Group Inhibition of Bystander Intervention in Emergencies PDF

Journal at Personality and Social Psycholoty 1968, Vol. 10, No. 3, 215-221 GROUP INHIBITION OF BYSTANDER INTERVENTION IN EMERGENCIES l BIBB LATANfi 2 AND JOHN M, DARLEY » Columbia Universi...

Journal at Personality and Social Psycholoty 1968, Vol. 10, No. 3, 215-221 GROUP INHIBITION OF BYSTANDER INTERVENTION IN EMERGENCIES l BIBB LATANfi 2 AND JOHN M, DARLEY » Columbia University New York University Male undergraduates found themselves in a smoke-filling room either alone, with 2 nonreacting others, or in groups of 3. As predicted, Ss were less likely to report the smoke when in the presence of passive others (10%) or in groups of 3 (38% of groups) than when alone (75%). This result seemed to have been mediated by the way 5s interpreted the ambiguous situation; seeing other people remain passive led Ss to decide the smoke was not dangerous. Emergencies, fortunately, are uncommon viduals heard a person undergoing a severe events. Although the average person may read epileptic-like fit in another room. In one about them in newspapers or watch fictional- experimental condition, the subject thought ized versions on television, he probably will that he was the only person who heard the encounter fewer than half a dozen in his life- emergency; in another condition, he thought time. Unfortunately, when he does encounter four other persons were also aware of the one, he will have had little direct personal seizure. Subjects alone with the victim were experience in dealing with it. And he must much more likely to intervene on his behalf, deal with it under conditions of urgency, un- and, on the average, reacted in less than certainty, stress, and fear. About all the indi- one-third the time required by subjects who vidual has to guide him is the secondhand thought there were other bystanders present. wisdom of the late movie, which is often as "Diffusion of responsibility" seems the most useful as "Be brave" or as applicable as likely explanation for this result. If an indi- "Quick, get lots of hot water and towels 1" vidual is alone when he notices an emergency, Under the circumstances, it may seem sur- he is solely responsible for coping with it. prising that anybody ever intervenes in an If he believes others are also present, he may emergency in which he is not directly in- feel that his own responsibility for taking volved. Yet there is a strongly held cultural action is lessened, making him less likely to norm that individuals should act to relieve help. the distress of others. As the Old Parson To demonstrate that responsibility diffusion puts it, "In this life of froth and bubble, two rather than any of a variety of social influence things stand like stone—kindness in another's processes caused this result, the experiment trouble, courage in your own." Given the con- was designed so that the onlookers to the seiz- flict between the norm to act and an indi- ure were isolated one from another and could vidual's fears and uncertainties about getting not discuss how to deal with the emergency ef- involved, what factors will determine whether fectively. They knew the others could not see a bystander to an emergency will intervene? what they did, nor could they see whether We have found (Barley & Latane", 1968) somebody else had already started to help. that the mere perception that other people Although this state of affairs is characteristic are also witnessing the event will mark- of many actual emergencies (such as the edly decrease the likelihood that an indi- Kitty Genovese murder in which 38 people vidual will intervene in an emergency. Indi- witnessed a killing from their individual thank Lee Ross and Keith Gerritz for their apartments without acting), in many other thoughtful efforts. This research was supported by emergencies several bystanders are in contact National Science Foundation Grants GS 1238 and with and can influence each other. In these GS 1239. The experiment was conducted at Columbia situations, processes other than responsibility University. 2 Now at the Ohio State University. diffusion will also operate. 8 Now at Princeton University. Given the opportunity to interact, a group 215 216 BIBB LATAN£ AND JOHN M. DARLEY can talk over the situation and divide up the that this passive behavior will signal the helping action in an efficient way. Also, since individual that the other bystanders do not responding to emergencies is a socially pre- consider the situation to be dangerous. We scribed norm, individuals might be expected predict that an individual faced with the pas- to adhere to it more when in the presence of sive reactions of other people will be influ- other people. These reasons suggest that inter- enced by them, and will thus be less likely to acting groups should be better at coping with take action than if he were alone. emergencies than single individuals. We sus- This, however, is a prediction about indi- pect, however, that the opposite is true. viduals; it says nothing about the original Even when allowed to communicate, groups question of the behavior of freely interacting may still be worse than individuals. groups. Most groups do not have preinstructed Most emergencies are, or at least begin as, confederates among their members, and the ambiguous events. A quarrel in the street may kind of social influence process described erupt into violence, but it may be simply a above would, by itself, only lead to a con- family argument. A man staggering about vergence of attitudes within a group. Even may be suffering a coronary or an onset of if each member of the group is entirely guided diabetes; he may be simply drunk. Smoke by the reactions of others, then the group pouring from a building may signal a fire; on should