A THEORY OF, SELFESTEEM .IN A TWO-PERSON TASK SITUATEON Thesis for the Degree of M. A. MICHIGAN STATE UNIVERSITY JAMES waLLIAM BALKWELL 1968 1 . Livyl .l J“ 3 .g, Etc :9 rr . .'...._. . 1197-?" 4w $4 \ ' MW THESIS ’J:T‘; I IE 9 A n V l 59‘! ABBTRACT A IHEORI OF SELF-ESTEEM IN A TWO-PERSON IABK SITUATION by James William Balkwell Ihis:paper:presents a formal theory of self-esteem which we believe to be applicable to all two-person task situations which require.decisions on.the part of group members. It is addressed to social.psychologists.interested in decision- making and social influence; it will perhaps be.pertinent to the.interests.of others as well. The.most,clear+cut applications of our theory are to two»step decision situations which have the following char- acteristics: (1)" Each person makes an initial task-relevant decision, there being two.possib1e responses. (2) Each.perscn then.rensives information that the other person.has chosen the opposite response. (3) After receiving this information, each person makes a giggg decisimn. We.began by examining a study of this type of situation by Santa-F. Gamilleri and Joseph Berger-.1 may have developed a model for.predicting the probability of a “self-response" for a standardized set of experimental conditions which have the three.characteristics that we've:statede a{u1 + u3) + us P(8) = where ul+u3+u5 James William Balkwell 111 = ma value to Actor of the experimenter's apprnyal. The. value. to Actor of his partner's approval. the: value. to Actor. of self‘-c:onsistency,. e.i., self;- esteem. 1.13 = u. = E = Actor's estimate of the» probability that. he: is- correct given. that he; and his; partner disagree on their initial choices. through a detailed analysis of the: empirical findings of Camilleri and Berger, we arrived at some. ideas. which may be expressed in propositional form as. follows: Assumption 1: Assumption 2: Pagosi'ticn. 1: Proposition 2: Prgposi‘tion 3:: Proposition 4:: me: effects. involving individual difference»- intelligence, stew-remain fixed at. a given level. The; effects involving socio-cultural factors. remain; fixed. at a given level. The value to Actor of self-esteem varies directly with the. degree of equality of abil- ity (other situational factors: held constant?» the; value: tor-Actor of self-eaten varies directly with the degree of equality of control (”others situational factors held constant). the value tor. Actor of. self-esteem. varies inversely with the; amount: of control which Actor has over. group decisions (iother situ- ational factors. held constant). Ihe. effects-which determine the: value to- Actor of self-esteem are additive. the, scepee of our- theory encompasses any the. very important-:- but much neglectedncatagory. of. factors, situational factors. Algebraically, our thecry' may be stated thus: “'5 = tb(E'(A))‘+ bosom). + comm) + 81% where; to, ho, co, and the d1 are apprcpriate constants, and James William Balkwe 11 E.(‘.) = The equality function Of ability. _ l ' $53. " no 100 (lote—-the symbol 7&3 means: "percentage ability of self“, and and. the symbol filo means ”percentage ability of other”.5 3(0); = lh‘es. equality function: of control (fe.i. power). _ aw. - zoo - - loo 0(a) = The standardized quantity function of control. _ so. " T65 1‘1 = Various: functions of individual differences and analog-cultural conditions which take. on constant values in a well-planned set. of laboratory experiments. films, the sum of the 6.111 is equal to some constant, do. through various estimation proceduresumainly through. solving sets of simultaneous equationsp-we developed; the following elaboration and extension of the Oamilleri-Borger model for Oami-lleri-Berger- type experimental siwationsz: C(ul + 113) '0' 115 P(8 )' = where- u1+u3+u5 .85 in all cases- 2(Q(0-))1 1.552(E(1)) + .343(E(G)). - .28‘0(Q(G)) - .884 ULJ. “3 . 115 James William Balkwell .75 for all “high self, low other" expect. states. g = .50 for all “high self, high other" expect. states. .50 for all “low-self, low. other" expect. states. .25 for all “low self, high other” expect. states. An independent study was. done which indicated. that. the. model is capable; of giving us accurate predictions. We believe that the success. of our model. lends.» credibilijy to- our theory, as well as. to; a whole host of theoretical ideas which form the basis of. the Camilleri-Berger model. In science, preemption is. no: virtue. Enough theory is at stake so.- that a replication of our study,. using larger group sizes, would be desirable. lsmto. F. Oamilleri and Joseph Berger, “Decision-Making and; Social Influence: AZM'odel‘ and an Experimental lest“, Socialism, December 1967}, pp. 365-378 A THEORI OF SELF—ESTEEM IN A TWO-PERSON TASK SITUATION By. James William Balkwell A'THESIB Submitted‘to Michigan State:University' in partial fulfillment of the requirements for the degree of MASTEEEOFUARTB Department of Sociology 1968 ACKNOWLEDGMENTS I would like to thank some of those who have markedly influenced my thought, especially as expressed in the present essay. Dr. Santo F. Camilleri has, over the past two-and; one-half years, greatly affected my approach to social psy- chology. He.read the present paper, and his:comments were most helpful. A'debt of gratitude is also owed to Dr. Hans Lee, who read this essay and made valuable suggestions. My greatest debt is to Dr. Thomas L. Conner, my committee chairman, for his interest and encouragement, for his count- less suggestions and criticisms, and for his guidance throughout the thesis process. the research.which I report in my subsection, "An Inde- pendent Tsst”, was supported by NSF grant.lGS-1310 to the- Michigan State.Department of Sociology for the study of authority and evaluation structures (Santo F. Oamilleri, principle investigator). J. W. Balkwell March 27, 1968 11 II.. III. IV. V. VI L IN THODUO TION THE OAMILLERI-BERGER. RESEARCH THE “EILEEN-BERGER EXPERIMEN T3,: A". RE-ANALYSIS TABLE OF CONTENTS The: Basic Plan of Attack Some preliminary Matters A Workahle. Set: of Values. An. Analysis of 3 THE THEORY HE-STLTED An Independent: Test; A' Crucial Experiment A4- ODNOLUDIN G NOTE BIBLIOGRAPHY 111 A‘NOTE10N INTEGRITY, ESTEEM, AND VALUE TO. ACTOR. INVESTIGATING THE THEORY P383 24 27. 