18
Contradiction and Change of Performance Expectations* M A R T H A F 0 S C H I / University of British Columbia L‘auteur propose deux hypothbes pour justi6er le fait que les anticipations de e performance D changent lorsqu’elles sont mises en prksence de faits objectifs qui les contredisent. On peut pr6dire que ce changement dans les anticipations Cvolue en fonction de la variCt6 et du niveau des contradictions. Le cas de celui dont les attentes pour soi sont infCrieur aux attentes pour les autres est utilis6 pour des fins exphimentales de vCri6cation des hypothtses. A la fin de l’article I’auteur apporte des suggestions dans le but d’amkliorer l’arrikre-plan thCorique des hypothhses. Two hypotheses are proposed on the way specific performance expectations change when confronted with objective evidence that contradicts them. The change in expectations is predicted to be a function of the number and the degree of the contradictions they receive. The case of initial lower expectations for self than for other is used in an experimental test of the hypotheses, and these are confirmed by the results of the test. Some suggestions for the refinement of the hypotheses’ theoretical background are discussed at the end. INTRODUCTION Heider (1944) suggested that the evaluation of an act and the evaluation of the person who did the act are not independent. Since then, several authors (Harvey, 1953; Sherif, White, and Harvey, 1955; Berger and Snell, 1961; Berger, Cohen, and Zelditch, 1966; Lewis, 1966; Camilleri and Berger, 1967; Moore, 1968, 1969; Berger, Conner, and McKeown, 1969; Webster, 1969) have either tested empirically or included as an assumption the follow- ing proposition: when subjects are motivated to make correct evaluations of performance, but are left with no objective criteria with which to make them, they use the expectations they hold about the performers as substitute criteria. Thus, under these conditions, a performance by a person who is expected to be good tends to be highly evaluated and a performance by a * This paper was presented at the annual meetings of the Canadian Sociology and Anthropology Association, St John’s, Newfoundland, June 6-9, 1971. Its theoretical section represents part of a doctoral dissertation submitted to the Department of Sociology at Stanford University. Financial support for the experiments was pro- vided by the Canada Council under grant 68-0842. The author gratefully acknowl- edges the advice and support received from Bernard P. Cohen, Bo Anderson, Reginald A. H. Robson, and Morris Zelditch, Jr. She is also grateful to Susan Clark and Frances M. Reimer for their invaluable assistance as experimenters, and for Mrs Clark’s critical comments on earlier drafts of this manuscript. 205 Rev. canad. SOC. & AnthJCanad. Rev. SOC. & Anth. 8(4)1971

Contradiction and Change of Performance Expectations

Embed Size (px)

Citation preview

Page 1: Contradiction and Change of Performance Expectations

Contradiction and Change of Performance Expectations*

M A R T H A F 0 S C H I / University of British Columbia

L‘auteur propose deux hypothbes pour justi6er le fait que les anticipations de e performance D changent lorsqu’elles sont mises en prksence de faits objectifs qui les contredisent. On peut pr6dire que ce changement dans les anticipations Cvolue en fonction de la variCt6 et du niveau des contradictions. Le cas de celui dont les attentes pour soi sont infCrieur aux attentes pour les autres est utilis6 pour des fins exphimentales de vCri6cation des hypothtses. A la fin de l’article I’auteur apporte des suggestions dans le but d’amkliorer l’arrikre-plan thCorique des hypothhses.

Two hypotheses are proposed on the way specific performance expectations change when confronted with objective evidence that contradicts them. The change in expectations is predicted to be a function of the number and the degree of the contradictions they receive. The case of initial lower expectations for self than for other is used in an experimental test of the hypotheses, and these are confirmed by the results of the test. Some suggestions for the refinement of the hypotheses’ theoretical background are discussed at the end.

I N T R O D U C T I O N

Heider (1944) suggested that the evaluation of an act and the evaluation of the person who did the act are not independent. Since then, several authors (Harvey, 1953; Sherif, White, and Harvey, 1955; Berger and Snell, 1961; Berger, Cohen, and Zelditch, 1966; Lewis, 1966; Camilleri and Berger, 1967; Moore, 1968, 1969; Berger, Conner, and McKeown, 1969; Webster, 1969) have either tested empirically or included as an assumption the follow- ing proposition: when subjects are motivated to make correct evaluations of performance, but are left with no objective criteria with which to make them, they use the expectations they hold about the performers as substitute criteria. Thus, under these conditions, a performance by a person who is expected to be good tends to be highly evaluated and a performance by a

* This paper was presented at the annual meetings of the Canadian Sociology and Anthropology Association, St John’s, Newfoundland, June 6-9, 1971. Its theoretical section represents part of a doctoral dissertation submitted to the Department of Sociology at Stanford University. Financial support for the experiments was pro- vided by the Canada Council under grant 68-0842. The author gratefully acknowl- edges the advice and support received from Bernard P. Cohen, Bo Anderson, Reginald A. H. Robson, and Morris Zelditch, Jr. She is also grateful to Susan Clark and Frances M. Reimer for their invaluable assistance as experimenters, and for Mrs Clark’s critical comments on earlier drafts of this manuscript.

