Professional Documents
Culture Documents
Doing Good or Feeling Good Justice Concerns Predict Online Shaming Via Deservingness
Doing Good or Feeling Good Justice Concerns Predict Online Shaming Via Deservingness
iew
Doing good or feeling good? Justice concerns predict online shaming via deservingness
and schadenfreude
v
Anna C. Barrona
re
Lydia Woodyattb,c
Emma F. Thomasb
er
J. E. Katherine Lohb
pe
Katherine Dunningb
Author Note
Correspondence concerning this article should be addressed to Anna C. Barron, School of
Pr
Public and International Affairs, Princeton University, Princeton, NJ, 08542. Email:
anna.barron@princeton.edu.
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
ed
v iew
re
Doing good or feeling good? Justice concerns predict online shaming via deservingness
er
and schadenfreude
pe
ot
tn
rin
ep
Abstract: 199
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
Abstract
ed
Public shaming has moved from the village square and is now an established online
phenomenon. The current paper explores whether online shaming is motivated by a person’s
iew
desire to do good (a justice motive); and/or, because it feels good (a hedonic motive),
examine two key aspects of social media that may moderate these processes: anonymity
v
(Study 1) and social norms (the responses of other users; Studies 2-3). Across three
re
experiments (N = 225, 198, 202) participants were presented with a fabricated news article
concerns about social justice were not directly positively associated with online shaming and
er
had few consistent indirect effects on shaming via moral outrage. Rather, justice concerns
were primarily associated with shaming via participants’ perception that the offender was
pe
deserving of negative consequences, and their feelings of schadenfreude regarding these
consequences. Anonymity did not moderate this process and there was mixed evidence for
ot
the qualifying effect of social norms. Overall, the current studies point to the hedonic motive
in general and schadenfreude specifically as a key moral emotion associated with people’s
tn
shaming behaviour.
rin
ep
Pr
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
Doing good or feeling good? Justice concerns predict online shaming via deservingness
ed
and schadenfreude
COVID-19 once again highlighted the speed and scale at which instances of public shaming
iew
can develop online. In countries undergoing lockdown, photos and videos of people refusing
to socially distance or wear masks went viral on social networking sites (Max, 2020). On
Twitter, the term “#COVIDiot” was coined to refer to an individual seen to breach public
v
health advice (Tait, 2020). Pandemic shaming is just the most recent example in a growing
re
list of online shaming events where individuals are publicly condemned for violating
perceived norms or values – with sometimes devastating consequences for the individuals
er
targeted (Parr & Billingham, 2020). Victims of online shaming have reported experiencing
anxiety, depression, and post-traumatic stress disorder (Ronson, 2016). In 2015, a 13-year-old
pe
girl died of suicide days after her father posted a video online shaming her for sending a
Marwick, 2021; Muir et al., 2020; Ronson, 2016), experimental work examining the factors
that motivate people to shame others online is still relatively nascent. In the current paper we
tn
explore two possible explanations for engagement in online shaming. We propose that, on
one hand, online shaming may reflect people’s desire to right a wrong (what we term the
rin
justice motive). That is, online shaming may be motivated by people’s concerns over
violations of social norms or moral standards of behaviour (Brickman et al., 1981; Lerner,
ep
1981), a belief that the transgressor is deserving of punishment (Feather, 1999), and their
feelings of moral outrage regarding the wrongdoing (Haidt, 2003; Montada & Schneider,
1989). Alternatively, calling out others for their (real or perceived) transgressions may also
Pr
feel good, and thus may be motivated by (what we term) the hedonic motive. Specifically, we
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
explore the role of schadenfreude – the experience of malicious pleasure at witnessing
ed
another’s misfortune (Feather, 2012; van Dijk & Ouwerkerk, 2014) – as a potential predictor
of shaming behaviour. We examine these two possible processes in the context of an incident
iew
Further, we explore two key aspects of online environments that may amplify
shaming behaviour via a justice or hedonic pathway: the role of anonymity (Study 1) and the
behaviour of other online users (Studies 2-3). It may be that the relative anonymity afforded
v
to online users exacerbates the justice and hedonic pathways outlined above as there are no
re
perceived costs to the self in an anonymous environment (Crockett, 2017; Christopherson,
2007; Lapidot-Lefler & Barak, 2012). In addition, the behaviour of other users may increase
er
a person’s likelihood of engaging in online shaming when shaming is condoned, supported,
and even encouraged by other users (i.e., it becomes normative; Chia, 2019; Lea et al., 1992;
pe
Reicher et al., 1995).
mention being driven by a “desire to do good” (p. 116). Therefore, one possibility is that
tn
people engage in online shaming because they are motivated to restore a sense of justice
following their perception that an important moral value or standard has been violated by
rin
another’s actions (Wenzel et al., 2008; Wenzel & Okimoto, 2010). Often, demands for
punishment in response to such violations reflect people’s belief that an offender is deserving
ep
If this is the case, we would expect that perceived norm violations and perceptions of
Pr
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
of the transgressor. Moral outrage is defined as “anger produced by the perceived violation of
ed
a moral standard or principle” (Batson et al., 2007, p. 1273). Conceptually, moral outrage can
be distinguished from personal anger – which predominantly arises where one’s personal
interests have been harmed (O’Mara et al., 2011). In the context of online shaming, moral
iew
outrage should arise following the perception that an individual has deliberately violated an
important moral standard or principle, and that, as such, they are deserving of punishment
(Feather, 1999; Smith, 2002). For instance, Bastian et al. (2013) demonstrated that moral
v
outrage mediated the relationship between perceptions of harm and the desired severity of
re
punishment of the offender.