29 37. 4O table. I. II, III. LIST OF TABLES Predicted. and Observed Mean Proportions of S Responses. by Control a: Expectation Conditions A. Workama set Of. mlues o e e e e e e e e e E(c). = 11(Q(°)\) 0' O O O O O 0 O O O 0 O 0 Values of the. 03‘ and; Their. Components . . . . Pradict‘ed'é and Observed; Mean Proportions of. S Responses. by Control 8:: Expectation Conditions Predicted and Observed. Mean Proportions of S Responses. by Control & EXpectation Conditions iv page. 14 16 . 19 32 36 --- we are told of certain Polynesian chiefs, who, under the stress of good form, preferred to starve rather than carry their food to their months with their own hands; ... A'better illustration, or at least a more unmistakable one, is affbrded by a cer- tain king of France, who is said to have lost his life through an excess-of'moral stamina in the obser- vanes 01 good form. In the absense of the function- ary whose office it was to shift his master's seat, the king sat unoomplaining before the fire and suf- fered his royal person to be toasted beyond recovery. But in so doing, he saved his Most Christain Majesty from menial‘contamination." --Thorstein Veblen I. INTRODUCTION This paper presents a formal theory of self-esteem which we believe.to be applicable to all two—person task situations which require decisions on the part of group members. If we confine our attention to the most:usual cases, than the maintenance of Actor's self-esteem would'be, in Homans’ terms; a ”reward“ for making his preferred choices; and the less of self-esteem would be.a “cost" of being in- fluenced, by the other person, to forego his preferred cheices, and make the choices favored by the disagreeing other.1 In Festinger‘s terminology, self-esteem would'be a 'cegnition" which would steer.ictor in the direction of making his preferred chnices, and steer him.away from making the choices favored by the disagreeing other.2' Ihus; the contingencies which determine the veins to Actor of self- esteem.should be of theoretical interest to all sociar’psy- chologists who are concerned with the processes of decisions making and social influence in small group settings. It is 1George C. Humans, "Social Behavior as.Erchange”, Amerb ican Journal of Sociology, May 1958, pp. 597¥6073 and George Caspar Romans, Social Behavior: Its Elements Forms, New York: Harcourt, Brace, an ‘ , ‘9 ‘, pp. 9 - CO 2Leonl'estinger, i Ihen§¥ of Cnggétive:nissonance, Stan- ford:: Stanford Univers y' ess, , pp. 9- ; and Icon Restinger, Cbnflict Decision, and Dissonance, Stanford: Stanford UniversIty Press, I961; sections deaIIEg with the prehdecisional'state‘ 1 2 thesc social psychologists to whom this essay is addressed; it will perhaps be of interest to others as well. We shall begin by making a detailed.analysis of some recentlempirical‘findings reported by Camilleri and Berger.3 What, we.shall ask, would have to be true of the value to Actor of selfsesteem in order for the model employed by Camilleri-and Berger to fit.the data? This question will be our focus. of concern; we shall be 93;ng _f_a_ct_o explainers. However; our aim will be to develope some formal propositions which may be conclusively tested in the laboratory, and which, if proven correct, will have theoretical implications for the whole sub-area of social psychology which is concerned with decisionpmaking and social influence. Ihe ideas which we shall be.developing, if proven cor- rect, will force.many readers to alter their basic conceptions of self-esteem. It has long been observed that self-esteem can take on very high values for some persons at certain times and in certain places (see the Veblen.quotation at the beginning of this paper). We shall attempt to demonstrate that.self9esteem can also take on very low values, and, in fact, eyenznegative'walues. This idea may take some getting used to for many readers, since it is much more general than the common.sense notions which many of us accept without serious question-~although perhaps by default. 