BEHAVIORAL STRATEGIES IN WAGE DISTRIBUTION Thesis for the Degree of Ph. D. MICHIGAN STATE UNIVERSITY IRVING MARK LANE ' ' 19 70 ‘ LIBRARY Michigan State ' University THESIS This is to certify that the thesis entitled BEHAVIORAL STRATEGIES IN WAGE DISTRIBUTION presented by IRVING MARK LANE has been accepted towards fulfillment of the requirements for M; degree in Mg Y Date June 25, 19‘ 0-169 .I ABSTRACT BEHAVIORAL STRATEGIES IN WAGE DISTRIBUTION By Irving Mark Lane The present research investigated the effects of differences in work inputs on the way subjects with total power distribute rewards to themselves and co—workers for a task they both have performed. The two major theoretical positions in the area of wage distribution, the equity hypothesis and the expectancy hypothesis, make different predictions concerning how an individual would dis- tribute such rewards. The equity hypothesis predicts that persons try to obtain fair wages based on their perception of both their own and their co-workers' inputs and outcomes. On the other hand, the expectancy hypothesis predicts that persons try to obtain thegreatest possible rewards for themselves. In order to determine the conditions under which the equity and profit maximization motives tend to dominate, four research situations were used. Irving Mark Lane Conditon 1. The person determining the wages (the chooser) had high inputs, and his co-worker (the receiver)--who received whatever wages the chooser allotted to him--had low inputs. Condition 2. The chooser had high inputs and the receiver also had high inputs. Condition 3. The chooser had low inputs and the receiver also had low inputs. Condition 4. The chooser had low inputs and the receiver had high inputs. First, the subjects in the four conditions completed an Industrial Opinion Questionnaire. Half the subjects were given a form of the questionnaire that required 1% hours to complete; these subjects had high inputs. The remainder of the si1bjects, those with low inputs, were given a shorter form of the questionnaire requiring % hour to complete. After the subjects completed the Industrial Opinion Ques- tionnaire, they were given the Wage Distribution Booklet. This booklet contained six sets of two alternative distributions. In a given set, the two alternatives (the standard distribution and the alternative outcome) each designated payments to both the chooser and the receiver. The standard distribution always had one person Irving Mark Lane receive $4 and the, other person $1. In three of the sets (Series X), the chooser received the $4 payment if he selected the standard distribution; in the remainder (Series Y), the standard distribution awarded $4 to the receiver. The alternative outcome always dis- tributed the payments more equally than the standard distribution. In addition, the alternative outcomes differed systematically in terms of relativewelfare. (Is the payment of the chooser greater than, equal to, or less than that to the receiver?) All subjects--both choosers and receivers—~completed the Wage Distribution Booklet. Each chooser knew that one of his choices would be selected at random to determine both his payment and that of the receiver with whom he was paired. The receivers knew that their choices did not influence rewards. The four situational factors systematically varied were: (a) inputs of the subject, (b) inputs of his co-worker, (c) real versus hypothetical choices, and (d) order of presentation (whether the chooser responded first to the three Series X alternatives or first to the three Series Y alternatives). The responses to the sets of alternatives in the Wage Dis- tribution Booklet were combined to produce scores for each subject on three personal dispositions: (a) self—interest-—the degree to which the subject maximized his own reward, (b) altruism--the Irving Mark Lane degree to which the subject maximized his co-worker' s reward, and (c) role symmetry--the degree to which the subject was concerned with a general principle of allocation (e. g. , equality) rather than a specific distribution of rewards. Subjects made no altruistic responses. Therefore, role symmetry and self—interest were perfectly negatively correlated. Taken together, the results for inputs of the subject and inputs of his co-worker provided insight into those conditions which generate concerns with either fairness (the equity hypothesis) or profit maximization (the expectancy hypothesis). There was strong evidence that a person, in distributing rewards to himself and another individual, only considered the other' 5 inputs. Subjects tended to give themselves as much of a reward as possible unless they perceived that their partners had worked hard enough to deserve more than minimal pay. It was as if the inputs of the co—worker had to exceed a threshold value before a subject felt that his partner deserved more than this minimum. If this threshold value was exceeded, then equity motives tended to dominate. On the other hand, if the inputs of the co-worker were less than this value, sub- jects tended to maximize their profits. Thus, these findings have significant implications for both the equity and expectancy hypotheses. The variable of order of presentation had a significant effect on the distribution of wages. Subjects who were presented Irving Mark Lane Series Y first tended to be less role symmetric. On the other hand, the variable of real versus hypothetical choices had very little effect. In addition, the results of the Guttman scale relating to relative welfare indicated that it was easiest for subjects to give up some of their potential wages to their co—workers when the resultant distribution still gave them more money. Thus, the results indicated that co-worker input is an important determinant of the way in which rewards are distributed. However, they also indicated that other factors besides inputs must be considered if a comprehensive theory concerning the distribution of rewards is to be developed. Approved: ($6,th ”(/7 .. / Date: 725% #2742 Dissertation Committee: L. A. Messé, Chairman C. Hanley E. H. Jacobson J. L. Phillips J. H. Wakeley BEHAVIORAL S TRA TE GIE S IN WAGE DISTRIBUTION By Irving Mark Lane A THE SIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Psychology 1970 To my wife, Myrna ACKNOWLEDGMEN TS I sincerely wish to thank my chairman, Dr. Lawrence A. Messé, for the encouragement, guidance, and understanding he has provided while this research was being conducted. In addition to his very significant contribution to both this research and my professional training, his altruistic behavior has been a constant inspiration. I would also like to thank the other members of my disser- tation committee, Dr. Charles Hanley, Dr. Eugene H. Jacobson, Dr. James L. Phillips, and Dr. John H. Wakeley. Not only have they provided help during this research, but each one has made a significant contribution to some phase of my graduate education. I would also like to thank Dr. Lawrence 1. 0' Kelly for pro- viding the necessary funds to conduct this research. In addition, I would also like to thank Ronald Cohen and Melvin Goldstein for running the subjects. Finally, I would like to express appreciation to both my wife, Myrna, and to my chairman' 5 wife, Susan Messé, for their continued understanding regarding the long periods of time they were without their husbands as a result of this research. TABLE OF CONTENTS LIST OF TABLES Chapter I. / INTRODUCTION Review of the Literature / The Present Research II . ME THOD / The Sample / Instruments . / Variables and Experimental Design I Procedure . . . . III. RESULTS ,. Effects of Situational Variables on the Manifestation of the Personal Dispositions . . / Further Analyses of the Effects of Inputs of Subject and Inputs of Partner . - / The Effect of Relative Welfare . IV. DISCUSSION . 1 Inputs of Subject and Inputs of Partner Relative Welfare . / Effect of the Control Variables iv Page Vi 77 84 93 94 94 100 . 101 Chapter Page V. CONCLUSION . . . . . . . . . . . . . . . . . . 106 ,4 Implications of the Results for Wage Distribution Theory . . . . . . . 106 , Implications of the Present Methodology . . . . 107 , Suggestions for Further Research . . . . . . . 109 / A Concluding Statement . . . . . . . . . . . . 111 /REFERENCES...................... 112 Appendix I. WAGE DISTRIBUTION BOOKLET . . . . . . . . . 115 II. ATTITUDE QUESTIONNAIRE . . . . . . . . . . . 122 III. INDUSTRIAL OPINION QUESTIONNAIRE . . . . . . 125 IV. STUDENT BACKGROUND QUESTIONNAIRE . . . . 139 V. GUTTMAN SCALED ROLE SYMMETRY RESPONSES................... 140 Table LIST OF TABLE S The choice situations . A sample work situation Summary of the analysis of variance for role symmetry scores Cell means for role symmetry scores Percent scale-point responses for work quantity tabulated within each experimental condition Percent scale-point responses for work quality tabulated within each experimental condition Percentage of role symmetry and self- interest responses Number of subjects manifesting a majority of each personal disposition classified by experimental condition t_va1ues for comparison of percentages of role symmetry responses occurring between experimental conditions . Page 57 58 79 80 84 85 88 89 92 I. IN TRODU C TION Adams (1965) notes that the just distribution of wealth, power, goods, and services in society traditionally has been the concern of philosophers, political scientists, politicians, jurists, and economists. Through 1966, psychologists and sociologists had displayed little interest in this area. 1 Since then, however, a growing number of empirically—oriented social scientists have turned their attention to situations where rewards of a number of kinds, but especially wages, are distributed. This growing interest no doubt stems from some degree of appreciation of the possible benefits, both to basic theory and to applied problems, which knowledge in this area could provide. For example, such knowledge should contribute to the basic understanding of the process of social exchange which underlies all human interaction. On the other hand, such knowledge when applied to problems in industry should also provide guidelines for wage distribution plans that employees would consider fair. Valid guidelines would be of great benefit in this 1 Homans and Adams are two notable exceptions to this generalization. era where employee satisfaction has become a management objec- tive. The present study is part of this growing interest, since its focus was the investigation of the senses of fairness in wage dis- tribution. To help understand the nature of this research, imagine the following situation: An individual is asked to distribute the wages to himself and another individual for a task they both have performed. The question arises of how the individual would distrib- ute the wages if given the opportunity to do so. The present research was designed to provide an answer to this question. Such an answer is important for assessing the influence of the sense of fairness on the distribution of wages, a problem of some theoretical interest at the present time. As will be seen, the two major hypotheses in the area of wage distribution, the equity hypothesis and the expectancy hypothesis, make different predictions concerning how an individual would distribute the wages if given an opportunity to do so. The present research, however, is not seen as a definitive test of the two hypotheses. Rather, it was performed to determine to what extent and under what conditions each of the hypotheses contributes to the understanding of how persons distrib- ute rewards. In order to better understand the contribution of the present research to the psychology of wage distribution, a review of the theoretical work in this area is presented below, followed by a summary of the research relevant to each theoretical position. Review of the Literature Theoretical Work in the Area of Wage Distribution Relative deprivation Most of the research concerning fairness in wage distribu- tion is an outgrowth of the studies on relative deprivation. This concept was initially proposed as an explanatory tool in the first volume of The American Soldier (Stouffer e_t_a_11. , 1949). Stouffer et a1. did not formally define the concept, nor did Merton and Kitt (1950), who analyzed the implications of the concept for reference group theory. However, its essential meaning can be inferred from the example given below. Despite the objective fact that soldiers with a high school education had better opportunities for advancement in the Army, they were not as satisfied with their prospects for advancements as were less educated men. This apparent paradox can be explained by assuming that the better educated men had higher levels of aspi- ration. These levels of aspiration presumably were based on what would have been realistic advancement expectations in civilian life. Therefore, they were relatively deprived of opportunities for advancement . Adams (1965), in reviewing the survey and experimental data concerning relative deprivation, noted that four main conclu- sions emerge. They are: 1. Dissatisfaction is a response to perceived injustice, and is not a direct response to perceived relative deprivation. Relative deprivation is a condition which elicits the feelings of injustice. In turn, these feelings of injustice trigger the expression of dissatisfaction. Therefore, injustice medi- ates the effects of relative deprivation. 2. That which is perceived to be just is based on relatively strong expectations. 3. A comparative process is inherent in the development of expectations and the perception of injustice. This is implied by the term relative deprivation. 4. Felt injustice is a response to a discrepancy between what is perceived to be and what should be. Distributive justice The existence of relative deprivation raises the question of the fair share of rewards, or in other words, distributive justice. The concept of distributive justice was formulated in economic terms by Homans (1961). The concept of distributive justice is most easily understood in terms of a dyadic relationship between Person and m. Person is any individual. M is an individual with whom Person is in an exchange relationship, or with whom Person compares himself when both he and 9113 are in an exchange relationship with a third party (e. g. , an employer). Oier is usu— ally a different individual but may be Person in another social role. According to Homans, distributive justice for Person and OM is realized when the profits they each receive are propor— tional to their investments. A state of distributive justice for Person and Other can be represented by the following equation: Person' 8 Profits Otherl s Profits Person' 8 Investments : Other' 5 Investments Homans defined profit as that reward which is received in an exchange relationship minus the cost incurred. A cost is defined as those alternative rewards which are forfeited by participating in an exchange relationship. Investments in an exchange are the rele- vant attributes that each individual brings to the relationship. Examples of investments in an industrial setting are the employee' s previous job experience, educational background, and the amount of physical effort he expends while performing his job. Distributive justice for Person and @323, when they are two individuals being rewarded by a third party (e. g. , an employer), is realized when the profit each receives is proportional to his investments. Homans stated that when distributive justice is not realized, a state of injustice (or inequity) exists; and if this injustice is to Person' 3 disadvantage, Person will display the emotional behavior called anger. Adams (1965) claimed that what Homans meant by anger is really the feeling or expression of dissatisfaction. Further— more, Homans stated that if distributive justice is not realized and is to Person' 8 advantage, Person will feel guilty. Homans asserted that the threshold for displaying dissatisfaction is lower than that for displaying guilt. Therefore, conditions of inequity must be greater when it is favorable to Person before he realizes it than when it is to his disadvantage. The concept of distributive justice is a significant advance— ment over the concept of relative deprivation, since it specifies to a greater degree the variables which mediate injustice. According to distributive justice, Person feels injustice not merely because his rewards are less than he expects (or feels is fair), but because when he compares his profit—investment ratio to Other' 3, he per— ceives that his ratio is smaller. The equity hypothesis2 Adams' (1963a, 1965) equity hypothesis, although consid- erably more. elaborate, is built upon distributive justice, and is consistent with it. Adams noted that whenever two individuals exchange anything, there is the possibility that one or both of them will find that the exchange was inequitable. One frequent case of perceived inequity occurs when a man exchanges his services for pay. On the man' 8 side of the exchange are his work experience, intelligence, training, sex, etc. These are what the man perceives as his contributions to the exchange; and for them, the man expects a just return. Adams used the term ”inputs" to refer to what Homans called "investments"; that is, those qualities which are perceived to be a man' 8 contribution to the exchange. Adams empha- sized that these inputs are as perceived by their contributor; they are not necessarily identical with those perceived by the other party in the exchange. Adams introduced two distinct characteristics of inputs: recognition and relevance. The existence of an attribute may be recognized only by its possessor, only by the other individual in the 2There have been attempts on the part of some to elevate both this and the expectancy hypothesis (see below) to the status of formal theories. However, strictly speaking, both are of such limited scope that it seems reasonable to refer to both of them as hypotheses. exchange, or by both. Whether or not such an attribute is considered an input is dependent upon the possessor' s perception of its relevance to the exchange. If he perceives it to be relevant, and therefore, if he expects a just return for it, then it is an input. Problems of inequity arise if the possessor of an attribute considers it relevant to the exchange, but the other party considers it irrelevant and acts accordingly. On the other side of an exchange is an individual' 8 receipts. Adams called these outcomes. In an employee-employer exchange they include pay, job status, fringe benefits, etc. In a manner analogous to inputs, outcomes are as perceived by their receiver. They are also characterized in terms of recognition and relevance. Inputs and outcomes are correlated, but imperfectly so. It is because they are not perfectly correlated that there is a con- cern with inequity. There exists normative expectations of what is a "fair correlation” between inputs and outcomes. These expecta- tions are learned during the process of socialization and are based upon observation of the correlations for a reference person or group. Although Adams acknowledged that it is important to specify those variables which determine the appropriate reference group or person, he conceded that such a specification was beyond the scope of his hypothesis. When the normative expectations of the individual making social comparisons are violated, i. e. , when he finds his inputs and outcomes are not in balance in relation to those of his comparison individual (Qt—ha), feelings of inequity result. Adams (1965, p. 280) states that "inequity exists for E SE whenever he perceives that the ratio of his outcomes to inputs and the ratio of Other' s outcomes to Other' 5 inputs are unequal. " Schematically, inequity exists whenever: Perceived Outcomes Person Perceived Outcomes Other < Perceived Inputs Person Perceived Inputs Other or whenever: Perceived Outcomes Person Perceived Outcomes Other > Perceived Inputs Person Perceived Inputs Other Likewise, a condition of equity exists whenever: Perceived Outcomes Person Perceived Outcomes Other Perceived Inputs Person : Perceived Inputs Other Up to this point the equity hypothesis is largely a clarifica— tion of distributive justice. However, Adams also broadens the theoretical basis of distributive justice by specifying the conse— quences of perceived inequity in terms of dissonance theory (Fes— tinger, 1 957). 