563 162 (Job) Senior .623 120 Job Seniority .737 (Job) Junior .559 112 H.S. or More .574 139 Ed. of Self .228 Less than H.S. .553 93 More than G.S. .641 116 Ed. of Father .218 G.S. or Less .536 116 Size of com- Urban .593 142 munity where .036 raised Rural .596 90 Father's Skilled .659 112 414 Occ”pat‘°" Unskilled .714 120 Involved .642 136 Job Involvement .533 Not Involved .504 96 Very Identified .555 94 Job Identification .750 Less Identified .622 138 38 Table 8. Results of investigation of use of I.S. and CON. as moderator variables in I.S.-satisfaction and CON.-satisfaction rela- tionships. Potential Dichotomized r Moderator Label investigated N 2 High 1.5. rCON,JINV = -.147 91 Initiating Structure r 0.103 Low 1.5. CON.JINV = -.l6O 141 High 1.5. rcon, JIDENT = -.125 91 Initiating Structure r 0.184 Low I.S. CON, JIDENT = -.100 141 High con. '15, JINV = .013 93 Consideration r 1.052 Low CON. IS, JINV = -.129 139 High con. '15, JIDENT = -.064 93 Consideration r 0.096 Low con. IS, JIDENT = -.051 139 CHAPTER IV DISCUSSION Worthy of Note Throughout the discussion of the results of this thesis, the reader should keep in mind the work environment that each of the subjects in this study must cope with every working day. The factories they work in are all large, highly automated, multi-floored and noisy. In brief, the physical work conditions are not seen as pleasant. These factors, this author believes, significantly effect the results found. The CON.-I.S. Relationship Perhaps the most striking result of this study was the high positive correlation (.593) between the Consideration and the Initiating Structure scores. This result conflicts with Fleishman's claim that these scales are independent. This study is not the first to find results that conflict with Fleishman's claim as the review of the literature (see Chapter I, pp. 6-13) pointed out. Fleishman's defense [1969] generally states that the appearance of significant correlations between QQN;_and l;§;_represents a simple sampling distribution with a true value of r = O. The results of Weissenberg and Kavanagh's review [1972] of seventy-two studies using the CON. and I.S. dimensions seems 39 40 more convincing. They pointed out that over all those studies, there was a mean correlation of .23 and a median correlation of .36 between 'QQN;_and l;§, A further investigation of their data shoWs that studies using the LBDQ (subordinates' perception of leader behavior) rather than the LOQ (leaders rate themselves or ideal behavior) were more likely to show a positive correlation between CON; and l;§: Korman's review [1966] indicates the same trend leading this author to conclude that although the variables Consideration and Initiating Structure may be conceptually distinct, they are not perceptually independent to most subordinates. The results of this thesis lends further support to this conclusion. QQN;_and l;§;_were measured through the percep- tions of leader behaviors by their subordinates by using the LBDQ. Future research using the behavioral approach to investigate leader- ship concepts such as QQN;_and l;§;_should keep in mind the influence of perceptual bias on the results. Unfortunately, this thesis did not include other data measuring the behaviors of the supervisors. Thus, the extent to which perceptual bias influenced the findings concerning §9N4_and lLSL cannot be determined. The reader should therefore keep in mind that any other conclusions reached in this study pertain to perceived consideration and perceived initiating structure. House's Path-Goal Theory offers a possible explanation of the high §QN;:1;§;_correlation found in this thesis. House states that when the system controls the objectives and routines of the task, a leader's attempts to clarify path—goal relationships (to initiate structure) may be seen as redundant with existing conditions. Such behavior then may be perceived as dissatisfying behavior. Since all the subjects of 41 this study are factory workers, in large automated plants, one may assume that their tasks are routine and that the system controls their objectives. Such subjects may tolerate only a small amount of l;§;_ before they perceive the leader's behavior as dissatisfying. The.l;§; scores of this study were considerably lower than the averages reported by Stogdill in studies involving less routine jobs (see Table 4). EON; scores were also lower than found in previous studies, possibly indi- cating that even low levels of structuring by the supervisor was seen as relatively inconsiderate behavior. Besco and Lawshe [1959] offer an explanation very similar to House's. Their study also found a high positive correlation between Subordinate Perceived Consideration and Subordinate Perceived