Professional Documents
Culture Documents
Order in The Court: How Firm Status and Reputation Shape The Outcomes of Employment Discrimination Suits
Order in The Court: How Firm Status and Reputation Shape The Outcomes of Employment Discrimination Suits
2017
ASRXXX10.1177/0003122417747289American Sociological ReviewMcDonnell and King
Abstract
This article explores the mechanisms by which corporate prestige produces distorted legal
outcomes. Drawing on social psychological theories of status, we suggest that prestige
influences audience evaluations by shaping expectations, and that its effect will differ
depending on whether a firm’s blameworthiness has been firmly established. We empirically
analyze a unique database of more than 500 employment discrimination suits brought
between 1998 and 2008. We find that prestige is associated with a decreased likelihood of
being found liable (suggesting a halo effect in assessments of blameworthiness), but with
more severe punishments among organizations that are found liable (suggesting a halo
tax in administrations of punishment). Our analysis allows us to reconcile two ostensibly
contradictory bodies of work on how organizational prestige affects audience evaluations
by showing that prestige can be both a benefit and a liability, depending on whether an
organization’s blameworthiness has been firmly established.
Keywords
status, reputation, deviance, employment discrimination, lawsuits, punishment
2011; Kim and King 2014; Podolny 1994; be heuristically used to assess blameworthi-
Sine, Shane, and Di Gregorio 2003; Stuart, ness. Individual characteristics that prior work
Hoang, and Hybels 1999; Thye 2000; has linked to biased outcomes include race,
Waguespack and Sorenson 2011). age, gender, socioeconomic status, and per-
Of course, the advantages the corporate ceived character (Albonetti 1998; Nadler and
sector enjoys often come at the expense of McDonnell 2011; Steffensmeier and Demuth
employees and communities. Theoretically, 2006; Steffensmeier, Ulmer, and Kramer
legal action against corporations, whether 1998). Yet we know little about the character-
civil or criminal, operates as an external con- istics of organizational defendants that pro-
straint on corporate power by punishing busi- duce similarly biased outcomes in legal
nesses that engage in prohibited behaviors. In proceedings. The present article addresses this
this way, the law is meant to provide an issue by exploring whether and through what
autonomous means through which society mechanisms corporate prestige operates as a
holds business accountable for its actions. salient organizational characteristic that dis-
However, the legal system’s ability to pro- torts legal outcomes.
mote prosocial corporate behavior may be Our exploration of the relationship between
undermined by the Matthew effect, insofar as corporate prestige and legal outcomes provides
elite firms are able to garner more favorable an additional opportunity to reconcile two
outcomes. This possibility is supported by ostensibly contradictory findings about the
past work finding that elite corporations enjoy effects of organizational prestige in evaluative
considerable influence within the legal domain settings. The first, consistent with the Matthew
(Heinz and Laumann 1982; Shaffer 2009), and effect proposition, suggests that prestige leads
that litigation generally favors business inter- audiences to give organizations the benefit of
ests (Brace and Hall 2001; Dunworth and the doubt in evaluative situations (Davies et al.
Rogers 1996; Hadfield 2005; Kenworthy, 2003; Dowling 2002; Fombrun and van Riel
Macaulay, and Rogers 1996). In fact, in one 2003). For example, organizational status
of the earliest sociological treatments of the eases regulators’ concerns when a company
subject of corporate crime, Sutherland (1949) experiences product failures (Kim 2012), and
explicitly warned that status within the busi- investors react less negatively to earnings
ness sector distorted the enforcement of laws restatements when a firm belongs to a high-
applied to corporate offenders. status industry (Sharkey 2014).
Despite Sutherland’s proposition, the pre- Interestingly, a second body of research
cise mechanisms through which corporate supports a contrary proposition that prestige
prestige influences judicial outcomes remain can be a liability to firms in evaluative set-
unclear. Furthermore, research on corporate tings, insofar as it can translate into increased
litigation has to date “devot[ed] next to no scrutiny and harsher reactions from audience
attention to distinctions between litigant members. For example, reputable companies’
types” (Hadfield 2005:1277), which hampers actions are more consistently subject to scru-
our understanding of why certain corporations tiny from their peers and the media (Edelman
seem to receive more favorable treatment than 1992; Fombrun 1996; Pollock, Rindova, and
do others (Dunworth and Rogers 1996). The Maggitti 2008) as well as agitated stakehold-
lack of research in this area is surprising, con- ers (Bartley and Child 2014; Briscoe and Saf-
sidering the substantial body of work at the ford 2008; King 2008, 2011; King and
individual level devoted to exploring the char- McDonnell 2015; McDonnell and King
acteristics associated with sentencing biases. 2013). Kovács and Sharkey (2014) found that
This work provides robust evidence that eval- books that receive prestigious awards are
uators respond to the uncertainty and com- subsequently rated more negatively by users,
plexity of legal proceedings by searching for suggesting that status markers can produce
salient characteristics of defendants that can heightened expectations that lead to more
McDonnell and King 3
negative evaluations (for other examples, see argue that the positive expectations associated
Malmendier and Tate 2009; Rhee and Haun- with prestige ought to benefit the accused
schild 2006).1 In these cases, rather than company, creating a halo effect for the accused.
inducing more favorable evaluations as the Firms that are found blameworthy in the first
Matthew effect would predict, prestige stage of the lawsuit then move to the second
appears to amplify concerns about motives or stage: the punitive phase. Here, the company’s
behavior (Hahl and Zuckerman 2014). blameworthiness has been established with
We seek to reconcile these opposing sets certainty and the jury is asked to determine an
of findings by exploring the boundary condi- appropriate punishment. Under these condi-
tions under which an organization’s prestige tions, the positive expectations associated with
might shift from a benefit to a liability in the prestige have been clearly violated, which we
litigation process. Specifically, we propose expect will lead to a halo tax in the form of
that the influence of prestige on legal judg- harsher punishments.
ments depends on whether a firm’s blame- We test our hypotheses using a unique
worthiness for a charged transgression has archival dataset of jury verdicts from employ-
been firmly established. Drawing on social ment discrimination claims filed against a
psychological theories of status (Berger et al. sample of large U.S. companies between 1998
1977; Correll, Benard, and Paik 2007; Ridge- and 2008. We chose employment discrimina-
way 1991; Wagner and Berger 2002), we tion as our empirical context because it is a
claim that the mechanism through which clear example of an organizational transgres-
prestige influences stakeholder evaluations is sion, as indicated by highly institutionalized
elevated expectations (Kim and King 2014). efforts to suppress it, including anti-discrimi-
When an organization is initially accused of nation laws and ubiquitous sensitivity training
deviant behavior but its blameworthiness has programs (Nelson, Berrey, and Nielsen 2008).
not yet been firmly established, a positive Legal research documents that employment
reputation or high status will lead evaluators discrimination lawsuits are especially difficult
to expect that it behaved appropriately. But to win, partly due to biases held by judges and
once an attribution of guilt or blameworthi- juries (Selmi 2000), which emphasizes the
ness has been established, a positive reputa- uncertainty of these proceedings.
tion and high status become a liability, as Ultimately, our results provide evidence for
evaluators react harshly to violation of their both the halo effect and the halo tax. In our
expectations. Evaluators are more likely to model predicting a jury’s initial assessment of
feel betrayed when an organization they trust liability, we find that a firm’s status is nega-
and admire has deviated, which can elicit an tively associated with the likelihood of being
especially punitive response: what we call a deemed blameworthy for employment dis-
halo tax. Accordingly, once an attribution of crimination. However, in the second stage of
blameworthiness has been substantiated, we litigation, after a company’s blameworthiness
expect evaluators to punish a more prestig- has been established, we find that a firm’s
ious organization more harshly than its peers. status becomes a liability, especially when it is
Importantly, in the context of civil litiga- coupled with a good reputation in the domain
tion, our proposed boundary condition sug- in which the transgression occurred. We argue
gests that an organization’s prestige is likely to that these observed effects can be explained
affect legal outcomes differently at different by the benefit (in the former case) and the
stages of a lawsuit. In the first phase—the burden (in the latter) of positive expectations.
trial—the jury is asked to determine culpabil-
ity. During this phase, it is uncertain whether
the company’s actions merit blame. The Background And Theory
aggrieved party will provide evidence to show Development
that the company’s behavior was wrongful or Before delving into an analysis of how organi-
negligent, and the company will offer evidence zational prestige affects evaluative outcomes,
of its innocence. Under these conditions, we it is necessary to first clarify what we
4 American Sociological Review 00(0)
conceptualize as complementary components that actors benefit from having both, espe-
of prestige: status and reputation (Bitektine cially when they are aligned (Kim and King
2011; Sorenson 2014). Although they are 2014; Stern, Dukerich, and Zajac 2014). Kim
sometimes used interchangeably (Lang and and King (2014) argue that the evaluative
Lang 1988; Lange, Lee, and Dai 2011; Love benefits that come from status and reputation
and Kraatz 2009; McDonnell and King 2013; stem from the same underlying mechanisms:
Rindova, Williamson, and Petkova 2010; Rin- behavioral expectations. Status characteris-
dova et al. 2005), status and reputation are tics theory and expectation states theory pro-
different forms of prestige. Status refers to an pose that certain qualities are ascribed with
actor’s “relative social standing” (Sorenson status and subsequently shape expectations
2014:63) and usually denotes an actor’s posi- associated with actors who have those quali-
tion in a hierarchy (Graffin et al. 2013; ties (Berger et al. 1977; Correll et al. 2007;
Podolny 2001) or social rank (Washington and Ridgeway 1991; Wagner and Berger 2002).
