Content uploaded by Eliza Byington
Author content
All content in this area was uploaded by Eliza Byington on Jan 08, 2018
Content may be subject to copyright.
HOW, WHEN, AND WHY BAD
APPLES SPOIL THE BARREL:
NEGATIVE GROUP MEMBERS
AND DYSFUNCTIONAL GROUPS
Will Felps, Terence R. Mitchell and Eliza Byington
ABSTRACT
This paper presents a review and integrative model of how, when, and why
the behaviors of one negative group member can have powerful, detri-
mental influence on teammates and groups. We define the negative group
member as someone who persistently exhibits one or more of the following
behaviors: withholding effort from the group, expressing negative affect,
or violating important interpersonal norms. We then detail how these
behaviors elicit psychological states in teammates (e.g. perceptions of
inequity, negative feelings, reduced trust), how those psychological states
lead to defensive behavioral reactions (e.g. outbursts, mood maintenance,
withdrawal), and finally, how these various manifestations of defensive-
ness influence important group processes and dynamics (e.g. cooperation,
creativity). Key mechanisms and moderators are discussed as well as
actions that might reduce the impact of the bad apple. Implications for
both practice and research are discussed.
Research in Organizational Behavior: An Annual Series of Analytical Essays and Critical Reviews
Research in Organizational Behavior, Volume 27, 175–222
Copyright r2006 by Elsevier Ltd.
All rights of reproduction in any form reserved
ISSN: 0191-3085/doi:10.1016/S0191-3085(06)27005-9
175
Organizations are increasingly relying on the work team model to capture
efficiencies and create value, with estimates predicting that as much as half
of the U.S. workforce will be working in teams by the year 2010 (Stewart,
Manz, & Sims, 1999). Indeed, most models of the ‘‘organization of the
future’’, such as networked, clustered or horizontal forms, are implicitly or
explicitly based on teams as the central organizing unit. As groups have
become more common, so has the importance of scholarly efforts to un-
derstand their potentialities and limitations (see for reviews Cohen & Bailey,
1997;Hackman, 1987;Ilgen, 1999;Ilgen, Hollenbeck, Johnson, & Jundt,
2005;Kozlowski & Bell, 2003). However, all teams are not equal, and as the
literature continues to evolve, we are beginning to understand how and why
these differences emerge.
In this vein, researchers have noted that, while some teams achieve co-
hesion between members, a mutually supportive ethos, and high collective
efficacy, other groups exhibit divisiveness, conflict, as well as the tendency to
‘‘burn themselves up’’ (Kozlowski & Bell, 2003). As noted by Hackman
(2002) ‘‘Some project groups do turn out to be more frustrating that ful-
filling, more a source of angst than of learning y. Teams can stress their
members, alienate them from one another, and undermine their confidence
in their own abilities’’ (p. 29). Many groups fail, but our understanding of
how and why this occurs is limited.
To date, the academic literature tends to highlight group-level phenomena
(Kozlowski & Bell, 2003) such as group paranoia (Kramer, 2001), group
think (Janis, 1982;Moorhead, Neck, & West, 1998) and low group efficacy
(Gully, Incalcaterra, Joshi, & Beaubien, 2002) as the culpable forces behind
ineffective teams. While these group-level variables are surely important,
this paper argues that, in some cases, a single, toxic team member may be
the catalyst for group-level dysfunction. This is a perspective echoed in
Keyton’s (1999) review of dysfunctional teams, which states that in most
models of group process or performance ‘‘group members are [treated as]
equal or interchangeable’’ and that there is a paucity of ‘‘attention to diffi-
cult group members’’ (p. 492). He goes on to claim that ‘‘[s]ometimes the
source of the dysfunction is one individual’’ (p. 493).
Upon first blush, Keyton’s statement seems obvious. Indeed, the common
idiom ‘‘a bad apple spoils the barrel’’ captures the core idea of negative
individuals having an asymmetric and deleterious effect on others. In a
Harvard Business Review article, Wetlaufer (1994) talks about ‘‘team de-
stroyers’’, taking for granted that persistent negative behavior can have
huge repercussions on group functioning. In an HR Magazine cover story
on ‘‘hard-core offenders’’, Andrews (2004) describes how ‘‘egregious
WILL FELPS ET AL.176
employee behavior can ycripple employee morale’’ (p. 43). Similarly, in an
article on training, Tyler (2004) urges, ‘‘[b]efore the whole bunch spoils,
train managers to deal with poor performers’’ and says these ‘‘bad apples’’
are ‘‘like a cancer that spreads throughout the entire workplace’’ (p. 77). But
despite this provocative rhetoric, the truth is that we currently know very
little about how, when, or why a negative member might have an asym-
metric effect on teammates, group processes, or group outcomes.
Moreover, academic theory is almost totally silent about these issues.
Indeed, given current accounts, it is unclear exactly how a negative indi-
vidual would persist in a group, or have powerful effects if they did.
For example, in his influential work on how groups influence individuals,
Hackman (1976) suggested that members co-regulate each other’s behavior
through ambient and discretionary stimuli to effectively produce uniformity
among members (p. 1473). Recently, Lepine and Van Dyne (2001) suggested
four potential peer responses to low performers: training, compensation,
motivation, or rejection. In both of these seminal and recent models, the
roseate conclusion seems to be that difficult teammates will be rehabilitated,
ousted, or teammates will compensate for them.
In contrast, we are interested in the instances when constructive responses
are not available or utilized and when negative behavior persists day after
day with little recourse. These scenarios may result when the harmful person
has seniority, political connections, task expertise, or when teammates
choose ineffective response strategies. We believe these scenarios describe
the circumstances under which the ‘‘bad apple spoils the barrel’’, through a
profound and harmful effect on the group. In other words, the focus of this
paper are those situations where the group functions poorly, and may al-
ternately fail or disband as a result of one member’s actions. By integrating
and extending prior work, we detail which negative behaviors are a threat to
effective group functioning, the conditions under which groups are able to
deal with negative behavior; how negative members influence the thoughts,
feelings, and behaviors of teammates; and the mechanisms by which these
‘‘bad apples’’ can provoke dysfunctional group dynamics. We conclude with
a discussion of what can be done to alleviate these negative effects and,
perhaps, ‘‘save the barrel’’.
EXTANT EVIDENCE OF BAD APPLE EFFECTS
The central goal of this paper is to explain how, when, and why negative
group members might have a powerful, asymmetric effect on the group. But
How, When, and Why Bad Apples Spoil the Barrel 177
first, it is important to firmly establish that this effect occurs at all. To date,
the primary evidence relevant to the ‘‘bad apple’’ phenomenon has been the
linkage between member personality and group outcomes. And indeed, the
evidence here is remarkably robust even if the causal explanations are sparse
or non-existent. This personality-based research has found that how low the
lowest teammate is on the variables of conscientiousness, agreeableness, and
emotional stability is usually a strong predictor of group-level variables. The
ostensible implication is that the ‘‘worst’’ group member can have important
effects. We briefly review the relevant studies below.
Across several companies, Barrick, Stewart, Neubert, and Mount (1998)
researched how members’ personalities affected group outcomes in 51 man-
ufacturing-related work teams. They were surprised to find that the lowest
team member’s score for conscientiousness, agreeableness, and emotional
stability was a good predictor of social cohesion (r¼0.14, 0.38, 0.34 re-
spectively), communication (r¼0.29, 0.50, 0.50), team conflict (r¼0.39,
0.51, 0.40), and perceptions of equitable workload sharing (r¼0.30,
0.62, 0.33). Moreover, across these group process variables and across
the three personality dimensions, these worst member correlations were
substantially stronger predictors than the team’s mean personality scores
or the highest (e.g. ‘‘best’’) person’s score. For the outcome variable of
task performance, the scores for the least conscientious and agreeable
member predicted team performance fairly well (r¼0.34 and 0.32 respec-
tively).
The findings of Barrick et al. (1998) are not isolates. Indeed, an increas-
ingly common practice is to actually operationalize ‘‘group personality’’
as the lowest member’s score. Theoretically, this is predicated on Steiner’s
(1972) argument that the weakest link is particularly important in conjunc-
tive tasks. In the laboratory study of Lepine, Hollenbeck, Ilgen and
Hedlund (1997), using the Team Interactive Decision Exercise (TIDE),
they test the role of the personality variable of conscientiousness on group
performance, and find that the lowest member’s score is an important pre-
dictor (r¼0.18), but that the mean score is not. They use this as evidence
that the task is a conjunctive one. Similarly, Neuman and Wright (1999)
conducted a study of teams of human resource professionals, and found
that the lowest member’s score for conscientiousness and agreeableness
predict group performance (r¼0.36 and 0.27 respectively), and to do so
over and above cognitive ability. Chatman and Barsade operationalized
collectivism as agreeableness and found that less agreeable members de-
pressed the cooperativeness of more agreeable members, but that the reverse
did not hold true. Again, this indicates an asymmetric effect of negative
WILL FELPS ET AL.178
teammates, as defined by their personality. Finally, in one of the few
studies linking emotional stability to group performance, Camacho and
Paulus (1995) compared the creativity of groups with different combina-
tions of member social anxiety. Teams composed of all socially anxious
(e.g. emotionally unstable) members came up with relatively few ideas
(M¼45.8); while teams composed of all socially calm members were
much more creative (M¼85.5); but most interesting and relevant to
our purposes, teams composes of two anxious and two stable members
performed about as badly (M¼53.2) as the group with all socially
anxious members – again indicating an asymmetric effect of negative in-
dividuals.