205

Rev. canad. SOC. & AnthJCanad. Rev. SOC. & Anth. 8(4)1971

Page 2: Contradiction and Change of Performance Expectations

person who is expected to be poor tends to receive a low evaluation. This proposition has received considerable support.

Berger and Snell's theory also predicts that each time an evaluation in line with the expectations is made under the conditions described above, the expectations tend to be maintained. Expectations are then very well pro- tected when they constitute the only criterion used to make the evaluations. But what happens when objective criteria are used to arrive at the evalua- tions, and these evaluations contradict the expectations? Are the expectations destroyed immediately, or is there a tendency to maintain them even in this case? If there is such a tendency, how resistant to change are the expecta- tions? How does the number and the degree of the contradictory evalua- tions affect them? The formulation of a problem suggested by these questions is presented in this paper, together with some theoretical considerations and two hypotheses proposed to answer the questions. In addition, an experi- mental test of one of the cases of the hypotheses is reported and its results examined.

T H E P R O B L E M

The scope of the problem is limited by (i) the various ways in which expec- tations are characterized,l and by (ii) the manner in which their contradic- tion takes place.

(i) Firstly, expectations are predictions which a person makes with the in- tention of being correct. In other words, the expectations do not refer to what he hopes would happen, nor to what he thinks should happen. They are beliefs with an empirical content which can be proved to be correct or incor- rect by subsequent events. Expressions of hope and normative statements cannot be so proven. In the second place, expectations are conceptualized as associated with abilities that a person has. These abilities, in turn, are said to have states, such as various levels of mathematical ability, or a large or small amount of physical strength. States of an ability are differentially evaluated if the performances that correspond to one of them are thought of as better or worse than the performances corresponding to other states. For example, if running a mile in four minutes is considered to be better than doing it in ten, then the states of the ability reflected in each performance are differentially evaluated. Expectations are, in turn, assumed to be asso- ciated with each of these states and to refer to future performances requiring the ability. Thirdly, the abilities and the expectations are specific, that is, they refer to a single and well-defined type of performance. For example, the specific expectations may be based on a person's ability to play chess, as opposed to the only broadly characterized performances that can be expected of someone who is described as being very intelligent. Finally, the expecta- tions are conceptualized as relative. That is, a person is not described with

1 This characterization is based on the work of Berger, Cohen and Zelditch, in par- 206 ticular on the paper cited above (1966).

Page 3: Contradiction and Change of Performance Expectations

respect to what he expects regarding his own performances only, but his expectations for self are said to be higher or lower than, or equal to, his expectations for others. Considering one person, self, who has only himself and another person, called other, as objects of orientation, three types of expectations are possible. These are indicated between brackets: [+-I, higher expectations for self than other; [-+I, lower expectations for self than other; [OO], equal expectations for self and other.

(ii) The scope of the problem on the contradiction of the expectations is limited as follows. The expectations refer to a series of future performances and each step of this series consists of independent performances by self and by other. When the performances actually occur the evaluations that are made at each step contradict self’s initial expectations. In addition, self considers these contradictory evaluations to be objective, that is, to reff ect accurately the performances.

The evaluations at each step are referred to as a “unit evaluationy’ and are indicated between parentheses. Like expectations, unit evaluations are rela- tive and three types are possible: (+-), higher evaluation for self than other; (-+), lower evaluation for self than other; (00), equal evaluations for self and other.

In addition, the following terminological conventions are used regarding the relationships between expectations and unit evaluations. When the two are of the same type, as for example in the case of [+-I and (+-), they are said to be in correspondence. On the other hand, when their types differ, they may contradict each other in one of two degrees: when the signs in one of them are the reverse of the signs in the other, as in [-+I and (+-) , they are considered to be in total contradiction; when either [OO] or (00) are in- volved the relationship is said to be one of partial contradiction, as in [OO] and (-+),or [+-]and (00).

In summary, the problem is the following. There are two persons, self and other, but the situation is looked at only from self’s point of view. Ini- tially, he holds one of the three expectation types. As indicated by the limita- tions set out above, self believes these expectations to be correct. On the other hand, he cannot dismiss as inaccurate the unit evaluations he begins to receive, which are in contradiction with his initial expectations. Furthermore, he is assumed to continue to be motivated to have correct expectations. How does the number and the degree of these contradictions affect his expecta- tions? In the next section two hypotheses are presented in an attempt to answer these questions.

T H E H Y P O T H E S E S A N D T H E I R T H E O R E T I C A L B A C K G R O U N D

In order to provide a rationale for the hypotheses a conceptualization of “expectations”’ presented by the author in another paper2 is used. The con- ceptualization assumes that when a series of performances is anticipated by

2 See Foschi (forthcoming). 207

Page 4: Contradiction and Change of Performance Expectations

self, his expectations will have a generalized character and will consist of predictions about a number of these future performances. In terms of the problem introduced in the previous section, expectations are seen as sets of more than one predicted unit evaluation. These sets are referred to as “a sets.”

