JOURNAL OF RESEARCH IN Necessary PERSONALITY versus 7, 197-207 Sufficient for Success ANNA (1973‘) KUN University AND and Causal Failure1 BERNARD of California, Schemata WEINER Los Angeles 197 subjects judged the perceived presence or absence of ability or effort, given information about task outcome (success or failure), the difficulty of the task (easy, intermediate, or difficult), and the state of the complementary cause (effort or ability). The data revealed that a multiple sufficient causal schema is used to explain common events. That is, the presence of ability OT effort is perceived as enough to produce success at an easy task, while the absence of ability or effort is perceived as sufficient to result in failure at a difficult task. On the other hand, a multiple necessary schema tends to be employed to explain uncommon events. Success at a difficult task is believed to require both ability and effort, while failure at an easy task tends to be perceived as caused by low ability and low effort. In addition, there were disparities in the causal judgments for success and failure: failure outcomes are more likely to elicit a multiple sufficient schema. Further, in achievement-related contexts ability and effort are perceived as negatively covarying causal determinants of typical successes and failures. In addition to supporting hypotheses from attribution theory, the data shed further light upon the perceived determinants of success and failure, and demonstrate the influence of cognitive strurtures on achievementrelated causal judgments. The most effort, task Reed, Rest ability, high to be primarily is perceived salient perceived causes of success and failure are ability, difficulty, and luck (Heider, 1958; Weiner, Frieze, Kukla, & Rosenbaum, 1971). That is, if one succeeds, then high effort expenditure, an easy task, or good luck are believed responsible for the outcome. as being caused by low ability, In a similar manner, failure lack of effort, a hard task, ‘This research wss supported by grant GS-35216 to the second author from the National Science Foundation. The first author was supported by a Training Grant to the Personality and Social Psychology program at UCLA from the United States Public Health Service, MH11696-04. 197 Copyright @ 1974 by Academic Press, Inc. All rights of reproduction in any form reserved. 198 KCN AND WEINER or bad luck. These judgments are, in part, based upon information such as the person’s past success history, social norms concerning the success of others, observed persistence at the task, and so on (see Frieze & Weiner, 1971). In addition to specific cues, causal preferences influence the judgment process. For example, individuals high in achievement needs are prone to infer that high or low effort is respectively responsible for success and failure (Kukla, 1972; Weiner et al., 1971). Finally, the type of cognitive structure that is used to derive causal judgments also may affect the perceived determinants of success and failure. One such cognitive structure that influences causal judgments is labeled a “causal schema.” This paper specifically examines the role of causal schemata in determining achievement-related causal judgments. A causal schema concerns the beliefs that a person holds about the relationship between an observed event (an effect) and the perceived causes of that event (Kelley, 1972). A cause may be related to an effect in at least one of two distinct ways: it can be necessary to produce the effect, or sufficient to produce the effect. When several causes are perceived as related to the same effect, many different combinations of these necessary and sufficient relations are possible. Here we will consider only two of these cases, namely, when all the causes are perceived as necessary for the effect, and when any one cause is seen as a sufficient determinant of the effect. Consider the simple situation in which two causes (X and Y) are associated with an effect (2). Further, let us distinguish only ,between the presence and the absence of causes and effects. For example, eating or not eating (the observed effects) may be associated with the presence or absence of hunger and/or the presence or absence of attractive food (the perceived causes of eating or not eating). An individual may believe that the presence of either hunger or attractive food will produce eating behavior. This disjunctive set of causal relations, coded linguistically as an ‘(or” relationship, is referred to as a multiple sufficient causal schema. Each cause in and of itself is capable of producing the effect. On the other hand, the person may believe that both hunger and the availability of desirable food are required to produce eating behavior. This conj?snctive set of