However, while these results are interesting, and provide broad support
for the ‘‘bad apple’’ phenomenon, they are not adequate. First, they are
theoretically inadequate in that most were post hoc findings that were not
central to the original questions under investigation. Second and more im-
portantly, the personality approach to understanding the bad apple phe-
nomenon is inherently problematic. There are many situational variables
which inhibit or enable the behavioral expression of personality in the
workplace (Tett & Burnett, 2003). For example, in many cases, a person
with low conscientiousness can force themselves to act thoughtfully and
carefully, at least for a while (Tett & Burnett, 2003). But it is the behavioral
expressions of negativity, not personalities, that upsets others and blocks
key group processes. A direct focus on the asymmetric influence hypothesis
requires moving away from distal personality measures to more proximal
causal variables of actual negative behaviors and dysfunctional group pro-
cesses. A recent review of the relationship between personality and group
outcomes says it better than we can:
‘‘Future research yshould focus on refining our understanding of how personality traits
are related to the task and interpersonal behaviors in group processes y. The inatten-
tion to mediating mechanisms is exacerbated in the literature by the tendency to focus on
desirable behaviors (e.g. helping, cooperation). For the most part, undesirable behaviors
such as malingering, social loafing, dishonesty, and sabotage have been ignored y.We
suspect, in short, that many of the process theories need to explicate the negative in-
dividual behaviors that cause poor group performance’’ (Moynihan & Peterson, 2001,
p. 340).
After briefly discussing the boundary conditions of this paper, we return to
this challenge of Moynihan and Peterson’s, and attempt to specify precisely
which negative behaviors cause which dysfunctional reactions, group proc-
esses, and group outcomes.
How, When, and Why Bad Apples Spoil the Barrel 179
BOUNDARY CONDITIONS
McGrath (1984) defines a group as ‘‘an entity that interacts, is interde-
pendent, mutually aware, with a past and an anticipated future’’ (p. 6).
We are employing this definition and narrowing the scope of our analysis
to small groups for several related reasons. First, we believe that destruc-
tive behavior will be particularly impactful in small groups, which are
often characterized by a high degree of interaction and interdependence
(Wageman, 2000), two factors that are predicted to make dysfunctional
behavior both more salient and disruptive. Second, and as a consequence of
their interdependence, small groups tend to be less tolerant of negative
behaviors than independent individuals (Liden et al., 1999). Members of
small groups have a greater motivation to identify and address behavior,
which threatens the group. The third reason for focusing on a small group
context is that these groups have properties that facilitate responses to neg-
ative group member behavior. Small groups build a consensual social reality
that is negotiated through reoccurring interaction and discussion (Hardin &
Higgins, 1996), which in turn facilitates other members responding as
a coordinated coalition (Lyons, Mickelson, Sullivan, & Coyne, 1998). In
sum, we delimit our focus to the small group simply because it is ‘‘where
the action is’’ – where a negative group member will have an increased
impact, but also where the group will have stricter standards, social
norms about appropriate behavior, and the potential to build coalitions.
While chronically dysfunctional people may have impacts in many settings,
small groups are a particularly appropriate venue for investigating their
effects.
We also limit our focus to a subset of the behaviors, which might be
considered ‘‘negative’’. A dysfunctional member’s behavior inhibits essential
group functions, processes, and goals. As such, we chose a pan-group defi-
nition of a bad apple member as individuals who chronically display behavior
which asymmetrically impairs group functioning. Three parts of this definition
bear noting. First, for the purpose of this analysis, who counts as a bad
apple is defined by their pattern of behaviors in a particular group setting.
These negative behaviors might variously be a function of dysfunctional
roles, dispositions, negative life events, substance abuse, some combination
of these, or something else entirely. By defining negative team members in
terms of clearly observable behavior – rather than these varied and more
distal contributors – much more specific predictions can be made. Second,
for the purposes of this paper, a group member is considered negative
only to the extent that their behavior violates norms that are empirically
WILL FELPS ET AL.180
supported as necessary for effective group functioning. Specifically, we are
investigating group members who violate norms of equity, positive affect,
and appropriate social functioning. We will elaborate on the support and
relevancy of these categories in our discussion on types of negative group
members. Finally, we would assert that this definition is not tautological
despite the fact that bad apple behaviors are defined as a function of their
effects on group performance. Tautologies are redundant statements that do
not add understanding and which are true by virtue of their logical form
alone. In contrast, our definition of what would constitute bad apples is
open to revision and disconfirmation and, as we will see, includes fairly
elaborate predictions of unfolding effects and underlying processes. More-
over, we would argue that our definition is completely consistent with other
prevalent theories. For example, work on organizational citizenship be-
havior is defined as a function of the contextual behaviors that contribute to
organizational functioning, and even more broadly, personality (defined as
tendencies to express behavior) is often empirically linked to expressions of
behavior.
BAD APPLE TEAM MEMBERS
Types of Bad Apple Team Members
In researching dysfunctional group dynamics, we identified three categories
of difficult team member behavior, which are especially likely to ‘‘spoil the
barrel’’ if left unchecked: withholding of effort, being affectively negative,
and violating important interpersonal norms. These categories emerged
from an analysis of the major categories of behavior that are needed for a
group to be successful. First, and most simply, members must contribute
adequate effort by working towards group goals with intensity and persist-
ence (Mitchell, 1997). Second, group members must perform ‘‘emotional
labor’’ by regulating their expressions of feelings to facilitate comfortable
and positive interpersonal interactions within the group (Hochschild, 1983;
Morris & Feldman, 1996). Finally, members must perform ‘‘contextually’’,
by not violating or detracting from the organizational, social, and psycho-
logical environment, which they inhabit (Motowidlo, Borman, & Schmit,
1997). Contextual performance is accomplished through expressions of
interpersonal respect and adherence to interpersonal norms (Tyler &
Blader, 2001). Our paper reviews evidence, which suggests that under cer-
tain circumstances, group members who persistently and consistently
How, When, and Why Bad Apples Spoil the Barrel 181
under-perform these three types of behavior can have a severe impact on
group functioning.
Withholders of effort intentionally dodge their responsibilities to the
group and free ride off the efforts of others. Behavioral examples of with-
holding effort consist largely of not doing something – of not completing
tasks or contributing adequate time, not taking on risks or responsibilities,
or not disclosing aptitudes in the hope that others will compensate. While
these behaviors have alternately been labeled shirking (by economists), free
riding (by sociologists), and social loafing (by psychologists), Kidwell and
Bennett (1993) convincingly argue that these terms just describe different
reasons and contexts in which people withhold effort from the collective. We
agree and refer to all three literatures when discussing withholders of effort.
Second, a person may continually express a negative mood or attitude.
We call this kind of member affectively negative, employing the broad usage
of affect to encompass the triumvirate of emotion, mood, and attitude (c.f.
Brief, 1998). To assess this construct, Furr and Funder (1998) combined
measures of depression, happiness, satisfaction, and self-esteem. Then, from
an analysis of a series of dyadic interactions, Furr and Funder constructed
behavioral profiles of this sort of individual, who they call personally neg-
ative. They found that ‘‘personally negative’’ individuals were more likely to
exhibit an awkward interpersonal style and to more frequently express pes-
simism, anxiety, insecurity, and irritation. Diverging from Furr and Funder,
we are interested in those individuals who are especially high in these di-
mensions. Moreover, as noted previously, the focus is behaviors rather
than the personality variables that underlie those behaviors, since it is be-
havioral expressions rather than internal states that will impact other group
members.
Finally, those that detract from the group’s contextual environment by
violating interpersonal norms of respect are called interpersonal deviants
(Robinson & Bennett, 1995;Bennett & Robinson, 2000). Bennett and
Robinson have conducted a series of studies to try to understand which
workplace behaviors are consistently considered deviant. They have found
seven common behaviors which are reliably assessed as deviant: making fun
of someone, saying something hurtful, making an inappropriate ethnic or
religious remark, cursing at someone, playing mean pranks, acting rudely,
and publicly embarrassing someone. For our purposes, these seven be-
haviors define the category of interpersonal deviance.
Note that these three categories are not all encompassing – not everyone
considered an ‘‘undesirable’’ group member is eligible for ‘‘negative mem-
ber’’ status. For example, many characteristics like shyness, lacking a sense
WILL FELPS ET AL.182
of humor, or being unpredictable do not enter into our definition because
they are unlikely to seriously disrupt important group processes. Instead,
the focus is on negative interpersonal behaviors, whose persistence would
have important harmful effects on the dynamics, processes, and team out-
comes. Other harmful behaviors like theft, cheating, sabotage, or vandalism
are excluded since they affect the organization rather than teammates (c.f.
Robinson & Bennett, 1995). Similarly, we do not include group members
with distinctive demographic backgrounds or those who have divergent
opinions about the best way to accomplish group goals (O’Leary-Kelly,
2005). Although some group members may consider these characteristics
difficult to deal with, both demographic diversity and divergent opinions
may improve group functioning, and are consequently of a qualitatively
different variety than our three destructive behaviors (e.g. Nemeth & Kwan,
1987). Further, we omit individuals who are motivated to achieve group
goals but do not have the requisite ability. While poor performance can
certainly diminish group performance, this low performance does not de-
pend on negative interpersonal reactions for its effect, and indeed tends to
evoke sympathy and compensation from teammates (Jackson & Lepine,
2003;Taggar & Neubert, 2004). Moreover, to the extent that these indi-
viduals have negative effects, they are likely to be additive rather than
asymmetric. Finally, given the focus on ‘‘spoiled barrels’’, there is little
reference to whistleblowers, positive deviants, change leaders, or exceptional
individuals who carry the group (c.f. Warren, 2003).
At this point, we can display Fig. 1, which depicts the organization of this
paper. We have described above the three categories of behavior that define
what we call a bad apple group member. Initially, when these behaviors
surface or are noticed they might be described as episodic (box 1). Our next
section described how the group will try to change the behavior or perhaps
oust the negative member. If that does no’t work, we are left with a more
persistent and chronic problem (box 2). It is at this point where negative
psychological reactions become more apparent (box 3) and we will discuss
the factors that may make this situation better or worse (the moderators in
box 4). The negative psychological states will lead to defensive behaviors by
group members (box 5) and through the mechanisms of aggregation, spill-
over, and sensemaking, these behaviors will come to influence group proc-
esses (box 6) and group outcomes (box 7).
