HELP THY NEIGHBOR: A STUDY OF BYSTANDER
INTERVENTION IN EMERGENCIES
ELLEN WEISS WILLIAMS
A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF
THE UNIVERSITY OF FLORIDA IN PARTIAL
FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
We finally made it. I'm coming home for good.
The writer wishes to express her gratitude for the guidance
and counsel given her by the members of her supervisory committee:
Dr. Lawrence J. Severy and Dr. Marvin E. Shaw, co-chairmen; Dr.
Franz Epting; Dr. Norman Markel; and Dr. Milan Kolarik. In
addition, the author would like to thank Ms. Sharon Spector and
Mr. William Cone for all the long hours they spent acting as
TABLE OF CONTENTS
ACKNOWLEDGEMENTS. . .. . . . . . . ... . .. iii
LIST OF TABLES. . . . . . . . . ... . . . vii
LIST OF FIGURES . . . . . . . . ... ..... .ix
ABSTRACT . . . . . . . . . . . . . x
I 'INTRODUCTION . . . . . . . . . . 1
Overview . . . . . . . . ... . .. 3
Bystander Intervention Paradigm. . . . . . 4
The Victim.. . . . . . . . . . 4
The Bystander. . . . . . . . . . 5
The Emergency. . . . . . . . . . 6
The Intervention Process . . . . . . 8
A Theory of Interaction Outcomes . . . . . 9
Analysis of Interaction. .. . . . . 9
Consequences of Interaction (Rewards/Costs). . 10
The Determinants of Rewards and Costs . . .. 10
Formation of the Relationship. . . . . ... 11
Evaluation of the Relationship . . . . .. 11
Power and Dependence . . . . . . .. 13
BIP: A Case for Interaction Outcome Theory. . 15
Literature Review . . . . . . . .. 18
Relationship Among Bystanders .. . .... 18
Situational Determinants of the BIP. .. ... 25
Demographic and Physical Characteristics of the
Victim . . . . . . . . . . 31
Summary . . . . . . . . . . .. 36
Background of the Present Experiment . . ... 37
Bystander/Victim Interaction . . . . .. 37
Attitude Similarity Between Bystander and Victim 40
Bystander/Environmental Interaction . . ... 42
Environmental Familiarity. . ....... .42
The Present Experiment ............. .43
Attitude Similarity. . . . . . . ... 43
Environmental Familiarity. . . . . . .. 43
II METHODOLOGY. .. .. . . . ..... . . . . 46
Subjects . . . . . . . .. . .46
Procedure. . . . . . . . . ... .. 46
Part I: Personality . . . . . ... .48
Part II: Visual Perception. . . . . ... 50
The Emergency. . . . . . . . . ... 51
III RESULTS. . . . . . . . ... ..... .54
Check on Manipulation. . . . . . . ... 54
Attitude Similarity . . . . . ..... 54
Environmental Familiarity. . . . . . .. 55
Major Hypotheses . . . . . . .... .55
Data Analysis for All Subjects . . . .... .57
First Help Latency . . . . . . . .. 59
Mode of Helping Behavior . . . . . .. 62
Differential Mode Latency. . . . . . .. 67
Summary. . . . . . . . . . .. 73
Satisfaction with Response . . ......... 75
Data Analysis for Helping Subjects . . . ... 81
Population . . . . . . . . . 81
First Help Latency . . . . . . . .. 81
Mode of Helping Behavior . . . ... .... 83
Satisfaction with Response . . . . .... .88
Perceived All Subject and Helping Subject Data
Analysis . . . . . . . . . . 88
Sunmary. . . . . . . . .. .. .91
Secondary Hypotheses . . . . . . .... .92
Secondary Hypothesis 1 .
Secondary Hypothesis 2 .
Secondary Hypothesis 3 .
Secondary Hypothesis 4 .
IV DISCUSSION AND CONCLUSION. ..
The Hypotheses . . . .
Support for the Theory . .
Additional Findings. . .
Implications of the Theory
Concluding Comment . . .
APPENDICES. . . . . . . .
A PERSONALITY QUESTIONNAIRE. ..
B SEMANTIC DIFFERENTIAL QUESTION
SUBJECTS ONLY. . .. .. .
C VISUAL PERCEPTION QUESTIONNAIRE.
D POST-EXPERIMENTAL QUESTIONNAIRE.
REFERENCES. . .. . . . ...
BIOGRAPHICAL SKETCH . . . . . .
. . .
. . .
. . .
. . .
. . .
. . .
. . . .
. .. o.
. . .
LIST OF TABLES
1 PERCENTAGE AND LATENCY OF HELP EXHIBITED BY
BYSTANDERS ACCORDING TO GROUP (MANIPULATION). .... . 58
2 ANALYSIS OF VARIANCE OF FIRST HELP LATENCY (IN SECONDS)
AS A FUNCTION OF FAMILIARITY, ATTITUDE SIMILARITY,
NUMBER (ALONE OR CONFEDERATE) AND SEX OF BYSTANDER. . 60
3 ANALYSIS OF VARIANCE OF BEHAVIORAL ONLY HELP AS A
FUNCTION OF FAMILIARITY, ATTITUDE SIMILARITY,
NUMBER (ALONE OR CONFEDERATE) AND SEX OF BYSTANDER. . 63
4 ANALYSIS OF VARIANCE OF VERBAL ONLY HELP AS A FUNCTION
OF FAMILIARITY, ATTITUDE SIMILARITY, NUMBER (ALONE
OR CONFEDERATE) AND SEX OF BYSTANDER. . . . ... 66
5 ANALYSIS OF VARIANCE OF BEHAVIORAL HELP LATENCY (IN
SECONDS) AS A FUNCTION OF FAMILIARITY, ATTITUDE
SIMILARITY, NUMBER (ALONE OR CONFEDERATE) AND
SEX OF BYSTANDER. . . . . . . . .... . 69
6 ANALYSIS OF VARIANCE OF VERBAL HELP LATENCY (IN SECONDS)
AS A FUNCTION OF FAMILIARITY, ATTITUDE SIMILARITY,
NUMBER (ALONE OR CONFEDERATE) AND SEX OF BYSTANDER. . 72
7 ANALYSIS OF VARIANCE OF SATISFACTION WITH RESPONSE AS
A FUNCTION OF FAMILIARITY, ATTITUDE SIMILARITY, NUMBER
(ALONE OR CONFEDERATE) AND SEX OF BYSTANDER . . 76
8 ANALYSIS OF VARIANCE OF FIRST HELP LATENCY FOR HELPING
SUBJECTS (IN SECONDS) AS A FUNCTION OF ATTITUDE
SIMILARITY, NUMBER (ALONE OR CONFEDERATE), AND
SEX OF BYSTANDER. . . . . . . . . .. 82
9 ANALYSIS OF VARIANCE OF BEHAVIORAL ONLY HELP FOR HELPING
SUBJECTS AS A FUNCTION OF ATTITUDE SIMILARITY, NUMBER
(ALONE OR CONFEDERATE) AND SEX OF BYSTANDER . . .. 84
10 ANALYSIS OF VARIANCE OF VERBAL ONLY HELP FOR HELPING
SUBJECTS AS A FUNCTION OF ATTITUDE SIMILARITY,
NUMBER (ALONE OR CONFEDERATE) AND SEX OF BYSTANDER. . 85
11 ANALYSIS OF VARIANCE OF SATISFACTION WITH RESPONSE
FOR HELPING SUBJECTS AS A FUNCTION OF ATTITUDE
SIMILARITY, NUMBER (ALONE OR CONFEDERATE) AND
SEX OF BYSTANDER. . . . . . . . . ... 89
12 PEARSON r CORRELATION BETWEEN HOMETOWN SIZE AND
BYSTANDER INTERVENTION. . . . . . . . .. 94
13 PEARSON r CORRELATION BETWEEN BYSTANDER DEMOGRAPHIC
VARIABLES AND BYSTANDER INTERVENTION. . .. . .. 96
14 PEARSON r CORRELATION BETWEEN SOCIAL DESIRABILITY SCORES
AND BYSTANDER INTERVENTION. . . . . . . . 99
15 PEARSON r CORRELATION BETWEEN SEX OF BYSTANDER AND
BYSTANDER INTERVENTION. . . . . . . . .. 101
16 PEARSONr CORRELATION BETWEEN BYSTANDER-VICTIM
INTERPERSONAL ATTRACTION AND BYSTANDER
INTERVENTION. . . . . .... ....... .102
LIST OF FIGURES
1 The Effect of First Help Latency in Seconds as a
Function of Attitude Similarity and Number
(All Subjects). . . . . . . . .... . 61
2 The Effect of Behavioral Only Help as a Function of
Attitude Similarity and Number (All Subjects) . .. 64
3 The Effect of Verbal Only Help as a Function of
Attitude Similarity and Number (All Subjects) . .. 68
4 The Effect of Behavioral Help Latency as a Function of
Attitude Similarity and Number (All Subjects) . .. 71
5 The Effect of Verbal Help Latency as a Function of
Attitude Similarity and Number (All Subjects) .... . 74
._6 The Effect of Satisfaction with Response as a Function
of Attitude Similarity and Number (All Subjects). . 77
7 The Effect of Satisfaction with Response as a Function
of Familiarity and Sex of Subject (All Subjects). . 79
8 The Effect of Satisfaction with Response as a Function
of Familiarity, Attitude Similarity, and Sex of
Subject (All Subjects) . . ... .. ... . 80
9 The Effect of Behavioral Only Help as a Function of
Attitude Similarity and Sex of Subject (Helping
Subjects) . . . . . . . .... ..... .86-
10 The Effect of Satisfaction with Response as a Function
of Attitude Similarity and Number (Helping Subjects).. 90
Abstract of Dissertation Presented to the Graduate Council
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy
HELP THY NEIGHBOR: A STUDY OF BYSTANDER
INTERVENTION IN EMERGENCIES
Ellen Weiss Williams
Chairman: Dr. Marvin E. Shaw
Major Department: Psychology
The area of bystander intervention, one of the numerous out-
growths of research in helping behavior, has generated considerable
interest in.the past few years. Initial investigations focused on
developing a conceptual framework of the bystander intervention
paradigm (BIP), the effects of number of bystanders or diffusion
of responsibility, and the effects of social influence processes
on helping. More recently, work has been done on the effects of
ambiguity and geographical location and their effects on the
The.present study was designed to incorporate previous research,
as well as to test two new hypotheses, within the framework of
Thibaut and Kelley's theory of interaction outcomes. The purpose
of reevaluating previous experimental BIP results within the
reward/cost framework was not intended to devalue their impli-
cations, but rather to eliminate some of the confusion generated
by conflicting interpretations.
Using this framework, diffusion of responsibility, for example,
became an instance of relationship evaluation; situational ambiguity
was redefined in terms of bystanders' desires for positive outcomes,
and differential helping rates based on geographic location were
reexamined in terms of bystander individual differences in CL.
Based on the notion of reward/cost, it was predicted that;
1. Bystander intervention would occur more frequently and
more rapidly when a bystander perceived himself to be
attitudinally similar rather than dissimilar to the
2. Bystander intervention would occur more frequently and
more rapidly when a bystander is familiarized with the
emergency setting prior to the emergency than when he
is not (Environmental Familiarity).
Data were collected on 128 college students at the University
of Florida. In addition to speed and mode of helping, demographic
characteristics and social desirability scores were obtained for
each subject. Comparisons were also made between male and female
bystanders as well as between bystanders who were alone with those
in the presence of two non-reacting confederates.
Results indicate that environmental familiarity and sex of
bystander do not affect bystander intervention. Attitude similarity
and group size, however, were strong determinants of helping. With
regard to all analyses, it was the single bystander who was both
most likely to aid the victim, and the one to do so in the shortest
amount of time. The data also indicate that increased attitude
similarity between the victim and bystander facilitated certain modes
of helping behavior.
An interesting attitude similarity by number interaction occurred
in a number of analyses. It was found that while high similarity
subjects were most efficient when alone, the low similarity bystander
was most likely to intervene when placed with two non-reacting con-
federates. In addition, no significant correlations were obtained
between bystander intervention and social desirability.
One of the major weaknesses of the bystander intervention area
is that, until recently, researchers have been concerned with why
people don't help (i.e., diffusion of responsibility). In contrast,
the present study attempted to focus on variables which might increase
the probability of helping. Further research along the lines of the
present investigation may open.up new avenues for increasing the
likelihood that bystanders will intervene.
It was 3:20 in the morning, March 13, 1964. Twenty-eight year
old Catherine (Kitty) Genovese was returning home from her job as
manager of a bar. She parked her car in the parking lot near her
Kew Gardens, New York City apartment where she had lived since her
move from Connecticut one year earlier. Miss Genovese noticed a
man at the far end of the parking lot and becoming somewhat nervous,
headed towards a nearby police call box. She got as far as a street
light before the man grabbed her. Miss Genovese screamed, "Oh my God,
he stabbed me! Please help me! Please help me!" Lights went on as
thirty-eight of Miss Genovese's neighbors came to their windows to
see what was happening. From one of the upper windows in the apart-
ment house a man called down, "Let that girl alone!"
