Understanding Consumer Panic: a Sociological Perspective

William M. Strahle, Rider College
E. H. Bonfield, Rider College
ABSTRACT - No consumer behavior theory directly addresses collective action such as fads and fashions, stock market movements, runs on nondurable goods, buying sprees, hoarding, and banking panics. Theory developed from panic literature in sociology has potential for adding to the understanding of collective consumer actions. A panic paradigm is presented in this paper along with a preliminary test of that paradigm.
[ to cite ]:
William M. Strahle and E. H. Bonfield (1989) ,"Understanding Consumer Panic: a Sociological Perspective", in NA - Advances in Consumer Research Volume 16, eds. Thomas K. Srull, Provo, UT : Association for Consumer Research, Pages: 567-573.

Advances in Consumer Research Volume 16, 1989      Pages 567-573


William M. Strahle, Rider College

E. H. Bonfield, Rider College


No consumer behavior theory directly addresses collective action such as fads and fashions, stock market movements, runs on nondurable goods, buying sprees, hoarding, and banking panics. Theory developed from panic literature in sociology has potential for adding to the understanding of collective consumer actions. A panic paradigm is presented in this paper along with a preliminary test of that paradigm.


Most of the effort toward developing theory in consumer behavior has centered on single decision makers faced with making economic choices in relative social isolation (Granbois 1977). That is, either no situational anchoring will be specified in the data gathering instrument or the research will take place in a laboratory setting. Even theories which explicitly link choice context and decision making processes have little to say about the impact of social context per se on consumer behavior (Bettman 1979; Howard 1977; Nicosia 1966). Thus, there is little consumer and economic research about abrupt market transitions such as fads and fashions, stock market movements, runs on nondurable goods, buying sprees, hoarding, and banking panics.

Understanding panic is of interest in its own right, and is reflected in the behavioral economic approaches used to explain collective consumer responses to monetary crises (Katona 1974; Katz 1972; Strumpel 1972; Wicker 1980, 1982) and consumer responses to crises of confidence in the marketplace such as the recent Tylenol scares. It can be found in the communication literature focusing on the consumption, effects, and regulation of mass media (Cantril 1952; McQuail 1983). Interest in panic can also be found in the area of building regulations--particularly in public buildings such as department stores, hospitals, apartment complexes, hotels, and theaters (Canter and Matthews 1976; Canter, et al. 1980).

Finally, panic, as historically conceived, has been represented as a polar case of collective disorganization (Smelser 1963), clearly resting beyond the explanatory power of economic theories which depend on the rationality assumption. If it can be demonstrated that this limitation does not hold and that a purposive action approach is applicable even in the extreme case, panic, the effort may also pave the way for including less extreme and more frequent types of collective behavior (e.g., abrupt market transitions) in theories of consumer behavior.


The model presented here conceptualizes individuals as reward seeking, punishment avoiding (Homans 1974), and operating in a social situation approximating that of an economic market (Coleman 1975). Each individual is characterized as making decisions based on cost-reward calculations, comparisons of alternative courses of action, and anticipation of future rewards (Scanzoni 1979; Kelly and Thibaut 1978). Further, all social relationships-including participation in collective action--are initiated by individuals with given perceptions (Gergen 1971), situational definitions (Abelson 1981) and expectations (Fox 1974), with an eye towards achieving a variety of tangible and/or intangible goals (Kelly and Thibaut 1978). It is the profit, or potential profit, attached to specific behaviors that motivates the individual to invoke them in the appropriate situations. Blau (1964), Homans (1974), and Kelly and Thibaut (1978) suggest other factors that facilitate or inhibit decision-making in a social context and which are incorporated as structural factors in the figure.

The model proposed (and depicted in the figure) has three components: First, a panic cannot occur without the participant's perception of the danger stimulus, their inability to cope with the threatening situation, and the existence of potential--but closing-escape routes. Individually, each of these three conditions is necessary but not sufficient. When all three conditions occur, taken together, they are both necessary and sufficient. Second, the behavioral outcome of panic is either physical or mental withdrawal from the threatening situation. Finally, the structural factors will facilitate or inhibit the decision making process leading from perception to behavior. A complete literature review leading to the development of the model is beyond the scope of the present paper. For a more complete review and development, see Strahle and Bonfield (1988a).