Note again that the underlying message and contribution of this paper is
not that one bad group member can cause groups to fail or disband. We
already know that a bad apple can sometimes spoil the barrel (see Barrick
et al, 1998;Chen & Bachrach, 2003;Camacho & Paulus, 1995;Dunlop &
How, When, and Why Bad Apples Spoil the Barrel 183
aggregation, spillover, sensemaking
1
Episodic
Negative
Behavior
Relative Power
3
ΨReactions
- Inequity
- Negative Emotions
- Damaged Trust
6
Group Processes
- Motivation
- Cooperation
- Conflict
- Creativity
- Learning
7
Group Outcomes
- Low performance
- Low wellbeing
- Low viability
4
Moderators
- Intensity of Negative Behaviors
- Team Interdependence
- Valence of recent outcomes
- Personal Coping Abilities
5
Defensive Behaviors
-Explosions
- Covert revenge
- Mood maintenance
- Denial
- Slacking off
- Withdrawal
2
A Negative Member
(i.e. Chronic Behavior)
- Withholder of Effort
- Affectively Negative
- Interpersonal Deviant
Fig. 1. The Bad Apple Phenomenon Aggregation, Spillover, Sensemaking.
WILL FELPS ET AL.184
Lee, 2004;Haythorn, 1953;Neuman & Wright, 1999). Instead, our analysis
shows how this process evolves over time, how individual reactions become
group dysfunction, and describes the major steps involved. It confronts the
questions of why, when, and how this happens. And in the process we will
discover some research areas where our knowledge is solid and some other
areas where more work needs to be done. These are the focus and contri-
bution of the paper.
Responses to Negative Members
Several research efforts have investigated initial responses to the sort of
people we designate as withholders of effort, affectively negative, and in-
terpersonal deviants. The following section concerns itself with a description
of these responses.
Across disparate literatures, the same reactions to negative behavior crop
up again and again under different labels. We believe that these reactions
can be parsimoniously collapsed into three classes of teammate response –
motivational intervention, rejection, and defensiveness. Each of these three
responses have a common foundation; the desire to improve an aversive
experience. However, where these responses differ is in their aims – e.g.
towards either changing the negative person’s behavior (motivational
intervention), removing negative people (rejection), or protecting one’s
own self (defensiveness). If either the motivation intervention or rejection is
successful, the negative member never becomes a bad apple or spoils the
barrel. But it is still important to review these three responses in greater
detail.
We define the motivational intervention as those acts of teammates which
intend to change negative behavior though the application of influence tactics
(Orcutt, 1973). The literature provides evidence that the motivating response
is a common reaction to both withholders of effort (Jackson & Lepine,
2003) and interpersonal deviants (Taggar & Neubert, 2004;Schachter,
1951), but is used less frequently with affectively negative individuals. It
seems as though teammates lack efficacy in boosting a teammate’s negative
moods, and so tend to reject affectively negative individuals rather than
attempt to motivate them (Helweg-Larsen, Sadeghian, & Webb, 2002). This
is an example of the broader finding from the attributional research liter-
ature that motivating responses are particularly likely when the focal per-
son’s behavior is ascribed to controllable causes (Jackson & Lepine, 2003;
Green & Mitchell, 1979;Sampson & Brandon, 1964;Taggar & Neubert,
2004;Weiner, 1993).
How, When, and Why Bad Apples Spoil the Barrel 185
In any case, when team members do believe change is possible, motivating
actions may include the withholding of praise, respect, or resources until
behavior changes (Hackman, 1976), subtle and not so subtle confrontations
(Lepine & Van Dyne, 2001;Lubit, 2004), formal administration of punish-
ments (Liden et al., 1999;Hackman, 1976), or demands of apology and
compensation (Bies, Tripp, & Kramer, 1997). A classic example of team-
mates motivating a negative member can be found in the Hawthorne studies
(Homans, 1950). When a person was not working hard enough (what the
men at the plant called a ‘‘chiseler’’) co-workers would ‘‘bing’’ the man on
the upper arm and criticize his laziness. This was remarkably effective, more
so than managerial supervision or incentives. In another early ethnography,
Rosabeth Moss Kanter (1972) reviews how the Oneida community used
‘‘public criticism’’ as a formal mechanism to ensure that those who deviated
from the norm were provided ‘‘enlightening’’ feedback. Of course, these
formal and informal punishments might be coupled with positive reinforce-
ment for more desirable behaviors. Whether explicit or implicit, punish-
ments or rewards, a motivational response means that teammates will try to
bring negative members back into the fold by changing their behavior.
Multiple taxonomies also identify rejection as a common response to
negative members, especially after motivational attempts fail (Orcutt, 1973).
For our purposes, rejection can be defined as those acts which intend to
minimize or eliminate interaction with the negative member. There is evidence
that rejection is a common response for withholders of effort (Lepine et al.,
1997), for affectively negative individuals (Coyne, 1976;Furr & Funder,
1998;Helweg-Larsen et al., 2002), and for interpersonal deviants (Taggar &
Neubert, 2004;Schachter, 1951). Like motivational responses, research on
attributions has been instrumental in predicting when rejection will occur –
namely when negative behavior is ascribed to stable and uncontrollable
causes (Jackson & Lepine, 2003;Green & Mitchell, 1979;Sampson &
Brandon, 1964;Schachter, 1951;Taggar & Neubert, 2004;Weiner, 1993).
The most prototypical example of rejection would involve ejecting a neg-
ative individual from the group. Lacking this option, members of groups with
a fixed constituency will change the ‘‘psychological composition’’ (Festinger,
1950) of the group by ostracizing negative members, reducing social inter-
action, talking at rather than with, exclusion from decisions, or removing
responsibilities that require them to interact with others (Hackman, 1976;
Lepine et al., 1997). Alternately, when ostracism is unfeasible due to organ-
izational constraints such as seniority or formal role sets, the difficult person
may be ‘‘rejected’’ in more subtle ways. Teammates can restructure work to
decrease task interdependence, or segment responsibilities so that goals and
WILL FELPS ET AL.186
rewards are less interdependent. As a concrete example, faculty at a university
might decide to forego an integrated curriculum in order to avoid having to
interact with a frustrating individual. In summary, this response type entails
rejecting the negative individuals through expulsion, psychological distancing,
or altering task interdependence to reduce the impact of the negative be-
havior.
If they work, both motivational interventions and rejection are fairly
constructive responses to a negative individual. They represent what is
probably a minor distraction from task performance; a bump in the group’s
unfolding path towards goal attainment. It could even be argued that these
two responses might serve as mastery experiences (Bandura, 1986) that
could strengthen members’ efficacy in dealing with difficult social situations,
and reaffirm the group’s normative order (Dentler & Erickson, 1959). While
little empirical evidence exists about the net effect of motivating or rejecting
a negative individual, we would suggest that the ultimate consequence will
be modest, either way. However, more severe effects can be expected if
motivation or rejection isn’t possible – that is if the social context is con-
strained in such a way that group members are powerless to motivate or
reject.
Accordingly, the final category of response is defensiveness. For our pur-
poses, defensiveness is defined as those acts which intend to protect and repair
one’s own sense of autonomy, status, self-esteem, or wellbeing. Manifestations
of defensiveness can include lashing out, revenge, unrealistic appraisals, dis-
traction, various attempts at mood maintenance, and withdrawal. When
motivation and rejection fail, groups are faced with the dilemma of a neg-
ative member who they cannot change or get rid of, the primary condition
under which a ‘‘bad apple’’ might ‘‘spoil the barrel’’. As such, defensiveness
will be a major focus of our analysis and is discussed in much greater detail
as we proceed.
Antecedents to Defensiveness
As mentioned above, a motivation intervention or rejection requires that
teammates have some power. When unempowered, teammates become
frustrated and defensive. According to Janis and Mann’s (1977) model of
decision-making, members of groups become defensive when all decision
alternatives have low probabilities for success. In the case of the bad apple,
frustration is caused by an individual who behaves in dysfunctional ways,
has a negative impact on personal well-being, impedes performance – and
yet, due to organizational constraints on acceptable social action – cannot
How, When, and Why Bad Apples Spoil the Barrel 187
be easily reformed or rejected. When there’s no viable way to deal with a
harmful person, but members are still strongly influenced by them, the only
recourse is defensive self protection.
The inclusion of defensiveness as a reaction to a negative member rec-
ognizes that peoples’ reactions to difficult circumstances (especially if at-
tempts to change the situation fail or cannot be tried) are often less than
rational. Moreover, in contrast to responses like rejection or motivation,
defensiveness does not resolve the negative member problem; rather, it can
intensify the problem as teammates either withdraw or lash out in emo-
tionally motivated attempts to protect themselves. In the following section,
we discuss the two key factors that promote defensiveness: a lack of power
and the basic psychological tendency to react strongly to negative behavior.
In conjunction, these two answer the question of why bad apples can have
asymmetric negative effects on others.
Low Power Situations
Group members can be relatively powerless either because the negative
member has power or because the group member in question does not. The
negative member’s power may originate from social resources, such as per-
sonal connections to higher ups, prestigious degrees, or knowledge of ‘‘where
the skeletons are buried’’ (Morrill, 1995). Power could also originate from
structural characteristics, such as instances when others are highly dependant
on the negative individual for unique knowledge or skills (Robinson &
O’Leary-Kelly, 1998), or when the negative individual is placed at a critical
juncture in workflow (i.e. a secretary or facilitator) (Doerr, Mitchell,
Schriesheim, Freed, & Zhou, 2002). Finally, power can be formal, such as
whenever the negative individual has direct control over the allocation of
rewards and punishments. Whether leaders are more or less likely to be bad
apples is an unanswered empirical question. Organizations will probably
attempt to avoid hiring or promoting difficult individuals for leadership
positions, but research suggests that dysfunctional people do hold leader-
ship positions with some frequency (Ashforth, 1994;Pearson, Andersson, &
Porath, 2000).
Finally, teammates themselves may not have the power needed to respond
to a negative member. In many cases group members may look to their
leader to punish a deviant group member (Butterfield, Trevino, & Ball,
1996). Poor leadership may allow a negative person to persist in their
destructive activity. Relatedly, the group members may lack the resources
or empowerment to enact change. Kirkman and Rosen (1999) suggest
that members of the groups with low empowerment will not have the
WILL FELPS ET AL.188
decision-making authority, responsibility, adequate experience, or confi-
dence to take decisive action. Thus, powerlessness constrains the available
response behaviors. But paradoxically, this powerlessness in the face of
threat is also extremely frustrating and is actually likely to intensify psy-
chological reactions to bad apple behavior.