10 Adams notes that the cognitive changes, "anger, " ”guilt, " or ”dissatisfaction, " generated by perceived inequities should be accompanied by some behavioral manifestation. Moreover, the intensity of these manifestations is determined by mechanisms sum— marized in two postulates (derived from dissonance theory). 1. The presence of inequity in Person creates tension in him. The tension is proportional to the magnitude of the inequity present. 2. The tension created in Person will motivate him to elimi- nate or reduce it. The strength of the motivation is pro— portional to the tension created. From these postulates, and from Festinger' s theory of cognitive dissonance, Adams has identified six methods by which Person can reduce or eliminate the dissatisfaction associated with inequity. 1. Person altering his inputs. Person may vary his inputs, either increasing them or decreasing them, depending on whether the inequity is advantageous or disadvantageous. When the inequity is to Person' 3 disadvantage, he may, for example, engage in restriction of output. When the inequity is to Person' 3 advantage, he may increase his productivity or the quality of his work. Person altering his outcomes. Person may attempt to vary his outcomes, either decreasing or increasing them depend- ing on whether the inequity is advantageous or disadvanta— geous. An example of an attempt to increase outcomes is the wage demands in union-management negotiations. Adams stated that decreasing outcomes to reduce inequity is an unlikely occurrence. Person distorting his inputs and outcomes cognitively. Person may cognitively distort his inputs and outcomes. The direction of the distortion would be the same as if he had actually altered his inputs and outcomes. Adams stated that since most individuals are influenced by reality, sub- stantial distortion is generally difficult. However, there is substantial literature relating to cognitive dissonance that has demonstrated that individuals do distort reality in order to mediate cognitive conflict. Person leaving the field. Leaving the field involves cutting off the social relationship. Quitting a job or absenteeism are two frequently used forms. Person acting on M. Person may attempt to alter or cognitively distort Other' 3 inputs and outcomes, or he may try to force Other to leave the field. 12 6. Person changing the object of his comparison. Person may change 9m with whom he compares himself when he experiences inequity and he and 913% are in an exchange relationship with a third party. This-method of reducing inequity is not applicable when Person and M are in a direct exchange. Changing the object of comparison when Person and M are in a direct exchange reduces to Person leaving the field. The expectancy hypothesis Lawler has proposed the expectancy hypothesis as an alter- native to the equity hypothesis. The expectancy hypothesis, de— scribed below, is considerably different from distributive justice and the equity hypothesis, and the predictions derived from it are in most cases directly opposite to those derived from the other two. Lawler (1968b) stated that the expectancy hypothesis and drive theory are the two most widely accepted theoretical positions in the area of human motivation. Drive theory stems from the work of Hull and Spence, while the expectancy hypothesis can be traced back to the works of Tolman and Lewin. More recent, refined forms of the hypothesis have been advanced by Porter and Lawler (1967), and Vroom (1964). Drive theory and the expectancy hypothe— sis are very similar. Atkinson (1964), after an extensive literature review, concluded that in almost every situation they lead to similar predictions. As Lawler (1968b) notes, the expectancy hypothesis empha- sizes that the motivation to perform an act is a function of both the strength of the expectancy that the act will be followed by a given outcome (reward or goal) and the value (valence) of that outcome to the individual. To him expectancies are subjective probabilities. They have values ranging from 0 (no likelihood that the act will be followed by outcome) to 1 (certainty that the act will be followed by outcome). Expectancies are subjective in nature and, therefore, may not be in accord with actual probabilities. Outcomes take values ranging from +1 (very desirable) to -1 (very undesirable); like an expectancy, valence is subjective in nature. Therefore, a particular outcome may very well have different valences for dif— ferent people, and it may vary in valence for the same person from time to time. Expectancy and valence combine multiplicatively in determining motivation. Thus, both expectancy and valence must be greater than zero in order for motivation to be present. It is also important, as Lawler notes, to consider all the possible out- comes that might be perceived to depend upon a given act. The products of all the Expectancy X Valence combinations need to be summed in order to determine the total amount of motivation present in a given situation. 14 Lawler (1968b) presented an example of the kind of predic— tion the expectancy hypothesis makes in a specific situation. Accord- ing to the expectancy hypothesis, a person should be motivated to work hard and produce to the extent that he feels working hard is likely to result in his receiving outcomes that he values. Therefore, in a properly designed piece rate situation, where large rewards are related to productivity, a person should be motivated to work hard provided that (a) he values the reward (money) and (b) he sees that there is a connection between his hard work and his receiving the reward. However, it is necessary to consider all the outcomes that might be related to working hard. The effects of a piece rate system might be negative if undesirable outcomes (e. g. , social ostracism) are seen as being related to high productivity. This point is critical for the expectancy hypothesis. In a piecework situation, an individual can be predicted to be a high producer if money has a positive valence for him or a low producer if money has a negative valence for him. Therefore, contrasting results can be explained post hoc in such a way that they all appear consistent with the expectancy hypothesis. From this it would seem that no evidence could be found which would invalidate the expectancy hypothesis, and therefore, its usefulness as a predictive tool is highly questionable. However, Lawler (1968) attempted to increase 15 the usefulness of the expectancy hypothesis, at least in the area of wage distribution, by providing more structure to the concept of valence, which he points out has been notoriously vague. He assumed that an increase in the amount of a reward (like money) should always increase its attractiveness, and therefore, an incre— ment of reward can never have negative valence. As an overall summary, the expectancy hypothesis empha— sizes a motivation to maximize positive outcomes. In situations where rewards are being distributed, this motivation leads persons to attempt to maximize their wages. In contrast, as noted above, the equity hypothesis emphasizes persons trying to obtain fair wages based on their perception of an input-outcome ratio compari- son. Research Relevant to the Hypotheses in the Area of Wage Distribution Research support for the equity hypothesis There is a considerable body of research related to both Homans' concept of distributive justice and Adams' equity hypothe— sis. These studies fall into two broad categories: naturalistic studies conducted in real life industrial settings; and experimental studies, for the most part conducted in the laboratory. 16 Naturalistic studies An early study relevant to distributive justice was conducted by Homans (1953). The study was concerned with two jobs within one division in a company. One was cash posting, a single repeti— tive job requiring little thought but some degree of physical mobility. The other job was that of a ledger clerk which entailed performing a number of non—repetitive tasks requiring some thought but little physical mobility. Ledger clerks were older than cash posters and had worked longer with the company. The company considered being a ledger clerk a higher level job, although being promoted from cash poster to ledger clerk brought no increase in pay. The study was conducted in three phases. Phase 1. The researcher received permission from both union and management to observe activity in the division. He then moved into the back of the room in which the division worked and spent a month allowing the workers to become used to his presence. Phase 2. A systematic observation of interaction in the room was undertaken for a two—week period. Phase 3. Each member of the division was interviewed for approximately two hours. The results indicated that the ledger clerks felt that they deserved to have higher pay than the cash posters, since their jobs 17 were more responsible. In other words, the ledger clerks felt inequity. Their inputs were higher than those of the then cash posters. They were older, more experienced and had a more re- sponsible job. However, some of their outcomes, such as salary, were not higher. The ledger clerks attempted to reduce their per- ceived inequity by complaining to their union representative. In other words, they tried to alter their outcomes by asking for an increase in salary. Clark (1958) studied checkout counters in a chain of super— markets. These counters were manned by two persons: ringers, who operate the cash register; and bundlers, who put the goods in bags. Within the social structure of the supermarket, ringing was considered a higher status, better paid job that was usually handled by full-time employees. Bundling was of lower status, and this task was usually performed by part—time employees, frequently students. During rush hours in the store, a bundler often would be directed in his work by a ringer, whose status in non-job related areas, such as education, was lower than his own. This created in many bun- dlers feelings of inequitable treatment. These bundlers felt they had higher inputs than the ringers but that their outcomes were lower. Through interviewing, Clark determined that the inequity that existed was actually perceived by both ringers and bundlers. 18 To reduce the inequity, the higher status individuals purposely slowed down their rate of bundling. By bundling slowly they reduced their inputs and by doing so succeeded in reducing their feelings of inequity. Sayles (1958) stated that fairness of wage distribution for industrial work groups was figured on the basis of the following equation. Our importance in the plant _ Our earning§_ Any other group' s importance - Their earnings When equality was attained, satisfaction was experienced; on the other hand, an inequality between the ratios produced dissatisfaction and pressures for redress. Sayles' findings provide support for the equity hypothesis. Patchen (1961) conducted a field study in which he inter- viewed approximately five hundred non-supervisory workers. He performed the research with the belief that comparisons of pay are not made in a cognitive vacuum. Rather, he felt that such compari- sons become meaningful to individuals only when they also compare themselves to others on dimensions that are believed to be bases of pay. Examples of these dimensions are skill, seniority, education, etc. In effect, Patchen believed that the individual makes a cognitive relation of the following type: My position on dimensions related to pay His (their) position on dimensions related to pay My pa His (their) pay compared to Within the framework of Festinger' s theory of cognitive dissonance, two major hypotheses were generated. Hypothesis 1. Wage comparisons which are objectively dissonant will be judged as unsatisfactory by the person who makes such comparisons. Hypothesis 2. Men who are satisfied with specific wage comparisons will explain their satisfaction in terms of a consonance between relative wage standing and relative standing on attributes relevant to pay; men who are dissatisfied with specific wage com— parisons will explain their dissatisfaction in terms of a dissonance between relative wage standing on attributes relevant to pay. These two hypotheses were confirmed. Another part of Patchen' 5 study was concerned with deter- mining the factors that influence an individual' 3 choice of a wage comparison individual. While this concern is relevant to the equity hypothesis, his results in this area were not statistically significant, but were promising enough to indicate future research would be desirable. Jacques (1961, 1962), in conducting interviews with indus— trial workers, noted a characteristic pattern of responses as a 20 function of their level of payment. He stated that a worker whose compensation is at the equitable level feels he is receiving fair pay- ment. By contrast, deviations in payment below the equitable level are usually accompanied by feelings of dissatisfaction which become stronger as the deviation increases. At ten percent below the equitable level of payment, there are active grievances and an actual desire to leave the job or take collective action. On the other hand, deviations above the equitable level are accompanied by feelings of being relatively well off as compared with others. At ten to fifteen percent above the equitable level of payment, there is a sense of preferential treatment and a feeling of unease about the relatively advantageous position. These results suggest that it is not neces- sarily true that each worker is out to get as much as he can for himself. Instead, workers appear to have a strong desire that each one should earn the right amount, a fair and reasonable amount relative to others. The studies cited above offer preliminary evidence to sug- gest that the desire to achieve an equitable state appears to also hold for wage distribution in industrial settings. Experimental studies The next three studies to be discussed were performed by Adams, the formulator of the equity hypothesis. Since the method 21 employed in each was quite similar (and has since been adapted by many other researchers in the field), only the first will be described in considerable detail. Adams and Rosenbaum (1962) studied the consequences of perceived inequity on quantity of productivity. The experiment was performed to test two hypotheses derived from the equity hypothesis. Hypothesis 1. If Person is paid by the hour, his produc- tivity will be greater when his outcomes are perceived as too great than when identical outcomes are perceived as equitable. The rationale for hypothesis 1 is as follows: if other means of reducing inequity are unavailable, increasing productivity will increase Person' 5 perceived inputs and thus bring them into line with both his perceived outcomes and Other' s perceived inputs and outcomes. Hypothesis 2. If Person is paid on a piecework basis, his productivity will be smaller when his perceived outcomes are too large than when the same conditions are perceived as fair. The rationale for hypothesis 2 is as follows: because inequity is associated with each unit produced, inequity will increase as work proceeds. Therefore, Person will strive not so much to reduce inequity as to avoid increasing it and he will, therefore, restrict production. 22 The two hypotheses were tested in the following manner. Students were hired through the Placement Service for part—time temporary work as interviewers. The subjects were given the impression that interviewing would take place for an extensive period of several months. When the subjects reported to their prospective employer they were randomly assigned in equal num- bers to the experimental condition (inequity condition) or to the control condition. They were asked to fill out a questionnaire requesting, among other things, their previous employment history. The experimenter then studied the subject' s background in his presence and, depending on the condition to which he had been assigned, induced the perception that the subject was qualified or that he was not. The subjects in the inequity condition were told, ”You don' t have any (nearly enough) experience in interviewing or survey work of the kind we're engaged in here. I specifically asked the Place— ment Service to refer only people with experience. . .I can' t under- stand how such a slip—up occurred. . .I guess I' ll hire you anyway . . .I' 11 just have to pay you at the advertised rate of $3. 50 per hour. On the other hand, the subjects in the control condition were told, "We can use you for this work. You meet all the qualifications for the job. . .We pay $3.50 per hour. This rate is standard for work of this kind performed by people of your qualifications. " 23 The experimenter then informed each subject about the interviewing task. Interviewing was chosen because it was brief enough so that many interviews could be obtained during the sched- uled work period of 2. 5 hours. When the subjects returned from the interviewing they were questioned by the experimenter. The questioning revealed that the subjects uniformly believed that they had been hired and had worked on a real job. The results of the experiment confirmed the predictions of hypotheses 1 and 2, thereby providing substantial support for the equity hypothesis. Adams (1963b) noted that the subjects in the piecework condition in the previous study (Adams and Rosenbaum, 1962) could have produced less when overpaid than when equitably paid for a dif- ferent reason than originally believed. It was believed that the lower productivity was an attempt by the overpaid pieceworkers to avoid increasing inequity by restricting output. It is entirely pos— sible, however, that subjects reduced inequity by increasing their effort on the production of each unit; that is, they increased the quality of their work. This would have the effect of increasing the production time per unit and, therefore, overall productivity would decrease. In terms of the equity hypothesis, this explanation assumes that pieceworkers reduce their inequity by increasing their inputs , in a manner similar to the hourly workers in the Adams and Rosenbaum study. 24 However, the task used in the Adams and Rosenbaum study did not permit measuring quality of work. Adams (1963b) replicated Adams and Rosenbaum' 5 study except that the interview was designed to permit measurement of both quantity and quality of work. The specific hypothesis tested was: Pieceworkers who per- ceive that they are overpaid will produce less work of better quality than pieceworkers who are paid the same rate and perceive they are equitably paid. The results supported the hypothesis. Adams explained the findings of the two previous studies in terms of the arousal of inequity and its subsequent reduction. How- ever, Adams and Jacobsen (1964) stated that the experimental design previously used did not rule out the possibility that the results obtained were a function of perceived job insecurity and subsequent attempts to reduce it. Therefore, they designed an experiment to assess the importance of feelings of job insecurity. One independent variable manipulated by Adams and Jacobsen was dissonance. This variable has three levels. 1. High dissonance (HD). Subjects in this condition were induced to perceive that they were comparatively unquali- fied. They were told that they would nevertheless be paid at the advertised piece rate of $. 30. 2. Reduced dissonance (RD). Subjects in this condition were told the same thing as subjects in the HD condition except 25 that they were told that they would be paid at a piece rate of $. 20. 