Although feelings of outrage are often cast in a destructive light (for discussion see
er
Spring et al., 2018), current theorising in social psychology suggests that moral outrage is
activism (see Akfırat et al., 2020; Thomas et al., 2020; 2022). Importantly, moral outrage has
ot
been implicated as a justification for certain forms of online harassment following perceived
norm violations (Marwick, 2021). Marwick (2021) describes how even extreme forms of
tn
online shaming (e.g., disseminating personal information about an offender and/or sending
death threats) are justified on the basis of concerns about moral norms being violated.
rin
Shaming allows people to communicate that they oppose such norm violations, as well as
signalling to others in their given network to do the same (thereby triggering further outrage
ep
and harassment; Sawaoka & Monin, 2018). Given this, one possibility is that people’s
concerns about social justice will influence online shaming via perceptions that the offender
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
Feeling Good: Online Shaming Arising from the Hedonic Motive
ed
However, people may punish for motives beyond “doing good”. An alternative view
is that people participate in online shaming because they get an emotional payoff – that is,
because it feels good to do it. Indeed, Jensen (2012) argues that people are less motivated by
iew
altruistic concerns for social justice and more motivated by the gratification derived from
witnessing another receiving negative consequences (Leach et al., 2015). Schadenfreude has
v
been distinguished from related concepts such as gloating, where one aims to flaunt or boast
re
about an opponent’s misfortune – usually in the context of direct competition (Leach et al.,
experiencing schadenfreude).
pe
Like moral outrage, schadenfreude also arises following perceived norm violations
(e.g., Feather & Sherman 2002; Powell & Smith 2013; van Dijk & Ouwerkerk 2014).
of receiving punishment (Feather, 2006; Wang et al., 2019). However, unlike moral outrage,
punished (Feather, 2008; 2012). Indeed, research has shown that the experience of
al., 2020). In the context of online interactions, Wei and Lui (2020) found that disclosures of
misfortune on social media elicited people’s expressions of schadenfreude, and that these
ep
effects were mediated by perceived deservingness. Further, Cecconi et al.’s (2020) analysis
issues (i.e., Brexit) showed that many posts expressed what they term “injustice
Pr
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
deserved because of that person’s own (past) unjust behaviour (p. 2). Taken together, these
ed
findings suggest that justice concerns and deservingness perceptions may influence online
shaming via people’s feelings of pleasure regarding the offender’s punishment (i.e., their
experience of schadenfreude).
iew
The (Moderating) Role of the Online Environment
There are many aspects of online shaming that distinguish it from its historical
analogues. Unlike the town square of centuries past, information online can be easily
v
distributed to a global audience via a couple of clicks (Klonick, 2015). Online, what may
re
begin as one user calling out a transgressor for their behaviour often snowballs into more
malevolent forms of online shaming (e.g., “doxxing” – a practice where online users work
er
together to distribute private identifying information about a target; Eckert & Metzger-
Riftkin, 2020, p. 2). Thus, in the current paper we aim to understand both the individual
pe
motives as well as the contextual factors that may influence the nature of shaming that people
engage in on social media. We therefore focus on two aspects of social networking sites that
may exacerbate online shaming via a justice or hedonic path: the relative anonymity of
ot
relatively anonymous (Brady et al., 2020). A wealth of research has implicated anonymity as
an important motivator in the display of more hostile online behaviour (Barlett et al. 2016,
rin
Fox et al., 2015; Moore et al., 2012). For example, Wilton and Campbell (2011) found that
untouchable” (p. 2). Accordingly, anonymity may work to make either the justice motive or
the hedonic pathway stronger, as restoring justice or feeling malicious pleasure via shaming a
target online does not entail any perceived consequences for oneself (Brady et al., 2020;
Pr
Crockett, 2017).
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
Alongside the affordances of anonymity, social media provides a key platform for
ed
intra- and intergroup interaction. The Social Identity Model of Deindividuation Effects
(SIDE; Reicher, Spears & Postmes, 1995; Postmes & Spears, 1998a) proposes that when
immersed in an online group, people make a shift from their personal identity to their social
iew
identity. This shift has been shown to lead to behaviour that is strongly susceptible to
normative influence (i.e., people attend more to norms operating within the online group or
network that they identify with; Postmes et al., 2000; Reicher et al, 1995). Thus, online
v
shaming behaviour could be understood in terms of peoples’ increased susceptibility to the
re
thoughts, feelings, and actions of those around them (the social norms present in the given
context), and, in this way, provide an account for how online shaming can emerge as a
er
normalized behaviour within online networks.