Bad is Stronger than Good
As reviewed by Baumeister, Bratslavsky, Finkenauer, and Vohs (2001),
‘‘bad is stronger than good’’ in many areas of human psychology. Negative
cognitions, feelings, and events will usually produce larger, more consistent,
and long-lasting effects as compared to equivalent positive thoughts, feel-
ings, and events. Manifested in varied and subtle ways, this pervasive phe-
nomenon holds across information interpretation, impression formation,
relationship maintenance, experiencing emotions, memory, learning, and
health (Baumeister et al, 2001;Lewicka, Czapinski, & Peeters, 1992;
Rozin & Royzman, 2001;Taylor, 1991). Lewicka et al. (1992) and Skowron-
ski and Carlston (1989) have found that the strength of bad over good also
holds in social environments, where negative interpersonal interactions elicit
uncertainty, anxiety or fear, such that processing these events becomes a
high priority.
Adaptability is the rationale underlying Baumeister’s arguments for the
relative salience and influence of negativity. Generally, negative events have
greater survival implications and denote more information than positive
events about the environment. According to Baumeister et al., the strength
and salience of bad over good ‘‘may in fact, be a general principle or law of
psychological phenomena possibly reflecting the innate predispositions of
the psyche or at least the almost inevitable adaptation of each individual to
the exigencies of life’’ (p. 323).
The ‘‘bad is stronger than good’’ effect is especially noticeable in the
social realm. Studying romantic relationships, Gottman and coworkers
(Gottman & Krokoff, 1989;Levenson & Gottman, 1985) found that the
frequency, intensity, and reciprocity of negative interactions are much more
predictive of marital satisfaction and divorce than are positive interactions.
Gottman’s (1994) rule of thumb is that positive interactions must outnum-
ber negative ones by a ratio of 5:1 if the relationship is to have a good
chance of success. Additionally, Baumeister et al. (2001) review nine studies
which compare the effects of social support and social undermining across
diverse populations. They summarize their findings by saying that ‘‘[t]aken
together, these studies suggest that helpful aspects of one’s social network
bear little or no relation to depression, well-being, and social support
How, When, and Why Bad Apples Spoil the Barrel 189
satisfaction, while upsetting or unhelpful aspects do y. Bad interactions
have stronger, more pervasive, and longer lasting effects’’ (p. 340).
Recent research in organizations has also explored the topic of negative
relationships and behavior, confirming that bad is often stronger than good
in this setting. Gersick, Bartunek, and Dutton (2000) conducted numerous
interviews with academics about relationships that influenced their careers.
While positive relationships were more frequent according to the academics’
self-reports, the negative ones were reported to be very important with a
substantial impact on career success. A recent paper by Labianca and Brass
(in press) finds that while negative relationships may be rare (constituting
between 1–8% of ties), they have greater impact on job satisfaction and
organizational commitment than do positive or neutral associations. These
scholars also find that negative effects are most pronounced in high density,
high interdependence situations (e.g. teams). Finally, in a study of fast food
restaurants, Dunlop and Lee (2004) compared the effects of organizational
citizenship behaviors and deviant workplace behaviors. They found that
deviant behaviors explained considerably more of the variance in subjective
and objective work group outcomes than did the citizenship behaviors.
A lack of power is what prevents reform or rejection, and the ‘‘bad is
stronger than good phenomenon’’ is what allows negative team members to
have an asymmetrically strong effect on others. By extension, this asym-
metric effect explains why dysfunctional individuals are an important con-
cern for groups. In interdependent teams where people depend on each
other, these intense psychological reactions are more likely to spillover be-
yond dyadic interactions to influence the broader social environment. As
noted by Baumeister et al. (2001), ‘‘in order for a system to function effec-
tively, each component of the system must do its part.’’ At the level of the
individual’s relation to the group, bad is undeniably stronger than good; any
individual part can prevent the system from functioning; but no individual
part can by itself cause the system to succeed. This is especially true of social
groups ymarked by a division of labor’’ (p. 358). In summary, the conjoint
of intense psychological reactions at the individual level, and spillover
effects onto group dynamics underlies the assertion that a ‘‘bad apple can
spoil the barrel’’.
NEGATIVE PSYCHOLOGICAL STATES
In this case, we are confronted with a situation where a member’s behavior
is persistently and consistently negative. The bad behavior is noticed and
WILL FELPS ET AL.190
influential in its effects on group members who do not have the power or
wherewithal to enact change. What happens now? We will review the likely
psychological states that emerge in response to each of the three negative
member behavioral categories.
The Withholder of Effort. A bad apple who withholds effort from the
collective triggers some undesired cognitions. If free riding persists, team-
mates face the challenge of correcting equity imbalances in input to outcome
ratios relative to others (Adams, 1963). Research finds that the most com-
mon referent that people look to for social comparison (the ‘‘other’’ in
the equity formulation) are the peers one works with every day (Kulik &
Ambrose, 1992). It follows that social loafing by a teammate can be a major
source of distress, as it violates effort norms and takes advantage of other
members’ good-faith contributions. It is also important to note that being
under-rewarded, as is the case here, produces stronger psychological effects
than being over-rewarded (e.g. Bloom, 1999) – another example of ‘‘bad
being stronger than good’’. As such, perceptions of inequity will arise when
group members compare their own contributions to those of a withholder of
effort in their team, and will result in a desire to restore equity by reducing
contributions (Jackson & Harkins, 1985;Schroeder, Steel, Woodell, &
Bembenek, 2003). However, due to task interdependence, this scenario cre-
ates a dilemma for contributing group members in which they are motivated
to avoid being a ‘‘sucker’’ and decrease their own contributions to the group
– but in doing so they risk rupturing relations with other members and
compromising group outcomes themselves. Thus, withholders of effort pro-
duce feelings of inequity with no easy resolution in a team environment.
The Affectively Negative Individual. Affectively negative individuals in-
fluence their teammates’ affect (including attitudes, moods and emotions).
Empirical work has shown that simply observing another person’s expres-
sions of affect can generate those feelings in others. Hatfield, Cacioppo, and
Rapson’s (1994) book Emotional Contagion describe how the diffusion of
affect is ‘‘unintentional, uncontrollable and largely inaccessible to aware-
ness’’ (p. 5), picked up unconsciously through facial expressions, vocaliza-
tions (e.g. tone, intensity, volume), postures, and movement. Using a
confederate trained to display positive and negative affect, Barsade (2002)
found that subjects working together on a task partially adopted the con-
federate’s mood. Even more simply, subjects observing angry facial expres-
sions quickly become angry themselves (Dimberg & Ohman, 1996). The
negative emotions engendered by bad apple behavior may also be long
lasting. Whereas a positive emotion (i.e. compassion) wears off relatively
quickly, researchers find that when they give someone a negative feeling
How, When, and Why Bad Apples Spoil the Barrel 191
(i.e. anger) to concentrate on, the physiological effects last over 5 h (Rein,
McCraty, & Atkinson, 1995). An extension of the negativity bias would
suggest that individuals will pay more attention to negative others and are
therefore prone to use negative others as a referent for social comparisons,
give negative emotional information more credibility, experience negative
emotions for a longer period, and ruminate more on negative events (Ba-
umeister et al., 2001;Rozin & Royzman, 2001). However, this hypothesis is
tempered by the lack of support for Barsade’s (2002) hypothesis that neg-
ative affect would spread more completely through the group than positive
affect. Clearly, more research is needed to understand if and when negative
affect will have an asymmetric effect.
The Interpersonal Deviant. As described earlier, the interpersonal deviance
category is defined by seven behaviors (e.g. making fun of a teammate,
acting rudely, saying something hurtful, etc). It is therefore somewhat
broader than the withholding effort and affectively negative categories.
Despite that breadth, we believe that these behaviors have similar goals and
mort importantly, similar consequences. More specifically, the main effect
of an interpersonal deviant is to undermine trust in that individual.
In teams, this can be problematic, since members depend on each other to
take advantage of division of labor efficiencies or develop transactive mem-
ory models (Wageman, 2000). Conversely, distrust in a group member re-
quires increased monitoring of the interpersonal deviant, and can distract
from task performance. Like inequity and negative emotions, trust is
also asymmetric, easier to damage than it is to build (Lewicki & Bunker,
1995).
More Complex Psychological Effects of Negative Teammates
The above discussion suggests some simple, direct effects of each type of
negative behavior – namely that withholding effort produces perceptions of
inequity, affective negativity spreads contagiously to teammates, and inter-
personal deviance engenders distrust. However, beyond direct effects, each
of these states can also have a secondary impact on the other two. With
respect to inequity, although Adams’ original focus was on cognition, other
research has clearly demonstrated that inequity also produces strong emo-
tional reactions (Goodman, 1977), and one can expect trust in a difficult
team member to deteriorate. With respect to emotions, negative feelings
trigger the search for mood-congruent cues (Meyer, Dayle, Meeham, &
Harman, 1990), and ambiguous social information is more likely to be
interpreted as inequitable or signaling untrustworthiness. Finally, since
WILL FELPS ET AL.192
trustworthy behavior is generally expected, a secondary consequence of
distrust is negative feelings such as anger, anxiety, and fear (Kramer & Wei,
1999). The ‘‘collateral damage’’ is potentially extensive.
Moreover, to fully consider the effect of any one specific negative member
requires other considerations. For example, imagine a person who is se-
verely depressed. They are highly likely to be affectively negative, but they
might also be unmotivated to put forth much energy into tasks – e.g. with-
holding effort from the group. Or consider the interpersonal deviant who
yells and bullies at the slightest provocation while concomitantly expressing
pessimistic attitudes. A benefit of understanding the primary and secondary
effects of each class of bad apple behavior is that these combinations can be
addressed. At the current time, little evidence exists to guide predictions of
how these behaviors might interrelate. However, at least three theoretical
possibilities exist. One alternative is that multiple behaviors will be largely
independent (i.e. be additive) such that someone who displays two catego-
ries of behaviors will have double the impact of a member who engages in
only one. Another possibility is that there is a limit to how upsetting one
individual can be, with multiple types of negative behavior drawing from the
same reservoir of defensiveness. A third option is that different types of
negative behavior will interact to reinforce and compound each other, re-
sulting in ultimately larger impacts on teammates.