3. Low dissonance (LD). Subjects in this condition were induced to perceive that they were fully qualified for the advertised piece rate of $. 30. The other independent variable manipulated was job pros- pects. This variable had two levels. 1. High prospects (HP). Subjects in this condition were told that their work would continue for several months and therefore they might be employed again. 2. Low prospects (LP). Subjects in this condition were told that the job was a one time only affair-with the implication that the subject' 5 service would not be required again. The experimental method used in this study was very similar in all respects to that used by Adams in his two studies described above. The major difference was that the task the sub- jects performed was changed to correcting page proofs. Work quality was measured by the number of errors detected. The hypotheses tested were as follows: Hypothesis 1. Subjects in the HD conditions will produce higher quality work than subjects in either the LD or RD conditions. Subjects in the RD and LD conditions will produce work of similar quality. 26 This first hypothesis is consistent with the equity hypothesis. The rationale behind hypothesis 1 was previously discussed in ref— erence to the Adams and Rosenbaum (1962) study. Hypothesis 2. Subjects in the HP conditions will produce higher quality work than subjects in the LP conditions. Also, there will be a significant interaction between job prospects and disso- nance. Specifically, the HD —HP condition subjects will produce a higher quality product than the HD—LP condition subjects. This second hypothesis is consistent with the belief that the subjects were motivated to reduce their feelings of job insecu- rity. Adams and Jacobsen stated that subjects reported that the manipulations of the two independent variables were effective. The results of the study were clear cut. Hypothesis 1 was confirmed. Subjects in the HD conditions found a significantly greater number of errors per page than the subjects in the RD or LD conditions. Hypothesis 2 was not confirmed. No significant main effect of job prospects was found, and no significant interaction effects were found. The results of this study support an equity hypothesis interpretation of the results obtained by Adams in his other studies, and failed to support the alternative interpretation that the results obtained were a function of the subjects' insecurity about job tenure. 27 Arrowood (1961) designed an experiment in which subjects were either overpaid or equitably paid on an hourly basis, and per- formed their work under either public or private conditions. The equitable-inequitable pay distinction was achieved using the same method as Adams. The public-private distinction was achieved by having subjects either submit their work to the ”employer" (the experimenter) or mail it in a preaddressed envelope to New York. In the latter case, the subjects were under the impression that the experimenter would never see their work. The results obtained were entirely consistent with those of Adams and Rosenbaum. In both public and private conditions, overpaid subjects produced sig- nificantly more than equitably paid subjects. There was no signifi— cant interaction between the inequity—equity and public-private dimensions. Unlike the previously cited studies which produced inequity by the manipulation of perceived job inputs (qualifications for the job), Andrews (1967) produced inequity by the manipulation of assigned piece rate. Through pretesting, equitable, overpay and underpay piece rates were determined. Two major hypotheses derived from the equity hypothesis were tested. They were: Hypothesis 1. As compared to workers assigned an equitable piece rate, workers assigned to a lower rate will tend to 28 produce more work of lower quality, whereas workers assigned to a higher piece rate will tend to produce less work of higher quality. Hypothesis 2. Within groups, the more that present wage potential exceeds previous high wage, the lower the work quantity and the higher the work quality. The more the present wage poten- tial is less than previous high wage, the higher the work quantity and the lower the work quality. The basis for hypothesis 2 was the assumption that a worker' s perception of his present pay is influenced by pay rates he has received on other jobs. This emphasizes Adams' position that inputs and outcomes must be evaluated in terms of the worker' s perception of them, rather than in terms of reality. The results of Andrews' study generally confirmed his two hypotheses. The results were in the predicted direction, but not all the differences were significant. Goodman and Friedman (1968) tested three major hypotheses derived from the equity hypothesis. Hypothesis 1. Subjects who perceive they are overpaid and who perceive quantity of production as the basis for equity resolu- tion will produce more units in the hourly condition than subjects who perceive they are equitably paid. This is Adams' original hypothesis, but the task used, questionnaire scoring, was different from that used by him. 29 Hypothesis 2. Subjects who perceive that they are overpaid will increase quality (or quantity) of their inputs if quality (or quantity) is perceived to be the basis for equitable payment. Hypothesis 3. The introduction of ”equitable" production rates will reduce production variance in both overpaid and equitably paid groups relative to comparable groups without production rates. Goodman and Friedman' s method was very similar to the one Adams employed. In general, the results supported the hypothe- ses. Messé (1968) conducted an experiment which tested the predictions of the equity hypothesis. This study was the basis for the present research and so is described below in detail. Messé identified three types of personal dispositions an individual can follow in distributing wages to himself and another individual. These personal dispositions were adapted from Sawyer' s (1966) interpersonal orientations. These personal dispositions are mani— fested in the form of behavioral strategies. The personal disposi- tions and their behavioral counterparts, the stragegies, are pre- sented below. For the following discussion let X represent the wages of Person and Y the wages of 9m. Self—interest. Person may be primarily concerned with how much reward he can get for himself. If Person followed this 30 disposition of self-interest, his strategy for distributing wages would be to maximize X. Altruism. Person may be primarily concerned with how much reward he can get for %. His personal disposition would be altruism and his strategy for distributing the wages would be to maximize Y. Role symmetry. Person may equally weigh the reward of himself and Other. This personal disposition is role symmetry and the major role symmetric strategy is equality. Motivated by role symmetry, Person may feel that both M and himself should receive the same reward. If Person followed his disposition of equality his strategy in distributing the wages would be to minimize Ix - YI . In Messé' 5 study the investments of Person and M were identical. As a result, equity in his study was equivalent to the personal disposition of role symmetry. Therefore, Messé hypothesized that the most frequently manifested personal di sposi- tion in his study would be role symmetry. Messé' s procedure was somewhat complex. First, he had groups of subjects fill out personality inventories. When they were through with this task, each subject filled out a "wage distribution booklet. " The 18 items of this instrument asked each subject to 31 choose between two ways of distributing wages to himself and another individual who had also filled out the personality inven- tories. A sample item from the wage distribution booklet follows: Alternative A Alternative B You get $4.00 You get $2.50 and and He gets $1. 00 He gets $2. 50 Which alternative set of payments would you rather see used? A or B? Half the subjects in Messé' s study were told before they responded that their choices on the wage distribution booklet would actually determine their own wages and the wages of one other individual. The responses of these subjects were called real choices. The remaining subjectswere told that their choices on the wage distribution booklet would have no effect on their own wages or the wages of anybody else. The choices of these subjects were said to be hypothetical. Responses on the wage distribution booklet were used to compute self—interest, altruism and role symmetry scores for each person. The results confirmed Messé' s hypothesis. The most frequently manifested personal disposition was role symmetry, followed by self—interest and then altruism. In addition, subjects making hypothetical choices made significantly more role symmetric 32 responses. He concluded that behavior guided by the personal dispositions can be explained and predicted by the equity hypothesis. The studies cited above offer evidence which suggests that the desire to achieve an equitable state is a motivation present in a broad range of settings, both naturalistic and laboratory. However, the equity hypothesis has been criticized on a number of grounds by both Weick and Lawler. Weick (1966) has pointed out some possible limitations to the equity hypothesis. These limitations can be classified into three major areas. 1. The degree to which inputs and outcomes are independent, both in a theoretical and empirical sense, is ambiguous. Often it is difficult to classify a given behavior as an input or outcome. The nature of the experiments relevant to the equity hypothesis have drawn attention away from this simple but important problem. The settings used in these studies have been sufficiently controlled that subjects have focused on their effort as the salient input and on their salary as the salient outcome. However, in everyday work situations there are considerably more behaviors which can be recognized in determining the input—outcome ratio, and many of these can be assigned to either side of the ratio. This is certainly an issue to which the empirical 33 tests of the equity hypothesis have not addressed them- selves. Adams himself has noted that inputs and outcomes often are intercorrelated. Even when this relationship is only moderate, it would be difficult to reduce inequity. This difficulty arises because any attempt to change one side of the ratio also changes the other side to the extent the two are intercorrelated. Thus, changing the inputs of a person in an attempt to make them more congruent with those of a comparison individual could also change his outcomes. Therefore, a new source of inequity would be created. This is another issue which has been ignored by the proponents of the equity hypothesis. Rather, it seems that they as— sumed that equity can be achieved quite easily when an individual makes an attempt to do so. The equity hypothesis assumes that the standard by which a given person evaluates his input—outcome ratio is always external. However, as Weick points out, it is possible for an individual to evaluate his ratio in terms of some com- posite internal standard which is not directly tied to any external group or person. If an individual uses internal standards in place of social standards, then he might feel 34 his outcomes are not commensurate with his inputs, even though his ratio is congruent with that of the external refer— rent. Thus, feelings of inequity could result even when the input-outcome ratio seems congruent with external criteria. Weick and Nesset (1968), in elaborating on this last point, introduced two terms: Own equity and Comparison equity. Own equity occurs when Person' 8 inputs and outcomes are identical. Comparison equity occurs when Person' 5 input—outcome ratio is identical with his social referrent' s input-outcome ratio. Weick used a questionnaire to present his subjects with pairs of hypotheti— cal work situations. The subjects were asked to select the work situation in which they would be the most comfortable. Among Weick' s findings was that Comparison equity was preferred to Own equity. However, he found that Own equity was also a significant factor that could not be ignored. To summarize, although open to some criticism, the equity hypothesis has a substantial amount of empirical support which attests to its predictive value across a number of settings. Research support for the expectancy hypothesis The experimental support for the expectancy hypothesis is somewhat less than that for the equity hypothesis. However, those 35 studies which have examined the expectancy hypothesis, in respect to the distribution of rewards, are summarized below. Lawler and O' Gara (1967) report research that was more supportive of the equity hypothesis than the expectancy hypothesis but did provide some support for the latter. They noted that most research in inequity ignored underpayment. According to the equity hypothesis, people underpaid on a piece rate basis should produce large amounts of low quality work when compared with people paid an equitable piece rate. Producing large amounts of low quality work would raise the underpaid workers' outcomes (wages) without raising their inputs (effort), and therefore this would help establish equity. On the other hand, the expectancy hypothesis would predict no difference in the quality or quantity of work between equitably paid and underpaid piece rate workers. According to the expectancy hypothesis, equitably paid and underpaid piece rate workers should both be expected to produce as much as possible in order to maxi- mize their wages. As previously reported, Andrews (1967) found that subjects underpaid on a piece rate basis when compared to subjects equitably paid on a piece rate basis produced somewhat more of a lower quality product. However, only the difference in quantity was sig- nificant. Lawler and O' Gara felt that the lack of a significant 36 difference in quality may have been due to the relatively small piece rate difference that Andrews paid his two groups (equitably paid and underpaid). Therefore, the major objective of the Lawler and O' Gara study was to provide a more definitive test of the equity hypothesis by increasing the piece rate differences between the two groups. Unlike Adams' studies which produced inequity by the manipulation of perceived job inputs (qualifications for the job), Lawler and O' Gara, using the same procedure as Andrews, pro-— duced inequity by the manipulation of assigned piece rate. Through pretesting, they determined that $. 10 per interview would be per— ceived as too low a wage and $.25 per interview as an equitable wage. However, the $. 15 difference in piece rate payment between the two groups was much larger than the $. 05 difference that Andrews used. The results obtained by Lawler and O' Gara were consistent with the equity hypothesis. The underpaid subjects produced a sig~ nificantly greater amount of a significantly lower quality product. Further support for the equity hypothesis was obtained in a post experimental attitude questionnaire. Lawler and O' Gara concluded that the equity hypothesis was helpful for understanding the effects of piece rate underpayment. However, the expectancy hypothesis was of little value. 37 The only support for the expectancy hypothesis in the Lawler and O' Gara study was the finding that the importance of money to the subject was significantly associated with his trying to increase his productivity when underpaid. Lawler (1968a) also investigated the effects of hourly over- payment. He noted that Adams had found that overpaid subjects do produce more during a 2%—hour work period than do equitably paid subjects. These findings are consistent with the equity hypothesis. However, these findings are inconsistent with the expectancy hypothe— sis, which predicts that equitably paid and overpaid subjects should produce the same amount in a hourly pay rate situation. The ra- tionale for this is as follows: In an hourly pay rate situation the amount an individual produces is not related to the pay he receives. Therefore, there is no reason to expect overpaid subjects to pro— duce more than equitably paid subjects. Thus, according to the expectancy hypothesis, the high productivity found by Adams for his overpaid subjects should be a transitory phenomenon, if it exists at all. To provide a more definitive test of the two hypotheses, Lawler replicated Adams' overpayment hourly situation. However, he collected data beyond the initial 2%-hour period that Adams used. Instead, his subjects worked three 2—hour periods. 38 Furthermore, Lawler noted that even if the effects of overpayment did disappear after one session, the problem of accounting for the initial high productivity for the overpaid group would still remain. One explanation for their high productivity that is congruent with the expectancy hypothesis and does not require the equity hypothesis has to do with the method Adams has used to pro— duce the feelings of overpayment in his subjects. Adams told his overpaid subjects that they were unqualified for the job. Presumably, the subjects who are told they are unquali— fied will feel inequity because they will see their input—outcome ratio as higher than their comparison individual' 8 ratio. However, Lawler stated that instead of producing feelings of overpayment, Adams might have challenged his subjects' self-esteem. If this is so, then the high initial productivity that Adams found can be explained by assuming that feelings of esteem have positive valence. Thus, it might have been that overpaid subjects, who were insulted, used high productivity as a means of raising their self-esteem. On the other hand, equitably paid subjects, who were not insulted, found this unnecessary. Some evidence to support this interpretation was provided by Friedman and Goodman (1967), who found that their sub- jects were motivated to maintain their self-image. One way to determine the cause of productivity differences between the overpaid and equitably paid subjects is to try to induce 39 overpayment in a way that does not attack the subjects' qualifications. Lawler did this. He told his subjects that, because of a grant he had received, he was required to pay them at a higher rate than that received by people like themselves who did comparable work. Ac— cording to the equity hypothesis, if these subjects felt inequitably overpaid, they should be as high producers as the subjects who are made to feel overpaid by being told they are unqualified. However, according to the expectancy hypothesis, it would be expected that they should behave the same way as the equitably paid subjects since their feelings of esteem were not attacked. As a result, no addi- tional rewards were related to good performance. The method Lawler used was very similar to that of Adams, with the addition of a group of subjects who were overpaid but not told they were unqualified. Also, as previously stated, he continued collecting data beyond the 2%—hour period that Adams had used. Lawler found little difference in productivity between the equitably paid group and the overpaid—by—circumstance group. This result supports the expectancy hypothesis and is contrary to the equity hypothesis. Lawler also found a tendency for the unqualified over— paid group to produce more than either of the other two groups. This was a replication of Adams' findings. The high productivity of the overpaid unqualified group was not a transitory phenomenon, 40 but it continued over all three sessions. However, after the first session, the high productivity was accompanied by a decrease in work quality, a result contrary to the prediction of the equity hypothesis. However, this decrease is consistent with the expec- tancy hypothesis which states that unrewarded effort should decrease over time. It appears that in an hourly overpayment situation the decrease in effort occurs in the form of a decrease in work quality. As a result of this study, Lawler concluded that the expec- tancy hypothesis explains the effects of hourly