Indeed, Brady et al. (2020) propose that expressions of moral emotions (e.g., outrage,
pe
disgust, contempt, and shame) are more likely to “go viral” on social media (p. 980). Brady et
al. (2020) suggest that context specific norms (i.e., the norms operating within a particular
online network) may shape the extent to which emotions are diffused on social media. For
ot
example, following the widespread expression of moral outrage online, other users may shift
their emotions from positive (e.g., happiness) to negative (e.g., anger) to match the emotions
tn
of similar others (Brady et al., 2020; Goldenberg et al., 2020). In line with this work, Chia
(2019) found that individuals’ cyber vigilantism (that is, searching for information to “name
rin
and shame” wrongdoers) was influenced by the behaviour of others – an effect mediated by
In the current research, we propose that social norms may influence online shaming
via both a justice and hedonic pathway. We suggest that how people express their concerns
around social justice, perceptions of deservingness, moral outrage, and schadenfreude – that
Pr
is, the type and severity of shaming they engage in – may be particularly susceptible to
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
normative influence (i.e., shaped by the responses of others). Specifically, we test whether
ed
presenting shaming as a normative response (in the form of comments left by other users)
enhances the links between justice concerns, deservingness, moral outrage, and shaming (the
iew
schadenfreude, and shaming (the hedonic motive).
In three studies, we explore two pathways through which online shaming may occur:
v
1) a justice motive, in which justice concerns and perceptions of deservingness facilitate
re
online shaming via participants’ moral outrage, and/or 2) a hedonic motive, whereby the
Figure 1 for our hypothesized conceptual model). In Study 1 we test the moderating role of
pe
anonymity by manipulating how identifiable participants feel during the study. In Studies 2-3,
we manipulate social norms by varying the severity of comments left by perceived other
users.
ot
tn
rin
ep
Pr
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
Across all three studies participants were told that they would be taking part in a
ed
survey interested in people’s responses to information that they encounter online, and that
their task would be to evaluate an article that had been published on an online news website.
iew
Islamophobia on Australian public transport. The article described an incident in which a
male university student (‘Benjamin’) verbally abused a woman wearing a hijab on a Sydney
bus. Participants were told that below the news article they would be able to view responses
v
from other participants who had previously participated (or were currently participating) in
re
the study. In reality, these responses were preprogramed by the researchers. Participants were
next provided with the opportunity to respond. We systematically varied the information
er
conveying the reactions of other participants (Studies 2-3); and took behavioural measures of
our outcome variables (‘likes’, Study 1; ‘reacts’, Studies 2-3; ‘shares’; and comments,
pe
Studies 1-3). We then took measures of justice concerns, moral outrage, deservingness, and
violations on social media and, it follows, that online shaming can include several online
tn
actions – from sharing images, to posting statuses, to leaving comments (Muir et al., 2010).
In the current study we therefore operationalized online shaming using two behavioural
rin
measures. The first includes participants’ overall engagement with the news article, that is,
the extent to which they commented, “liked”, “reacted”, or “shared” the article. While these
ep
actions may appear relatively harmless in isolation, interacting with or “sharing” content
online can increase its visibility and thereby contribute to a “pile on” effect whereby
thousands of users are exposed to the content in a short period of time; Thompson & Cover,
Pr
10
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
The second shaming measure was created to explore the severity of participants’
ed
responses by analysing the content of participants’ comments. Here we drew on Braithwaite’s
iew
shaming intends to label an offender as “evil” in an attempt to cast them out from the moral
community (p. 101). In contrast, reintegrative shaming condemns an act as immoral but does
not label the transgressor as such. Thus, consistent with Braithwaite’s concept of
v
“stigmatizing” shaming, severe shaming responses were defined as those that targeted the
re
offender as a person, while comments that targeted the offender’s behaviour were considered
less severe.
er Study 1
and/or commenting). Participants were told that they could choose to make any (or all) of the
available responses and that they were also free to not respond. Anonymity was manipulated
ot
via instructions that told participants that their responses (and the responses of the “other
names (this data was not collected and only observable to the participant). We tested our
theoretical model using multigroup structural equation modelling (SEM) with anonymity
rin
(anonymous, identifiable) as the grouping variable. Multigroup SEM allows for testing
complex mediation models across experimental groups (Yuan & Bentler, 2001).
ep
Method
(79% female; Mage = 22.20), 169 of whom participated for course credit and 56 volunteers
Pr
from the wider student population who were reimbursed 10.00AUD for their time. The study
11
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
adopted a 2-cell between-groups design (anonymous vs. identifiable) design. Sample size was
ed
determined based on prevailing standards at the time that we collect the data of
iew
link and were randomly assigned to one of the two experimental conditions (anonymous vs.
identifiable). Participants were told that their task would be to read a news article that had
recently been published on an online news site. The news article was created to emulate
v
articles typically found on online news websites (the content was adapted from a series of
re
Islamophobic incidents that had occurred and were published online between 2015-2016).