Finally, it seems to us that while negative affect can definitely cause un-
constructive outcomes, the withholding of effort and particularly interper-
sonal deviance can cause even more acute negative effects. The interpersonal
deviant directly and powerfully threatens other members and challenges the
normative integrity of the group as a whole. Given the interdependence of
groups, the sense of inequity produced by a withholder of effort will likely
also be quite distressing. In contrast, affective negativity may have a smaller
effect size since it operates through the less direct (and arguably less pow-
erful) mechanism of contagion. But again, these are conjectures for future
research. To the best of our knowledge, no studies have compared the effect
sizes of these negative behaviors against each other. Next, the discussion
elaborates on the consequences of teammate psychological states on be-
haviors.
Defensive Behavioral Reactions
Generally, defensive responses are self-protective efforts to cope with a
negative internal state. This negative state might arise from threats to au-
tonomy (Ashforth, 1989), identity (Aquino & Douglas, 2003), self-esteem
How, When, and Why Bad Apples Spoil the Barrel 193
(Baumeister, Dale, & Sommer, 1998) or general well-being (Berkowitz,
1989). Persistent and consistent harmful behavior by a negative member
challenges these core concerns and leads to ongoing perceptions of threat.
These threats can be countered in two ways – externally or internally. Ex-
ternally directed responses include acting against the negative member to
restore feelings of autonomy, identity, self-esteem, and well-being. Internally
directed responses involve taking steps to change one’s own moods, emo-
tions, or appraisals. Our subsequent discussion will include external forms
of defensiveness, such as emotional explosions or revenge, as well as more
internally focused efforts, such as mood maintenance, distraction, denial,
and withdrawal from the group. However, while different, both types of
defensiveness are caused by the same psychological states, and both lead to
dysfunctional group processes and outcomes.
When experiencing aversive events, people often react emotionally
(Berkowitz, 1989). Following Bies et al. (1997), we call this defensive re-
sponse ‘‘exploding’’. Exploding is a direct and intense release of negative
feelings, and is usually motivated by the desire to dominate or attack a
frustrating person (Aquino, Galperin, & Bennett, 2004). However, explo-
sions often lead to retaliation from those who are the target of these emo-
tional releases. As such, they can sometimes result in an escalating tit-for-tat
spiral of retaliation (Andersson & Pearson, 1999).
Additionally, rather than emotionally exploding, a person can defend
themselves through the more controlled act of revenge. Revenge is moti-
vated by a desire to restore perceptions of equity and justice. As noted by
Bies et al. (1997) ‘‘Any perceived inequities on the job or violations of
fairness norms can motivate revenge’’ (p. 21). Using their extensive inter-
views, they go on to note what kinds of things provoke revenge and uncover
precisely what we would call bad apple behaviors. ‘‘Violations include
bosses or co-workers who shirk their job responsibilities, take undue credit
for a team’s performance, or outright steal ideas’’ (p. 21). Morrill’s (1995)
ethnography, The Executive Way documents that managers are often loath
to confront each other directly, but are still ingenious in the ways they
sabotage those who frustrate them. For example, Morrill tells of coworkers
who enact revenge by giving the ‘‘perpetrator’’ wrong information, distorted
files, or sending them on ‘‘wild goose chases’’. However, experiments in the
lab point out an inherent difficulty of revenge in the team settings. Using a
social dilemma framework, Chen and Bachrach (2003) found that when a
single individual free rides across experimental trials it led to an asymmetric
and precipitous decline in teammate contributions. One interpretation of
this finding is that offended members wanted to restore equity perceptions,
WILL FELPS ET AL.194
but could not get even without also harming themselves and their group.
This prevented the group as a whole from provisioning the social good
and meant that all members were worse off. Chen and Bachrach’s study
underscores that in interdependent teams, confining the effect of revenge
acts is often difficult. Next, we turn to internal manifestations of defen-
siveness.
When feeling emotionally negative, people often take action to improve
their mood. Mood maintenance behaviors are efforts to improve one’s affect
and can be either consciously or unconsciously motivated (Baumeister,
Heatherton, & Tice, 1994;Thayer, 1996). For group members, examples
may include the seeking out of positive social interactions – i.e. lunch out-
ings, happy hour, etc. – or more individual mood elevators like taking
breaks, eating, or smoking. While perfectly functional for the individual,
mood maintenance may have an adverse affect on the group. Indeed, a
laboratory study by Tice, Bratslavsky, and Baumeister (2001) found that
repairing negative emotions takes precedent to considerations of task per-
formance when people are emotionally depleted. As such, people at their
wits end might socialize with others, eat a treat, or surf the internet, but tend
to direct attention away from the task performance.
Said another way, a negative member can be a distraction. In an article by
Andrews (2004), one interviewee stated: ‘‘If you’ve ever been in a situation
where you feel offended by the behavior of a coworker – you know that you
can’t bring your best effort to work. Emotionally, intellectually and be-
haviorally, you’re just not going to be all there’’ (p. 45). Supporting this
assertion, field work by Pearson et al. (2000) found that over one half of
those who experienced incivility at work reported that they lost time wor-
rying about the uncivil incident and its future consequences. Other research
on affect also confirms that feelings of anxiety, anger, or sadness tend to
distract and demotivate (George & Brief, 1996).
A fourth form of defensiveness is denial, a strategy by which an individual
avoids dealing with negative events by behaving as if group problems are
not occurring, significant, or the result of the negative member. Denial has
been evocatively described as ‘‘a primitive and desperate method of coping
with otherwise intolerable conflict, anxiety, and emotional distress or pain’’
(Laughlin (1970, p. 57), originally cited in Brown (1997)). However, the
interdependence of group work and the persistence of negative behavior
conspire to make denial at best only a temporary stop-gap to the negative
group member problem. One can only override genuine emotions for so
long before becoming emotionally depleted (e.g. Baumeister et al., 1994) and
suffering the explosive effects mentioned above.
How, When, and Why Bad Apples Spoil the Barrel 195
The final defensive response we will explore is withdrawal from the group.
Social interactions are often stressful, and are likely to be more so in the
presence of a negative teammate. As such, a particularly easy, and hence
probable, response is to withdraw into oneself by not fully engaging in the
group (Bergman & Volkema, 1989;Bies et al., 1997). Pearson and Porath
(2005) document that 20% of the workers they interviewed report that they
reduced their rate of work as a result of incivility and 10% said they de-
liberately cut back the amount of time they spent at work. Pearson et al.
(2000) find that over 25% of individual who were targets of incivility
acknowledged withdrawing from work situations. They summarize their
findings by noting,
Through all phases of our study, people told us that after being targets they ceased
voluntary efforts. Some stopped helping newcomers; others stopped offering assistance
to colleagues. Additionally, targets reduced their contributions to the organization as a
whole, whether by pulling themselves off task forces and committees, or by reducing
efforts to generate or inspire innovation (p. 130).
More extremely, teammates might even exit the group to escape the negative
thoughts and feelings induced by a negative member. Pearson et al.’s data is
instructive, finding that half of the individuals interviewed contemplated
leaving their jobs after being the target of incivility, and a full 12% reported
actually quitting.
We have reached a point in our discussion where the negative mem-
bers’ behaviors have undermined perceptions of equity, mood, and trust.
Members may respond defensively to these psychological states via explo-
sions, revenge, mood maintenance, distraction, denial, and withdrawal. In
sum, withholding effort, affective negativity, and interpersonal deviance
can each trigger defensive thoughts and behaviors with powerful conse-
quences.
MODERATORS OF THE BAD APPLE EFFECT
Thus far, we have reviewed the factors that motivate members of teams
to respond defensively to a difficult individual. However, this response
is moderated by several factors, which influence when bad behavior will
impact the psychological reactions and subsequent actions of teammates.
Specifically, four variables emerge from the literature that seem espe-
cially important in determining perceived impact severity – (1) intensity
of the negative behaviors exhibited, (2) the group’s interdependence,
WILL FELPS ET AL.196
(3) whether outcomes are successes or failures, (4) and the teammates’ cop-
ing abilities.
Intensity of Negative Behaviors. The potency and frequency of negative
behavior will determine its perceived intensity. First, of the three classes of
behavior that have been identified as likely to elicit a group response (e.g.
withholding effort, affective negativity, and interpersonal deviance), each
has a range of severity. One affectively negative individual might be ex-
tremely pessimistic, while another might be only mildly depressed. Indeed,
the widely employed ‘‘circumplex’’ model of emotion is based on an inten-
sity dimension (Larsen & Diener, 1992), as is Ajzen’s (2001) conceptual-
ization of attitude. Similarly, the withholder of effort might slack off a little
or do next to nothing. The literature on social loafing recognizes this and
measures free riding as a continuous variable (Karau & Williams, 1993).
Further, the interpersonal deviant might purposefully sabotage other’s
efforts or display the milder behavior of mean-spirited criticism. Robinson
and Bennett’s (1995) inductive typology of interpersonal deviance is sup-
portive, finding that people naturally categorize deviance from mild to se-
vere. In sum, potency is a central part of theories of effort, affect, and
deviance. Second, in addition to the behaviors exhibited, the frequency of
those actions is likely to play a role in perceptions of intensity. In an in-
teresting analogy, Cunningham, Barbee, and Druen (1997) suggest that
aversive behaviors can be thought of as ‘‘social allergens’’, where increased
exposure leads to increased sensitivity. However, this fascinating hypothesis
has yet to be tested. Regardless, more potent and frequent negative member
behaviors will have a greater impact on teammates.
Interdependence. If the group is highly interdependent, then dysfunctional
behavior is of more consequence. Groups can be interdependent to varying
degrees in terms of tasks, goals, feedback, or rewards (Wageman, 2000).
Highly interdependent groups have more interaction and the content of that
interaction is more central to accomplishing the work task. As such, high
interdependence means there are more opportunities for affect to contagiously
spread to others and a greater chance for interpersonal attacks. In addition,
the inequity caused by shirking is more noticeable and meaningful when
members are interdependent and receive rewards based mainly on group ac-
complishment. Whereas a group that is not interdependent allows members to
‘‘do their own thing’’, a highly interdependent group provides less opportunity
for avoidance. The experience of threat is ever-present, and so is the chance
of acrimonious interpersonal conflict. This is especially problematic since in-
terdependent tasks necessitate that a group maintains higher quality social
relationships in order to effectively coordinate their activities (Gittell, 2003).