The assailant looked up at him, shrugged and walked down toward
a car parked a short distance away. Miss Genovese struggled to her
feet as lights went out. Moments later, the killer returned to Miss
Genovese, now trying to make her way around the side of the building
to get to her apartment. The assailant stabbed her again. "I'm
dyingl" she shrieked. "I'm dying!"
Windows :.ere opened again and lights went on in many apartments.
The murderer got into his car and drove away as Miss Genovese
staggered to her feet. It was 3:35 A.M.
Once again, the assailant returned. By then Miss Genovese had
crawled to the back of the building where the doors to the apartment
house held out hope of safety. The killer tried the front door, but
found that she wasn't there. At the second door he saw her slumped
on the floor of the stairs. He stabbed her a third time, fatally.
One half hour and thirty-eight witnesses later, Kitty Genovese was
It was 3:50 by the time the police received their first call
from a man who was a neighbor of Miss Genovese. In two minutes
they were at the scene. The man explained that he had called the
police after much deliberation. He had phoned a friend in Nassau
County for advice and then he had crossed the roof of the building
to the apartment of an elderly woman to get her to make the call.
"I didn't want to get involved," he told the police (New York
Times, March 28, 1964).
The bizarre behavior of these thirty-eight murder witnesses
would be much easier to accept if it were unique. Unfortunately,
however, it Is not. Observe:
Eleanor Bradley was walking on crowded Fifth Avenue
in New York City when she suddenly tripped and broke
her leg. She lay on the sidewalk crying for help while
hundreds of people passed by, but for 40 minutes no one
stopped. (Freedman, Carlsmith and Sears, 1970, p. 413)
Seventeen year old Andrew Mormille was stabbed in the
stomach in a New York subway. His assailants fled,
leaving him bleeding badly. None of the 11 passengers
in the subway car helped him and he bled to death.
(Ibid, p. 414)
An 18 year old switchboard operator was raped and beaten in
her office in the Bronx (New York City). She eluded her
assailant and rushed out into the street, naked and bleeding.
It was during the day and a crowd of 40 people gathered. No
one, however, helped her when the rapist attempted to drag
her back into the building. (Ibid, p. 414)
Why then, didn't these bystanders help? Most explanations offered
have centered around the idea of bystander apathy. Some critics blame
alienation caused by industrialization. Others cite depersonalization
resulting from urbanization. Still others claim we have been dehumanized
by our cold society. While these explanations may satisfy some, they do
little in helping to understand the nature of bystander non-intervention
and are, therefore, of little help to the behavioral researcher. The
question remains, what factors affect bystander intervention?
All cases of bystander apathy have been subsumed under the general
heading bystander intervention. Despite the differences in circum-
stances surrounding each incident, certain features remain constant.
A breakdown delineation of the components in this paradigm, therefore,
In an attempt to understand the nature of bystander apathy, many
researchers have applied existing social psychological theories to the
bystander paradigm. In a second section of this chapter, a brief
introduction to Thibaut and Kelley's interaction outcome theory, a
theory not yet applied to this area, will be reviewed so that one may
better understand its relevance to bystander intervention.
Following this, the bystander intervention paradigm will be
reformulated under Thibaut and Kelley's theoretical framework. The
present author believes that this theory can help account for some of
the conflicting interpretations arrived at by previous research in this
In the fifth section, a comprehensive literature review is
presented, reevaluating prior findings within the interaction-
outcome framework. The research presented falls into three major
areas with reference to bystander intervention; factors affecting
the relationship among bystanders, situational determinants of the
bystander intervention paradigm, and demographic and physical
characteristics of the victim.
The sixth section reviews previous research leading directly to
the present investigation; specifically, bystander/victim interaction,
attitude similarity between the bystander and the victim, bystander/
environmental interaction, and environmental familiarity.
Finally, hypotheses and expectations relevant to the present
experiment are formally stated.
Bystander Intervention Paradigm
Kitty Genovese, Andrew Mormille, Eleanor Bradley, and others all
suffered in seemingly senseless tragedies. Although each incident was
unique, it is possible to extrapolate and examine a number of factors
which were common to all, and have come to be known as the bystander
intervention paradigm (BIP). Each incident involved four basic com-
ponents: there is the victim of the incident, the bystander(s), the
incident itself, generally referred to as the emergency, and the
intervention of help-giving process.
The victim is the person injured or harmed. All injury or harm
implies either physical or psychological abuse or discomfort. This
injury can be brought about by some act (eg., falling off of a high
object), condition (e.g., seizure caused by epilepsy), agency (e.g.,
some other person physically harming the victim as in an assault),
or circumstance (e.g., smoke pouring into a room through a vent).
Whether the victim expects and prepares for such an emergency (e.g.,
possesses a fire extinguisher) or not is irrelevant. While ex-
pectation and preparation can minimize the damage caused by the
emergency, it cannot keep it from occurring. In traditional by-
stander intervention studies, however, (Darley and Latane, 1968;
Latane and Darley, 1968) the victim is generally unprepared for the
The victim and the bystander (to be discussed shortly) may or
may not be acquainted prior to the emergency. They may become
acquainted with one another prior to the actual emergency, they may
be well acquainted with one another, or they may be total strangers
to one another, and still be viewed as victim and bystander. It is the
role rather than the relationship that delineates the victim from the
The bystander is an individual who is near to but is not part of '
the initial emergency. In the intervention paradigm, he is the person
witnessing the emergency situation. Witnessing refers to being
physically in the vicinity (not necessarily within sight) of the
victim so that he is aware that something unusual is occurring. The
bystander, prior to the emergency, is unaware that such is about to
One of the most distinctive aspects of an emergency is that it
involves threat or harm. Even if an emergency is handled successfully,
usually it is the victim (not the bystander) who is the better off
after intervention. Therefore, from the bystander's point of view,
there are few rewards for successful action in an emergency. At
worst, an emergency can bring physical harm not only to the original
victim but also to anybody who gets involved in the situation, namely,
the bystander himself. At best, the bystander will receive some rein-
forcement in the form of either financial gain, public recognition
(via the news media), or some intrinsic feeling of having done well.
This notion of possible harm puts pressure on bystanders to ignore
the emergency, to distort their perceptions of it, or to underestimate
their responsibility for getting involved.
A second distinctive feature .of an emergency is that it is an
unusual or rare event. An unusual or rare event is one that has
either never or hardly ever occurred in the life of the bystander
before. In addition to being rare, emergencies differ widely from
one another. That is, even if a bystander had been present during
some prior emergency (an automobile accident) it may bear no rela-
tion to the present one. Each emergency presents a different problem
and each requires a different type of action. Having observed a fire
would not necessarily equip one with the knowledge of how to act
during an epileptic seizure.
A fourth characteristic of an emergency is that it usually is
unforeseen; that is, it occurs suddenly and without warning. Neither
the bystander nor the victim have any prior knowledge that the
emergency might occur. Therefore, the bystander does not have the
opportunity to think through in advance what course of action should
be taken during an emergency. Since the victim becomes incapacitated
during the emergency, the bystander, even if he decides not io ;ct,
must do his thinking in the immediacy of the situation. If there is
more than one bystander present (as there is in many emergency
situations) these bystanders have no opportunity to consult others
not present as to the best course of action or to alert experts who
are especially equipped to deal with emergencies (e.g., police,
firemen, doctor). The bystander to an emergency is oftentimes left
to his own resources.
The fifth and final feature of an emergency is that it requires
immediate action. If the seizure, or fire, or assault is not dealt
with at that moment; the situation may deteriorate, the seizure may
get worse, the fire may spread, the victim may be murdered. The
necessity for instant action prevents any bystander confronted with
it from slowly and carefully weighing the different courses of action.
In summary, an emergency situation is one which involves harm
or threat. It is a distinctively .unusual or rare event which often-
times occurs suddenly and without warning, and requires immediate action.
It is, therefore, natural that most emergencies appear somewhat ambiguous
to the bystander.
The Intervention Process
To intervene means to come between two things or to interfere
some ongoing action. Within the BIP, intervention refers to any
helping action taken by the bystander. The two things a bystander
could come between might be the victim and his assailant, or simply
the victim and his pain. Similarly, a bystander who extinguishes a
fire is interfering with some ongoing action whether a victim is
present or not.
A bystander always has the choice of whether or not to inter-
vene. If the bystander does decide to intervene, he must also decide
what form of assistance to give. Should he rush in directly and try
to help the victim or should he go and call someone better qualified
for that job. Intervention, therefore, requires singling out a
particular course of action from a rather wide choice in a very limited
amount of time. Latane and Darley, two of the earliest researchers in
this area, observed, "...failure to intervene may result from failing
to notice an event, failing to realize that the event is an emergency,
failing to feel personally responsible for dealing with the emergency,
or failing to have sufficient skill to intervene" (1969, p. 248).
In summary, the bystander intervention paradigm involves a victim
and a bystander, both possibly unprepared and unaware of the emergency
that is about to take place. The victim, in what might be interpreted
as an ambiguous situation, is in the process of being physically or
psychologically harmed. The bystander, if he so chooses, intervenes by
personally giving aid or sending for the same.
A Theory of Interaction Outcomes
Many theories have been applied to account for bystander behavior
(e.g., social responsibility, reciprocity, and interdependence theory).
One relevant social psychological theory which has not been applied to
explain the bystander effect is Thibaut and Kelley's reward/cost theory
of interaction outcomes. Thibaut and Kelley (1959) developed the theory
of interaction outcomes to account for the behavior of two or more
individuals in an interaction relationship where each member is depen-
dent on every other member for the achievement of positive outcomes.
Analysis of Interaction
According to Thibaut and Kelley, "...the essence of any inter-
personal relationship is interaction. Two individuals may be said
to have forried a relationship when on repeated occasions they are
observed to interact. By interaction it is meant that they emit
behavior in each other's presence, they create products for each
other. In every case we would identify as an instance of interaction
there is at least the possibility that the actions'of each person
affect the other" (1959, p. 10). In addition, an interaction will be
continued only if all participants in that interaction have been
reinforced as a function of having participated (positive outcomes).
Such reinforcements may take the form of material gain (money,
objects) or psychological gain (more power, status, etc.). For
interaction to continue, not only must participants achieve positive
outcomes, there must also be the maximization of positive outcomes for
each of the participants. That is, each individual tries to achieve
the most reinforcement possible for interacting. Such a goal is
functional both to the individual as well as the group as a whole.
Consequences of Interaction (Rewards/Costs)
Although interaction outcomes can be described in a variety of
ways, Thibaut and Kelley chose the rewards an individual receives and
the costs he incurs, as their measures. Rewards were defined as a
reduction in drive or in need fulfillment usually thought of in terms
of satisfactions and pleasures derived from the interaction. Costs
referred to any factors which made the performance of a behavior
sequence more difficult or impossible. The greater the difficulty
the individual has in exhibiting a particular set of behaviors, the
greater the cost of interaction. They cited anxiety, embarrassment,
physical or mental strain, and competing response tendencies as factors
that increase cost. Therefore, the outcomes or consequences of inter-
action can be stated in terms of rewards received and costs incurred for
each member of the dyad.
The Determinants of Rewards and Costs
Thibaut and Kelley proposed two broad classes of determinants
which together establish the rewards and costs of interaction.
Exogenous determinants referred to factors that are more or less
external to the interaction. Thibaut and Kelley believed that each
individual brings with him into all of his relationships certain
values and abilities which will affect the interaction. They cite
four exogenous determinants which appear to be most significant:
abilities, similarity, propinquity, and complementarity.
The second class of determinants, endogenous determinants, refers
to those factors which are inherent to the relationship itself. For
every response each participant is capable of enacting, there are
other responses which are incompatible (i.e., disturbing, or dis-
tracting) with it. If one member of the dyad enacts behavior al
while the second one enacts an incompatible response b2, the
results will be an increase in costs to produce one or both sets.
In general, whether they take the form of anxiety, embarrassment,
annoyance or increased effort in responding, incompatible response
tendencies increase the cost of behavior and, hence, interaction.
Formation of the Relationship
Obviously, a relationship will never begin unless there is some
initial contact between the dyadic members. In its most simple form,
two people come in contact with one another if they are physically
near one another (proximity). :As in the case of rewards and costs,
there are two broad headings of factors which influence the formation
of a relationship; production of behavior and perception of behavior.
Thibaut and Kelley posited four major factors that affect which
behaviors an individual will produce during the preliminary stages of
interaction. These factors are strangeness, accessibility and cultural
norms, autistic hostility, and autistic friendliness.
In addition to those factors which affect behavior production,
Thibaut and Kelley focused on four factors which affect how behavior
is interpreted. These factors are; availability of cues, the primacy
affect, organization of perception, and states of the observer.