Zajac 2005). Although different sociological Similarly, an actor’s reputation in a particular
literatures vary in their definitions of status, a domain can shape expectations for future per-
common theme is an emphasis on a hierarchy formance in that domain (Raub and Weesie
that emerges within a group and allows actors 1990). As evidence of this, Kim and King
with certain status characteristics to obtain (2014) find that high-status baseball pitchers
greater power and influence (Berger et al. and pitchers who have reputations for throw-
1977; Correll and Ridgeway 2003). Among ing with great accuracy receive more favora-
organizations, status is often made visible in ble calls from umpires than do low-status
the form of rankings or patterns of deference pitchers or pitchers who have reputations for
that elevate one organization over another being wild and inaccurate.
(Espeland and Sauder 2007; Podolny 2001). Lawsuits in the United States provide a
Reputation, in contrast, refers to shared natural setting in which to test the effects of
perceptions of an actor’s unique and distin- reputation and status on audience evaluations.
guishing qualities (Fombrun and Shanley Key to our theory is the notion that these
1990; King and Whetten 2008) and frequently prestige markers ought to matter differently
invokes an audience’s perceptions of a firm’s when an actor’s blameworthiness is uncertain
past actions or performance in a particular versus when it has been clearly established.
domain (Fombrun 1996; Roberts and Dowling The U.S. legal system bifurcates a lawsuit
2002; Sorenson 2014). Reputation is multidi- into two discrete phases that vary in terms of
mensional and can be rooted in a variety of established blameworthiness. In the next sec-
different performance criteria (Rao 1994). The tions, we discuss how the enhanced expecta-
same organization can have a positive reputa- tions triggered by organizational prestige are
tion in one domain, such as product quality, likely to influence audience evaluations
and yet have a weak or negative reputation in within these two separate phases.
another domain, such as treatment of employ-
ees.2 In evaluative settings such as legal pro-
The Trial Phase: How Prestige
ceedings, we expect a firm’s reputation in the
Influences Evaluations Made Under
precise domain in which it is being evaluated
Conditions Where Blameworthiness
will have the strongest effect on outcomes.
Although status and reputation differ con-
Has Not Been Established
ceptually, they are both thought to confer The positive expectations associated with
advantages. The benefits that come from prestige ought to be especially beneficial
being associated with a high-status position when actors face controversial situations for
are sometimes treated as equal to the positive which their level of responsibility is unclear
outcomes associated with having a good rep- (Lange et al. 2011; McDonnell and King
utation. In fact, recent research demonstrates 2013; Rindova, Petkova, and Kotha 2007).
McDonnell and King 5
For example, some past research finds that such that the actions of high-status and repu-
audiences are less likely to assign a reputable table companies are likely to be interpreted in
organization responsibility for a crisis, such a more favorable light. For example, one
as when its products are deemed harmful or experimental study found that participants
dangerous (Fombrun 1996; Klein and Dawar evaluating a price increase were more inclined
2004). Other work finds that key external to infer that the action stemmed from a posi-
audiences, like the media, are less critical of tive and fair motive when the company was
prestigious organizations implicated in a cri- reputable, but they tended to infer a negative
sis (Balzer and Sulsky 1992; Coombs 1999; and unfair motive when the company was less
Ulmer 2001). Critical stakeholders like politi- reputable (Campbell 1999). Applied to the
cians are also less likely to cut ties with high- context of employment discrimination trials,
status organizations accused of bad behavior this research suggests that juries tasked with
(McDonnell and Werner 2016). evaluating a defendant firm’s culpability for
Reputation and status inform the holistic employment discrimination are less likely to
assessments evaluators make of firms by deem a prestigious organization’s actions as
shaping expectations (O’Donnell and Schultz blameworthy. Thus, we expect the following:
2005). Specifically, people tend to infer less
blame and responsibility for actors’ bad Hypothesis 1a: In the first stage of a lawsuit
actions when they have pre-established posi- where a company is charged with a trans-
tive expectations about the actor. This ten- gression, the defendant firm’s status will be
negatively associated with its likelihood of
dency stems from various social psychological
being found liable.
theories, including status characteristics and Hypothesis 1b: In the first stage of a lawsuit
expectation states theories (Berger et al. 1977; where a company is charged with a trans-
Correll and Ridgeway 2003), as well as theo- gression, the defendant firm’s reputation
ries of motivated reasoning (Traut-Mattausch will be negatively associated with its likeli-
et al. 2004). The former theories propose that hood of being found liable.
an actor’s position in a status hierarchy pro-
duces performance expectations, and these We also contend that the positive effects of
expectations lead to reproduction of the hier- status and reputation on jury evaluations will
archy. The concept of motivated reasoning, in be greater when the two forms of prestige are
contrast, emphasizes that people tend to inter- aligned. Past research indicates that alignment
pret evidence presented to them in a way that between status and reputation magnifies the
allows them to confirm their priors (Ditto and potential for an actor to receive advantages
Lopez 1992; Kunda 1987). In explorations of (e.g., Stern et al. 2014). One reason for this is
motivated reasoning within the context of that, as Kim and King (2014) found, status
jury decisions, prior work shows that decision- and reputational alignment sharpens behavio-
makers are likely to interpret the evidence ral expectations, magnifying innate biases.
presented in a way that confirms their preex- Another reason is that whereas status is a dif-
isting beliefs derived from racial stereotypes fuse signal of prestige and quality, reputation
(Rachlinski et al. 2009) or initial impressions is domain-specific and informs evaluators’
of a defendant’s moral character (Nadler and choices about when and how status should
McDonnell 2011). influence their judgments. Accordingly, align-
When the quality or appropriateness of ment between status and reputation reinforces
performance is difficult to evaluate, observers expectations about a particular kind of behav-
will fall back on previously held beliefs and ior, which may be directly relevant to evalua-
expectations about an actor. In turn, these tions of blameworthiness for a particular
expectations color observers’ judgments and transgression. Thus, if a high-status company
inferences when they are faced with evidence is also known for being a good employer, the
of a potential organizational transgression, jury in an employment discrimination lawsuit
6 American Sociological Review 00(0)
may believe that the company’s high status is a phenomenon they call “betrayal aversion,”
partly a result of being such an outstanding Koehler and Gershoff (2003) found that peo-
employer, giving the jury more reason to expect ple recommend more severe punishments
the company treats its employees responsibly. when a person or entity they trust behaves in
a way that violates their trust. Relatedly,
Hypothesis 2: When charged with a transgression, Rosoff (1989) discovered that high-status
a firm’s status will interact with its domain- physicians found guilty of a serious crime
specific reputation to negatively affect the were evaluated more negatively than low-
likelihood that it will be found liable.
status physicians who committed the same
crime, and Skolnick and Shaw (1994) found
The Punitive Phase: How Prestige that high-status criminals were judged more
Influences Evaluations Made Under harshly when their crime was deemed to be
Conditions When Blameworthiness related to their profession.
Has Been Established Extended to the organizational setting,
betrayal aversion suggests that the transgres-
The same positive expectations associated with sions of high-status or reputable organiza-
high status and good reputations can, however, tions are especially likely to provoke feelings
have detrimental effects once an organization’s of outrage from their audiences, as audience
blameworthiness for a transgression has been members may feel duped, jilted, or betrayed
established. Prestigious organizations are held given the strong, positive expectations they
to a higher standard precisely because more is had for those organizations. This should, in
expected of them. When it is proven that they turn, provoke an especially punitive response.
have engaged in actions that do not accord with
their proffered image, they are more likely to be Hypothesis 3a: Once an organization has been
pegged a hypocrite, provoking a more punitive found blameworthy for a transgression,
response from evaluators (Harrison, Ashforth, higher status will be associated with harsher
punishments.
and Corley 2009).