Initiating Structure (.40) in a highly automated production facility. They sug- gest that the factors of such a situation (automated machines, fixed work pace, fixed production schedules, static job assignments, etc.) tend to restrict the range of possible l;§;_behavior of the foremen. They state, . in this situation the only variation in Initiation of Structure scores was due to error variation and the halo effect. Behavior of the Consideration type was the only leadership behavior that really varied between the foremen [Besco and Lawshe, 1959, p. 581]. The present study did result in less variance in the l4§;_scores com- pared to the ggfl;_scores but the difference is not statistically significant. However, the point of the above explanations is to question the appropriateness of using an l;§;_scale to measure leader- ship behavior in a situation where the leader has little opportunity to display such behavior. A look at the content of the individual 42 items of these subscales raises further doubts about the appropriateness of these scales to measure leader behavior in a highly automated work situation. In particular, items 4 ("He gives advance notice of changes."), 13 ("He assigns group members to particular tasks."), and 15 ("He schedules the work to be done,") deal with behaviors that are typically out of the control of a first-level supervisor in a large auto plant such as was the case in this study. Table 9 shows the results of a factor analysis (using a computer program called FACTOR, developed by John Hunter of Michigan State University) with a varimax rotation. Note that all the QQN;_items (items l-10) loaded on the same factor with the exception of item 4 (which loads about equally on both scales). Also, note that two l;§; items (items 13 and 15) load on the factor the CQNL_items did while the rest of the I.S. items loaded on Factor 2. One might conclude from this that items 4, l3, and 15 are "bad" items. A cluster analysis subsequently performed (using Hunter's program "Package") lends further support to this con- clusion. Appendix E supplies the interested reader with the complete intercorrelation matrix of all 17 items and two scales. No wonder why much inconsistency has appeared in previous studies using ggfl;_and l;§, One might speculate on the confusion sub- jects must face when asked to respond to such inappropriate items. This could result in random responses to such items (causing error variance in scale scores), the introduction of a third factor guiding such responses, and/or an overall general (Halo effect) factor account- ing for responses. A principal components solution was elicited from the same factor analysis program used earlier. Table 10 displays the 43 Table 9. Factor analysis with varimax rotation (aSterisk indicates the factor on which each item had its highest loading). Item Number Factor 1 Factor 2 1 60* 31 2 58* 27 3 67* 18 4 43 43* 5 38* ll 6 74* 35 7 53* 27 8 58* 19 9 49* 4 10 73* 20 ll 10 58* 12 19 63* 13 35* 9 14 23 68* 15 40* 26 16 27 70* 17 21 63* Proportion of variance accounted for by factor. Factor 1: 23 percent Factor 2: 17 percent 44 Table 10. Factor analysis with principle components solution (the factor on which each variable has its highest loading is marked by an asterisk). Item Number Factor 1 Factor 2 1 66* -13 2 62* -15 3 64* -27 4 60* 7 5 36* -15 6 80* ~18 7 58* -ll 8 58* -21 9 41* -27 10 70* -29 ll 44* 40 12 54* 38 13 33* -14 14 60* 39 15 47* - 5 16 64* 39 17 56* 37 EIGENVALUES Factor 1: 5.59 Factor 2: 1.15 Proportion of variance Factor 1: 33 percent Factor 2: 7 percent 45 results. All items were submitted and programmed to form as many factors as possible with the criterion that the analysis will stop once the Eigenvalues fell below Kaiser's criterion of 1.00. Only two factors thus resulted with the highest loading for each of the seventeen items on the first factor. Also, one might note that of the 40 percent of the variance this two-factor solution accounts for, 33 percent of the variance is accounted for by this first factor. This points out two important aspects of the data used in this study: 1) There is evidence of a halo effect or some general factor effecting the data (all items, QQNL_and l;§;_loaded on the same factor) and 2) that a large proportion of the variance (60 percent) must be accounted for by small factors (with Eigenvalues less than 1.00 and accounting for a very small per- centage of variance each) and error variance. The form of the LBDQ used in this study was Form XII developed by Stogdill. An article by Stogdill, Goode, and Day [1962] describes the development of this form. Upon close inspection, one will see that in each of the populations used in the development, a factor analysis of the results was performed. The results show the emergence of large general factors in these populations as well. The authors state "Despite a strong general factor which accounts for 45 percent of the total factor variance, several of the subscales exhibit sufficiently high loadings on specific factors to suggest that they have some value as measures of discrete aspects of leader behavior [p. 264].