Participants were told that the comments, ‘likes’, and ‘shares’ of other participants (described
er
as people who, like them, were currently participating or had previously participated in the
study) would be displayed beneath the news article. In Study 1 the responses of the “other
pe
participants” were held constant across condition. Consistent with Braithwaite’s (1989)
targeted the offender: e.g., “Disgusting this guy shoud be ashamed of himself”; “this is awful,
assigned a personal user ID (i.e., H38119) and told that their response would appear
rin
alongside this username. The comments from the supposed other participants in this
112), participants were required to enter their first and last name and were led to believe that
their responses would appear next to their name. In this condition, the comments from the
Pr
1 Study 1 involved a 2 x 2 design. We report the results of another (unsuccessful) manipulation unrelated to the
focal variables in the online supplementary materials.
2 All of the spelling and grammatical errors in the comments are intentional
12
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
supposed “other users” had a series of first and last names attached to them, all fabricated by
ed
the researchers (Figure 2 depicts the news article and the comments shown to participants in
v iew
re
er
pe
Figure 2. User interface for Study 1 for participants in the Identifiable condition.
ot
and/or “sharing” the article. Participants then completed an anonymity manipulation check
tn
Participants also completed additional exploratory measures unrelated to the focus of the
rin
current paper. Finally, participants completed demographic variables and were debriefed
Dependent Measures
All items were measured on a 7-point Likert scale anchored at 1 (strongly disagree) to
7 (strongly agree) unless otherwise specified. Measures are scored such that higher scores
Pr
13
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
Manipulation check. One item was used to assess how anonymous participants felt
ed
when responding to the article, “I felt anonymous when choosing how to respond to the news
article”.
Justice concerns. Adapted from Wenzel & Okimoto (2010), four items were used to
iew
assess participants’ perception that the offender in the news article had violated important
values, “I feel the person in the article acted in a way that is inconsistent with my values and
beliefs”, “The person in the article disregarded the values we should all share”, “The person
v
in the article has violated common decency”, and “I have similar values and beliefs as the
re
person in the article” (reverse-coded), α = .77.
Deservingness. Three items were created to capture participants belief that the
er
offender in the news article was deserving of negative consequences, “I feel that the person in
the article deserves any negative outcomes they may receive”, “The person in the article
pe
deserves to be punished for their actions”, and “The person in the article deserves to be called
Moral outrage. Three items were used to capture participants’ emotional experience
ot
of moral outrage: “I feel angry because of the actions of the person in the article”, “I feel
outraged by the actions of the person in the article”, and “I feel a sense of injustice about
tn
Schadenfreude. Schadenfreude was assessed using two items adapted from Feather
rin
& Sherman (2002), “It was gratifying to see how others responded to the person in the
article” and “I felt pleased by the responses of others to the person in the article”, r = .77.
ep
were created using the three behavioural measures (whether participants commented,
“reacted”, or “shared”).
Pr
14
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
Total number of shaming responses. A total shaming response score was calculated
ed
by adding the number of behavioural responses participants selected out of the available
iew
all blind to participant condition – who coded for the level of condemnation expressed in
shaming that targets the offender’s behaviour; 4 = stigmatizing shaming that targets the
v
offender as a person). As anticipated above, stigmatizing shaming responses (coded 4) were
re
those that targeted the offender as a person, consistent with Braithwaite (1989). See
his upbringing or lack thereof. Hope he is not representative of Aussies at large”, “This guy
pe
needs to be dismissed from his Law degree and never be allowed to study Law at any Uni. He
of course should be charged with verbal assault”. An example of shaming that targeted the
offender’s behaviour (coded 3) included: “A very horrible thing for someone to say, I feel so
ot
sorry for the girl”. An example of a comment that was coded as not expressing any
condemnation regarding the actions of the transgressor: “I feel bad for that lady”. Inter-rater
tn
reliability was high, ICC = .96. Accordingly, a total severity of comment score was
Results
more anonymous than participants in the identifiable condition (M = 3.29, SD = 1.62), t(223)
= -5.79, p < .001, d = 0.77. Table 1 displays the scale inter-correlations and descriptive
Pr
15
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
statistics for the key dependent variables across the two experimental (anonymous vs.
ed
identifiable) conditions.