Hypothesis 3b: Once an organization has been
The scope condition of this halo tax is that found blameworthy for a transgression,
an organization’s blameworthiness has previ- more positive reputations in the domain of
ously been established, such that observers the transgression will be associated with
have no choice but to revisit and question their harsher punishments.
prior beliefs and expectations. Evaluators who
previously believed that an organization main- We also expect that the alignment of status
tained a high standard are motivated to punish and reputation will magnify the harshness of
the organization for failing to live up to that punishment. Inasmuch as alignment sharpens
very standard. Illustrative of this, King and expectations associated with status in a
McDonnell (2015) found that organizations domain-specific area, juries will see employee
with better reputational standing are more discrimination by high-status companies that
likely to be targeted by activist groups when have a reputation for being a good employer
they fail to live up to their reputation. Simi- as an even greater betrayal.
larly, Rhee and Haunschild (2006) found that
firms with higher reputations for product qual- Hypothesis 4: Among firms found blameworthy
ity suffer greater market penalties in response for a transgression, a firm’s status will in-
to a product recall. teract with its domain-specific reputation to
Psychological research on the inclination positively affect the harshness of punishment.
to punish suggests there is a strong relation-
ship between the anger elicited by an offense Research Setting
and the extent to which people are inclined to
punish the offender (Bies 1987; Kahneman, Employment discrimination lawsuits in the
Schkade, and Sunstein 1998). Demonstrating United States result when an employee or
McDonnell and King 7
prior employee of a company alleges that the different cases of deviance. For this reason,
company treated them adversely (e.g., by fir- punitive damages have been used in other
ing or refusing to promote them) because of research exploring the mechanisms that affect
their membership in a protected class. Feder- the severity of punitive responses (e.g., Nor-
ally recognized protected classes include dgren and McDonnell 2011).
those based on gender, age, disability, reli- The structure of employment discrimina-
gion, race, or nationality. This context pres- tion suits makes them especially useful for
ents an ideal setting for exploring our theory. testing our predictions about the disparate
The United States has formally prohibited role of prestige in attributions of blamewor-
discrimination since the passage of Title VII thiness (when there is uncertainty about the
in the 1964 Civil Rights Act (Sutton et al. firm’s culpability) versus allotments of pun-
1994), which remains the primary federal ishment (when a firm’s deviance has been
cause of action that gives victims of discrimi- established). Employment discrimination
nation a right to sue their employer. However, lawsuits are bifurcated into two discrete
discrimination continues to be a pervasive phases. The first phase, the trial, determines
problem in the U.S. workplace (Edelman whether the company is liable, or blamewor-
2016); in recent years, the U.S. Equal Employ- thy, for a proscribed discriminatory action.
ment Opportunity Commission has received Aggrieved employees will provide evidence
nearly 100,000 discrimination charges per to show that a company wrongfully discrimi-
annum (USEEOC 2013). Employment dis- nated against them, and companies will offer
crimination is therefore a clear example of evidence of their innocence. During this first
misconduct that is nevertheless common stage, a company’s blameworthiness is uncer-
enough to enable empirical investigation. tain. Some charges of employment discrimi-
These cases also represent an ideal oppor- nation are supported by overwhelming
tunity to study how prestige affects organiza- evidence, such that blameworthiness is clearly
tional punishment because of the manner in apparent before a formal declaration by a
which punishment occurs. Organizations jury, but such clear-cut cases are unlikely to
found liable for discrimination have to pay a be in our sample of cases that go to trial and
monetary sum to the employee, referred to as are ultimately decided by a jury. We inter-
the employee’s “damages award.” One por- viewed a prominent corporate defense attor-
tion of that sum, the “compensatory damages ney who explained this as follows:
award,” is meant to compensate injured par-
ties for the direct financial or emotional harms There are very high expenses associated
they suffered as a consequence of the dis- with litigating a case through trial, so it is not
criminatory action. This aspect of damages is in the interests of either party to litigate a
legally limited to the wronged individual’s trial where the outcome of the case is clear.
real or projected losses established in the evi- If there is clear and compelling evidence of
dence (e.g., the value of their lost salary), discrimination, the economic incentives of
giving the jury little discretion over the the defendant are to settle the case, as
awarded amount. However, the jury has more opposed to waiting for a jury verdict. The
discretion to determine a separate portion of cost of a jury verdict is typically greater than
the plaintiff’s damages, the “punitive dam- a settlement . . . and precedential conse-
ages award,” which reflects a sum that the quences of having an adverse judgment that
jury can make the corporation pay for purely could be applied in future cases strongly
punitive reasons, meant to demonstrate soci- push companies to not litigate the case if the
ety’s disapproval of its deviant behavior. The evidence is clear. So, really, it is only the
punitive damages award, which can vary hard cases that go to trial.
extensively from case to case, provides an
easily observable and quantifiable proxy for As this quote illustrates, the cases typically
the severity of audience reactions across litigated before a jury are those for which the
8 American Sociological Review 00(0)
outcome is difficult to predict, such that the rankings between 1998 and 2008. The Fortune
first stage of a trial is properly characterized rankings are regularly used in organizational
as having a considerable degree of uncer- scholarship as an indicator of a company’s
tainty about whether the defendant compa- prominence in the corporate field (e.g., Fomb-
ny’s actions were in fact deviant. run and Shanley 1990; King 2008; McDonnell
Firms found liable for discrimination in the and King 2013; Roberts and Dowling 2002;
first stage of a trial then move on to the second, Staw and Epstein 2000). To create the sample
punitive, stage: the damages phase. At this each year, Fortune surveyors begin with the
stage, the court holds a separate proceeding to Fortune 1,000, from which they select the 10
ascertain the appropriate amount to make the largest firms (based on asset size) from each
company pay the aggrieved employee. Again, industry. They then collect survey data about
both the employee and the company will pre- these firms from industry executives, directors,
sent evidence to encourage the jury to award and security analysts. Fortune aggregates the
an amount in their favor. But here, the compa- survey results for each company, then assigns
ny’s blameworthiness for a deviant action is no each company a reputation score that ranges
longer uncertain because the company has from 0 to 10. Fortune ultimately uses this data
already been found liable. The two stages of a to create and publish a list of the 50 Most
lawsuit thus represent two evaluative settings Admired Companies, but our sample includes
that vary in terms of whether blameworthiness every company surveyed (including those that
has been established, providing an ideal con- were not ultimately selected for the final pub-
text to test our hypotheses. lished list), providing a sample of organiza-
Past research on corporations and litiga- tions with a wide variation of scores. By
tion points to numerous advantages that com- limiting our analysis to companies surveyed
panies have in the process. For one, the “size for the rankings, we are able to more accu-
and complex interrelationships of large cor- rately quantify and compare the relative status
porations make it difficult [for ordinary plain- of all companies in our analysis. However, this
tiffs] to compete with them on an equal basis sampling strategy does limit the scope of our
in investigations and in litigation” (Clinard findings to firms that are fairly large and visi-
and Yeager 1980:313). Moreover, because ble, which are likely to have the kind of well-
corporations are often “repeat players,” they established reputations and status orderings
accumulate more legal expertise and relation- that are perceptible to lay audience members.
ships with judges that improve their chances As past research shows, the Fortune ranking
of successfully navigating the litigation pro- has become a commensurable status hierarchy
cess when compared to “one shot” plaintiffs within the corporate realm (Bermiss, Zajac,
(Galanter 1974). We propose prestige as an and King 2014), and maintaining one’s posi-
additional source of power that corporations tion in the ranking is greatly valued by presti-
uniquely possess over relatively unknown, gious firms (McDonnell and King 2013).
individual plaintiffs, but we also account for Thus, the ranking exhibits similar qualities to
the relative influence of corporate resources other status rankings (see Espeland and Sauder
and past experience in our analyses. 2007).
Having established our sample, we searched
the Westlaw jury verdicts database for all
Methods employment discrimination cases brought
Sample Creation against sampled firms. This database is a com-
prehensive legal resource that includes a
To test our hypotheses, we built a dataset of all searchable archive with information on the
employment discrimination cases against any jury verdicts or outcomes of civil (noncrimi-
of the 826 unique companies that were surveyed nal) cases brought in all 50 states and the Dis-
for Fortune’s Most Admirable Companies trict of Columbia. Ultimately, we identified a
McDonnell and King 9
total of 922 employment discrimination cases For companies found liable of employ-
brought against companies included in the ment discrimination, we capture the extent of
Fortune rankings that were decided between punishment through the punitive damages
1998 and 2008. From these, we culled a subset award, a continuous count variable indicating
(n = 176) that settled out of court and never the amount (in dollars) that the company was
went to trial, as well as a subset (n = 112) that assessed as punishment for its discriminatory
were dismissed by the trial judge for failure to action. A total of 303 cases in our sample
provide a colorable case or for lack of adequate resulted in a verdict of liability and produced
evidence (i.e., dismissal or summary judg- a damages award. Federal guidelines do not
ment). We ran several robustness checks to allow punitive damages to be awarded for
ensure that culling of this subset does not intro- violations of the Age Discrimination in
duce selection bias for our ultimate analysis. Employment Act (ADEA), so we omit 24
Unpaired t-tests of means comparing the firms cases that rested solely on claims brought
in the top 50 percent of status and domain- under the ADEA. Data on punitive damages
specific reputation with the firms in the bottom were not reported in three cases, and data on
50 percent yielded no evidence of significant incorporated controls were missing for 38
differences in the incidence of settlement or firms, reducing the sample size to 238 cases
dismissal between these two groups. Supple- in models predicting punishment.