Evaluation of the RelationshiD
Once initial contact has been made between dyadic members, and
they have sampled some outcomes, there will arise a need within them
for some sort of standard by which to evaluate the acceptability of
of interaction outcomes. Thibaut and Kelley postulate two such
standards called the comparison level (CL)*and the comparison level
for alternatives (CLalt). The CL is used by each member to determine
the attractiveness of the relationship while the decision of whether
to remain in the relationship or not is determined by his CLalt
(dependency on relationship). According to Thibaut and Kelley (1959),
"...CL is a standard by which the person evaluates the rewards and
costs of a given relationship in terms of what he feels he 'deserves.'
Relationships, the outcomes of which fall above CL, would be rela-
tively 'satisfying' and attractive to the member; those entailing
outcomes that fall below CL would be relatively 'unsatisfying' and
unattractive" (p. 21). Similarly, they state (1959) that, "...CLalt
can.be defined...as the lowest level of outcomes a person will accept
in the light of available alternative opportunities. It follows...
that as soon as outcomes drop below CLalt the member will leave the
relationship" (p. 21). The major purpose for postulating two standards
is based on the fact that an individual may be forced to remain in what
he considers an unattractive relationship. Therefore, although outcomes
fall below his CL, he will remain in the relationship based on his CLalt'
In addition, because interaction requires that the two participants be
interdependent, a relationship will be formed only if both members
experienced outcomes are above their respective CLalts'
Thibaut and Kelley viewed the CL as a neutral point on a continuous
scale ranging from dissatisfaction to satisfaction. Any time the inter-
action outcomes fall above this neutral point the interaction will be
viewed as attractive. Conversely, when outcomes drop below the neutral
point the interaction will be unsatisfactory. According to Thibaut
and Kelley (1959), the CL is defined, "...as being some modal or
average value of all outcomes known to the person (by virtue of
personal or vicarious experience), each outcome weighted by its
salience (or the degree to which it is instigated for the person at
that moment)" (p. 81). We see, then, that an individual's CL will be
affected (i.e., change) as he experiences new consequences which
changes his hypothetical average value of outcomes. Likewise, the
CL will be affected by situational variables which change the salience
of specific outcomes.
Individual differences and comparison level.--what individual
differences and comparison levels refers to here is the differences
between individuals in power and control and how this affects their CL.
An individual who in the past has.experienced positive outcomes related
to power and control will view presently unattainable goals as within
his reach, and as such will tend to stress its reward aspects.
Conversely, an individual who sees himself as powerless will continue
to view unattainable goals as such and will tend to stress the high
costs involved in its attainment. In addition, the CL will be higher for
the former than the latter individual.
Power and Dependence
The final major area to be dealt with is concerned with power and
dependence in dyadic relationships. It will be remembered that the
CLalt was defined as the minimum level of outcomes an individual will
accept in view of the alternative opportunities that are available to
him. In the same way that evaluation was described in terms of an
individual's CL, so must power and dependence be explained in terms of
his CLalt. "CL, then is crucial in his attraction to the dyad, but CLalt
is crucial in determining his dependency upon or, conversely, his power
within it" (Thibaut and Kelley, 1959, p. 101). Stated differently,
since interaction implies interdependence, it can be assumed that each
dyadic member, in some way, controls the rewards and costs of the other
member. The CLalt is the standard by which an individual can measure
his degree of power or dependency in the present relationship compared
to that of alternative ones. If the individual discovers that his power
in this relationship is minimal compared to his best alternative rela-
tionship, he will probably dissolve the present partnership, assuming
Thibaut and Kelley (1959) defined power as, "...A's ability to
affect the quality of outcomes attained by B" (p. 101), and as such
described two types of power; fate control and behavior control.
Fate control.--If A, by altering his own behavior can change B's
outcomes, no matter what B does, A is said to have fate control over B.
For example, A is a professor and B is a student in his class. If
grades are based solely on A's personal evaluation of B, A is said to
have fate control over B.
Behavior control.--If A, by altering his own behavior can make it
desirable for B to alter his behavior, A is said to have behavior
control over B (or vice versa). A and B are husband and wife. If A,
who is usually sloppy, begins to clean up the house, it becomes
desirable for B to stop nagging him to do so. Similarly, if B starts
cooking tasiter meals, she makes it desirable for A to come home for
dinner. Each is therefore said to have behavior control over the other.
Thibaut and Kelley discuss the consequences of power in terms of
both the dyad as a whole and the individual members. They believe
that if a dyad is highly interdependent, the members will wield high
and equal amounts of power over one another and as such will be highly
cohesive. The individual member of a dyad who controls the greater
power can look forward to considerable influence over the values and
attitudes of his partner. Some of the strategies Thibaut and Kelley
suggest for increasing power in a dyad include; improving one's ability
to deliver rewards to his partner, building up the value of one's
product (i.e., rewards one can offer), and reducing the alternatives
available to the other member of the dyad.
Having spent considerable time describing and exploring Thibaut
and Kelley's reward/cost theory of interaction outcomes, it is now
appropriate to return to the original impetus for such a discussion;
that is, to incorporate the paradigm of bystander intervention into
BIP: A Case for Interaction Outcome Theory
It will be remembered that the bystander intervention paradigm
involves a victim and a bystander, both possibly unprepared and unaware
of the emergency that is about to take place. The victim, in what
might be interpreted as an ambiguous situation is in the process of
being physically or psychologically harmed. The bystander, if he
so chooses, intervenes by giving aid or sending for the same.
It will also be remembered that according to Thibaut and Kelley.
C1959), "...The essence of any interpersonal relationship is inter-
action. By interaction it is meant that they (the dyadic members)
emit behavior in each others presence or they communicate with each
other (where)...there is at least the possibility that the actions
of each person affect the other" (p. 10). In other words, for inter-
action to occur, each participant in an interdependent relationship
must have the ability to affect the other's outcomes (i.e., rewards,
payoffs, reinforcements, etc.). If the relationship between the victim
and the bystanders in the bystander intervention paradigm is truly one
of interaction, then this relationship must meet two criteria; the
members must emit behavior in each other's presence or communicate
with each other, and each member must have the ability to affect the
The BIP meets the first criteria of interaction; that of emitting
behavior in each other's presence or communicating with one another.
In all naturally occurring instances (i.e., Kitty Genovese, Andrew
Mormille, Eleanor Bradley) of this phenomenon, both the victim and
the bystander were responding in each other's presence. In the case
of Andrew Mormille, all of the subway passengers watched as he was
stabbed in the stomach. Although he pleaded for help, none was offered.
The reaction of these bystanders, although it was one of non-intervention,
was emitted in his presence. Similarly, the New York Times account of
the Kitty Genovese murder (March 28, 1964) attests to the fact that
although no one actually helped, all of the thirty-eight neighbors
came to their windows to see what was happening and one man actually
called down to the killer to "Let that girl alone!" In essence, then,
each bystander by being at his window was reacting in the presence of
Miss Genovese. In addition, since Miss Genovese was calling to these
bystanders for assistance, she was both aware of and reacting to
In most experimental variations of the BIP, the same is true.
That is, the bystander and the victim are responding in each other's
visual presence (Piliavin, et al., 1969; Latane and Darley, 1968).
A number of other bystander effect studies are designed in such a way
that although behavior is not emitted in each other's visual presence,
the bystander and the victim do communicate with one another verbally
(Darley and Latan6, 1968; Latane and Rodin, 1969; Clark and Word, 1972;
Levy, Lundgren, Ansel, Fell, Fink and McGrath, 1972). In all instances
of the BIP then, whether real or experimental, the first criteria of
interaction is met.
The second criteria, that of interdependency for the attainment of
positive outcomes is somewhat more subtle. It is obvious that the
bystander has the ability to affect the victim's rewards. If the by-
stander comes to the aid of the victim, he is either eliminating or at
least alleviating the victim's suffering. It is also true, however, that
the victim can affect the positive and negative outcomes (i.e., rein-
forcements) of the bystander. Thibaut and Kelley include as reinforcement
of interactions both material loss or gain (money, objects) or psycho-
logical loss or gain (power, status). Therefore, the victim at least
has the possible power to reward or punish bystander intervention via
financial reimbursement or other forms of material gain as well as
affecting the bystander's self-image for his actions. For interaction
to continue, according to theory, not only must participants achieve
positive outcomes, there must also be the maximization of positive
outcomes for each of the participants. That is, each individual tries
to achieve the most reinforcement possible for interacting. Therefore,
the greater the likelihood of reinforcement as perceived by the by-
stander, the greater the possibility that he will continue interacting
with the victim via intervention. We see then, that for all intents
and purposes, the BIP is an instance of interaction. As such, the
consequences of bystander/victim interaction can be viewed in terms
of rewards and costs.
In order to better understand the scope of Thibaut and Kelley's
interaction outcome theory for the BIP, a reevaluation of previous
experimental research is attempted within the reward/cost framework.
The purpose for such an endeavor is not to devalue the implications of
such studies, but rather to eliminate some of the confusion generated
by conflicting interpretations.
Since its inception in 1968, considerable attention has been
directed to the study of helping behavior in the context of bystander
intervention. Research can be divided into three major areas: factors
that affect the relationship among the bystanders, situational deter-
minants of the BIP, and demographic and physical characteristics of
Relationship Among Bystanders
The effect of the number of bystanders present on subsequent
helping as well the degree of intimacy among bystanders can be viewed
as information related to the relationship among bystanders. As such,
both diffusion of responsibility among bystanders and intervention
by friend versus stranger bystanders are included here.
Diffusion of responsibility.--The earliest research in the field
of diffusion of responsibility was carried out by Darley and Latane
(1968). They claimed that diffusion of responsibility was the major
cause of non-action in an emergency situation.
Diffusion of responsibility takes its origins from two major
mechanisms that explain mob action (Brown, 1965). The first is the
feeling of anonymity that comes with being part of a crowd. This
results in a diffusion of the individual's sense of personal respon-
sibility. Because he blends into a large mass of others, each member
of a crowd can feel relatively anonymous and therefore less res-
ponsible for his actions (or inactions), than otherwise would be the
The second mechanism to explain mob action was termed an "impression
of universality" by Allport (1924). According to this view, the member
of the crowd receives the approval of everyone. As a result, whatever
he does seems to him to be the right thing to do.
Applying this to the Kitty Genovese murder, Darley and Latan6
made these observations. Each of the thirty-eight observers by seeing
lights and figures in the other apartment house windows knew that others
were reacting. These facts provide several reasons why any individual
may have delayed or failed to help. The responsibility for helping was
diffused among all thirty-eight observers. There was also a diffusion
of any potential blame for not taking action. And, finally, it was
possible that somebody unperceived, had already initiated helping action.
S Darley and Latane state that when only one bystander is present
in an emergency, if help is to come, it must come from him. Although
he may choose to ignore it (out of concern for personal safety or
desires not to get involved), any pressure to intervene focuses
uniquely on him. When there are several observers present, however,
the pressures to intervene do not focus on any one of the observers;
instead the responsibility for intervention is shared among all the
onlookers and is not unique to anyone. As a result, no one helps.
For these reasons, Darley and Latane hypothesized that the more
bystanders there are to an emergency, the less likely, or the more
slowly any one bystander will intervene to provide aid. In their
study (1968), college students agreed to participate in an intercom
discussion where each participant would remain totally anonymous.
Subjects were led to believe that they were either part of a two,
three, or six member panel. In each case, there was only one true
subject and one victim. All other members were.simply voices pro-
vided on tape by the experimenter. Midway into the discussion, the
victim began to have an epileptic seizure. Presumably the experimenter
was not present and it was up to the real subject to either summon
him or help the victim directly. The experimenter, as his dependent
measure, clocked the speed of the subject's response. The results
supported Darley and Latane's hypothesis; 85 percent of the subjects
in the alone condition (subject-victim) went for help as compared to
62 percent in the three-man groups and only 31 percent in the six-man
groups. In addition, subjects in the alone condition responded more
quickly (52 seconds) than their six-man counterpart group (166 seconds).
In this same study, Darley and Latane found no significant
difference in helping behavior between males and females, or between
experts and non-experts (pre-med vs. arts and sciences students). In
addition, there were no reliable correlations between helping behavior
and personality traits as measured by scales of Machiavellianism,
anomie, and authoritarianism (Christie, 1964), Crowne-Marlowe's social
desirability scale (1964), and Daniel and Berkowitz' social responsi-
bility scale (1964). They also found no correlations between helping
behavior and socio-economic background.
One interesting aspect of this study was the reactions of those
subjects who did not help. The subjects exhibited great concern about
the victim when the experiment was over. Invariably such subjects
checked with the experimenter to make sure the victim was all right.
Why then, didn't they respond? Darley and Latan6 contend that such
subjects had not decided not to respond; rather they were still in a
state of indecision and conflict concerning whether or not to respond.