How, When, and Why Bad Apples Spoil the Barrel 197
Outcomes. Work team outcomes can exert a powerful influence on the
perceived severity of negative member behavior. After a team failure occurs,
the negative member behaviors are more salient, and thus more influential.
According to attribution theory (Weiner, 1980, 1995), failure triggers the
process of determining causal factors, and relatively innocuous behavior can
be reclassified as a significant threat to team functioning. If unchangeable,
this newly salient dysfunction provokes the defensive reactions we have de-
tailed. In addition the severity of the outcome can influence the response. This
assertion is supported by Mitchell and Wood’s (1980;Mitchell, Green, &
Wood, 1981) research, which gave nurse managers scenarios of offenses that
nurses had actually committed. In one condition, the nurse had left down a
bed rail and the patient fell out and broke a hip, while in another the nurse
had made the same mistake, but the patient did not fall. The punishments
that managers recommended in the first condition were quite severe, includ-
ing dismissal and probation. The punishments were much milder in the sec-
ond condition, with the most common response being a verbal reminder of
hospital procedure. Accordingly, reactions by group members to negative
behavior will be more extreme when the behavior results in failure outcomes,
and when those failure outcomes are more consequential.
Coping Skills. Finally, individuals are also likely to differ in their personal
coping skills. A high locus of control would lead to beliefs that life events
and reactions to life events are controlled internally. If teammates have high
self-esteem, they know that their essential needs will be met. If they have
high generalized self-efficacy, then they are likely to have confidence that
either the negative member or the situation can be changed. Further, if they
are calm (low neuroticism), then their reactions will be extreme. Notably,
the work by Judge and his coworkers on core self-evaluations integrates and
aggregates these four classic psychological variables – providing compelling
reasons and evidence for conceptualizing and measuring a single underlying
construct (Erez & Judge, 2001;Judge, Locke, Durham, & Kluger, 1998;
Judge, Van Vianen, & De Pater, 2004). These self-attributes are useful be-
cause they change the meaning of threatening situations. For example,
someone with a highly positive core self-evaluation might interpret inter-
personally deviant behavior as merely a nuisance rather than a substantial
threat. Or they might find a silver lining to the situation, such as a chance to
learn conflict management skills. Using such mental techniques, those with
high core self-evaluations are likely to be motivated and able to reconstruct
the meaning of the bad apple’s behaviors to be less negative. In summary, if
a teammate has extensive coping resources then negative behaviors will have
less intense psychological impact.
WILL FELPS ET AL.198
GROUP TRANSITION MECHANISMS
Thus far, we have defined the behaviors that make someone a negative
group member and described how chronic display of those behaviors can
subsequently influence other individuals to feel and act defensively. So far,
this description has been initially unidirectional, then dyadic. However, we
mentioned at the beginning of this paper that most of the research on team
effectiveness has focused on how team attributes and processes result in
effective team performance. At this point in our analysis, we will explore
how individual states and actions transition to group constructs and be-
havior, and move from one conceptual level to the next.
One of the major shifts in team research documented by Ilgen et al. (2005)
is that more emphasis is being placed on multilevel theoretical and analytical
contributions. Ilgen elaborates on the fact that organizations are multilevel
and that many of the variables central to understanding teams appear at the
group level as well as the individual level. He also points out that there are
many parallel constructs, ones that have both an individual and team
counterpart. For example, motivational constructs such as efficacy and
emotional constructs like mood can be construed at both these levels. The-
oretically, these collective constructs are usually assembled from individual
interactions. When A talks to B, and B responds in some way, we have what
Weick (1979) calls a ‘‘double interact’’. It is the structure and function of
these double interacts that are the building blocks of collective constructs.
These ‘‘[c]ollective structures emerge, are transmitted, and persist through
the actions of members of the collective’’ (Morgeson & Hofmann, 1999,
p. 53). We support Morgeson and Hofmann’s notion that ‘‘[i]ntegrating
across levels may provide a more veridical account of organizational phe-
nomena’’ (p. 249). The question for the moment is how these individual
interactions, which we have described are translated into group constructs
and then into group action. We describe three mechanisms below: addition,
spillover, and sensemaking.
Additive Defensiveness. The simplest and most obvious transition occurs
using an additive mechanism. Obviously, the more types of negative be-
havior, and the more interactions with team members, the more negative
psychological states and defensive behaviors will accrue. Brass, Butterfield,
and Skaggs (1998) discuss how the impact of a negative member on a team
depends on the ratio of contacts the person has with group vs. non-group
members. Duffy, Ganster, and Pagon (2002) summarize their discussion of
social undermining behaviors by commenting that ‘‘their efforts add up over
time’’ (p. 233).
How, When, and Why Bad Apples Spoil the Barrel 199
Spillover Effects. A different mechanism for moving from dyadic ex-
change to group level constructs is caused by what we call a spillover effect.
The subtle and automatic form of spillover occurs through the process of
modeling behaviors. Seeing others act antisocially makes those behaviors
more mentally accessible and lowers inhibitions about behaving in a similar
fashion. Bandura’s famous ‘‘Bobo the Clown’’ studies demonstrate that
even strangers can be influential models of antisocial behavior (Bandura,
Ross, & Ross, 1963). These social learning effects are likely to be even
stronger in groups. Indeed, a paper entitled ‘‘Monkey See, Monkey Do’’ by
Robinson and O’Leary-Kelly (1998) found precisely that; the more in-
terdependent the social context, the greater the effects of social learning.
Keaton (1999) even suggests that these other team members can become
‘‘secondary provokers’’ or negative members themselves. In short, through
mimicry and modeling, spillover effects of negative thoughts, feelings, and
actions can move from individual to group level characteristics.
Spillover can also be seen in the phenomenon of displaced aggression.
While we are often able to use regulatory skills to control frustration in the
moment, as those resources are expended, group members become more
likely to lash out at others (Muraven & Baumeister, 2000). Sometimes those
others are entirely removed from the situations and people who are the
source of frustration (Marcus-Newhall, Pederson, Carlson, & Miller, 2000).
Research shows that provoked participants readily displace aggression onto
blameless individuals (e.g. Worchel, Hardy, & Hurley, 1976), especially
when social and status hierarchies constrain direct expression of aggression
– e.g. in comparatively low power situations (Marcus-Newhall et al., 2000).
Folger and Skarlicki (1998) describe this sort of spillover as a ‘‘popcorn
model’’ of aggression, where aggression or violence can ricochet throughout
a group; setting off one individual after another and lowering everyone’s
inhibitions.
Just as contagion serves as a mechanism for spreading mood from A to B,
it can also spread from B to C, C to D and so on; spillover occurs when team
members’ individual responses to the bad apple start to have an impact on
other team members, an ‘‘interaction breeds similarity’’ effect (Brass et al.,
1998, p. 25). In one of the more definitive pieces of evidence to date,
Barsade’s (2002) article on the ‘‘ripple effect’’ found that a confederate
displaying physical manifestations of negative affect (e.g. posture, manner-
isms, facial expressions) was able to engender negative moods in groups, and
multi-level modeling techniques (HLM) affirmed that these effects perme-
ated the group above and beyond dyadic contagion. Bartel and Saavedra
(2000, p. 197) describe this effect in their research, stating that ‘‘Group
WILL FELPS ET AL.200
members come to develop mutually shared moods and emotion’’. Evidence
of these affective spillover effects has accumulated in recent years (Bakker &
Schaufeli, 2000;George, 1990;Totterdell, Kellett, Teuchmann, & Briner,
1998). The transfer of affect is largely automatic and subconscious, occur-
ring through mimicry and psychological feedback (Hatfield et al, 1994).
Sensemaking Effects. More conscious processes can occur as well. In
many cases a negative member may act out in a public context (e.g. bully a
teammate, refuse to contribute in a social problem solving context) or be-
have so egregiously that it requires sensemaking by one or more team
members (Weick, 1995). The recipient of an attack, or an observer of one,
may seek out the advice and interpretation of other team members or even
outsiders. Social communication can be an important part of individual
sensemaking (Hardin & Higgins, 1996). Pearson and Porath found that
over 90% of people who were treated badly (i.e. uncivilly) say they sought
the counsel of someone else. Moreover, research by Rime, Finkenauer,
Luminet, Zech, and Philippot, (1998) describes the process of ‘‘secondary
social sharing’’ where those who have heard about frustrating interactions
themselves share it with others. Rime’s research indicates that this secondary
social sharing occurs with surprising frequency – around two thirds of the
time negative events are shared a second time. Finally, their studies show
that such sharing is especially likely to happen when the event is intense or
negative (Christophe & Rime, 1997;Luminet, Bouts, Delie, Manstead, &
Rime, 2000).
An obvious outcome of this sensemaking process is that people agree that
the negative member is different and dysfunctional and the group tries to
change or reject this person. However, it is also possible that neither re-
sponse is viable (described earlier), and under these circumstances the neg-
ative effects are likely to have a wider and more substantial impact on the
team. Lacking power to enact change prompts group member sensemaking
about one’s own relationship to the group. When a group has lost its in-
strumental ability to effectively enforce norms, elicit cooperation and
achieve goals, members may no longer recognize the team as a desirable
entity with which to be associated. When members loose faith in the groups
of which they are a part, it is called de-identification (Dutton, Dukerich, &
Harquail, 1994). One of the major drives behind identifying with a collective
is the desire to be part of something positive that enhances one’s own self
concept (Dutton et al., 1994). As the group loses its positive ethos, members
de-identify from the collective and categorize themselves more as an indi-
vidual and less as a part of the group. As members physically and psycho-
logically disengage, the character of the group is marked by decreasing
How, When, and Why Bad Apples Spoil the Barrel 201
commitment to group goals and dissatisfaction with team membership
(Ouwerkerk, Ellemers, & de Gilder, 1999). In closing, it is sufficient to say
that the individual actions of a negative member can spread in various ways
to the group – through aggregation, spillover, and sensemaking – and that it
is through these transformational mechanisms that dyadic effects come to be
a group level phenomenon – i.e. a spoiled barrel.
