ArticlePDF Available

To Act or not to Act: The Dilemma Faced by Sexual Harassment Observers

April 2005
Academy of Management Review 30(2)

DOI:10.5465/AMR.2005.16387886

Authors:

Western New England University

Efforts to end sexual harassment that rely primarily on target reporting are unlikely to be successful because most targets do not report their experiences. Thus, we explore an alternative mechanism for controlling sexual harassment - observer intervention. We examine observer intervention in sexual harassment using the literature on bystander intervention for guidance. We describe the concept of observer intervention, develop a taxonomy of intervention types, and discuss factors promoting and inhibiting its occurrence.

…

Figures - uploaded by Lynn Bowes-Sperry

Content may be subject to copyright.

Content uploaded by Lynn Bowes-Sperry

Content may be subject to copyright.

TO ACT OR NOT TO ACT: THE DILEMMA

FACED BY SEXUAL HARASSMENT OBSERVERS

LYNN BOWES-SPERRY

Western New England College

ANNE M. O’LEARY-KELLY

The University of Arkansas

Efforts to end sexual harassment that rely primarily on target reporting are unlikely to

be successful because most targets do not report their experiences. Thus, we explore

an alternative mechanism for controlling sexual harassment—observer intervention.

We examine observer intervention in sexual harassment using the literature on

bystander intervention for guidance. We describe the concept of observer interven-

tion, develop a taxonomy of intervention types, and discuss factors promoting and

inhibiting its occurrence.

One of the most interesting aspects of sexual

harassment (SH) phenomena is that victimized

employees often respond passively—for exam-

ple, by denying the harassment, avoiding the

harasser, or treating the harassment as a joke—

rather than directly—for example, by confront-

ing the harasser or reporting the behavior

(Knapp, Faley, Ekeberg, & Dubois, 1997; U.S.

Merit Systems Protection Board, 1995). On the

surface, passive coping strategies seem odd,

given the very negative consequences targets of

harassment often suffer, including physical vio-

lation, psychological harm, lower job satisfac-

tion and organizational commitment, and dete-

rioration of work relationships (Gutek & Koss,

1993). Recent research, however, identifies plau-

sible explanations for this seeming preference

for passive rather than direct coping. Targets

juggle competing goals following a harassment

incident, with their desire to end the harassment

weighed against such objectives as avoiding

reprisal by the harasser and maintaining their

reputation and status in the workgroup (Bing-

ham, 1991; Knapp et al., 1997; Gutek, 1985;

O’Leary-Kelly, Paetzold, & Griffin, 2000; Ragins

& Scandura, 1995). This implies that organiza-

tional efforts to end SH that rely primarily or

exclusively on target reporting are unlikely to

be successful.

Given this, it is useful to explore other preven-

tion or control mechanisms, such as observer

intervention. Observers are individuals who see

harassment occurring but are not directly in-

volved in the incident. Of course, SH is not al-

ways witnessed. Some incidents, particularly

quid pro quo harassment (which requires sexual

compliance in exchange for the retention or at-

tainment of some employment opportunity), may

unfold with only the harasser and target having

knowledge of its occurrence. However, espe-

cially in the case of hostile environment harass-

ment, in which the work climate itself becomes

poisoned, there often are individuals present

who observe the harassment and who might

take action to stop it or prevent future incidents

(O’Leary-Kelly et al., 2000). Specifically, observ-

ers may intervene in varied ways, such as re-

porting cases of witnessed harassment, stop-

ping an unfolding event, or providing negative

feedback to harassers regarding their behavior

(Bowes-Sperry & Powell, 1996, 1999).

Although bystanders often have the power to

change a situation (Clarkson, 1996), their influ-

ence largely has been overlooked in SH re-

search (Cleveland, 1994; Kulik, Perry, &

Schmidtke, 1997). We address this gap by devel-

oping a theoretical framework that identifies (1)

the types of intervention behaviors initiated by

SH observers and (2) the conditions under which

SH observers are likely to choose intervention

over nonintervention. In the following sections

we describe the concept of observer interven-

tion, develop a typology of intervention strate-

gies, and propose a model that describes how

the decision to intervene might occur. Finally,

We thank Gary N. Powell and two anonymous reviewers

for their helpful comments on earlier drafts of this article.

姝Academy of Management Review

2005, Vol. 30, No. 2, 288–306.

288

we present a research agenda based on our

theory and discuss some of the challenges it

poses.

OBSERVER INTERVENTION: EXPLAINING OUR

TERMS

Before we discuss specific forms of observer

intervention, it is necessary to clarify the terms

observer and intervention. Depending on how

broadly it is defined, observer could include

both individuals who hear about an incident of

harassment and those who actually witness it.

Witnessing an incident of SH generally will be a

more compelling experience compared to ac-

quiring second-hand knowledge of it, suggest-

ing that witnesses may experience stronger and

more complex reactions. In addition, individuals

who witness SH have a broader range of actions

available to them, because they can intervene

not only to prevent future incidents of harass-

ment but also in the unfolding event itself.

Therefore, we focus our discussion around ob-

servers who have witnessed SH behavior. This

conceptualization of observers is consistent

with the notion of “bystanders” in the bystander

intervention (BI) literature (e.g., Latane´ & Darley,

1970).

Intervention generally may be conceptualized

as helping behavior. That is, intervention often

will reflect a desire by the observer to provide

assistance. Although helping is not the only mo-

tivation for observer intervention in SH (we dis-

cuss situations in which the observer is acting

in a more self-interested mode later), it is likely

to be a dominant motive. Given this, it is impor-

tant to place observer intervention within the

domain of this other, related concept.

Research categorizes helping behavior along

several underlying dimensions, including role

characteristics and intended beneficiary. In re-

gard to the former, helping behavior can be ei-

ther inrole or extrarole. For example, organiza-

tional citizenship behavior (OCB; Organ, 1988)

is, by definition, extrarole (Van Dyne, Cum-

mings, & McLean Parks, 1996). Alternatively,

other helping (e.g., prosocial behavior; Brief &

Motowidlo, 1986) can include both inrole and

extrarole actions (Van Dyne et al., 1995). Simi-

larly, we expect that observer intervention in SH

can occur when a person’s work role necessi-

tates action or when the individual chooses to

act even when no role requirement exists. Es-

sentially, then, SH observers sometimes help be-

cause they have to and sometimes because they

want to.

In regard to the latter dimension—intended

beneficiary—we also expect observer interven-

tion in SH to reflect the breadth evident in the

more general helping literature. Some helping

behaviors are intended to benefit the organiza-

tion (e.g., OCBs), but others (e.g., prosocial be-

haviors) may be intended to benefit an individ-

ual, a group, or the organization (Van Dyne et

al., 1995). Similarly, the BI literature distin-

guishes multiple motives for intervention. Al-

though observer action often is attributed to

“pure” motives such as altruism (where the in-

tended beneficiary is the target of the negative

event that is witnessed), this literature also of-

fers hedonistic motives such as a desire to re-

lieve one’s own empathic suffering or to look

good in front of others (Latane´ & Darley, 1970).

Taken together, this literature suggests that ob-

servers of SH may intervene with varying bene-

ficiaries in mind. At times their actions may be

intended to assist the harassment target, at

times to build a stronger organization, at times

to alleviate a guilty conscience, and even at

times to help the harasser. It is our supposition,

then, that a variety of beneficiaries and role

expectations can influence observer action.

A TYPOLOGY OF OBSERVER INTERVENTION

IN SH

Given this introduction to observer interven-

tion, we now specifically define the forms that

this behavior might take in SH incidents. As

shown in Figure 1, observer intervention is cat-

egorized along two dimensions: (1) immediacy of

intervention and (2) level of involvement.

The first dimension (immediacy of interven-

tion) distinguishes cases in which intervention

occurs in a current situation (high immediacy)

from interventions that take place at a later

point in time (low immediacy). In some in-

stances observers’ actions are instantaneous

and incident focused—that is, they attempt to

manage the unfolding SH event. In other cases

the observer’s involvement may be delayed (it

occurs after the harassing episode has ended).

For example, the observer may not take action

during the incident but may later intervene by

advising the target or by reporting the incident

(i.e., whistle-blowing). In general, high-immedi-

2005 289Bowes-Sperry and O’Leary-Kelly

acy interventions focus on interruption of a spe-

cific ongoing harassment incident, whereas

low-immediacy interventions might be thought

of as efforts to prevent future harassment from

occurring. Of course, these categories are not

mutually exclusive; a particular observer may

intervene in an unfolding SH incident and then

follow up with postincident intervention.

The second dimension (level of involvement)

reflects the degree to which individuals im-

merse themselves in the SH incident. The lowest

level of involvement is noninvolvement. How-

ever, assuming that some intervention occurs, it

can vary based on how much observers publicly

embroil themselves. In high-involvement inter-

vention, observers insert themselves into the SH

episode or issue, increasing the potential for

resultant personal risks and benefits. Alterna-

tively, in low-involvement intervention strate-

gies (e.g., private support for the target), ob-

server assistance does not involve a strong

public connection to the incident or issue. As

evident in our examples, we define involvement

in a public, social sense. It is the willingness to

take action on the “social stage” of the organi-

zation (Gardner & Martinko, 1998: 69)—not ob-

servers’ readiness to be involved emotionally or

cognitively—that defines “involvement.”

Taken together, these dimensions indicate the

range of different behavioral options available

to harassment observers. First, observers may

choose “strong” intervention actions, such as

high-immediacy– high-involvement behaviors,

which publicly immerse them in the unfolding

SH episode. In such situations observers may

opt for actions such as challenging the harasser

(telling him/her to stop the harassment), publicly

naming the conduct as SH, or publicly encour-

aging the target to report the incident. The com-

mon denominator is that observers take an ac-

tive and identifiable role in the SH episode.

Similar high-involvement actions taken after

the incident (low-immediacy– high-involvement

behaviors) might include whistle-blowing (i.e.,

reporting the SH) or offering to accompany the

target as reporting occurs.

Alternatively, lower-profile interventions in

the unfolding situation (high-immediacy–low-

involvement behaviors) might include actions

like redirecting the harasser’s attention away

from the harassing conduct or removing the tar-

get from the situation. In both cases observers

FIGURE 1

A Typology of Observer Intervention Behaviors in Sexual Harassment

290 AprilAcademy of Management Review

will want to achieve these goals without nam-

ing the conduct as harassing (because they do

not want to be embroiled in the harassment mat-

ter). Essentially, they want to make the SH inci-

dent end, but not via any confrontational stand

that pulls them into the conflict. Similarly, after

the incident, observers may undertake low-

involvement efforts such as covert efforts to

keep the target and harasser apart (e.g., ensur-

ing they are not in the same workgroup or en-

suring they do not travel together), privately

sharing advice with the target, or providing so-

cial support behind the scenes. The key charac-

teristics of these low-immediacy–low-involve-

ment behaviors are that observers do not

become party to the harassment incident and

they occur after the SH incident.

This typology is useful for conceptualizing the

types of behaviors SH observers might enact,

thereby providing the reader a clearer concep-

tualization of observer intervention. The typol-

ogy itself, however, does not provide insight into

when (if at all) observers will intervene or what

forms of intervention will occur. We address

these issues in the remainder of the article.