The a sets may be composed of any combination of the three types of unit evaluations; thus expectations are regarded as frequency distributions of these evaluations. One can then process the unit evaluations in a given a set by means of certain rules and determine what types of expectations the set constitutes. One such set of rules is proposed in the paper mentioned above.3 In addition, the use of these rules permits the representation of expectations on a continuum, according to the various combinations of the three types of unit evaluations which constitute their a sets. At one extreme of this con- tinuum would be the [+-] expectations with an a set containing only (+-) unit evaluations; at the other extreme, the [-+I expectations formed ex- clusively of (-+) unit evaluations. At the centre would be the [OO] expectations formed of either (00) unit evaluations only, or an equal num- ber of (+-) and (-+) in addition to any number of (00) unit evaluations. These a sets at both extremes and the centre of the continuum will be said to be “pure” cases of expectations; all other instances of a sets would fall closer to or farther from these three cases.

Finally, the following assumptions are made regarding the relationship between the type of expectations that results from a given a set and m e actual4 unit evaluation of performance. It is postulated that when the expec- tations and the Unit evaluation are of the same type or in correspondence, the situation is balanced and the expectations remain unchanged. On the other hand, imbalance is created if they contradict each other. This im- balance is assumed to produce tension which, in turn, generates a ten- dency towards balance.

There are various ways in which this balance may be restored. In order to simplify the problem it is assumed that self has only two possibilities avail- able, and that other balancing solutions are blocked.6 The two possibilities

3 These rules are formulated in terms of four types of unit evaluations: (+-), (-+), (++), and (--). In order to use the rules for the present problem, simply substitute (00) for either (++) or (--).

4 Notice that it is possible to talk about the relationships between the types of expec- tations and unit evaluations in two different contexts and that the terms “predicted’ and “actual” are used to identify these contexts. When a unit evaluation is part of the (I set which makes up expectations of a given type, it is a predicted unit evalua- tion. Only in the “pure” cases may all the predicted unit evaluations be of the same type as the resulting expectations. On the other hand, one may refer to the relation- ship between the type of expectations that results from a given a set and a unit evaluation of performances by self and other which have actually taken place.

5 Other solutions which could be followed to resolve the imbalance are given below. The list is not exhaustive, mainly because “tension” and “balance” have been left undefined. However, it would seem that these two are major alternatives for the solution of the imbalance: ( a ) self may decide that his initial expectations were about a different ability; ( b ) self may redefine the ability as unimportant. Accord- ingly, he will neither differentially evaluate its states nor hold the corresponding expectations for them. In. the design of the experiment attention was given to 208

Page 5: Contradiction and Change of Performance Expectations

are either to maintain the composition of the a set intact or to alter it. The alteration may leave the type of expectations unchanged or may modify it.

Two hypotheses are proposed to account for the way expectations will change under these conditions. Only general trends are predicted at the present time, although it would be more desirable to specify how the two possibilities are used at each step of the series of contradictions. Such a specification, however, would be very complex6 and not practicable since the conceptualization of expectations as a sets has not been tested before. On the other hand, the results from the present test should indicate whether such a refinement of the theory would be worthwhile and, if so, how it should be carried out.

The rationale for the first hypothesis is that self will change his expecta- tions by introducing into the a! set the unit evaluations in contradiction with the initial expectations. As this rationale can only be used to state the hy- pothesis as a general trend, no specific form is proposed for the relationship. Thus : Hypothesis I. The larger the number of actual unit evaluations in contradic- tion with the initial expectation type, the closer the a set will become to the “pure” case in correspondence with those unit evaluations.

If Hypothesis 1 is considered together with the previously made assump- tion regarding expectations and an actual unit evaluation of their same type, a corollary can be derived which limits the scope of the hypothesis. Corollary I. If the expectations become of the same type as the actual unit evaluations the situation is balanced and the type of the expectations ceases to change.’

Thus the type of expectations in correspondence with the unit evaluations becomes a bound for the trend described in Hypothesis 1.

The second hypothesis compares the magnitude of the change in the two degrees of contradiction. Using the representation of expectations on a con- tinuum and assuming that change occurs according to Hypothesis 1, it can be concluded that a given a! set is altered more by a totally contradictory unit

blocking these alternatives. The nature of the experimental task did not change and the subjects were told that having the ability to perform this task was positively evaluated.

6 For example, the assumptions refer only to the imbalance created by one contra- dictory unit evaluation, not to the effect of a series of such contradictions, and are not specific enough about how the composition of the a set will be affected by this single contradiction. Let us assume that, as the result of a contradictory unit evaluation, self will alter the composition of the (Y set by incorporating this evalua- tion to it. But it does not seem to be a realistic assumption to state that he will do so at each step. For example, in all instances where the expectations are not “pure” cases it is likely that self will not feel any need to make changes in the a set after one contradiction. The reason for this would be that this contradictory unit evalua- tion is already contained in the set and thus “expected.” But unless one knows at least the number of elements in the initial a set and can establish the degree of accuracy with which self attributes a size and a composition to this set, it is im- possible to say exactly what will happen to it after each contradiction.