3Santo F. Camilleri and Joseph Berger, ”Decision-Making and Social Influence: A Model and an EXperimental Test", Sociometgz, December 1967, pp. 365-378 3 II. THE GAMILLERI-BERGER RESEARCH It would be desirable for the reader to become familiar with the Camilleri-Berger research before reading the pres- ent essay; we can summarize this research only very briefly at this time. The original inspiration for the Camilleri-Berger study had to do with some ideas relating the performance expectations of Actor to the relative frequency of Actor's acceptance of influence. Somewhat later, Camilleri and Berger became interested in a costpgain interpretation of the acceptance of influence phenomenon; and the model which they present in their article draws upon both sets of ideas. The experimental situation of the Camilleri-Berger research, which will be of continuing interest to us, is essentially as follows: A pair of subjects are asked to make binary choices. Specifically, they are asked to deter- mine whether a pattern projected on‘a screen by means of a 35 mm. slide is predominantly black in color or white. Actually, the pattern is fifty percent of each color--it is clearly an ambiguous perceptual stimulus. Either answer would seem plausible, and the subjects are led to believe that.a correct answer does exist. The decision-making process involves two steps. After the subjects have made their initial choices, they are fed information via an electronic apparatus that the other person has disagreed with them. Then, after a period of 4 five seconds, they are asked to make a second choice; and they may stick with their initial Judgments or change, as they see fit. However, they have agreed beforehand to try to attain the highest.tg§m score which they can--they are working together as a team--and they have been told that gal: their final decisions will be recorded. To complicate things somewhat, each person is led to believe that he has certain level of ability at correctly identifying the predominant color of this type of pattern; this is accomplished by a preliminary manipulation phase. Each person is "assigned", by a random process, high ability (+) or-low ability (-); thus, there are four possible abil- ity combinations. Also, there are three control conditions. A‘subject may have complete responsibility for the team score; he may have half the responsibility; or he may have no responsibility. These, too, are randomly assigned. When the ability combinations and control conditions are. taken together, it can be readily seen that.there are twelve distinct experimental conditions; and Camilleri and Berger ran, over a period of years, enough experiments so that they had data from approximately thirty subjects for each of these twelve conditions. Let us summarize so as to be able to appreciate the experimental conditions from the standpoint of Actor. Actor is faced with a binary choice situation, and he has certain ”information" at his disposal: (1) He has a certain level of ability. 5 (2) His partner has a certain level of ability. (3) He and his partner differed on their initial Judgments. (4) He.has a certain amount of reSponsibility for making the team's final decisions. Will he stick with his initial judgment, or will he change his.mind? That is, will he make a “self-response" or an “other-response” for his final choice? That is the question of interest for Camilleri and Berger, and the experiments reported in their article were structured so that each sub- ject participated in twenty trials which,.for purposes at hand, may be-regarded as independent. The proportion of selfyresponses, according to Camilleri and Berger, is theor- etieally predictable by the following algebraic.function: a + 0’: 1.15 NS) = where. c = —— and l + c ul + u3 u1 = The value to Actor of the experimenter's approval. u3 = The value to Actor of his partner's approval. “5 = The value tozActor of self-consistency. a_ = Actor's estimate of the probability that he is correct given that he and his partner disagree on their initial choices. Let us hasten to point out that this formula is a reasonably straightforward consequence of the model which Camilleri and Berger present in their article. It should be apparent from our summary that the 6 Camilleri-Berger research is an extremely creatiVe piece of work; our description does not do justice to the ingenuity and the elaborate;procedures and controls used by professors Camilleri and Berger in implementing this research.4 Now let us report the findings of the Camilleri-Berger study by reproducing a table from Camilleri and Berger's Sociometgy article. (See table I, page 7.) The.number'§ of the Camilleri-Berger equation was computed from a simple algebraic formula, and g was determined from the.(++) eXpectation state for each control condition. That is the reason for the 1's in the.table. We can see that the predictions were.very good for the (+-) and (—-) conditions, but not adequate for the;(-+)' condition. This last;observation could lead us to a number of conclusions, but two in particularrcome readily.to mind: (1) perhaps the model'does not apply to the:(-+) condition for one.reason or another, or (2) perhaps some incorrect assumptionS'were.made.in the analysis which led to the predictions.. WO-GhOOBOUtOib8119V6 the latter. III. THE GAMILLERI-BERGER EXPERIMENTS: A RE-ANALIBIS The Basic Plan of Attack The basic-premise upon which we shall work is that the less accurate,of the.Camilleri-Berger predictions resulted 4The most elaborate.exposition of the model, we might add, may be found in Santo F. Camilleri, Joseph Berger, and ghomas L. Conner, “A'Formal Theory of Decision-Making“, ociolggical Theories in Progress Berger' Zelditeh and Anderson (eds.), Boston: ought é Mifflin, 1968 (vol. II) TABLE I Predicted and Observed Mean Preportions of S Responses by Control & Expectation Conditions Expect. Full Control Equal Control A No Control State: Pred. Obs. N Pred. Obs. N (Pred. Obs. (+-). .75 .73 32 .77, .78. 