The emotional behavior of these non-responders, according to Darley
and Latane, was a sign of their continuing conflict, a conflict that
other subjects resolved by helping.
If Darley and Latane's study is analyzed within the framework
of Thibaut and Kelley's interaction outcome theory, diffusion of
responsibility can be seen as an instance of relationship evaluation.
In the case of the lone bystander, the only relationship available
to him involves the victim. Therefore, although his evaluation of
the relationship may fall below his CL, he will remain in the rela-
tionship and continue interaction (i.e., intervention) based on the
fact that this relationship exceeds his CLalt. This is not the case,
however, for multiple bystanders. Once again, interaction with the
victim may not be very attractive, but now each bystander has the
option of engaging in an alternative interaction with any other
bystander. Such an interaction would incur lower costs since it does
not involve the possibility of threat, harm, or increased effort in
responding and, therefore, raises their CLalt. As stated previously,
if an individual's CL falls above outcomes and below his CLalt, he
will discontinue the relationship. What we see then is that in the case
of multiple bystanders, each bystander chooses as his partner in inter-
action any or all of the other bystanders rather than the victim.
The idea of alternative relationships would also help to explain
the results obtained by Levy et al. (1972) involving diffusion of
responsibility in a non-threatening situation. The experiment was
designed to determine if the bystander effect could be produced in a
non-emergency situation; one where neither the bystander nor the victim
was threatened, but one which nonetheless demanded action from the
bystander.. A subject waiting either alone or with one or two confederates
heard an intruder demand entry into the experimental room where subjects
were filling out a questionnaire. Treatments were varied as to whether
intruder demands increased in intensity or remained the same, and
whether confederates appeared to notice the demands or not. Focusing
on response latencies, Levy et al. found helping rate was most rapid
for the alone subjects, followed by subjects participating with one
confederate. Subjects working with two confederates took the most time
in responding. In other words, the bystander effect found by Darley
and Latane was demonstrated for each of the present experimental
treatments, none of which involved fear or a real emergency situation
but each of which involved some intrusion or demand for response on
the part of the naive subject. As with Darley and Latane, bystanders
in this experiment had the option of continuing interaction with one
another or entering into a new interaction with a relative stranger
(victim). Although possible injury to the bystander was not a factor
in helping, interaction did involve an increased effort in responding
(getting up and opening the door). The lone bystander, having no
alternative relationship from which to choose, interacted with the
victim. The decision to help on the part of a bystander, therefore,
may have had less to do with the number of other bystanders present
as it did with the attractiveness and dependency such other relationships
offered compared to one with the victim. In addition to attractiveness
and dependency, the guilt aroused in the lone bystander for failure to
help might actually have increased the cost of not interacting. Helping
the victim eliminates this guilt, thus increasing the reward/cost ratio.
Another case in point concerns the results obtained by Ross (1971).
He found that a bystander paired with two children responded to a
victim's plea more rapidly than one who was paired with two confederate-
peers. He interpreted these results in terms of an increase in respon-
sibility. However, he could not explain why a single bystander responded
faster than his adult/child subject. If looked at in terms of CLalt
the results become clearer. Once again, the alone bystander had no
alternative relationship to consider, hence he helped. The adult/child
subject did have an available other relationship albeit not as appealing
as the one offered to the adult/peer subject; hence the former may even-
tually have chosen the victim for his partner and did so more quickly
and more often than the latter, but not as often or quickly as the
Friends vs. strangers.--Assuming there is more than one bystander
present to an emergency, will the degree of intimacy between bystanders
affect their probability of intervention?
This question of whether friends or strangers are more responsive
bystanders was addressed by Latane and Rodin (1969). Subjects were
asked to wait either alone, with a friend, or with a stranger (con-
federate or naive) to participate in a market research study. As they
waited, subjects heard someone .fall and apparently injure herself in
the room next door. Whether the bystanders tried to help and how long
they took were the main dependent variables. Results indicated that
two friends helped the same percent of time (70 percent) as single
subjects, as compared to 40 percent for two strangers and only 7 percent
in the confederate condition. This experiment may also be viewed within
a reward/cost framework. A naive bystander, paired with a stranger, must
choose between a continued interaction with this stranger or a new inter-
action with a victim. This new victim interaction involves a higher cost
and no guarantee of greater rewards than the subject's present rela-
tionship. The same might not be true of two friends. Friends might not
feel forced to choose between each other or the victim. Their relation-
ship to one another may be independent of victim-interaction and as such
they may consider themselves responding as a single individual. As such,
they are similar to the alone bystander. Considered this way, it is not
surprising that Latane and Rodin (1969) found identical degrees of
helping (70 percent) between two friends and single bystanders. Rather
than interpreting this percentage as inhibition due to friendship as
Darley and Latane did, it indicates joint action for a single response.
Latane and Rodin themselves state (1969), "...Friends...often dis-
cussed the incident and arrived at a mutual plan of action" (p. 200).
The second explanation of why friends are more willing to inter-
vene than strangers concerns the notion of cost incurred in interaction.
Thibaut and Kelley assert that the greater the difficulty an individual
has in exhibiting a particular set of behaviors, the greater the cost
of interaction. They cite anxiety, embarrassment, physical or mental
strain and competing response tendencies as factors that increase cost.
Referring back to the results of Latane and Rodin, one finds these
experimenters saying, "...It may be that people are less likely to fear
possible embarrassment in front of friends than before strangers" (1969,
p. 200). In other words, a decrease in possible embarrassment would
decrease the relative costs of victim interaction and would increase
the probability of intervention. If fears of embarrassment is less of
an issue between friends than strangers, one would expect, based on the
desire for positive outcomes, that friends would interact with a victim
more often than two strangers. Such is the case.
Situational Determinants of the BIP
Factors inherent to the BIP which affect the probability of by-
stander intervention can be discussed as situational determinants.
Two of the major areas studied deal with the effect of situational
ambiguity and geographic location on subsequent helping.
Ambiguity.--As noted earlier, ambiguity is defined as an event
that is uncertain or vague. With reference to the BIP, Latan6 and
Darley (1968) became interested in how bystanders interpret an
ambiguous event. When faced with such an event, they suggested the
individual bystander is likely to look at the reactions of the people
around him and be influenced by them. It was predicted that the sight
of other non-responsive bystanders would lead the individual to inter-
pret the emergency as non-serious and therefore not act. It was also
predicted that the dynamics of the interaction process would lead each
of a group of naive onlookers to be misled by the apparent inaction of
the others into adopting a non-emergency interpretation of the event
and a passive role. Darley and Latane termed this "pluralistic
ignorance" (p. 216). These experimenters had subjects fill out a
questionnaire either alone, in groups of three, or with a non-responsive
confederate. Midway through the questionnaire, the experimenter began
to introduce smoke through a vent such that by the end of the experi-
mental period, vision was obscured. Darley and Latane's results support
their predictions. An individual exposed to a room filling with smoke
in the presence of passive others, themselves remain passive (only 10
percent sought help). Groups of three naive subjects were also less
likely to report the smoke (38 percent) than solitary bystanders
Clark and Word (1972) also studied the effects of ambiguity on
intervention. In their study, either one, two, or five naive male
subjects were led to believe that they would be discussing sexual
topics with a female student under the guise that the experimenter
was interested in observing which combination of males would be more
effective in changing her attitude. While waiting for the experiment
to begin, subjects overheard a maintenance man, in the adjoining room,
fall. In the low ambiguity condition he cried out in agony, whereas
no verbal cues of injury were emitted by the victim in the high ambiguity
condition. Clark and Word hypothesized that bystanders would be more
likely to respond to the pleas of an individual in distress than when
no such pleas were heard. Their results supported their hypothesis.
Helping behavior occurred regardless of group size in every case in
which subjects were exposed to an unambiguous emergency. In the high
ambiguity condition, however, bystanders responded in only 30 percent,
20 percent, and 40 percent of the time in the one, two, and five person
groups respectively. In addition, the mean reaction time in seconds
was significantly lower in the low vs. high ambiguous conditions.
Clark and Word interpreted their results as indicating that the
characteristics of the emergency situation are major determinants of
whether an individual or group is likely to respond to the pleas of
an individual in distress. Individuals who are confronted.with a
highly ambiguous emergency situation are less likely to help a victim
than are bystanders who are exposed to a less ambiguous situation,
especially when in the presence of non-reacting others. According to
Clark and Word, "...w'en an emergency is non-ambiguous involving
severe negative consequences to another person with minimal negative
consequences for the persons) who help(s), and when the amount of
effort required for intervention is minimal, derogation of the victim
is not an appropriate response, and diffusion of responsibility is
not likely to occur, individuals will intervene in an emergency
situation" (p. 399).
Although the experiments are quite different, both sets of
researchers (Latane and Darley, 1968; Clark and Word, 1912) reported
experimental results indicating an inverse relationship between
ambiguity and helping behavior. They interpreted these findings in
terms of the demand or emergency stimulus. All four experimenters
believed the more cues of distress emitted by the victim, the less
ambiguity of the demand stimulus, and the greater the degree of
intervention. However, as reported earlier, the results of Levy
et al., (1972) indicated that even in situations involving low ambi-
guity of demand stimuli, multiple bystander intervention is minimal.
The discrepancy between these studies can be somewhat eliminated
when reactions are explained in terms of the desire for positive outcomes.
In the experiment by Clark and Word (1972), subjects overheard
a maintenance man fall and apparently injure himself in an adjacent
room. In the low ambiguity condition, the victim called out for
assistance., high ambiguity subjects did not hear these pleas. For a
naive subject to intervene, he would have to discontinue interacting
with the confederate bystanders. This would involve analyzing the
new victim relationship in terms of rewards and costs. We have already
mentioned that the costs involved in such an interaction can be high
(harm, embarrassment, etc.). What about the rewards? By calling out
for assistance (low ambiguity) the victim is cueing the subject that
he will greatly appreciate interaction and as such, he (the victim)
is increasing the probability of rewards while removing the cost of
embarrassment, thus making intervention more salient. In highly
ambiguous situations this is not the case. While costs are obvious,
comparable rewards are not. Similarly, the increased effort in res-
ponding necessary on the part of the subjects in the Levy et al.
experiment might not be counterbalanced by the rewards offered by
the victim for opening the door. Indeed, perhaps this stranger does
not belong there in the first place, therefore increasing costs even
Despite the finding of these experiments, the question still
remains as to why all of the thirty-eight witnesses to the Genovese
slaying, after hearing her pleas for help, and from the safety of
their own apartments, failed to telephone for help.
The answer may be that in the case of the Genovese bystanders,
in order to eliminate any possibility of retribution of the part of the
assailant, assistance would have to be anonymous eliminating any
possibility of rewards. In essence, then, ambiguity may be less a
matter of the social situation or the emergency stimulus as it is a
means of increasing or decreasing the salience of rewards for inter-
Geographic location.--Another topic related to situational deter-
minants and helping deals with geographic location. When comparing the
results of BIP studies from Darley and Latane with those of Clark and
Word, we find that although the designs were identical and utilized a
comparative population (college students), they observed differential
helping responses. More specifically, intervention occurred more
frequently among Clark and Word's subjects than those of Darley and
Latane. It is important to note that the former's subjects lived in
Tallahassee, a small southeastern university town, compared to the
home of the latter's subjects, New York City. This difference in
geographic location may affect bystanders' individual differences in
comparison level. It is not earth shattering to hypothesize that life
in a large urban center is different from life in a small town.
Therefore, it is not surprising that people who live in such divergent
atmospheres will respond differently to similar situations. One such
situation concerns perception of personal power. Milgram (1970) in
writing about urban life states, "...The interposition of institutions
between the individual and the social world, a characteristic of all
modern society and most acutely present in the large metropolis, has
its negative side. It deprives the individual of a sense of direct
contact and spontaneous integration in the life around him. It
simultaneously protects and estranges the individual from his social
environment" (p. 154).
Similarly, Thibaut and Kelley state that an individual who in the
past has experienced positive outcomes related to power and control
will view presently unattainable goals as within his reach, and as
such will tend to stress the reward aspects of such goals. Conversely,
an individual who sees himself as powerless will continue to view
unattainable goals as such and will tend to stress the high costs
involved in its attainment.
In other words, individuals from large metropolitan areas are
accustomed to having power controlled by anonymous institutions with
whom interaction is very frustrating and hence costly. Therefore, it
is not surprising that they view themselves as powerless in reference to
attaining presently unattainable goals. This in turn would cause them
to stress the cost aspect of ever getting involved. Small town dwellers,
on the other hand, do not have this problem of "overloading" (Milgram,
1970, p. 153) and as such feel more powerful and personally responsible
for their actions. Since interaction in such an environment is more
successful such individuals would tend to stress the rewards rather than
the costs of relationships. Since past experiences based on geographic
location can affect an individual's perception of present outcomes
(Milgram, 1970), one would expect a city dweller to be less prone to
help a person in trouble. If one could empirically measure an individual's
perception of interaction in terms of his emphasis on rewards and costs,
one would expect to find a significant difference in interaction approaches
between these two populations.