causal relations, coded as an “and” relationship, is referred to as a multiple necessary causal schema. Both causes are needed to produce the effect (see Kelley, 1972). The significance of causal schemata is that they permit the individual to predict effects from the presence or the absence of certain causes. Furthermore, given an effect, they generate inferences about the underlying causes. That is, a schema permits hypothetico-deductive statements about the causal structure of a situation. For example, if an individual CAUSAL SCHEMATA FOR SUCCESS 199 TABLE 1 AND THE: PRESENCE Multiple OF AN EFFECT (X) sufficient schema present Cause Y absent absent, present Cause S SLlultiple necessary schema present Cause Y absent absent present Cause X observes someone eating, using a multiple necessary schema would lead him to deduce that the person was hungry and that the food was attractive. This inference obviously has functional significance. Necessary versus sufficient schemata generate different attributions and beliefs about the world (see Tables 1 and 2). As already indicated, if the attributor knows that 2 has occurred, then using a multiple sufTABLE 2 INFERENCES ABOUT L4 CAUSE (X OR Y), GIVEN INFORMATION ABOUT EFFECT (2) AND THE STATE OF THE ALTERNATIVE CAUSE, USING MULTIPLE SUFFICIENT AND MULTIPLE Nscessa~~ SCHEMATA Given source of informat,ion Inferred cause Multiple sufficient schema Multiple necessary schema 1. 2. 3. 4. X,ZQ Y,Z 4, Zb Y, 2 Y x I-r x Uncertain ‘, Certain L‘ Certain “ Uncertain <c 5. 6. 7. 8. z, z Y, Z x, 2 Y,Z Y X Y X Certain ,( Incompatible “ Incompatible ‘I Certain L‘ - The cause is present and the effect is present. The PRIIS~ ia ahsent, snd t,he effect, is n.hsent. b AN ficient schema allows him to infer that either X or Y were present. Further, if it is known that X was present when 2 occurred, then there should be uncertainty about the presence of Y, provided that X and Y are independent. As shown in Table 1, 2 occurs given X irrespective of the presence or absence of Y. Thus, for example, if a person eats when hungry, we should be uncertain about the attractiveness of the food if we use a sufficient schema relating eating behavior and the presence of hunger. On the other hand, if a multiple necessary schema is used, then the occurrence of 2 in the presence of X provides unequivocal evidence for the presence of cause Y. As depicted in Table 1, with a necessary schema 2 occurs only when both X and Y occur. Thus, if a necessary schema is used, we infer that the food is attractive if eating behavior is observed. Now consider the situation in which Z has not occurred (e.g., the person has not eaten). Using a multiple necessary schema, the absence of 2 given the absence of X results in uncertainty about the state of Y; 2 cannot occur whether Y is present or absent. Continuing with our eating example, if the person does not eat and is not hungry, we are unsure about the attractiveness of the food if a multiple necessary schema is used. But with a multiple sufficient schema the absence of 2 and X provides unequivocal evidence that Y also must be absent. That is, if an individual does not eat, even though he is not hungry we infer that the food is not attractive. In sum, knowledge about’ the presence of an event in the presence of one cause leads to most uncertainty about the state of the other cause when a multiple sufficient schema is used. But knowledge about the absence of an event in the absence of one of the causes leads to most uncertainty about the state of another cause when a multiple necessary schema is employed. These attributional disparities in judgment certainty as a function of the schema employed and the presence or absence of causesare summarized in Table 2. The upper patterns of data shown in Table 2 can be incorporated into either the multiple necessary or multiple sufficient causal schema. However, t’he degree of certainty concerning the attributional inference varies as a function of the type of schema that is employed. This disparity in certainty permits the identification of an individual’s causal schema by assessingthe confidence he places in his attributions. Thus, certainty of judgment, an “outward criterion,” provides one behavioral index of an individual’s causal perceptions, an “inner process” (see Wittgenstein, 1953). In addition, the four conditions listed in the lower portion of Table 2 show that there are patterns of information incompatible with one . ? . .I 1 .