REACTIONS TO VICTIMS UNDER CONDITIONS OF SITUATIONAL DETACHMENT: THE EFFECTS OF RESPONSIBILITY, SEVERITY, AND EXPECTED FUTURE INTERACTION

The research of Learner and others suggests that people tend to derogate innocent victims. A crucial assumption underlying Lerner's approach is that observers, unable to ascribe some misdeed to a victim, will attempt to con- vince themselves that the victim deserved his suffering by attributing personal unworthiness to him. The analysis developed in the present study, however, suggests certain circumstances under which innocent victims would not be devalued by strangers. The experimental design holds constant the observer's situational removal from the victim's misfortune and thereby facilitates a separation and assessment of factors that have been interlocked in previous research. Subjects learn about the victim's problem through examination of a clinical data folder; they either do, or do not, expect to meet the victim at a subsequent experimental session. The victim is portrayed as either innocent or responsible for her misfortune, and her suffering is described as either severe or mild. Consistent with predictions, innocent victims were not derogated vis-a-vis responsible ones; victims of severe suffering were evaluated less favorably than those of mild misfortune; and subjects who expected to meet the victim rated her more favorably than those who did not anticipate future interaction with her.

former, the observer will tend to impute the latter to the victim. According to Lerner, this phenomenon explains the rejection of innocent victims by detached observers.
The most direct evidence in support of the above assumption stems from an experiment by Lerner and Matthews (1967) in which a pair of female subjects faced the prospect of one of them having to perform a learning task under negative reinforcement while the other performed it under positive reinforcement. Subjects were told that they would determine their respective reinforcement conditions by choosing from a bowl containing two slips of paper, one labeled "reward" and the other labeled "shock." It was found that when subjects drew a reward slip first and thereby assigned their partner to the shock condition, they evaluated her less favorably than when the partner chose a shock slip first and thereby assigned herself to the undesirable condition. This finding was interpreted as being consistent with the just-outcome hypothesis since derogation of the victimized partner by subjects in the self-picks-first condition appeared to reflect their need to 199 justify the innocent victim's misfortune in terms of her personal unworthiness, so as to maintain the notion of a just world.
It is plausible, however, that subjects' devaluation of the victim resulted from certain pressures inherent in the experimental situation, rather than their need to believe in a just world. In the Lerner and Matthews format, subjects are situationally implicated in the victim's misfortune. More specifically, they participate in an experiment that imposes a competitive, zero-sum relationship upon themslves and their partner. It is conceivable that the subjects' direct participation in a system administering negative reinforcements to their partner and relatively favorable outcomes to themselves may have engendered a tendency on their part to view the system as just and its victim as unworthy. Moreover, this attributional tendency would have been especially strong in the self-picksfirst condition where subjects felt that they had personally contributed to the victim's misfortune and to their own good fortune.
Given circumstances, though, in which observers are situationally removed from a victim's suffering and their investment in the punishing system is low, it is quite probable that their reaction to an innocent victim (i.e., someone who has not caused his own problems) will be sympathetic rather than derogatory. It is our contention, then, that the absence of an observable causal connection between a victim and his misfortune will not necessarily prompt observers to attribute personal unworthiness to the victim as an explanation for his suffering. Other attributions on the part of observers, regarding the victim and his misfortune, are possible and in fact predictable, given that they are judging the situation from a relatively objective perspective.