mental regressions of the likelihood of settling
out of court and of early dismissal also showed
Independent Variables
no indication of a company’s reputation or
status being significantly associated with either Our measure of organizational status is based
outcome. on the general score of favorability a firm
To focus on cases that were decided by received in the Fortune Most Admirable
comparable juries, we omit from our analysis Companies rankings. This Most Admired
a subset of 17 cases that were decided by a index is a proxy for overall audience assess-
judge (i.e., bench trials) or an arbitrator. Addi- ments, being founded, in part, on surveys
tionally, because class action suits, as opposed capturing industry leaders’ and analysts’ per-
to suits brought by individuals or small groups, ceptions Each organization included in the
are likely to present idiosyncratic incentives list is given a raw score that ranges from 0 to
against litigating due to the potentially extreme 10. Inherent in this process are comparisons
damages awards they invite when successful, that executives and analysts make between
we omit a further four cases that were peer firms. Surveyed individuals are asked to
expressly labeled as class action lawsuits.3 rate the 10 largest firms in their industry
Of the remaining 613 employment dis- along various dimensions, allowing Fortune
crimination cases in our sample that went to to construct a positional hierarchy of firms
jury, the company was found liable for dis- within and across industries (for details about
crimination in 303 cases and was deemed not the ranking process, see Bermiss et al. 2014).
blameworthy in 310. Data on incorporated Because the Fortune rankings are collected
controls were missing for 94 of these firms, from surveys administered in the year prior to
reducing our final sample size to 519 cases. that in which they are reported, the rankings
have a natural one-year lag. We therefore base
our status variable on each company’s score
Dependent Variable
in the same year that the verdict in a given
In our analysis predicting the outcome of the case was decided.
first stage of employment discrimination suits, We chose to use the Fortune Most Admired
we use a binary dependent variable coded “1” rankings to capture status because they (1) are
if the jury found the company liable and “0” if positional, (2) convey relative standing of
the company was found not to have engaged general favorability to an organization, as
in employment discrimination. opposed to the domain-specific judgment of
10 American Sociological Review 00(0)
quality associated with reputation (Rhee and their assessment of status (Brown and Perry
Haunschild 2006; Sorenson 2014), and (3) are 1994). Substantiating this concern, the raw
available for a large sample of firms across reputation scores in our sample show strong
the full panel of our archival analysis. Impor- correlations with firm size and performance.
tantly, we are not arguing that jurors are We address this issue by implementing Brown
aware of the Fortune rankings per se, but and Perry’s (1994) method for removing the
rather that the Fortune rankings provide a financial performance halo from the Fortune
colorable and quantifiable proxy for percep- ranking, which involves a two-stage empiri-
tions of organizational status in the general cal strategy. First, we ran a regression in
population from which jurors are drawn (Ber- which we predicted reputation as a function
miss et al. 2014). Although the Fortune rank- of four indicators of size and performance:
ings are constructed from surveys of industry liquid assets, Tobin’s Q, total employees, and
insiders, prior work provides evidence that logged assets. The models show that each of
the rankings reliably align with perceptions these variables significantly predict Fortune
held by external audience members. For rankings. Next, we computed the residuals
example, the Fortune rankings predict the from this model, which substantively corre-
willingness of a broad set of actors to associ- spond to that portion of the Fortune ranking
ate with a given firm, including audience that cannot be explained by size and recent
members within its corporate community, like performance. We then used these residuals as
business partners (Sullivan, Haunschild, and the measure of status in our models, rather
Page 2007), as well those outside its corpo- than the raw Fortune ranking.4
rate community, like politicians (McDonnell In contrast to status, reputation scholars
and Werner 2016). Other work demonstrates argue that most organizations’ reputations rest
that a firm’s Fortune ranking predicts how it on a set of specific practices or characteristics
will respond to social activists and manage its that distinguish them from competitors in a
public image in the media (e.g., King 2008; particular domain (e.g., Bitektine 2011; King
McDonnell and King 2013). These findings and Whetten 2008; Lange et al. 2011). We
suggest that the Fortune rankings accord with examine the effect of domain-specific reputa-
a firm’s own beliefs about the general pub- tion by utilizing the annual ratings of firms’
lic’s assessment of its status. We acknowl- social reputations provided by Kinder, Lyden-
edge that the Fortune ranking is an imperfect berg, Domini & Co (KLD) STATS (Statistical
proxy for specific juries’ assessments of cor- Tool for Analysis of Trends). The KLD data
porate status, and it does introduce potential draw from diverse sources, including media
measurement error. But given that prior reports and firm disclosures, to construct yearly
research demonstrates that the ranking has numerical ratings of firms’ strengths and weak-
validity among activists and other outsiders to nesses in seven social domains: community,
the corporate community and maps on to their corporate governance, diversity, employee rela-
strategic interactions with companies (e.g., tions, environment, human rights, and product.
King and McDonnell 2015), we deem it a Employment discrimination naturally touches
defensible assumption that the rankings align on two of the social domains covered in the
with jury members’ perceptions of status. KLD STATS: diversity and employee relations.
One concern about the Fortune rankings as To create our proxy for domain-specific reputa-
a proxy for general audience assessments of tion, we computed a net composite score of the
status is that the measure has been historically number of strengths minus the number of con-
dominated by recent performance (Bermiss cerns in these two areas.5
et al. 2014). This could introduce measure- Like our measure for status, KLD data are
ment error in our setting insofar as industry an indirect proxy for an individual jury’s per-
insiders are more likely than the general pub- ceptions of domain-specific reputation. How-
lic to prioritize performance indicators in ever, KLD data are a common proxy for
McDonnell and King 11
jury compositions would vary systematically the effect of prior experience, we include a
with respect to firm prestige. However, because variable capturing the total number of times a
legal and cultural responses to discrimination firm was charged in a discrimination lawsuit
may differ by region, we do include fixed in the prior three years.9
effects for the circuit in which each case was To account for each firm’s general size and
brought. We account for potentially meaning- performance, each of which have been shown
ful differences in venue in two additional to correlate with the commission of pro-
ways. First, federal and state venues can differ scribed conduct (Clinard and Yeager 1980),
in their discrimination statutes, rules of discov- we include controls for logged assets and
ery, and standards for summary judgment, Tobin’s Q. In the organizations literature,
each of which could affect case outcomes. We Tobin’s Q is a common proxy for market per-
thus include a binary control capturing whether formance that is operationalized as the market
the case was brought in federal, as opposed to value of assets divided by the replacement
state, court. Second, to address the possibility cost of assets. This variable is superior in this
that companies may enjoy a “home field setting to other common metrics for perfor-
advantage” when they face a jury in their home mance (e.g., ROA), because it is less suscep-
state, we include a binary variable, home state tible to managerial manipulation, which is
case, that is coded “1” when the case is in the common during periods of social scrutiny
state where the defendant company is head- (McDonnell et al. 2015; Watts and Zimmer-
quartered, and “0” otherwise. man 1986). To account for variance in the
In addition to these case-level controls, we available resources that firms can devote to
account for several firm-level characteristics their defense, we include a control for the
that may be associated with legal outcomes. total amount of a firm’s cash stores. Given
One variable likely to play a central role in that firms are naturally more likely to be
attributions of blameworthiness and punish- accused of employment discrimination when
ment is the defendant’s prior experience with they employ more people, we also control for
similar cases. Interestingly, the organizational each firm’s number of employees.
and criminological literatures suggest two We account for potential temporal factors
competing mechanisms through which prior that might affect jury outcomes through fixed
experience could influence case outcomes. effects for the year in which each verdict was
The organizations literature predicts that prior decided. Finally, to address the possibility
experience should benefit firms through a that jury members may have different expec-
“repeat player” effect, whereas the criminol- tations and responses to discrimination com-
ogy literature suggests that prior experience mitted by companies in different industries,
could provoke a disadvantageous “repeat we include fixed effects for each company’s
offender” effect. The repeat player effect, first major SIC industry division (i.e., two-digit
posited by Galanter (1974), suggests that SIC code), as reported in Compustat. Table 1
firms acquire expertise and develop critical provides summary statistics and correlations
relationships over the course of repeated liti- of these variables.
gation in an area, improving their ability to
successfully navigate the complex litigation
Model Specification
process. The repeat offender effect, supported
by robust evidence in the individual criminal We test Hypotheses 1 and 2 using a probit
context, suggests that a lengthier record of model that predicts the likelihood a firm
deviance can disadvantage defendants in the charged with employment discrimination is
legal process by indicating a pattern of devi- found liable. The probit regression is appro-
ance that makes the defendant appear more priate for models with a binary dependent
blameworthy and less capable of rehabilita- variable. Every case that went to jury was
tion (Steffensmeier et al. 1993). To control for included in this first stage.