OBSERVER INTERVENTION IN SEXUAL

HARASSMENT

Latane´ and Darley (1970) introduced what has

become a widely cited framework for conceptu-

alizing BI. This framework presents intervention

as the last step in a decision-making process

where observers intervene if they notice an un-

folding event, and then (1) interpret the situation

as one requiring action, (2) decide that it is their

personal responsibility to act, and (3) decide on

a specific form of assistance to provide. We use

this framework for our discussion of observer

intervention in SH. Specifically, and as shown in

Figure 2, we identify conditions likely to pro-

mote or inhibit intervention at each of these

decision stages.

Prior to explaining Figure 2, we should note

that this model explains intervention in SH spe-

cifically; it is not intended to explain interven-

FIGURE 2

Potential Routes to Observer Intervention in Sexual Harassment

2005 291Bowes-Sperry and O’Leary-Kelly

tion in other negative work conduct. We expect

the factors that influence intervention will de-

pend on the nature of the conduct itself. For

example, SH is an interpersonal form of harm,

indicating that dyadic relationship and social

identity issues may become central in ways they

do not for other negative actions (e.g., vandalism

of company property). In addition, there is sig-

nificant ambiguity around people’s understand-

ings and definitions of SH, suggesting that ob-

server sensemaking may be a more challenging

task than with other negative conduct (e.g., reck-

less use of company equipment). Other forms of

intervention, then, may require quite different

theoretical explanations.

General Nature of the Intervention Model

Figure 2 illustrates a decision tree model that

depicts the stages inherent in observers’ deci-

sion making regarding SH intervention. Specif-

ically, we pose several decision-making stages

and behavioral outcomes. Consistent with pre-

vious BI research (Latane´ & Darley, 1970), the

decision-making stages (illustrated by boxes in

Figure 2) include a determination of whether the

situation requires action, the degree of personal

responsibility for action, and the form of inter-

vention that will be undertaken. Two general

outcomes of the decision-making process (illus-

trated by circles in Figure 2) also are included in

the model: (1) observer intervention (which can

take the four different forms identified in Figure

1) and (2) observer nonintervention. Observers

are expected to choose between these behav-

ioral alternatives based on their determinations

(depicted as yes/no assessments) at each of the

decision-making stages.

An important aspect of the model is that un-

less observers develop affirmative answers to

the questions posed at each decision-making

stage, their intervention is unlikely. Also in-

cluded in the model are factors expected to be

especially relevant to observers’ assessments at

each decision stage. These factors (presented in

bullet-point format in Figure 2), therefore, are

fundamental to whether or not intervention oc-

curs. It is important to note that each of these

factors is expected to have an independent and

significant influence on the relevant decision-

making stage—that is, there is no expectation

that all of the factors identified at each decision

stage are necessary to an observer’s assessment

at that stage.

This model implies a rational, controlled de-

cision-making process. However, it seems inev-

itable that, at times, this process will be more

automatic than depicted in the model. For exam-

ple, when individuals witness a very egregious

form of harassing conduct or when they have

witnessed multiple similar incidents in the past,

their reactions to the event may proceed more

immediately and with less conscious cognitive

processing. Although it will be important for

researchers to classify conditions that prompt

controlled versus automatic decision making at

some point, the more fundamental research is-

sue now is the identification of factors that will

influence observers’ perceptions regarding in-

tervention. It is worth noting, however, that our

emphasis on the latter issue does not imply an

assumption that all intervention decisions will

involve this high level of cognitive processing.

Does This Situation Require Action?

Witnesses are unlikely to interpret a situation

as one requiring intervention unless they regard

the witnessed conduct as harmful behavior.

This, then, is the first assessment faced by SH

observers. Research on BI and ethical decision

making implies that SH often will not be per-

ceived as harmful because of (1) the ambiguous

nature of some SH conduct, (2) the low moral

intensity of this conduct, and (3) social influence

effects. These factors, therefore, are predicted to

lead subjects to a “No” response to the question

“Does this situation require action?”

Ambiguous nature of some sexual harass-

ment. Witnesses will be more likely to intervene

in an observed event they regard as harassment

than in an interpersonal exchange they perceive

as benign. If the exchange is perceived as ha-

rassment, it becomes an event for which behav-

ioral norms or codes of conduct exist. Using the

language of accountability theory (Cummings &

Anton, 1990; Frink & Klimoski, 1998; Schlenker,

Britt, Pennington, Murphy, & Doherty, 1994), the

incident becomes one that is governed by “be-

havioral prescriptions” to which individuals are

accountable. This initial framing of an event,

then, determines which prescriptions are acti-

vated, thereby affecting the type of behavioral

response likely.

292 AprilAcademy of Management Review

SH is a phenomenon, particularly in its less

severe forms, that can be ambiguous in that

individuals define it quite differently. For exam-

ple, there is evidence that men perceive fewer

actions as harassing than do women (e.g., Blu-

menthal, 1998), that the degree of work experi-

ence influences harassment perceptions (e.g.,

Baker, Terpstra, & Cutler, 1990), and that per-

sonal experiences with harassment can influ-

ence awareness (Bowes-Sperry & Powell, 1996).

Even the legal standards that define hostile en-

vironment harassment are open to subjective

interpretation (Gutek et al., 1999; O’Leary-Kelly

et al., 2000; Paetzold & O’Leary-Kelly, 1994), with

varied perspectives on such questions as what

constitutes “severe or pervasive” harassment,

when conduct is “unwelcome,” and what a “rea-

sonable woman” or “reasonable victim” would

do. This contrasts with some other forms of neg-

ative work behavior (e.g., unsafe use of company

equipment, on-premise drug abuse) in which

most observers would have similar interpreta-

tions of the situation as problematic. However,

given the differential interpretations around SH,

observers may not be confident that their own

definition of the situation is accurate, and this

lack of confidence associated with ambiguous

incidents has been found to work against gen-

eral bystander intervention (Clark & Word, 1972,

1974). Therefore, we offer the following proposi-

tion.

Proposition 1: Observer intervention

in SH is more likely when observers

perceive the witnessed incident as

low (versus high) in ambiguity.

An interesting question, then, is what charac-

teristics of interpersonal exchanges at work

lead them to be clearly interpreted as SH? A few

characteristics seem particularly important in

regard to SH (in the interest of brevity, we do not

share research propositions related to each).

Perhaps most obvious, situations in which the

target directly appeals to observers for help are

likely to be interpreted as necessitating inter-

vention. Because SH is an interpersonal form of

aggressive work behavior (as opposed to ac-

tions like vandalism or sabotage), there is a

person who can seek help from the observer.

When a target directly asks for help, much am-

biguity is removed from the incident, because

the target has labeled it as one that requires

action. The BI literature provides support for this

in that direct verbal appeals and eye contact

from targets appear to facilitate intervention

(Fehr, Dybsky, Wacker, Kerr, & Kerr, 1979; Shot-

land & Johnson, 1978).

In the absence of a direct appeal for help from

the target, the existence of power differentials

between harassers and targets seems important

in observer interpretations. Research on SH

shows that behavior initiated by a higher-level

employee toward a lower-level employee (as op-

posed to SH between peers) is more likely to be

interpreted as SH (e.g., Anderson & Hunsaker,

1985; Cochran, Frazier, & Olson, 1997; Kenig &

Ryan, 1986). This finding may be attributed to

the ability of high-level harassers to impact the

job-related conditions of targets. For example,

there is little doubt that conduct such as requir-

ing women to engage in sexual activity with

supervisors to obtain promotions, as occurred in

the District of Columbia prison system (Neal v.

Director, District of Columbia Dept. of Correc-

tions, 1995), constitutes SH.

Social identity perceptions also may influence

the interpretation of SH conduct. Although both

men and women likely perceive egregious be-

havior as SH, men appear less likely to regard

ambiguous social sexual behavior as harassing

(e.g., Blumenthal, 1998; Gutek, Morasch, & Co-

hen, 1983; Hartnett, Robinson, & Singh, 1989). So-

cial categorization theory, which explains why

individuals behave on behalf of groups, pro-

vides an explanation. Proponents of this per-

spective argue that individuals organize their

social worlds into ingroup and outgroup catego-

ries; that they regard themselves in terms char-

acteristic of the ingroup (prototypes); and that

these prototypes influence self-concept, cogni-

tions, and behavior (Hogg & Terry, 2000).

Since many SH situations involve a male ag-

gressor acting on a female target (U.S. Merit

Systems Protection Board, 1995),

sex-based

identities are likely to be salient (Ashforth &

Johnson, 2001). Social categorization principles

(Hogg & Terry, 2000), then, suggest that male

observers are motivated to interpret the situa-

tion in ways that facilitate positive attitudes

about their ingroup member (i.e., the male ha-

Although other types of harassment certainly occur (e.g.,

females as harassers, same-sex harassment), popular wis-

dom about SH, as well as social science knowledge (Leng-

nick-Hall, 1995) and legal theory (MacKinnon, 1979), has de-

veloped primarily based on a male-on-female model.

2005 293Bowes-Sperry and O’Leary-Kelly

rasser). In other words, they are motivated to

interpret ambiguous social sexual behavior as

something other than SH, which, in turn, influ-

ences their decision regarding intervention.

This argument is evidenced in recent reports

concerning the U.S. Air Force Academy, where it

appears that scores of female cadets who com-

plained to male academy officials of sexual as-

sault by male cadets received no assistance

from these officials (Denver Post, 2003). Although

this is not a case of direct observer intervention

(in that officials did not witness the incidents), it

suggests that even official, premediated re-

sponses to severe forms of harassment may be

influenced by efforts to protect members of sa-

lient identity groups.

Moral intensity of sexual harassment. Wit-

nesses also are unlikely to believe that an event

is harmful, and therefore necessitating interven-

tion, when they do not regard the event as in-

volving a moral issue. Moral intensity captures

the degree to which a decision maker perceives

a moral imperative in an event. This moral im-

perative stems from characteristics of the event

itself—that is, the immediacy and seriousness of

consequences and degree of social consensus

on the issue (Jones, 1991). Individuals who do not

regard an event as having a moral imperative

are less likely to engage ethics schemata and to

behave ethically around this event (Bowes-

Sperry & Powell, 1996, 1999; Jones, 1991; O’Leary-

Kelly & Bowes-Sperry, 2001). Therefore, individ-

uals who do not regard SH as a high moral

intensity issue are unlikely to intervene.

Although the moral imperative inherent in

many forms of negative work conduct (e.g., phys-

ical assault of a coworker, abusive supervision)

is not open to debate, this may not be true of all

SH. Previous research suggests that there are

numerous aspects of SH scenarios that discour-

age individuals from perceiving them as high in

moral intensity (see O’Leary-Kelly & Bowes-

Sperry, 2001, for a review). The tendency for tar-

gets to respond passively (Knapp et al., 1997)

may create the misperception that SH has few

serious or immediate consequences—a condi-

tion associated with low moral intensity (Jones,

1991). Further, differing perceptions about what

constitutes SH may result in divergent assess-

ments of its harm, again leading to low moral

intensity perceptions. In other words, many

forms of SH are not perceived as morally in-

tense. Indeed, it appears SH may be viewed as a

“normal” occurrence in organizations such as

Mitsubishi Motor Manufacturing of America,

where hundreds of female employees repeat-

edly were groped and exposed to sexually ex-

plicit pictures and gestures. In such environ-

ments SH may no longer be regarded as

“morally wrong” but, in fact, may become nor-

mative.

Proposition 2: Observer intervention

in SH is more likely when observers

perceive the witnessed incident as

high (versus low) in moral intensity.