>I. Ll.. -AL-- ^-L--77-m n.,nm,ln thn CAUSAL SCHEMATA FOR SUCCESS 201 occurrence of Z which specifies the absence of X is incompatible with a multiple necessary schema that both X and Y must be present for 2 to occur. Such incompatibility is likely to produce structural .change. On the other hand, the occurrence of 2 in the absence of X is c’ompatible with a multiple sufficient achema. Determinants of Causal Schemata Attribution theorists specify that perceived antecedent-consequent covariation provides the foundation for the attribution process (Heider, 1958; Kelley, 1967). Further, general inductive rules concerning the appropriateness of various causal schemata are likely to be formed on the bases of prior experiences. One such general rule has been identified by Kelley (1972). Kelley proposes that, a multiple sufficient schema is generally employed to account for common events, while unusual or extreme events are likely to elicit a multiple necessary schema to explain their occurrence. For example, considter the extreme event of over- or undereating on a specific occasion. Based upon Kelley’s analysis, if one eats much more than normally, then it may be inferred that he was hungry and that the food was particularly attractive. In a similar manner, if very little food was consumed, then it is likely to be inferred that the person was not hungry and that the food was unattractive. On the other hand, if a typi’cal amount, of food is consumed, which is a usual event, then the presence of only hunger or attractive food is likely to be deduced. Causal Schemata and Achievement Outcomes In the present investigation, Kelley’s suppositions concerning causal schemata were employed to examine the perceived determinants of success and failure. Information concerning success and failure outcomes at tasks varying in perceived level of difficulty was provided to the subjects. It is assumed that success at an easy task is a more common event than success at a difficult task. In a similar manner, it is assumed that failure at a difficult task is perceived as a more common event than failure at an easy task. After receiving the task difficulty and outcome information, the subjects made attributions concerning the certainty of the presence or absence of either high or low ability or high or low effort,, given knowledge about the presence or absence of the other of these two causes. Ability and effort were specified as the two causes because they have been identified as the most salient personal determinants of success and failure (Frieze, 1973; Heider, 1958; Weiner et al., 19711. Following Kelley (1972), it was expected that attributions for common events elicit a mul- 202 KUN AND WEINER tiple sufficient schema. Thus, success at an easy task will be perceived as due to the presence of high ability or high effort. In a similar manner, failure at a difficult task will be due to the presence of low ability or low effort. On the other hand, success at a difficult task and failure at an easy task are likely to elicit a multiple necessary causal schema to explain their occurrence, for these are unusual events. The presence of high ability and high effort for success, and low ability and low effort for failure, were expected to be perceived as the causes of these unusual events. In sum, given a success, it was hypothesized that as task difficulty increases the use of a necessary schema indicating the presence of multiple causes will increase. In addition, given a failure, as task difficulty increases the use of a sufficient schema indicating the presence of only one cause will increase (see Kcpka & Brickman, 1971; Lanzetta & Hannah, 1969; and Weiner & Kukla, 1970, for indirect confirmation of these hypotheses) . It also has been suggested that there are differences in the analysis of positive and negative events (Kanouse & Hanson, 1971). If this supposition is correct, then the judgment data in the success and failure conditions will not be equivalent. METHOD Subjects were 197 male and female students enrolled in an upper division psychology course at UCLA. College students are an especially good population for the present investigation, for achievement performances are frequent and salient occurrences in their lives. Thus, students are likely to have well-articulated causal schemata for success and failure. Each subject. received a questionnaire with instructions to imagine that, he or she is a high school teacher, making judgments about the contributions of ability (A) and effort (E) to exam performance. The subjects were given information about the difficult,y of an exam, a hypothetical student’s performance at that exam, and the state (high or low) of one of the two persona1 causes of success and failure (.4 or E). The subjects’ task was to infer whether the complementary cause (E or A) contributed to the student’s exam performance. Task difficulty was conveyed by social norms indicating the percentage of students (lo%, 50%, or 99%) who scored excellent (success) on the exam. Excellent. satisfactory, or