Table 1. Summary Statistics and Correlation Matrix
Mean SD Min. Max. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
13
14 American Sociological Review 00(0)
The models testing Hypotheses 3 and 4 between a firm’s status and its likelihood of
include only firms that have been deemed being found liable for employment discrimi-
blameworthy for employment discrimination, nation. Post-estimation margins analysis of
as only firms found liable will go on to be Model 2 indicates that firms with a status
assigned a punishment by the court. Our score one standard deviation above the mean
dependent variable, punitive damages, is a are around 14 percent less likely to be found
highly overdispersed count variable, so we liable in a discrimination trial than are firms
use a generalized linear model with negative with a status score one standard deviation
binomial errors and a log-link function, using below the mean.
the glm command in Stata. One common way Model 3 of Table 2 introduces the proxy
to correct for potential bias in a skewed count for domain-specific reputation to test Hypoth-
dependent variable is by log-transforming it esis 1b. This hypothesis is not supported, as
and then running an OLS regression. The the model provides no evidence that a com-
alternative we use, a log-link function in a pany’s reputation, as indicated by its net KLD
generalized linear model, has the advantage score for employee relations and diversity, is
of returning predictions directly on the related to the likelihood of liability. Model 4
dependent variable’s original measured scale, of Table 2 tests Hypothesis 2 by including an
negating the need for back-transformation interaction of status and domain-specific rep-
and simplifying interpretation (Cox et al. utation. Again, we find no support for the
2008; McDonnell and Werner 2016). hypothesized effect. Taken as a whole, these
We expect that results within a state are results suggest that holistic, generalized sig-
more likely to be correlated than those nals of prestige, such as status, may carry
between states due to localized laws as well more weight for evaluations of corporate con-
as potential regional differences in attitudes duct in ambiguous situations than do more
toward discrimination. Additionally, our data applied indicators of prestige, like domain-
include multiple within-firm observations, specific reputations.
which are also likely correlated with one Table 3 shows models predicting the extent
another. Accordingly, we cluster standard of punishment through assessed punitive
errors at both the state and firm levels. All damages. These models yield several interest-
models were run using Stata 14.0. ing results for control variables that warrant
mentioning. First, across all models we find
that past involvement in discrimination suits
Results appears to broadly benefit a defendant firm,
Results from the models predicting the likeli- as the number of past discrimination suits a
hood that a company charged with employ- firm has faced is negatively related to the
ment discrimination was found liable are level of punitive damages a liable firm is
shown in Table 2. Model 1 includes only the assessed. This finding supports Galanter’s
control variables. The models provide mixed (1974) notion that firms are advantaged by
evidence that larger firms (in terms of number being repeat players in the legal system: past
of employees) are less likely to be found lia- experience confers enhanced capabilities for
ble than smaller firms. Models 1 and 2 pro- achieving more favorable outcomes in future
vide additional suggestive evidence that firms suits. This also highlights that punitive pro-
do enjoy a home court advantage, as firms are cesses are distinct for organizations and indi-
significantly less likely to be found liable viduals, given that the latter tend to be
when a case is brought in the state in which punished more harshly when they have prior
they are headquartered. charges on their record that mark them as
Model 2 introduces our measure of status “repeat offenders.” Because the offensive
to test Hypothesis 1a. In support of this actions at issue in discrimination cases are
hypothesis, the model demonstrates a statisti- normally undertaken by different deviant
cally significant and negative relationship individuals within a firm, juries might not
McDonnell and King 15
Table 2. Probit Regression Predicting the Likelihood of Being Found Liable in Cases of
Alleged Employment Discrimination, 1998 to 2008
Note: Robust standard errors, clustered by company and state, are in parentheses. Omitted
discrimination category is age discrimination.
*p < .05; **p < .01; ***p < .001 (two-tailed tests).
16 American Sociological Review 00(0)
Table 3. Generalized Linear Models Predicting Punitive Damages Levied Against Companies
Found Liable for Employment Discrimination, 1998 to 2008
Note: Robust standard errors, clustered by firm and state, are in parentheses. Omitted discrimination
category is age discrimination.
*p < .05; **p < .01; ***p < .001 (two-tailed tests).
McDonnell and King 17
perceive firm-level offenses as contributing Model 2 in Table 3 introduces the proxy for
to an aggregated “record” in the same way as status. In this second stage of litigation—once
they accumulate for individuals. This may be an organization’s culpability has been estab-
remiss, as some scholars argue that corporate lished—we find that organizational status
recidivism signals a firm-level problem, sug- becomes a liability, in support of Hypothesis
gesting “a corporate atmosphere favorable to 3a. Notably, the effect of status on punitive
unethical and illegal behavior” (Clinard and damages is net of the effects of firm size and
Yeager 1980:117). profitability. Post-estimation margins analysis
Results for the other control variables vary of Model 2 indicates that a firm with status one
markedly across the models and should be standard deviation below the mean would pay
interpreted with caution. Here we discuss only a predicted value of $183,143 in punitive dam-
the variables that show a significant associa- ages, whereas a firm with status one standard
tion with punishment in the final model deviation above the mean would pay $674,860.
(Model 4 in Table 3), which incorporates all Model 3 in Table 3 introduces the proxy
hypothesized effects. This model suggests that for domain-specific reputation. This model
cases alleging racial discrimination, as com- produces no evidence of reputation having a
pared to age discrimination, draw significantly significant independent effect. Hypothesis 3b
more punitive damages, whereas those alleg- is thus not supported. The interaction of status
ing disability discrimination yield signifi- and domain-specific reputation in Model 4 is
cantly smaller awards. Cases that produced statistically significant, suggesting initial sup-
more costly harm—indicated by greater com- port for Hypothesis 4. However, interaction
pensatory damages—draw significantly larger terms in a nonlinear model must be inter-
awards, supporting the notion that discrimina- preted with caution (Ai and Norton 2003;
tory actions that caused more harm are pun- Mood 2010), and post-estimation margins
ished more aggressively. The significant effect analysis of this model yielded no evidence
of the female plaintiff variable suggests that that the interaction produces a significant
juries may be more punitive when at least one effect at any particular point in the observed
of the victims of discrimination is female. distribution of our data. Given this mixed
This finding complements criminological evidence, we explore the interaction effect in
work that shows violent crimes with female more detail through a battery of additional
victims are sentenced more stringently (Curry models (see the online supplement). As a key
et al. 2004). Finally, the significant effect of robustness check, we replicate the interaction
Tobin’s Q suggests that better-performing effect in a linear model (an OLS regression
firms may be punished more harshly for trans- with a logged dependent variable), which
gressions. Predicted values from the final allows for a more straightforward interpreta-
model suggest that a firm with performance tion. In the linear model, the interaction term
one standard deviation below the mean would remains statistically significant and positive.
pay around $23,000 in punitive damages, This evidence provides clearer support for
whereas a firm with performance one standard Hypothesis 4, insofar as it demonstrates that
deviation above the mean would pay over 20 the interaction has a significant average effect
times more: $470,000. One possible explana- across the full distribution of status and repu-
tion for this is that outperformance might tation. However, we are unable to identify
carry a connotation of corruption or injustice exactly where in the distribution the effect is
when it is won during periods of objectionable driven. Future work is needed to more pre-
conduct. This resonates with the more general cisely illustrate the moderating role that repu-
claim that stakeholders respond negatively to tation (and the associated mechanism of
outperformance that appears to have been hypocrisy) plays in driving the halo tax.
achieved through illicit means (Harris and We ran a number of additional alternative
Bromiley 2007; Mishina et al. 2010). models to probe the robustness of our findings.
18 American Sociological Review 00(0)
First, we acknowledge the potential for selec- We discuss the theoretical implications of
tion bias to affect our results, as only the firms these findings in the next section.
found liable for employment discrimination
are selected into the second stage of a lawsuit
and assigned punishment. To mitigate con- Discussion And
cerns about selection bias, we replicated our Conclusions
models as a two-stage Heckman selection Many observers argue that corporations
model. In this estimation, the probit regression receive deferential treatment in society. One
predicting liability is used as a first-stage of the arenas in which they are advantaged is
model to estimate a selection effect coefficient the legal domain (Nielsen, Nelson, and Lan-
(referred to as the inverse Mills coefficient, or caster 2010). In the United States, legal inter-
λ) that is included as a control in the second- vention is a primary mechanism through
stage model predicting punishment. We use the which social proscriptions of employment
win ratio for similar suits in prior year varia- discrimination are enforced, and yet past
ble (described earlier) as a selection instru- research shows that legal means are often
ment. Supporting its efficacy as a selection quite ineffective at improving workplace con-
instrument, this variable is highly correlated ditions for the underrepresented (Edelman
with a firm’s likelihood of being found liable 2016; Hirsh 2009; Kalev and Dobbin 2006).
for employment discrimination, but it is not One key reason for this seems to be that pres-
significantly associated with punishment lev- tigious companies are less likely to be found
els or the underlying error term of the second- liable when discrimination charges are
stage model. Results for all key variables are brought against them. Interestingly, we also
substantively identical in these models. Fur- found mixed evidence that larger firms (in
thermore, the lambda coefficient in the second- terms of number of employees) and firms
stage model does not approach statistical headquartered in the state in which a suit is
significance, which suggests that selection is brought are less likely to be found liable.10
not a serious concern in this context. Our analysis confirms that the legal arena is
Our data also include a handful of signifi- not an even playing field. The most powerful
cant outliers vis-a-vis punitive damages, firms—larger and more prestigious ones—are
introducing concerns that these outliers may more likely to avoid sanctioning by courts
be unduly biasing the effects we observe. To when charged with discrimination. Firms are
rule this concern out, we replicated the mod- also less likely to be held liable for discrimi-
els in Table 3 with models winsorizing the nation when cases are brought in their home
dependent variable at the top and bottom 10 state, where their status and political power
percent of its distribution. Although the size are likely augmented.
of the coefficients for our independent varia- However, our analysis provides a twist to
bles is expectedly smaller in these models, the story of corporate legal favorability.