7 Notice that expectations may be balanced with the unit evaluations without their (Y

sets becoming the “pure” case. The reference to this case in Hypothesis 1 is meant to indicate only the direction of change. 209

Page 6: Contradiction and Change of Performance Expectations

evaluation than by one that only contradicts the expectations partially, thus: Hypothesis 2. The greater the degree of contradiction between the initial expectation type and the actual unit evaluations, the faster the rate of change in the expectations.

EXPERIMENTAL PROCEDURES8

An experiment was designed in order to test the hypotheses in the case of initial [-+I expectations. Two persons participated in each of the runs: self, a naive unpaid subject and other, a paid confederate of the experimenter. The subjects were female undergraduates at a Canadian university who had volunteered for a “study involving a perceptual task.” In addition, they had indicated no previous participation in experiments in sociology or psychology and had taken at most an introductory course in these disciplines. The study consisted of two experimental conditions and a control condition, and the subjects were assigned at random to a run of one of the three.

Once in the laboratory, the two persons saw each other briefly9 before sitting behind individual tables, separated from each other by a partition. They did not interact directly with each other before or during the experi- ment. They were read a set of instructions by the experimenter and then performed a task described to them as requiring an intuitive ability called ‘‘Pattern Recognition.” These instructions were aimed at forming certain beliefs in the subject, for example, having this ability was said to be positively evaluated and the task was described as a reliable test of Pattern Recognition Ability.

The task involved a series of patterns projected on to a screen. First, a single pattern was shown for five seconds. After it had been removed, a re- sponse set consisting of four distinctively labelled patterns, the original pattern plus three similar ones, was presented for ten seconds. The task was to select from the four patterns the one which was exactly the same as the original. There were thirty different patterns and their corresponding re- sponse sets.

Each original pattern was composed of thirty punches in a section of a data processing card. The four patterns in each response set also had thirty punches each. The instructions indicated that in each of the sets there was one pattern corresponding exactly to the original which had to be identified, one corresponding by twenty punches, one by ten and one which had no cor- respondence at all with the original. The higher the degree of correspondence between the original pattern and a person’s choice, the better the perfor- mance and the higher the score.

8These procedures are presented and discussed in detail in Foschi (1970:2041, 105-151), where basically the same design was used to test the hypotheses with respect to initial [+-I expectations.

9 A total of four persons served as confederates, the main criterion for their selection being that their appearance would be typical of that of the students who would

210 serve as subjects.

Page 7: Contradiction and Change of Performance Expectations

Each person had a card file on her desk containing four cards per pattern. They were told that these cards were identical to the four patterns projected in the response sets. After a response set had been shown, the two persons made their selections by taking from their files the card labelled the same as the pattern that they chose. Each person then gave her card to the experi- menter, who projected it on top of the original pattern in order to determine the score. Each person counted her own score, and the card was not removed until the other person had indicated her agreement with the reported score. At all times the procedure was such that the subjects believed that objective evaluations of their performances were being made. Neither person had, however, any control over the scores they received, as the cards in their files had been punched to result in prearranged correspondences. This could be done without arousing the subject’s suspicion because the task was actually ambiguous: the four patterns in each response set were very similar to each other and they were projected for only a short time.

After the scores for the two persons had been counted, they wrote their individual predictions of the scores each would obtain in the next pattern on a special form. These forms were to be placed face down in a given box. This was done to avoid the possibility that the subject would mechanically copy her previous predictions and thus not evaluate self and other. For a similar reason the scores were only called aloud and no records of the perfor- mances were given; thus the subject’s expectations depended on what she remembered of the scores obtained by both.

The subjects were to make their predictions with the intention of being correct and they could express their estimates in terms of any numbers between zero and thirty, not necessarily in terms of the four possible scores. It was expected that the subject’s predictions for each pattern would thus reflect not only the type but also the composition of her a set. The score predicted for self less the score predicted for other would indicate the type of the expectations: a positive value for [+-I, a negative value for [-+I and a zero for [OO] expectations. The place of the a set on the continuum of expec- tations would be given by the size of these predicted differences.

Once the predictions were recorded, the next pattern and its response set were projected and the entire sequence repeated. When the task was completed the subject was interviewed in order to determine how effective the experimental procedures had been in creating certain beliefs, in particular the existence of Pattern Recognition Ability. The interview also served to find out if there was any suspicion and, at the end, to reveal the real nature of the study.

Each run of the experiment consisted of two parts. The scores during the first part were used to form [-+I expectations and were the same for the three conditions.1o The second part varied for each case: in the experimental

10 In order to ensure that the subjects in the three conditions would be as similar as possible in the expectations they formed several steps were taken to remove other possible sources of expectations. To this effect, the instructions indicated that only 21 1

Page 8: Contradiction and Change of Performance Expectations

212

conditions this part was used to contradict the initial expectations, totally in one case and partially in the other; in the control condition it served as a test of whether these expectations had been formed.