Social influence effects. A final condition that

may lead witnesses to perceive a SH event as

harmless, and therefore not requiring interven-

tion, is based in social influence effects. Re-

search on BI suggests that the reactions of other

bystanders (especially to ambiguous events)

can determine an observer’s beliefs regarding

the necessity of intervention (Clark & Word,

1974; Latane´ & Darley, 1970). Specifically, ob-

servers seek cues regarding intervention from

other bystanders and are less likely to intervene

if these individuals do not appear anxious about

the situation (Clark & Word, 1972; Darley, Teger,

& Lewis, 1973; Harrison & Wells, 1991; Latane´&

Nida, 1981). This effect is particularly pro-

nounced when other observers are perceived as

similar to oneself (Smith, Smythe, & Lien, 1972).

Therefore, observers of a SH situation seem

likely to scrutinize the reactions of other observ-

ers in order to determine the appropriate per-

ceptual framing of the situation. If other wit-

nesses appear anxious or uncomfortable, the

observer is more likely to perceive the situation

as one requiring action. Alternatively, if other

witnesses do not appear anxious, this signals to

the observer that nothing is amiss, resulting in a

low probability of intercession. This effect, re-

ferred to as pluralistic ignorance in the BI liter-

ature (Latane´ & Darley, 1970), may be quite sim-

ilar to social influence effects found in

groupthink situations (Janis, 1982).

Proposition 3: The intervention deci-

sions of SH observers are influenced

by the behavior of other observers

such that intervention by any one in-

dividual is more likely when other ob-

servers display stronger (versus

weaker) concern regarding the inci-

dent.

294 AprilAcademy of Management Review

Is It My Personal Responsibility to Act?

If observers believe that a witnessed situation

requires action, their next decision (as shown in

Figure 2) involves beliefs regarding intervention

as a personal responsibility (Latane´ & Darley,

1970). It is quite possible for bystanders witness-

ing a harmful event to believe that someone

should take action but also to assume that some-

one means someone else. The determination of

personal responsibility to act is a complex issue

and one that requires consideration of numer-

ous factors in the social environment within

which the event is unfolding. We begin with a

discussion of a well-established effect in the BI

literature and then discuss how this effect is

complicated by social environment factors.

The most common finding in the BI literature

is that “the presence of other people serves to

inhibit the impulse to help” (Latane´ & Darley,

1970: 38). A review of dozens of studies across

thousands of subjects suggests the robustness

of this general finding (Latane´ & Nida, 1981).

There are several theoretical explanations for

this effect, perhaps the most common of which is

diffusion of responsibility—an explanation pos-

iting that as the number of bystanders in-

creases, individual observers feel less person-

ally responsible to act and less concerned that

they will be blamed for their inaction (Gottlieb &

Carver, 1980; Latane´ & Nida, 1981). The number

of other observers, then, distributes responsibil-

ity across a wider range of individuals, thereby

decreasing perceptions both of individual liabil-

ity and of the risks of inaction for a given indi-

vidual.

Based on this widespread finding, it would be

reasonable to predict that as the size of the

observer group increased, the likelihood of a

particular observer intervening in SH would di-

minish. However, recent criticisms of the BI lit-

erature suggest that belief in an automatic in-

verse relationship between observer group size

and helping is “overly reductionist” (Levine,

1999) in explanation and ignores the impact of

social factors on intervention decisions. This

criticism raises the possibility that “failure to

intervene can be seen not as the product of deg-

radation of feelings of individual responsibility

brought on by the presence of others, but as a

socially meaningful act in its own right” (Levine,

1999: 1151). In particular, Levine (1999) predicts

that the perceived relationship between actor

and victim, the social appropriateness of interven-

tion, and the social identity categorizations of the

observer influence intervention decisions (Le-

vine, 1999).

Relationship between actor and target. From a

social categorization perspective, individuals’

decisions regarding personal responsibility will

be influenced by the social categories they in-

voke when making sense of the witnessed event

(Levine, 1999). We find evidence for this in the

tragic episode of nonintervention illustrated by

the famous James Bulger case, in which two-

year-old James was abducted, beaten, and

killed by two ten-year-old boys who first walked

the injured and crying toddler through the

streets of Liverpool, England, for over two hours.

In this case some witnesses’ self-reports indi-

cate that these individuals chose inaction not

because of diffusion of responsibility (as the tra-

ditional BI literature suggests) but because they

categorized James’ attackers as family members

and were reluctant to intervene in domestic mat-

ters, which are accorded a strong right to pri-

vacy in Western societies (Levine, 1999). This

suggests that observers’ perceptions of personal

responsibility depend, in part, on the social cat-

egorizations used to define the relationship be-

tween the actor and victim.

Some early findings in the BI literature are

consistent with this explanation. This research

demonstrates that when bystanders perceive a

personal relationship between two parties in a

witnessed exchange, the bystanders are less

likely to become involved (Levine, 1999; Mori-

arty, 1975; Shotland & Straw, 1976). For example,

when witnessing a violent argument between a

man and a woman, bystanders are much less

likely to intervene if they believe the individuals

are married—and the default assumption ap-

pears to be that they are married (Shotland &

Straw, 1976). Thus, it seems that an inferred in-

timate relationship somehow legitimates vio-

lence, causing it to be perceived as less harmful

and as “none of my business” (Levine, 1999:

1136).

In SH situations we expect that witnesses of-

ten will assume a preexisting relationship be-

tween the harasser and target. The target and

harasser typically are coworkers (e.g., U.S. Merit

Systems Protection Board, 1995), which implies

at least some level of interaction and involve-

ment. At times, this relationship may be closer,

such as when individuals belong to the same

2005 295Bowes-Sperry and O’Leary-Kelly

workgroup or when the exchange involves a su-

pervisor-subordinate dyad. In addition, BI re-

search (Levine, 1999) suggests that witnesses

tend to interpret exchanges between male ag-

gressors and female targets, which is the most

common SH scenario (U.S. Merit Systems Protec-

tion Board, 1995), as private or personal matters.

Observers, then, may rationalize their inaction

in a SH event with several social category–

based justifications. For example, they may dis-

miss harassing actions as “horsing around be-

tween friends” or as a private matter between a

man and a woman who have some previous

history that is unknown to the observer.

Regardless of the perceived form of relation-

ship, observers who suspect a preexisting rela-

tionship between two parties in a SH event are

more likely to assume they do not have the con-

textual information to determine what the wit-

nessed conduct represents, and less likely to

intervene. In addition, the BI literature suggests

that observers often perceive male aggressor–

female victim aggression as less damaging in

cases of a personal relationship (Shotland &

Straw, 1976), again reducing the likelihood of

intervention. Finally, shrewd harassers may ac-

tually work to create the impression that there is

some legitimate personal relationship that jus-

tifies the behavior and that suggests the inap-

propriateness of observer intervention. These

messages, regardless of their truth, create am-

biguity for observers, who then are less likely to

intervene.

Proposition 4: Observer intervention

in SH is more likely when the observer

perceives no personal relationship be-

tween the harasser and target.

Social appropriateness of intervention. Tradi-

tional BI literature suggests that the presence of

multiple observers leads not only to diffusion of

responsibility but also to audience inhibition, in

which bystanders fear embarrassment if they

intervene inappropriately (Clark & Word, 1972,

1974; Latane´ & Nida, 1981). Determination of the

appropriateness of personal intervention is

partly a function of the perceived relationship

between the actor and target (as discussed pre-

viously) but also partly of other social factors,

such as the observer’s role requirements.

In many organizational settings, certain indi-

viduals are formally assigned—through role ex-

pectations—the task of preventing SH (e.g., su-

pervisors, human resource professionals).

Similarly, legal theory (e.g., Andrews v. City of

Philadelphia, 1990; Burlington Industries v. El-

lerth, 1998; Paetzold & O’Leary-Kelly, 1994) estab-

lishes role expectations for “agents of the orga-

nization”—that is, people with supervisory

authority or who exercise significant control

over hiring, firing, or conditions of employment

(Paroline v. Unisys Corporation, 1990; Paetzold &

O’Leary-Kelly, 1994). Given this, it seems likely

that these agents, on observing an incident of

harassment, will be more likely than other ob-

servers (those without role responsibilities) to

intervene. Because of both legal and organiza-

tional expectations, nonintervention has greater

costs for these individuals (e.g., potential legal

liability, potential disciplinary action by the em-

ployer).

Using the language of accountability theory,

the SH incident is governed by “behavioral pre-

scriptions” to which these individuals are ac-

countable (Cummings & Anton, 1990; Frink &

Klimoski, 1998; Schlenker et al., 1994). Research

on BI supports this expectation, in that when

subjects are asked to accept “focused responsi-

bility” (i.e., asked to provide assistance in the

event of a particular type of emergency situa-

tion), they are more likely to act (Shaffer, Rogel,

& Hendrick, 1975). Similarly, research on peer

reporting of unethical behavior and whistle-

blowing indicates that observers are more likely

to intervene if such behavior is formally estab-

lished as a responsibility of their work role (e.g.,

Miceli, Near, & Schwenk, 1991; Trevin˜o & Victor,

1992).

Proposition 5: Observer intervention

in SH is more likely when observers’

organizational role expectations in-

clude SH prevention than when they

do not.

The existence of these harassment-related

role expectations influences not only the behav-

ior of the role holder but also that of other ob-

servers. Some BI research suggests that respon-

sibility for helping may not be perceived as

uniform among all bystanders (Cramer, McMas-

ter, Bartell, & Dragna, 1988; Schwartz & Clausen,

1970). Specifically, observers appear to be less

likely to act when they believe there are others

in the bystander group who are more competent

to intervene (Cramer et al., 1988; Pantin &

Carver, 1982; Ross & Braband, 1973; Schwartz &

296 AprilAcademy of Management Review

Clausen, 1970). If the group of witnesses to a

harassing event includes an individual whose

organizational role expectations include SH pre-

vention (e.g., a supervisor, a human resources

professional), other observers are less likely to

intervene. The presence of an observer with

such role expectations both enhances the audi-

ence inhibition effect (in which the embarrass-

ment associated with inappropriate interven-

tion is heightened) and minimizes felt

responsibility (given the formal role expecta-

tions attached to the other observer).

Proposition 6: Observers for whom in-

tervention is extrarole will be more

likely to intervene when the observer

group does not include an individual

for whom such behavior is considered

inrole.

Social identity categorizations. Earlier, we

mentioned social identity effects on the observ-

er’s interpretation of the sexual harassment sit-

uation. Identity issues are also relevant in de-

termining one’s personal responsibility to act.

Social identity involves an awareness of the self

as belonging to a unique social unit that shares

a common identity—for example, “I am a female

professor” or “I am a black American” (Tajfel &

Turner, 1979). This awareness is associated with

a depersonalization of the self—a recognition of

the self as a representative of this social unit

(Turner, 1985). These social connections lead in-

dividuals to develop “an empathic altruism”

(Turner, 1985: 111), in which they feel obliged to

support the needs and goals of other ingroup

members. Given this, observers who share a

salient identity grouping with the SH target will

experience greater felt responsibility to provide

assistance than will individuals who do not re-

gard the target as a member of their identity

group. Of course, the specific nature of the iden-

tity-based connection could vary in that individ-

uals may share a grouping with the target

based on many factors, such as gender or race,

work position or role, or even friendship. It

seems likely, though, that gender may be an

especially salient identity grouping in SH situ-

ations.