In the present experiment, observers remain situationally removed from the victim's suffering. Hence, it is less likely that certain factors (e.g., competitive feelings) will operate to distort their evaluation of the victim and the circumstances surrounding her misfortune. The experimental design incorporates a direct manipulation of victim's responsibility for her misfortune, and it is predicted that subjects will not derogate innocent victims, vis-a-vis responsible ones. This prediction stems from equity theory (Adams, 1963(Adams, , 1965 which contends that people operate on the basis of commonly held notions of justice and tend to recognize and resist conditions of inequity. Given that an innocent victim has suffered, observers will perceive an inequitable discrepancy between the victim's actual outcomes and those that he deserves since the unfortunate person did nothing to merit his unpleasant fate. Hence, rather than reject the victim in order to convince themselves that an injustice does not exist, detached observers, viewing the situation from an objective perspective, will be inclined to respond sympathetically and favorably toward the suffering person. Moreover, it is assumed that detached observers will tend to be more critical of victims who caused their misfortune through some misdeed of their own, especially if their inappropriate behavior suggests some underlying characterological flaw. Thus, between conditions, innocent victims should be rated more favorably than responsible ones. In addition to perceived responsibility, two other factors likely to influence observers' attributions about victims are manipulated in the present study in order to examine their effects under conditions of situational detachment. One factor pertains to an important aspect of victimization, namely, the severity of the victim's misfortune. According to the notion of self-protective attribution (Shaver, 1970;Walster, 1966), observers who are potentially vulnerable to the victim's fate tend to assign responsibility for serious misfortune to the victim rather than attribute the unfortunate circumstances to chance. This attributional tendency enables the observers to avoid the threatening implication that they, too, could experience a similar misfortune due to chance alone. Though perceptual distortion of the victim's innocence would be more likely in situations of high rather than low outcomeseverity, defensive attribution theory does not predict that attributed responsibility will necessarily lead to devaluation of the victim.
In this experiment, though, the assumption that outcome-severity would elicit a tendency to view the victim as responsible for his problems led to the prediction that victims of severe suffering would be rated less favorably than those of mild misfortune. This expecta-tion is in line with the prediction regarding responsibility, that innocent victims will be rated more attractively than careless ones. Thus, it is anticipated that high outcomeseverity will heighten the observers' tendency to view the victim as the cause of his misfortune and to evaluate him unfavorably, even when evidence is provided that the victim was innocent.
The third factor examined in the present study relates to the subjects' anticipation of future interaction with the victim. Several studies have shown that the expectation of being in contact with a person leads to an increase in that person's perceived attractiveness (Berscheid, Boye, & Darley, 1968;Darley & Berscheid, 1967;Davis & Jones, 1960;Mirels & Mills, 1964). Furthermore, Schopler and Stokols (1970), using a modified version of Lerner and Matthews' experimental format, serendipitously observed that subjects in the self-picks-first condition rated the victim significantly more favorably than did subjects in the other-picks-first condition. It was later shown that this reversal of Lerner and Matthews' results occurred only when subjects expected to occupy a common room with their partner throughout the learning period and test phase of the experiment. The implication of this finding was that the mere expectation of meeting someone who has suffered may promote an empathetic and uncritical orientation toward that person, even in situations where the observer realizes that he has personally harmed the victim.
The factor of expected future interaction, then, is included in the present design in order to examine the generalizability of Schopler and Stokols' findings, as well as those of the previous studies, to situations in which the observer is removed from the victim's suffering. On the basis of balance theory (Heider, 1958), it is expected that the victim's attractiveness rating will be higher in those conditions in which observers expect to interact with him than in comparable situations in which this expectation is absent. That is, the anticipation of a unit relation will evoke a positive sentiment relation.
Although observers in this experiment are spatially and temporally removed from the victim's suffering, it is plausible that their sentiments toward the reinforcement-administering agents will exert some influence on their evaluation of the victim. To the extent that the observer is aligned with the system and perceives it to be just, he will tend to rate its victims critically. An initially critical view of the punishing agents, though, will promote a favorable orientation toward the victim. These assumptions are supportable from a balance theory perspective.
In order to control for the effect of the observer's sentiments toward the system, two variables which indirectly express this relationship are assessed. Since the victim is portrayed as someone who has violated one of society's norms, the observer's opinion about this norm is assumed to be an index of his relative support for either the system or the victim. As a further indication of his alignment, the observer's perceived similarity to the victim is measured. It is expected that both the observer's opinion about the norm and his perceived similarity to the victim will be highly correlated with his evaluation of the unfortunate person.
As an experimental framework for examining the reactions of people to victims under conditions of situational detachment, a variation of the paradigm developed by Darley and Berscheid (1967) was employed. The manipulation of experimental variables was accomplished through the subject's examination of a folder containing clinical data relating to the misfortune of a victim. The subject either did, or did not, expect to participate in a future discussion session with the person described in the folder.