29 .80 .82 3o (++) x .60 31 x .67 31 x .71 37 (--) .60 .52 3o .67‘ .65 32 .71 .73 33 (-+); .47 .24 31 .55 .44 28 .60 .43 35 8 directly from a mistaken assumption. We noted above that Camilleri and Berger determined the value of g.for each control condition from the (++) expectation state. They then assumed that this value of‘g would remain constant over the other expectation states, We take.this as our point of departure from the Camilleri-Berger analysis--we do;ggt assume that the value.of g remains.constant.over the four expectation states for each control condition. We.submit that'g is a function of several variables, and that there is good theoretical reason why it:should vary considerably from one expectation state to the next. The remainder of this.essay will, in a sense, be an argument for some formal propositions which will allow us to actually determine a formula for'g Which will account for the.CamilleriLBerger results, as well as for the results of experiments.which are yet to be done. We wish to present.an argument, and this argument will begin on a very intuitive level. Our rigor will pick up as: we go along, so we ask the reader to bear with us. Some Preligigggy Matters It will aid our later discussion to present some concepts. at this point. We shall be interested in subjective--that is, perceived, as opposed to objective--percentage abilities, and in subjective.probabilities of self or other being correct or incorrect on a given decision-making reSponse. We shall also be concerned with the percentages of control of self and other. In all cases, our concern will be with the mental 9 states of Actor, and these may or may not correspond to "objective” reality. XAS = Subjective.percentage ability of self. fiAo = Subjective percentage ability of other. $08 = Percentage.control of self. 100 = Percentage control of other. an. P(sr) = -—— = Subjective.probability of self being right. 100 P(sw) = l - P(sr) = Subjective-probability of self wrong. P(Or)' = —_°. = Subjective: probability of other being right. 100 P(ow) = l - P(or) = Subjective probability of other wrong. $03 . ‘ Q(O) = ——— = Standardized.quantity of control of.self. 100 Leteus~bear.in mind that all these concepts relate to twow person task situations.which call for binary decisions on the part of group members. Furthermore, let us-make clear that: we do not suppose XAS, for example, to have a generalized value; rather, the value which it'has is Specific to a given person, a given task, and a given point.in time. The same is true of 1A6 and the various probabilities. The reader will perhaps think that concepts which are so subjective in nature will lead us to serious methodological difficulties. Howevmr, we will, in the Camilleri-Berger research, be dealing with the averages of groups of approximately thirty subjects; so the values of these.variables may be taken to be quite stable 10 in our analysis. The values of the.variables which have to do.with control are, of course, determined by the experimenter in.the Camilleri-Berger research. They have a strong object- ive basis, and it is not unreasonable for us to assume that the objective reality is virtually identical to Actor's sub- jective reality with reapect to these variables. It will be helpful for our analysis to estimate the values of fins and flib for the four expectation states of the Camilleri-Berger research. These.estimates which we:are about to present, although we believe them to be highly accur- ate, were not arrived at via an a priori method. Xi: = 75 in the (++) and (+-) states. = 50 in the (--) and (-+) states. flAo = 75 in the (++) and (-+) states. = 50 in the (4-) and (+-) states. The probabilities which we defined above may, of course, be computed from these values. The percentage abilities which we have presented are not identical to the.percentages which Actor and Other go "correct” on the manipulation pre-test. The "true“ percentage abilities were 85 and 40 for the high and low subjects, respectively. We might offer the following explanation for the discrepancies: The subjects.were led to believe, before the manipulation phase, that a percentage ability of 62.5 was about average. Because the task was unfamiliar, each subject's initial expectations would have been based upon this figure; and each subject's ability would have been sub- ll jectively conceived, after phase one was over, as a percentage which reflected b23h_his initial eXpectations and the “evidence”. An explanation of this sort is consistent with the notion of selective perception which is central in the work of George Herbert.Mead, the.great philosopher and social psychologist whom we associate with symbolic interaction theory and itssoutgrowths.5 Now let.us define a procedure for obtaining the expected values of ggfor each expectation state: P(sr)P(ow) E(a) = P(sr)P(ow) + P(sw)P(or) This function is based upon the conditional probability laws which may be found in any textbook on- probability theory.6 We recognize that the relationship between objective and subjective probabilities is, in general, not a simple one; however, we assume that the.two approximate each other in uncomplicated situations, where "reality" is well-defined. Our procedure for determining E(a) is different from the procedure used by Camilleri and Berger in their analysis. They assumed that.E(a) was equal to the number that Actor got "correct” on the manipulation pre-test divided.by the number that.both got ”correct”, e.i., the sum ofictor's and Other's scores. The values.obtained by their procedure are, of SGeorge Herbert.Mead, Mind