Proposition 7: Observer intervention

in SH is more likely when an observer

and target are (versus are not) mem-

bers of the same salient identity

group.

What Type of Intervention Should I Undertake?

If observers have determined that the SH in-

cident requires intervention and that it is their

personal responsibility to take action, then they

are primed to act. The remaining question con-

cerns what form their action will take. As shown

in Figure 2, observers may choose to act imme-

diately and/or to act later, after the current SH

event has unfolded. Further, they must deter-

mine their level of involvement. We now exam-

ine conditions that influence the immediacy of

intervention and the level of involvement that

observers choose (i.e., the two dimensions used

in Figure 1 to identify types of intervention be-

haviors).

Simon argues that if rationality in decision

making is even to be approached, “a period of

hesitation must precede choice” such that deci-

sion makers can consider a broader range of

behavioral alternatives and their consequences

(1997: 101). Alternatively (and more likely, ac-

cording to Simon) is the case in which decisions

involve relatively limited information process-

ing (such as habitual responses). This suggests

that SH observers who intervene immediately in

an unfolding event will be more limited in their

information processing than will those who de-

lay intervention. Different factors, then, are

likely to influence immediate versus delayed SH

intervention.

Should I take action now? Immediate inter-

vention involves rapid decision making, which

seems most likely when observers (1) hold pre-

existing cognitive scripts for action and (2) are

driven by emotions rather than controlled cog-

nitive processing.

Intervention scripts. Scripts are cognitive rep-

resentations of sequences of events learned

from direct or vicarious experience (Abelson,

1976). If an observer personally has intervened

in the past, intervention in the present seems

more likely, given this precedent. More interest-

ing, however, is the case where individuals re-

call intervention modeled by others. Having wit-

nessed a role model, individuals may face fewer

significant cognitive hurdles in determining

whether and how to intervene, which will influ-

ence their ability to respond quickly. This role

modeling might occur in formal venues (e.g.,

2005 297Bowes-Sperry and O’Leary-Kelly

organization-sponsored training programs) or

more informally (e.g., an individual witnessing

a coworker intervention).

Modeling influences observers in multiple

ways (Bandura, 1986). First, it teaches new be-

havioral patterns, as well as “generative rules”

that determine when behavior should be initi-

ated (Bandura, 1986). Second, it minimizes an

individual’s inhibitions toward performing pre-

viously learned actions (Bandura, 1986). Model-

ing can occur through physical demonstration,

such as witnessing another observer interven-

ing in a SH incident, through pictorial represen-

tation, such as witnessing intervention in a

training video, or through verbal description,

such as hearing a colleague’s account of ob-

server intervention (Bandura, 1986).

This suggests that individuals who have had

role models for intervention are more likely to

intervene themselves, both because they have

learned how and when to take such actions and

because their inhibitions toward intervention

have been lowered by the role model’s previous

behavior. For example, inhibitions toward inter-

vention are likely to be lower for observers who

have been provided with formal role models

through training programs. We take this point

one step further, to predict that the influence of

role models on the immediacy of intervention

will be profound, because individuals who have

observed role models are better able to quickly

assess generative rules and access knowledge

of intervention behavior.

Proposition 8: Observers who have ex-

perienced role modeling around SH

intervention (versus those who have

not) are more likely to intervene in the

unfolding SH incident (i.e., display

high-immediacy intervention behav-

ior).

Emotional reactions. We also expect that indi-

viduals who have strong emotional reactions to

harassing incidents (compared to those who do

not) are more likely to take immediate interven-

tion action. Affective events theory (Weiss &

Cropanzano, 1996) depicts emotions as event-

driven phenomena—that is, their proximal

cause is some “happening” (Weiss & Cropan-

zano, 1996). Most theories of emotion (e.g., Frijda,

1986; Lazarus, 1991) associate an “action readi-

ness” with emotions. Because emotions are in-

tense affective experiences, individuals become

controlled by the emotional state and are pre-

disposed to act in ways that help them cope with

the emotions (Weiss & Cropanzano, 1996). This

coping can take varied forms, including “prob-

lem-focused coping,” which involves dealing

with the emotion-eliciting situation (Lazarus,

1991). Further, Frijda (1986) argues that action

tendencies (which can be driven by emotions)

and intentions are different reactions, with the

former involving efforts to change a given cur-

rent situation and the latter involving efforts to

achieve a future state.

This suggests, then, that individuals who

have strong emotional reactions to a SH incident

may be more prone to intervene in the unfolding

incident than those who do not. Because these

individuals are experiencing strong affective

states, their emotions have “control precedence”

(Frijda, 1986; Weiss & Cropanzano, 1996) and will

predispose them to action. It seems logical to

assume that because these individuals already

have characterized the situation as problematic

and requiring action (as depicted in Figure 2),

their emotional reactions to the episode will

tend to be negative (e.g., anger, exasperation,

anxiety) rather than positive (e.g., affection, con-

tentment, pride). Although it is beyond the scope

of this article to discuss all relevant factors, it

should be noted that numerous factors may lead

individuals to experience strong emotions on

witnessing a SH incident—for example, previ-

ous experience with SH, feminist orientation,

gender, relationship with the target, direct tar-

get appeals, shared social identity with the tar-

get, and perceptions regarding harm experi-

enced by the target.

Proposition 9: Observers who have

strong negative emotional reactions to

the SH incident (versus those who do

not) are more likely to intervene in the

unfolding SH incident (i.e., display

high-immediacy intervention behav-

ior).

Should I take action later? As shown in Figure

2, it is possible a SH observer who feels personal

responsibility to intervene will choose inaction

during the event (e.g., if the event unfolds

quickly or if he or she does not have time to

think through an appropriate response). This

person may choose to act after the incident, but

he or she now has the opportunity to consider

more carefully the advantages and disadvan-

298 AprilAcademy of Management Review

tages of intervention. As suggested by Simon,

this “period of hesitation” (1997: 101) allows for

more careful consideration of alternatives. Sev-

eral considerations seem especially critical to

subsequent decision making: (1) beliefs about

SH recurrence, (2) perceived harm of the SH, and

(3) perceived welcomeness of intervention.

Recurrence beliefs. If SH observers regard the

witnessed harassment as a “one-time only” ep-

isode, it is less likely they will intervene after

the event if they remained inactive during the

event (unless their motive is to punish the ha-

rasser). Low expectations of recurrence may ex-

ist, for example, when the precipitating incident

occurs outside of typical work conditions or re-

lationships (e.g., away from the work site, with a

visiting client). However, if observers expect

similar SH to recur (e.g., if it involves two co-

workers who see each other daily, if it is inher-

ent in the work environment), they will more

likely take subsequent action. Given that the

observer has identified the event as harmful

and as involving personal responsibility (as in

Figure 2), he or she is likely to feel accountable

for future incidents. Accountability theory sug-

gests that expectations about accountability are

a fundamental constraint on human behavior

and that individuals strive to act in ways they

can justify to themselves and others (Tetlock,

1992). It is likely, then, that substantial disso-

nance will be caused by continued inaction if

the observer continues to witness (or expects to

witness) SH.

Proposition 10: Observers who believe

that SH will recur (versus those who do

not) are more likely to intervene after

the SH incident (i.e., display low-

immediacy intervention behavior).

Perceived harm from SH. Both the BI (e.g., Do-

zier & Miceli, 1985) and the whistle-blowing (e.g.,

Near & Miceli, 1987) literature suggest that seri-

ous harmful acts are more likely to lead to ob-

server action. Observers who are deciding

whether to intervene after a witnessed event

will assess the harm caused by the SH they

observed, as well as the harm that might result

from future incidents. Earlier, we predicted that

when SH is viewed as an issue high in moral

intensity, intervention is more likely. This argu-

ment resurfaces here, in that when individuals

regard SH as causing significant harm, moral

standards become relevant and individuals

more motivated to take action (Bandura, 1999;

Jones, 1991). It should be noted that the assess-

ment of harm may involve a broad range of

objects. For example, observers may consider

harm to the target or to other observers from the

witnessed incident, potential harm to others

who might be future targets (including the self),

and harm to the organization (its reputation and

its functioning) that occurs as a result of past or

future SH.

Proposition 11: Observers who believe

that SH has caused or will cause sig-

nificant harm (versus those who do

not) are more likely to intervene after

the SH incident (i.e., display low-

immediacy intervention behavior).

Perceived welcomeness. We have argued that

observers not intervening in the unfolding SH

incident but considering delayed intervention

have time to ponder the wisdom of subsequent

intervention. Another factor they seem likely to

consider is the welcomeness of their interven-

tion actions. Observers contemplating delayed

intervention that involves the target (e.g., en-

couraging the target to report the harassment)

probably will assess whether the target wel-

comes such actions. If the observer believes that

a target will respond negatively (e.g., question-

ing why the observer did not provide assistance

during the incident, telling the observer to mind

his or her own business), delayed intervention

seems unlikely.

Similarly, observers contemplating a delayed

intervention that involves the organization (e.g.,

reporting the incident to management) will also

consider how their actions will be received. Or-

ganizations that take SH seriously—for exam-

ple, Deluxe Specialty Manufacturing Company,

which discusses SH during orientation and in

periodic meetings throughout the year, posts its

SH policy in full view of all employees, allows

complaints to be brought directly to the firm’s

owner, and encourages reporting of even minor

complaints (Cole, 1999)—have a low organiza-

tional tolerance for sexual harassment (OTSH;

Hulin, Fitzgerald, & Drasgow, 1996). In such or-

ganizations the perceived risk for employees

who report SH is low and the likelihood that

claims of SH will be taken seriously and that

harassers will be punished is high (Hulin et al.,

1996). Therefore, observers who are contemplat-

ing delayed intervention and who work in low

2005 299Bowes-Sperry and O’Leary-Kelly

OTSH organizations seem more likely to inter-

vene.

Proposition 12: Observers who believe

that their intervention is welcomed by

the target and/or the organization

(versus those who do not) are more

likely to intervene after the SH inci-

dent (i.e., display low-immediacy in-

tervention behavior).

How Involved Should I Get?

Research on whistle-blowing and BI indicates

that individuals engage in cost-benefit analyses

when making decisions about involvement in a

situation they witness (e.g., Clark & Word, 1974;

Dozier & Miceli, 1985; Miceli & Near, 1988; Pili-

avan, Piliavan, & Rodin, 1975). In other words,

observers choose the response that is most ef-

fective in dealing with the problem while mini-

mizing net costs (Piliavan et al., 1975; Piliavin,

Rodin, & Piliavin, 1969).

Not surprisingly, high-involvement strategies

have greater potential costs for observers than

low-involvement strategies. The most obvious

resistance to public involvement by the observer

will come from the harasser. When an observer

becomes involved in a situation in which the

harasser’s conduct is criticized, strained work

relationships and even retaliation from the ha-

rasser and other employees who support him or

her become possible (Loy & Stewart, 1984). Par-

ticularly in the case of powerful harassers, the

costs of high involvement by the observer may

be quite severe (e.g., in terms of pay raises,

assignments, termination). For example, even

though many managers at Astra AB’s U.S. phar-

maceutical subsidiary were appalled by the SH

they witnessed in their organization, they chose

not to intervene because the harasser was the

president of the company (Maremont, 1996).