failure were denoted as the possible exam outcomes. The exam performance of the hypothetical student was either excellent or failure, and the state of the one given cause was consistent with the outcome (high A or high E given success, low A or low E given failure). Thus, the possible incompatible situations outlined in the lower portion of Table 2 were not examined in this experiment. There were 12 experimental conditions (3 levels of task difficulty X 2 levels of outcome X 2 types of causes). Each subject made judgments concerning the importance of the unstated cause in all 12 conditions. More specifically, each situation presented to the subjects included data about one personal cause (A or E) and the difficulty level of the task. The subjects were asked to rate whether the other personal cause (E or A) may “have been an CAUSAL SCHEMATA FOR SUCCESS 203 important, determinant of the outcome.” For example, the subjects were informed that 10% of the pupils performed excellently on an exam. The pupil under consideration failed, and was described as low in ability. The subjects then had to judge whether low effort also was an important determinant of the outcome. In this example it was expected that the subjects would be unsure about the effort expenditure. Failure at a difficult task is a usual event, and should elicit a sufficient schema. Low ability is a sufficient cause of the failure. Thus, there should be uncertainty concerning the presence of low effort. Subjects marked their judgment anywhere on a line that had five equidistant labels as anchor points. The five labels were “Yes for sure,” “Probably yes,” “Maybe yes, maybe no, ” “Probably no,” and “No for sure.” For scoring purposes it was decided a priori to divide the scale into 17 equal intervals that were correspondingly assigned scores from +8 to -8. The zero point represented greatest uncertainty (“Maybe yes, maybe no”) concerning the importance of the judged cause. To ease the judgmental process, the 12 experimental conditions were presented on three separate pages. Each page contained information about one of the three task difficulty levels. Four judgments were made on each page (2 levels of outcome x 2 kinds of causes). The sequential order of task difficulty levels was completely permuted and the order of the four conditions within each task difficulty level was partially permuted, yielding 24 different judgment orders. The orders were randomly distributed to the subjects. RESULTS AND DISCUSSION The mean certainty judgments in the 12 experimental conditions are shown in Figure 1. Examining the extreme task difficulty levels, the figure Easy I l”+ermed,O+e I Difficult TASK DIFFICULTY FIG. 1. Mean certainty judgments of high ability and high effort (Success condition) and low ability and low effort (Failure condition) given information concerning the complementary cause (effort or ability), task outcome, and the difficulty of the task. The labels indicate the cause being judged. 204 KUN AND WEINER shows that given success at the most difficult task, which is an uncommon event, and the presence of high ability (or high effort), the belief is expressed that high effort (or high ability) also was present. Thus, a multiple necessary schema is used. Conversely, given success at the easiest task, which is a common event, subjects infer that the complementary cause was not present. That is, a sufficient schema is employed. Turning to the failure condition, given failure at the most difficult task, which is a common event, and low ability (or low effort), subjects express the belief that low effort (or low ability) was not a cause. Thus, a sufficient schema is used. And given failure at the easiest task, an uncommon event, there is a slight tendency to infer the presence of both low ability and low effort. In sum, when performance outcomes are un-ommon (success at difficult tasks and failure at easy tasks), attributions tend to be made to multiple causes; when performance outcomes are common (success at easy tasks and failure at difficult tasks), attributions are made to only one cause. Summarizing the data in Figure 1 so that the intermediate difficulty level judgments also are included, it can be stated that, given success,the likelihood of employing a multiple necessary rather than a multiple sufficient schema inferring the presence of high ability and high effort increases as the task becomes more difficult. In addition, given a failure, the likelihood of using a multiple necessary rather than a multiple sufficient schema inferring the presence of low ability and low effort decreases as the task becomes more difficult. Thus, as predicted, there is a Task Difficulty X Outcome interaction (F(2,392) = 220.27, p < ,001). Although the