Self and Societ , Charles W. Morris (ed.), Chicago: The fintversity of GHIcago Press, 1934, especially the section on mind‘ 68ee.e.g. Emanuel Parzen, Modern Probabilit Theo and Its Applications, New York:' Wiley, I951, chapter 3 12 course, different from the values obtained by ours. Using our formula and appropriate values of its compon- ents, we.can obtain the expected values of §.for each expec- tation state.of the Camilleri-Berger study: .75 .50 .50 3(a+-) .25 E(a--) E(a++) E(a-+) These values are to be interpreted as the average values which therrandOm variable g_wou1d take on in a study using an infin- ite number of subjects-for each expectation state. That is, they are limits. In a study using approximately thirty sub- jecte for each expectation condition, we would anticipate slight deviations from these expected values. Although it seems a paradox, we would.npt expect our expectations to be met: Oh.the.other.hand, we would.not anticipate.large discrep- ancies, and our suspicions would be;aroused if a pronounced lack of correspondence were;to occur. A Workable Set of values. After examining the. data from the Camilleri-Berger exper- iments, the writer determined.by a long, tedious process of. trial and'error some workable values of §,and g,for each of the twelve experimental conditions. These values were chosen in such a way as to have certain properties, one of which is related to an assumption which we must now discuss. We have stated that we do n23 assume, as Camilleri and Berger do, that.the.va1ue of 3 remains constant over the four- 13 expectation states for each control condition. On the contrary, ‘we assumt that.g_varies within control conditions. Further- more, we.assume.that the operative characteristic within-con- trol conditions is-the degree.g§_eguality‘g§ ability of Actor and Other. Thus, g_has not one value, as professors Camilleri and Berger assumed, but Egg values within each control condi- tion. The values of §_and g; which we have determined are. presented in table.II on page fourteen. There are.three things which we would like to have the reader observe concerning these.valuesz (l)' The values of g are all very close to the expected values which we arrived at above. (2) Each value of.g goes with exactly two values of g, (equal ability or unequal ability): (3) These values, if plugged into the Camilleri-Berger equation (see page.5), give us-the desired 2(8) tn every case. Thus, it is plausible that the.valuesrwhich we have presented could be:very close to the true values of these.variables. The meanxdiscrepancy between our values.of.§_and the.expected values is.slightly less than .023, which for the moment we will not consider to be cause for alarm.7 This.mean discrep- 7The; true.discrepancy may be much less than this.figure. We are absorbing all of our error into a whereas the vast. majority of it should no doubt be absorbed into c, since the components ofig would--their values would, that_Is--be affected by the employment of different experimenters, by slight ir- regularities of procedure, etc. However, it will be conven- lent to treat g;as if it were fixed for purposes of our analysis, even though this is not literally the case. No harm should result. 14 TABLE II A Workable Set of values w—wfi—w‘ a++ a++ a++ Full Control .5455 .4545' .7300 .2400 Equal Control .5147 .4853 .7226 .2938 No Control .4821 .5179 .7607 .2422 .1362 .0000 .4706 g .2611 = .7859 I = .3294 .__..____J ll ‘Icll 1'11ch (I. l‘ [.4 1'. i a II I 15 ancy is Just a shade larger than what we would anticipate if our anticipations were based upon the Bernoulli model. New let.us-leave g and.take a closer look at the.variable g, which is, as we have said, the major concern of this essay. in Analysis of c When.we.examine the values of g which have.been determ- ined, it becomes clear that g is a function of gt‘;gg§t two things: (1)? The degree;of equality of ability of Actor and Other. (This should not surprise us, of course, since we'picked'g in such a way as to have this property.)T (2) The amountgf control which Actor has over group deoisicns. If we consider the two cases of ability, equal and unequal, separately, then we may graph 3 as.a function of the stand- ardized quantity of control. This we have done, and our results are.presented.on.page.sixteen. Let us point out that these curves are somewhat misleading. One appears to be a straight line, and the other appears to be perhaps a quadratic curve. We are getting ahead of‘our- selves, but we shall see later that these curves are determ- ined by the same function, and that they differ only in the values of two ability parameters. Now let us analyze g.in greater detail. It is intuitive- ly clear that'Actor will have to take Other into account a 16 ILLUSTRATION III Me): = rims» E(c), A 1.00 i Equal Ability: __________ Unequal Ability: .75 .50 .25 0(0). 