In relation to previous research, there are several advantages to the present format. Most important, the subject's evaluation of the victim is more completely separated from his reaction to the experiment since the victim's misfortune (a premarital pregnancy), described in the data folder, occurred totally outside the experimental situation. This feature permits a separation of the experimental treatments from factors related to an observer's situational involvement.
Furthermore, providing a written account of the victim's ordeal facilitates an examination of the observers' reactions to certain circumstances surrounding the victim's suffering.
More specifically, subjects' sentiments toward the persons or systems (e.g., family and society) administering punishment to the victim, as well as their perceived similarity to her, can be measured. This permits a more complete understanding of the subjects' acceptance or rejection of the victim in terms of their reactions to the reinforcement-generating agents.

METHOD Subjects
One hundred and twenty-eight female undergraduates at the University of North Carolina participated in an experiment concerning "female liberation and changing sexual mores" as part of the expected requirements for an introductory psychology course. The sign-up sheet on which subjects registered for participation indicated that the experiment consisted of two separate sessions, an initial introductory session of approximately 20 to 25 minutes and a final discussion session lasting for half an hour. Subjects registered for the initial meeting and during that meeting were told that they would be able to arrange an appointment for the discussion session at a time convenient to them, if they wished to continue their participation in the experiment after the introductory session.

Procedure
Subjects arrived at the experiment in groups of four to eight. Each subject, upon her arrival at the introductory session, was ushered into one of eight separate experimental rooms. These rooms opened onto a larger central area from which the experimenter read a set of standard instructions and distributed experimental forms. On the desk of each room was a xeroxed copy of a letter, allegedly sent by the Guidance and Testing Center, that briefly explained the purpose of the present study. Subjects were instructed to take note of the letter.
The experimenter introduced himself as a departmental research assistant. He explained that the university's student counseling service had recently initiated a program of research designed to obtain normative information on the attitudes of college women toward female liberation and changing sexual mores. This research was being conducted in conjunction with the department of psychology, which was assisting by gathering data pertaining to these issues and by providing data that had been collected at an earlier time but were relevant to the issues currently under study.
The experiment further explained that it had been decided that the information could be gathered most efficiently by tape-recording two-person discussion groups in which each student would discuss her attitudes with another student. Subjects were told that if they agreed to participate in the discussion session, they would be randomly paired with another girl to form a discussion group. It was emphasized that they would not be paired with anyone attending the current meeting and that their partner might or might not be enrolled in the psychology course that they were taking.
The experimenter then described the discussion sessions, which were portrayed as involving an intimate association with the discussion partner. Subjects were informed that discussions would take place in an experimental room, equipped with two microphones and a tape recorder. The subjects would remain isolated from other persons during the entire session and would be expected to discuss specific topics listed on a piece of paper that they would be given just before the discussion began. The closeness of association between discussion partners was further stressed by telling the subjects that the counseling service and the department of psychology were firmly committed to maintaining the anonymity of subjects participating in the discussions and the complete confidentiality of any information disclosed by the participants. Furthermore, subjects were told that if they wished to participate in the discussions, they would be required to sign a "Confidentiality Guarantee Form." At that point, it was explained that the psychology department, in addition to assisting the student counseling service, hoped to obtain information from these sessions concerning processes of interpersonal judgment. In order to achieve this purpose, the department of clinical psychology had provided data that were obtained in a previous study from a random selection of subjects. Several of the persons who took part in that study had agreed to participate in the current experiment as well. These files contained personality profiles and interview reports written by clinical psychologists.
Subjects were told that if they agreed to participate in the discussion sessions, they would be given, at the present meeting, a data folder pertaining to an unidentified girl participating in this study. They would be asked to complete an impression-formation questionnaire based upon the information contained in the folder. It was emphasized that none of the girls who were described in the data folders were present at the current session, but that some of these girls would serve as discussion partners for subjects who were evaluating them at the present meeting. If the girl described in the folder was to be the partner of the person evaluating her, this would be indicated on a card contained in the folder; otherwise, the subject would be paired with a different girl. In either case, the identity of the prospective partner would be kept strictly anonymous until the time of the discussion session.