Participant responses. There was variation in the way participants responded to the
article. 49.3% of participants left a comment, 23.6% “shared” the article, and 18.7% “liked”
iew
the article. Of those that commented (N = 111), 39 participants left comments that were
classified as shaming or above, while 42 participants left a comment that was classified as
stigmatizing shaming (category 4). Thus, the majority of comments were classified as the
v
highest possible severity – shaming that targeted the offender as a person.
re
er
pe
ot
tn
rin
ep
Pr
16
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
e d
w
Table 1. Summary of means, standard deviations, and 95% confidence intervals (across conditions), and intercorrelations for key variables
(Study 1). Note. *p < .05. **p < .01. ***p < .001
v i e 3 4 5 6
1. Moral outrage
2. Justice concerns
5.40 (1.10)
[5.18, 5.60]
6.12 (1.04)
5.40 (1.30)
[5.18, 5.60]
6.04 (1.05)
-
.46**
r e -
3. Perceptions of deservingness
[5.92, 6.31]
4.79 (1.25)
[4.55, 5.03]
er
[5.83, 6.23]
4.68 (1.27)
[4.44, 4.90]
.45** .37** -
4. Schadenfreude
p e4.43 (1.35)
[4.18, 4.69]
1.91 (0.75)
.30**
.05
.18**
.08
.44**
.13*
-
.26** -
t
[1.76, 2.05] [1.76, 2.05]
6. Comment severity 2.16 (1.29) 2.08 (1.21) .11 .10 .07 .23** .61** -
n o
[1.93, 2.39] [1.11, 1.28]
ir n t
e p
P r 17
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
Testing the hypothesized model. Structural equation modelling (SEM) was
ed
conducted using Amos 25.0 (Arbuckle, 2007) to test our hypothesised model. Justice
deservingness influenced shaming behaviour (comment severity and total number of shaming
iew
responses) via two separate pathways: moral outrage and schadenfreude. We included direct
paths from justice concerns to both shaming outcomes to explore whether there was any
direct effect of justice concerns on shaming. The error terms of moral outrage and
v
schadenfreude were allowed to correlate, as were the error terms of the two shaming
re
variables. To test the moderating role of anonymity we used the anonymity manipulation
between schadenfreude and the two shaming outcome variables, and moral outrage and the
pe
shaming outcomes, differed depending on whether participants were anonymous or
We report several widely accepted goodness-of-fit indices to judge model fit: the
ot
comparative fit index (CFI), the root mean square error of approximation (RMSEA), and the
standardized root mean square residual (SRMR). Good fit for CFI is indicated by values of
tn
higher than .95; and values of .08 or lower for both RMSEA and SRMR (Bentler, 2007). As
can be seen in Table 2, the initial hypothesized model did not fit the data well (Table 2,
rin
Model 1). Modification indices suggested that there was additional variance between justice
concerns and moral outrage that needed to be accounted for; adding this path significantly
ep
improved model fit (see Table 2, Model 2). This path accounted for concerns about justice
influencing moral outrage directly, over and above participants’ perceptions of deservingness.
Pr
18
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
Table 2. Fit statistics for the hypothesized structural equation models.
ed
Chi-square difference
Model description χ2 (df) CFI RMSEA SRMR
test
Model 1: Theoretical model χ2 (8) = 33.84*** .91 .12 .07 -
iew
Model 2: Extra path from justice
χ2 (6) = 5.14 1.00 .00 .03 -
concerns to moral outrage
Model 3: All paths constrained χ2 (16) = 13.18 1.00 .00 .05 -
Model 4: Path from schadenfreude –
χ2 (15) = 13.06 1.00 .00 .05 χdiff2(1) = 0.12
comment severity released
Model 5: Path from schadenfreude –
χ2 (15) = 12.25 1.00 .00 .04 χdiff2(1) = 0.93
total responses released
Model 6: Path from moral outrage –
v
χ2 (15) = 13.18 1.00 .00 .05 χdiff2(1) = 0.00
comment severity released
Model 7: Path from moral outrage –
χ2 (15) = 13.18 1.00 .00 .04 χdiff2(1) = 0.00
re
total responses released
Note. *** p <.001.
We then fixed all the paths in the modified model to be the same across conditions, to
er
test the structural relationships between variables (to test for moderation; Table 2, Model 3).
pe
This model with standardized path (beta) weights is shown in Figure 2. We then compared
this constrained model with a series of models in which the paths from schadenfreude to
comment severity (Model 4) and total shaming responses (Model 5), and the paths from
ot
moral outrage to comment severity (Model 6) and total responses (Model 7), were released
(free to vary) across condition. Releasing each path separately would allow us to isolate the
tn
moderation effect to a specific path (or paths). The unconstrained models were compared
with the constrained model using the chi-square test statistic, appropriate for testing nested
rin
models (Kline, 2011). A non-significant chi-square value supports retention of the more
parsimonious model (i.e., with the constrained paths). That is, support for the moderating role
of anonymity would be obtained if the released model(s) fit better than the model in which
ep
the paths were constrained to be the same (i.e., there is a significant difference between the
chi-square value of Model 3 and the released models). All other paths remained constrained
Pr
19
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
Table 2 shows that the difference between the constrained (Model 3) and
ed
unconstrained models (Models 4-7) were not significant – indicating that the models did not
fit the data better when the hypothesized paths were allowed to differ. Thus, there was no
evidence of moderation. The relationships between schadenfreude and shaming, and moral
iew
outrage and shaming, were unaffected by the presence (vs. absence) of the anonymity
manipulation.