The first fifteen patterns were used to create the required expectations. In order to avoid suspicion neither person was given the same scores for all iif- teen patterns, instead, self received scores of 10 (a “fair” level of the ability) for most of these patterns and other received mostly scores of 20 (a “good” level). The scores for self and for other were paired so that they would result in a majority of (-+) unit evaluations (i.e., self received a lower score than other) ; the differences between the two scores at each of these patterns are shown in broken lines in Figures I, 11, and III.

The scores received during the last fifteen patterns served in the experi- mental conditions to contradict the expectations initially formed. No separa- tion between the two parts was made explicit to the subjects and the pro- cedure was exactly the same as during the first fifteen patterns. In both con- ditions the scores for other remained the same as during the initial trials. In the total contradiction condition self received scores of 30 (“perfect scores indicating an excellent level of Pattern Recognition Ability”) for most of the patterns; in the partial contradiction condition most of self’s scores climbed only to the level of 20, that is, the same level as other’s. The arrange- ment of these scores resulted in a majority of (+-) unit evaluations in the total contradiction condition and in a majority of (00) unit evaluations in the partial contradiction condition, as shown in broken lines in Figures I and 1 1 . ~ ~ For each pattern the values for the total contradiction condition were ten points larger than for the partial contradiction condition.

In the control group the two persons did not receive any scores during the second fifteen patterns. Instead of selecting a card from the file, for each pattern they wrote their choices on the appropriate form and gave it to the experimenter, who read them aloud in a pre-arranged fashion. Although self’s choices were reported accurately, the confederate “choices” were arranged each time so that for most of the patterns self and other were said to have differed in their selections. After the choices were announced, they predicted for the next pattern as usual.

These predictions served to test whether the required [-+I expecta-

pattern recognition ability was needed to perform well, that this specific ability had been recently discovered, that it was unrelated to sex, age, level of education, grades, various skills and IQ, and that most probably it was an innate ability. It was realized that, in spite of these precautions, variables such as the subject’s general self- concept or cues she may have picked up from the other person could have acted as a source of expectations. To prevent such from occurring, certain procedures, d e scribed at the beginning of this section, were followed: the subjects were drawn from a relatively homogeneous population, the confederate was chosen so that she would appear to the subject to be of equal status in many respects (and thus be less likely to serve as a source of differential expectations), the two persons were not allowed to interact directly with each other and, finally, the subjects were assigned at random to the three conditions of the experiment.

11 Although the problem specifies that all the unit evaluations contradict self‘s initial expectations, the design included only a majority of such unit evaluations in order to avoid suspicion.

Page 9: Contradiction and Change of Performance Expectations

20

10

0

- 10

*---

--

SC

OR

E

RE

CE

IVE

D

BY

S

EL

F

LE

SS

S

CO

RE

R

EC

EIV

ED

B

Y

OT

HE

R

AT

EA

CH

P

AT

TE

RN

- MEA

N S

CO

RE

P

RE

DIC

TE

D

FO

R

SE

LF

L

ES

S

ME

AN

S

CO

RE

P

RE

DIC

TE

D

FOR

O

TH

ER

F

OR

E

AC

H

PA

TT

ER

N

1 3

5

7

9

11

13

15

17

19

2

1

23

25

27

29

31

-X

PA

T T

ER

N

NU

MB

ER

5 w

FIG

UR

E 1

. RE

CE

IVE

D A

ND

PR

ED

lCT

ED

SC

OR

ES.

PA

RT

IAL

CO

NT

RA

DIC

TIO

N C

ON

DIT

ION

N =

15

Page 10: Contradiction and Change of Performance Expectations

20

10

---"--

SCO

RE

R

EC

EIV

ED

B

Y

SE

LF

L

ES

S

SC

OR

E

RE

CE

IVE

D

BY

O

TH

ER

A

T E

AC

H

PA

TT

ER

N

- -

-ME

AN

S

CO

RE

P

RE

DIC

TE

D

FO

R

SE

LF

L

ES

S

ME

AN

S

CO

RE

P

RE

DIC

TE

D

FO

R

OT

HE

R

FO

R

EA

CH

P

AT

TE

RN

PA

T T

ER

N

NU

MB

ER

FIG

UR

E 11.

RE

CE

NE

D A

ND

PREDICTED SC

OR

ES.

TO

TAL CONTRADICTION CONDITION

N =

14

Page 11: Contradiction and Change of Performance Expectations

Y

20

10

11 - 111

D : D

ISA

GR

EE

ME

NT

A:

AG

RE

EM

EN

T

------

SCO

RE

R

EC

EIV

ED

B

Y

SE

LF

L

ES

S

SC

OR

E

RE

CE

IVE

D

BY

O

TH

ER

A

T E

AC

H

PA

TT

ER

N

-

ME

AN

S

CO

RE

P

RE

DIC

TE

D

FO

R

SE

LF

L

ES

S

ME

AN

S

CO

RE

P

RE

DIC

TZ

D

FO

R

OT

HE

R

FO

R

EA

CH

P

AT

TE

RN

DD

DA

DA

DD

AD

AD

D O

D

9

11

13

15

17

19

21

23

25

27

29

31 <

1

3

5

7

PA

T T

ER

N

NU

MB

ER

FIG

URE m. RE

POR

TE

D C

HO

ICES

AN

D R

EC

EIV

ED

AN

D P

RE

DIC

~D

SCORES. CONTROL C

ON

D~

ON

N =

14

VI

Page 12: Contradiction and Change of Performance Expectations

tions had been created. The studies mentioned in the introduction, in par- ticular the theory by Berger and Snell, and the conceptualization of expecta- tions as (Y sets provided the rationale for the test. If self knew only that the other person’s choice had differed from hers, and if she had no objective criteria with which to evaluate the performances, she would use her [-+I expectations as substitute criteria. More specifically, if these expectations had been formed, a disagreement in their selections would be interpreted by self as (-+) rather than as (+-). This, in turn, would result in the main- tenance of the expectations for the next pattern. It was also assumed that the scores received during the initial fifteen patterns would result in an a set with a large enough number of (-+) predicted unit evaluations to enable the subject to interpret all the disagreements in the second part of the task as a poorer performance by self than by other. On the other hand, if the dis- agreements were all that the subject needed to entertain the idea that she was equal to or better than the other person, this would be sufficient indica- tion that no [-+I had been formed.

The control treatment included four agreements for the purpose of allaying any suspicion. The order in which agreements and disagreements occurred is indicated in Figure 111. The [-+I expectations, however, could not provide a criterion to interpret these agreements; they implied equal ability on the part of both persons and could actually change the expectations into [OO] . Nevertheless, the four agreements were included because it was considered that they would not be sufficient to accomplish this change.

RESULTS AND DISCUSSION

A total of sixty subjects participated in the study, twenty in each of the con- ditions. For several reasons, seventeen had to be discarded. One subject failed to give predictions for one of the patterns and another never predicted a score for the other person. A third misunderstood the instructions and thought that a low score indicated a high degree of Pattern Recognition Ability. When interviewed upon completion of the task, seven subjects re- vealed suspicion about the real nature of the study. Finally, seven others were eliminated because the scores they predicted for the first fifteen patterns indicated that they had not formed lower expectations for self than for other. The rule adopted to decide whether or not the subjects had developed the appropriate expectations was as follows: only those who had predicted at least a ten point difference in favour of other for eight or more of these patterns would be retained.12 This rule ensured that the [-+I expectations

12Since the existence of a given type of initial expectations was a condition of the hypotheses, the subjects who did not meet this condition had to be eliminated. The predicted scores were assumed to indicate the expectations the subjects held for each pattern, but no specific theoretical guidelines had been set regarding their expec- tations over all of the initial fifteen patterns. The rule used to classify the subjects was then constructed to involve a minimum of arbitrariness and to be as simple as possible. Firstly, rather than arbitrarily selecting some of the predictions during these fifteen patterns, all of these were considered in the rule. Secondly, the ten 216

Page 13: Contradiction and Change of Performance Expectations

had been well established for each subject before contradiction began, and at the same time served to control for the similarity of these initial expecta- tions across subjects.

The results from the remaining forty-three subjects appear in Figures I,

11, and 111, and in Table I. The mean difference between the scores predicted for self and for other for each pattern appear in solid lines in the three figures. It will be noticed that the curve in solid lines is one pattern ahead of the broken-line curve. The reason for this is that the subjects received scores at patterns 1 through 30, while their first prediction was for pattern 2 and their last for pattern 3 1. Table I presents the results in a more concise form. Here the differences between the scores predicted for self and for other during the first fifteen patterns were averaged first over patterns and then over subjects. The same over-all average was also obtained for the second fifteen patterns.

TABLE I

MEAN DIFFERENCES BETWEEN THE SCORES PREDICTED FOR SELF A N D THE SCORES PREDICTED FOR OTHER

First fifteen Last fifteen patterns patterns

Number of Condition subjects Mean s2 Mean s2

Total contradiction 14 -8 .57 1 . 1 3 4.19 9.20 Partial contradiction 15 -8 .55 1.22 -2.89 6.01 Control 14 -8 .77 1.96 -9.20 1.48

The three figures show very similar solid curves for the first fifteen pat- terns. This reflects that the scores received during these patterns had been the same in the three conditions and shows, to a certain extent, the operation of the rule used to classify the subjects. A different result would have pre- vented any comparisons between the three conditions with respect to the last fifteen patterns. In addition, the ordinates of the three solid curves are always negative. This is a good indication that the subjects had indeed formed lower expectations for self than for other and had not simply met the requirements of the rule. It will be recalled that for seven predictions the rule allowed for any positive differences in favour of self consistent with the task, i.e., between zero and thirty. Although exhibiting fluctuations, these three curves have a noticeable downward trend, that is, as the number of (-+) unit evaluations increased, the subjects became more convinced of the other person’s superi- ority at the task. The fluctuations tend to disappear in the last three patterns and the predicted differences stay around the -10 figure, which was the

point difference was the one most frequently received during these patterns. Finally, this difference was required for at least eight or the simple majority of patterns, in consideration of the fact that other differences had also been received. For a de- tailed discussion of the rationale for a similar rule, see Foschi (1970), pp. 49-56. 217

Page 14: Contradiction and Change of Performance Expectations

most frequently received difference in scores. The results in Table I restate the similarity of the three groups: not only are the mean values very close but the variances are low and do not differ markedly. The subjects saw them- selves from 8.55 to 8.77 worse than the other person in the task, a very adequate result considering that the rule used to classify them was not particularly restrictive.