Subjects were subsequently requested to decide whether or not they wished to participate in the discussion sessions. Each participant remaining in the experiment was given a "Preliminary Questionnaire" that contained items designed to assess the subject's opinions on various issues, for example, premarital sex and job discrimination against women. Upon completion of the questionnaire, each subject was provided with a data folder containing information about an unidentified girl. Before opening the folders, subjects received a Confidentiality Guarantee Form which they read carefully and signed. After signing this form, they were asked to read through the material contained in the folder, paying particular attention to the interview report and less attention to the more technical data of the enclosed psychological testing forms.

Experimental Manipulations
The experiment used a 2X2X2 factorial design with two levels of expected future interaction between subject and victim, two levels of victim's responsibility for her misfortune, and two levels of victim's outcome-severity. The future-interaction manipulation was accomplished by informing the subject that the person whose data folder she was observing either would, or would not, be her partner for the discussion session. This information was indicated on a card contained in the data file and was read by the subject upon opening the folder. The responsibility and severity manipulations were woven into the interview report within each folder. The psychologist's summary stated that the data obtained from the girl described in the folder was collected as part of an experiment conducted by the department of clinical psychology. The purpose of that experiment had been to investigate the reactions of subjects to various methods of clinical assessment, particularly those dealing with the discussion of traumatic life events. During an interview session, the psychologist learned that the interviewee had experienced a premarital pregnancy during her last year of high school. The report further explained that the girl's pregnancy eventuated in a miscarriage during the summer preceding her enrollment in college.
In the "not-responsible" condition, the interview report indicated that the girl's pregnancy was the result of her being raped by a boy she had been dating for less than a week. In the "responsible" condition, the report indicated that the girl's pregnancy resulted from her carelessness in using contraceptive methods. In order to avoid a possible confounding of the responsibility manipulation with a difference in attitude toward premarital sex (between the victims in the two responsibility conditions), evidence was provided that neither the victim of carelessness nor the one of coercion was opposed to premarital sex as long as the persons involved were deeply concerned with each other's welfare.
For each level of responsibility, there were "high" and "low" conditions of outcome-severity. In the high-severity condition, the report indicated that the girl experienced much physical discomfort during the course of her pregnancy, which led to her frequent absence from school. Furthermore, she became a target of rumors and scandal that sprung up as a consequence of her pregnancy. To make matters worse, her miscarriage, occurring late in the sixth month of pregnancy, was fraught with medical com-plications and required prolonged hospitalization. In the low-severity condition, though, the interview report indicated that the victim's pregnancy eventuated in a medically uncomplicated miscarriage after she had been pregnant for just six weeks. Moreover, the subject was informed that only the girl's closest friends and advisers ever learned about the pregnancy and that they all kept the matter quite confidential.

Measures
After all subjects had examined the contents of their folders, the experimenter explained that the girls whose data files were being used in the present study shared certain common attributes. First, they were all unmarried, female undergraduates in a local university. Furthermore, in the earlier study it was learned that each of the girls had experienced some type of medical or emotional problem prior to her enrollment in college. The data files obtained from these girls, therefore, were particularly useful for the purposes of the present experiment since they provided an opportunity to study the ways in which people form impressions about an individual based upon knowledge of his past or present problems. In order to gain information pertaining to this type of impression formation, subjects would be asked to complete a questionnaire concerning the circumstances surrounding the problem of the girl whose folder they examined.
At this point, each subject received the "Participant Reaction Inventory" which included checks on the experimental manipulations. The items contained in this questionnaire also assessed the subject's sympathy toward the victim, the degree to which the subject believed that the victim deserved her misfortune, and the subject's perceived adequacy of the victim's family in minimizing her suffering.
After completing the reaction inventory, each participant was asked to fill out an "Impression Formation Questionnaire." This form required that the subject rate the girl whose folder she observed on eight dimensions of personality (e.g., sincere-insincere, warm-cold). She was also asked to indicate how probable it was that she would like the girl and the extent to which she would want to work with her on a task. Another item assessed the subject's perceived similarity to the girl described in the folder in relation to political beliefs. Finally, the subject was asked to rate the girl in terms of how "interesting" and "unlucky" a person she appeared to be, based upon the information provided in the folder.