“ Thus, the large general factor found in this study apparently should have been expected. 46 The high correlation between QQN;_and l;§;_found in this study could possibly be explained as the effect of the existence of a third factor interacting with these concepts. In a recently completed dissertation, Johnson [1973] posits that the degree of participative decision making allowed will effect the Egfl;:l;§; relationship. He concludes that "empirical research supports the notion that the initiat- ing structure, participative decision making and consideration are oblique [correlated] dimensions of leadership behavior [Johnson, 1973, p. 23]." The results of his own empirical study lends further support to his statement. Studies done by Stogdill, Goode, and Day [1962, 1963, 1964] and Beer [1966] also support this notion, at least partially, in that consideration was found to correlate with measures of participa- tive decision making. However, the results of the factor analysis previously mentioned discounts this as an explanation of the results of this study. Finally, despite Fleishman's claim of independence of Cgfl;_and .l;§;9 the evidence does seem to leave much doubt of that claim. It may be helpful to point out that Fleishman's claim is anchored into the development of the LBDQ, [Stogdill and Coons, 1957] of which he played a major part. The ggfl;_and l;§; dimensions came out as orthogonal factors in a factor analysis of potential LBDQ items. Given that a factor is a linearly weighted combination of all items being factor analyzed, and that orthogonality refers to zero correlation between these weighted composites, it is easily seen why the actual scale scores tend to be correlated. The orthogonal- ity is a mathematically imposed constraint which may not be psychologically meaningful [Johnson, 1973, p. 20]. In conclusion the high positive correlation found in this study in combination with the explanations of this result offered above,’ 47 leads this researcher to conclude that the Consideration and Initiating Structure dimensions of leadership behavior are not necessarily inde- pendent, especially when measurement of the dimensions are obtained through perceptions of subordinates and when situational variable (like automation) may make items on the scales (if not the whole l;§;_scale) meaningless. Investigation of Potential Moderator Variables The second purpose of this study was to investigate potential moderator variables of the QQN;:1;§L_relationship. All nine of the variables investigated failed to moderate the QQN;31;§L_relationship significantly. Several factors may have led to this result. In order for a variable to moderate a relationship between two other variables, one must have two other variables. The previous discussion in this chapter supplied much evidence of a general factor possibly operating on the data. QQN;_and l;§;_were not perceived as independent variables and this is true of every subgroup of subjects produced through the dichotomizing of the variables here used. Thus, one might conclude that none of the nine variables tested, differentially effected QQN;_and l;§, A look at the results displayed in Table 5 con4 firms this notion. With the exception of Company Seniority, any variable that correlated significantly with ggfl;_also correlated significantly with l;§, The test for the hypothesis that rxy = rxz for the same population [Walker and Lev, 1953, p. 256] was performed for each of the potential moderator variables of the QQNL:LL§L_rela— tionship. No significant differences were found. Furthermore, the 48 largest correlation between these variables and QQN;_or l;§;_was r = .223. Such a correlation could only account for 5 percent of the common variance. Thus, none of the variables used as moderators in this study provide an important avenue of understanding of leadership behavior as here measured. The extent of measurement error in the instruments used in this study may have influenced the results of this study. The items used as data for this thesis were embedded in a larger questionnaire (neces- sary for another study). The chances for error in measurement due to fatigue, rater bias, misunderstanding of questions, and even random responding are quite possible. The design of this thesis may also have contributed to the lack of dinding significant moderating