still respond with a likelihood equal to the other hand, it may be simply steam or the average of the individuals. air-conditioning vapor. Before a bystander is An additional factor is involved, however. likely to take action in such ambiguous situa- Each member of a group may watch the tions, he must first define the event as an others, but he is also aware that the others emergency and decide that intervention is the are watching him. They are an audience to proper course of action. his own reactions. Among American males it In the course of making these decisions, is considered desirable to appear poised and it is likely that an individual bystander will collected in times of stress. Being exposed to be considerably influenced by the decisions public view may constrain an individual's he perceives other bystanders to be taking. actions as he attempts to avoid possible If everyone else in a group of onlookers seems ridicule and embarrassment. to regard an event as nonserious and the The constraints involved with being in proper course of action as nonintervention, public might in themselves tend to inhibit this consensus may strongly affect the per- action by individuals in a group, but in con- ceptions of any single individual and inhibit junction with the social influence process de- his potential intervention. scribed above, they may be expected to have The definitions that other people hold may even more powerful effects. If each member be discovered by discussing the situation with of a group is, at the same time, trying to them, but they may also be inferred from appear calm and also looking around at the their facial expressions or their behavior. A other members to gauge their reactions, all whistling man with his hands in his pockets members may be led (or misled) by each obviously does not believe he is in the midst other to define the situation as less critical of a crisis. A bystander who does not respond than they would if alone. Until someone acts, to smoke obviously does not attribute it to each person only sees other nonresponding by- fire. An individual, seeing the inaction of standers, and, as with the passive confeder- others, will judge the situation as less serious ates, is likely to be influenced not to act than he would if he were alone. himself. In the present experiment, this line of This leads to a second prediction. Com- thought will be tested by presenting an emer- pared to the performance of individuals, if we gency situation to individuals either alone or expose groups of naive subjects to an emer- in the presence of two passive others, con- gency, the constraints on behavior in public federates of the experimenter who have been coupled with the social influence process will instructed to notice the emergency but re- lessen the likelihood that the members of the main indifferent to it. It is our expectation group will act to cope with the emergency.  GROUP INHIBITION op BYSTANDER INTERVENTION IN EMERGENCIES 217 It has often been recognized (Brown, 1954, a small vent in the wall. The "smoke" was finely 1965) that a crowd can cause contagion of divided titanium dioxide produced in a stoppered bottle and delivered under slight air pressure through panic, leading each person in the crowd to the vent.2 It formed a moderately fine-textured but overreact to an emergency to the detriment of clearly visible stream of whitish smoke. For the everyone's welfare. What is implied here is entire experimental period, the smoke continued to that a crowd can also force inaction on its jet into the room in irregular puffs. By the end of members. It can suggest, implicitly but the experimental period, vision was obscured by the amount of smoke present. strongly, by its passive behavior, that an All behavior and conversation was observed and event is not to be reacted to as an emergency, coded from behind a one-way window (largely dis- and it can make any individual uncomfort- guised on the subject's side by a large sign giving ably aware of what a fool he will look for preliminary instructions). If the subject left the experimental room and reported the smoke, he was behaving as if it is. told that the situation "would be taken care of." If the subject had not reported the presence of smoke METHOD by 6 minutes from the time he first noticed it, the The subject, seated in a small waiting room, faced experiment was terminated. an ambiguous but potentially dangerous situation as a stream of smoke began to puff into the room RESULTS through a wall vent. His response to this situation was observed through a one-way glass. The length Alone condition. The typical subject, when of time the subject remained in the room before tested alone, behaved very reasonably. Usu- leaving to report the smoke was the main dependent ally, shortly after the smoke appeared, he variable of the study. would glance up from his questionnaire, notice Recruitment of subjects. Male Columbia students living in campus residences were invited to an inter- the smoke, show a slight but distinct startle view to discuss "some of the problems involved in reaction, and then undergo a brief period of life at an urban university." The subject sample indecision, perhaps returning briefly to his included graduate and professional students as well questionnaire before again staring at the as undergraduates. Individuals were contacted by smoke. Soon, most subjects would get up from telephone and most willingly volunteered and actu- ally showed up for the interview. At this point, they their chairs, walk over to the vent, and were directed either by signs or by the secretary investigate it closely, sniffing the smoke, to a "waiting room" where a sign asked them to waving their hands in it, feeling its tempera- fill out a preliminary questionnaire. ture, etc. The