their statistical significance increases, which Although prestigious firms are less likely to
provides strong evidence that the effects be found liable of discrimination charges,
observed in Table 3 are not being driven by once juries rule against them and find them
the presence of outliers. blameworthy, those firms are punished more
In summary, our results suggest that an harshly for their wrongdoing. In the case of
organization’s status generates a halo effect employment discrimination suits, we see that
during the first stage of litigation. However, juries assign greater punitive damages to
in the second stage, when blameworthiness higher status companies, especially when their
has been unambiguously established, high high status is coupled with a strong reputation
status can provoke a halo tax, especially when in the same domain as their transgression.
it is paired with an admirable reputation in the We assert that the same social psychologi-
domain in which the transgression occurred. cal mechanism—expectations associated with
McDonnell and King 19
(Benson, Madensen, and Eck 2009; Simpson status perceptions of the general populous
2002, 2013). A large body of work in crimi- from which jurors are drawn. Given that prior
nology, and in white-collar crime more spe- work shows these measures meaningfully pre-
cifically, examines the attributes of defendants dict how organizations interact with activists
that affect legal outcomes, such as race, age, and outsiders to the corporate community
gender, socioeconomic status, and perceived (e.g., King 2008; McDonnell and King 2013;
character (e.g., Albonetti 1998; Nadler and McDonnell and Werner 2016; Werner 2015),
McDonnell 2011; Steffensmeier and Demuth we think this assumption is defensible. But we
2006; Steffensmeier et al. 1998). To our acknowledge that individual jurors’ familiar-
knowledge, this work has not considered how ity with the firms in our sample is likely to
the status of the employer of individual white- vary considerably, which does introduce
collar defendants affects the outcomes of their measurement error. Further work is needed to
prosecutions. Does working for a high-status shed light on the precise manner in which
company convey the same halo effect and information about a defendant’s status and
halo tax for a deviant employee as it would for reputation is shared with jurors over the course
the company? This represents an interesting of a trial, as well as how prestige signals might
question that might be pursued in future be manipulated by legal counsel to their cli-
research. ents’ advantage.
Although it helps establish the practical Another limitation of our study is that we
validity of our theoretical propositions, our only measure reputation that aligns with the
decision to empirically pursue our research domain of the transgression, rather than repu-
questions using archival data drawn from real tation in unrelated domains. Work in impres-
legal cases does limit our ability to establish sion management finds that firms that face
with certainty the causal role of our proposed social challenges often try to buffer their
mechanisms. This is because our data are public images by bolstering their reputation
subject to a first-order selection problem: with charitable endeavors in other, less con-
discrimination charges are not assigned ran- tentious domains (McDonnell and King
domly, and we cannot rule out the possibility 2013). A firm’s reputation along unrelated
that an employer’s status and reputation may dimensions might prime evaluators to see
affect the circumstances in which employees them as less blameworthy, without triggering
who are discriminated against will bring as strong a repulsion to hypocrisy when they
charges, or whether employees themselves are deemed blameworthy. Future work is nec-
will perceive that an adverse employment essary to assess our theory and the generaliz-
action was taken with discriminatory intent. ability of our findings with respect to other
Future work might address these limitations dimensions of reputation.
by exploring the halo effect and halo tax in a In summary, our study provides evidence
controlled context, such as through an experi- that prestige confers a beneficial halo effect
ment, where evaluators’ expectations could when audience members make attributions of
be measured directly and our proposed mech- guilt under conditions of uncertainty, but pres-
anisms could be manipulated to provide more tige produces a detrimental halo tax when
robust evidence of causality. audience members allot punishment after
Our use of archival cases also imposes data blameworthiness has been established. Our
limitations, as we have no ability to directly findings demonstrate the unevenness of the
observe jurors’ perceptions or access full case legal playing field and provide evidence that
transcripts. Instead, we rely on common prox- prestige plays an important role in shaping
ies for prestige from the organizational litera- corporate legal outcomes. But our study also
ture. The construct validity of these proxies in provides insight into the limits of corporate
the context of discrimination suits rests on an prestige by showing that in instances of proven
assumption that they meaningfully align with blameworthiness, prestige becomes a liability
22 American Sociological Review 00(0)
for corporations. Thus, prestige both elevates reputations). Additionally, while the KLD data cover
the full Fortune 500 for the period we study, firms
corporate influence and creates an inherent
had to apply to be considered for the Fortune rank-
risk for firms that engage in deviant behavior. ing, such that the latter proxy for reputation poten-
tially suffers from unobservable selection bias.
6. To avoid overestimating our models, we only
Acknowledgments include controls for adverse actions that occurred in
The authors would like to thank their four anonymous at least 5 percent of the cases in our sample. These
reviewers, as well as Adam Cobb, Zeke Hernandez, include, in order of frequency, termination or con-
Joshua Margolis, Nathan Miller, Mike Pfarrer, Robert structive discharge, harassment, hostile work envi-
Prentice, Jim Wade, Tim Werner, participants in seminars ronment, failure to accommodate, loss of benefits
at the Wharton School of the University of Pennsylvania, or pay, failure to promote, failure to hire or rehire,
the Academy of Management Meeting in Orlando, and the closer supervision or scrutiny, and demotion.
Strategic Management Society Annual Meeting in Prague 7. As noted earlier, we exclude class actions from our
for helpful comments on earlier versions of this paper. sample, so our findings with regard to this variable
are limited to cases with single plaintiffs or smaller
groups of plaintiffs. This exclusion is meaningful
Notes insofar as prior work suggests that cases with very
1. Status research at the individual level also indicates large numbers of victims may actually provoke a
that status may be a liability in certain situations. less punitive response from jurors, because jurors
Graffin and colleagues (2013), for example, found have more difficulty emotionally identifying with
that high-status members of the British Parliament larger and more abstract groups of victims (Nor-
who were implicated in an expense scandal expe- dgren and McDonnell 2011).
rienced greater pressure from journalists and were 8. In the case of compensatory damages, we add 1
consequently more likely to leave Parliament. And before taking the natural log so as not to lose obser-
Wheeler, Weisburd, and Bode (1982) found that vations with a value of 0.
the severity of punishment meted out to individu- 9. We also ran models where we instead controlled for
als prosecuted for white-collar crime was positively the number of times a firm was found guilty in the
related to their occupational status. prior three years, which is a more accurate proxy
2. In a review of theoretical work on the reputation for the repeat offender effect. We control instead
construct, Lange and colleagues (2011) said that for total cases in the prior three years, because this
researchers tend to think of reputation in one of three produced better fitting models and both variables
ways: (1) as being known, (2) as being known for produced substantively similar effects.
something, and (3) generalized favorability. Reputa- 10. Insofar as size may be considered a proxy for status,
tions certainly consist of different dimensions, but one might argue that the observed effects of employ-
our purpose is not to offer a new operationalization ees at each stage of a trial (negatively influencing
but rather to note that underlying all these definitions the likelihood of liability, but exaggerating the
is a belief that organizational reputations consist of extent of punishment) is further evidence of Hypoth-
shared perceptions about quality and standing. eses 1a and 3a. However, these effects might also
3. Only six of the 53 class action suits in our data went be explained by two other potential mechanisms
to trial. Of these six, two were dismissed on sum- we cannot rule out. First, given that larger firms
mary judgment and two ended in a verdict for the have more employees who might sue, they may be
defense. The two that resulted in a verdict of liabil- especially motivated to establish legal precedents
ity produced an average damages award of nearly that dissuade their employees from bringing suits,
30 million dollars. leading them to spend more money on legal defense,
4. We also ran the models with the simple raw score which could lower their likelihood of being found
from the Fortune ratings; these models produced liable. Second, juries might be motivated to punish
substantively identical findings as those incorpo- firms that employ many people more severely when
rating the variable produced by Brown and Perry’s they discriminate as a deterrence measure, given that
(1994) method. more people would be affected by such firms’ dis-
5. In separate models, we used an alternative measure criminatory behaviors if left unchecked.
of domain-specific reputation by utilizing Fortune 11. Interestingly, the distinction between civil and
magazine’s annual rankings of the Best Places to criminal cases is rarely made in sociological work
Work. The results for this alternative measure are in this area. Rather, definitions of corporate crime
substantively similar to those using the KLD-derived in the sociological literature generally encapsu-
proxy. We opt for the KLD proxy in our primary late both, casting corporate crime as “conduct of a
models because it allows for consideration of both corporation, or of employees acting on behalf of a
strengths and weaknesses in the domain at issue (pro- corporation, which is proscribed and punishable by
viding a score that encapsulates both good and bad law” (Braithwaite 1984:6; Simpson 2002).
McDonnell and King 23
Fombrun, Charles J., and Cees B. M. van Riel. 2003. Jonsson, Stefan, Henrich R. Greve, and Takako Fuji-
Fame & Fortune: How Successful Companies Build wara-Greve. 2009. “Undeserved Loss: The Spread
Winning Reputations. New York: Prentice-Hall of Legitimacy Loss to Innocent Organizations in
Financial Times. Response to Reported Corporate Deviance.” Admin-
Fombrun, Charles J., and Mark Shanley. 1990. “What’s istrative Science Quarterly 54(2):195–228.