The whistle-blowing literature supports the

notion that power issues are important, in that

employees who blow the whistle on wrongdoing

tend to have more power or credibility than

those who choose not to intervene (e.g., Miceli &

Near, 1988; Miceli et al., 1991). Costs could poten-

tially originate from the employer itself. In firms

with high OTSH (Hulin et al., 1996), intervention

may be regarded as “meddling” or “causing

trouble,” rather than as a positive act that serves

the organization.

However, there may be costs associated with

noninvolvement as well. These costs are most

likely to originate from the target and/or the self.

Specifically, observers who choose noninterven-

tion may find that targets resent their inaction.

This reaction is especially likely when the tar-

get and observer share a salient social identity.

For example, a female employee who reports SH

may be more angry over dismissal of these con-

cerns by a supervisor if the supervisor shares

the employee’s gender identity, racial identity,

or religious identity. This lack of “empathic al-

truism” (Turner, 1985) by ingroup members may

cause especially strong negative reactions.

Further, individuals who choose inaction may

experience self-initiated costs, such as guilt or

shame. Individuals maintain moral standards

that regulate their behavior, and when these

standards are violated, they experience strong

dissonance (Bandura, 1999). Given that observ-

ers who have progressed to this stage of the

decision-making process have recognized SH as

harmful and have accepted personal responsi-

bility to take corrective action (as shown in Fig-

ure 2), this dissonance is likely to exist for ob-

servers who choose nonintervention.

The reverse side of this argument deals with

the benefits associated with public involvement.

In general, the benefits of becoming involved

are likely to accrue from the target, from the self,

and perhaps from the organization. With the

first, when an observer becomes publicly in-

volved, the target now has an ally. Certainly,

observers with significant organizational power

are the most ideal allies, but even in cases

where the observer has power equal to or less

than the target, observer support is likely to be

appreciated and rewarded by the target. Simi-

larly, when an individual takes action to correct

a situation he or she has defined as harmful,

there are likely to be positive internal benefits,

in that internal moral standards have been re-

spected and cognitive dissonance is thereby

minimized (Bandura, 1999). Finally, in organiza-

tions with low OTSH (Hulin et al., 1996), we expect

that organization-initiated benefits will accrue

from public involvement. Such organizations are

likely to view SH intervention as a positive action

on behalf of the organization, rather than as an act

of dissent.

Proposition 13a: Observers are more

likely to choose high-involvement in-

300 AprilAcademy of Management Review

tervention strategies when the per-

ceived costs associated with involve-

ment are low (versus high).

Proposition 13b: Observers are more

likely to choose high-involvement in-

tervention strategies when the per-

ceived benefits associated with in-

volvement are high (versus low).

Before concluding our discussion of Figure 2,

it is important to emphasize an aspect of the

model that may get lost in the detailed discus-

sion we have presented to this point. The out-

come of this decision tree model is a prediction

about the specific type of observer intervention

(or nonintervention) that will occur. That is, the

pattern of observer assessments that result from

multiple decision stages is expected to deter-

mine the type of action taken by the observer.

For example, the model predicts that low-

involvement/high-immediacy intervention is

most likely for observers who believe that the

situation requires action, they have personal re-

sponsibility, action should be taken now, and

the costs of involvement are high. We reiterate

this point here to ensure that our detailed dis-

cussion of factors influencing judgments at each

decision stage does not overshadow this point.

DISCUSSION

Observer intervention to minimize or prevent

SH is a largely unexplored topic. In this article

we raise the issue as one worthy of research

attention, provide a typology of specific types of

intervention behaviors, and develop a process

model of how the decision to intervene might

occur. As depicted in Figure 1, intervention can

be characterized in terms of its immediacy and

level of involvement, resulting in four primary

types of intervention. Furthermore, Figure 2 il-

lustrates our predictions regarding the process

and factors leading to each type of intervention

described in Figure 1. In this final section we

discuss the research questions made salient by

our model and the challenges to conducting em-

pirical research on this topic.

Research Questions

The framework presented here raises numer-

ous research questions. Most obvious, the thir-

teen propositions indicate specific opportunities

for beneficial research. Indeed, there are numer-

ous ways to operationalize constructs inherent

in these propositions, suggesting a broad future

research agenda.

For example, the first proposition suggests

that observer intervention is unlikely when wit-

nessed conduct appears ambiguous. There are

likely to be multiple factors that can make an

individual’s conduct appear ambiguous to the

observer. For instance, ambiguity can be cre-

ated by a lack of social information regarding

whether the situation is an emergency (e.g., the

target does not ask for help, other observers do

not seem to be distressed), by nonconformity of

the situation to stereotypical SH situations (e.g.,

the harasser is not a supervisor but, rather, a

coworker), or even by more self-generated forms

of ambiguity (e.g., observers do not really try to

minimize ambiguity if the harasser is a member

of their own identity group, because the ambi-

guity allows them to excuse the negative behav-

ior of ingroup members).

Essentially, then, this one proposition allows

for numerous research questions, depending on

how “ambiguity of conduct” is operationalized

and explored. This is true of many propositions

in the model, leading all thirteen to generate a

significant research agenda.

Beyond the direct research implications of the

model in Figure 2, several additional research

issues seem particularly interesting and timely.

First, we are intrigued by the role that social

identity may play in reactions to SH. We argued

here that observers will interpret misconduct in

ways most beneficial to members of their in-

group and that they will feel the greatest per-

sonal responsibility to act when members of

their ingroup are threatened. The nagging ques-

tion, however, concerns which identity groups

are salient in sexual harassment situations. It

seems likely that groupings such as gender and

organizational role will be highly relevant, but

this opens the possibility for multiple identities

to be in play.

How, then, does an observer decide which

identity will drive reaction to SH? For example,

in the case of Harris v. Forklift Systems, Inc.

(1992), a high-ranking executive frequently

tossed objects on the floor so he could watch

women pick them up. On the one hand, another

male executive who witnesses such behavior

may feel a responsibility, based on his identity

2005 301Bowes-Sperry and O’Leary-Kelly

as a company official, to end it, yet, on the other

hand, he may be overly generous in his inter-

pretation of this situation if he wishes to protect

a member of his gender identity group (e.g.,

“He’s a good guy underneath it all,” “He just

doesn’t understand that times have changed”).

How the executive resolves this identity conflict

and what factors influence his decision making

are very interesting issues. These types of ques-

tions are particularly timely, given recent theo-

retical advances in social identity theories (Ash-

forth & Humphrey, 1995; Ashforth & Johnson,

2001; Hogg & Terry, 2000; Tyler, 1999).

Second, we suggest research on role modeling

effects in SH. Here, we suggested that role mod-

eling plays an important part in the observer’s

ability to quickly intervene, a prediction that

requires empirical verification. More broadly,

there is little research that explores role model-

ing effects in SH generally—that is, how it influ-

ences a target’s likelihood of reporting harass-

ment and how it influences a harasser’s

likelihood of harassing (see Pryor, LaVite, &

Stoller, 1993, for an exception). These are impor-

tant omissions in current research knowledge,

but they are quite easily explored, given the

well-established theory around role modeling

(Bandura, 1986).

Third, we suggest research on the emotions

that surface around SH. We suggested here that

emotional reactions may determine whether

and when observers will intervene. Again, a

broader issue is how emotions influence targets

of harassment generally. Although there is a

great deal of research addressing target inter-

pretations of SH and a growing body of research

addressing the types of coping behaviors that

targets undertake, there is little research on how

SH makes targets feel and how these emotions

influence subsequent coping. Again, this strikes

us as a particularly timely topic, given recent

interest in emotions in work organizations (e.g.,

Ashkanasy & Davis, 2002; Huy, 1999; Jordan, Ash-

kanasy, & Hartel, 2002; Pugh, 2001).

Fourth, we encourage research on observer

intervention in other forms of aggressive work

behavior—for example, abusive supervision

(Tepper, 2000), social undermining (Duffy, Gan-

ster, & Pagon, 2002), retaliatory behaviors (Skar-

licki & Folger, 1997), and theft (Greenberg, 1998).

Although all of these (and SH) are forms of ag-

gressive work behavior, the frameworks that ex-

plain observer intervention in each may differ.

As mentioned earlier, SH differs from other types

of work-related misconduct in ways that may

limit the generalizability of our framework.

First, SH is interpersonal and intended to harm

an individual, whereas other types of deviance

(such as sabotaging equipment or stealing from

the organization) involve an organizational tar-

get (Robinson & Bennett, 1995). This suggests

that factors in our model (e.g., the relationship

between the actor and target) may have little

relevance. Second, there is more definitional

ambiguity and controversy surrounding SH than

many other forms of workplace misconduct

(O’Leary-Kelly & Bowes-Sperry, 2001; O’Leary-

Kelly et al., 2000), such as theft or supervisor

bullying. Again, this suggests that issues raised

here—ambiguity of conduct, for example—may

not be generalizable. Finally, targets of SH often

adopt passive coping strategies, rather than the

more active coping (such as reporting the inci-

dent to management) likely when individuals

are targets of other forms of workplace miscon-

duct (e.g., nonsexual physical assault). Again,

factors central to our model (e.g., emotional re-

actions, perceived welcomeness) may therefore

be less relevant. Taken together, these differ-

ences suggest that although research on ob-

server intervention in other forms of workplace

aggression is important, different frameworks

may be needed to drive this research.

Finally, we suggest research exploring ob-

server intervention from a legal perspective (as

opposed to the social psychological perspective

adopted here). Inherent in our discussion is the

assumption that observer intervention is bene-

ficial to organizations; however, we recognize

that this assumption is open to debate. Observ-

ers may at times intervene in ways that create

difficulties within the organization (e.g., unjus-

tified accusations, severely damaged work rela-

tionships, creation of divisions between work-

groups or individuals) and even may create

legal liabilities for themselves or for their employ-

ers. Research that untangles functional observer

intervention (from an organizational perspective)

from dysfunctional observer intervention would

be beneficial. In addition, given the ambiguity

surrounding the definition of SH, observer in-

volvement creates the possibility for targets to

have one interpretation of the incident and ob-

servers to have another. From a legal stand-

point, this creates interesting complications

concerning whose testimony the courts will

302 AprilAcademy of Management Review

value and concerning whether organizations

are responsible should an observer (e.g., a su-

pervisor) label an incident as SH when a target

does not.

Research Challenges

As with most research related to human be-

havior, there are challenges inherent in re-

search on observer intervention in SH. Perhaps

most obvious is the tradeoff between assess-

ment of observer intervention in actual inci-

dents of SH (which generally necessitates field

research) and examination of the social and

psychological mechanisms through which ob-

server intervention occurs (which generally is

easier in laboratory settings). In regard to the

latter, many of the constructs mentioned in our

propositions involve individual perceptions—

for example, whether conduct is regarded as

ambiguous, whether other observers are per-

ceived to be alarmed by the behavior, and

whether emotional reactions occur, along with

recurrence beliefs and perceived costs of action,

among others. Given this, the most direct tests of

our propositions will involve assessment of ob-

server perceptions. This assessment is most eas-

ily obtained in controlled settings where observ-

ers’ beliefs can be probed immediately after an

incident has occurred. This type of controlled

setting, however, typically involves a “manufac-

tured” SH incident (e.g., observers witnessing a

scripted harassment episode, watching a video-

taped episode, or reading about a SH incident).

Eventually, researchers will want to test the

propositions around actual incidents of harass-

ment.