GROUP CONSTRUCTS
We have argued that the individual and dyadic effects of the negative
member can be transmuted into group constructs – what Cohen and Bailey
(1997) call group psychosocial traits – through the mechanisms off aggre-
gation, spillover, and sensemaking. In the abstract, group constructs
are mental heuristics to think about qualities of a collective (Morgeson &
Hofmann, 1999). However, when recognized and internalized by group
members, group psychosocial traits come to have a life of their own and
exist apart from individuals. As Weick and Roberts (1993) point out, people
‘‘construct their actions while envisaging a social system of joint action’’
(p. 363). In short, we act as if social groups have a character of their own,
and so, in a way, it comes to be true.
NEGATIVE MEMBER’S EFFECTS ON GROUP
PROCESSES AND OUTCOMES
Effective groups have two meta-skills – their members produce as individ-
uals, and together as a group they effectively coordinate and integrate in-
dividual action into a coherent whole constituting a group output
(Hackman, 1987). This first skill, the ability to produce, depends on hav-
ing a team that is motivated, capable, and able to learn and change. These
are the basic building blocks for performance, without which there would be
little to integrate. The second skill, group integrative actions, includes the
group processes of productive conflict and cooperation (Smith et al., 1994).
Having a bad apple in a group will have a negative impact on the group
production related processes of motivation, creativity, and learning and on
the integrative processes of cooperation and conflict. Without these proc-
esses in place, groups fail.
Motivation. Motivation to perform is central to work behavior (Mitchell,
1997). We have already discussed how motivation at the individual level
WILL FELPS ET AL.202
could suffer and, in addition, influence collective motivational constructs
such as group efficacy (Gully et al., 2000). Teams with lower efficacy exert
less effort, set lower goals, and perform less well than group with higher
efficacy (Gully et al., 2000). Beyond efficacy, a negative group affective tone
also has a deleterious affect on group performance (George, 1990). Negative
moods and emotions engendered by the negative member will distract other
team members from focusing on the task. This distraction might take
the form of ruminating on the negative interactions or gossiping about
them with others (Burt & Knez, 1995;Rimes et al., 1998). This assertion is
consistent with the findings of Grawitch, Munz, and Kramer (2003) that
negative group moods focus attention on interpersonal issues and away
from task concerns. Lastly, recent work by van Knippenberg (2000;van
Knippenberg & van Schie, 2000) suggests that since the prototype of a
‘‘good’’ employee is usually a motivated employee, group members who
categorize themselves as part of a healthy group will conform to that iden-
tity by displaying more task motivation. Thus, if a destructive group mem-
ber causes de-identification, there is likely to be a decrease in task effort and
persistence as the team members deviate from the ‘‘good worker’’ prototype
(see also Hogg, 2000 and Shamir, 1990). In summary, having a negative
member in the group will decrease motivation through the processes of
lowered efficacy, distraction (e.g. gossiping, affective rumination, and mood
maintenance), and de-identification.
Creativity and Learning. Creative problem solving is seen to be increas-
ingly important in groups (Paulus, 2000). In a recent article (Amabile,
Barsade, Mueller, & Staw, 2005) shows that positive affect facilitates cog-
nitive variation and yields new associations, thereby enhancing creativity in
a linear fashion. But creativity also depends on several fragile conditions,
including the free exchange of ideas, confidence that innovation is possible,
and the motivation to create (West, 2002). Further, the creative process of
coming up with new ideas is intimately related to the group’s ability to learn.
The same safe and motivated environment that allows groups to come up
with new ideas also allows them to learn and remember effective methods of
action (West, 2002). While learning and creativity are not synonymous, both
involve an intellectual openness to new possibilities, and are consequently
coupled together here.
The negative member’s behavior can have a major effect on the creative
and learning processes in groups. In inequitable situations, such as with a
withholder of effort, teammates are unlikely to be motivated to contribute
to the collective pool of ideas or to teach and learn from others (West, 2002).
In addition, numerous empirical studies have found that negative feelings
How, When, and Why Bad Apples Spoil the Barrel 203
have a chilling effect on creativity for individuals (see for a review Isen,
2000) and on groups (Grawitch et al., 2003). Specifically, research exploring
the contagion of the negative emotion of social anxiety has discovered that
the worst (i.e. most socially anxious) group member exerts a powerful
asymmetric effect on team creativity (Camacho & Paulus, 1995). Similar to
our affectively negative individual, the most socially anxious person par-
alyzed other members’ ability to creatively perform. Finally, threat generally
hinders inventiveness by restricting one’s behavior to well-established pat-
terns (West, 2002;Staw, Sandelands, & Dutton, 1981). A similar logic holds
true for learning in groups. A perception of threat triggers defensive reac-
tions aimed towards self-protection (Aquino & Douglas, 2003). Groups
composed of self-protective members will not feel safe, and so will be re-
luctant to do things like admit a knowledge deficit or ask for help in de-
veloping competencies (Edmondson, 1999, 2002), which will impede
learning. Finally, given that knowledge can be a source of power, those
who do not identify with the group are more likely to hoard information
and ideas for political purposes (Jones & George, 1998). If, by engendering a
hostile atmosphere, a negative member may cause the group to be mute
about problem areas and engage in political use of knowledge. Again, group
learning is likely to suffer. In sum, equity perceptions, group affective tone,
feelings of safety, and identification each play an important role in prompt-
ing creativity and learning but will be undermined by the behaviors of a
negative group member.
We now shift our attention to the ways that a negative member may
influence the integrative processes necessary to coordinate various members’
efforts. These integrative processes may be especially compromised as
team members rush meetings to hasten their escape from negative interac-
tions, and succumb to the common bias of coordination neglect (Heath &
Staudenmayer, 2000).
Cooperation. Cooperation is perhaps the most quintessentially ‘‘integra-
tive’’ component of group work. One way bad apples inhibit cooperation is
by undermining what has been called ‘‘depersonalized trust’’ or the ‘‘positive
expectation or presumption that interpersonal risks can be assumed with a
reasonable degree of confidence that others [in the group] will not betray or
violate the trust’’ (Kramer & Wei, 1999, p. 146). A central facet of deper-
sonalized trust is the knowledge that others will abide by norms of civil
behavior. When a negative member steals credit or spreads negative gossip,
other employees’ begin to lose confidence (i.e. decrease their expectations)
that cooperation will result in mutually beneficial outcomes. Kramer and
Wei note that a violation ‘‘may create problems that undermine the smooth
WILL FELPS ET AL.204
exchanges, disclosures, affirmations, and validations associated with group-
based trust (p. 147). According to rational models of human behavior, as
expectancies worsen, so will the motivation to cooperate (Bommer, Miles, &
Grover, 2003). Identity theory makes similar predictions along less ca-
lculative premises of human behavior. Lind and Tyler’s (1988) group value
model of behavior argues that cooperation is an expressive sign of feeling
respected and respecting others. When people identify with the group, they
feel a moral duty to cooperate (Kramer & Goldman, 1995) and sometimes
do so even when it is not in their best interest (Brann & Foddy, 1988;Dawes,
van de Kragt, & Orbell, 1990). On the other hand, when people categorize
themselves as individuals rather than as members of a group, they withdraw
from collective life by thinking and acting more selfishly (Kramer, Brewer, &
Hanna, 1996). In sum, decreased perceptions of depersonalized trust provide
an instrumental rationale for avoiding cooperation; and de-identification
produces expressive reasons for eschewing cooperation.
Conflict. Group conflict was once considered anathema (Robbins, 1974).
However, recent thinking and research indicates that under certain circum-
stances, conflict can benefit groups. Specifically, a distinction is drawn be-
tween relational conflict (i.e. about the person) and task conflict (i.e. about
how to work). While relational conflict indeed detracts and distracts, task
conflict can actually serve to reinforce social responsibilities, enhance de-
cision quality by checking assumptions, and clarify group members’ mental
models (Jehn, 1995;Tjosvold, 1998). It seems likely that the interpersonal
deviant and the withholder of effort are likely to provoke both immediate
and sustained relational conflict by breaking important norms such as mu-
tual respect and parity of effort. Evidence suggests that even the affectively
negative individual may prompt conflict by causing reactions of irritation,
condescension, and humorlessness (Furr & Funder, 1998). And as other
group members rebuke or retaliate against this member, relational tensions
will escalate (Andersson & Pearson, 1999). Moreover, some of the resulting
hostility is likely to be ‘‘displaced’’ towards other group members (Marcus-
Newhall et al., 2000), increasing overall relational conflict. Finally, by cre-
ating a threatening psychological environment, a negative member could
also cause people to retreat inwards, resulting in hesitance to engage in
constructive task conflict, since it may result in unpleasant acrimony. As
such, the groups with a negative member might experience relatively more
interpersonal conflict along with relatively less task conflict – a doubly
counter productive state of affairs. However, this is a place where our
knowledge is somewhat speculative and more empirical evidence would be
useful.
How, When, and Why Bad Apples Spoil the Barrel 205
In conclusion, through various individual cognitions (e.g. inequity, neg-
ative mood, and distrust) and group level constructs (e.g. lower mood, po-
tency, safety, and group-based trust), the key processes that make groups
effective (e.g. motivation, creativity, learning, cooperation, and task con-
flict) will be undermined.
Group Outcomes. These individual and group effects mean that the
ultimate outcomes for the group include poor performance, low viability
(e.g. a weakened social structure), and an unhappy team. Group perform-
ance will suffer as measured in terms of quantity, quality, and timeliness.
The link between group processes and group outcomes is a rich and well-
researched topic (see Cohen & Bailey, 1997; Campion, Medsker, & Higgs,
1993;McGrath, 1984). So as not to reinvent the wheel, we will merely
reiterate that group behavioral variables such as motivation, creativity, co-
operation, and conflict are central mediators between inputs such as group
member’s abilities and the key outcomes of performance, worker well-being,
and group viability. However, one interesting long-term consequence of the
negative member invites further elaboration. Since members of dysfunc-
tional groups are likely to be dissatisfied and to de-identify, we would expect
increased absenteeism and turnover (Pelled & Xin, 1999), each of which
have significant negative impacts on group functioning (Mitchell & Lee,
2001). In fact, the desire to avoid a negative member may even explain
additional variance in turnover that would not surface in traditional pre-
dictors like job satisfaction. For example, Mitchell and Lee (2001) note that
events like fights with a coworker may act as a ‘‘shock’’ that precipitates
leaving. Moreover, since the best employees have greater job mobility, they
are often the most likely to leave (Mitchell & Lee, 2001). As the best group
members jump ship, one can imagine a downward spiral in group perform-
ance, unfolding over time.