covariation of effort attributions with the task difficulty norms is greater than the corresponding ability ascription covariation, the main effect of the type of cause does not approach significance, F < 1. The task difficulty effect is highly significant (p < .OOl) in all four of the functions shown in Figure 1. SuccessVersus Failure The Task Difficulty X Outcome interaction was decomposed into simple effects for successand for failure. The main effects of task difficulty for both success and failure are highly significant (respectively, F(2,232) = 272.96, p < .OOl; F(2,232) = 52.39, p < .OOl). But it is evident that social norms are more powerful in producing differential attributional patterns for successthan for failure. The disparity in judgments as a function of outcome also is revealed in a significant Outcome main effect (F(1,196) = 71.23, p < .OOl). The mean certainty judgments are greater in the successthan in the failure condition. Further, Figure 1 shows that over an extended range of task difficulty levels both ability CAUSAL SCHEMATA FOR SUCCESS 205 and effort are believed to be necessary to attain success. This is evident in the figure because both the intermediate and difficult task causal judgments are above the uncertainty point in the success condition. On the other hand, over an extended range of task difficulty levels either low ability or low effort is believed to be sufficient to produce a failure. This is observed in the figure because both the intermediate and difficult task causal judgments are below the uncertainty point, thus indicating an absence of the complementary causal factor. Success Condition As already indicated, the results for the success condition support the broad hypotheses tested concerning the linkages between the type of causal schema employed and the commonness of an event. But further examination of mean certainty judgments reveals an unanticipated finding. Given success at an easy task and the presence of one cause, subjects infer the absence of the other cause. That is, the subjects feel somewhat certain that the complementary factor was not an important determinant of the outcome. Both the ability and the effort judgments are significantly less than the zero or uncertainty point (respectively, t = 3.20, df = 196, p < ,001; t = 9.18, df = 196, p < .OOl). These data are consistent with the hypothesis that multiple sufficient schemata are employed to account for common events. However, in the introduction section of this paper and in Table 2 it was deduced that subjects would be uncertain about the state of the complementary cause given a common event and the presence of the alternative independent cause. The likely explanation for the inferred absence of ability or effort given success at an easy task and the presence of the complementary cause is that ability and effort are not perceived as independent, but rather subjectively covary negatively for any particular positive outcome. If one works hard at an easy task, then he or she is not likely to be perceived as able. And if one has ability, then he or she evidently need not expend effort to succeed at an easy task (see Heider, 1958; Weiner et al., 1971). As Heider (1958) notes, exertion is inversely related to power. In the derivations shown in Table 2 it was assumed that X and Y are independent causes. If X and Y are indeed independent, then an individual should have no increment in knowledge about cause Y when given information that sufficient cause X is present and that the outcome is successful. But if X and Y are negatively related, then information about the presence of one cause also provides evidence concerning the absence of the other cause. Dependent causes can be readily incorporated into the general framework provided by Kelley (1972)) although the derivations shown in Table 206 KUN AND WEINER 2 would be somewhat altered. Rather than being uncertain about the presence of one cause given success at an easy task and the complementary cause, the individual should feel somewhat certain that the other cause was absent. The degree of cert’ainty would, of course, be a function of the perceived magnitude of the negative covariation. Failure Condition Examination of the judgments for failure reveals additional complexities. If one fails at a hard task and lacks ability (or effort), then the belief is expressed that low effort (or low ability) is not an important causal factor (t(ability) = 6.72, df = 196, p < .OOl; t(effort) = 9.34, df = 196, p < .OOl) . These findings are consistent with the logically equivalent causal judgments for success at an easy task, and support the hypothesis that a multiple sufficient schema is employed to account for a common event. However, the data again are not in