overpayment better than the equity hypothesis. Gordon and Lowin (1965) performed a study in which subjects in one group were told that they were unqualified and therefore their pay was being reduced. Subjects in another group were told that they were unqualified but their paywas not reduced. Both groups pro- duced at a similar rate. Taken together, the results of this and Lawler' s (1968a) study indicate that, for hourly overpayment, just feeling overpaid does not lead to high productivity. On the other hand, having one' s qualifications threatened does, and this occurs even when feelings of overpayment are not aroused. These results suggest the attack on qualifications is the crucial variable producing differences, a position consistent with the expectancy hypothesis. Lawler _et_a_l_. (1968) investigated the effects of piece rate overpayment. As previously stated, Adams had found that piece 41 rate overpayment leads to a lower level of productivity of a higher quality product than does equitable piece rate payment. This is contrary. to the expectancy, hypothesis, which predictsthat people equitably paid and people overpaid on a piece rate basis should both be high producers, since, for both, each additional unit produced leads to higherwages. Lawlergt_al. replicated Adams' overpayment piece rate situation, but continued to collect data from the subjects'past the 2-hour time period Adams used. Lawler et al. had their subjects work for three 2-hour'sessions. They found that for session 1 the overpaid piece rate subjects produced significantly less of a sig- nificantly higher quality product than the equitably paid subjects. This replicated Adams' finding and it is consistent with the equity hypothesis. However, they also found that these differences in work quantity and work quality dissipated over the next two sessions. By session 3 the productivity of the two groups was very similar. This result is consistent with the expectancy hypothesis. Therefore, Lawler 3321. concluded by stating that neither the expectancy hypothesis nor the equity hypothesis alone can explain the results in the piece rate overpayment situation. Rather, a combination of the two hypotheses is necessary. Thus, there is some experimental support for the expec- tancy hypothesis. However, Adams (1968) has challenged Lawler' s vr 42 (’1968a)‘ study. Adams (1968) stated that the thrust of Lawler' s challenge to equity is based principally on the performance of sub- jects in his overpaid-by-circumstance group. As previously stated, the overpaid-by-circumstance groupwas made to believe that they were qualified for the job. In addition, these subjects were told that their wage rate was to be considerably higher than that paid to I similar subjects on past projects. The subjects were told their high wage rate was due to a grant the researchers had received. The results of Lawler' 5 study indicated that the overpaid-by- circumstance group produced no more and no better a product than did subjects in an equitably paid group. In replying to this study, Adams conceded that the overpaid-— by-circumstance group did feel overpaid. However, he stated that feelings of overpayment are not necessarily the same as feelings of inequity. Lawler assumed that the comparison individuals for the overpaid—by-circumstance group were less well-paid subjects on past projects. Adams clained that it seems more likely that the comparison individuals for the overpaid subjects were other over- paid subjects.working on the same project. Adams felt that these other overpaid individuals provided a more relevant and immediate comparison. If Adams' contention is correct, then the overpaid-by- circumstance subjects would experience no feelings of inequity, and 43 their performance would, therefore, not be expected to differ much from that of the subjects in the equitably paid group. Adams stated that the reasoning he had used to explain Lawler' s findings in terms of the equity hypothesis was based on his own speculative judgments. However, he claimed that Lawler' s explanation of the results was also based on speculative judgments. Clearly, further experimentation is needed to assess the contribu- tion of both the equity hypothesis and the expectancy hypothesis to the understanding of hourly rate overpayment. Adams (1968) concluded his rebuttal with an appeal to the adherents of both major hypotheses. He stated that the contrast which has been drawn between the equity hypothesis and the expec- tancy hypothesis has implied that performance in work situations must be accounted for by m one or the other. Multiple motiva— tional states may influence behavior, though one state may be dominant in an individual at a particular time. Under particular conditions the motivation to achieve equity may dominate; under others maximizing gain may be salient. Adams doubted if anyone could seriously question the fact that the desire to maximize out- comes is a powerful determinant of behavior, and he stated that now there is also sufficient evidence that the desire to achieve justice has a substantial influence on behavior. According to him, the 44 question of greatest importance is not whether the equity hypothesis is better or worse than the expectancy hypothesis, but rather, under what conditions do equity motives and gain-maximizing motives account for certain proportions of observed performance variance. The Present Research Following Adams' suggestion, the purpose of the present study‘was not to determine which of the two hypotheses is best, but rather to determine what each of the two hypotheses contributes to understanding behavior in a multitude of situations. As previously stated, the present research investigated how an individual would distribute an allotted sum of wages to himself and a co-worker when given the opportunity to do so. In Messé' 3 study cited above, there was one major experi- mental group, where all the subjects had identical inputs. For reasons described in Chapter II, it was felt the method Messé used was in some ways superior to those of Adams and Lawler. There- fore, it was decided to use Messé' 3 basic method, but to employ a wider range of conditions for the study than he had. To facilitate the following discussion, Person and 91313, which have already been used by Homans, will be defined in a slightly different manner. The term Person will represent an individual 45 who distributes wages to himself and a co-worker of his. This co-worker, the one who receiveswhatever wages Person allots to him, will be referred to as Other. 3 The two major variables in the present study were the inputs of Person and the inputs of Other. Each variable had two levels, low and high, systematically varied such that four experi- mental conditions were created. Condition 1. Person had high inputs and Other had low inputs. Condition 2. Person had high inputs and Other had high inputs. Condition 3. Person had low inputs and Other had low inputs. Condition 4. Person had low inputs and Other had high inputs. Condition 2 was a replication of the situation Messé used in his study. The other three conditions were an extension to his research. Hypotheses The hypotheses tested in the present study are presented in two sections. Both sections are concerned with the predictions 3Note that the number of underlines is used throughout the present report to differentiate the Person of Homans from the Person in the present research who distributes the wages. Thesame 1s also true of Other and Other. 46 that the equity hypothesis and the expectancy hypothesis make con- cerning the way Person‘will distribute an allotted sum of wages to himself and a co-worker, Other. Specifically, the hypotheses are concerned with predicting which personal dispositions Person will manifest while distributing the wages. The first section of hypothe- ses are predictions about the manifestation of the personal dispo- sitions within each of the four experimental conditions. The second section are predictions about the manifestation of the personal dis- positions between each of the four experimental conditions. Within gfliup hypotheses Condition 1 The equity hypothesis. For Persons in Condition 1, equity and self—interest are equivalent. Therefore, according to the equity hypothesis, the most frequent personal disposition should be self- interest, followed by role symmetry and then altruism. 4 The expectancy hypothesis. For Persons in Condition 1, maximum wages for themselves can be obtained by manifesting the 4The reason why role symmetry was predicted to be more prevalent than altruism under Condition 1 was that role symmetry responses have a self—interest component to them, while altruism responses have none. Therefore, role symmetry should always be more prevalent than altruism when self-interest is predicted to be the dominant response. 47 personal disposition of self-interest. Therefore, according to the expectancy hypothesis, the most frequent personal disposition should be self-interest, followed by role symmetry and then altruism. Under this condition, both the equity hypothesis and the expectancy hypothesis make identical predictions about the way Persons will distribute a given amount of potential'wages. Conditions 2 and 3 The equity hypothesis. For Persons in Conditions 2 and 3, equity and role symmetry are equivalent. Therefore, according to the equity hypothesis, the most frequent personal disposition should be role symmetry, followed by self—interest and then altru- ism. 5 The expectancy hypothesis. For Persons in Conditions 2 and 3, maximum wages for themselves can be obtained by manifest- ing the personal disposition of self-interest. Therefore, according to the expectancy hypothesis, the most frequent personal disposition should be self-interest, followed by role symmetry and then altruism. 5The reason why self-interest was predicted to be more prevalent than altruism under Conditions 2 and 3 is that the equity hypothesis postulates a differential threshold between inequity in one' s favor and inequity against oneself. This implies that a person will beconcerned with the welfare of both himself and others, but self-interest will be somewhat stronger than interest in the other. In addition, Messé (1968) found empirical support for this aspect of the equity hypothesis. 48 For'Persons in Conditions 2 and 3, the equity hypothesis and the expectancy hypothesis make different predictions concerning the distribution of wages. Condition 4 The equity hypothesis. For Persons in Condition 4, equity and altruism are equivalent. Therefore, according to the equity hypothesis, the most frequent personal disposition should be altruism, followed by role symmetry and then self-interest. 6 The expectancy hypothesis. For Persons in Condition 4, maximum wages for themselves can be obtained by manifesting the personal disposition of self-interest. Therefore, according to the expectancy hypothesis, the most frequent personal disposition should be self-interest, followed by role symmetry and then altru- ism. For Persons in Condition 4, the equity hypothesis and the expectancy hypothesis make different predictions concerning the distribution of wages. 6The reason why role symmetry was predicted to occur more frequently under Condition 4 than self-interest was that role symmetry responses have an altruistic component while self- interested responses have none. 49 Between conditions hypotheses Self-interest The equity hypothesis. According to the equity hypothesis, self-interest is equivalent to equity for Persons in Condition 1. Therefore, self-interest should be most frequent in Condition 1, less-in Conditions 2 and 3, and least in Condition 4. The expectancy hypothesis. According to the expectancy hypothesis, self-interest should occur with equal frequency in all four experimental conditions. The rationale for this is that in all four groups Person should be equally likely to try to maximize his own wages. Role symmetry The_equity hypothesis. According to the equity hypothesis, role symmetry is equivalent to equity for Persons in Conditions 2 and 3. Therefore, role symmetry should be most frequent in these conditions, less in Condition 4, and least in Condition 1. The expectancy hypothesis. According to the expectancy hypothesis, role symmetry should occur with equal frequency in all four experimental conditions. The rationale for this is that there is no reason to believe that Persons in the four experimental con- ditions will distribute the wages differently. 50 Altruism The equity hypothesis. According to the equity hypothesis, altruism is equivalent to equity for Persons in Condition 4. There- fore, altruism should be most frequent in Condition 4, less in Con- ditions 2 and 3, and least in Condition 1. The expectancy hypothesis. As with self-interest and role symmetry, the expectancy hypothesis predicts that altruism should occur with equal frequency in all four experimental conditions. The rationale for this is that there is no reason to believe that Persons in the four experimental conditions will distribute wages differently. The following chapter presents a detailed account of the research situation, the instruments used to test these hypotheses, and the variables examined within this situation. II. METHOD The present chapter is a discussion of the method used to test the hypotheses presented in Chapter I. The four sections of this chapter are (a) recruitment and characteristics of the subjects, (b) instruments used, (c) variables and experimental design, and (d) procedure. As noted above, the method used was very similar to that employed by Messé (1968). The Sample Recruitment of Subjects It was essential that the subjects were interested in a monetary reward so that their responses, which determined how much they and a co-workerwould be paid, were meaningful for them. Therefore, it was necessary that the subjects not only volun— teered, but volunteered to make some money. The experimenter told the students in his introductory industrial psychology class that an industrial opinion survey was being conducted on campus, and that male volunteers were being sought. They were told the survey would take from one to two hours of their time, and that they would 51 52 receive no extra credit for the course by participating in the research; instead each subject would be paid up to four dollars for paritipating. Further, anyone interested in making some money should sign his name on one of the sign-up sheets that were passed around. The sign-up sheets were then distributed. They listed several convenient periods of time for which a subject could volun— teer. From the sign-up sheets, four times to run subjects were selected in terms of the number of subjects that would be available. In the class period following the one in which sign-up sheets had been passed out, those subjects who had volunteered were given a card indicating the time and room they should be at for the study. Characteristics of the Subjects Fifty-eight male students at Michigan State University served as subjects. The reason that all the subjects were of one sex (male) was that the sex of an individual' s co-worker has been shown to affect the way in which the individual distributes the wages to himself and that co-worker (Messé, 1968). An investigation of the combined effects of sex of the co-worker and differential inputs did not seem worth the cost of running additional groups of subjects. The subjects ranged in age from 19-35, with the large majority being either 20 or 21 years old. Most of them were either 53 juniors or seniors, and they had a wide range of majors. However, more than half were majoring in either social science or business. Instruments There were fourinstruments used in the present research: a wage distribution booklet, an attitude questionnaire, an industrial Opinion questionnaire, and a student-background questionnaire. Wage Distribution Booklet The subjects' responses to the Wage Distribution Booklet (presented in Appendix I) were the dependent variable. The Wage Distribution Booklet used in the present research was a shortened version of the one used by Messé. Since it is the instrument used to measure the dependent variable, it is described in considerable detail below. 1 The first part of the discussion is concerned with the selection of the six items used in the Wage Distribution Booklet. Let two persons perform an identical task. Their employer could then say to one of them, "Suppose for the work you both have done, we are to pay you $4.00 and him $1. 00. " This base set of payments to which an alternative set is to be compared will be called the standard distribution. "Would you rather have me give you each 1The following discussion borrows heavily from the one which Messé used to describe his booklet. 54 $2. 50 instead?" This is the alternative outcome to the standard distribution. "What would you say if the choice-were between $4. 00 for you and $1.00 for him (standard distribution) versus $2.00 for you and $3.00 for him (alternative outcome)?" In both cases, an individual motivated by either altruism or role symmetry would choose the alternative outcome, while a person motivated by self-interest would choose the standard distri- bution. In the first case, the standard distribution provides the chooser with a greater reward ($4.00 versus $2. 50). On the other hand, the‘alternative outcome gives a greater reward to his co- worker ($2. 50 versus $1.00), and yields a more equal distribution of rewards ($0. 00 difference versus a $3.00 difference) than does the standard distribution. In the second case, the standard distri- bution gives the chooser a greater reward ($4. 00 versus $2. 00) than the alternative outcome, and a greater reward than that of the co- worker ($4. 00 versus $1.00). The alternative outcome gives the co-worker a greater reward ($3. 00 versus $1.00), and distributes the rewards more equally (a $1.00 difference versus a $3.00 dif- ference) than does the standard distribution. However, the alter- native outcome gives the co-worker more than the chooser ($3. 00 versus $2.00). These differences in rewards to both are due to two inde- pendent dimensions of the payoff structure. A person--if he is to 55 make a rational choice—-must address himself to these dimensions, since their various combinations determine just what his alternatives are. These two dimensions are: 1. Relative welfare of the standard distribution Does the standard distribution give the chooser: (a) more than the co—worker? (b) the same as the co-worker? (c) less than the co-worker? 2. Relative‘welfare of the alternative outcome If the chooser selects the alternative outcome, does he obtain: (a) more than the co-worker? (b) the same as the co—worker? (c) less than the co-worker? Since these dimensions are independent, nine unique situa- tions can be generated by varying the possible combinations of dif— ferent cases which occur in each. The number of situations can be reduced to six by eliminating case (b) of the standard distribution dimension. Case (b) of the standard distribution was eliminated because it was felt that the standard distribution case which gave both persons the same reward was, by far, the least interesting; 56 the more extreme cases of being either ahead or behind the other were thought to present the subjects with more meaningful and more difficult choices. Table 1 presents these six situations modified by the follow- ing restrictions which further define their structure. (a) The sum of the standard distribution rewards is always $5. 00, divided into either $4. 00 for the chooser and $1. 00 for the co-worker, or $1. 00 for the chooser and $4.00 for the co-worker. (b) The sum of the alternative outcome rewards was also always $5.00, divided into either $3.00 for the chooser and $2. 00 for the receiver, $2. 50 for the chooser and $2. 50 for the receiver, or $2. 00 for the chooser and $3. 00 for the receiver. (c) When the two rewards in an alter— native outcome are unequal, they always differ by $1. 