During the lust fifteen patterns of the control condition the solid curve continued at approximately the same level as during the initial fifteen pat- terns, thus conlirming that [-+] expectations had been created. It is interest- ing to notice that although there was an increase after each agreement, this was never sufEcient for all the subjects to change their expectations into [OO]. On the other hand, the two experimental treatments had the expected effect of increasing the values of the predicted differences in line with the received ones. The two solid curves differ in the magnitude of their values but fluctuate in similar ways, reflecting in both of these respects the curves of the received scores. As to the over-all values, the means for the three groups shown in Table I are ordered as expected. In addition, the control treatment had the effect of slightly lowering the initial values of both the mean and the variance, thus not only making the subjects feel even more inferior than the other person but also more uniform in this belief. On the other hand, both experimental groups show marked increases in their variances, the subjects in the total contradiction varying more than those in the partial contradiction. An analysis of the raw data suggested the following interpretation. The expected effect of the experimental treatments was to make the subjects change their expectations, while no change was involved in the control group. In the total contradiction condition some of the sub- jects made this change by holding first the intermediate [OO] type for a few patterns, while others changed directly to [+-I expectations. One subject moved only from [-+I to [OO]. The larger discrepancy between the scores received in each half of the task thus permitted the subjects in this condition to be more diverse in their reactions than those in the partial contradiction. The subjects in the control group did not undergo any changes and accord- ingly have the lowest variance of the three conditions.

A visual inspection of the right-hand side of Figures I and II shows sup- port for the hypotheses and the corollary. Hypothesis 1 predicted that the solid curve should change from negative towards positive values in the total contradiction and towards zero in the partial contradiction, as a function of the number of contradictory unit evaluations, and both figures show these increasing functions. The corollary predicted bounds indicating that the ex- pectations have become of the same type as the unit evaluations. As this can be satisfied by any positive value in the total contradiction case, the differ- ences in scores most frequently received during the last fifteen patterns are used for a more exact specification of the bounds. These differences were 10 in the total contradiction and 0 in the partial contradiction. An upper bound very close to this expected value is clearly shown in Figure 11. In Figure I the 218

Page 15: Contradiction and Change of Performance Expectations

bound is not as definite and its value is not as close to the specified one, sug- gesting that more contradictory unit evaluations are needed in this condition to complete the change. Finally, as predicted by Hypothesis 2, the solid curve does appear to change faster in the total than in the partial contra- diction.

In order to test the hypotheses and the corollary in a more precise way the experimental results were associated with continuous functions. From the trends indicated in the figures an exponential of the form y = a + bk-" was chosen for the regressions. Several values of k were tested and for each of them the values of a and b were computed to obtain the best fit in the least squares sense. For each k and for each condition the coefficient of cor- relation was found. The highest correlations, in the order of 0.90, corre- sponded to k = 1.40. The resulting regression equations and the data they fit are shown in Figure IV. Notice that the differences in scores predicted

0 TOTAL CONTRADICTION CONlllTtOff

0 PARTIAL CONTRADICTION COUOITION

10 1 0

o r

5 -

0 0 0 0

Y = -1.408 - 8.841 (1.40)

0

1 3 5 7 9 11 13 15 X

NUMBER OF PATTERNS DURING THE CONTRAOICTIOM OF EXPECTATIONS

FIGURE N. REGRESSION CURVES AND EXPERIMENTAL RESULTS 219

Page 16: Contradiction and Change of Performance Expectations

220

immediately before the contradiction of expectations began were used for x = 0 in each condition. The derivatives of the regression equations are given by :

dy/& = 18.277 (log. 1.40) ( 1.40) --z

dy/& = 8.841 (log 1.40) (1.40)-"

Hypothesis 1 is now supported by the fact that for any x > 0 the rates of change are positive. For x + CQ the regression equations have an upper bound given by 7.842 in the total contradiction and -1.408 in the partial contra- diction. These values do not differ markedly from those specified earlier in reference to the corollary. Finally, a comparison between the derivatives supports Hypothesis 2, since the rate of change in the total contradiction is twice that of the partial contradiction.

(total contradiction) (partialcontradiction).