effects. The moderator variables were dichotomized. This procedure, although often practiced in moder- ator variable research, could mask any differences if the variable actually effects the relationship in a curvilinear fashion rather than a linear one. An improvement in the design to correct this possible flaw might involve trichotomizing the moderator variables. Others might argue against the method by which the moderator variables were dichotomized. Still others would point out the lack of data collected on the foremen's actual behavior seriously limits any conclusions con- cerning moderation of the relationship of QQN;:1;§;_leader behavior. It is possible that the proposed moderator variables did not result in significant effects because the subjects merely accurately described their foremen. 49 A recent study by Runyon [1973] also found no significant relationship between an individual difference variable (Rotter's con- cept of locus of control-~Internal vs. External) and a subordinate's perception of management style. This same study, though. found that subordinates' satisfaction with supervision was very significantly affected by the interaction of a subordinate's personality trait and his foreman's management style. The results of Runyon's study might lead one to believe that even if the flaws in design and other criti- cisms listed above were corrected, personal background characteristics would still be found to have no moderating effects on the perceived CON;_and l;§;_relationship. However, the effects of these background characteristics might be seen in the subordinates' attitudes towards their foremen. This could explain the evidence of halo effect in the Qgfl;_and ILSL scores found in this thesis. CON., I.S., and Satisfaction Measures Finally, this thesis dealt with the use of CON;_as a moderator of the l;§;75atisfaction relationship and l;§;_as a moderator of the C9N;7Satisfaction relationship. Again, no significant moderator effect was found. Most of the above criticisms of the section above also hold for this section. In particular, the fact that CON;_and l;§;_scores were dichotomized rather than trichotomized when used as moderator variables in an appropriate criticism. The classic study of Fleishman and Harris [1962] established that curvilinear relationships exist between CON. and satisfaction measures (Grievance and Turnover rates) 50 and between l;§;_and the same satisfaction measures. Fleishman trichotomized the QQN;_and the j;§;_data to investigate the effects of interaction of QQN;_and l;§;_on satisfaction measures. The failure of this study to replicate the significant results that Fleishman and Harris found may be due to this lack of trichotomizing. Another possible criticism of this study is the use of Job Involvement and Job Identification as measures of satisfaction. More appropriate measures might have been measurements of the subordinates' satisfaction with supervision. Conceivably any subject could have been very satisfied with his supervisor but very dissatisfied with his job or vice-versa. Interestingly enough, however, negative correla- tions were found between QQN;_and both satisfaction measures as well as between l;§;_and the satisfaction measures. Previous literature would indicate a positive relationship between CON;_and satisfaction but would be less clear about the l;§;:satisfaction relationship. The results of this study show a high Job Identification score average (3.92), average Job Involvement scores (3.18), and relatively low CON; and l;§;_scores (2.49 and 2.02, respectively). All of these scores have a possible range of l to 5. If one assumes a halo effect is operating on the EON;_and l;§;_data, one might predict a negative rela- tionship between satisfaction with supervision and satisfaction with the job for this sample. Data is not available to test this hypothesis. One might also note that in all eight of the subgroupings (as shown in Table 8), the relationship between either QQN;_or_Lg§L and either of the satisfaction measures is very small. In fact the largest correlation of any of them is an r of -.l6. Thus, the importance of 51 any moderating effect found would have been very small in terms of the amount of variance that could have been accounted for. Summary This thesis lends further support to the notion the Considera- tion and Initiating Structure behaviors of first-line supervisors are not perceived independently by their subordinates. Situational vari- ables (e.g., work situation) do seem to effect the results of leadership studies. Future researchers should keep this in mind and question the appropriateness of the_;g§; scale or at least the appropriateness of certain items on that scale. One must question