In a Name? Reputation Building and Corporate Strat- Kahneman, Daniel, David Schkade, and Cass R. Sun-
egy.” Academy of Management Journal 33(2):233–58. stein. 1998. “Shared Outrage and Erratic Awards: The
Galanter, Marc. 1974. “Why the ‘Haves’ Come Out Psychology of Punitive Damages.” Journal of Risk
Ahead: Speculations on the Limits of Legal Change.” and Uncertainty 16(1):49–86.
Law & Society Review 9(1):95–160. Kalev, Alexandra, and Frank Dobbin. 2006. “Enforce-
Garrett, Brandon L. 2016. Too Big to Jail: How Pros- ment of Civil Rights Law in Private Workplaces: The
ecutors Compromise with Corporations. Cambridge, Effects of Compliance Reviews and Lawsuits Over
MA: Harvard University Press. Time.” Law & Social Inquiry 31(4):855–903.
Graffin, Scott D., Jonathan Bundy, Joseph F. Porac, Kenworthy, Lane, Stewart Macaulay, and Joel Rogers.
James B. Wade, and Dennis P. Quinn. 2013. “Falls 1996. “‘The More Things Change…’ Business Liti-
from Grace and the Hazards of High Status: The 2009 gation and Governance in the American Automobile
British MP Expense Scandal and its Impacts on Par- Industry.” Law & Social Inquiry 21(3):631–78.
liamentary Elites.” Administrative Science Quarterly Kim, Jerry W. 2012. “Arbiter of Science: Institutionaliza-
58(3):313–45. tion and Status Effects in FDA Drug Review 1990–
Greve, Henrich R., Donald Palmer, and Jo-Ellen Pozner. 2004.” Strategic Organization 10(2):128–57.
2010. “Organizations Gone Wild: The Causes, Pro- Kim, Jerry W., and Brayden G King. 2014. “Seeing
cesses, and Consequences of Organizational Miscon- Stars: Matthew Effects and Status Bias in Major
duct.” Academy of Management Annals 4(1):53–107. League Baseball Umpiring.” Management Science
Hadfield, Gillian K. 2005. “Exploring Economic and 60(11):2619–44.
Democratic Theories of Civil Litigation: Differences King, Brayden G. 2008. “A Political Mediation Model of
between Individual and Organizational Litigants in Corporate Response to Social Movement Activism.”
the Disposition of Federal Civil Cases.” Stanford Law Administrative Science Quarterly 53(3):395–421.
Review 57(5):1275–1327. King, Brayden G. 2011. “The Tactical Disruptiveness of
Hahl, Oliver, and Ezra W. Zuckerman. 2014. “The Social Movements: Sources of Market and Mediated
Denigration of Heroes? How the Status Attainment Disruption in Corporate Boycotts.” Social Problems
Process Shapes Attributions of Considerateness 58(4):491–517.
and Authenticity.” American Journal of Sociology King, Brayden G, and Mary-Hunter McDonnell. 2015.
120(2):504–554. “Good Firms, Good Targets: The Relationship between
Hahl, Oliver, Ezra W. Zuckerman, and Minjae Kim. Corporate Social Responsibility, Reputation, and
2017. “Why Elites Love Authentic Lowbrow Culture: Activist Targeting.” Pp. 430–54 in Corporate Social
Overcoming High-Status Denigration with Outsider Responsibility in a Globalizing World, edited by K.
Art.” American Sociological Review 82(4):828–56. Tsutsui. Cambridge, UK: Cambridge University Press.
Hamdani, Assaf, and Alon Klement. 2008. “Corpo- King, Brayden G, and David A. Whetten. 2008. “Rethink-
rate Crime and Deterrence.” Stanford Law Review ing the Relationship between Reputation and Legiti-
61(2):271–310. macy: A Social Actor Conceptualization.” Corporate
Harris, Jared, and Philip Bromiley. 2007. “Incentives to Reputation Review 11(3):192–207.
Cheat: The Influence of Executive Compensation and Klein, Jill, and Niraj Dawar. 2004. “Corporate Social
Firm Performance on Financial Misrepresentation.” Responsibility and Consumers’ Attributions and
Organization Science 18(3):350–67. Brand Evaluations in a Product-Harm Crisis.”
Harrison, Spencer H., Blake E. Ashforth, and Kevin G. International Journal of Research in Marketing
Corley. 2009. “Organizational Sacralization and 21(3):203–217.
Sacrilege.” Research in Organizational Behavior Koehler, Jonathan J., and Andrew D. Gershoff. 2003.
29:225–54. “Betrayal Aversion: When Agents of Protection
Heinz, John P., and Edward O. Laumann. 1982. Chicago Become Agents of Harm.” Organizational Behavior
Lawyers: The Social Structure of the Bar. New York: and Human Decision Processes 90(2):244–61.
Russell Sage Foundation. Kovács, Balázs, and Amanda J. Sharkey. 2014. “The
Hirsh, C. Elizabeth. 2009. “The Strength of Weak Paradox of Publicity: How Awards Can Negatively
Enforcement: The Impact of Discrimination Charges, Affect the Evaluation of Quality.” Administrative Sci-
Legal Environments, and Organizational Conditions ence Quarterly 59(1):1–33.
on Workplace Segregation.” American Sociological Kunda, Ziva. 1987. “Motivated Inference: Self-Serving
Review 74(2):245–71. Generation and Evaluation of Causal Theories.” Jour-
Jensen, Michael. 2006. “Should We Stay or Should nal of Personality and Social Psychology 53(4):636–47.
We Go? Accountability, Status Anxiety, and Cli- Lang, Gladys Engel, and Kurt Lang. 1988. “Recognition
ent Defections.” Administrative Science Quarterly and Renown: The Survival of Artistic Reputation.”
51(1):97–128. American Journal of Sociology 94(1):79–109.
McDonnell and King 25
Lange, Donald, Peggy M. Lee, and Ye Dai. 2011. “Orga- Nordgren, Loran F., and Mary-Hunter McDonnell. 2011.
nizational Reputation: A Review.” Journal of Man- “The Scope-Severity Paradox: Why Doing More
agement 37(1):153–84. Harm Is Judged to Be Less Harmful.” Social Psycho-
Love, E. Geoffrey, and Matthew Kraatz. 2009. “Charac- logical and Personality Science 2(1):97–102.
ter, Conformity, or the Bottom Line? How and Why O’Donnell, Ed, and Joseph J. Schultz Jr. 2005. “The Halo
Downsizing Affected Corporate Reputation.” Acad- Effect in Business Risk Audits: Can Strategic Risk
emy of Management Journal 52(2):314–35. Assessment Bias Auditor Judgment about Accounting
Malmendier, Ulrike, and Geoffrey Tate. 2009. “Super- Detail?” The Accounting Review 80(3):921–39.
star CEOs.” Quarterly Journal of Economics 124(4): Peterson, Richard A., and Narasimhan Anand. 2004. “The
1593–1638. Production of Culture Perspective.” Annual Review of
McDonnell, Mary-Hunter, and Brayden G King. 2013. Sociology 30:311–34.
“Keeping Up Appearances: Reputational Threat Podolny, Joel M. 1994. “Market Uncertainty and the
and Impression Management after Social Move- Social Character of Economic Exchange.” Adminis-
ment Boycotts.” Administrative Science Quarterly trative Science Quarterly 39(3):458–83.
58(3):387–419. Podolny, Joel M. 2001. “Networks as Pipes and Prisms
McDonnell, Mary-Hunter, Brayden G King, and Sarah A. of the Market.” American Journal of Sociology
Soule. 2015. “A Dynamic Process Model of Private 102(1):33–60.
Politics: Activist Targeting and Corporate Receptivity Pollock, Timothy G., Violina P. Rindova, and Patrick
to Social Challenges.” American Sociological Review G. Maggitti. 2008. “Market Watch: Information and
80(3):654–78. Availability Cascades among the Media and Investors
McDonnell, Mary-Hunter, and Elizabeth Pontikes. 2017. in the U.S. IPO Market.” Academy of Management
“Bad Company: The Reputational Implications of Journal 51(2):335–58.
Cross-Sector Interactions with a Stigmatized Firm.” Rachlinski, Jeffrey J., Sheri Lynn Johnson, Andrew J.
Working Paper, University of Pennsylvania. Wistrich, and Chris Guthrie. 2009. “Does Uncon-
McDonnell, Mary-Hunter, and Timothy Werner. 2016. scious Racial Bias Affect Trial Judges?” Notre Dame
“Blacklisted Businesses: Social Activist Challenges Law Review 84(3):1195–1246.
and the Disruption of Corporate Political Activity.” Rao, Hayagreeva. 1994. “The Social Construction of
Administrative Science Quarterly 61(4):584–620. Reputation: Certification Contests, Legitimation, and
Merton, Robert K. 1968. “The Matthew Effect in Sci- the Survival of Organizations in the American Auto-
ence.” Science 159(3810):56–63. mobile Industry: 1895–1912.” Strategic Management
Mishina, Yuri, Bernadine J. Dykes, Emily S. Block, and Journal 15(S1):29–44.