After all subjects had completed this questionnaire, the true nature of the experiment was revealed.

Manipulation Checks
A semantic differential item, as well as an open-ended question, were utilized to measure the success of the responsibility manipulation. The first item required that the subjects indicate, on a 7-point scale, the degree to which they perceived the girl to be responsible for her problems. A 3-way analysis of variance on this item (see Table 1) revealed that subjects in the responsible condition attributed significantly more responsibility to the victim than did subjects in the not-responsible condition (F = 190.31, df -1/120, p < .001). The second question asked subjects to indicate who they believed to be primarily responsible for the girl's misfortune. Three alternative answers were provided: the girl, her family, or some other person. If subjects chose the latter response, they were requested to identify or describe the person they believed to be primarily responsible. Virtually all subjects in the responsible condition attributed primary responsibility to the girl, while subjects in the not-responsible condition identified the rapist as the person most accountable for the girl's suffering.
In order to examine the impact of the severity manipulation, two semantic differential items were employed. Subjects were first asked to rate the unpleasantness (general severity) of the girl's problem on a 7-point scale. They were then asked to rate the severity of the consequences experienced by the girl, on a 7-point scale, in contrast to how unpleasant the consequences could have been for the type of problem she experienced (specific severity). A 3-way analysis of variance was performed for each item. The first analysis revealed that high-severity subjects perceived the girl's problem to be significantly more unpleasant for her than did lowseverity subjects (F= 10.72, df = 1/120, p < .001). Similarly, the second analysis disclosed that subjects in the high-severity condition perceived the consequences of the premarital pregnancy to be maximally unpleasant for the girl, while those in the low-severity condition perceived them to be minimally unpleasant for her (F = 186.64, df= 1/120, p < .001).
Finally, to assess the effectiveness of the future-interaction manipulation, subjects were asked to indicate whether or not the girl whose folder they observed would be their partner for the discussion session. Virtually all subjects in the expected-future-interaction condition responded that the girl would be their partner, while those in the no-expectedfuture-contact condition indicated that she would not be their partner.

Effects of the Experimental Manipulations
Responsibility. From Table 2, it is apparent that the predicted main effect for respon- sibility did not occur. Although the difference between the mean attraction ratings for the responsible and not-responsible conditions approached significance in the predicted direction (F=2.67, dj =1/120, p< .105), the evaluations of the innocent victim were not significantly more favorable than those of the careless one. Two measures dealing with subjects' sympathy toward the victim (see Tables  3 and 4), however, were significantly affected by the responsibility manipulation. Subjects in the responsible condition viewed the victim as more deserving of her misfortune (F= 37.88, dj = 1/120, p < .001) and as a less unlucky person (F = 17.28, dj = \J2Q, p < .001) than did subjects in the notresponsible condition.
Severity. Analyses revealed a significant main effect for severity on two of the items concerning victim's attractiveness. The victim was rated more attractively (see Table 2) when her outcomes had been mild than when they had been severe (F = 7.20, df = 1/120, p < .008). Also, the probability that subjects would like the victim was greater in the lowseverity condition, than in the high-severity condition (F = 6.22, dj = 1/120, p < .014).
A severity main effect appeared on three other items. The victim's family was perceived to have been less adequate in mini-  mizing her suffering by subjects in the highseverity condition than by those in the lowseverity condition (F -10.66, df = 1/120, p < .001). Furthermore, high-severity victims were viewed as significantly more unlucky (see Table 4) than low-severity victims (F = 28.56, df = 1/120, p < .001). Finally, subjects in the low-severity condition perceived themselves to be more similar to the victim, in terms of political beliefs, than did subjects in the high-severity condition (F = 9.30, df -1/120, p < .003). Expected future interaction. All dependent measures involving the subjects' rating of the victim's attractiveness were significantly affected by the manipulation of expected future interaction. As predicted, subjects who expected to be the discussion partner of the victim rated her as significantly more attractive (see Table 2) than did those who did not hold that expectation (F -6.27, dj = 1/120, p < .014). Also, they were more apt to like the victim, in general (F = 4.67, df= 1/120, p < .033), and to enjoy working with her on a task (F = 24.16, df = 1/120, p < .001). Finally, subjects in the expected future-interaction condition viewed the victim as a more interesting person (F = 17.06, df = 1/120, p < .001) than did subjects in the no-expected-future-interaction condition.