CONCLUDING REMARKS

In summary, the results show that when [-+I expectations are contradicted by evaluations made with the use of objective criteria, the expectations change as a function of the number and the degree of the contradiction. As a first test using the proposed theoretical apparatus the present study did not specify the exact form of these functions;13 instead only their general trends were predicted by the hypotheses and the corollary. Both the inspection of the experimental data and the test by means of the regression curves have confirmed these predictions in the case of initial [-+I expectations. The re- sults, in turn, gave indirect support to the conceptualization of expectations as a! sets which had been used to formulate the rationale for the hypotheses. Some of the scores predicted by the subjects also provided support for the conceptualization in a different way. As it will be recalled, they could make their predictions in terms of any numbers between zero and thirty, not neces- sarily in terms of the four possible scores. An examination of the raw data revealed that, at one time or another in the series of patterns, twenty of the forty-three subjects took advantage of this allowed flexibility and gave, as their predictions, scores that they could not actually get, such as 12 for self and 22 for other. If the subjects' expectations had consisted of only one Unit evaluation no such predictions would have been made. Instead, it seems reasonable to interpret these predictions as indicating that a! sets had been formed and as giving the contents of these sets in a summarized way. How- ever, an independent test of the formation of a! sets would be needed to substantiate this interpretation.

The results of the present test justify refining the hypotheses14 to account

13 The regression curves are simply a device to present the data in a way that makes the testing of the hypotheses more precise, and should not be taken for theoretical curves of change of expectations.

14 The results reported in Foschi (1970) also supported the hypotheses. However, they cannot be used as a test of the conceptualization because in this case the sub- jects had been instructed to form (I sets.

Page 17: Contradiction and Change of Performance Expectations

for the change of expectations step by step. Some of the directions in which this modification should be carried out have already been suggested when proposing the hypotheses. The variances of the two experimental conditions during the last fifteen patterns showed another line for investigation. Refined hypotheses should take into account that although the subjects in both con- ditions followed the general trend predicted by Hypothesis 1, the analysis of their individual predictions had shown that those in the total contradiction had more paths available to them in changing their expectations. Finally, a closer look at the curves of the experimental results suggested further direc- tions for research. These curves exhibit noticeable fluctuations throughout the entire series of patterns. Are these fluctuations a feature of the theoretical curves, refI ecting how the information received at each pattern reorganized the composition of the a! set? On the other hand, the subjects appear to have responded very closely to the scores received at the immediately previous pattern. Does this mean that no a sets had been formed and that the expec- tations consisted of only one unit evaluation reflecting the scores just re- ceived? Perhaps it is more justifiable to interpret these fluctuations as indicat- ing that the last unit evaluation should be given more weight than the others when characterizing an a! set. These questions will be examined in detail in

Finally, a few words about possible extensions of the study can be said. future reports in which step-by-step hypotheses will also be proposed. This line of research could be followed to try to understand how beliefs, other than expectations associated with specific abilities, are affected by in- formation that contradicts them. For example, it has frequently been ob- served that prejudices are maintained even when objective evaluations of performances are in contradiction with them. Prejudices can be concep- tualized in terms of the theory by Berger, Cohen and Zelditch (1966) re- lating diffuse status characteristics such as age and sex to general expectation states. This theory, together with an extension of the present study, could then be used to provide an explanation of why prejudices are maintained and how they may be changed.

R E F E R E N C E S

Berger, J., B. P. Cohen, and M. Zelditch, Jr. 1966 “Status characteristics and expectation states.” Pp. 29-46 in J. Berger,

M. Zelditch Jr., and B. Anderson (eds.) , Sociological Theories in Progress. Boston: Houghton Mifflin Co. Vol. 1.

Berger, J., T. L. Conner, and W. L. McKeown 1969 “Evaluations and the formation and maintenance of performance

Berger, J. and J. L. Snell 1961 “A stochastic theory for self-other expectations.” Stanford, California:

expectations.” Human Relations 22 :481-502.

Technical Report no. 1, Laboratory for Social Research, Stanford Univer- sity.

Camilleri, S. F. and J. Berger 1967 “Decision-making and social influence: a model and an experimental test.”

Sociometry 30: 365-378. 22 1

Page 18: Contradiction and Change of Performance Expectations

Foschi, M. 1970 “Contradiction of specific performance expectations: an experimental

Foschi, M. Forthcoming “On the concept of ‘expectations.’ ” Acta Sociologica. Harvey, 0. J. 1953 “An experimental approach to the study of status relations in informal

Heider. F.

study.” Stanford University: unpublished PHD dissertation.

groups.” American Sociological Review 18 : 357-367.

1944 “Social perception and phenomenal causality.” Psychological Review 5 1 : 3 58-3 74.

Lewis, G. H. 1966 “Performances, evaluations, and expectations: an experimental study.”

Moore, J. C., Jr. 1968 “Status and influence in small group interactions.” Sociometry 3 1 :47-63. Moore, J. C., Jr. 1969 “Social status and social influence: process considerations.” Sociometry 32:

Sherif, M., B. J. White, and 0. J. Harvey

Stanford University: unpublished PHD dissertation.

145-1 58.

1955 “Status in experimentally produced groups.” American Journal of Sociology 60:370-379.

Webster, M., Jr. 1969 “Source of evaluations and expectations for performance.” Sociometry 32:

243-258.

222