Eight structural factors have emerged from the literature which can have an impact on the time between the perception of a panic stimulus and the resultant behavior.

Following the Kelly and Thibaut (1978) notions of comparison levels, comparison levels of alternatives, and cost-reward calculations, individuals tentatively rank behaviors in terms of their perceived value--the most salient behaviors being invoked more often than the less salient. This assessment process, or activity, is a continuous one. The Kelly and Thibaut framework is predictive--based on profit calculations, i.e., a behavior set would be enacted that would maximize the rewards attained. The availability of rewards, the knowledge of alternative (and perhaps more attractive) behaviors and their corresponding costs, and the calculation and realization of profit enable individuals to behave in a fashion following the principle of profit maximization.

Exceptional caution must be exercised in the utilization of such a "rational man" model in panic producing situations. The time available, or perceived to be available, for considering alternative behavior patterns in terms of their survival value becomes the major factor affecting outcomes (McArthur 1981). In their exhaustive review of decision making in general, as well as in crisis situations, Shapiro and Gilbert (1975) found evidence supporting the following propositions (p. vii, see also pp. 32-35, 72, 73):



1. "In a crisis situation, there is a breakdown in the intellectual abilities of the individual in terms of processing information, assessing the environment, and analyzing alternatives."

2. "The greater the perceived time pressure, the smaller the number of alternatives considered, the greater the likelihood that decisions will be made before necessary, and the greater the likelihood of incorrect choice of alternatives"

While these results do not support a notion of individuals in panic as being either mindless or incompetent, they do suggest that the study of decision making in panic situations must be approached differently than more frequent forms of decision making. While they do not imply that the decision making process in panic is unique, they do suggest that in panic situations, time configurations have a unique effect on the individual's decision making abilities in terms of the costs attached to these alternatives.

Individuals' notions regarding the time patterning or history of previously rewarded behaviors also affect the potential responses of individuals in these situations (Homans 1974; Kelly and Thibaut 1978). Shapiro and Gilbert (1975) found that subjects in panic or crisis situations were likely to use prior experiences--not necessarily in similar situations--as behavioral referents.

According to Smith (1975), emerging norm theory postulates a disregarding of conventional norms, and development of new ad hoc norms by the group in a panic situation. However, emerging norms implies a group consensus regarding appropriateness of behavior. The time factor alone suggests this application would be more suitable to situations such as the Andean plane crash where the survivors voted to consume the bodies of those killed in the crash.

Norms not only define the rules of exchange, but the appropriate behaviors within the relationship as well (Blau 1964). The individual's decision making process, then, can be influenced by the norms accruing to the situation (Smith 1976). The rules regarding execution of soldiers deserting under fire or passengers failing to follow orders while at sea (Lord 1955) are examples of attempts to control this factor through regulation.

Schultz (1964) found that organized groups such as military units warranted separate treatment from unorganized groups such as a typical theater audience when attempting to account for whether or not panic occurred. Subsequent studies focused on the effects of high and low group morale in both military (Shapiro and Gilbert 1975) and non-military groups (Dawes, Shaklee, and Talarowski 1977).

Smelser (1963) maintained that, for a variety of reasons, women, children, and blacks tended to panic first. Smith (1976), on the other hand, found that in primary group relations, children look to their parents for appropriate responses, and generally, wives look to their husbands for the same definitions of the danger stimuli, danger objects, and crisis situations. For example, in the Tower of London bombing, children tended to panic if their parents did, and not to panic if their parents did not (Smith 1976). This implies that the primary group bond may supersede the definition process in unorganized panic behavior and that a panic leader, or model, must appear in these cases for panic to occur (Canter, et al. 1980).

The reason so many believe and follow the cues of panic models may be that their non-normal behavior is inexplicable except as a flight for self survival. The immediacy of the crisis situation and the appreciation of the high cost of remaining in the situation-- frequently personal annihilation--causes others to be receptive to the panic model. Thus, for some, the communication link becomes a deciding factor in selecting withdrawal. Panic leaders, themselves, would not have observation of a panic model as a factor affecting their own decision making process (Shapiro and Gilbert 1975).