Only one other dependent variable reflected an expected-future-interaction main effect. Subjects who expected to be the victim's discussion partner perceived her family to have been significantly more adequate in minimizing her suffering than did subjects who did not expect to be her partner (F = 4.74, dj-1/120, p < .031). Alignment with the system: Attitude toward premarital sex and perceived similarity to the -victim. It was expected that subjects' attitudes toward premarital sex as well as perceived similarity to the victim would be positively correlated with their evaluation of the victim and would exert an influence on their attraction ratings above and beyond the effects of the experimental manipulations. Hence, it was predicted that, when attitude toward premarital sex and perceived similarity to the victim were used as covariates in analyzing the attraction scores, the effects of the experimental manipulations would become more robust.
An examination of the within-cells correlations among the dependent variables revealed that the attraction scores were significantly correlated with perceived similarity to the victim, in terms of political beliefs (r-.297, df= 120, p < .005). No significant correlation between subjects' evaluation of the victim's attractiveness and their attitudes toward premarital sex was found. Two separate analyses of covariance were performed. In the first analysis, attitude toward premarital sex was employed as the covariate, while in the second, perceived similarity to the victim in terms of political beliefs served as the covariate. For all analyses, the items concerning victim's attractiveness, probability of liking her, and willingness to work with her on a task were used as the criterion measures. The analyses of covariance disclosed that the effects of the experimental treatments on the attraction scores were not changed substantially by removing the variance attributable to either attitude toward premarital sex or perceived similarity to the victim.

DISCUSSION
The manipulation checks provide ample evidence that the experimental treatments were effective. The results concerning the per-ceived causal relation between the victim and her problems lend greater support to the predictions derived from equity theory than those based upon Lerner's just-outcome hypothesis. First, careless victims were perceived as significantly more deserving of their misfortune than innocent ones. Second, innocent victims were rated as significantly unluckier than careless ones. These two findings are interpretable in terms of equity theory which suggests that subjects would react more sympathetically to an innocent victim than a careless one since the former's suffering would be attributed to circumstances beyond her control (e.g., "bad luck"), while the latter's misfortune would be viewed as the result of inappropriate behavior. Hence, the innocent victim would be perceived as less deserving of her misfortune than the careless victim.
Because subjects would tend to be sympathetic toward an innocent person and critical of a careless one, it was further hypothesized that the former would be evaluated more attractively than the latter. The results indicated that the evaluations of the innocent victim were not significantly more favorable than those of the careless one. The difference between the mean attraction ratings for the two types of victims, however, approached significance in the predicted direction (p < .105). While this finding does not offer clear-cut support for the prediction that innocent victims would be judged more favorably, it clearly represents evidence that contradicts the prediction, derived from Lerner's just-outcome hypothesis, that innocent victims would be devalued. It appears, then, that when innocent victims are evaluated by situationally detached observers, they are perceived to be at least as attractive as careless victims, if not more so.
The above results are somewhat qualified, though, by certain features of our experimental setting. 3 As stated earlier, the crucial assumption of Lerner's hypothesis is that observers will attribute personal deficiencies to a victim in the absence of some evidence that he was the behavioral cause of his suffer-ing. This assumption, perhaps, is best tested in situations where there is no available characterological information about the victim that might swamp out a rather fragile just-world effect. In the present experiment, however, subjects are initially alerted to the fact that they will provide normative data on attitudes toward sex, and later realize that they must evaluate victims of either rape or carelessness. There are a variety of character inferences that might be made about these two types of victims, and it is possible that subjects may have responded in terms of how they were supposed to feel toward raped or careless people, according to society's norms.