This challenge, of course, is nothing new in

social science research, and the solution in-

volves careful triangulation around research

questions using multiple data collection meth-

ods. For example, we might begin with research

examining observers’ recollections of SH inci-

dents in which intervention either occurred or

did not occur. Observers might report on those

factors (from Figure 2) that are most easily mea-

sured retrospectively. For example, witnesses

might recount the organizational roles (a proxy

for perceived power differentials) of harassers

and targets, the gender and organizational po-

sition matches of observers with targets and

harassers (proxies for social identity groupings),

or the organizational roles of other observers (a

proxy for social appropriateness). These vari-

ables, which are relatively objective and, thus,

likely to be valid (despite the fact that they are

collected retrospectively), allow for direct tests

of propositions in our framework (Figure 2). Be-

cause this research involves “real” incidents of

SH and intervention, if findings are supportive

of Figure 2, this suggests that research into un-

derlying cognitive mechanisms would be bene-

ficial.

Research into these underlying cognitive

mechanisms, then, might occur in more con-

trolled settings where researchers can manipu-

late constructs that either are difficult to evalu-

ate in the field or should be assessed at a more

perceptual level. As an example of the former, it

might be difficult to assess emotional reactions

retrospectively in the field. Observers may not

accurately recall their emotional reactions,

and/or their recall may be biased toward subse-

quent experiences (e.g., the observer whose in-

tervention was punished by management may

have an overall impression of being angry, even

if the anger did not occur around the incident

itself).

As an example of the latter (assessing con-

structs at a perceptual level), factors like moral

intensity or perceived welcomeness can be cre-

ated through experimenter manipulations (e.g.,

manipulate moral intensity by varying the seri-

ousness or immediacy of consequences; Jones,

1991), these manipulations can be verified by

assessing subject perceptions (e.g., asking sub-

jects how serious or immediate the conse-

quences of SH were for the target), and then the

effects of this manipulated variable can be as-

sessed. This approach allows for greater under-

standing about how observers are processing

information around their intervention decisions

(Bowes-Sperry & Powell, 1999). An understand-

ing of these factors important to intervention

decision making is difficult to achieve in field

settings, yet it is essential for establishing sup-

port for the theoretical predictions in Figure 2.

Conclusion

It is important to note that if observer nonin-

tervention in SH represents the status quo in an

organization, as we suspect it does in many

organizations, this can be a perilous condition.

When observers are excused from responsibility

for SH prevention, this enhances the ambiguity

2005 303Bowes-Sperry and O’Leary-Kelly

around defining SH and diminishes the moral

intensity of the issue. Indeed, over time, nonin-

tervention actually may create an environment

that encourages SH. Careful attention to SH ob-

servers and to the management of their inter-

vention, therefore, is critical.

REFERENCES

Abelson, R. 1976. Script processing in attitude formation and

decision-making. In J. Carroll & J. Payne (Eds.), Cogni-

tion and social behavior: 290 –303. Hillsdale, NJ: Law-

rence Erlbaum Associates.

Anderson, C. I., & Hunsaker, P. L. 1985. Why there’s romanc-

ing at the office and why it’s everyone’s problem. Per-

sonnel, 62(2): 57– 63.

Andrews v. City of Philadelphia, 895 F.2d 1469 (3d Cir. 1990).

Ashforth, B. E., & Humphrey, R. H. 1995. Labeling processes in

the organization: Constructing the individual. Research

in Organizational Behavior, 17: 413– 461.

Ashforth, B. E., & Johnson, S. A. 2001. Which hat to wear?: The

relative salience of multiple identities in organizational

contexts. In M. A. Hogg & D. J. Terry (Eds.), Social identity

processes in organizational contexts: 31– 48. Philadel-

phia: Psychology Press.

Ashkanasy, N. M., & Davis, C. S. 2002. Emotion in the work-

place: The new challenge for managers. Academy of

Management Executive, 16(1): 76 – 86.

Baker, D. D., Terpstra, D. E., & Cutler, B. D. 1990. Perceptions

of sexual harassment: A reexamination of gender differ-

ences. Journal of Psychology, 22: 17–37.

Bandura, A. 1986. Social foundations of thought and action: A

social cognitive theory. Englewood Cliffs, NJ: Prentice-

Hall.

Bandura, A. 1999. Moral disengagement in the perpetuation

of inhumanities. Personality and Social Psychology Re-

view, 3: 193–209.

Bingham, S. G. 1991. Communication strategies for manag-

ing sexual harassment in organizations: Understanding

message options and their effects. Journal of Applied

Communication Research, 19: 88 –115.

Blumenthal, J. A. 1998. The reasonable woman standard: A

meta-analytic review of gender differences in percep-

tions of sexual harassment. Law and Human Behavior,

22: 33–57.

Bowes-Sperry, L., & Powell, G. N. 1996. Sexual harassment as

a moral issue: An ethical decision-making perspective.

In M. S. Stockdale (Ed.), Sexual harassment in the work-

place, vol. 5: 105–124. Thousand Oaks, CA: Sage.

Bowes-Sperry, L., & Powell, G. N. 1999. Observers’ reactions

to social-sexual behavior at work: An ethical decision-

making perspective. Journal of Management, 25: 779 –

802.

Brief, A. P., & Motowidlo, S. J. 1986. Prosocial organizational

behaviors. Academy of Management Review, 11: 710 –

725.

Burlington Industries v. Ellerth, 524 U.S. 742 (1998).

Clark, R. D., III, & Word, L. E. 1972. Why don’t bystanders

help? Because of ambiguity? Journal of Personality and

Social Psychology, 24: 392– 400.

Clark, R. D., III, & Word, L. E. 1974. Where is the apathetic

bystander? Situational characteristics of the emer-

gency. Journal of Personality and Social Psychology, 29:

279 –287.

Clarkson, P. 1996. The bystander: An end to innocence in

human relationships? London: Whurr.

Cleveland, J. N. 1994. Women and sexual harassment: Work

and well-being in U.S. organizations. In M. J. Davidson &

R. J. Burke (Ed.), Women in management: Current re-

search issues. London: Chapman.

Cochran, C. C., Frazier, P. A., & Olson, A. M. 1997. Predictors

of responses to unwanted sexual harassment. Psychol-

ogy of Women Quarterly, 21: 207–226.

Cole, J. 1999. Sexual harassment: New rules, new behavior.

HR Focus, 75(3): 1, 14 –15.

Cramer, R. E., McMaster, M. R., Bartell, P. A., & Dragna, M.

1988. Subject competence and minimization of the by-

stander effect. Journal of Applied Social Psychology, 18:

1133–1148.

Cummings, L. L., & Anton, R. J. 1990. The logical and appre-

ciative dimensions of accountability. In S. Sivastra,

D. Cooperrider, & Associates (Eds.), Appreciative man-

agement and leadership: The power of positive thought

and action in organizations: 257–286. San Francisco: Jos-

sey-Bass.

Darley, J. M., Teger, A. I., & Lewis, L. D. 1973. Do groups

always inhibit individuals’ responses to potential emer-

gencies? Journal of Personality and Social Psychology,

26: 395–399.

Denver Post. 2003. Colorado Senator acts on rape, sexual

harassment charges at Air Force Academy. February 27.

Dozier, J. B., & Miceli, M. P. 1985. Potential predictors of

whistle-blowing: A prosocial behavior perspective.

Academy of Management Review, 10: 823– 836.

Duffy, M. K., Ganster, D. C., & Pagon, M. 2002. Social under-

mining at work. Academy of Management Journal, 45:

331–351.

Fehr, M. J., Dybsky, A., Wacker, D., Kerr, J., & Kerr, N. 1979.

Obtaining help from strangers: Effects of eye contact,

visible struggling, and direct requests. Rehabilitation

Psychology, 26: 1– 6.

Frijda, N. H. 1986. The emotions. Cambridge: Cambridge

University Press.

Frink, D. D., & Klimoski, R. J. 1998. Toward a theory of ac-

countability in organizations and human resources

management. In G. R. Ferris (Ed.), Research in personnel

and human resources management: 1–51. Greenwich,

CT: JAI Press.

Gardner, W. L., & Martinko, M. J. 1998. An organizational

perspective of the effects of dysfunctional impression

management. In R. W. Griffin, A. M. O’Leary-Kelly, &

J. M. Collins (Eds.), Dysfunctional behaviors in organiza-

304 AprilAcademy of Management Review

tions: Violent and deviant behavior, vol. 23: 69 –125.

Stamford, CT: JAI Press.

Gottlieb, J., & Carver, C. S. 1980. Anticipation of future inter-

action and the bystander effect. Journal of Experimental

Social Psychology, 16: 253–260.

Greenberg, J. 1998. The cognitive geometry of employee theft:

Negotiating “the line” between taking and stealing. In

R. W. Griffin, A. M. O’Leary-Kelly, & J. M. Collins (Eds.),

Dysfunctional behavior in organizations: Violent and de-

viant behavior, vol. 23: 147–194. Stamford, CT: JAI Press.

Gutek, B. A. 1985. Sex and the workplace. San Francisco:

Jossey-Bass.

Gutek, B. A., & Koss, M. P. 1993. Changed women and

changed organizations: Consequences of and coping

with sexual harassment. Journal of Vocational Behavior,

42: 28 – 48.

Gutek, B., Morasch, B., & Cohen, A. G. 1983. Interpreting

social-sexual behavior in a work setting. Journal of Vo-

cational Behavior, 22: 30 – 48.

Gutek, B. A., O’Connor, M., Melancon, R., Stockdale, M. S.,

Geer, T. M., & Done, R. S. 1999. The utility of the reason-

able women legal standard in hostile environment sex-

ual harassment cases: A multi-method, multi-study ex-

amination. Psychology, Public Policy and Law, 5: 596 –

629.

Harris v. Forklift Systems, Inc., No. 92–1168 (6th Cir., 1992).

Harrison, J. A., & Wells, R. B. 1991. Bystander effects on male

helping behavior: Social comparison and diffusion of

responsibility. Representative Research in Social Psy-

chology, 19: 53– 63.

Hartnett, J. J., Robinson, D., & Singh, B. 1989. Perceptions of

males and females toward sexual harassment and ac-

quiescence. Journal of Social Behavior and Personality,

4: 291–298.

Hogg, M. A., & Terry, D. J. 2000. Social identity and self-

categorization processes in organizational contexts.

Academy of Management Review, 25: 121–140.

Hulin, C. L., Fitzgerald, L. F, & Drasgow, F. 1996. Organiza-

tional influences on sexual harassment. In M. S. Stock-

dale (Ed.), Sexual harassment in the workplace: 127–150.

Thousand Oaks, CA: Sage.

Huy, Q. N. 1999. Emotional capability, emotional intelli-

gence, and radical change. Academy of Management

Review, 24: 325–345.

Janis, I. L. 1982. Groupthink. Boston: Houghton Mifflin.

Jones, T. M. 1991. Ethical decision making by individuals in

organizations: An issue-contingent model. Academy of

Management Review, 16: 266 –295.

Jordan, P. J., Ashkanasy, N. M., & Hartel, C. E. J. 2002. Emo-

tional intelligence as a moderator of emotional and

behavioral reactions to job insecurity. Academy of Man-

agement Review, 27: 361–372.

Kenig, S., & Ryan, J. 1986. Sex differences in level of tolerance

and attribution of blame for sexual harassment on a

university campus. Sex Roles, 15: 535–549.

Knapp, D. E., Faley, R. H., Ekeberg, S. E., & Dubois, C. L. Z.