usual alone subject would hesi- Experimental manipulation. Some subjects filled tate again, but finally walk out of the room, out the questionnaire and were exposed to the poten- tially critical situation while alone. Others were part look around outside, and, finding somebody of three-person groups consisting of one subject and there, calmly report the presence of the two confederates acting the part of naive subjects. smoke. No subject showed any sign of The confederates attempted to avoid conversation as panic; most simply said, "There's something much as possible. Once the smoke had been intro- duced, they stared at it briefly, made no comment, strange going on in there, there seems to but simply shrugged their shoulders, returned to the be some sort of smoke coming through the questionnaires and continued to fill them out, oc- wall...." casionally waving away the smoke to do so. If ad- The median subject in the alone condition dressed, they attempted to be as uncommunicative had reported the smoke within 2 minutes of as possible and to show apparent indifference to the smoke. "I dunno," they said, and no subject persisted first noticing it. Three-quarters of the 24 in talking. people who were run in this condition In a final condition, three naive subjects were reported the smoke before the experimental tested together. In general, these subjects did not period was terminated. know each other, although in two groups, subjects reported a nodding acquaintanceship with another Two passive confederates condition. The subject. Since subjects arrived at slightly different behavior of subjects run with two passive times and since they each had individual question- confederates was dramatically different; of 10 naires to work on, they did not introduce themselves people run in this condition, only 1 reported to each otBer, or attempt anything but the most 2 rudimentary conversation, Smoke was produced by passing moisturized air, Critical situation. As soon as the subjects had under pressure, through a container of titanium tetra- completed two pages of their questionnaires, the chloride, which, in reaction with the water vapor, experimenter began to introduce the smoke through creates a suspension of tantium dioxide in air. 218 BIBB LATANE AND JOHN M. DASLEY the smoke. The other 9 stayed in the waiting "groups" of three scores from the alone room as it filled up with smoke, doggedly condition to serve as a base line.8 working on their questionnaire and waving In contrast to the complexity of this pro- the fumes away from their faces. They cedure, the results were quite simple. Subjects coughed, rubbed their eyes, and opened the in the three naive bystander condition were window—but they did not report lie smoke. markedly inhibited from reporting the smoke. The difference between the response rate of Since 75% of the alone subjects reported 75% in the alone condition and 10% in the the smoke, we would expect over 98% of the two passive confederates condition is highly three-person groups to contain at least one significant (p <.002 by Fisher's exact test, reporter. In fact, in only 38% of the eight two-tailed). groups in this condition did even 1 subject Three naive bystanders. Because there are report (p <.01). Of the 24 people run in three subjects present and available to report these eight groups, only 1 person reported the the smoke in the three naive bystander smoke within the first 4 minutes before the condition as compared to only one sub- room got noticeably unpleasant. Only 3 ject at a time in the alone condition, a people reported the smoke within the entire simple comparison between the two condi- experimental period. tions is not appropriate. On the one hand, Cumulative distribution of report times. we cannot compare speeds in the alone con- Figure 1 presents the cumulative frequency dition with the average speed of the three distributions of report times for all three con- subjects in a group, since, once one subject ditions. The figure shows the proportion of in a group had reported the smoke, the pres- subjects in each condition who had reported sures on the other two disappeared. They the smoke by any point in the time following legitimately could (and did) feel that the the introduction of the smoke. For example, emergency had been handled, and any action 55% of the subjects in the alone condition on their part would be redundant and poten- had reported the smoke within 2 minutes, but tially confusing. Therefore the speed of the the smoke had been reported in only 12% first subject in a group to report the smoke of the three-person groups by that time. After was used as the dependent variable. However, 4 minutes, 75% of the subjects in the alone since there were three times as many people condition had reported the smoke; no addi- available to respond in this condition as in tional subjects in the group condition had the alone condition, we would expect an done so. The curve in Figure 1 labeled increased likelihood that at least one person "Hypothetical Three-Person Groups" is based would report the smoke even if the sub- upon the mathematical combination of scores jects had no influence whatsoever on each obtained from subjects in the alone condi- other. Therefore we mathematically created tion. It is the expected report times for groups in the three-person condition if the members of the groups had no influence upon each other. It can be seen in Figure 1 that for every point in time following the introduction of the smoke, a considerably higher proportion of subjects in the alone condition had reported the smoke than had subjects in either the two passive confederates condition or in the three naive subjects condition. The curve for the latter condition, although considerably below "The formula for calculating the expected propor- tion of groups in which at least one person will have acted by a given time is 1 — (!