Although the experimental setting used in this study is qualitatively different from the one employed in most of Lerner's studies, it is nonetheless comparable to a variety of situations in which observers are situationally removed from a victim's suffering yet still have access to at least some information about the circumstances surrounding his misfortune. From our perspective, the interesting question is how independent observers arrive at divergent attributions on the basis of identical information. In order to explore this issue, we incorporated a number of assessments designed to supplement the subjects' ratings of the victim's attractiveness. These additional assessments enabled us to understand more fully the kinds of attributions made about victims along various dimensions (e.g., responsibility, deservingness, luckiness) and to examine subjects' evaluations of a victim in light of their reactions to circumstances surrounding her suffering (e.g., perceived adequacy of the victim's family). It was found, for example, that subjects' rating of the victim's attractiveness was uninfluenced by their attitude toward premarital sex but was positively correlated with their perceived similarity to her. Our main point, then, is that observers' reactions to victims can be considered more fruitfully as a composite of attributions rather than as a single attractiveness rating.
With regard to the severity of victim's misfortune, the data revealed that lowseverity victims were perceived to be significantly more attractive than high-severity victims. Furthermore, subjects indicated that there was a greater probability that, upon meeting them, they would like the lowseverity victims more than the high-severity victims. These results are in accord with the prediction that victims who had suffered severely would be perceived as generally less attractive than those who had experienced a mildly unpleasant fate.
It was assumed that the more favorable evaluations of low-severity victims would be mediated by subjects' self-protective attribution of greater responsibility to high-severity victims. The occurrence of this phenomenon seemed likely in the present experiment since subjects were quite similar to the victim in both age and status and, therefore, would probably feel potentially vulnerable to the victim's fate. Thus, rather than believe that a person similar to themselves would experience severe misfortune due to chance, subjects would tend to attribute responsibility for extreme suffering to the victim herself.
From the responsibility manipulation checks (see Table 1), however, it is evident that high-severity victims were perceived to be no more responsible for their misfortune than low-severity victims. Two additional items provide further evidence that increased attribution of responsibility to high-severity victims did not occur. First, victims of severe suffering were rated as significantly unluckier than those of mild misfortune. Second, the family of a low-severity victim was perceived to be more adequate in ameliorating her suffering than the family of a high-severity victim. These findings suggest that subjects attributed partial responsibility for severe misfortune to the inadequacy of the victim's family or to chance factors (e.g., bad luck) rather than assigning greater responsibility to the victim herself.
The most plausible interpretation of the results appears to be one derived from balance theory. In the present context, three related elements can be identified, namely, the victim, her misfortune, and the detached observer. Furthermore, three primary relationships can be identified that link the various elements to each other: (a) a unit relation between the victim and her misfortune, (b) a sentiment relation between the observer and the victim's unpleasant experience, and (c) a sentiment relation between the observer and the victim. It is assumed that the vie-tim's experience of an unpleasant fate represents a unit relation between herself and her misfortune. Moreover, it is expected that observers would develop more negative sentiment toward an extremely unfortunate experience than toward a less pleasant one. Finally, it is predicted that the more negative the sentiment felt toward the victim's misfortune, the less positive will be the sentiment felt toward the victim herself. Thus, given a unit relation between the victim and her misfortune, observers would tend to evaluate high-severity victims less favorably than lowseverity victims, in order to maintain a balanced set of relations between the various elements.
Although in all treatment conditions a unit relation exists between the victim and her misfortune, the strength of that relation presumably would be greatest where the victim had been behaviorally responsible for her own problems. Thus, from an extension of balance theory, it is reasonable to assume that severity effects on ratings of the victim's attractiveness would be most pronounced in those conditions where the victim had personally caused her misfortune.
The pattern of means presented in Table 2 provides partial support for the above interpretation. The four cells within expected future interaction reflect values that are congruent with the assumptions derived from balance theory. Ratings of the victim in the no-expected-future-interaction conditions, however, are discrepant with the severity predictions based on balance theory. Most important, within high-severity-no-expectedfuture-interaction, the victim is rated equally unattractive (relative to victims in the other conditions) whether she was responsible for her misfortune or not. Though this pattern does not reflect a clear-cut devaluation of the innocent victim, it does suggest that some pressure to explain the victim's misfortune in terms of personal unworthiness may have been felt where subjects did not expect to meet her and realized that an innocent person had suffered severely. The means presented in Table 3 further suggest that innocent victims in the high-severity-noexpected-interaction condition were viewed as slightly more deserving of their misfortune vis-a-vis innocent victims in the other cells.