0 17 great deal when Actor has full control; less so when Actor and Other have equal control; and not at all when Actor has no control over group decisions. For the sake of clarity, let us state.this as a formal assumption, albeit an obvious one: Assumption: The value to Actor of “ether approval“ varies in preportion to the amount of control which Actor has over group decisions. It could be argued that Other's ability would affect the value to Actor of "other approval". We reject this argument, however, and maintain that, with reapect to thisvariable, the operative characteristic is Actor’s Quantity 22 control. Let us arbitrarily assign the values (2,1,0) to u3 for the full control, equal control, and no control conditions, respectively. We could have chosen the values (4,2,0) or (6,3,0) or (8,4,0) or any number of others. Our choice merely determines the scale of measurement of our utilities, and is otherwise inconsequential. If we had chosen the values (4,2,0), then the values of our other utilities would have to have been doubled in the discussion which follows. Actor must take the esperimenter into account about the same.amount irrespective of the eXperimental condition; the value.which we get for ul is .85 for all conditions. We must.ask the reader to accept this value on faith for the present. It is.derived by solving a system of equations, which is a consequence of another system of equations, whose theoretical basis we have not yet discussed. A difficulty of the English language is that we can.say merely one thing 18 atha time! It:is.interesting to note that Actor values the approval of the.experimenter slightly less than he values. the approval of Other in the equal control conditions. We know the.value.of'g for each experimental condition, and the.value of two of its three components we also know in each case. Thus, it is.a simple matter to solve for the value of us for each condition. This we have done, and we present our results in table;IV’on page ninetemm. After studying_thevvalues.of the;u53 which are.presented, the writer was struck by the.presence.of three fascinating regularities. Specifically, the variable uSJ appears to be a function of the following three.things: (l) The degree of equality of ability. (2) The degree of equality of control. (3) The amount of control. Looking at table IV, we.see thatu51 is greater than u52, “53 is greater than u54, and u55 is greater than “56° These observations suggest that the variable u53 varies directly with the degree of equality of ability when the amount of control is held constant. Iboking at table IV again, we see that “53 is greater than both u51 and u55, and u54 is greater than both u52 and u55. These observations suggest.that.the-variable.u53 varies directly with the degree of equality of control when the equality of ability and the amount of contrhl‘are.held'constant. Looking at table IV one more time, we see that u51 is lags than u55, and u52 is $555 than u55. This third set of observations suggests that values of the.c3 and Their Components 19 TABLE IV o .. “5.1 ontrol—Ability cJ- - ' 111. + 1.133 .388 Full- Equal C 1 = m = e 136 . + .000 Full-Unequal c2 = —é-5———2- = .000 . + .871 Equal-Equal C3 : —8-5-—]—. 3 .471 . 4 .483 Equal-Unequal C4 = -gg———i' = .261 0 § .668 None-Equal c5 = -g;--S’ = .786 e + t .280 NonepUnequal °6 = -§§+——6. = .329 J e + 20 the.variable u53 variesinversely with the.amount of control when.the equality of ability and the equality of control are held constanta Based-upon these three sets of observations, we would suggest;that.u5J is a function of the following kind (the reader may wish to refer back to pages 9 and 10 during parts of this presentation): “53 = «Locum + boom) + comm) + 21121 where: so, b0, co, and the di are constants, and: E(A) E(O) NO) The equality function of ability. are, 100 The equality function of control. $03 - $00 100 1 - The standardized quantity function of control. $03 100 Various functions of individual differences and socio-oultural conditions which take on constant'values in a well-planned set of.1ab— oratory experiments. Thus, the sum of the d1T1 is equal to some constant do. For the sake of convenience, let:us determine the values of our functions for the different ability and control con- ditions of the Camilleri-Berger study. This will aid us in 21 the computations which we will be making. ( ) {1‘00 for all equal ability conditions. E'A = 0.75 for all unequal ability conditions. ( ) 0.00 for all full control conditions. E O = 1.00 for all equal control conditions. 0.00 for all no control conditions. “(0) 1.00 for all full control conditions. 0.50 for all equal control conditions. 0.00 for all no control conditions. Using these values,.we:may set up the following system of equations for the uSJ: u51 = 1.00 a0 + 0.00 bo + 1.00 00 + 1.00 do = .388. u52 = 0.75 a0 + 0.00 bo + 1.00 co + 1.00 do = .000 u53 = 1.00 ab + 1.00 bo + 0.50 00 + 1.00 db = .871 u54 = 0.75 a0 + 1.00 be + 0.50 co + 1.00 do = .483 u55 = 1300 ab + 0.00 ha + 0.00 co + 1.00 do = .668. u55- = 0.75 so + 0.00 be + 0.00 co + 1.00 db = .280 The values on.the far right hand sidezcoms from table IV on page.nineteen, and the various coefficients.are.appropriate values; as given above, for the different ability and control conditions. If we have been reasoning correctly so far in this.essay, then there:should exist.a solution to these equa- tions. The equatiOns should be consistent with one.another, and four of them should be linearly independent. The easiest 22 way to.cheok these things is to, first, express our system of equations as a matrix product in the following manner: \ \ / \ 1.00 0.00 1.00 1.00 so .388 0.75 0.00 1.00 1.00 ho : .000 1.00 1.00 0.50 1.00 co .871 0.75‘ 1.00 0.50 1.00 do) .483' 1.00 0.00‘ 0.00 1.00 .668; l 0.75‘ 0.00 0.00 1.00 .280; \ l x / Then we take the augmented matrix (A,U) and try to put it: into rowareduoed'form. Fortunately, this can be done, and we find that the values of.our unknowns are as follows: W 0 II 1.552‘ .343' -.280 -.884 8" D O H 9! o u Putting these values