1997. Determinants of target responses to sexual harass-

ment: A conceptual framework. Academy of Manage-

ment Review, 22: 687–729.

Kulik, C. T., Perry, E. L., & Schmidtke, J. M. 1997. Responses to

sexual harassment: The effect of perspective. Journal of

Managerial Issues, 9: 37–53.

Latane´, B., & Darley, J. M. 1970. The unresponsive bystander:

Why doesn’t he help? New York: Appleton-Century-

Crofts.

Latane´, B., & Nida, S. 1981. Ten years of research on group

size and helping. Psychological Bulletin, 89: 308 –324.

Lazarus, R. S. 1991. Emotion and adaptation. New York: Ox-

ford University Press.

Lengnick-Hall, M. L. 1995. Sexual harassment research: A

methodological critique. Personnel Psychology, 48: 841–

894.

Levine, M. 1999. Rethinking bystander nonintervention: So-

cial categorization and the evidence of witnesses at the

James Bulger murder trial. Human Relations, 52: 1133–

1155.

Loy, P. H., & Stewart, L. P. 1984. The extent and effects of

sexual harassment of working women. Sociological Fo-

cus, 17: 31– 43.

MacKinnon, C. A. 1979. Sexual harassment of working

women. New Haven, CT: Yale University Press.

Maremont, M. 1996. Abuse of power: The astonishing tale of

sexual harassment at Astra USA. Business Week, May

13: 86 –98.

Miceli, M. P., & Near, J. P. 1988. Individual and situational

correlates of whistle-blowing. Personnel Psychology, 41:

267–282.

Miceli, M. P., Near, J. P., & Schwenk, C. R. 1991. Who blows the

whistle and why? Industrial and Labor Relations Re-

view, 45: 113–130.

Moriarty, T. 1975. Crime, commitment, and responsive by-

stander: Two field experiments. Journal of Personality

and Social Psychology, 31: 370 –376.

Neal v. Director, District of Columbia Dept. of Corrections,

U.S. Dist. LEXIS 11461, 11469, 11515 (D.C. 1995).

Near, J. P., & Miceli, M. P. 1987. Whistle-blowers in organiza-

tions: Dissidents or reformers? Research in Organiza-

tional Behavior, 9: 321–368.

O’Leary-Kelly, A. M., & Bowes-Sperry, L. 2001. Sexual harass-

ment as unethical behavior: The role of moral intensity.

Human Resource Management Review, 11: 73–92.

O’Leary-Kelly, A. M., Paetzold, R. L., & Griffin, R. W. 2000.

Sexual harassment as aggressive behavior: An actor-

based perspective. Academy of Management Review,

25: 372–388.

Organ, D. W. 1988. Organizational citizenship behavior: The

good solider syndrome. Lexington, MA: Lexington Books.

Paetzold, R. L., & O’Leary-Kelly, A. M. 1994. Hostile environ-

ment sexual harassment in the United States: Post-

Meritor developments and implications. Gender, Work

and Organizations, 1: 50 –57.

Pantin, H. M., & Carver, C. S. 1982. Induced competence and

2005 305Bowes-Sperry and O’Leary-Kelly

the bystander effect. Journal of Applied Social Psychol-

ogy, 12: 100 –111.

Paroline v. Unisys Corporation, 900 F.2d 27 (4th Cir. 1990).

Piliavin, I. M., Piliavin, J. A., & Rodin, J. 1975. Costs, diffusion,

and the stigmatized victim. Journal of Personality and

Social Psychology, 32: 429 – 438.

Piliavin, I., Rodin, J., & Piliavin, J. A. 1969. Good Samaritan-

ism: An underground phenomenon? Journal of Personal-

ity and Social Psychology, 13: 289 –299.

Pryor, J. B., LaVite, C. M., & Stoller, L. M. 1993. A social

psychological analysis of sexual harassment: The per-

son/situation interaction. Journal of Vocational Behav-

ior, 42: 68 – 83.

Pugh, S. D. 2001. Service with a smile. Academy of Manage-

ment Journal, 44: 1018 –1027.

Ragins, B. R., & Scandura, T. A. 1995. Antecedents and work-

related correlates of reported sexual harassment: An

empirical investigation of competing hypotheses. Sex

Roles, 32: 429 – 455.

Robinson, S. L., & Bennett, R. J. 1995. A typology of deviant

workplace behaviors: A multidimensional scaling

study. Academy of Management Journal, 38: 557–572.

Ross, A. S., & Braband, J. 1973. Effect of increased responsi-

bility on bystander intervention, II: The cue value of a

blind person. Journal of Personality and Social Psychol-

ogy, 25: 254 –258.

Schlenker, B. R., Britt, T. W., Pennington, J., Murphy, R., &

Doherty, K. 1994. The triangle model of responsibility.

Psychological Review, 101: 632– 652.

Schwartz, S. H., & Clausen, G. T. 1970. Responsibility, norms,

and helping in an emergency. Journal of Personality and

Social Psychology, 16: 299 –310.

Shaffer, D. R., Rogel, M., & Hendrick, C. 1975. Intervention in

the library: The effect of increased responsibility on

bystanders’ willingness to prevent a theft. Journal of

Applied Social Psychology, 5: 303–319.

Shotland, R. L., & Johnson, M. P. 1978. Bystander behavior

and kinesics: The interaction between the helper and

the victim. Environmental Psychology and Nonverbal

Behavior, 2: 181–190.

Shotland, R. L., & Straw, M. K. 1976. Bystander response to an

assault: When a man attacks a woman. Journal of Per-

sonality and Social Psychology, 34: 990 –999.

Simon, H. A. 1997. Administrative behavior (4th ed.). New

York: Free Press.

Skarlicki, D., & Folger, R. 1997. Retaliation in the work-

place: The roles of procedural, distributive, and inter-

actional justice. Journal of Applied Psychology, 82:

434 – 444.

Smith, R. E., Smythe, L., & Lien, D. 1972. Inhibition of helping

behavior by a similar or dissimilar nonreactive fellow

bystander. Journal of Personality and Social Psychology,

23: 414 – 419.

Tajfel, H., & Turner, J. C. 1979. An integrative theory of inter-

group conflict. In W. G. Austin & S. Worchel (Eds.), The

social psychology of intergroup relations: 33– 47.

Monterey, CA: Brooks/Cole.

Tepper, B. 2000. Consequences of abusive supervision. Acad-

emy of Management Journal, 43: 178 –190.

Tetlock, P. E. 1992. The impact of accountability on judgment

and choice: Toward a social contingency model. Ad-

vances in Experimental Social Psychology, 25: 331–376.

Trevin˜ o, L. K., & Victor, B. 1992. Peer reporting of unethical

behavior: A social context perspective. Academy of

Management Journal, 35: 38 – 64.

Turner, J. C. 1985. Social categorization and the self-concept:

A social cognitive theory of group behavior. Advances in

Group Processes, 2: 77–121.

Tyler, T. R. 1999. Why people cooperate with organizations:

An identity-based perspective. Research in Organiza-

tional Behavior, 21: 201–246.

U.S. Merit Systems Protection Board. 1995. Sexual harass-

ment in the federal workplace: Trends, progress, continu-

ing challenges. Washington, DC: Office of Policy and

Evaluation.

Van Dyne, L., Cummings, L. L., & McLean Parks, J. 1995.

Extra-role behaviors: In pursuit of construct and defini-

tional clarity (a bridge over muddied waters). Research

in Organizational Behavior, 17: 215–285.

Weiss, H. M., & Cropanzano, R. 1996. Affective events theory:

A theoretical discussion of the structure, causes, and

consequences of affective experiences at work. Re-

search in Organizational Behavior, 18: 1–74.

Lynn Bowes-Sperry is an associate professor in the Department of Management at

Western New England College. She received her Ph.D. in management from The

University of Connecticut. Her research interests include the study of employee reac-

tions to wrongdoing in the workplace (sexual harassment, aggression) and organiza-

tional justice.

Anne M. O’Leary-Kelly is a professor in the Department of Management at The

University of Arkansas. She received her Ph.D. in organizational behavior from Mich-

igan State University. Her research interests include the study of aggressive work

behavior (violence, sexual harassment) and individual attachments to organizations

(psychological contracts, identification).

306 AprilAcademy of Management Review

A Spectrum of Bystander Actions: Latent Profile Analysis of Sexual Harassment Intervention Behavior at Work

Article

Full-text available

Mar 2025

Sexual harassment bystander intervention (SHBI) has been deemed critical to addressing persistent incidents in the workplace, yet scholarly knowledge of this behavior remains sporadic and limited. To move this field of research forward, the present study departs from the traditional variable-centered approach and instead adopts a latent profile approach to answer three key questions: (1) Which combinations (profiles) of actions do bystanders take to intervene? (2) When do bystanders intervene with specific profiles of SHBI? and (3) What happens when bystanders intervene with different behavioral combinations? We first developed and validated a scale to measure five distinct SHBI behaviors (i.e., confronting, distracting, supporting, reporting, and discussing) with two scenario-based pilot studies. Then, using this scale and latent profile analysis, we identified three distinctive profiles (i.e., active intervention, low-risk intervention, and no/limited intervention) in a field survey study (N1 = 381). In two additional field survey studies (N2 = 312; N3 = 326), by integrating social cognitive theory with the moral lens, we not only replicated the three similar profiles but also examined antecedents (i.e., organizational norms about sexual harassment and gender, and bystanders’ anger, empathy, and harassment-curbing expectancy) and outcomes of the profile memberships (i.e., aggression from the harasser, target gratitude, third-party elevation, and bystander guilt and pride). Overall, this research provides new insights into the nature of SHBI, its distinct patterns in the workplace, and potential organizational practices related to SHBI profiles.

Punish or help? Third parties’ constructive responses to witnessed workplace incivility: the role of political skill

Article

Full-text available

Apr 2025

Existing workplace incivility research from the perspective of third parties has been found to be limited in explaining the specific boundary conditions of their constructive responses, such as punishing the instigator and helping the target. The lack of relevant knowledge hinders a comprehensive understanding of the coping strategies favoured by third parties. Building on the deontic model and the value protection model, this research involved two sub-studies with a total of 710 Chinese employees, employing a scenario experiment (Study 1) and a time-lagged survey (Study 2). The results showed that witnessed incivility positively predicted third parties’ workplace ostracism against the instigator, with moral anger acting as a mediator. The research did not identify a direct link between witnessed incivility and third parties’ organisational citizenship behaviour towards the target. However, the mediating role of moral anger between these two variables was found in Study 1. Moreover, Study 1 indicated that political skill strengthened the relationship between witnessed incivility and moral anger, but weakened the relationship between moral anger and third parties’ workplace ostracism against the instigator or their helping behaviour towards the target—findings partially supported by Study 2. These insights provide a practical and theoretical understanding of how organisations can utilise the role of third parties to intervene in workplace incivility effectively.