what gain in the advancement of knowledge on leadership can be accomplished by asking subordinates to rate the leader on behaviors over which the leader has no control. None of the variables tested significantly moderated the QQN4; _;§§L relationship. The large general factor (probably a halo effect) operating on the ggfl;:l;§;_data probably is the main reason for the lack of findings here. Runyon's [1973] findings offer an important explanation, i.e., that perception of £9N4_and l;§;_are not differ- entially effected by these variables but attitudes toward supervision may be. Future researchers might do well to keep this in mind and also consider other individual difference variables (e.g., personality variables) as moderators of such relationships. Finally, QQN;_failed to moderate the lLSL7Satisfaction rela- tionship, and l;§;_failed to moderate the QQN;7Satisfaction 52 relationship. Again, one must question just what are the CON. and I.S. scales measuring in this study. Future researchers might do well to look at the relationship of leadership variables with satisfaction with leadership rather than overall job satisfaction. LIST OF REFERENCES LIST OF REFERENCES Anderson, L. R. Leader behavior, member attitudes, and task performance of intercultural discussion groups. Journal of Social Psychology, 1966, 69, 305-319. Beer, M. Leadership, employee needs, and motivation. Columbus: Bureau of Business Research, Ohio State University, Monograph No. 129, 1966. Besco, R. 0., and Lawsche, C. H. Foremen leadership as perceived by supervisors and subordinates. Personnel Psychology, 1959, 12, 178-193. Blum, M. L. and Naylor, J. C. Industrial psychology: Its theoretical and social foundations. 'New York: Harper and wa,TPublishers, Inc., 1968, 414-450. Bruning, J. L. and Kintz, B. L. Computational handbook of statistics. Glenview, Illinois: Scott, Foresman and Company, 1968, 191-192. Campion, G. E., Jr. Effects of managerial style on subordinates' attitudes and performances in a simulated organizatiOnal settipg. Unpublished doctoral dissertation, university of Minnesota, 1968. Cronbach, L. J. Coefficient Alpha and the internal structure of tests. Psychometrika, 1951, 16, 297-334. Cummins, R. C. Relationship of initiating structure and job performance as moderated by consideration. Journal of Applied Psycholpgy, 1971, 55 (5), 489-490. Day, 0. R. Basic dimensions of leadership in a selected industrial organization. Doctoral dissertation. Columbus: The Ohio State university Library, 1961. Dawson, J. E. Toward a more general theory of leadership effectiveness. Doctoral dissertation. E. Lansing: Michigan State University Library, 1972. Evans, M. G. The effects of supervisory behavior upon worker perception of their path-goal relationships. Unpublished doctoral dis- sertation, Yale University, 1968. Fleishman, E. A. The description of supervisory behavior. Journal of Applied Psychology, 1953, 36, 1-6. 53 54 Fleishman, E. A. The leadership opinion questionnaire. In R. M. Stogdill and A. E. Coons (Eds.), Leader behavior: Its descri - tion and measurement. Columbus, Ohio: Bureau of Business Research, Ohio State University, 1957. Fleishman, E. A. Manual for leadership opinion questionnaire (1969 revision), Science Research Associates, Inc., 1969. Fleishman, E. A. Twenty years of consideration and structure. Paper presented at the Southern Illinois University Centennial Event Symposium, Contemporary development in the study of leadership, April 29-30, 1971. Fleishman, E. A., and Harris, E. F. Patterns of leadership behavior related to employee grievances and turnover. Personnel Psychology, 1962, 15, 43-56. Fleishman, E. A., Harris, E. F., and Burtt, H. E. Leadership and sppervision in industry: An evaluation of a superViSory trainin ro ram. Columbus, Ohio: Bureau of Educati6ha1 ResearcE, Ohio State University, 1955. Fleishman, E. A. and Peters, 0. A. Interpersonal values, leadership attitudes, and managerial "success." Personnel Psychology, 1962, 15, 127-143. Galbaith, J. and Cummings, L. An empiric investigation of the motiva- tional determinants of past performance: Interactive effects between instrumentality-valence, motivation, and ability. gggagggational Behavior and Human Performance, 1967, 2_(3), Georgopoulous, B. S., Mahoney, G. M. and Jones, W. Jr. A path-goal approach to productivity. Journal of Applied Psychology, 1957, 41, 345-353. Gillmore, B. M. Cronbach's alpha inner-consistency test reliability. Technical report 89, Computer Institute for Social Science Research, Michigan State University, East Lansing, Michigan, 1968. Guilford, J. P. Fundamental statistics in psychology and education. New York: McGraw Hill Book Company, 1965, 163-164. Halpin, A. W. The leadership behavior and combat performance of air- plane commanders. Journal of Abnormal and Social Psychology, 1954, 42, 19-22. Halpin, A. W. The leader behavior and effectiveness of aircraft com- manders. In R. M. Stogdill and A. E. Coons (Eds.), Leader behavior: Its description and measurement. Columbus, Ohio: Bureau of Business Research, Ohio State University, 1957. 