back into our general equation for the u53>gives-us.the;following result: **| u53 = 1.55213(L)) + .343(E(0)) - .230(Q(09) - .334 This, we believe, is a highly significantzderivation, and we shall return to it in the next section. It.is a trivial, althoughatime consuming, matter to verify that our function gives the appropriate values of the “53 for each of the:exper- imental conditions studied by professors Camilleri.and Berger. One interesting feature of our system of equations is 23 that this system gives us a means of evaluating ul, the value to.Actor of "experimenter.approva1“. It will be recalled thatrwe asked the reader to accept as an article of faith the factthatu1 was equal to .85 for all the experimental conditions. It is.now possible to demonstrate that.u1 must have this value if our theoretical assertions-about the compos- ition of theuusJ are correct. Suppose,.for the moments that.we had not attempted a matrix solution of our system of equations. If we had not been concerned with matters of consistency and'lrneer indepenp dence, then we could have used §g_hgg_methods. How would we have:solved'for the;value of the constant, so? We could have subtracted the second equation from the first; we could have subtracted the fourth from the third; or we could have sub- tracted the sixth from the fifth. In each case, we would have gotten:.25 ao equal to some other constant k1. Since this ig_a-consistent.set of equations, the k1 would have to all be equal to each other. This means that u51 minus “52 must equal “53 minus u54 which must equal u55 minus.u55. Erom.table IV we see that the us: can be expressed in terms of the c3, “1' and the u3J. Thus, we may set up a second system of equations as follows: c1(u1 + u31)'- c2(u1 + u32) 03(u1 4‘. 1.133) ' 04(111 + 1134) c5(u1 + u35) - c5(u1 + u35) We know the.values of the c3 and the u33; these may be read 24 from table IV on page nineteen. Thus, we may modify this sys- tem of equations in the following way: .136(u1 + 2) - .000(u1 + 2) .4Z1(u1 + l) - .261(u1 + l) .786(u1 + o) - .329(u1 + o) If we multiply everything out and collect terms, then we have three simple, consistent equations and two unknowns. Solving for “1 gives us .85, whiCh, of course, we’ve.known for-some time now! The point, however; is that this value is not arbi- trary; it is a mathematico—logicai consequence of our theoret- ical.assertions. These equations allow the reader a good Opportunity to-prove:to:hhmself that our choice of values of the “33 determined merely ourrscalerof'measurement for:the three;utilities. Now it is time to present.in a moretsystematic.faahion the theoretical ideas.which we’ve been developing. IV. THE‘THEORIfRE-BTATEDA Let‘ussbegin with a few remarks about the scope of the theory. We recognize, with the.reader, that, in general, the value to Actor of selfAesteem would be a function of many things. It seems reasonable to the writer that this value could be influenced by factors in each of the following three categories: (1) Idiosyncratic Factors: These would include such "individual differences” as inherited.mental.poten- 25 tials (related to plasticity of the nervous system), and certain temperamental qualities which are be- lieved by=some' 0 have an endocrinous basis. (2) Socio-cultural‘ Factors: These would include a large number of things, but'especially cultural definitions relating to social rank and to various status characteristics. For a graphic illustration, see the Veblen quotation preceding this essay. (3) Situational Factors: These would include ability and control, or power, relative to the ability and control of others in one's immediate lifepspace. These factors would be most evident in a task situation. The reader will recall that our original hypothesis about the value to Actor of self-esteem was-embodied in the fellowb ing equation: ”‘51 = acme): + bow-(an + comm): + 241:, He stated very briefly that; in*a well-planned set of labora- tory experiments, the sun of the diTi would be equal to some constant, do. This amounted to saying that we assumed that. the idiosyncratic and socio-cultural aspects could be treated as.constants when we were dealing with the averages of groups of approximately thirty subjects drawn from the same popula- tion. Any theory must take certain things as "givon‘; we have.nOW'made our concerns explicit. The scope of our theory anludea‘gg y the.vory importantA-but much neglected--class of factors, situations; factors. 26 Let us state-~we have, until now, only implied--that we assume that the idiosyncratic and socio-cultural factors, and any interactions among or between these factors, could be expressed by a finite number of mathematical functions, and we have chosen to call these the T1. It is not necessary that we actually determine these functions: we must only acknowledge that.they do, theoretically, exist. Each of the T1, as well as E(A), E(C), and 0(0), is the mathematical characterization of what we shall call an “effect". So when we Speak of an effect, we are speaking of the substantive counterpart of a mathematical function of one or more variables, that is "factors". The reader should keep this conceptual terminology in mind in order to avoid'confusion with reSpect to the statements Which follow. Before stating the theory, let us refresh our memories about equation (**), whiCh we derived in the last section. n 1153 = +1.552‘(E(A.