We computed indirect effects using the indirect effects command in Amos (Gaskin &
v
Lim, 2018) with 10,000 bootstrap samples (95% confidence intervals) following Preacher
re
and Hayes (2008). The indirect effect of justice concerns on shaming via perceptions of
deservingness and moral outrage was not significant (see Table 3 for the indirect effects for
er
each pathway). However, there were significant indirect effects of justice concerns on
punishment, which was associated with feelings of pleasure seeing them suffer
20
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
ed
v iew
re
Figure 3. Study 1 structural model (Table 2, Model 3: paths constrained across conditions).
Values represent standardized regression coefficients. Values to the left of the slash represent
er
the standardized weights in the identifiable condition, while values to the right of the slash
represent standardized weights in the anonymous condition. *** Denotes that the path is
pe
significant at p <.001. Dashed lines represent non-significant paths. N.B. Residual error terms
for moral outrage and schadenfreude, and both shaming variables, respectively, were allowed
to covary.
ot
The direct paths from justice concerns to the two shaming outcomes were not significant.
Interestingly, justice concerns did not influence shaming via increasing moral outrage –
rin
moral outrage had no significant relationship with either shaming outcome. Rather, the results
suggest that it was via participants’ perceptions that the offender in the news article was
ep
provide some initial evidence that “feeling good” at the suffering of others may play a
Pr
21
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
Contrary to our expectations, we found anonymity had no moderating effect on the
ed
relationships of interest. Although the manipulation successfully induced a difference in
anonymity across conditions, the mean for participants in the “identifiable” condition was
just over the midpoint of the scale, indicating that participants may have felt relatively
iew
anonymous even in the identifiable condition. According to the predictions of the SIDE
model, the relative anonymity of online environments should increase the likelihood that
participants attend to the social norms within a given context (Postmes & Spears, 1998b;
v
Postmes et al., 2000). Thus, in Study 2 we varied the responses of other users to assess
re
whether people’s participation in shaming was a result of their endorsement of wider societal
norms that condone shaming individuals who violate moral standards of behaviour – or the
er
norms operating in situ (the norm presented in the form of “other participants” responses and
comments).
pe
Study 2
Method
Mechanical Turk (MTurk) (43.1% female; Mage = 35.24, SDage = 10.93) and received
USD$1.20 for participating. Participation was limited to residents in the United States,
tn
Canada, United Kingdom, and Australia. As in Study 1, this study utilized a 2-cell design.
Participants were shown the same news article used in Study 1 and were randomly allocated
rin
to a condition that contained comments beneath the news article shaming the transgressor, or
Procedure. The procedure and set up were similar to that of Study 1. In this study,
participants in the shaming comments condition were shown fourteen comments below the
news article that targeted the offender as a person (“stigmatizing shaming” according to
Pr
Braithwaite; 1989; see Figure 4 for a screenshot of the comments in this condition).
22
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
Participants were then provided with the opportunity to submit a comment, “react”, or
ed
“share” the article. In the no comments condition, participants did not see any responses
below the news article (they were presented with the news article and directly underneath
iew
In this study, participants were given the option to “react” to the news article – as well
as “sharing” and commenting (as in Study 1). “Reacts” were introduced to Facebook in 2016,
letting users hold on the “Like” button to respond to content with 5 emotions: “Love”,
v
“Haha”, “Wow”, “Sad” or “Angry”. Participants were again led to believe that the comments,
re
“reactions” and “shares” displayed were from participants who had participated (or were
currently participating) in the study. Following this, all participants completed measures of
er
justice concerns (3 items, α = .80), perceptions of deservingness (2 items, α = .76),
Figure 4. User interface for the shaming comments condition. Full comments were viewed
Pr
23
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
As in Study 1, three coders (2 researchers and one independent rater – all blind to
ed
participant condition – rated the severity of participants’ comments in line with the coding
criteria established for Study 1 (Appendix A). Inter-rater reliability was again high (ICC
=.97) and thus a total severity of comment score was calculated by averaging scores across
iew
the three raters.
Results
Participant responses. Table 4 shows the means, standard deviations, and zero-order
v
correlations between the key variables. In this study, 76.8% of participants left a comment
re
and 89.9% “reacted” to the article. Of the participants who reacted, 0.6% “love” reacted,
2.2% “haha” reacted, 19.1% “wow” reacted, 40.4% “sad” reacted, and 33.7% “angry”
er
reacted. 3.9% “liked” the article, and 37.4% shared the article to the supposed “other
24
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
Table 4. Summary of means, standard deviations, and 95% confidence intervals (across conditions), and intercorrelations for key variables,
e d
w
Study 2. Note. *p < .05. **p < .01. ***p <.001.