55 Halpin, A. W., and Winer, B. J. A factorial study of leadership behavior descriptions. In R. M. Stogdill and A. E. Coons (Eds.), Leader behavior: Its description and measurement. Columbus, 0323: Bureau of Business Research, Ohio StateUniversity, Hemphill, J. K. Leadership behavior associated with the administrative reputation of college departments. Journal of Educational Psychology, XLVI (1955), 385-401. Hemphill, J. K., and Coons, A. E. Development of the leadership behavior description questionnaire. In R. M. Stogdill and A. E. Coons (Eds.), Leader behavior: Its description and measurement. Columbus, Ohio: BUreau 6f Business Researdh, Ohio State University, 1957. House, R. J. A path goal theory of leader effectiveness. Administra- tive Science Quarterly, 1971, 1§_(3), 321-338. House, R. J. A path-goal contingency theory of leadership. Paper presented at symposium, Contemporary Development in the Study of Leadership, Southern Illinois University, April 30, 1971. House, R. J. Some applications and tests of the path-goal theory of leadership. Paper presented at the First National Behavioral Organizational Conference, University of Toronto, April, 1972. Hunter, J. E., and Cohen, S. H. Package, Version 1.0. East Lansing: Michigan State University, Computer Institute for Social Science Research, June 1, 1971. Johnson, R. H. Initiating structure, consideration, and participative decision making: Dimensions of leader behavior. Doctoral dis- sertation. E. Lansing: Michigan State University Library, 1973. Katzell, R. A., Barrett, R. S., and Parker, T. C. Job satisfaction, job performance, and situational characteristics. Journal of Applied P§ychologys XLV (1961), 65-72. Korman, A. K. "Consideration," "initiating structure," and organiza- tional criteria--a review. Personnel Psychology, 1966, 12_(4), 349-361. Litwin, G. and Stringer, R. Organizational climate and motivation. Division of Research, Graduate School of Business Administra- tion, Harvard University, 1968. Lowin, A., Hrapchak, W. J., and Kavanagh, M. J. Consideration and initiating structure: An experimental investigation of leader- ship traits. Administrative Science Quarteriy, 1969, 14, 238-253. 56 Marder, E. Leader behavior as perceived by subordinates as a function of organizational level. Master's thesis. Columbus: The ' Ohio State University Library, 1960. Newport, G. A. A study of attitudes and leadership behavior. Personnel Administration, 1962, 2§_(2), 42-46. Parker, T. C. Relationships among measures of supervisory behavior, group behavior, and situational characteristics. Personnel Psychology, 1963, 16, 319-334. Rambo, W. W. The construction and analysis of a leadership behavior rating form. Journal of Applied Psychology, 1958, 42, 409-415. Rizzo, J. R., House, R. J., and Lirtzman, S. I. Role conflict and ambiguity in complex organizations. Administrative Science Quarterly, 1970, 15, 150-163. Ruh, R., Johnson, R., and Scontrino, M. P. The Scanlon Plan, participa- tive decision making, and job attitudes. Journal of Industrial and Organizational Psychology, 1973, 1_(l), 36-40. Runyon, K. E. Some interactions between personality variables and management styles. Journal of Applied Psychology, 1973, §1_(3), 288-294. Shartle, C. L. Introduction. In R. M. Stogdill and A. E. Coons (Eds.), Leader behavior: Its description and measurement. Columbus, Ohio: Bureau of Business Research, Ohio State University, 1957. Stogdill, R. M. Manual for the leader behavior description question- naire. Form XII. Columbus: Ohio State University, Bureau of Business Research, 1963. Stogdill, R. M. Managers, employees, and organizations. Columbus Ohio: Bureau of Business Research, Division of Research, College of Commerce and Administration, Ohio State University, 1965. Stogdill, R. M. and Coons, A. E. (Eds.) Leader behavior: Its descrip- tion and measurement. Columbus, Ohio: Bureau of Business Research, Ohio State University, 1957. Stogdill, R. M., Goode, O. S., and Day, 0. R. New leader behavior description subscales. Journal of Psychology, 1962, 51, 259-269. Stogdill, R. M., Goode, O. S., and Day, 0. R. The leader behavior of corporation presidents. Personnel Psychology, 1963, 16, 127-132. Stogdill, R. M., Goode, O. S., and Day, 0. R. The leader behavior of presidents of labor unions. Personnel Psychology, 1964, 11, 49-57. 