00. It was hoped these restrictions would give each subject a standard frame of reference, and therefore enable him to make more meaningful choices. Thus, Table 1 indicates the standard distributions and alternative outcomes presented to the subjects. For example, in the first row of Table 1 (series X situation (1)), if the individual determining the wages selected the standard distribution, he would receive $4. 00 while the co-worker whose wages he was determining would get $1.00. However, if the individual selected the alternative 57 outcome instead, he would get $3. 00 and his co-worker would receive $2. 00. Table 1. -- The choice situations Standard Distribution Alternative Outcome Series X (1) $4.00 $1.00 $3.00 $2.00 (2) $4. 00 $1. 00 $2. 50 $2. 50 (3) $4.00 $1.00 $2.00 $3.00 Series Y (1) $1.00 $4.00 $3.00 $2.00 (2) $1.00 $4.00 $2.50 $2.50 (3) $1.00 $4.00 $2.00 $3.00 The outcomes presented in Table 1 in terms of standard distribution and alternative outcome were given more neutral labels --Alternative A and Alternative B, respectively--and put in the form of a six page Wage Distribution Booklet. This booklet is presented in Appendix I. It had two forms, A and B. In Form A the Series X items were presented first, while in Form B the Series Y items were presented first. For convenience, a sample page of the Wage Distribution Booklet is presented as Table 2. The order that the 58 Table 2. -- A sample choice situation Alternative A You get $4. 00 and He gets $1. 00 Alternative B You get $3. 00 ' and He gets $2. 00 \I/ 1. Which alternative gives you more? A or B? (Check one) 2. Which alternative, if either, gives the greatertotal to the two of you, together? Check A or B. If the two give the same total, check "Equal. " \V 3. Which alternative gives the two of you more nearly the same? A or B? (Check one) SKIP THE QUESTION BELOW: GO ON TO THE NEXT PAGE CHOICE X1: Which alternative set of payments would you rather see used? A or B? (Check one) Remember: If this choice is the one drawn, you and he will be paid just the amounts you now choose. (X1) 59 Series X and Series Y items appeared in the Wage Distribution Book— let was the same order in which they appeared in Table 1. As can be seen from Table 2 and Appendix I, each page of the booklet contained one of six choice situations. Underneath each choice situation was a set of three questions which were identical on all pages of the booklet. The questions served two purposes: (a) they insured that the subject was aware of the specific structure of each choice; (b) they allowed the experimenter to check if each subject understood the structure of the choices. None of the answers given by any subject on these three questions were in error. Beneath each set of questions was another question. This asked each subject to select which of the two sets of payments (Alternative A or Alterna- tive B) he preferred to see used. From a person' 5 choices on the Wage Distribution Booklet it was possible to infer the strengths of the various personal dispo- sitions: self—interest, role symmetry, and altruism. This was done by comparing the individual choices in the three item pairs which were mirror images of each other. Item 1 in Series X and Item 3 in Series Y are a mirror image pair. Likewise, Item 2 in Series X and Item 2 in Series Y are a mirror image pair, as are Item 3 in Series X and Item 1 in Series Y. These three pairs of items are mirror images because for each pair the values in the 60 standard distributionand alternative outcome are interchanged from one to the other. The method by which the strength of each of the personal dispositions was computed is presented below. The index of self-interest If a person chooses Alternative A in the Series X member of a mirror image pair, but Alternative B in the Series Y member, he is displaying self-interest. This is because his choices indicate that he is concerned only-with obtaining as much money as he can for himself. Each such display contributes one point to his self- interest score. Since there are three mirror image pairs, a per- son's self-interest score can range from zero to three. The index of role symmetry If a person chooses either Alternative A or Alternative B for both members of a mirror image pair, he is displaying role symmetry. This is because his choices indicate that he is concerned with both the wages of himself and the co-worker for whom he is determining the wages. As with self-interest, a person' s role sym- metry score could range from zero through three. The index of altruism If a person chooses Alternative B in the Series X member of a mirror image pair, but Alternative A in the Series Y member, X 61 he is displaying altruism. This is because his choices indicate that he is concerned only with obtaining as much money. as he could for the co—worker for whom he was determining the wages. As with both self-interest and role symmetry, a person' s altruism score could range from zero through three. Relative welfare In addition to providing a measure of the strength of each of the personal dispositions, the six items on the Wage Distribution Booklet permitted the examination of the effects of the reward structure variable of relative welfare on the manifestation of the personal dispositions. Relative welfare refers to whether an individual receives more than, less than, or the same amount of money as the co-worker whose wages he is determining. As can be seen in Table 1, for Series X item (1), both the standard distribu- tion and the alternative outcome give the chooser more money than his co—worker. Therefore, both alternatives have positive relative welfare for the chooser. However, for Series X item (3), the standard distribution gives the chooser more money than his co- worker, while the alternative outcome gives the co-worker more than him. Therefore, the standard distribution has positive relative welfare for the chooser and the alternative outcome has negative relative welfare for him. 62 Attitude Questionnaire The Attitude Questionnaire (see Appendix II) contained five rating scale items and two open—ended questions requiring brief answers. The main purpose of this questionnaire was to determine the degree of success that had been attained in manipulating the independent variables. Two of the questions (items 2 and 3) are concerned with the degree to which subjects perceived differences in work quantity and the extent to which these differences affected choices on the Wage Distribution Booklet. Two other questions (items 4 and 5) asked for the same information, but with work quality as the relevant variable. The remaining questions were concerned with other variables that might have affected choices on the Wage Distribution Booklet, as well as the student' 5 interest in and per- ception of the study. Indust rial Opinion Que sti onnai re The Industrial Opinion Questionnaire (see Appendix III) had two forms, Research Form 1 and Research Form 2. Form 1 con- tained four short essay questions which asked the subjects about their opinions concerning areas in the business world. Form 2 con- tained twelve such questions, including the four from Form 1. The specific number of questions in each form, four and twelve, was 63 selected so that the subjects would work the desired length of time, 30 minutes or 90 minutes. The main criterion for selecting the particular twelve items used was the apparent easerwith which written responses could be made to them. Student Background Questionnaire The Student Background Questionnaire (see Appendix IV) contained several biographical items, and a short essay item asking for the subject' s attitude toward labor unions. As discussed below, the subject' 5 responses to this instrument were used as the reason why they were given tasks of different lengths. Variables and Experimental Design Four situational conditions were manipulated. Two of these, (a) hypothetical versus real choices and (b) order of presentation of the choices series, were manipulated mainly for reasons of control. The other two variables, (0) inputs of the subject and (d) inputs of his partner, were the main focus of the research. Situational Variables Hypothetical versus real choices All subjects, after completing the Industrial Opinion Ques- tionnaire, filled out the Wage Distribution Booklet. Half the subjects 64 were each told that their responses on the booklet would determine not only their'wages but thewages of a co-worker. These subjects were told that they would be making r_ea_l choices on the booklet. The other half of the subjects were told that their responses. on the Wage Distribution Booklet would have no effect on the wages they received nor on the-wages anyone else received. These subjects were told that they would be making hypothetical choices. Order of presentation Messé (1968) found that the order of presentation of the six items in the Wage Distribution Booklet (whether the Series X or Series Y items were presented first) had both a significant effect on the responses overall and interacted with a number of other variables. Although-this phenomenon was more or less irrelevant to the specific theoretical concerns of the present research, Messé' s findings implied that this variable should be manipulated for reasons of con- trol. Therefore, half the subjects were given booklets which had the three Series Xitems first, followed by the three Series Y items. The remaining subjects were presented the series in reverse order. Inputs of the subject Before each group of subjects began, they were informed, supposedly on the basis of their responses to the Student Background 65 Questionnaire, if they were expected to produce high or low quality answers to the items on the Industrial Opinion Questionnaire. The subjects were told that those persons expected to produce high quality answers were being given a longer questionnaire than those expected to produce low quality responses. The more extensive questionnaire required three times as long to complete. Therefore, the variable inputs of the subject had two levels, low and high. Those subjects who had low inputs worked one-third as long as those with high inputs, and they were told that their answers were expected to be of low quality. Those subjects who had high inputs worked three times as long as those with low inputs, and they were told that their answers were expected to be of high quality. Inputs of partner Subjects were not told the specific identity of the other indi- vidual (their partner), but they were told whether those in the other group had high inputs (long Industrial Opinion Questionnaire and ex- pected to produce high quality answers) or low inputs (short Industrial Opinion Questionnaire and expected to produce low quality answers). Experimental design As noted above, four situational conditions were system- atically manipulated. This produced a factorial design whose 66 dimensions were 2 (inputs of subject—-high versus low) X 2 (inputs of partner--high versus low) X 2 (choices--real versus hypothetical) X 2 (order of presentation——Series X first versus Series Y first). Thus, the design required sixteen cells. Procedure There were two major features of the experimental pro- cedure: (a) one subject determined the reward for himself and another subject; (b) the subjects were run in groups, and not indi- vidually, as has been customary in this area of research. The rationale for these procedures is presented below. It is rare in ”real life” for one individual to freely deter- mine the rewards for himself and another individual. In real life most encounters take the form of interactions. In an interaction, each person involved partially determines the rewards of all. Therefore, the factor of reciprocity enters into real life situations. However, if an individual is given the opportunity of determining rewards for both, he can more easily express any personal dispo- sition he wishes, since he does not have to be concerned with the actions of the other person. In the present research, the individual distributing the wages can do what he wants without considering the other' 8 reactions. Thus, as with most experimental studies, the justification for the sacrifice of realism is that it makes the 67 behavioral expression of the personal dispositions more directly accessible to investigation. As previously stated, the subjects were run in groups. The procedure generally used in equity and expectancy research involved the subjects working on a task in the absence of co—workers. The subject, in previous research, was first told that he was either qualified or unqualified for a job, and therefore, entitled or not entitled to the pay associated with the job. The subject was then hired for the job and he went out and performed it in the absence of co-workers. Running subjects in groups has some distinct assets. First, it is more efficient in terms of data collection. Second, it provides a different experimental setting for the research than those pre— viously used. The use of different experimental settings is desirable in terms of testing the generality of theories. Consistent results obtained using different procedures give additional support for a theory beyond that provided by a simple replication. Inconsistent results from studies with different experimental settings frequently generate reasons for these discrepancies which often lead to a bet- ter understanding of the variable in question. Third, in real life individuals often work in the presence of co-workers. Therefore, the procedure used in the present research provides an element of realism not present in previous studies. 68 Perhaps the most important advantage of running subjects in groups pertains to the area of reference individuals. In Adams' and Lawler' 5 research, the reference individual for the subject was hypothesized to be a co-worker described to the subject by the experimenter. This co—worker was never seen by the subject. In the present research the reference individual (the individual for whom the subject is determining the wages) is a visible co-worker. In the present research the subject is able to see the amount of work his co—worker actually does, while in previous research the behavior of the hypothetical co—worker had to be imagined. As previously stated, there were four experimental condi— tions. The procedure used in the four conditions was identical except for two features. First, the part of the oral instructions per— taining to what quality answers the experimenter said he expected the subjects to produce was different for the four conditions. Also, the time the subjects spent filling out the Industrial Opinion Ques— tionnaire was different for the four conditions. The procedure used for the group where the individuals working 1% hours determined the wages for themselves and for their co—workers who worked % hour is described in detail, below. The experiment was run in a classroom at Michigan State University. The room was arranged so that the 48 seats were 69 placed in two groups of 24. Each group of 24 seats was arranged into three rows of eight. A large aisle separated the two groups of 24 «seats. In each group, subjects were seated in the two rows closest to the aisle. This was done so that each subject could see the remaining subjects in the room. When a subject arrived for the study, the experimenter asked the subject for his name. He then told the subject which group of seats he was to sit in, and he also told the subject to sit in one of the two rows of seats closest to the aisle. The subjects had been randomly assigned to one of the two groups. Prior to arriving for the experiment, one group of subjects had been told they were to fill out a questionnaire for 1% hours, while the other group had been told they were to fill out a questionnaire for % hour. When the subjects arrived for the study they did not know that there would be other subjects who were filling out a questionnaire for a different length of time than they. 2 Five minutes after the time the subjects were told to arrive the experimenter began with the instructions. 3 The oral instruc- tions for the subjects were as follows: 2The individual running the subjects was not the experi- menter. The reason the experimenter did not run the subjects was that they were his students and his presence might have had an effect on the subjects' responses. 3One subject arrived an hour late. He was told to go home since the experiment was almost over. 70 ”Each of you has volunteered to participate in this industrial opinion study. This morning you will fill out a questionnaire asking your opinions about some important industrial issues. Because you have varied backgrounds pertaining to your work experiences, grade point, etc. , you have been separated into two groups. ”We have two different questionnaires. People on the left side of the room will fill out their questionnaire for 1% hours. People on the right side will fill out a shorter questionnaire for % hour. "Based on the questionnaire you filled out in your Psychology 255 class, we feel that the people in the 1% hour group should be better able to answer the longer questionnaire, and to produce higher quality answers on it. However, we are also interested in the responses of the people in the % hour group. "I will periodically indicate the amount of time remaining on the board in front of the room. When the time elapses, I will collect the questionnaire. You will be required to remain the full time. The group on the left will remain 1% hours and the group on the right % hour. Please answer all questions and return at one o' clock when you will be paid. Please put your name on the first page of the questionnaire. Do you have any questions ?” When % hour had elapsed the experimenter said, ”The group on the right please stop and hand in your papers. I will see you back 71 here at one o' clock when you will be paid. The group on the left still has an hour remaining. " When the remaining hour had elapsed the experimenter then said, "Please stop writing and hand in your papers. I will see you at one ol clock when you will be paid. " When the subjects arrived at one o' clock they were told to sit in the same group of seats as they did in the morning. At 1:05 all the subjects had arrived. The experimenter then read the fol— lowing instructions, 4 "Now we are going to determine what your wages will be. Your wages will be determined by your answers on the Wage Distribution Booklet which I am now passing out. Please put your name on the front of the booklet. Now open your booklet and you will notice that there are six pages. Each page contains two alternative ways of distributing the wages. You will notice that you are asked to distribute wages to yourself and one other individ- ual. The answers on this booklet of those individuals who worked 1% hours this morning will determine the wages for themselves and one individual who worked % hour this morning. Even though the choices of the people on the right side of the room (experimenter pointed to those who had worked % hour) will not have any effect on 4The oral instructions were pretested by having 20 people listen to them one at a time and then answer the question, "What is unclear about these instructions ?" Many modifications of the original instructions were made. 72 the way the wages are to be distributed, we are interested in how you would have distributed the wages if you were given the oppor- tunity to do so. Therefore, you should also fill out the booklet. " The experimenter then went through the following example which indicated how the wages were to be distributed. 1. The subjects were told to look at each page of the booklet and to notice that they were asked to select one of two ways of distributing wages to themselves and another individual. They were also told that the other individual they were asked to determine the wages for was someone in the room, but he was in the other group. They they were informed that the experimenter had paired each subject in one group with another. individual in the other group for the purpose of distributing the wages. The specific identity of the sub- ject' s "partner" was not told to him, but as previously stated, he knew this individual was seated across the room from him. The subjects were reminded that the responses on the Wage Distribution Booklet of those who had worked 1% hours in the morning would determine their wages and the wages of their "partner" in the other group, while the responses of those who had worked % hour would have no effect on the way the wages were to be distributed. 