Variable
1. Moral outrage
No comments (N = 95)
5.42 (1.15)
Shaming norm (N = 103)
5.19 (1.45)
1
-
v i
2
e 3 4 5 6
2. Justice concerns
[5.19, 5.67]
5.87 (1.12)
[5.65, 6.09]
[4.93, 5.45]
5.85 (1.22)
[5.61, 6.07]
r e
.56** -
3. Perceptions of deservingness
4. Schadenfreude
5.29 (1.16)
[5.05, 5.54]
4.77 (1.42)
er 5.10 (1.27)
[4.81, 4.38]
4.87 (1.58)
.51**
.61**
.43**
.49**
-
.70** -
p e [4.57, 5.18]
3.11 (0.93)
[2.93, 3.28]
.19* -.01 .09 .22** -
6. Comment severity
t
2.47 (1.04)
[2.26, 2.66]
o
2.75 (1.17)
[2.54, 2.97]
.24** .17* .21* .31** .65** -
t n
ir n
e p
P r 25
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
Testing the hypothesized model. As in Study 1, we tested for mediation using
ed
structural equation modelling in IBM SPSS Amos 25. Modification indices again suggested
that model fit would significantly improve when a direct path was added from justice
iew
We next used multi-group structural equation modelling to test whether the presence
of shaming comments influenced the relationships between schadenfreude and shaming and
moral outrage and shaming, respectively. Condition (shaming comments, no comments) was
v
the grouping variable. Model 3 shows the results for the constrained model (displayed in
re
Figure 4). We compared this constrained model to four unconstrained models in which the
key paths from schadenfreude and shaming outcomes (Models 4 and 5) and moral outrage
er
and shaming outcomes (Models 6 and 7) were released (free to vary) across condition. As in
Study 1, the remaining paths were held constant as these relationships were predicted to
pe
remain stable across conditions. Support for moderation would be obtained if the released
models had better model fit than Model 3 (evidenced by a statistically significant difference
Table 5. Fit statistics for the structural equation models (Study 2).
Chi square
Model description X2 (df) CFI RMSEA SRMR
difference test
Model 1: Theoretical model χ2 (10) = 94.52*** .81 .21 .09 -
rin
26
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
Results revealed that the relationships between schadenfreude and shaming and moral
ed
outrage and shaming were unaffected by the presence or absence of stigmatizing shaming
comments.
iew
Table 6. Indirect effects Study 2.
v
Outcome: Total number of shaming responses
re
Justice concerns deservingness schadenfreude .30*** .02, .10
Justice concerns deservingness moral outrage .16* .00, .05
Note. *** p <.001, p <.05.
Contrary to the results of Study 1, there was a significant indirect effect of justice
er
concerns on total number of shaming responses via deservingness and moral outrage (see
pe
Table 6). However, there was no indirect effect of justice concerns on the severity of
participant’s comments via deservingness and moral outrage. In line with the results of Study
1, there were significant indirect effects from justice concerns to shaming (both comment
ot
severity and total number of shaming responses) via perceptions of deservingness and
schadenfreude.
tn
rin
ep
Pr
27
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
ed
Figure 5. Path model for Study 2 (Table 5, Model 3 – all paths constrained). Values reflect
standardized regression weights. Values to the left of the slash represent the standardized
iew
weights in the no comments condition. Values to the right of the slash represent the
standardized weights in the shaming comments condition. ***Denotes that the path was
significant at p <.001. ** Denotes that the path was significant at p <.01. * Denotes that the
v
path was significant at p <.05. N.B. Residual error terms for moral outrage and
re
schadenfreude, and both shaming variables, respectively, were allowed to covary. Dashed
er
Replicating the results of Study 1, Study 2 found evidence for perceptions of
deservingness and schadenfreude as key predictors of online shaming. Concerns about justice
pe
indirectly predicted shaming via participant’s perceptions that the offender was deserving of
Intriguingly, in this study, participants’ concerns about justice were associated with a
reduction in total shaming responses. Given that the zero-order correlations between justice
tn
and shaming (Table 4) were non-significant and/or positive, this finding represents a
suppressor effect. That is, once the effects of deservingness, outrage and schadenfreude on
shaming are accounted for, it reveals a “true” negative effect of justice concerns in these data.
rin
Further, in this study, moral outrage was a weak positive predictor of shaming responses –
Results showed that the presence or absence of stigmatizing shaming comments did
not influence the relationships between schadenfreude and moral outrage on participants’
Pr
28
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
given the ubiquitous nature of such comments in online networks. That is, in the absence of
ed
any comments at all below the article, participants may have defaulted to thinking about
severe forms of shaming typically seen in response to such transgressions on social media. To
account for this possibility, in Study 3 we manipulated group norms more concretely by
iew
providing both conditions with comments below the news article which, depending on
Study 3
v
Method
re
Design and participants. Study 3 again used a one-factor between-subjects design
and manipulated the comments participants viewed to create the presence of a shaming norm
er
(shaming comments vs. non-shaming comments). Two hundred and two participants from the
USA were recruited online via Amazon’s Mechanical Turk (MTurk) (49.5% female; Mage =
pe
37.55, SDage = 11.53) and reimbursed US$1.20 for their participation.
Procedure. The procedure and set up were similar to the first two studies.