Demographic and Physical Characteristics of the Victim
This third area deals with the effect of victim/bystander simi-
larity along demographic lines on bystander intervention. Research
has focused on the state of the victim, his race, and sex.
State of the victim.--The state of the victim refers to his physical
appearance at the time of the emergency. A victim may look conservative
or liberal, rich or poor, sick or healthy, drunk or sober. Each of
these states may affect the bystander's perception of the victim, which
in turn, may influence his intervention decision.
An experiment by Piliavin, Rodin, and Piliavin (1969) focused on
the effect of the state of the victim (drunk vs. ill) on intervention.
In addition, race was a variable. Theirs was a field experiment
conducted in a New York City subway train by teams of four college
students. The emergency occurred within a seven-minute interval
between two subway stations. Each time, the victim began to stagger
and eventually collapsed to the floor of the car. In half of the
cases, a confederate came to the victim's aid. In the other half,
the fate of the victim was solely in the hands of the naive bystanders.
One of the findings of Piliavin et al. was that an individual who
appeared to be ill was more likely to receive aid than one who appeared
to be drunk. In addition, given mixed groups of men and women and a
male victim, men were more likely to help than were women. Given
mixed racial groups, there was some tendency for same race helping
to be more frequent (especially if the victim was drunk rather than
ill). And, finally, they found no relationship between the number of
bystanders and the speed of helping (i.e., no diffusion of responsibility).
These results appear to be contradictory with those obtained by Darley
and Latan6 (1968). There were, however, three major differences between
this study and both the study of Darley and Latan6 and the Genovese
murder. First, the bystanders in this study were literally trapped
in the subway car with the victim for seven minutes. The victims in
previous bystander intervention studies were always in rooms adjoining
that of the bystanders. Therefore, the emergency was always out of
sight and bystanders were free to leave if they so desired. Similarly,
the witnesses to the Genovese murder did so from the safety of their
own apartments, quite a distance away from the street slaying. Second,
there could be no uncertainty (ambiguity) about the victim's needing
help in this study. This was not true for the Darley and Latane study.
And third, there was no possibility of personal injury to a bystander
who helped an unconscious victim, whereas in the Darley and Latan4
study, at least, personal harm was not totally eliminated (perhaps
the victim would attack the helper while having his seizure). It is
even possible that the "I don't want to get involved" attitude of the
Genovese bystanders was an indication of the fear of retaliation by
the murderer or one of his convict acquaintances. Any one of these
differences, and all three acting together, could easily account for
the contradictory results.
A second study on victim state (Emswiller, Deaux, and Willits,
1971) was carried out to determine whether similarity in appearance
between helper and victim would increase the rate of helping behavior.
Victims were dressed either as "hippies" or "straights" (quotes added).
They approached prospective helpers who were deemed as either similarly
or dissimilarly dressed and requested a small favor. Results indicated
a significantly greater number of persons willing to aid someone who
resembled them in appearance. The authors hypothesized that dress
styles provide a basis for assumptions about other areas of similarity
and create a greater willingness to help a similar other.
Race and sex.--The effects of race and sex of the victim on
bystander intervention have typically been studied simultaneously and
as such, will be reported together.
Gaertner and Bickman (1971), in their study of the effects of race
on helping behavior, hypothesized that an individual of one race might
view another individual of his race as similar, and as such, would be
willing to aid him more often than one of a different race. Their
research was conducted in the field and included over one thousand
black and white subjects (helpers). Each subject received what was
supposedly a wrong number telephone call. The caller (victim) was
clearly identifiable as either black or white, by his voice char-
acteristics. The victim explained that he was attempting to reach
his auto mechanic from a pay phone booth which was located on a major
parkway (where his car had just broken down). The victim further
claimed that he had just spent his last dime making this phone call.
The bystander could aid the victim by contacting the car mechanic
for him. Gaertner and Bickman's results indicated that while black
bystanders extended approximately equal levels of assistance to both
black and white victims, white.bystanders helped black victims less
frequently than their white counterparts. Ignoring the race of the
bystander and the victim, the experimenters found that male subjects
helped more often than female subjects.
Wispi and Freshly (1971) also studied the race and sex effect on
degree of helping in a naturalistic setting. A total of 176 black
and white, male and female subjects, aged 20-60 found themselves in
a position to help or not help a young black or white female con-
federate whose bag of groceries had just broken in front of a super-
market. Their results showed that significant sex differences occurred
in helping behavior for the black but not for the white sample and that
women tended to be less helpful toward women of the same race. With
this one exception, there was no racial difference in helping behavior.
These findings run counter to the results of the Piliavin et al. study
(1969) where helpers tended to aid victims of the same race more than
those of a different race.
Bryan and Test (1967), in two field studies, found that black
Salvation Army solicitors received significantly fewer contributions
than did white solicitors, although there was no difference in the
amount of money donated. The effect of race of the victim solicitorr)
would also help to explain Gaertner and Bickman's (1971) results that
white bystanders help black victims less frequently than chance.
Despite the diversity, all of these studies stress the importance
of the role of external cues to helping behavior. In the initial
stages of the bystander/victim relationship, the availability of cues
is limited. Both the bystander and the victim evaluate one another based
on superficial external cues such as physical appearance. Then, working
from this first impression, each individual begins to evaluate the
relationship. Thibaut and Kelley believe that despite cultural norms
against initial self-disclosure, similarity of attitudes, abilities,
needs, etc. facilitate communication and increase the probability of
Returning to the BIP, Emswiller et al. (1971), concluded that
similarity along dimensions such as dress styles and race provide a
basis for assumptions about other areas of similarity and create a
greater willingness to help a similar other. In essence then, our
initial impression of another based on his physical appearance leads
us to either a closer or more distant identification with him. If
someone looks similar to another in dress, race, language, etc. the
probability in the eyes of the first that the second is similar to him
along other dimensions such as values and interests is increased.
Thus, both members are more likely to anticipate a pleasant interaction
and would be more likely to pursue one.
It has already been stated that similarity between interacting
partners is both rewarding and cost reducing at the same time. Therefore,
the results of these studies can be reinterpreted in terms of inter-
action outcomes. When a bystander is confronted by a victim, an
initial first impression is formed based on outward appearance. The
more similar the victim is to the bystander along the various physical
dimensions, the more likely the bystander will assume intrinsic simi-
larities. Similarity, in turn, increases interpersonal attraction
(Byrne and Nelson, 1965) and the salience of positive outcomes for the
bystander (Thibaut and Kelley, 1959). The result, then, is an increase
in the possibility of intervention.
In looking back over the literature reviewed here, one finds that
all of the research has focused on either the bystander, the victim, or
the environmental situation as separate entities. With reference to
the bystander, experimenters have varied; the number of bystanders to
an emergency (alone vs. two or more), the age of the bystanders (adult
vs. children), the sex of the bystander, and their race (black vs.
white). Victims were varied as to their race (black vs. white), and
physical state (drunk vs. ill). The situational determinants that have
been studied range from: geographic location (urban vs. rural), to
degree of ambiguity (high vs. low). Theorists have searched in vain to
account for bystander behavior (i.e., social responsibility, reciprocity,
and interdependence theory). One begins to wonder then, why researchers
have focused on the victim, the bystander, and the environment as
separate entities when their interaction is so eminent and vital to the
entire paradigm? In order for intervention to occur, there must be
some degree of interaction between the victim and the bystander, and
between the bystander and the environment. Since research has shown
that both quantity and quality of human interaction (Festinger,
Schachter, and Back, 1950) are important determinants of behavior,
attention will be focused on bystander/victim interaction first.
Background of the Present Experiment
The earliest reference to bystander/victim interaction can be found
in an article in Nation by Milgram (1964). The article was prompted by
the same incident which prompted all subsequent research in this area,
the Kitty Genovese murder.
Referring to the relationship between Miss Genovese and the thirty-
eight bystanders, Milgram states, "...Many facts of the case have not
been made public such as the quality of the relationship between Miss
Genovese and the community, the extent to which she was recognized
that night and the number of persons who knew her" (p. 604). He goes
on to say, "...in New York City it is not unusual to see a man sick
with alcohol, lying in a doorway; he does not command the least bit of
attention or interest from those who pass by. The trouble here, as in
Kew Gardens, is that the individual does not perceive that his interests
are identified with others or with the community at large" (p. 604).
Unfortunately, the literature pertaining to this phenomenon, lack
of interaction and hence lack of interpersonal attraction between
victim and bystander, is scarce. Subjects in one experiment in this
area (Epstein and Hornstein, 1969) believed that they were involved
in an impression-formation and decision-making task. They were each
confronted with a dilemma; they had to choose between earning money
for themselves while allowing another to be shocked, or foregoing
the profit in order to prevent the victim from being harmed. One-
half of the helpers anticipated a penalty if they chose to allow the
victim to be shocked; the rest held no such anticipation. In addition,
there were three conditions of interpersonal attraction; like, dislike
and (a control) no manipulation. Epstein and Hornstein's results
indicated that subjects (helpers) who liked the victim chose to help
more often when they anticipated punishment for themselves, but chose
to help less often when no punishment was anticipated. As expected,
the reverse was true for helpers who disliked the victims. In other
words, a bystander will help a victim if at the same time he avoids
hurting himself. However, if there are no unpleasant consequences in
store for a bystander who neglects to help a victim, the victim will
receive little or no assistance from the would-be helper.
In another study, working within the framework of the Zeigernik
effect, Hornstein, Masor, Sole and Heilman (1971) found that persons
were more likely to complete the interrupted goal attainment of liked
than disliked others. And, Emswiller et al. (1971) found that by-
standers were more interpersonally attracted to and therefore more
likely to aid victims who were culturally or physically similar than
dissimilar to themselves.
Interpersonal attraction has also been shown to be affected (i.e.,
increased) by having the bystander simply meet the victim prior to
the emergency. Darley and Latan6 (1968), in their epileptic victim
study observed the effects of prior acquaintance with the victim on
degree of bystander intervention. The results were impressive.
Subjects who had met the victim, even though it was for less than
a minute, were significantly quicker to report his attack than any
other subjects in the six-person condition. When questioned later,
these subjects explained their reactions to the emergency. As opposed
to subjects in any other group, some of the bystanders who had pre-
viously met the epileptic victim revealed that they had actually been
able to see him (in their own minds) having the seizure. It appears
that if one can picture a certain specific individual in distress, the
likelihood of bystander intervention increases.
Although the literature on the effects of interpersonal attraction
between the bystander and his victim are scarce, a number of points can
be made regarding the previous section. It will be remembered that
bystander intervention is a particular case of helping behavior. In
all such instances there is a helper (bystander) and a prospective
helper (victim). Previous research on helping behavior has shown that
the level of interpersonal attraction, as measured by physical char-
acteristics, attitude similarity, race, and prior acquaintance affects
the probability of help being offered. Bystander intervention is
different from helping behavior in that the emergency situation is often
ambiguous for the bystander. However, since both require one individual
to come to the aid of another individual, it is not unlikely that
increased interpersonal attraction should affect the bystander much the
same way it affects the helper; that is the greater the interpersonal
attraction between the bystander and his victim, the greater the
likelihood of intervention. One means of enhancing bystander/victim
interpersonal attraction is through increased attitude similarity.
Attitude Similarity Between Bystander and Victim
One hypothesis which has continually been supported is that people
like those who possess attitudes similar to their own (Newcomb, 1947).
It has also been shown that similarity of attitudes increases the degree
of interpersonal attraction (Byrne and Nelson, 1965). Within the frame-
work of helping behavior, attitude similarity functioning as an exogenous
determinant of interaction outcomes, serves to increase the salience of
rewards while decreasing its relative cost. This is accomplished by
the fact that interaction with a similar other decreases the likelihood
of competing response tendencies, strangeness and inaccurate first
impressions while increasing the probability of social support and
autistic friendliness. Similarly, it has been shown (Smith et al.,
1972) that attitudinal dissimilarity hampers continued interaction.
The variable of attitude similarity has been applied to the BIP
with reference to inter-bystander influence (Smith et al., 1972).
As was predicted, perceived attitudinal similarity between bystanders
significantly reduced emergency intervention. Only when bystanders
believed themselves to be attitudinally different from one another
was victim interaction in the form of help giving initiated. It is
surprising, therefore, that the relationship between victim/bystander
attitude similarity and subsequent bystander intervention has not
been explored. It will be remembered that the relative cost of
interrupting an ongoing emergency is high. There is the danger of
threat or harm in addition to possible embarrassment or anxiety
caused by overreacting. The rewards for intervention are less obvious.