57 Vroom, V. H. Some personality determinants of the effects of participation. Journal of Abnormal and Social Psychology, 1959, 55, 322-327. Vroom, V. H. Work and motivation. New York: Wiley, 1964. Wahba, M. A. and House, R. J. Expectancy theory in work and motiva- tion: Some logical and methodological issues, Unpublished paper, The City University of New York, 1972. Walker, H. M., and Lev, J. Statistical inference. New York: Holt, Rinehart and Winston,'1953, 256-258. Weissenberg, P., and Kavanagh, J. The independence of initiating structure and consideration: A review of the evidence. Personnel Psychology, 1972, 25_(l), 119-130. APPENDICES \lOlUl-hOON oxooo APPENDIX A CONSIDERATION SUBSCALE OF LBDQ--FORM XII He does little things to make it pleasant to be a member of the group. He puts suggestions made by the group into operation. He He He He He He He He treats all group members as his equals. gives advance notice of changes. keeps to himself. looks out for the personal welfare of group members. is willing to make changes. refuses to explain his actions. acts without consulting the group. is friendly and approachable. 58 11. 12. 13. 14. 15. 16. 17. APPENDIX B INITIATING STRUCTURE SUBSCALE OF LBDQ-~FORM XII* He lets group members know what is expected of them. He makes his attitudes clear to the group. He assigns group members to particular tasks. He makes sure that his part in the group is understood by the group members. He schedules the work to be done. He maintains definite standards of performance. He asks that group members follow standard rules and regulations. *Slightly modified version (See Dawson, 1972). 59 APPENDIX C CHRYSLER FOREMAN TRAINING RESEARCH PROJECT-- BIOGRAPHICAL DATA BLANK Age Sex Name of Supervisor to whom you report directly How long have you been employed at Chrysler? Years Months How long have you worked at your present job? Years Months INSTRUCTIONS: Simply circle the number in front of the answer that is closest to being true for you. How many years of school have you completed? (circle)highest grade com- pleted 6 or less 7 8 9 10 ll 12 some college college graduate work In what type of community were you raised (before age 18)? 1. In the country 2 Town of less than 2,000 3 Town of 2,000 or more but less than 10,000 4 City of 10,000 to 100,000 5 City of more than 100,000 What was (or is) your father's chief occupation? l Unskilled worker -- day laborer 2. Semi-skilled worker -- factory worker, for example 3. Skilled worker -- machinist, tool and die maker, for example 4 Office worker or clerk in store 5 Sales work 6 Foreman or supervisor 7 Subprofessional -- bookkeeper, pharmacist, surveyor, draftsman, etc. 8 Scientist or professional ll lawyer or engineer, for example 9 Small business man 0 l : Executive in a large company 60 Job APPENDIX D SATISFACTION SCALES Involvement My job means a lot more to me than just money. I'm really interested in my work. I would probably keep working even if I didn't need the money. The major satisfaction in my life comes from my job. The most important things that happen to me involve my work. I actually enjoy the daily activities that make up my job. I look forward to coming to work each day. Identification I feel bad when I make mistakes in my work. I really want to work hard at my job. I really want to do a good job. I'm really a perfectionist about my work. My personal satisfactions are related to how well I do my job. The welfare of my company is related to my own personal welfare. The company's achievement of its goals helps my achievement of my own personal goals. 61 m..Nm n Op..Oczesou wmmem>< ....-.. OOOO. OO sum. OLOOO O.OOO OLOOOOLOO OO.OO.O.O. u NOOO. .Ao.-. mama. Oo Ezmv meoom w.Oom cowpmemOOO=ou u .ome OO. O. NO O. .O ON ON NO .O .O .N OO OO OO ON OO OO NO OO O. OO. OO OO OO .O OO OO OO O. OO NO NO OO NO OO OO .O OO NO OO .OOO NO O. OO O .O OO OO N. ON OO OO O. NO ON ON OO O. OO NO .NO. OO OO .N OO OO .O O. ON OO OO N. OO OO .O .O .O OO O. OO ...O ON NO ON N. ON .. O. ON OO O. OO .N .O OO O. .O OO OO ON .OO. O. OO .O OO .. .N ON OO O. OO .N OO OO ON OO O .N NO O. .OO O O OO N N. ON .N N. NN O. ON ON O .OO. .O ON .. OO ON OO O. OO NO NO ON .O OO OO OO N. .O O .O .ON. ON O O. NN ON ON ON N. ON .N .O O. OO .O ON OO OO ON ON .OO. .O OO OO NO ON OO OO OO OO .N OO N. O. N. N. N .. O NO .ON. OO ON OO O. ON ON OO NO OO NO ON ON O. .N N. OO O. OO OO .OOO OO OO ON OO .O OO OO OO OO OO OO OO OO .N OO ON NO OO OO OO ...O OO .O OO OO OO ON .O O. N. O. O. N. O. ON ON O. ON ON OO .O.. O. .N .N ON OO OO NO OO OO OO NN OO ON OO ON OO NO .O O. .ONO NO OO .O OO OO ON OO .N .N O. NO N. OO ON .O OO OO .N .O .NO. OO OO NO .O ON .O .O OO ON NO ON OO OO OO NO OO .N OO OO .OO. .O OO OO OO .O OO OO ON ON .N OO NO OO OO OO ON .O OO .O .OO. Nom Pom N— o. m. OF m. N. O. op m w n m m O m N F onH