Following Schultz (1964), this would be the point where the individual becomes survival oriented in terms of the costs of remaining in the situation.


Withdrawal, as panic behavior, can mean withdrawal into a state of catatonia, as well as physical flight (Merloo 1950). Indeed, as Homans (1974) has pointed out, doing nothing is still a form of behavior.

In summary, panic, which can now be defined as ego-centered survival-oriented withdrawal, is clearly of interest to consumer research. While it may not warrant separate theoretical consideration from other-more frequent--decision making processes, qualitatively it is a highly complex phenomenon on both the individual and group, or social, levels. What makes panic behavior a social phenomenon are the structural factors affecting the individual decision making process: group goals, the apparent necessity for panic leadership, and emergent norms. What makes panic an individual phenomenon are social psychological factors: each individual's panic threshold, perception of the immediacy of the panic stimulus, calculation of rewards and costs attached to alternatives, and perception of potential--but closing-escape routes.

In light of the presence of the panic stimulus and the immediacy of the threatening situation, each individual decides whether the costs of continued association are fair (Wicker 1980). The time for this calculation is momentary, at best, as the decision must be made under the externally imposed condition of a potential, but rapidly closing, escape route. On one hand, all parties might assume the maximum joint profit could be realized by continued association; that by cooperation, the panic stimulus could either be dealt with or avoided, and the cost of potential personal annihilation averted (i.e., "united we stand, divided we fall"). On the other hand, the evidence supports the conclusion that during crisis situations trust and cooperation are, in fact, rare (e.g., Merloo 1950, Quarantelli 1957, Schultz 1964, Wicker 1980, 1982). The net result is the perception of others as obstacles to the individual's safety, or continued existence, and the individual's withdrawal from continued association with others in the panic situation. In a sense, continued interaction becomes too costly as the situation becomes perceived in zero-sum terms: "either I-win-you-lose or I-lose-you-win (Shapiro and Gilbert 1975)." In terms of positions and interests, the individual indicates by withdrawal an unwillingness to be subordinate to, or sacrifice for, the collectivity (Lang and Lang 1961). This decision point is the individual's panic threshold.

The emergence of panic can be seen as affected by the factors bearing on the individual's decision making process under the given time constraints. These factors act as facilitators to the emergence of panic behavior in individuals and, ultimately, in groups (Brown 1954). The presence of a panic stimulus posing the threat of immediate annihilation for the individual, the feedback from others regarding the inability to achieve group goals by coping with the panic stimulus, uncertainty and distrust directed toward the individual and by the individual toward others, the individual and others' panic threshold, and panic scripting (Abelson 1981) all converge to induce the individual to withdraw from the situation by changing his or her goal orientation from maximum joint profit to zero sum.


The question now becomes a methodological one: How is the complex formulation in the figure to be tested? If a panic could be predicted, should the researcher allow it to occur in the interests of observation; or should the researcher take steps to prevent the panic? This ethical dilemma faced by those who may have identified such conditions appears to have channeled research into either laboratory simulations (Mintz 1951) or interviews with surviving participants (Quarantelli 1957). However, attempts to simulate panic situations in laboratories generally have been considered failures (Milgram and Toch 1969) because the treatments seemed to lack realism, "mundane" or otherwise (Fromkin and Streufert 1976). The more frequently employed post facto interviews with participants have often yielded misleading or conflicting results (Shapiro and Gilbert 1975). Researchers have found that many subjects-tend to rationalize their behavior in accounts given following the actual episode. Indeed, a number of subjects would not even tolerate an interview by social scientists or journalists-particularly when they were viewed as persons in a position to pass judgment on the participant's behavior during the episode (Scanlon, et al., 1976).

Testing the complete model found in the figure would pose similar problems. Because the effects of the structural factors on the emergence of panic are mediated through each individual's decision making process during the episode which is, itself, of short duration, evaluation of the contribution or salience of these factors is not likely to either be included or accurately communicated in the retrospective accounts presently available. While collecting protocols from individuals as they experience a panic episode would seem ideal, the logistics in fielding a tea n of depth interviewers prior to the episode, as well as the difficulties of these observers in retaining their objectivity during the episode (ethical considerations aside), render this approach neither practical nor manageable.