Therefore, unless the victim can, in some way, cue the bystander of
forthcoming rewards, the probability of bystander intervention is
minimal. This can be viewed in another way. When the victim and the
bystander have little in common, the bystander perceives that the
victim has little or nothing to offer him in terms of a rewarding
interaction. Hence, the bystander has complete fate control over the
victim. If the bystander helps, the victim is rewarded; if he does
not help, the victim suffers. If, however, the bystander perceives
that the victim has something to offer him (in the sense of being
alike) similarity acts as a trump card for the victim and the rela-
tionship becomes one of behavior control. That is, the victim, by
holding similar attitudes, is potentially more rewarding and can make
it desirable for the bystander to intervene.
Clark and Word (1972) observed that cries for help by the victim
cued the bystander in such a way as to reduce ambiguity, thereby
increasing the likelihood of reward salience. The present author
believes that attitude similarity with the victim as perceived by the
bystander functions much the same way. And, likewise, dissimilarity
in attitudes has the reverse affect. That is, once the bystander
perceives the victim to be similar to himself, interaction becomes
more attractive since the victim now has a greater potential for
rewardingness. When the bystander perceives himself to be dissimilar
to the victim, however, interaction becomes more costly since the
victim, by being different, has lost much of his reward power.
As stated previously, in order for intervention to occur, there
must be some degree of interaction between the victim and the by-
stander and between the bystander and the environment in which the
emergency occurs. Having discussed, in detail, the interaction
between the bystander and the victim it is appropriate to turn our
attention now to bystander/environmental interaction.
Very little research has been done on bystander/environmental
interaction. Those experiments done, exist under the heading of
One study, by Granet (1971) took place in both a New York City
airport and subway station, and involved helping a young man on
crutches. Granet hypothesized that persons who were familiar with
the physical location in which the emergency took place would be more
likely to help the victim. The results on helping supported his
prediction. In both the airport and the subway, there was a significant
correlation (r = .29, p<.05; r = .31, p<.05, respectively) between
familiarity and responding to the emergency.
Granet interpreted these findings to indicate that a person who
is more familiar with the environment is more aware of the way in which
the environment works. He is not overloaded with stimuli, and his fears
of embarrassment or, in the case of the subway, actual physical harm,
have moderated. He may have a greater stake in keeping the environment
safe. Thus, he is more likely to help.
Three developmental studies (Weizman, Cohen, and Pratt, 1971;
McCall and Kagen, 1967; Moseley, Faust, and Reardon, 1970) have also
examined the effects of environmental familiarity. Their results
suggest that subjects are more willing to engage in novel experiences
within familiar settings than when placed in an unfamiliar environment.
Although the research on environmental familiarity and helping
behavior is limited, the implications are clear. Environmental
familiarity, by decreasing a bystander's fears of embarrassment and
actual physical harm, decreases the potential costs of helping. In
addition, the salience of rewards for intervention increase as his
stake in keeping the environment safe increases, thus facilitating a
desire to help.
The Present Experiment
Based on theoretical as well as empirical data, the following
hypothesis has been generated in reference to the affect of bystander/
victim attitude similarity on bystander intervention:
I. Bystander intervention will occur more frequently and more
rapidly when a bystander perceives himself to be attitudinally
similar rather than dissimilar to the victim.
Based on the research of Granet (1971), the following hypothesis
has been generated with respect to bystander/environmental interaction
and its effect on subsequent bystander intervention:
II. Bystander intervention will occur more frequently and more
rapidly when a bystander is familiarized with the emergency
setting prior to the emergency than when he is not.
In addition, the following secondary hypotheses have been
1. Bystander intervention is negatively correlated with
bystander hometown size.
This hypothesis is based on the notion that individual differences
and CL will be affected by a bystander's hometown size, as measured by
population density, and as such, urban bystanders should be less likely
to render aid to a victim than rural bystanders.
2. Bystander intervention is not correlated with age of bystander,
family income, educational background of parents, number of brothers,
number of sisters, and number of siblings in the bystander's family.
This hypothesis is based on the belief that these variables do
not affect an individual's CL and therefore would not enter into the
3. There is no relationship between bystander intervention and
bystander social desirability needs.
Although some experimenters (Darley and Latan6, 1968) have
inferred a relationship between bystander intervention and social
desirability, this author does not expect to observe a significant
correlation between these two factors since social desirability is not
systematically related to interaction outcomes.
4. Bystander intervention is not affected by the sex of the
This hypothesis is based on the results of Gruder and Cook (1971).
In an experiment on sex, dependency, and helping, Gruder and Cook
found no helping effects due to the sex of the potential helper.
The results of their experiment imply that sex differences observed
in previous research were probably due to the sex of the person
receiving help and not to the sex of the person giving it. Since
the victim in the present experiment is a female, equal help from
both male and female bystanders is anticipated.
Subjects participated either alone, or with two confederates
in a study on personality and visual perception. During the
experiment, they heard someone fall and apparently injure herself
in the room next door. Whether they tried to help and how long
they took to do so were the main dependent variables of the study.
One hundred twenty-eight introductory psychology students from
the University of Florida served as subjects (Ss). There were an
equal number (64) of males and females. All Ss were recruited via
a sign-up sheet in the psychology department under the experimental
title Personality and Visual Perception. In order to avoid suspicion
on the part of the Ss, two additional names were added to the sign-
up sheet in the confederate conditions, making it appear that some
students were being tested in groups of three.
The experiment took place at the University of Florida. The
rooms were set up according to the following floor plan.
Room A Room B
All Ss, regardless of condition, were met by the experimenter
in room 217 and led back into the experimental rooms. Room "A"
which became known as the equipment room, contained an old book-
case stacked high with books and papers, a large movie camera on
a hugh tripod (the visual perception apparatus), and an Akai tape
recorder (which was concealed behind books and folders) in addition
to the table and chairs at which the Ss worked. Overall, the room
appeared crowded and disorganized. Room "B" contained three chairs
which faced the one-way mirror. In all conditions, the naive S
sat in the chair designated "S" on the floor plan.
The experiment involved the manipulation of three variables;
familiarity (high versus low) with the emergency setting, attitudinal
similarity (high versus low) with the victim, and number of by-
standers (S alone versus S+ two confederates) to the emergency.
The study, therefore, was a 2 x 2 x 2 design involving a total of
eight conditions. A total of sixteen Ss (eight male and eight
female) were run in each condition. Each S, then, was involved in
one of the following eight conditions:
1 High Familiarity (F)- High Similarity (S)- S Alone
2 High F- High S- S+ 2 Confederates
3 High F- Low S- S Alone
4 High F- Low S- S+ 2 Confederates
5 Low F- High S- S Alone
6 Low F- High S- S+ 2 Confederates
7 Low F- Low S- S Alone
8 Low F- Low S- S+ 2 Confederates
Part I:. Personality
SHigh familiarity vs. low familiarity.--In the High F condition Ss
were directed from room 217 into room A. Upon entering the experimenter
(E) said: "Hello. My name is Ellen Weiss. I am Dr. Severy's graduate
assistant and although he is in charge of this experiment, I run the Ss
for him. (The graduate assistant was dressed casually to give a non-
professional appearance). By the way, please excuse the mess in here
but quite a few of us have to use this room and it tends to get
cluttered after a while. The name of this study is Personality and
Visual Perception. The first phase of the experiment deals with
Personality. I've given you a questionnaire to fill out. In addition
to the cover sheet, this questionnaire contains a series of statements
about different issues and ideas. Each statement is followed by a
scale ranging from 'strongly agree' to 'strongly disagree'. What we
want you to do is to indicate your agreement or disagreement with
each statement by marking the appropriate word or words on each scale.
Take your time and read each item carefully before indicating your
attitude towards it. Are there any questions? In addition, in
accordance with department rules, you are free to leave at any time
if you desire not to continue the experiment."
This final comment was added to give the S the choice of leaving
during the subsequent emergency. The E then went on to say: "While
you're filling out your questionnaire, I'll be setting up this equip-
ment for the Visual Perception phase of the experiment. I'll try and
make as little noise as possible." (While Ss were filling out the
questionnaire, the E subtly commented on the general disorder and
instability of the bookcase, table, etc.).
The questionnaire that all Ss completed contained a cover sheet
and 53 items (Appendix A). In addition to the complete Crowne-Marlowe
Social Desirability Scale, items were selected which pertained to
attitudes toward Negroes, friends, work, etc. Ss in the High F
condition also rated the emergency room on a Semantic Differential
scale, forcing them to attend to the room (Appendix B).
Low F.--Ss in the Low familiarity condition were led directly
into the second room (room B). E made the same comment about the
messiness of room A as they walked through it. Instructions were
identical except this time E said, "Go ahead and fill out the
questionnaire. I'm going next door (room A) to set up the equipment
for the second phase of the experiment. I should be back before
you finish. If you finish before I return- just knock on the door for
High similarity vs. low similarity.--All Ss completed the Person-
ality questionnaire. In the High- S condition, E took the completed
questionnaire from S and said, "I have to wait for the camera to warm
up before we begin phase two." (E then began to flip through the
completed questionnaire. After doing so she comments) "That's really
unreal. I filled out this questionnaire once for Dr. Severy and our
answers are almost identical. Far out!"
In the Low- S condition the procedure was identical except in
this condition E said, "You know, I bet that we're very different
kinds of people. I filled out this questionnaire once for Dr. Severy
and I think we answered lots of these questions differently."
Confederate condition.--Two undergraduate psychology majors
served as confederates. The confederates were instructed to always
position themselves in the chairs designated "C" on the floor plan.
In addition, the same-sex confederate always sat next to S during the
emergency. The confederates were also told to answer the personality
questionnaire at a slower pace than S in order to give E time to make
the appropriate comments.
Part II: Visual Perception
All Ss were instructed to sit in their chairs in room B facing
the one-way mirror. (Ss in the High- F condition were led from:
room A into room B). The exit door was left slightly ajar to alert
Ss as to a means of escape, if that was their choice of action. After
S (and confederates if applicable) were seated, Eproceeded to say,
"First let me close this hallway door. This second phase involves
Visual Perception. I've given each of you a Visual Perception
questionnaire and an answer sheet (Appendix C). In a few minutes a
series of two lights will appear on the mirror along this opening.
What we would like you to do is to write down the approximate distance
between each pair of lights. For example, if you believe the first
two lights are six inches apart, write 'six inches' next to trial one
on the answer sheet in front of you. Do the same for each pair of
lights. There will be a total of 100 trials in all. I'll be working
the lights from the equipment room next door (room A). On the other
side of this panel is a speaker which is hooked up to a microphone
in my room, so I'll be talking to you through a mike. I'll say 'ready,
begin' before each trial to cue you that the lights are coming on. The
lights will be on for one second with a ten second intervel between
trials. Are there any questions? If not, I have a few more things to
do in the equipment room. Why don't you begin answering the Visual
Perception questionnaire and when you're finished with it, we'll begin
E then went back into room A and turned on a pre-recorded tape.
It began with E saying, "I'll be ready in a minute. I just want to
clear some of this junk off the table."
If S listened carefully, he would have then heard E climb on
a chair to stack a pile of books on the bookcase. Even if he were
not listening carefully, he would hear a loud crash and a scream as
the chair collapsed and E fell to the floor..."Oh, my God, my foot...
I...I can't move it. Oh...my ankle." Then E moaned. "I...can't...
get this...thing...off me." She cried and moaned for about a minute
longer, but the cries gradually got more subdued and controlled.
Finally, she muttered something about getting outside, knocked around
the chair as she pulled herself up and thumped to the door. The
entire incident took 125 seconds (Latane and Rodin, 1969).
If S intervened, the post-experimental questionnaire and inter-
view began immediately, If S did not intervene within the 125 seconds,
E waited an additional minute and then entered room B and began the
post-experimental session. In.the confederate conditions, the con-
federates had been instructed as how to react during the emergency.
They were to look up, shrug their shoulders and then continue writing.
If'S asked them for advice or suggestions, confederates always turned
the question back to S. For example, Subject:- "What do you think is
going on in there?" Confederate: "I don't know. What do you think
is going on in there?" Subject: "Do you think we ought to go look?"
Confederate: "Do you think we ought to go look?", etc.
The main dependent variable was whether the bystander (S) took
action to help the victim (E) or not. When E began the tape recorder,
she also began a stopwatch to time the speed of S's response. In
addition to speed of response, Ss were observed for type of intervention.
Ss. could have opened the door dividing rooms A and B; S could have left
room B to find someone else to help; or most simply, S could have called
out to see if E needed help.
After the emergency ended, Ss were given a post-experimental
questionnaire to complete. In it, Ss were asked to rate: the two
rooms (A and B), the E, and the experiment itself on a semantic
differential; their attitudinal similarity to E on a Likert-type
scale; and to describe their reactions to the emergency in a series
of open-ended questions (Appendix 0).
All Ss were then told the true nature of the experiment. Great
care was taken to relieve any anxiety experienced by those Ss who
did not intervene in the emergency by modifying the debriefing to
fit individual needs. Ss were then given their experimental credit
and asked not to discuss the experiment with anyone. All Ss were
also thanked for their participation.