Perceiving the Inevitable: Understanding Observer Reactions to Workplace Mistreatment Through the Lens of System Justification Theory

Article

Jan 2025
J ORGAN BEHAV

System justification theory posits that individuals tend to justify and maintain the status quo. For workplace mistreatment, we argue this tendency can elicit psychological processes in observers that may further disadvantage targets of mistreatment. We propose that organizational climates that are perceived to tolerate mistreatment increase the likelihood that observers perceive specific instances of mistreatment as inevitable. This can activate system justification tendencies in which observers evaluate the mistreatment incident as more legitimate and the target as less legitimate, prompting harmful observer reactions (e.g., minimizing the incident, negatively gossiping about the target). To investigate system justification in observer reactions, we validate a measure of perceived mistreatment inevitability and conduct a multiwave survey to test our hypotheses. Our findings indicate that organizational climates that tolerate mistreatment increase observers' perceptions that specific instances of mistreatment are inevitable, thereby activating processes that prompt observers to justify and maintain the status quo. Theoretical implications include identifying what activates system justification, why observers justify mistreatment, and how these tendencies elicit harmful reactions further disadvantaging targets. Practically, our findings highlight the importance of addressing organizational climates that tolerate mistreatment, avoiding reliance on observers to intervene constructively, and effectively addressing mistreatment to prevent further harm to targets.

Evaluating a dialogue-based bystander intervention for work-related gender-based violence and harassment: a mixed method pilot study

Article

May 2025
INT J CONFL MANAGE

Purpose This study aims to evaluate the experiences and perceived effects of the dialogue-based primary-level bystander intervention “Intervene: Sexual Harassment” in preventing work-related gender-based violence and harassment within a Danish public organisation. Design/methodology/approach A longitudinal mixed method approach was used across six care facilities. Group interviews were analysed using thematic analysis to assess participants’ experiences, while dependent sample t -tests measured changes in bystander behaviours, attitudes and perceived barriers. Two proportion z -score tests were used to evaluate changes in gender-based violence and harassment (GBVH) prevalence from baseline to follow-up. Findings Thematic analysis indicated that participants experienced increased knowledge and awareness of GBVH, increased personal and shared responsibility for addressing it and greater confidence in doing so. However, dependent sample t -tests showed no significant changes in quantitative outcomes. Contextual factors obstructing (e.g. lack of time) and facilitating (e.g. leadership endorsement) the implementation and effects of the intervention were also identified. Results suggest that “Intervene: Sexual Harassment” may positively affect factors associated with constructive bystander behaviour. Research limitations/implications The quantitative results are limited by a small sample size. Qualitative results indicate that participants perceived the interventions as impactfult. These contradictory findings warrants further investigation. Practical implications Leadership endorsement and addressing time constraints are essential for enhancing bystander interventions’ effectiveness. Tailored vignettes can further support contextual relevance and engagement. Originality/value This study is one of few empirical evaluations of bystander intervention programmes in organisational settings, emphasising the importance of contextual factors in implementation and the need for further research on long-term effectiveness.

Bystander behavior in workplace bullying: a vignette study exploring how organizational space and situational strength influence intentions to intervene

Article

Full-text available

Apr 2025

Purpose The study explores how organizational space and situational strength influence bystanders’ intentions to intervene in workplace bullying within the healthcare sector. Design/methodology/approach The research employs vignette-based interviews with 24 healthcare professionals from various medical fields in Sweden. The vignettes depict different bullying scenarios occurring in various spatial contexts, analyzed through the lens of situational strength theory. Findings The study identifies three main categories of bystander behavior: active constructive, passive constructive and passive destructive. Most participants described constructive behaviors, with no active destructive behaviors reported. The analysis reveals that situational strength does not consistently predict bystander behavior, suggesting that personal factors and the specific context of each vignette play significant roles. Research limitations/implications The small sample size ( N = 24) limits the generalizability of the findings. Social desirability bias may have influenced participants to report more constructive behaviors. Future research should explore conditions under which bystanders might engage in active destructive behaviors and the long-term effects of bystander interventions. Practical implications Organizations should develop clear policies and procedures on acceptable behavior and bullying, enforce these consistently and provide support systems to encourage bystander intervention. Understanding the influence of different workplace spaces on bullying and bystander behavior can help in designing effective anti-bullying strategies. Originality/value This study provides a nuanced understanding of how spatial aspects and situational strength influence bystander behavior in workplace bullying, particularly in the healthcare sector. It highlights the complexity of bystander intervention and the need for comprehensive organizational strategies to address workplace bullying.

“Did You See That Too?” Who Are the Observers of Workplace Mistreatment

Article

Apr 2025

Bystander intervention (BI) is an approach that has been used widely for addressing sexual and dating violence and has more recently been extended to other settings and types of harm, including workplace mistreatment. Understanding more about the opportunity to intervene—including the types of observed behaviors, who is observing them, and the frequency and breadth of those observations—is an important step to help inform the development of workplace BI programs. As such, the current study involved an anonymous online survey about observing workplace mistreatment, distributed by Qualtrics panel service to adult employees in workplaces throughout the United States ( n = 1,484). A total of 64.5% of participants indicated that they witnessed at least one type of mistreatment behavior against a work colleague within the past 12 months, and almost two-thirds of the behaviors were observed more than once. Most behaviors were observed in-person at work and were predominantly seen and seen/heard about. Employees who were nonbinary gender, had an advanced degree, were younger, and were shift workers were more likely to witness mistreatment. The findings indicate that witnessing mistreatment is common and that employees observe multiple and co-occurring behaviors, which suggests that BI strategies need to be multifaceted. The results also suggest that BI training should acknowledge people’s social identity, work status, type of position, and access to power, while also being accompanied by good organizational responses and structures.

A Mediated Model of Female Navy Personnel’s Personality Traits, Sexism, and Bystander Attitude Toward Sexual Harassment in Taiwan

Article

Mar 2025
ARMED FORCES SOC

This study investigated the mediating effects of personality traits, sexism, and bystander attitudes toward sexual harassment (BASH) among female navy personnel in Taiwan. A cluster sampling method was used to recruit 281 participants, with data collected on age, sexism perceptions, and BASH via a structured questionnaire. Descriptive analyses, analysis of variance (ANOVA), and structural equation modeling (SEM) revealed that personnel aged 21 to 30 and with 1 to 4 years of service reported significantly lower sexism scores. Older personnel, officers, and personnel with more years of service had significantly higher BASH scores than their younger, lower-level, and less experienced counterparts. Agreeableness and openness personality traits and sexism can explain female BASH. SEM results indicated that openness was the strongest explainer of BASH (β = .30), followed by age (β = .18), with sexism perceptions fully mediating the relationship between agreeableness and openness traits and BASH. Instructional implications and recommendations are discussed.

Sexual Misconduct in the Military: The Impact of Situational Factors on Bystander Intervention Strategies

Article

Feb 2025
ARMED FORCES SOC

Intervening is frequently encouraged to prevent or respond to sexual misconduct. However, due to the characteristics of military organizations (e.g., hierarchical structure), intervening may be challenging in military contexts. The aim of this study is to examine situational factors present in militaries (e.g., bystander’s rank relative to the perpetrator’s) that may impact the use of direct or indirect intervention strategies. A sample of Canadian Armed Forces members completed a scenario-based experiment. The results revealed that rank of the bystander, gender of the target, and severity of the situation impacted the use of direct intervention strategies, and the bystander’s rank relative to the perpetrator’s, gender of the target, and severity of the situation impacted the use of indirect intervention strategies. These findings highlight where direct and indirect interventions are unlikely to occur and situations that warrant greater focus in training programs and in communications from leadership.

Sexual Assault in a ‘Just World’: An Immersive VR Study of Bystanders

Preprint

Full-text available

Feb 2025

This paper explores how the need to believe in a just world (NBJW) affects bystanders’ intentions to intervene when witnessing a sexual assault. More specifically, we explored how non-rational strategies used to resolve a threat to the belief in a just world (BJW) are related to Latané and Darley’s (1970) bystander decision-making process, and whether empathic variables mediate this. In Study 1, the threat to the BJW was indirectly manipulated by varying the sexual assault severity described in a text vignette. Measures assessing victim derogation, blame, psychological distancing from the victim, empathy, empathic concern and the bystander process were completed by 294 participants. The assault severity had no significant effect on BJW strategies, but increased intervention intentions. There was a lack of evidence for the role of victim derogation and blaming, but greater psychological distancing indirectly predicted lower general intervention intentions via reduced empathic concern, and a reduced ability to succeed in several of the bystander stages, especially accepting responsibility to intervene. We replicated this design in Study 2 ( N = 117), but using filmed virtual reality (VR) stimuli. We again found a lack of effect of the severity manipulation beyond increasing the likelihood of intending to intervene. In Study 2, we also found that victim blaming predicted general intentions, but with no clear mediators. Implications are discussed, including how factors like the emotional impact of stimuli and the empathic nature of VR could help to explain some of the differing results found across our studies.

Sexual Harassment in an Organizational Context: Ethical Decision-Making Process

Chapter

Feb 2025

This chapter examines the problem of sexual harassment in organizations from the perspective of ethical decision-making, addressing the challenges and responsibilities of organizations and individuals when dealing with this sensitive issue, seeking to highlight the application of Thomas Jones' contingency model and its multidimensional component—Moral Intensity. Initially, the chapter characterizes sexual harassment as a complex ethical problem, whose consequences go beyond the individual impact, affecting organizational culture and the work environment. Next, key ethical decision-making process models that offer frameworks for evaluating and responding to ethical dilemmas in the business context are explored. These models guide managers and employees in analyzing the consequences, duties, and virtues involved in decisions about sexual harassment. Emphasis is given to the approach adopted by Jones, which provides a more comprehensive and adaptable understanding, emphasizing how different situations and contexts influence the ethical response. The dimensions of Moral Intensity are explored, which allow the severity of ethical dilemmas to be assessed based on factors such as magnitude of consequences, proximity, and probability of effect, facilitating understanding of the ethical and emotional impacts involved. The study contributes to reinforcing the importance of a robust and responsible ethical approach in organizations, emphasizing the need for preventive policies and support for victims to promote a safe and ethical work environment.

A Typology of Deviant Workplace Behaviors: A Multidimensional Scaling Study

Article

Full-text available

Apr 1995

Extra-role behaviours: In pursuit of construct and definitional clarity (a bridge over muddied waters)

Article

Full-text available

Jan 1995
RES ORGAN BEHAV

Which hat to wear? The relative salience of multiple identities in organizational contexts

Chapter

Full-text available

Jan 2001

LABELING PROCESSES IN THE ORGANIZATION: CONSTRUCTING THE INDMDUAL

Chapter

Full-text available

Jan 1995

Drawing on categorization theory and labeling theory, we argue that individuals working in or transacting with the organization tend to be perceived largely in terms of social categories (e.g., gender, job title)-though more individuated perceptions are possible if the perceiver has sufficient motivation and time. The primary functions of this categorization are to structure and simplify the social environment and to foster social control. Characteristics perceived to be prototypical of category members are often attributed to the individual such that he or she comes to be seen as exemplifying the category. The perceiver's frame of reference is a product of the organizational context, role and task demands, and certain individual characteristics. The labeling process tends to become more automatic with task experience as the perceiver develops a repertoire of expected categories. However, given the ambiguity of social stimuli and the multiplicity of potential labels, the act of labeling is often a negotiated process between the perceiver and the individual. The effects of labeling include altered interpersonal interactiolIS, formation of outgroups and ingroups, changes in social identity, and self-fulfilling prophecies that seemingly validate the labels. Tactics that may be utilized to cope with the more negative effects include embracing, distancing, passing, and repudiating, as well as more collective approaches. This perspective on labeling may help explain such diverse phenomena as stereotyping, tokenism, deifying, scapegoating, group polarization, and social identity formation.

Sexual Harassment as a Moral Issue: An Ethical Decision-Making Perspective

Chapter