Nonetheless, a partial test of the model is possible. The three initial conditions can be tested for their presence or absence in a sample of historical accounts of panic using the method of analytical induction (Denzin 1970). Briefly, analytical induction is a "strategy of analysis that directs the investigator to formulate generalizations that apply to all instances of the problem with which he is concerned (Denzin 1970)." Using theoretical sampling to select cases that explicitly place the hypotheses under investigation in greatest jeopardy, the researcher confronts his or her model with observations that could lead to its rejection or force a reformulation.


The following three hypotheses were derived from initial conditions of the model:

H1: Panic will not occur without the perception of a danger stimulus or the panic leader as a danger stimulus.

H2: Panic -will not occur without the perception of the inability of the individual or group to deal with the panic stimulus.

H3: Panic will not occur without the perception of potential but closing escape routes.

It is important to reiterate the necessary but not sufficient nature of the individual factors in the emergence of panic. That is, before a panic can occur, all three must be present.

The problems inherent in historical research, such as secondary data analysis, should not be minimized (Aydelotte 1971). While panics are clearly newsworthy items, journalists are not renowned for their careful, objective analyses of these occurrences (Sime 1980). Rarely are more than a few participants interviewed, and these clearly cannot give the unwary social scientist any but a disjointed and incomplete picture of the panic development process. In subjecting the hypotheses, and therefore the model, to historical test, the researcher must seek accounts of panic where any one of the three factors is absent rather than cannot be found.

The panic episode selected for initial test of the hypotheses was that occurring during Orson Welles' broadcast of H. G. Wells' "War of the Worlds" on October 30, 1938. Consideration was given to this particular episode because (cf: Cantril 1952):

(a) This episode was studied in detail by Paul Lazarsfeld, Frank Stanton, and Hadley Cantril under the auspices of the ongoing Princeton Radio Project. It was "the first panic that has been studied with the research tools now available (p. vii)."

(b) The study itself involved in-depth interviews of 135 subjects, most of whom were known to have been at least "upset" by the broadcast (p. xi).

(c) The original Mercury Theater script was included in the study, and many of the respondents referred to the substance of the play in accounting for their behavior (pp. 87102).

(d) The episode was an interesting one in its own right, as the panic involved listeners of all ages and social backgrounds connected only, in most instances, by exposure to a common panic stimulus and panic models-- the announcers (pp. 55-63). This panic was clearly aggregate behavior.

It is not possible to estimate the actual number of respondents who panicked, but inferences from the interviews and Cantril's own conclusions suggest the number was quite substantial, even across the nation (Cantril 1952).

Because all of the participants were exposed to the same panic stimulus--the broadcast, a substantive test of the model vis-a-vis the hypotheses would be to attempt to identify the presence or absence of information (i.e., stimuli) communicated to listeners that could have been interpreted as a danger stimulus, as an inability of individuals or aggregates to cope with the situation, and as proposing a potential but closing escape route. If there were no information provided regarding these interpretations, then one or more of the hypotheses would be rejected, as would the model.


Content analysis of the broadcast script provides support for each of the three hypotheses. The table contains a sampling of accounts of the panic stimuli from the script of the radio account of the invasion from Mars. The radio listeners were gradually acquainted with the nature of "the enemy" and its destructiveness in terms of lives and property. What at first was a nebulous danger became identified as "a machine" carrying Martians "with saliva dripping from [their] rimless lips" and sporting "heat-rays," that turned men into flame, as well as deadly gas producing equipment that had them "dropping like flies."

As the listeners stayed tuned, they were informed that these creatures possessed scientific knowledge "far in advance of our own" and that this knowledge had been used to "wipe out" the army and air force.

Concomitantly, anxious listeners were directed by a series of announcers to avoid fleeing by particular rail services and highways as they were either disrupted by the invaders themselves or were "clogged with frantic human traffic."

In total, there were ten specific references to the nature of the danger object and the human destruction it caused; seven accounts of the failure of the army, air force, and scientific community to deal with the danger object and stop the loss of life and destruction of property; and three clues regarding potential but closing escape routes for the listening audience.



There was ample information provided in the broadcast regarding the identity of the danger object or panic stimulus, the inability of individuals and aggregates to cope with the danger object, and potential but closing escape routes. The initial three hypotheses cannot be disconfirmed.