73 They were then shown that each page of the Wage Distribution Booklet had a letter (X or Y) and a number (1, 2 or 3) at the bottom right hand corner. They were told that all the Wage Distribution Booklets had the same six pages (X1, X2, X3, Y1, Y2 and Y3) but that they were in different orders. The experimenter then held up six cards. Each card con— tained either X1, X2, X3, Y1, Y2 or Y3. These six cards each corresponded to a page in the Wage Distribution Book— let. The subjects were told that after they had all completed their Wage Distribution Booklets that someone who had worked % hour would select at random without looking one of the six cards (X1, X2, X3, Y1, Y2 or Y3). The experi- menter held up the six cards again. The subjects were told that the card selected would determine which choice situa- tion would actually be used to distribute the wages. This one card would serve as the indicator of the choice situa— tion for all "partners" but the actual distribution of wages for each pair of subjects would be determined by the re— sponses of each individual who had worked 1% hours in the morning. It was then pointed out that only one choice situation would , be used to determine the wages for a pair of subjects, but 74 that each situation had the same probability of being picked. Therefore, all the choices were important. 7. The experimenter then selected one of the six cards at random and he then showed what the possible wages would be if this choice were selected. 8. He then asked for and answered questions concerning the distribution of wages. The subjects completed the Wage Distribution Booklet, after which the experimenter then collected the booklets and passed out the Attitude Questionnaire, which the subjects completed. A subject was then called up to select one of the six cards at random, and his choice was announced to all subjects. The experimenter explained what this choice represented in terms of the alternative ways the wages could be distributed. The experi— menter examined the Wage Distribution Booklets of the individuals whose choices determined how the wages were to be distributed. The wages were then given to all the subjects in sealed envelopes. The entire procedure——the experimenter reciting the instructions, the subjects filling out both the Wage Distribution Booklet and the Attitude Questionnaire, and the experimenter deter- mining and distributing the wages-—required only about 25 min- utes. 75 As previously stated, the procedure used in the four experi— mental groups was very similar. The procedure cited above was for the group in which the subjects working 1% hours determined the wages for themselves and their co—workers who worked % hour. The only difference in procedure in the four experimental groups involved the length of time the subjects worked and the part of the oral instruc- tions involving what the experimenter said about the quality of the answers he expected the subjects to produce. These differences are further described below. To the group of subjects where everyone worked 1% hours the experimenter said, ”Based on the questionnaire you filled out in your Psychology 255 class, we feel that all the people in this group should be able to answer this long questionnaire and to pro— duce high quality answers on it. ” To the group of subjects where everyone worked % hour the experimenter said, "Based on the questionnaire you filled out in your Psychology 255 class, we feel that all the people in this group will produce low quality answers and therefore we are giving you a short questionnaire. " The procedure for the group in which the subjects who worked % hour determined their wages and those of a co-worker who worked 1% hours was identical to that of the group in which the subjects 76 who worked 1% hours determined their wages and those. of a co- worker‘who worked% hour, except the individuals determining the wages were the ones who had worked % hour. The following chapter presents the results obtained by using‘the method described in this chapter. III. RESULTS The present. chapter is divided into three sections. The first section presents an examination of the situational variables discussed in Chapter-II. The second section is concerned with further analyses of the effects of inputs of subject and inputs of his partner. The third section is concerned with the effect of the variable of relative welfare. Effects of Situational Variables on the Manifestation of the Personal Dispositions Since there were no altruistic responses, role symmetry and self-interest were perfectly inversely related. For this reason, . for the most part, only role symmetry is referred to in the follow- ingsections. The reader should keep in mind, however, that the opposite of any statement made concerning role symmetry is true of self-interest. The effects of the four situational variables, inputs of the subject, inputs of his partner, real versus hypothetical choices, and order of presentation, on the manifestation of role symmetry were 77 78 determined by a four-way unweighted means analysis of variance, summarized in Table 3. The summary table indicates that the third order inter- action was significant at the .01 level. Therefore, tests of simple effects were performed to determine under which conditions each of the four situational variables had a significant effect on the mani- festation of role symmetry. Table 4 presents the cell means for role symmetry scores; for convenience and ease of identification in later discussion, each cell mean has a letter next to it. Inputs of Subject The main effect for inputs of subject was essentially zero (F = .07). Also none of the first or second order interactions involving the variable were significant, but as noted earlier, the third order interaction was significant. Exploration of the contri- bution of inputs of subject to this interaction through analyses of the simple effects revealed one significant difference between means ((1) and (1) (see Table 4). This difference indicates that, when a subject' 3 partner had high inputs and the subject made hypothetical choices starting with Series Y, the subject tended to be more role symmetric when he had high inputs than when he had low inputs (F = 5.59; d.f. = l, 42; p< .05). 79 Table 3. -- Summary of the analysis of variance for role symmetry scores Source of Variation d. f. M. S. F Inputs of subject 1 O. 07 O. 07 Inputs of partner 1 17. 06 17. 58** Real versus hypothetical choices 1 2. 66 2. 74 Order of presentation 1 4. 71 4. 85* Inputs of subject X inputs of partner 1 0. 00 O. 00 Inputs of subject X real versus hypothetical choices 1 2' 34 2' 41 Inputs of subject X order of presentation 1 1. 42 1. 46 Inputs of partner X real versus hypothetical choices 1 0' 46 0' 47 Inputs of partner X order of presentation 1 6. 16 6. 35>:< Real versus hypothetical ch01ces X order 1 O. 42 0. 43 of presentation Inputs of subject X inputs of partner X . . 1 1. 56 1. 61 real versus hypothetical ch01ces Inputs of subject X inputs of partner X 1 0. 14 0. 14 order of presentatlon Inputs of subject X real versus hypothetical choices X order of 1 0. 56 O. 58 presentation Inputs of partner X real versus hypothetical choices X order of 1 4. 89 5. 04* presentation Inputs of subject X inputs of partner X real versus hypothetical choices X 1 9. 84 10. 14** order of presentation Error 42 . 97 * p < . 05 >I‘*p< .01 80 .m 3 0 89C 53> Eco omHBox: 3:00 3an wit 5 swim monoom hypogahm 39H smog 9: .m 8. 0 89¢ Eco hawk, 3:00 HooEHHm Some .HoM mouoom hugger? 20h o5 oochw 3:9: 33 A3 rm .o on oo H A5 rm .0 Ads hm .o was hoauhmnm $59: .33 93 H835. . . . . 9.:qu HHNHHH E 8 H 3 on N 3 8 H 3 ms N 92H Satan . . . . 3:9: 33 3 E o AB 8 H CH 8 H H8 8 0 m2 esteem 335 SE was H836 . . . . mBSHHHmHHH :3 3 N 3 ON N 3H 8 o 3 N N was sunfish Hmsa poEomonQ V moEwm «mum popcommsm N moiom “mum Uwuqomoug W moflwm “mum pmucomona M moEom moo Hone Hmo 3.230%: moxma Hocflnsm mooHoHHo Boa mmxma wooflnsm .m monoom buofifihm 29H .HocH madman Sou I .v 2an 81 Inputs of Partner The main effect for inputs of partner was significant (F = 17.58; d. f. = 1, 42; p < .01). One first order interaction, inputs of partner X order of presentation, was also significant (F = 6. 35; d.f. = 1, 42; p < .05), as was one second order interaction, inputs of partner X real versus hypothetical choices X order of presenta- tion (F = 5.04; d. f. = 1, 42; p < .05). Exploration of the significant third order interaction through analyses of the simple effects revealed four significant differences (see Table 4). They were between means (a) and (e), (d) and (h), (i) and (m), and (k) and (o). The difference between (a) and (e) indicates that when subjects have high inputs and they make real choices starting with Series X, they tend to be more role symmetric when their partners have high inputs than when their partners have low inputs (F = 9. 24; d. f. = 1, 42; p < .01). The difference between ((1) and (h) indicates that when sub- jects have high inputs and they make hypothetical choices starting with Series Y, they tend to be more role symmetric when their partners have high inputs (F = 7. 87; d.f. = 1, 42; p < .01). The dif— ference between (i) and (m) indicates that when subjects have low inputs and they make real choices starting with Series X, they are more role symmetric when their partners have high inputs (F = 7. 87 d. f. = 1, 42; p < .01). The difference between (k) and (0) indicates . 1 82 that when subjects have low inputs and they make hypothetical choices starting with Series X, they tend to be more role symmetric when their partners have high inputs (F = 4. 09; d. f. = 1, 42; p < .05). Real Versus Hypothetical Choices The main effect for real versus hypothetical choices was non-significant (F = 2. 74; d.f. = 1, 42; p > .05). None of the first order interactions were significant. One second order interaction, inputs of partner X real versus hypothetical choices X order of pre- sentation, was significant (F = 5.04; d. f. = 1, 42; p < .05). Explo- ration of the significant third order interaction through analyses of the simple effects revealed one significant difference between means (b) and (d). This difference indicated that when subjects have high inputs, their partners have high inputs, and they start with Series Y, these subjects tend to make more role symmetric responses when their choices are hypothetical (F = 13. 79; d. f. = 1, 42; p < .01). Order of Presentation The main effect for order of presentation was significant (F = 4. 85; d.f. = 1, 42; p < .05). One first order interaction, inputs of partner X order of presentation, was significant (F = 6. 35; d.f. = 1, 42; p < .05), as was one second order interaction, inputs of partner X real versus hypothetical choices X order of presentation 83 (F = 5.04; d. f. = 1, 42; p < .05). Exploration of the significant third order interaction through analyses of the simple effects revealed three significant differences (see Table 4). They were between means (a) and (b), (i) and (j), and (k) and (l). The difference between means (a) and (b) indicates that when subjects have high inputs, their partners have high inputs, and they make real choices, those who start with Series X are more role symmetric than those who start with Series Y (F = 9.24; d. f. = 1, 42; p < .01). The dif- ference between means (i) and (j ) indicates that when subjects have low inputs, their partners have high inputs, and they make real choices, those who start with Series X are more role symmetric than those who start with Series Y (F = 5.59; d. f. = 1, 42; p < .05). The difference between means (k) and (1) reveals that when subjects have low inputs, their partners have high inputs, and they make hypothetical choices, those who start with Series X are more role symmetric than those who start with Series Y (F = 4. 09; d. f. = 1, 42; p < . 05). The situational variables of real versus hypothetical choices and order of presentation were included in the present research mainly as controls. The focus of the present research, as noted in Chapter I, was on the variables of inputs of the subject and his partner. Therefore, both the degree to which their manifestation 84 was successful and their effects on choices were explored through additional analyses, the results of which are presented below. Further Analyses of the Effects of Inputs of Subject and Inputs of Partner Validity of Manipulati ons As noted in Chapter II, the degree of success attained in manipulating the variables of inputs of the subject and his partner was measured by the subject' 8 responses to questions 2 and 4 of the Attitude Questionnaire. These responses are summarized below in Tables 5 and 6. Table 5. -- Percent scale —point responses for work quantity tabulated within each experimental condition Scale—point responses As compared to his partner, Experimental a subject reported working cond1t1on 1 2 3 4 5 4 times about 2 about about % as much to 3 times the % to% as much or more as much same as much or less 1 (1:30 — :30)a 14 86 o o o 2 (1:30 — 1:30) 0 6 94 0 0 3 ( :30 - :30) 0 0 100 0 0 4 ( :30 — 1:30) 0 7 33 60 0 aThis indicates that the 1% group determined the wages for themselves and their partners in the % group. 85 Table 5 presents the percent of scale-point responses for work quantity tabulated within each experimental condition. Table 6 pre- sents the same information pertaining to work quality. Table 6. -- Percent scale-point responses for work quality tabulated within each experimental condition Scale—point responses As compared to his partner, a subject felt the quality of his work was Experimental cond1tion 1 2 3 4 5 Consider- Some- About Some— Consider— ably what the what ably better better same worse worse 1 (1:30- :30) 14 50 36 0 0 2 (1:30 — 1:30) 0 17 71 12 0 3 ( :30 — :30) 8 0 92 0 0 4 ( :30 — 1:30) 0 7 86 7 0 These results are then discussed below for each experimental con- dition. Condition 1 (Subject had high inputs and his partner had low inputs) All fourteen subjects in this condition indicated that they felt they did at least two to three times as much work as their partners. In addition, nine of the fourteen subjects felt that the 86 quality of their answers was either somewhat or considerably better than that of their partners. These results indicate that all the subjects in Condition I felt that they had higher inputs than their partners and, therefore, the independent variables were success- fully manipulated. Condition 2 (Subject had high inputs and his partner had high inputs) Sixteen of the seventeen subjects indicated that they felt they worked about the same amount as their partners. Also, twelve of these seventeen subjects felt that their answers were of the same quality as those of their partners. These results indicate that most of the subjects in Condition 2 felt that they had approximately the same inputs as their partners and, therefore, the independent vari— ables were successfully manipulated. Condition 3 (Subject had low inputs and his partner had low inputs) A11 twelve subjects in this condition indicated that they felt they had worked about the same amount as their partners and eleven of these subjects felt the quality of their answers was the same as their partners. Therefore, it seems that almost all the subjects in Condition 3 felt that they had almost identical inputs as their partners 87 and, as a result, the independent variables were successfully manipulated. Condition 4 (Subject had low inputs and his partner had high inputs) Nine of the fifteen subjects indicated that they felt they worked about % to% as much as their partners, while five of the remaining six subjects felt they worked as long as their partners. On the other hand, almost all (13 or 15) the subjects felt that their answers were of the same quality as those of their partners. There— fore, about half the subjects in Condition 4 felt their inputs were as high as their partners, while the other half felt their inputs were only somewhat lower than their partners. Since the independent variables were successfully manipulated in the other three experi— mental conditions, the difference in Condition 4 seems to reflect a cognitive distortion on the part of the subjects. Tests of the Predictions Generated from the Equity and Expectancy Hypotheses The hypotheses tested in the present research were pre- sented in two sections of Chapter I. The first section was concerned with the manifestation of the personal dispositions within each of the four experimental conditions, while the second section was concerned 88 with the manifestation of the personal dispositions between the experimental conditions. The results obtained to test these hypothe- ses are presented below. Tests of the within group hypotheses The results obtained relating to the within group hypotheses are summarized in Tables 7 and 8, presented below. Table 7 sum— marizes the percentage of role symmetry and self-interest responses found in the four experimental conditions, while Table 8 classifies each subject in a given experimental condition by the personal dis- position which was his majority responses. Table 7. —- Percentage of role symmetry and self—interest responsesa Percentage Percentage Condition Inputs of Inputs of of role of self— the subject his partner symmetry interest responses responses 1 High Low 21 79 2 High High 61 39 3 Low Low 25 75 4 Low High 62 38 aThe category of altruism has been omitted from this and subsequent tables because no altruism responses occurred in any condition. 89 sition classified by experimental condition Table 8. -- Number of subjects manifesting a majority of each dispo— Number of subjects Number of subjects manifesting manifesting Condition self—interest role symmetry p more frequently than more frequently than role symmetry self—interest 1 12 2 . 01 2 5 12 14 3 8 4 28 4 5 10 30 The following discussion is concerned with the results pre- sented in the two tables above. Condition 1 Seventy-nine percent of the responses made on the Wage Distribution Booklet by subjects in Condition 1 were self-interest responses, while the remaining twenty—one percent were role sym- metry responses. In addition, a significant number (12 of 14) of subjects acted in a self-interested manner on at least two of the three mirror—image pairs of items on the Wage Distribution Book- let (p < .01; sign test). The results obtained in Condition 1 were consistent with both the equity and expectancy hypotheses. 90 C onditi on 2 Sixty-one percent of the responses made on the Wage Distribution Booklet by subjects in Condition 2 were role symmetric, while the remaining thirty-nine percent were self—interested. Twelve of the seventeen subjects in the present study acted in a role symmetric manner on at least two of the three mirror-image pairs of items on the Wage Distribution Booklet (p < . 14,- sign test). These results are more consistent with the equity hypothesis than with the expectancy hypothesis. Condition 3 Seventy—five percent of the responses made on the Wage Distribution Booklet by subjects in Condition 3 were self-interest responses, while the remaining twenty-five percent were role sym- metry responses. In addition, eight of the twelve subjects acted in a self-interested manner on at least two of the three mirror-image pairs of items on the Wage Distribution Booklet (p < .28; sign test). These results are more consistent with the expectancy hypothesis than with the equity hypothesis. C onditi on 4 Sixty-two percent of the responses made on the Wage Dis- tribution Booklet by subjects in Condition 4 were role symmetric, 91 while the remaining thirty-eight percent were self-interested. In addition, ten of the fifteen subjects tended to act in a role symmetric manner on at least two of the three mirror-image pairs of items on the Wage Distribution Booklet (p < . 