Timothy G. Pollock. 2010. “Why ‘Good’ Firms Do Raub, Werner, and Jeroen Weesie. 1990. “Reputation
Bad Things: The Effects of High Aspirations, High and Efficiency in Social Interactions: An Example
Expectations, and Prominence on the Incidence of of Network Effects.” American Journal of Sociology
Corporate Illegality.” Academy of Management Jour- 96(3):626–54.
nal 53(4):701–722. Rhee, Mooweon, and Pamela R. Haunschild. 2006. “The
Mizruchi, Mark S. 2004. “Berle and Means Revisited: Liability of a Good Reputation: A Study of Product
The Governance and Power of Large US Corpora- Recalls in the U.S. Automobile Industry.” Organiza-
tions.” Theory and Society 33(5):579–617. tion Science 17(1):101–117.
Mizruchi, Mark S. 2013. The Fracturing of the American Ridgeway, Cecilia. 1991. “The Social Construction of
Corporate Elite. Cambridge, MA: Harvard Univer- Status Value: Gender and Other Nominal Characteris-
sity Press. tics.” Social Forces 70(2):367–86.
Mood, Carina. 2010. “Logistic Regression: Why We Rindova, Violina P., Antoaneta P. Petkova, and Sureth
Cannot Do What We Think We Can Do, and What Kotha. 2007. “Standing Out: How New Firms in
We Can Do About It.” European Sociological Review Emerging Markets Build Reputation.” Strategic
26(1):67–82. Organization 5(1):31–70.
Nadler, Janice, and Mary-Hunter McDonnell. 2011. Rindova, Violina P., Ian O. Williamson, and Antoaneta
“Moral Character, Motive, and the Psychology of P. Petkova. 2010. “Reputation as an Intangible Asset:
Blame.” Cornell Law Review 97(2):255–304. Reflections on Theory and Methods in Two Empirical
Nelson, Robert L., Ellen C. Berrey, and Laura Beth Studies of Business School Reputations.” Journal of
Nielsen. 2008. “Divergent Paths: Conflicting Con- Management 36(3):610–19.
ceptions of Employment Discrimination in Law and Rindova, Violina P., Ian O. Williamson, Antoaneta P.
the Social Sciences.” Annual Review of Law and Petkova, and Joy Marie Sever. 2005. “Being Good
Social Science 4:103–122. or Being Known: An Empirical Examination of the
Nielsen, Laura Beth, Robert L. Nelson, and Ryon Lan- Dimensions, Antecedents, and Consequences of
caster. 2010. “Individual Justice or Collective Legal Organizational Reputation.” Academy of Manage-
Mobilization? Employment Discrimination Litiga- ment Journal 48(6):1033–49.
tion in the Post Civil Rights United States.” Journal Roberts, Peter W., and Grahame R. Dowling. 2002.
of Empirical Legal Studies 7(2):175–201. “Corporate Reputation and Sustained Superior
26 American Sociological Review 00(0)
Financial Performance.” Strategic Management Sullivan, Bilian N., Pamela Haunschild, and Karen Page.
Journal 23(12):1077–93. 2007. “Organizations Non Gratae? The Impact of
Rosoff, Stephen M. 1989. “Physicians as Criminal Unethical Corporate Acts on Interorganizational Net-
Defendants: Specialty, Sanctions, and Status Liabil- works.” Organization Science 18(1):55–70.
ity.” Law and Human Behavior 13(2):231–36. Sutherland, Edwin Hardin. 1949. White Collar Crime.
Selmi, Michael. 2000. “Why Are Employment Dis- New York: Dryden Press.
crimination Cases So Hard to Win?” Louisiana Law Sutton, John R., Frank Dobbin, John W. Meyer, and W.
Review 61(3):555–75. Richard Scott. 1994. “The Legalization of the Work-
Shaffer, Gregory. 2009. “How Business Shapes Law: A place.” American Journal of Sociology 99(4):944–71.
Socio-Legal Framework.” Connecticut Law Review Thye, Shane R. 2000. “A Status Value Theory of Power in
42(1):147–83. Exchange Relations.” American Sociological Review
Sharkey, Amanda J. 2014. “Categories and Organiza- 65(3):407–432.
tional Status: The Role of Industry Status in the Traut-Mattausch, Eva, Stefan Schulz-Hardt, Tobias Gre-
Response to Organizational Deviance.” American itemeyer, and Dieter Frey. 2004. “Expectancy Con-
Journal of Sociology 119(5):1380–1433. firmation in Spite of Disconfirming Evidence: The
Simpson, Sally S. 2002. Corporate Crime, Law, and Case of Price Increases Due to the Introduction of
Social Control. Cambridge, UK: Cambridge Univer- the Euro.” European Journal of Social Psychology
sity Press. 34(6):739–60.
Simpson, Sally S. 2013. “White-Collar Crime: A Review Ulmer, Robert R. 2001. “Effective Crisis Management
of Recent Developments and Promising Directions through Established Stakeholder Relations at Malden
for Future Research.” Annual Review of Sociology Mills as a Case Study.” Management Communication
39:309–331. Quarterly 14(4):590–615.
Sine, Wesley David, Scott Shane, and Dante Di Gregorio. U.S. Equal Employment Opportunity Commission
2003. “The Halo Effect and Technology Licensing: (USEEOC). 2013. “Charge Statistics: FY 1997
The Influence of Institutional Prestige on the Licens- through FY 2012.” Retrieved October 26, 2013
ing of University Inventions.” Management Science (http://www.eeoc.gov/eeoc/statistics/enforcement/
49(4):478–96. charges.cfm).
Skolnick, Paul, and Jerry I. Shaw. 1994. “Is Defendant Useem, Michael. 1984. The Inner Circle. New York:
Status a Liability or a Shield? Crime Severity and Oxford University Press.
Professional Relatedness.” Journal of Applied Social Wagner, David G., and Joseph Berger. 2002. “Expecta-
Psychology 24(20):1827–36. tion States Theory: An Evolving Research Program.”
Sorenson, Olav. 2014. “Status and Reputation: Syn- Pp. 41–77 in New Directions in Contemporary Socio-
onyms or Separate Concepts?” Strategic Organiza- logical Theory, edited by J. Berger and M. Zelditch Jr.
tion 12(1):62–69. New York: Rowman & Littlefield.
Staw, Barry M., and Lisa D. Epstein. 2000. “What Waguespack, David M., and Olav Sorenson. 2011. “The
Bandwagons Bring: Effects of Popular Management Ratings Game: Asymmetry in Classification.” Orga-
Techniques on Corporate Performance, Reputation nization Science 22(3):541–53.
and CEO Pay.” Administrative Science Quarterly Washington, Marvin, and Edward J. Zajac. 2005. “Status
45(3):523–56. Evolution and Competition: Theory and Evidence.”
Steffensmeier, Darell, and Stephen Demuth. 2006. Academy of Management Journal 48(2):282–96.
“Does Gender Modify the Effects of Race-Ethnicity Watts, Ross, and Jerold L. Zimmerman. 1986. Positive
on Criminal Sanctioning? Sentences for Male and Accounting Theory. Edgewood Cliff, NJ: Prentice-
Female White, Black, and Hispanic Defendants.” Hall.
Journal of Quantitative Criminology 22(3):241–61. Werner, Timothy. 2015. “Gaining Access by Doing Good:
Steffensmeier, Darrell, John Kramer, and Cathy Streifel. The Effect of Sociopolitical Reputation on Firm Par-
1993. “Gender and Imprisonment Decisions.” Crimi- ticipation in Public Policy Making.” Management
nology 31:411–46. Science 61(8):1989–2011.
Steffensmeier, Darrell, J. Ulmer, and J. Kramer. 1998. Wheeler, Stanton, David Weisburd, and Nancy Bode.
“The Interaction of Race, Gender, and Age in Crimi- 1982. “Sentencing the White-Collar Offender: Rhet-
nal Sentencing: The Punishment Cost of Being Young, oric and Reality.” American Sociological Review
Black, and Male.” Criminology 36(3):763–98. 47(5):641–59.
Stern, Ithai, Janet M. Dukerich, and Edward Zajac. 2014.
“Unmixed Signals: How Reputation and Status Affect Mary-Hunter McDonnell is an Assistant Professor of
Alliance Formation.” Strategic Management Quar- Management at the Wharton School of the University of
terly 35(4):512–31. Pennsylvania. Her work explores political interactions
Stuart, Toby E., Ha Hoang, and Ralph C. Hybels. 1999. between corporations and their stakeholders, as well as
“Interorganizational Endorsements and the Perfor- how these interactions shape corporate social activity and
mance of Entrepreneurial Ventures.” Administrative non-market strategy. Her research additionally explores
Science Quarterly 44(2):315–49. the mechanisms that stakeholders use to enforce
McDonnell and King 27
behavioral expectations for corporations and to punish department of sociology at Northwestern University.
corporate transgressions. She holds a JD from Harvard Brayden’s research focuses on how social movement
Law School and a PhD in management and organizations activists influence corporate social responsibility, organi-
from Northwestern University’s Kellogg School of zational change, and legislative policymaking. He also
Management. studies the antecedents and consequences of reputation
and status for organizations and other social actors. He is
Brayden G King is the Max McGraw Chair of Manage- currently a senior editor at the journal Organization Sci-
ment and the Environment at the Kellogg School of ence. He received his PhD in sociology from the Univer-
Management and holds a courtesy appointment in the sity of Arizona.