This partial testing of a model of individual decision making in panic situations represents a beginning in the search for a parsimonious approach to this periodic, yet recurrent form of collective behavior. Additional panic situations must be studied to provide further evidence as to whether any of the initial three hypotheses can be disconfirmed. A test of the complete model is also required. While panic induced behavior, withdrawal, can be easily observed; and the stimulus condition hypotheses can sometimes be tested; the temporal restraints affecting the structural factors make it difficult to test hypotheses for this part of the model.

From a social policy standpoint, the model has several implications. Given the severely limited time available for interventions, efforts to arrest the panic process would need to be preventive in nature. That is, public consumer contexts could be made safer by making exits more visible rather than irregularly spaced based on some arbitrary density rule or regularly spaced but hidden for decorative reasons. Conducting emergency exit drills in public contexts other than schools, such as in hospitals and hotels, to, at a minimum, ensure that the exits are open and working, seems worthwhile. Public service advertising could be designed to ensure that people know what they can do in a variety of public and private contexts in a variety of emergency situations (hurricanes, tornadoes, home fires, blizzards) in much the same way that airline crews demonstrate flotation devices and what to do in the case of sudden cabin decompression.

Much remains to be done in the study of panic induced decision making and behavior. Still needed is a more exhaustive--in terms of variety of panic types-set of historical tests of the model, as well as the design of a simulation that would allow for the assessment of the mediating structural factors in terms of their influence on individual decision making under the constraints accruing to panic situations.

Panic does not warrant separate treatment from other, more common, collective consumer phenomena such as hoarding behavior and buying sprees--nor does it require separate, substantive theoretical treatment. Financial market panic, such as occurred in October 1987, is relatively common even in our high technology economy. Understanding panic can lead to the development of steps to forestall panic behavior in these markets. It is time to turn our attention to collective behavior and bring it into our theories of consumer behavior.


Abelson, Robert (1981), "Psychological Status of the Script Concept," American Psychologist, 36 (Number 7, July), 715-729.

Aydelotte, W. (1971), Quantification in History, Reading, MA: Addison Wesley.

Bettman, James R. (1979), An Information Processing Theory of Consumer Choice, Reading, MA: Addison Wesley.

Blau, Peter (1964), Exchange and Power in Social Life, New York: Wiley.

Brown, Roger (1954), "Mass Phenomena," in Handbook of Social Psychology, ed. Gardner Lindzey, Reading, MA: Addison-Wesley, 833-876.

Canter, David and R. Matthews (1976), Behaviour in Fires: The Possibilities for Research, CP11/76 BRE Fire Research Station, Borehamwood, England.

Canter, David, J. Breaux, and J. Sime (1980), "Domestic, Multiple Occupancy, and Hospital Fires," in Fires and Human Behaviour, ed. David Canter, London: Wiley.

Cantril, Hadley (1952), The Invasion from Mars, Princeton, NJ: Princeton University Press.

Coleman, J. (1975), Social Structure and a Theory of Action, in Approaches to the Study of Social Structure, ed. Peter Blau, New York: Free Press

Dawes, Robyn M., Harriet Shaklee, and F. Talarowski (1977), "On Getting People to Cooperate when Facing a Social Dilemma," Oregon Research Institute, Department of Defense mimeo.

Denzin, Norman K. (1970), The Research Act, Chicago: Aldine.

Fox, A. (1974), Beyond Contract: Work, Power and Trust Relations, London: Hogart

Fromkin, Howard L. and Siegfried Streufert (1976), "Laboratory Experimentation," in Handbook of Industrial and Organizational Psychology, ed. Marvin D. Dunnette, Chicago: Rand McNally, 415-465.

Gergen, Kenneth J. (1971), Concept of Self, New York: Holt, Rinehart, and Winston.

Granbois, Donald H. (1977), "Shopping Behavior and Preferences," in Selected Aspects of Consumer Behavior: A Summary from the Perspective of Different Disciplines, ed. Robert Ferber, Washington: U.S. Government Printing Office, 259-298.

Homans, G. C. (1974), Social Behavior: Its Elementary Forms, New York: Harcourt, Brace and World.