‘)') + .343(s(c)) - .280(o(c)) .. .884 The reader should pay particular attention to the signs, positive or negative, of the coefficient terms. With equation (fit) at.the front of our minds, we state the theory: Assumption 1: The effects involving individual differences remain fixed at a given level. Assumption 2: The effects involving socio-cultural factors remain fixed at a given level. PrOposition l: The value to Actor of self-esteem varies directly with the degree of equality of ability (other situational factors held constant). 27 Proposition.2: The,value to Actor of selfsesteem varies directly with the degree of equality of control ether situational factors held. constant e Proposition 3: The value to Actor of self-esteem varies inversely-with the amount of control which Actor has (other situational factors held constant). Preposition.4: The effects which determine the value.to Actor of‘selfaesteem are additive. It is interesting_to note.that:the two effects having to do with control work in opposite directions. In concluding this section, let us remind the.reader that.our theory is intended to apply to two-person task situ- ations. We require that it be important to both persons that, the task.be.performed wellp-both persons must have some stake in the results. V.. A‘ NOTE ON. UTEGRITI', ESTEEM, AND VALUE TQ ACTOR It will be recalled that:we said early in this essay that. selfpesteem can take on, for Actor, very low values, and even negative values (see page:two). We shall now briefly comment upon this. In.ordinary discourse, people frequently lump ideas to- gether which,.for scientific purposes, mightrbemter be kept separate. This seems.to be how it is with reSpect to the concepts of integrity, esteem, and value to Actor. To clarify the relationships between these ideas, let us consider a' simple illustration. Suppose we have a two-person task situ- ation which calls for a decision. Suppose that there are two 28 possible responses, R1 and R2, and that it is.known that.Actor believes.Rl to be “correct”, and Other believes R2 to be.the "correct? responsel Suppose Actor selects 32? Then he.main- tains the;integrityeof.the other person, and he.reoeives Otherls esteem. Common senee_tells us that this esteem would normally be rewarding to Actor, but we can imagine cases: where Actor would negatively value the esteem of the other persons-for-example, the.case.where Actor disliked Other. Now, suppose Actor selects R1 instead; Then he maintains‘ his personal integrity; he receives self-esteem; and he normally finds this rewarding. But.there:g£g,cases;where. Actor would negatively value self-esteem. These cases.are harder to imagine because they are not recognized by the con- ventional wisdom, but our theory makes clear that such cases are possible. He.shall discuss one such casezin theenext. section of our presentation. A truly definitive analysis of the relationships between integrity, esteem, and value to Actor is obviously outside the scope of this essay. We would informally suggest, however, that the value to Actor of self-esteem.varies directly with the degree.of perception of.legitimacy, by Actor, of the power:(oontrol) relations-as, for example, in a bureaucratic organization.8 We would also informally suggest the following: The.value to Actor of self-esteem varies inversely with the degree to which Actor perceives his activity as being import- 8For a discussion of this general tOpic, see Max Weber, The Theor of Social'and Economic Or anization, Talcott FErsons (ed.), New York: “TE? Free Pgess, I947, pp. 324-424 29 ant to a "causey (e.g. science). A proposition of this sort would account.for the negative_value offdo in equation (**). The truth or falsity of our informal propositions in no way affects the truth or falsity of our theory. Our.only' purpose in stating them is to suggest the.complex:nature of the;relationshipeabetween.personal integrity, self-esteem, and.valhe to Actor. Nowzlet;us return.to matters.which are- within the scope of'our theory. VI. INVESTIGATING THE THEORY‘ An Independent Test. Our theory, when used in conjunction with the.Oemilleri- Berger~donner theory, describes the data from whence it.came admirably. ButLth19;1B not enough: A theory, if it is to» be;useful,.must:be able to:predict events which have not yet occurred. Hence, we shall.now.report a study.s1milar.to the one done by professors Camilleri.and Berger, but with Actor's quantity.of control altered so as to form eight new experi- mental conditioned Subjects.--Vblunteers from introductory sociology, classes were used, and were:paid a small amount for their participation. Subjects.drawn in this way tend to be reason- ably naive, and reasonably alike with respect to personal and social characteristics. These considerations are, of course, important. TWenty subjects.were,used for each of the eight experimental conditions. 30 gypotheses.--The.specific hypotheses of the study may be.taken to-be the basic proposition of the Gamilleri-Berger» 00nner.decision~making theory and the four propositions of our own theory. The.model.--By performing some algebraic manipulations, and combining some of the results we have obtained in previous. sections of this essay, we.can express the.numberwg of the 0amilleri-Bergerrequation as a function of 0(0); Two cases aresneededs’ fl is appropriate.for the equal'ability conditions, and‘f2_is appropriate.for the unequal ability conditions; 1.011 - .343 {|2(Q(o)) - 1|} - .28(Q(0)) flown. .625 - .343 {|2(e(0)) - 1.L} - .28(o(0)) .85 + attic)? r2