30; sign test). These results are not consistent with either the equity or the expectancy hypotheses. Tests of the Between Group Hypotheses Role Symmetry As noted above, Table 7 indicates the percentage of role symmetric and self—interested responses occurring in the four ex- perimental conditions. To determine if the frequency of personal dispositions differed as a function of experimental conditions, sub- jects in each condition were classified according to whether the majority of their responses were role symmetric or self-interested (see Table 8). A chi square test performed on these data indicated a significant difference in the majority response as a function of experimental condition (X2 = 12. 96, p < . 005). To explore this finding further the mean role symmetry score for a given condition was compared with the mean score for each other condition by means of individual t_ tests. 1 The results of the t_ tests are summarized in 1Individual t_ tests were performed instead of one of the multiple comparison methods because of the unequal number of sub- jects in the four experimental conditions. 92 Table 9, below. A positive t_ value in Table 9 indicates that the equity hypothesis correctly predicted the experimental condition in which role symmetric responses would be more frequently mani- fested, while a negative t_ value indicates that the equity hypothesis generated an incorrect prediction. Table 9. --_t_ values for comparison of percentages of role-symmetry responses occurring between experimental conditions Condition 1 2 3 4 Condition 1 3. 52>:=>:< _ 35 3. 17*);< Condition 2 ~2. 32>:< . 13 Condition 3 2. 35* Condition 4 * p < . 05 ** p < . 01 On five of the six comparisons, equity theory correctly pre— dicted in which condition role symmetry would be more frequently manifested. Three of the differences were statistically significant, two at the . 01 level and one at the . 05 level. 93 " Altruism As noted earlier, no altruism responses occurred in any of the four conditions. The Effect of Relative Welfare As noted above, relative welfare was the one reward struc- ture variable systematically examined. As previously stated, the relative welfare of the reward structure is the difference between the payment for the individual distributing the wages and the payment of his partner; that is, does the individual distributing the wage receive more than, the same as, or less than his partner? An attempt was made to Guttman scale the data in respect to relative welfare. Results of this attempt (presented in Appendix V) indicate that the data order on this dimension quite well. All but one of the fifty—eight subjects'i responses scaled perfectly. These results are discussed in the following chapter, especially in respect to the equity and expectancy hypotheses. IV. DISCUSSION [The present chapter is divided into three sections. The first section focuses on the variables of inputs of the subject and his partner, especially in respect to the equity and expectancy hy- potheses. The last two sections present an examination of the variables less central to the focus of the study: (a) relative‘welfare and (b) order of presentation and real versus hypothetical choices. Inputs of Subject and Inputs of Partner As noted above, the present research was mainly concerned with the effects of the variables of inputs of the subject and his part- ner. The equity hypothesis postulates that these variables are major determinants of the way in which wages are distributed. Because of this, these variables were the focus of the present study. Inputs of Subject The equity hypothesis states that persons should attend to the inputs of all relevant people in a work situation, including them- selves. This is so because the fair distribution of wages is a function 94 95 of every individual' 3 input-outcome (ratio. On the other hand, the expectancy hypothesis states that one' 3 inputs should be irrelevant since each person should attempt to maximize his own wages regard- less of differences in inputs. The (results of the present research, for the most part, support the expectancy hypothesis. In the analysis of variance, only one group of subjects was influenced by the manipulation of their own inputs, and these made hypothetical choices. Moreover, this one difference, which was attributable to the effects of inputs of the sub- ject, was contrary to the predictions of the equity hypothesis. In this instance, subjects with high inputs were more role symmetric than subjects with low inputs, a finding which neither theoretical position predicts nor can readily explain. This lack of support for the equity hypothesis is further emphasized by the additional analyses performed to examine in detail the effects of inputs of the subject and his partner. Individual com- parisons of the experimental conditions through t_ tests (Table 9, page 92) indicates that there was no difference between those con- ditions where inputs of the partner were the same but the inputs of the subject were different. Comparison of Conditions 1 and 3 yielded a t_ Value of only . 35. Likewise, a comparison of Conditions 2 and 4 yielded a t_ value of . 13. 96 These findings, taken together, imply that the complexity of the equity hypothesis is not justified, at least for the range of input conditions examined in the present research. Inputs of the subject just did not matter. Contrary to the equity hypothesis, subjects, in choosing among distributions of wages, did not take into account their own inputs. However, they did, as noted below, take into account the inputs of their partner. A partial explanation of why inputs of the subject failed to have the effect predicted by the equity hypothesis lies in the ability of the subjects to distort reality. Adams (1965), as noted in Chap- ter 1, indicated that distortion of reality is one mechanism by which persons can alleviate feelings of dissatisfaction produced by per- ceptions of inequity. This mechanism seemed to be operating in Condition 4, where subjects worked % as long as their partners. These subjects' answers to questions 2 and 4 of the Attitude Question- naire (seen in Tables 5 and 6) indicated that at least 40 percent of them felt that they did at least as much work of the same quality as did their partners. Since the amount of role symmetry displayed in this condition was about the same as that displayed in Condition 2-— where both the subject and his partner had high inputs--subjects in Condition 4 seemed to have overestimated their inputs. Comparison of the role symmetry scores for subjects in Condition 4, who distorted their inputs, with those who were realistic 97 provides additional support for the contention that such distortion influences choice behavior. Those subjects who reported an accurate perception of their inputs had a, mean role symmetry score of 2. 1 (out of a possible mean of 3.0). On the other hand, subjects who perceived their inputs as higher than they actually were had a mean role symmetry score of 1. 4. Moreover, the mean role symmetry score for subjects in Condition 2 was 1. 8. Although none of these differences is significant--probably due to the low number of sub- jects in each comparison--these results tend to support the equity hypothesis. Such a distortion, however, at most can be only a par- tial explanation of why the variable of inputs of the subject did not have the effect predicted by the equity hypothesis. First, the equity hypothesis also did not correctly predict the behavior of subjects in Condition 3, where both the subject and his partner had low inputs. In this condition the perception of inputs was not distorted. Second, as discussed in detail below, the subjects did attend to the inputs of their partner; that is, they tended to be more role symmetric when their partners had high inputs. Thus, it is difficult to explain why, if subjects distorted at all, they only distorted the perception of their own inputs. In summary, the results of the present research-—including the findings which indicate that some subjects distorted the perception 98 of their inputs—-imply that the variable of inputs of the subject has less of an effect on choice behavior than predicted by the equity hypothesis. On thewhole, these results supported the expectancy hypothesis. Inputs of Partner As noted a number of times, the equity hypothesis postulates that subjects should attend to the inputs of their partners when dis- tributing wages. On the other hand, the expectancy hypothesis states that this variable should be irrelevant. The findings of the present research provide strong support for the equity hypothesis. The results of the analysis of variance (Table 3, page 79) indicate that of the seven comparisons which involve inputs of part— ner, four were statistically significant. Moreover, only one addi— tional comparison (the main effect for order of presentation) was also significant. These results imply that inputs of partnerwas the crucial variable examined in the present research. Further, these findings, when examined in greater detail through analyses of the simple effects, indicate that all of the four significant differences attributable to inputs of partner were in the direction predicted by the equity hypothesis. That is, subjects tended to be more role symmetric when their partner' 3 inputs were high than when they 99 were low. In addition, differences in three of the fournon-significant simple effects were also in the direction predicted by the equity hypothesis. Further support for the equity hypothesis is provided by the additional analyses performed to examine in detail the effects of inputs of the subject and his partner. Examination of Tables 7 and 8 reveals that when inputs of the subject are the same, more role symmetry is displayed when the inputs of the partner are high than when they are low. That is, there is significantly more role sym— metry in Condition 2 than in Condition 1, and in Condition 4 than in Condition 3. Moreover, when just inputs of partner are considered, these results, presented in Table 8, also support the equity hypothe- sis. In Conditions 1 and 3, where inputs of the partner are low, the number of subjects whose major response is self-interest exceeds those whose majority response is role symmetry. (For Conditions 1 and 3 combined, X2 = 7. 54; d. f. = 1; p < .01). In Conditions 2 and 4, where inputs of the partner are high, the number of subjects whose major response is role symmetry exceeds those whose majority, response is self—interest. (For Conditions 2 and 4 com- bined, X2 = 4.50; d.f. = 1; p< .05). In summary, the present research provides strong support for the contention that persons both attend to the inputs of others and 100 weigh differences in these inputs when choosing among distribution of rewards. This is precisely the prediction that the equity hypothe- sis makes. On the other hand, the findings in respect to inputs of partner are contrary to the predictions generated by the expectancy hypothesis. Taken together, the results for inputs of subjects and inputs of partner provide insight into those conditions which generate con- cerns with either fairness (the equity hypothesis) or profit maximi- zation (the expectancy hypothesis). There. is strong evidence that a person, in distributing wages to himself and another-individual, only considers the other' 3 inputs. Subjects tend to give themselves as much money as possible unless they perceive that their partners have worked hard enough to deserve more than minimal pay. It is as if the inputs of the other must exceed a threshold value before a subject feels that his partner deserves more than this minimum. If this threshold value is exceeded, then equity motives tend to dominate. On the other hand, if the inputs of the other are less than this value, subjects tend to maximize their profits. Thus, these findings have significant implications for both the equity and expectancy hypotheses. Relative Welfare The results of the attempt to Guttman scale responses as a function of relative welfare indicate that it was easiest for subjects 101 to give some of their potential wages to their partners when the resultant distribution still gave them more money. Moreover, sub— jects found it easier to give uprsome money-when the resultant dis- tribution gave both persons the same than when it gave their partner more than themselves. These findings are consistent with those of Messé (1968). These results imply that most subjects have a competitive orientation which affected their concerns with equity and profit maximization. That is, in making their choices, they seemed to take. into account the difference in rewards to themselves and the other that each distribution of wages provided. This implies that subjects were motivated to stay ahead or at least not let the other receive more than they. Thus, there is at least one, and probably more motivating factors, in addition to concerns with equity and profit maximization, which affect the distribution of rewards. Therefore, those theorists who exclude these other potential factors from consideration may be presenting a somewhat simplistic model of what happens in those situations where rewards are distributed. Effect of the Control Variables As noted above, two of the situational variables, order of presentation and real versus hypothetical choices, were manipulated 102 primarily for reasons of control. However, these variables were found to affect choice behavior. Although these results are beyond the focus of the present research, they are discussed below since they havesome interesting implications. Order of Presentation The results of the analysis of variance (Table 3, page 79) indicate that of the seven comparisons which involve order of pre- sentation, four-were statistically significant. Moreover, only one additional comparison (the main effect for inputs of partner) was also significant. These results imply that although order of presen- tation was a control variable, it still had a major-impact on choice behavior. Results of the analyses of the simple effects indicate that, for those differences which were significant, subjects were more role symmetricwhen they began with Series Y choices. These findings are congruent with those of Messé (1968). A possible explanation for these results is related to concepts of response set and equity. For the subjects in the conditions where significant simple effects occurred, equity is equivalent to role symmetry, since most of these subjects perceived that they had approximately the same inputs as their partners. When Series X was presented 103 first to these subjects, choosing the alternative outcome was con- sistent with equity and role symmetry but not with self—interest. Consistent with this set, most subjects then chose the alternative outcome in Series Y, the choice consistentwith role symmetry and equity. However, when Series Y was presented first, selecting the alternative outcomewas consistent not only'with equity and role symmetry but also withself-interest. Therefore, in some subjects the set of self-interest may have been aroused and continued into Series X choices. As a result, presenting Series Y first more likely aroused the set of self-interest. Real Versus Hypothetical Choices The control variable real versus hypothetical choices had less of an effect on responses than did order of presentation. How- ever, one simple effect was significant. For pairs where both members had high inputs, subjects making hypothetical choices on Wage Distribution Booklets "which began with Series Y were more role symmetric than subjectsmaking real choices. These results replicated the findings of Messé (1968). It is interestingto note that the situation in the present research where real versus hypothetical choices made a difference--where both subject and partner had high inputs--was the one which most closely approximated the conditions of his study. 104 Since hypothetical choices did not have an effect on the actual distribution of wages, any difference due .to this variable should have been in the direction found; that is, subjects making hypothetical choices should have beenmore role symmetric than subjects making real choices. However, in the present research this occurred only once. An explanation for this result is not readily apparent. However, it is possible that this variable affected choice behavior in the conditions it did because of the potential difference with which subjects making real or hypothetical choices attended to order of presentation. Under conditions of high inputs the equity hypothesis predictsthat role symmetry should be dominant. It is only-under these conditions that order of presentation had an effect. It could be that the set toward self-interest induced by exposure to Series Y first only occurred when subjects made real choices, since these subjects might have paid closer attention to such subtle differ- ences in the Wage Distribution Booklet. This explanation is highly speculative, especially since no such difference in role symmetry occurred as a function of real versus hypothetical choices when sub- jects had low inputs and started with Series Y but their partners had high inputs. Taken together, the results for the control variables indi- cate that othersituational factors besides inputs influenced wage 105 distribution behavior. , It seems reasonable that variables—~both motivational, such as the competitive orientation, and situational, such as the sequential order of relevant experiences-~otherthan those specifically-related to the equity or expectancy hypotheses also must be considered if a comprehensive-theory concerning wage dis- tribution is to be developed. V. CONCLUSION The concluding chapter contains four brief sections. They are (a) implications for wage distribution theory, (b) implications of the present methodology, (c) suggestions for further research, and (d) a concluding statement. Implications of the Results for Wage Distribution Theory The stated purpose of the present research was to delineate those circumstances in which equity motives and profit maximization motives tend to dominate. The results indicate that inputs of the subject' 3 co-worker is a major determinant of behavior in wage dis- tribution situations. When the inputs of the subject' 3 co-worker were high, equity motives tended to dominate; when inputs of the subject' 3 co-worker were low, profit maximization motives tended to dominate. It is as if a threshold value on the variable of inputs of the subject' 8 co-worker must be exceeded before a subject feels that his co-worker deserves more than minimum pay. Once the sub- ject feels that this threshold value is exceeded, equity motives come 106 107 into focus and then exert a significant effect on behavior. If the threshold value is not exceeded, profit maximization motives tend to dominate. However, despite the fact that equity motives tend to dominate in certain situations while profit maximization motives tend to dominate in others, in all four situations studied in the present research, a substantial quantity of responses consistent with the less dominant motive in that situation occurred. Therefore, it appears that neither the equity hypothesis nor the expectancy hypothe- sis alone can explain all the results obtained pertaining to wage dis- tribution. Moreover, results indicate that other factors, both motivational and situational, also affect choice behavior. Therefore, the results of the present research, taken as a whole, suggest that further theoretical work in the area of wage distribution should focus on the multi-dimensional nature of man' s motivations and the situa- tional cues which interact with them. Implications of the Present Methodology One finding in the present research questions the validity of the procedure used by Adams and Lawler. The results in Condition 4 (subject determining the wages has low inputs and his partner has high inputs) of the present research indicated that a 108 substantial number of subjects distorted their inputs in relation to those of their partner. In the procedure used by Adams and Lawler, subjects should find it very difficult to distort inputs. The compari- son individual is a hypothetical person described to the subject by the experimenter. The subject is told that this hypothetical individ- ual is more qualified than he is for a job the subject is hired to per- form. However, the subject never sees him. Therefore, it should be more difficult to distort the inputs of the other-individual in Adams' and Lawler' s procedure than in the present research or in ”real life" situations. Thus, there is evidence that the procedure used in the present study is a productive tool that could be used effectively in further research. An additional methodological point concerns the use of the post-session Attitude Questionnaire. Most research in experimental social psychology uses some sort of post-session measurement designed to assess the degree to which manipulations were success- ful. The studies of Adams and Lawler employed such a procedure. The present study extended this technique somewhat. For example, the subjects in Condition 3 accurately reported on the Attitude Questionnaire that they and their partners had the same inputs. However, they also reported that these inputs, since they both were low, had very little effect on their choices. These 109 responses seemed valid since subjects in Condition 3 tended to be self-interested. While the responses indicating the effectiveness of the manipulations were not formally presented in this respect, they did, however, aid in the understanding of choice behavior. There- fore, such questions might be productive additionsto post-session measurement devices, especiallysince they aid in formulating theoretical explanations for the observed behavior. Suggestions for Further Research The present study suggested a number of issues which could be investigated through further research; three possibilities are dis- cussed below. The present research indicated that when the subject' 3 partner worked 1% hours, even though equity motives tended to dominate, a substantial number of profit maximization responses occurred. Research could be conducted to see if extending the length of time that both the subject and his partner worked would lower the number of self-interested responses. Such a test would require two conditions. Condition 1. Both the subject and his partner would work 10 hours. Condition 2. Both the subject and his partner would work 1% hours. 