—/>)" where p is FIG. 1. Cumulative proportion of subjects reporting the proportion of single individuals who act by that the smoke over time. time and n is the number of persons in the group.  GROUP INHIBITION or BYSTANDER INTERVENTION IN EMERGENCIES 219 the alone curve, is even more substantially questionnaires, subjects in groups usually kept inhibited with respect to its proper compari- their eyes closely on their work, probably to son, the curve of hypothetical three-person avoid appearing rudely inquisitive. sets. Social inhibition of response was so great Postexperimental interview. After 6 min- that the time elapsing before the smoke was utes, whether or not the subjects had reported reported was greater when there were more the smoke, the interviewer stuck his head people available to report it (alone versus in the waiting room and asked the subject group p <.05 by Mann-Whitney U test). to come with him to the interview. After Superficially, it appears that there is a seating the subject in his office, the inter- somewhat higher likelihood of response from viewer made some general apologies about groups of three naive subjects than from sub- keeping the subject waiting for so long, hoped jects in the passive confederates condition. the subject hadn't become too bored and Again this comparison is not justified; there asked if he "had experienced any difficulty are three people free to act in one condition while filling out the questionnaire." By this instead of just one. If we mathematically point most subjects mentioned the smoke. combine scores for subjects in the two pas- The interviewer expressed mild surprise and sive confederates condition in a similar man- asked the subject to tell him what had hap- ner to that described above for the alone pened. Thus each subject gave an account of condition, we would obtain an expected likeli- what had gone through his mind during the hood of response of.27 as the hypothetical smoke infusion. base line. This is not significantly different Subjects who had reported the smoke were from the.37 obtained in the actual three- relatively consistent in later describing their subject groups. reactions to it. They thought the smoke Noticing the smoke. In observing the sub- looked somewhat "strange," they were not ject's reaction to the introduction of smoke, sure exactly what it was or whether it was careful note was taken of the exact moment dangerous, but they felt it was unusual enough when he first saw the smoke (all report to justify some examination. "I wasn't sure latencies were computed from this time). whether it was a fire but it looked like some- This was a relatively easy observation to thing was wrong." "I thought it might be make, for the subjects invariably showed a steam, but it seemed like a good idea to check distinct, if slight, startle reaction. Unexpect- it out." edly, the presence of other persons delayed, Subjects who had not reported the smoke slightly but very significantly, noticing the also were unsure about exactly what it was, smoke. Sixty-three percent of subjects in the but they uniformly said that they had re- alone condition and only 26% of subjects in jected the idea that it was a fire. Instead, they the combined together conditions noticed the hit upon an astonishing variety of alternative smoke within the first 5 seconds after its explanations, all sharing the common charac- introduction (p <.01 by chi-square). The teristic of interpreting the smoke as a non- median latency of noticing the smoke was dangerous event. Many thought the smoke under 5 seconds in the alone condition; the was either steam or air-conditioning vapors, median time at which the first (or only) sub- several thought it was smog, purposely intro- ject in each of the combined together condi- duced to simulate an urban environment, and tions noticed the smoke was 20 seconds (this two (from different groups) actually sug- difference does not account for group-induced gested that the smoke was a "truth gas" inhibition of reporting since the report laten- filtered into the room to induce them to an- cies were computed from the time the smoke swer the questionnaire accurately. (Surpris- was first noticed). This interesting finding can probably be ingly, they were not disturbed by this con- explained in terms of the constraints which viction.) Predictably, some decided that "it people feel in public places (Goffman, 1963). must be some sort of experiment" and stoicly Unlike solitary subjects, who often glanced endured the discomfort of the room rather idly about the room while filling out their than overreact. 220 BIBB LATAN£ AND JOHN M. DARLEY Despite the obvious and powerful report- Two of these alternative explanations stem inhibiting effect of other bystanders, subjects from the fact that the smoke represented a almost invariably claimed that they had paid possible danger to the subject himself as little or no attention to the reactions of the well as to others in the building. Subjects' other people in the room. Although the pres- behavior might have reflected their fear of ence of other people actually had a strong fire, with subjects in groups feeling less and pervasive effect on the subjects' reactions, threatened by the fire than single subjects they were either unaware of this or unwilling and thus being less concerned to act. It has to admit it. been demonstrated in studies with humans (Schachter, 1959) and with rats (Latane", DISCUSSION 1968; Latan6 & Glass, 1968) that together- ness reduces fear, even in situations where it Before an individual can decide to intervene does not reduce danger. In addition, subjects in an emergency, he