The strength of a just-world interpretation, however, is reduced by the relatively high degree of unluckiness attributed to innocent victims in the high-severity-no-expectedinteraction condition.
Aside from the effects of outcome-severity on the aforementioned variables, one other main effect for severity was observed. Subjects perceived themselves to be more similar, in terms of their political beliefs, to victims of mild misfortune than to those of extreme suffering. This rinding is not surprising, in view of the fact that low-severity victims generally were rated as more attractive than high-severity victims. According to balance theory, the relatively positive sentiment felt toward low-severity victims should have led the observers to infer sentiment relations between the victim and certain topics (e.g., political issues) that were similar to their own feelings toward the same issues.
Considering the results pertaining to expected future interaction, it is evident that the girl described in the psychologist's interview report was rated more favorably by subjects who expected to be her discussion partner than by those who did not, over all dependent measures pertaining to the perceived attractiveness of the victim. This evidence extends the reliability of Schopler and Stokols's (1970) observation, that anticipation of meeting a victim will evoke an accepting orientation toward her, to situations in which the observers are clearly removed from the circumstances surrounding the victim's misfortune.
Tn the present experimental context, there are at least three possible interpretations as to why an expected unit relation with the victim would elicit positive sentiment toward her. First, perceiving the victim as an attractive person may have provided a means of insuring, on the part of the observers, that their discussion with the victim would be provocative and useful rather than boring and uninformative. Second, assuming that people desire to be evaluated favorably by others and tend to act friendly to those persons who express liking for them, it is reasonable that the subjects' adoption of a friendly orientation toward their future partner may have represented an attempt to increase the likelihood of gaining a favorable evaluation from her. Third, the anticipation of participating with the victim in a discussion that would concern issues very much related to her past misfortune may have increased the salience of her recovery. Thus, the expectation of meeting someone who had "bounced back" from her problems would have led subjects to focus upon the victim's heroic qualities (e.g., courage, competence, maturity), thereby promoting a favorable evaluation of her. This possibility may also explain why subjects who expected to meet the victim perceived her family as more adequate in minimizing her suffering. The heightened salience of the victim's recovery, then, may have led subjects to enhance not only her own courage and competence, but also her family's warmth and emotional support.
Since, in the current study, subjects were removed from the scene of the victim's suffering, they were under no pressure to justify their implication in her plight, as might have been the case, for example, in the Lerner and Matthews experiment. It seemed plausible, however, that subjects' perceived similarity to the victim and their commitment to the social norm that she had violated would exert some influence on their evaluations of her. Despite the observed correlation between perceived similarity and ratings of the victim's attractiveness, though, it can be surmised from the data that subjects' attitude toward premarital sex did not affect their evaluation of the victim. In order to gain more direct information about the effects of observers' commitment to the punishing system on their ratings of the victim, this dimension must be explored further in subsequent studies.
In conclusion, data concerning the manipulation of responsibility suggest that under conditions of situational detachment the reaction of a detached observer to a victim is based more upon an objective consideration of the circumstances surrounding the target person's misfortune than an attempt to maintain the notion of a just world. This finding suggests the applicability of equity notions to an analysis of observers' reactions to victims. The data regarding the factors of severity and expected future interaction, however, demonstrate the extent to which pressures toward cognitive balance can influence the observer's objectivity in evaluating a victim.
The anticipated relation between themselves and the victim, as well as the extremity of the misfortune, largely determine whether observers construe an innocent victim as an attractive or unattractive person. Since the present experiment did not directly manipulate the dimensions of situational involvement and alignment with the punishing system, their effects on an observer's evaluation of a victim can be inferred only through a consideration of the differences between separate experimental situations. The results of the present experiment are sufficiently interesting to warrant further exploration of these factors through direct manipulation.