accord with our expectations that subjects would be uncertain about the complementary cause in this situation. A perceived interdependence between effort and ability also can explain these results, although the interpretation is not as evident in the failure as in the success condition. Apparently, subjects believe that only an able person would not exert effort at a difficult task, and that one tries particularly hard at a difficult task if he or she is low in ability. Turning attention now to the judgments pertaining to the easy task in the failure condition, it can be seen in Figure 1 that the reported certainty levels are quite low. Given failure and low ability (or low effort) as one causal factor, the subjects do not firmly believe that the judged pupil was also low in effort (or low in ability). The ascriptions concerning low effort are significantly greater than zero (t = 2.95, df = 196, p < .OOl) , and provide some evidence that at least the rudiments of a multiple necessary schema are being used. But the judged certainty of low ability given failure at an easy task and low effort is not significantly greater than zero (t < 1). Further, the magnitudes of the judgments are far below the logically equivalent uncommon event judgments in the success condition given a difficult task. There are a number of plausible psychological interpretations of the data in the failure condit’ion showing that multiple necessary schemata are, at best, only weakly employed, and that multiple sufficient schemata dominate the judgments for failure. There is a large amount of evidence demonstrating that failure elicits attributions to external causes, such as bad luck or the difficulty of the task (see Frieze & Weiner, 1971; Weiner & Kukla, 1970). Thus, individuals apparently have generalized defensive attributions about failure and are loathe to infer that internal CAUSAL SCHEMATA FOR SUCCESS 207 factors such as low ability or low effort are responsible for unsuccessful performance. Another possible interpretation of the success-failure differences derives from a suggestion by Kanouse (1971). Kanouse contends that negative instances convey more information than positive instances. For example, a number of positive attributes are needed to form a positive impression, while a single negative attribute is sufficient to undermine a positive attitude. In a simiIar manner, our data indicate that in achievement contexts the absence of ability or the absence of effort (negative instances) are perceived as sufficient to explain most failures. And finally, there is suggestive evidence from the concept attainment literature that conjunctive rules (multiple necessary schemata) are difficult to employ with negative instances (see Bourne, 1967). Thus, there are both motivational and informational interpretations that might account for the disparate use of multiple necessary schemata in the success and failure conditions. REFERENCES L. E., JR. Learning and utilization of conceptual rules. In B. Kleinmutz (Ed.) Concepts in the structure of memory. New York: Wiley, 1967. Pp. l-32. FRIEZE, I. Studies of information processing and the attributional process. Unpublished doctoral dissertation, University of California, Los Angeles, 1973. FRIEZE, I., & WEINER, B. Cue utilization and attributional judgments for success and failure. Journal of Personality, 1971, 39, 591-605. HEIDER, F. The psychology of interpersonal relations. New York: Wiley, 1958. KANOUSE, D. Language, label&g, and attribution. New York: General Learning Press, 1971. KANOUSE, D., & HANSON, R., JR. Negativity biases in the attribution of value. New York: General Learning Press, 1971. KELLEY, H. H. Attribution theory in social psychology. In D. Levine (Ed.), Nebraska symposium on motivation. Lincoln: University of Nebraska Press, 1967, 192-237. KELLET. H. H. Causal schemata and the attribution process. New York: General Learning Press, 1972. KEPJC~, E. J., & BRICKMAN, P. Consistency versus discrepancy in the attribution of intelligence and motivation. Journal of Personality and Social Psychology, 1971, 20, 223-229. KUKLA. A. Attributional determinants of achievement-related behavior. Journal of Personality and Social Psychology, 1972, 21, 166-174. LANZETTA, J. T., & HANNAH, T. E. Reinforcing behavior of “naive” trainers. Journal of Personality and Social Psychology, 1969, 11, 245-252. WEINER, B., 8: KUKLA, A. An attributional analysis of achievement motivation. Journal of Personality and Social Psychology. 1970, 15, l-20. WEINER. B., FRIEZE, I., KUKLA, A., REED, L., REST, S., & ROSENBAUM, R. M. Perceiving the causes of success and failure. New York: General Learning Press, 1971. WITTGENSTEIN, L. Philosophical investigations. Oxford, England : BIackwell, 1953. BOLRNE,