Check on Manipulation
In the post-experimental interview, subjects were asked to
describe what they thought had taken place next door. All subjects
thought that the experimenter had fallen and hurt her foot. Less
than 10 percent of the subjects reported any suspicion that the
emergency was part of the experiment. Despite their suspicions, all
of these subjects said they had to check to be sure. All of the
subjects in the confederate condition reported that they believed
the two confederates to be other subjects.
A check was also performed on the attitude similarity and
environmental familiarity manipulations.
Subjects were asked in the post-experimental questionnaire to
rate the degree of attitudinal similarity between the victim and
themselves on a Likert-type scale. The means were 6.99 for high
similarity subjects and 5.52 for low similarity subjects. A t-test
performed to compare the means proved to be significant [at the
p<.001 level (t = 10.86, df = 1/126)] indicating the manipulation
was successful. Subjects in the high similarity condition perceived
themselves to be more attitudinally similar to the victim than subjects
in the low similarity condition.
Subjects were also asked to rate the emergency room on semantic
differential scales. The mean for high familiarity subjects was 4.50
compared to 4.41 for low familiarity subjects. A t-test performed to
compare the means proved to be non-significant (t = 0.49, df = 1/126),
thus indicating that the environmental familiarity manipulation was not
successful. Subjects in the high familiarity condition did not perceive
the emergency room to be any more familiar than subjects in the low
For purposes of data analysis, therefore, after analyzing mani-
pulation results, a subsequent empirical mean was obtained for familiarity
subjects. This mean (4.45) was then used to redivide subjects into high
and low familiarity groups. Identical analyses were then performed on
these new groupings. The rationale for this procedure was that helping
behavior would be affected more by the subjects' own perception of
familiarity with the emergency setting rather than by any type of
manipulation performed by the experimenter. This was considered a
distinct possibility during the planning stages of the experiment and
was therefore done on an apriori basis.
The major hypotheses of this experiment are:
Hypothesis 1. Bystander intervention will occur more frequently
and more rapidly when a bystander perceives himself
to be attitudinally similar rather than dissimilar
to the victim.
Hypothesis 2. Bystander intervention will occur more frequently
and more rapidly when a bystander is familiarized
with the emergency setting prior to the emergency
than when he is not.
In addition to investigating the effects of bystander/victim
similarity and environmental familiarity, differences in helping
behavior were observed for sex of bystander and for the alone bystander
as compared to one in the presence of two non-reacting confederates
(Darley and Latane, 1968).
The major hypotheses are addressed via four dependent measures.
1. First Help Latency. This score was determined by measuring,
in seconds, how long each subject took to respond to the
victim. Subjects' time was recorded as soon as he either
came into the emergency room or offered verbal assistance
to the victim.
2. Mode of Helping Behavior. Subjects were divided into two
groups; those who offered verbal assistance to the victim
(Verbal Only Help), and those who went directly into the
emergency room (Behavioral Only Help). This was done to
determine if any of the independent variables affect the
type of help offered the victim.
3.. Differential Mode Latency. This measure is similar to First
Help Latency in that subjects were clocked, in seconds, for
response time. Now, however, separate latencies were deter-
mined for both Verbal Only Help and Behavioral Only Help.
4. Satisfaction with Response. In the post-experimental inter-
view, subjects were asked to rate how satisfied they were
with their response, regardless of what it was.
A total of 105 subjects (of the original 128) helped the victim
in some way. It was therefore decided to analyze the data twice.
First, all 128 subjects were included in the analysis regardless of
whether they helped or not. Since the tape recorded emergency ran
for 125 seconds, all subjects who chose not to help were assigned
a latency score of 125 seconds. Similar analyses were then per-
formed excluding those 23 subjects who did not help. This was done
to determine if the independent variables, in addition to affecting
if a bystander helped or not, could also account for differences in
latency and mode of helping for those subjects who did help.
Data Analysis for All Subjects
A general overview of bystander intervention appears in Table 1.
It is apparent that the degree of helping behavior exhibited by subjects
in the present study is relatively high. Comparing these results to
those obtained by Darley and Latan6 (1968), we find that 97 percent
of our alone bystanders offered aid to the victim in less than 30 seconds,
whereas only 70 percent did so in the original design. Similarly, 65
percent of our confederate bystanders helped the victim compared to
only 7 percent tested by Darley and Latane.
The greatest amount of helping occurred in the high familiarity
alone, and the high similarity alone conditions (100 percent) followed
closely by the low familiarity alone and low similarity alone subjects
(94 percent). The least amount of help was exhibited by the high
PERCENTAGE AND LATENCY OF HELP EXHIBITED BY BYSTANDERS
ACCORDING TO GROUP (MANIPULATION)
Group Number Helping (Seconds)
All Alone Subjects 64 97% 22.99
All Confederate Subjects 64 65% 36.89
All High Familiarity Alone 32 100% 22.46
All High Familiarity
Confederate 32 69 39.04
All High Similarity Alone 32 100% 22.55
All High Similarity
*-Confederate 32 56% 34.83
All Low Familiarity Alone 32 94% 23.56
All Low Familiarity
Confederate 32 62% 34.53
All Low Similarity Alone 32 94% 23.46
All Low Similarity
Confederate 32 75% 38.43
similarity confederate bystanders. But here, too, the total inter-
vention, 56 percent, far exceeded any comparable Darley and Latane
group. The present results are more compatible with those obtained
in the Clark and Word (1972) experiment where helping behavior was
observed 100 percent of the time. While these percentages give a
brief overview of the results, more can be said concerning the nature
of these differences.
First Help Latency
Table 2 reports the results of the four-way ANOVA based on the
latency scores, in seconds, of bystander intervention, regardless of
There were no main effects observed for familiarity, similarity
or sex of bystander. Whether the bystander was alone or in the presence
of two non-reacting confederates, however, significantly affected their
speed in rendering aid (F = 44.48, p< .001, df = 1/112). The mean
latency for subjects in the alone condition was 26.18 seconds compared
to 67.18 seconds for confederate bystanders.
The Attitude Similarity X Number interaction approached signi-
ficance (p <.10, df = 1/112) and was the only such interaction to do so.
Figure 1 presents this data graphically.
It appears that although the alone bystander responds more
rapidly than the confederate bystander under both high and low
similarity conditions, it is the high similarity alone subject who
responds fastest (22.55 seconds). Under confederate conditions,
however, it is the low similarity bystander who aids the victim in
the least amount of time (60.07 seconds). It is important to note,
ANALYSIS OF VARIANCE OF FIRST HELP LATENCY (IN SECONDS) AS
A FUNCTION OF FAMILIARITY, ATTITUDE SIMILARITY,
NUMBER (ALONE OR CONFEDERATE) AND SEX OF BYSTANDER
Source df MS F
| 60.0-- Similarity
L 50.0 -
Figure 1. The Effect of First Help Latency in Seconds as a Function
of Attitude Similarity and Number (All Subjects).
however, that this is merely a trend since there is no significant
difference between the high and low similarity latencies.
Mode of Helping Behavior
Tables 3 and 4 report the results of the four-way ANOVA based
on the mode of helping behavior as a function of familiarity,
similarity, sex, and number of bystanders. It will be remembered
that any one bystander, if he is to help, can do so by calling out
to the victim (Verbal Only Help) or by going directly into the
emergency room (Behavioral Only Help). It was assumed by this
experimenter that walking into the emergency room was a qualitatively
stronger index of helping behavior than was calling out.
Behavioral help.--Looking.at Table 3 first, we see that once
again, familiarity, similarity, and sex of bystander were not signi-
ficant predictors, as main effects, of a bystander's decision to help
the victim by going directly into the emergency room. The fact that
a bystander was either alone or with confederates, however, was a
significant predictor of behavioral help (F = 13.22, p<.Ol, df =
1/112). The mean for alone bystanders as compared to confederates
was 1.11 and 1.38, respectively, where 1 indicated behavioral help
and 2 indicated no behavioral help. Based on these results one
would expect the alone bystander more than the confederate bystander
to aid his victim by proceeding directly into the emergency room.
An interaction affect was observed between Attitude Similarity X
Number. Here, as in first help latency, the interaction approached
significance (p <.10, df = 1/112). Figure 2 represents this data
graphically. When the bystander is alone, those subjects in the
ANALYSIS OF VARIANCE OF BEHAVIORAL ONLY HELP AS A FUNCTION OF
FAMILIARITY, ATTITUDE SIMILARITY, NUMBER (ALONE OR
CONFEDERATE) AND SEX OF BYSTANDER
Source df MS F
Familiarity (F) 1 .007
Similarity (S) 1 .070
Number (N) 1 2.258 13.22***
Sex (X) 1 .195
FS 1 .383
FN 1 .007 -
SN 1 .633 3.71+
FX 1 .070 -
SX 1 .383 -
NX 1 .070 -
FSN 1 .070 -
FSX 1 .070
FNX 1 .070
SNX 1 .070
FSNX 1 .007
within Groups) 112 .171
p 4 .10
***p < .01
(1 = Yes; 2 = No) NUMBER
Figure 2. The Effect of Behavioral Only Help as a Function of Attitude
Similarity and Number (All Subjects).
high similarity condition are more likely to go into the emergency
room than subjects in the low similarity groupings. However, when the
bystander is in the presence of two non-reacting others, it is the low
similarity bystander who is most likely to render assistance. Here,.
as in first help latency, the differences between the two are more of
a trend than a significant difference.
Verbal help.--What about those subjects who chose to call out
rather than proceeding into the emergency room? Table 4 examines
the effects of all four independent variables on a bystander's
likelihood to help verbally.
Two main effects are significant in determining verbal help. As
in the two previous analyses, whether a bystander is alone or with
others grossly affects his willingness to verbally help the victim
(F = 47.04, p<.001, df = 1/112). Here, too, it is the alone bystander
( = 1.42) who is more likely to give aid than his confederate counter-
part (X = 1.91). Therefore, it is the alone bystander rather than one
with confederates who is most likely to help both by calling out and
by walking into the emergency room. The differences in F's between
the two modes of helping (13.22 for behavioral help and 47.04 for
verbal help) leads to the hypothesis that the presence of non-reacting
others is more of an obstacle to overcome if one is to call out than
if one is to get up and leave them to aid the victim.
Attitude similarity was also a significant predictor of verbal
help (F = 3.96, p<.05, df = 1/112). It appears that the effect of
higK similarity is to increase the likelihood of calling out to the
victim to see if she is all right (X= 1.59. over low similarity
ANALYSIS OF VARIANCE OF VERBAL ONLY HELP AS A FUNCTION OF
FAMILIARITY, ATTITUDE SIMILARITY, NUMBER (ALONE OR
CONFEDERATE) AND SEX OF BYSTANDER
Source df MS F
Familiarity (F) 1 0.070
Similarity (S) 1 0.633 3.96*
Number (N) 1 7.508 47.04***
Sex (X) 1 0.383
FS 1 0.070
FN 1 0.383
SN 1 0.633 3.96*
FX 1 0.008
SX 1 0.070
NX 1 0.008
FSN 1 01070
FSX 1 0.383
FNX 1 0.008
SNX 1 0.383
FSNX 1 0.008
within Groups) 112 0.159
*p < .05
bystanders (X = 1.73, where 1 equals giving verbal help and 2 equals
no verbal help). This is not to be confused with behavioral help,
since there was no similarity main effect observed there. In other
words, while high attitude similarity between bystander and victim
does not affect behavioral help, it does significantly increase the
likelihood of verbal help.
Once again, the only interaction affect observed occurred between
Attitude Similarity X Number of bystanders (F = 3.96, p< .05, df =
1/112). Figure 3 graphically represents this data.
Here, as in the two previous analyses, it is the high similarity
alone bystander who is more likely to render aid than one who is in
the low similarity condition. :There was no difference observed
between high and low similarity under confederate conditions; neither
were likely to offer aid.
Differential Mode Latency
Since it was apparent that at least two factors (similarity and
number of bystanders, alone'vs. confederates) did affect the mode of
help chosen by the bystander, the next analyses determined if the
latency of help was different across modes.
Behavioral help.--Table 5 reports the results of the four-way
ANOVA based on the latency, in seconds, of bystander behavioral
As was anticipated, number of bystanders was a significant pre-
dictor of behavioral latency (F = 21.16, p <.001, df = 1/112). As
reported in Table 3, it was the alone bystander rather than the
confederate one who was most likely to offer behavioral assistance.
C = Yes; 2 = No)
Figure 3. The Effect of Verbal Only Help as a Function of Attitude
Similarity and Number (All Subjects).