DISCUSSION
Over the last half century, a clearer understanding has emerged about the
power of collectives to reconstruct the goals, behaviors, and perceptions of
the individual to serve the needs of the group. However, it is often over-
looked that people conform and converge largely because they want to; they
want to belong and have clear expectations about normatively appropriate
behavior (Baumeister & Leary, 1995;Salancik & Pfeffer, 1978;Sherif, 1935).
Sometimes individuals behave in ways that do not benefit the group; some-
times individuals are negative, refuse to contribute effort or break important
WILL FELPS ET AL.206
group norms. This behavior presents a challenge at both practical and the-
oretical levels. Practically, chronic expressions of harmful behaviors allow
these people to become a figurative thorn in the groups’ side – clearly a
distraction and possibly a ‘‘destroyer’’ of the group itself (Wetlaufer, 1994).
Theoretically, these negative behaviors threaten our standard assumptions
about groups as homogeneous structures capable of cohesive action (e.g.
Hackman, 1976). And yet, despite the importance of the topic, the field has
yet to find the theoretical traction that would allow for a complete and
coherent understanding of the key issues implicated by these negative group
members.
Our analysis and review attempts to fill that gap. We present an unfolding
model that describes the prototypical process by which one individual be-
having badly might have a profoundly negative impact on the group. We
suggest that the three most salient and important behaviors of a negative
member are the withholding of effort, the demonstration of negative affect,
and the violation of important interpersonal norms. At the beginning of this
process, team members will react by trying to change this negative behavior.
If that fails, the attribution becomes that the person’s behavior is stable and
intractable. Next, members will look to reject the person. But when this is
not possible due to social constraints, more defensive psychological reac-
tions and behaviors are likely to occur. Defensiveness is an especially intense
experience due to two factors – the aversiveness of not having the control
over the environment (i.e. low power), and due to the psychological prin-
ciple that bad experiences are hard to ignore, require attention and sense-
making, and consume large amounts of time and energy (i.e. bad is stronger
than good). The direct reactions to this persistent and unchangeable neg-
ative member are the feeling of inequity when confronted with someone
withholding effort, the spreading of negative affect to other members
through contagion, and the loss of confidence and trust in an interpersonal
deviant. These negative states lead to defensive behaviors.
Defensiveness is associated with dysfunctional behaviors such as explo-
sions, revenge, mood maintenance, distraction, denial, and withdrawal.
These reactions are especially likely to occur when the negative behaviors of
the negative member are intense, when the group is interdependent or ex-
periences bad outcomes, and when group members lack the coping skills to
deal with the situation. Moving forward in this unfolding process, it is
through additive, spillover and sensemaking mechanisms that these be-
haviors come to influence group psychosocial constructs such as group
mood, group potency, and psychological safety. As a result, group activities
such as motivated effort, cooperation, coordination, creativity, learning,
How, When, and Why Bad Apples Spoil the Barrel 207
and helpful conflicts are decreased and diminished, eventually resulting in
poor group performance, lower well-being, and possibly team collapse.
It is important to note, however, that the negative member phenomenon
does not explain every instance of group dysfunction. Other factors such as
lack of organizational support, work-family issues, inadequate member
competencies, or unclear directions provide a host of alternative causes. In
other words, there is reason to be cautious in applying a bad apple label to a
particular member when confronted with a dysfunctional group. The fun-
damental attribution error (Ross, Amabile, & Steinmetz, 1977) and the
sinister attribution error (Kramer & Wei, 1999) both argue that people have
a penchant for pinning ambiguous problems on an individual group mem-
ber, particularly those that are disliked (Naquin & Tynan, 2003). By doing
so, groups might incorrectly label someone a bad apple and blame them for
negative outcomes. Moreover, a group may succumb to the cognitive ‘‘per-
formance-cue’’ bias, where outcome success unduly influences judgment and
recollection of the event (Staw, 1975). For example, if a group’s project is
unsatisfactory to members, they are likely to look backwards and judge
ambiguous or marginal behavior as dysfunctional. Moreover, cognitive
psychology research finds that when someone is in a negative frame of mind,
negative behaviors will be more easily and clearly recalled (Meyer et al.,
1990). Finally, in these same situations, there is a motivational bias to blame
someone for bad outcomes. In order to protect the image of the group and
the member’s self-esteem, the least proto-typical member is often used as a
scapegoat for what was really a collective failure (Eagle & Newton, 1981;
Marques, Abrams, & Serodio, 2001).
This presents a troubling methodological conundrum – people who are
‘‘positive deviants’’ or ‘‘devil’s advocates’’ will likely be resented for not
conforming, and thus will be scapegoated and derided, particularly when
negative outcomes have recently occurred. That is, dissent will likely lead to
a negative halo which may increase reportage of the person as expressing bad
apple behaviors of withholding effort, negative affectivity, and interpersonal
deviance. This would seem to present a threat to the validity of survey
measures of the effects of bad apples. So, how is a researcher to know if bad
apples caused negative outcomes or if negative outcomes caused someone to
be labeled a bad apple? One admittedly imperfect resolution would be to
assess factors we already know to be associated with scapegoating – such as
opinion deviance and recent negative feedback – and show that bad apple
behaviors explain incremental variance. Another approach is to have a con-
federate display bad apple behaviors in a laboratory context and to show
asymmetric effects in a context where the performance-cue bias in not
WILL FELPS ET AL.208
operable. We should also add that while opinion deviance may lead to some
bad apple labeling, it is unlikely that such behavior will have the same
extreme effects. First, opinion deviance may in fact lead to positive outcomes
(Nemeth & Staw, 1989). Second, it is less likely to be taken personally and
result in the same negativity caused by bad apple actions. But clearly, the
relative effect of opinion deviants and bad apples is an issue needing more
research.
Our initial examination of the frequency of spoiled barrels suggested that
while negative members who persist over time and eventually produce dys-
functional groups are probably not ubiquitous,
1
their effects are substantial.
Teams may identify negative members and their destructive behaviors – but
organizational constraints may limit the group’s ability to remedy the sit-
uation. We have suggested that the negativity bias and various processes of
social interaction operate to make the negative member behaviors dispro-
portionately recognized, informative and influential.
But what explains why theorists have overlooked this fundamental dy-
namic about responses to negative individuals? One reason seems to be that
scholars have considered it ‘‘beyond the scope’’ of their own works. Mitchell’s
research looks at leader’s responses to poor performing workers, and con-
sequently did not need to contend with situations of low empowerment
(Mitchell et al., 1981;Mitchell & O’Reilly, 1983;Mitchell & Wood, 1980). In
addition, that research focused on individuals, not teams. Lepine and Van
Dyne (2001) are more overt, explicitly assuming that ‘‘the peer who notices
the low-performing coworker is competent and capable yis committed to
the group and the group’s goals yand that situational factors do not overly
constrain peer responses’’ (p. 69). In short, they assume away the problem
that we are interested in – e.g. when ‘‘bad apples might spoil the barrel’’. We
relax those assumptions, and propose that there are hosts of situations when
teammates are not powerful, competent, capable, committed, or uncon-
strained – in short, situations where teammates are unempowered.
A second reason is that most researchers have only examined parts of our
overall picture and have captured just a small portion of what unfolds over
time. The typical study may look at only two or three variables such as how
negative affect can spread through a group (Barsade, 2002) or how a co-
worker who withholds effort causes other team members to have feelings of
inequity (Jackson & Harkins, 1985). In addition, some authors focus on
immediate individual reactions (the front end of our analyses) like moti-
vational and isolation attempts by coworkers (Lepine & Van Dyne, 2001)
while others focus on the relationships between group psychosocial traits
like low-efficacy and outcomes like group motivation or performance
How, When, and Why Bad Apples Spoil the Barrel 209
(Gully et al., 2002); relationships that are the last step in our analysis. Still
others look at how personality variables (e.g. low conscientiousness or low
agreeableness) affect the very distal dependant variable of team performance
(Barrick et al., 1998;Haythorn, 1953), but confess ignorance when it comes
to explaining why negative individuals have such a large asymmetric effect
on the group.
In looking over the totality of our presentation we know that we have
introduced a number of ‘‘sets’’ of states and behaviors at the individual and
group level. Some things are included, some excluded. We have tried to be
precise about what is in or out, partly through our definition of what con-
stitutes a negative member (e.g. withholding effort, negative affectivity, and
interpersonal deviance). These three sets of behavior drive much of what
follows in terms of states and actions. However, it is also important to
recognize that our guide for inclusion or exclusion was the research liter-
ature itself. We focused on phenomena that people have written about and
empirically researched. Obviously, some things were omitted due to these
judgment calls but we are fairly confident that we have not overlooked any
major components for our review.
FUTURE RESEARCH
We have presented a model that captures how the effects of the behaviors of
a negative group member unfold over time and across conceptual levels.
While many of the pairwise relationships that adjoin neighboring stages of
our analysis (see Fig. 1) are well documented; it is the distal and mediating
aspects of our approach that need more work. In addition, we have little
idea about the combinational properties of our states and behaviors at both
the individual and group level. Which states are most important or when are
they important? How do they combine: additively, multiplicatively? Are
there thresholds which must be surpassed for effects to occur and if so what
are they? In addition, we present our analysis in a lock step fashion over
time. In reality both individual and group psychological actions and reac-
tions may occur simultaneously and interact over time. Some stages may
take longer, others shorter. There is lots of research left to be done.
However, there are major problems with conducting such research. Be-
cause we are describing offensive behaviors and intense reactions, field re-
search would seem to be most appropriate. Also, the dynamic nature and
extended time frame point to field investigation. Extreme behaviors and
lengthy periods of interaction are hard to ca