must, implicitly or ex- may have felt that the presence of others plicitly, take several preliminary steps. If he increased their ability to cope with fire. For is to intervene, he must first notice the event, both of these reasons, subjects in groups may he must then interpret it as an emergency, have been less afraid of fire and thus less and he must decide that it is his personal likely to report the smoke than solitary responsibility to act. At each of these pre- subjects. liminary steps, the bystander to an emergency A similar explanation might emphasize not can remove himself from the decision process fearfulness, but the desire to hide fear. To and thus fail to help. He can fail to notice the extent that bravery or stoicism in the face the event, he can fail to interpret it as an of danger or discomfort is a socially desirable emergency, or he can fail to assume the trait (as it appears to be for American male responsibility to take action. undergraduates), one might expect individuals In the present experiment we are pri- to attempt to appear more brave or more marily interested in the second step of this stoic when others are watching than when decision process, interpreting an ambiguous they are alone. It is possible that subjects event. When faced with such an event, we in the group condition saw themselves as en- suggest, the individual bystander is likely to gaged in a game of "Chicken," and thus did look at the reactions of people around him not react. and be powerfully influenced by them. It was Although both of these explanations are predicted that the sight of other, nonrespon- plausible, we do not think that they pro- sive bystanders would lead the individual to vide an accurate account of subjects' think- interpret the emergency as not serious, and ing. In the postexperimental interviews, consequently lead him not to act. Further, it subjects claimed, not that they were unwor- was predicted that the dynamics of the inter- ried by the fire or that they were unwilling action process would lead each of a group of to endure the danger; but rather that they naive onlookers to be misled by the apparent decided that there was no fire at all and the inaction of the others into adopting a non- smoke was caused by something else. They emergency interpretation of the event and a failed to act because they thought there was passive role. no reason to act. Their "apathetic" behavior The results of this study clearly support was reasonable—given their interpretation of our predictions. Individuals exposed to a room the circumstances. filling with smoke in the presence of passive The fact that smoke signals potential others themselves remained passive, and danger to the subject himself weakens another groups of three naive subjects were less likely alternative explanation, "diffusion of responsi- to report the smoke than solitary bystanders. bility." Regardless of social influence proc- Our predictions were confirmed—but this does esses, an individual may feel less personal re- not necessarily mean that our explanation for sponsibility for helping if he shares the these results is the correct one. As a matter of responsibility with others (Darley & Latane", fact, several alternatives are available. 1968). But this diffusion explanation does not  GROUP INHIBITION OF BYSTANDER INTERVENTION IN EMERGENCIES 221 fit the present situation. It is hard to see how have listened to but not answered a call for an individual's responsibility for saving him- help. Newspapers have tagged these incidents self is diffused by the presence of other people. with the label "apathy." We have become The diffusion explanation does not account indifferent, they say, callous to the fate of for the pattern of interpretations reported by suffering others. The results of our studies the subjects or for their variety of non- lead to a different conclusion. The failure to emergency explanations. intervene may be better understood by know- On the other hand, the social influence ing the relationship among bystanders rather processes which we believe account for the than that between a bystander and the victim. results of our present study obviously do not explain our previous experiment in which REFERENCES subjects could not see or be seen by each BROWN, R. W. Mass phenomena. In G. Lindzey other. Taken together, these two studies (Ed.), Handbook of social psychology. Vol. 2. suggest that the presence of bystanders may Cambridge: Addison-Wesley, 1954. affect an individual in several ways; including BROWN, R. Social psychology. New York: Free both "social influence" and "diffusion of Press of Glencoe, 1965. DARLEY, J. M., & LATANE, B. Bystander interven- responsibility." tion in emergencies: Diffusion of responsibility. Both studies, however, find, for two quite Journal of Personality and Social Psychology, different kinds of emergencies and under two 1968, 8, 377-383. quite different conditions of social contact, GOFFMAN, E. Behavior in public places. New York: that individuals are less likely to engage in Free Press of Glencoe, 1963. socially responsible action if they think other LATAN£, B. Gregariousness and fear in laboratory rats. Journal of Experimental Social Psychology, bystanders are present. This presents us with 1968, in press. the paradoxical conclusion that a victim may LATAN&, B., & GLASS, D. C. Social and nonsocial be more likely to get help, or an emergency attraction in rats. Journal of Personality and may be more likely to be reported, the fewer Social Psychology, 1968, 9, 142-146. people there are available to take action. It SCHACHTER, S. The psychology of affiliation. Stan- ford: Stanford University Press, 19S9. also may help us begin to understand a number of frightening incidents where crowds (Received December 11, 1967)