Howard, John A. (1977), Consumer Theory: Application of Theory, New York: McGraw-Hill.

Katona, George (1974), "Psychology and Consumer Economics," Journal of Consumer Research, 1 (June), 1-12.

Katz, Daniel (1972), "Psychology and Economic Behavior," in Human Behavior in Economic Affairs, eds. Burkhard Strumpel, James Morgan, and Ernest Zahn, Washington: Jossey-Bass, 57-81.

Kelly, Harold H. and John W. Thibaut (1978), Interpersonal Relations: A Theory of Interdependence, New York: Wiley.

Lang, Kurt and Gladys Lang (1961), Collective Dynamics, New York: T. Crowell.

Lord, Walter (1955), A Night to Remember, New York: Holt, Rinehart, and Winston.

McArthur, Leslie (1981), "What Grabs You? The Role of Attention in Impression Formation and Causal Attribution, in Social Cognition: The Ontario Symposium," eds. Tony Higgins, Peter Herman, and Mark Zanna, I, 201-246.

McQuail, Denis (1983), Mass Communication Theory, Beverly Hills, CA: Sage.

Merloo, Joost (1950), Patterns of Panic, New York: International Press.

Milgram, Stanley and Hans Toch (1969), "Collective Behavior: Crowds and Social Movements," in Handbook of Social Psychology, eds. Gardner Lindzey and Elliot Aronson, Reading, MA: Addison Wesley, IV, 507-610.

Mintz, Alexander (1951), "Non-Adaptive Group Behavior," Journal of Abnormal and Social Psychology, 46 (Number 1, January), 150-159.

Nicosia, Francesco M. (1966), Consumer Decision Processes: Marketing and Advertising Implications, Englewood Cliffs, NJ: Prentice-Hall. ,v, Quarantelli, E. (1957), "The Behavior of Panic Participants," Sociology and Social Research, 41 (January-February), 187-194.

Scanlon, Joseph, Jim Jefferson, and Debbie Sproat (1976), "The Port Alice Slide," Working Paper 76/1, Ottowa: Emergency Planning Canada.

Scanzoni, J. (1979), "Social Processes and Power in Families," in Contemporary Theories About the Family, Vol I, eds. Wesley R. Burr, Reuben Hill, F. Ivan Nye and I. L. Reiss, New York: Free Press.

Schultz, Duane P., ed. (1964), Panic Behavior: Discussion and Readings, New York: Random House.

Shapiro, Howard and Marcia Gilbert (1975), "Crisis Management: Psychological and Sociological Factors in Decision Making," Office of Naval Research Report: Advanced Research Projects Agency, Washington, DC: U.S. Department of the Navy.

Sime, Jonathan (1980), "The Concept of 'Panic,"' in Fires and Human Behavior, ed. David Canter, London: Wiley, 63-81.

Smelser, Neil (1963), Theory of Collective Behavior, New York: Free Press.

Smith, Don D. (1975), "Primary Group Interaction in Panic Behavior: A Test of Theories," Mimeo, Florida State University.

Smith, Don D. (1976), "Primary Group interaction in Panic Behavior: A Test of Theories," Mimeo, Florida State University, Paper presented to the Southern Sociological Society.

Strahle, William M. and E. H. Bonfield (1988a), "Consumer Panic: A Perspective on the Relevance of Collective Behavior for Consumer Behavior Theory," Rider College, School of Business Administration, Working Paper Number 75.

Strahle, William M. and E. H. Bonfield (1988b), "Understanding Consumer Panic: A Sociological Perspective," Rider College, School of Business Administration, Working Paper Number 76.

Strumpel, Burkhard. (1972), "Economic Behavior and Economic Welfare--Models and Interdisciplinary Approaches," in Human Behavior in Economic Affairs eds. Burkhard Strumpel, James Morgan, and Ernest Zahn, Washington, D.C.: Jossey-Bass, 83-108.

Wicker. Elmus (1980), "A Reconsideration of the Causes of the Banking Panic of 1930," Journal of Economic History, 40 (Number 3, September), 571-583.

Wicker, Elmus (1982), "Interest Rate and Expenditure Effects of the Banking Panic of 1930," Explorations in Economic History, 19 (October), 435-445.