ANALYSIS OF VARIANCE OF BEHAVIORAL HELP LATENCY (IN SECONDS) AS A
FUNCTION OF FAMILIARITY, ATTITUDE SIMILARITY, NUMBER (ALONE OR
CONFEDERATE) AND SEX OF BYSTANDER
Source df MS F
Familiarity (F) 1 140.07
Similarity (S) 1 219.19
Number (N) 1 30624.94 21.16***
Sex (X) 1 2800.33
FS 1 2403.83
FN 1 1195.13
SN 1 4667.09 3.23+
FX 1 342.24
SX 1 3381.50
NX 1 2.98
FSN 1 1517.17
FSX 1 26.91
FNX 1 528.36
SNX 1 151.01
FSNX 1 1837.02
within Groups) 112 1447.26
***p < .001
It follows then, as seen in Table 5, that the alone bystander should
and does render such aid at a more rapid speed than his confederate
counterpart. The average behavioral latency for the alone bystander
was 40.14 seconds as compared to 71.08 seconds for the confederate.
As was also expected, the Attitude Similarity X Number of by-
standers interaction approached significance (p<.10, df = 1/112)
and was the only such interaction to do so. Figure 4 represents this
Following the same pattern as behavioral help, the high similarity
alone bystander renders the most rapid behavioral help (35.41 seconds)
followed by the low similarity alone subject (44.87 seconds). Once
again, however, we find it is the low similarity bystander who helps
more quickly (63.73 seconds) than the high similarity subject (78.42
seconds) under confederate conditions. Although this is more of a
trend than a significant difference, the pattern reoccurs often enough
to warrant attention.
Verbal latency.--Table 6 reports the results of the four-way
ANOVA for verbal help latency as a'function of our four independent
The two variables (similarity and number) which accounted for
differences in amount of verbal help offered also account for the
differences in verbal latency observed. Here again, not only was
the single bystander more likely to offer verbal assistance (Table
4), he was also found to do it in less time (F = 46.37, p<.001,
df = 1/112). While the subject in the presence of non-reacting others
took an average of 115.31 seconds to respond verbally, the alone
> 55.0 /
Figure 4. The Effect of Behavioral Help Latency as a Function of
Attitude Similarity and Number (All Subjects).
ANALYSIS OF VARIANCE OF VERBAL HELP LATENCY (IN SECONDS) AS A
FUNCTION OF FAMILIARITY, ATTITUDE SIMILARITY, NUMBER (ALONE OR
CONFEDERATE) AND SEX OF BYSTANDER
Source df MS F
bystander did so in 65.07 seconds. Along the same lines, it was
observed in Table 4 that high similarity subjects were more likely
than low similarity subjects to help the victim verbally. Now it
is seen that he does so at a more rapid rate (F = 3.87, p<.10,
df = 1/112). High similarity bystanders took an average of 82.94
seconds to respond while low similarity subjects responded in
The expected Attitude Similarity X Number of bystanders inter-
action was observed approaching significance (F = 3.83, p <.10, df =
1/112). These data are represented graphically in Figure 5.
The same pattern observed in verbal help is once again seen in
its latency. A high similarity alone bystander is more likely to
help verbally and does so more rapidly than his low similarity alone
counterpart. However, no differences were observed, either in degree
orlatency, under confederate conditions.
Before proceeding to the last dependent measure, satisfaction
with response, it seems appropriate to summarize the findings thus
far. Data from all 128 subjects were analyzed to determine which,
if any, of the independent variables accounted for differences in
helping behavior. Familiarity with the emergency setting and sex
of bystander did not account for any significant proportion of the
variance of any of the dependent measures. The two most potent
independent variables were the number of bystanders present and
the degree of attitude similarity between victim and bystander. It
was apparent that whether one was measuring the first time a bystander
0 9. 0
Figure 5. The Effect of Verbal Help Latency as a Function of
Attitude Similarity and Number (All Subjects).
offered aid, or whether helping behavior was broken up into its
verbal and behavioral components, it was the alone bystander who
most often and most rapidly offered assistance. In addition,
although attitude similarity did not effect behavioral helping
behavior, it did significantly increase the likelihood and decrease
the latency of verbal help offered the victim.
A consistent Attitude Similarity X Number of bystanders inter-
action occurred. For first help and behavioral latency, high
similarity subjects helped more often and more rapidly when alone,
whereas low similarity bystanders did the same when in the presence
of non-reacting confederates. For verbal helping behavior, no
differences were observed between high and low similarity subjects
under the confederate condition, while under the alone condition,
the high similarity bystander was once again the most helpful.
Satisfaction with Response
Table 7 reports the results of the four-way ANOVA for satis-
faction with response as a function of the-four independent variables.
Three interaction affects were observed, all of which were
significant at at least the p< .05 level. These were: Attitude
Similarity X Number of bystanders (F = 6.79, df = 1/112), Familiarity
X Sex (F = 4.11, df = 1/112), and Familiarity X Similarity X Sex
(F = 4.11, df = 1/112). Figure 6 represents the Attitude Similarity
X Number of bystanders interaction graphically.
Once again, we observe the same interaction between these two
variables that occurred for first help latency and behavioral help.
When the bystander is by himself, the high similarity rather than
ANALYSIS OF VARIANCE OF SATISFACTION WITH RESPONSE AS A FUNCTION
OF FAMILIARITY, ATTITUDE SIMILARITY, NUMBER (ALONE OR
CONFEDERATE) AND SEX OF BYSTANDER
Source df MS F
Familiarity (F) 1 .031
Similarity (S) 1 .125
Number (N) 1 .031 -
Sex (X) 1 .000 -
FS 1 .031 -
FN 1 .000 -
SN 1 2.531 6.79**
FX 1 1.531 4.11*
SX 1 .000 -
NX' 1 .281 -
FSN 1 .125 -
FSX 1 1.531 4.11*
FNX 1 1.125
SNX 1 .281
FSNX 1 .125 -
within Groups) 112 .373
p < .05
**p < .01
(1 = Yes; 2 = No)
Figure 6. The Effect of Satisfaction with Response as a Function of
Attitude Similarity and Number (All Subjects).
the low similarity subject is more satisfied with his response.
When the bystander is in the presence of non-reacting others, however,
it is the low attitude similarity bystander who is most pleased with
his chosen course of action. As mentioned previously, possible
explanations for this continued interaction will be discussed in
the following section.
Figure 7 represents the Familiarity X Sex of bystander interaction
It is worthwhile to note that satisfaction with response is the
only dependent measure in which either environmental familiarity or
sex of bystander showed up as a significant effect.
According to the data, under conditions of low familiarity, female
bystanders rather than males were most satisfied with their response.
The opposite was true under high familiarity conditions; that is, it
was the male bystander who was most contented with his behavior.
Figure 8 represents the Familiarity X Attitude Similarity X
Sex interaction data graphically.
Of all the interactions observed on the data, this is the only
three-way interaction which was significant. It is also a most unusual
one. According to the data, high similarity male and female subjects
showed no difference in their satisfaction with their response, although
low familiarity bystanders were slightly less satisfied (1.31 vs. 1.36,
respectively). There was no significant difference between these two
scores. Low similarity, low familiarity females and low similarity,
high familiarity males were the most satisfied (X = 1.06 for both),
(1 = Yes; 2 = No)
Figure 7. The Effect of Satisfaction with Response as a Function
of Familiarity and Sex of Subject (All Subjects).
S- Low Familiarity
(1 = Yes; 2 = No)
Figure 8. The Effect of Satisfaction with Response as a Function
of Familiarity, Attitude Similarity, and Sex of
Subject (All Subjects).
while low similarity, low familiarity males and low similarity, high
familiarity females were the most dissatisfied (X = 1.50 for both).
Data Analysis for Helping Subjects
As mentioned in the beginning of this section, it was decided
to analyze the data in a second fashion. Any subject who did not
respond to the victim within the 125 second time limit was removed
from the analysis. The question was then asked; disregarding those
subjects who did not help, could our variables account for differences
in helping behavior latencies?
Because of statistical difficulties arising from the now unequal
group sizes, it was decided:
1. To exclude familiarity from these analyses since the familiarity
manipulation was unsuccessful in distinguishing between groups.
2. To limit the analyses to: first help latency, mode of helping
behavior (verbal or behavioral), and satisfaction with response.
A series of three-way ANOVAs were performed to answer the above questions.
From the original population of 128 subjects, 23 were dropped
because they chose not to help the victim in any way. The breakdown
of these 23 non-helping subjects was; 12 female and 11 male subjects,
14 high similarity and 9 low similarity subjects, 2 alone and 21 con-
federate subjects. This left a total of 105 subjects in the new
First Help Latency
Table 8 reports the results of the three-way ANOVA based on the
latency scores, in seconds, of bystander intervention, regardless of mode.
ANALYSIS OF VARIANCE OF FIRST HELP LATENCY FOR HELPING SUBJECTS
(IN SECONDS) AS A FUNCTION OF ATTITUDE SIMILARITY, NUMBER
(ALONE OR CONFEDERATE), AND SEX OF BYSTANDER
Source df MS F
Sex (X) 1 17.781
Similarity (S) 1 7.775
Number (N) 1 389.110 40.29***
XS 1 12.408
XN 1 8.316
SN 1 2.240
XSN 1 23.233
within Groups) 98 9.659
There were no main effects observed for attitude similarity or
sex of subject. Whether a bystander was alone or in the presence of
two non-reacting confederates, however, significantly affected their
speed in rendering aid (F = 40.29, p <.001, df = 1/98). The mean
latency for subjects in the alone condition was 22.99 seconds
compared to 36.89 seconds for confederate bystanders. This affect
was also observed in both two-way ANOVA's. Itappears, then, even
excluding the 21 sugjects who were in the confederate condition and
did not help, that number of bystanders significantly affects latency
of helping behavior.
Mode of Helping Behavior
Tables 9 and 10 report the results of the three-way ANOVA's
based on the mode of helping (either behaviorally or verbally), as
a function of attitude similarity, number of bystanders, and sex
Behavioral help.--Looking at Table 9 first, we see that, as opposed
to the all subject analysis where number significantly affected
behavioral help, none of the three variables alone affected whether
a bystander aided the victim by going directly into the emergency
The Attitude Similarity X Sex interaction approached significance
(p.lO1, df = 1/98) in the three-way ANOVA, and reached significance in
the Sex X Similarity two-way ANOVA (F = 5.09, p <.05, df = 1/103). It
was the only such interaction to do so for this measure. Figure 9
represents these data graphically.
According to the data, female bystanders rendered behavioral help
ANALYSIS OF VARIANCE OF BEHAVIORAL ONLY RELP FOR HELPING SUBJECTS
AS A FUNCTION OF ATTITUDE SIMILARITY, NUMBER (ALONE OR
CONFEDERATE) AND SEX OF BYSTANDER
Source df MS F
Sex (X) 1 .015
Similarity (S) 1 .003
Number (N) 1 .000 -
XS 1 .019 3.44+
XN 1 .005 -
SN 1 .000
XSN 1 .008
within Groups) 98 .006
Sex (X) 1 .005
Similarity (S) 1 .001 -
XS 1 .013 5.09*
within Groups) 103 .003
TABLE 10 -
ANALYSIS OF VARIANCE OF VERBAL ONLY HELP'FOR HELPING SUBJECTS AS
A FUNCTION OF ATTITUDE SIMILARITY, NUMBER (ALONE OR
CONFEDERATE) AND SEX OF BYSTANDER
Source df MS F
Sex (X) 1 .044
Similarity (S) 1 .039
Number (N) 1 .417 27.86***
XS 1 .002
XN 1 .004
SN 1 .025
XSN 1 .036
within Groups) 98 .015
Sex (X) 1 .013
Similarity (S) 1 .043 4.84*
XS 1 .006
within Groups) 103
(1 = Yes; 2 = No)
Figure 9. The Effect of Behavioral Only Help as a Function of
Attitude Similarity and Sex of Subject (Helping
under conditions of low attitude similarity (X = 1.00) but did not
always do so under high similarity conditions. Males, on the other
hand, although never helping as much as their female counterparts,
offered the most assistance under high similarity manipulations
(X = 1.04), helping the least when they were dissimilar to the victim
( = 1.18). The difference between these means, however, was not
Verbal help.--It will be remembered, that in the all subject
analysis, both number and similarity significantly affected a by-
stander's willingness to render aid. That is, the alone
bystander and the high similarity bystander were both more likely
to help by calling out and did so faster than their confederate or
low similarity counterparts, respectively. As indicated in Table 10,
it is the same two variables, number of bystanders and attitude
similarity, which again account for differences in verbal help.
Here too, it is the alone bystander who is more likely to render
verbal assistance than the subject in the presence of non-reacting
others (F = 27.86, p<.001, df = 1/98). On a scale where 1 =.verbal
help, and 2 = no verbal help, the mean for alone bystanders was 1.40
compared to 1.86 for the multiple bystander.
The effect of attitude similarity on verbal help is also con-
sistent with the all subject analysis. That is, high similarity
subjects more than low similarity subjects chose to help the victim
by calling out (7= 1.48 and 1.69, respectively). This difference
was significant at the p .05 level (F = 4.84, df = 1/103) in the