110 It is hypothesized that subjects in Condition 1 would be significantly more role symmetric. This would demonstrate the strength of the equity motive under the high input conditions typically found in industry. Subjects in the present research displayed no altruism. Therefore, a third condition might be introduced into this proposed study in an attempt to induce subjects to make at least some altruistic responses. In this condition, the subject would work 1%- hours and his partner 10 hours. If such responses did occur, as predicted by the equity hypothesis, they would seem to indicate that the equity motive is relevant to those situations, such as found in industry, where inputs are high. Another possibility for future research would be to have subjects perform a task requiring different degrees of co-operation between them. This would experimentally manipulate group cohe- siveness and simulate the work group atmostphere frequently found in industry. It is hypothesized that the equity motive would be mani- fested more frequently as cohesiveness increased. A third possibility for future research would be to present the Wage Distribution Booklet, modified appropriately, to workers in a number of industrial settings. Since the present study found very little difference between real and hypothetical choices, the 111 proposed research could present the alternative distributions in the questionnaire form typically used in industrial field research. Using an instrument, such as the Wage Distribution Booklet, in both field and laboratory settings would provide a degree of comparability between findings which has been lacking to date. A Concluding Statement The present research and those studies previously reported provide evidence that persons have two motivations which are to some degree incompatible. Clearly, man is to some extent moti- vated by self-interest and this is the basic premise of the expectancy hypothesis. On the other hand, under certain circumstances, as the equity hypothesis predicts, man is also concerned with the-welfare of others. It would be difficult for man to function as a social animal without the presence of norms such as equity. In the absence of such norms, it is doubtful that men could live and work together in a mutually productive and satisfying relationship. REFERENCES REFERENCES Adams, J. S. Toward an understanding of inequity. Journal of Abnormal and Social Psychology, 1963a, fl, 422-436. Adams, J. S. Wage inequities, productivity and work quality. Industrial Relations, 1963b, 3, 9-16. Adams, J. S. Inequity in social exchange. In L. Berkowitz (Ed.) Advances in experimental social psycholggy. Vol. 2. New York: Academic Press, 1965. Adams, J. S. Effects of overpayment: two comments on Lawler' 3 paper. Journal of Personality and Social Psychology, 1968, 12, 315-316. Adams, J. S. , and Jacobsen, P. R. Effects of wage inequities on work quality. Journal of Abnormal and Social Psychology, 1964, 29, 19-25. Adams, J. S., and Rosenbaum, W. B. The relationship of worker productivity to cognitive dissonance about wage inequities. Journal of Applied Psychology, 1962, Q, 161-164. Andrews, 1. R. Wage inequity and job performance: an experi- mental study. Journal of Applied Psychology, 1967, _5__1_, 39-45. Arrowood, A. J. Some effects on productivity of justified and unjustified levels of reward under public and private con- ditions. Unpublished doctoral dissertation, University of Minnesota, 1961. Clark, J. V. A preliminary investigation of some unconscious assumptions affecting labor efficiency in eight super- markets. Unpublished doctoral dissertation, Harvard University, 1958. 112 113 Festinger, L. Atheory of cogitive dissonance. Evanston: Row, Peterson, 1957. Friedman, A. , and Goodman, P. Wage inequity, self-qualifications, and productivity. Organizational Behavior and Human Per- formance, 1967, 3, 406-417. Goodman, P. , and Friedman, A. An examination of the effects of wage inequity in the hourly condition. Organizational Behavior and Human Performance, 1968, 3, 340-352. Gordon, B. F. , and Lowin, A. Qualifications, co-worker character- istics and productivity--An extension and critique of Adams' inequity theory. Technical Report, December 15, 1965, Bell Telephone Laboratories. Homans, G. C. Status among clerical workers. Human Organiza- tion, 1953, _13, 5-10. Homans, G. C. Social behavior: its elementary forms. New York: Harcourt, Brace and World, 1961. Jacques, E. Equitable payment. New York: Wiley, 1961. Jacques, E. Objective measure for pay differentials. Harvard Business Review, 1962, 4_0_, 133-138. Lawler, E. E. Effects of hourly overpayment on productivity and work quality. Journal of Personality and Social Psychology, 1968a, 12, 306-314. Lawler, E. E. Equity theory as a predictor of productivity and work quality. Psychological Bulletin, 1968b, 19, 596-610. Lawler, E. E. , et al. Inequity reduction over time in an induced overpayment situation. Organizational Behavior and Human Performance, 1968, _3_, 253-268. Lawler, E. E. , and O'Gara, P. W. Effects of inequity produced by underpayment on work output, work quality, and attitudes toward the work. Journal of Applied Psycholog, 1967, _5_1_, 403-410. 114 Merton, R. K. , and Kitt, A. S. Contributionsto the theory of reference group behavior. In Continuities in social research. R. K. Merton and P. F. Lazarsfeld (Eds.) Glencoe, Illinois: Free Press, 1950. Messé, L. A. Parameters of fair play: an investigation of differ- ences in the just distribution of rewards. Unpublished doctoral dissertation. University of Chicago, 1968. Patchen, M. The choice of wge comparisons. Englewood Cliffs, New Jersey: Prentice-Hall, 1961. _Porter, L. W. , and Lawler, E. E. Managerial attitudes andjer- formance. Homewood, Illinois: Irwin, 1968. Sawyer, J. The altruism scale: a measure of cooperative, individualistic, and competitive interpersonal orientation. The American Journal of Sociology, 1966, LXXI, 409-420. Sayles, L. R. Behavior'of industrial work groups: prediction and control. New York: Wiley, 1958. Stouffer, et al. The american soldier: adjustment during army life. Vol. 1. Princeton, New Jersey: Princeton Univer- sity Press, 1949. Vroom, V. H. Work and motivation. New York: Wiley, 1964. Weick, K. E. The concept of equity‘in the perception of pay. Administrative Science Quarterly, 1966, _13, 414-439. Weick, K. E. , and Nesset, B. Preferences among forms of equity. Organizational Behavior and Human Performance, 1968, 3, 400—416. APPENDICES APPENDIX I WAGE DISTRIBUTION BOOKLET 115 WAGE DISTRIBUTION BOOKLET 116 Alternative A Alternative B You get $4.00 You get $3.00 and and He gets $1.00 He gets $2.00 1 1. Which alternative gives you more? A or B? (Check one) 2. Which alternative, if either, gives the greater total to the two of you, together? Check A or B. If the two give the same total, check ”Equal." Equal 3. Which alternative gives the two of you more nearly the same? A or B? (Check one) SKIP THE QUESTION BELOW; GO ON TO THE NEXT PAGE CHOICE Xl: Which alternative set of payments would you rather see used? A or B? (Check one) Remember: If this choice is the one drawn, you and he will be paid just the amounts you now choose. 117 'Alternative A‘- You get $4.00 YOu get $2.50 and and He gets $1.00 He gets $2.50 I Alternative B 1. Which alternative gives you more? A or B? (Check one) 2. Which alternative, if either, gives the greater total to the two of you, together? Check A or B. If the two give the same total, check "Equal." Equal 3. Which alternative gives the two of you more nearly the same? A or B? (Check one) SKIP THE QUESTION BELOW; GO ON TO THE NEXT PAGE CHOICE X2: Which alternative set of payments would you rather see used? A or B? (Check one) Remember: If this choice is the one drawn, you and he will be paid just the amounts you now choose. (X2) 118 Alternative A Alternative B You get $4.00 You get $2.00 and and He gets $1.00 He gets $3.00 Which alternative gives you more? A or B? (Check one) Which alternative, if either, gives the greater total to the two of you, together? Check A or B. If the two give the same total, check "Equal.” Equal Which alternative gives the two of you more nearly the same? A or B? (Check one) SKIP THE QUESTION BELOW: GO ON TO THE NEXT PAGE CHOICE X3: Which alternative set of payments would you rather see used? A or B? Remember: If this choice is the one drawn, you and he will be paid (Check one) just the amounts you now choose. (X3) 119 Alternative A Alternative B You get $1.00 You get $3.00 and and He gets $4¥00 He gets $2.00 1. Which alternative gives you more? A or B? (Check one) 2. Which alternative, if either, gives the greater total to the two of you, together? Check A or B. If the two give the same total, check ”Equal." Equal 3. Which alternative gives the two of you more nearly the same? A or B? (Check one) SKIP THE QUESTION BELOW: GO ON TO THE NEXT PAGE CHOICE Yl: Which alternative set of payments would you rather see used? A or B? (Check one) Remember: If this choice is the one drawn, you and he will be paid just the amounts you now choose. (Y1) 120 Alternative A Alternative B You get $1.00 You get $2.50 and and He gets $4.00 He gets $2.50 1. Which alternative gives you more? A or B? (Check one) 2. Which alternative, if either, gives the greater total to the two of you, together? Check A or B. If the two give the same total, check ”Equal.” Equal 3. Which alternative gives the two of you more nearly the same? A or B? (Check one) SKIP THE QUESTION BELOW; GO ON TO THE NEXT PAGE CHOICE Y2: Which alternative set of payments would you rather see used? A or B? (Check one) Remember: If this choice is the one drawn, you and he will be paid just the amounts you now choose. (Y2) 121 Alternative A Alternative B You get $1.00 You get $2.00 and and He gets $4.00 He gets $3.00 . l 1. Which alternative gives you more? A or B? (Check one) 2. Which alternative, if either, gives the greater total to the two of you, together? Check A or B. If the two give the same total, check ”Equal.” Equal 3. Which alternative gives the two of you more nearly the same? A or B? (Check one) SKIP THE QUESTION BELOW; GO ON TO THE NEXT PAGE CHOICE Y3: Which alternative set of payments would you rather see used? A or B? (Check one) Remember: If this choice is the one drawn, you and he will be paid just the amounts you now choose. (Y3) APPENDIX II A TTI TUDE QUE STIONNAIRE 122 ATTITUDE QUESTIONNAIRE ,, 123 Please check the best answer for the following questions. 1. I found this study to be very somewhat neither somewhat very interesting interesting interesting uninteresting uninteresting nor unin- teresting 2. In comparison to the people in the other group how much work do you feel you did? 4 times as about 2 to about the about 1/3 to 1/4 as much much or more 3 times as same 1/2 as much or less much 3. How important was the difference in the amount of work between yourself and the people in the other group in influencing your choices on the wage distribution booklet? It was £§g_major It was one of It was a factor It was a factor factor influenc- several impor- but not as im- of little or no ing my choices tant factors portant as other importance in— influencing my ones in influ- fluencing my choices. encing my choices choices 4. In comparison to the peOple in the other grogp I feel the quality of my answers were considerably somewhat about somewhat considerably better better the same worse worse 124 5. How important was the difference in quality of the work between yourself and the people in the other group in influencing your choices on the wage distribution booklet? It was £23 major It was one of It was a factor It was a fac- factor influenc— several impor— but not as im- tor of little ing my choices tant factors portant as other or no impor— influencing my ones in influ— tance influ- choices encing my choices encing my choices 6. What other factors beside quantity and quality of the work af— fected your choices on the wage distribution booklet? Explain your answer. 7. Please write below any comments you may have concerning the study. APPENDIX III INDUSTRIAL OPINION QUESTIONNAIRE 125 INDUSTRIAL OPINION QUESTIONNAIRE Research Form 2 126 Name Age? Instructions: Please spend about 8 minutes on each question. Try not to spend too much time on any particular question. For your convenience, the time remaining will be written periodi- cally on the board in the front of the room. The questionnaire will be collected in l 35 hours. Please try to answer all 12 questions. 127 What do you feel will be your job when you finish your education? What were the major factors influencing your career choice? What college courses do you feel were the most beneficial to you? 128 How much confidence do you place in psychological tests as a device for selecting employees for a job? How would you feel about having to take a psychological test before you were hired? 129 If you were asked to take a psychological test for a job you wanted how likely is it that you would fake your answers to try to make a very favorable impression? What personality traits do you feel are generally desired in prospective employees? Why do you feel these qualities are desired in prospective employees? To what extent do you feel you possess these qualities? 130 While working on your job you are reprimanded by your super— visor for making an error in your work. The error was really your associate's. a. What would you do in such a situation? b. Why would you do this? c. What factors would influence your decision? d. wa do you feel the average workingman would react to such a situation? 131 While working for a large company you make a suggestion to your supervisor which he reports to a vice-president as his suggestion. The suggestion proves to be a valuable one. It saves the company a considerable amount of money. a. What would you do in such a situation? b. Why would you do this? c. What factors would influence your decision? d. How do you feel the average workingman would react to such a situation? 132 Yen are praised by your supervisor‘s boss for an idea that was your associate's. a. What would you do in such a situation? b. Why would you do this? c. What factors would influence your decision? d. How do you feel the average workingman would react to such a situation? 133 What factors do you feel are the most important ones for employees being satisfied with their jobs? Which factor of the ones mentioned in part a. do you feel is the most important one? Why did you select this one factor? Explain fully. 134 YOur company introduces a new plan. Whenever any suggestion by a worker saves the company money Eli the employees share in the benefits. a. Do you like this plan better than one in which only the person who thought of the suggestion gets the bonus? b. Explain the reasons for your answer in part a. c. Which plan do you feel would be more successful? 135 In deciding which of two companies to work for what three factors would be the most important ones influencing your decision? Why did you select these three? Explain fully. 10. 136 Some companies train their promising young executives by having them work two years on a job and then transferring them to another job. These job changes often involve moving to another part of the country; a. Would you accept such a job? b. Why? c. What were the most important factors influencing your decision? 11. 137 Some companies have unions that are not affiliated with the larger national unions. a. Do you feel these local unions can be as effective in representing their members as the ones who are affil- iated with national unions? b. What did you base your answer to part a. on? ———_______ 7” 12. 138 It is widely believed that most employees now place little pride in their work. a. Do you feel this is true? b. If you feel this is true what do you feel is responsible for this occurring? If you feel this is not true, why do you believe so many people feel that most employees place little pride in their work? APPENDIX IV STUDENT BACKGROUND QUESTIONNAIRE 139 Psychology 255 Instructor: Mr. Lane Student Background Questionnaire 1. Name 2. Sex 3. Age 4. Major 5. Grade Point Average 6. Previous Job Experience Type of Job How Long Employed a. b. c. d. e. f. 7. Please answer the following question. Spend about 10-15 minutes on your answer. Do you feel labor unions have become too powerful at the present time? Explain your opinion on this issue. Icheb 12-27-68 APPENDIX V GUTTMAN SCALED ROLE SYMMETRY RESPONSES 140 Personal Disposition Manifested When Series X Alternative Outcome Gave Subject Less Wages, Equal Wages, and More Wages Than His Partner Subj e ct Numbe r Series X Alternative Outcome Gave Subject Less Wages Equal Wages More Wages 141 Series X Alternative Outcome Gave Subject Subject Number Less Wages Equal Wages More Wages 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 NNNMNNNNNNN m NNNNNN 35 142 Subject Number Series X Alternative Outcome Gave Subject Less Wages Equal Wages More Wages 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 NNNNNNX‘NNNNN NMNNNN NNNNNNNNNNN NNNNNNNNNNNNNNNNNN 143 Series X Alternative Outcome Gave Subject Subject Number Less Wages Equal Wages More Wages SS4 X X X 855 X X X S56 X X X S57 X X X S58 X X X 3A blank box represents a self-interested response while an X represents a role symmetric response. S 42 responses were the only ones that did not Guttman scale. _!__ .< L ”71111131112! TATE UNIV 93 030 fllflillfllgfiflml“