Participants in both conditions were presented with the same doctored news article and the
ot
content shown beneath the article varied depending on condition. In the stigmatizing
comments condition, participants were shown the same comments used in Study 2. In the
tn
non-stigmatizing comments condition, the comments did not express disapproval towards the
offender, but rather sympathy for the victim or disapproval for the behaviour, rather than the
rin
offender as a person. Example comments include: “very troubling to know this happened on
a public bus”; “I tried wearing a scarf for 7 days in different styles and the harshest responses
ep
were when I looked Muslim”. Following the manipulation, participants were presented with
the full set of items used in Study 1 (all measures formed reliable scales, αrange = .78 -.95.
N.B. Moral outrage was captured by two items in this study). Comments were again rated for
Pr
29
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
their severity and a mean severity score was calculated by averaging scores across the three
ed
raters (ICC = .96).
Results
iew
Participant responses. 83.2% left a comment (N = 168). Of these, 50 comments were
rated as stigmatising, 29 contained features of shaming, and the remaining 20 fell in between
these two categories. Accordingly, 58.9% of comments were shaming the transgressor’s
v
behaviour/person). 63% of participants reacted “Angry”, 39.6% used the “Sad” react, 11.9%
re
er
pe
ot
tn
rin
ep
Pr
30
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
Table 7. Summary of means, standard deviations, and 95% confidence intervals (across conditions), and intercorrelations for key variables
e d
Variable Non-stigmatizing (N = 95) Stigmatizing (N = 103)
1
i
2
ew 3 4 5 6
v
1. Moral outrage 5.01 (1.67) 4.68 (1.63) -
[4.68, 5.33] [4.35, 5.01]
2. Justice concerns
3. Perceptions of deservingness
5.50 (1.32)
[5.25, 5.75]
5.02 (1.48)
5.53 (1.37)
[5.27, 5.75]
5.05 (1.44)
r e .55**
.60**
-
.63** -
4. Schadenfreude
[4.73, 5.32]
4.84 (1.69)
[4.50, 5.14]
er [4.77, 5.35]
4.52 (1.74)
[4.17, 4.84]
.68** .44** .61** -
6. Comment severity
3.23 (0.81)
[3.07, 3.38]
2.65 (1.01)
p e 3.10 (0.91)
[2.91, 3.29]
2.73 (1.08)
.21*
.37**
.01
.25*
.09
.29*
.28**
.40**
-
.56** -
t
[2.47, 2.85] [2.50, 2.96]
(Study 3). Note. *p < .05. **p < .01. ***p < .001.
n o
ir n t
e p
P r 31
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
Testing the hypothesized model. Consistent with the approach adopted in Studies 1-
ed
2, our modified hypothesized model (i.e., the model with an additional direct path from
justice concerns to moral outrage was added) fit the data well (Table 8). Error terms for both
shaming outcomes and schadenfreude and moral outrage were covaried. Condition was again
iew
the grouping variable (non-stigmatizing comments, stigmatizing comments) for our
multigroup structural equation model. Table 8 shows the fit statistics for the three models.
Table 8 shows that there was a marginal difference between Model 3 and Model 4, p = .057.
v
Figure 6 depicts the model with standardized path weights. Examination of the standardized
re
weights reveals that schadenfreude was a significant predictor of the severity of shaming
when participants were shown stigmatizing comments, but not significant when participants
er
were shown non-stigmatizing comments (see Figure 6).
pe
Table 8. Fit statistics for the structural equation models (Study 3).
Model description X2 (df) CFI RMSEA SRMR Chi square difference test
Model 1: Theoretical model χ2 (10) = 75.19 .86 .18 .08 -
Model 2: Extra path from justice -
χ2 (6) = 3.43 1.00 .00 .01
ot
released.
Note. *** p <.001.
ep
Tests of the indirect effects showed that there was no significant indirect effect of
justice concerns on comment severity via deservingness and moral outrage (in line with
Study 1 and 2), and no significant indirect effect of justice concerns on total shaming
Pr
responses via deservingness and moral outrage (confirming the results of Study 1). Consistent
32
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
with the results of Studies 1 and 2, there were again significant indirect effects of justice
ed
concerns on total number of shaming responses and comment severity via participants’
perceptions of deservingness and schadenfreude. See Table 9 for the indirect effects.
iew
Table 9. Indirect effects (Study 3).
Outcome: Comment severity IE 95% CI
Justice concerns deservingness schadenfreude .38*** .04, .13
Justice concerns deservingness moral outrage .28 -.00, .07
Outcome: Total number of shaming responses
v
Justice concerns deservingness schadenfreude .38*** .03, .11
Justice concerns deservingness moral outrage .28 -.00, .05
re
Note. ***p <.001.
er
pe
ot
tn
rin
Figure 6. Study 3 structural model (model 4: path from schadenfreude – comment severity
released). Values to the left of the slash represent the standardized weights in the non-
stigmatizing comments condition. Values to the right of the slash represent standardized
ep
weights in the stigmatizing comments condition. ***Denotes that the path is significant at p
<.001. **Denotes that the path was significant at p <.01. Dashed lines represent non-
significant paths. N.B. Residual error terms for moral outrage and schadenfreude, and both
shaming variables, respectively, were allowed to covary.
Pr
33
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
In line with Study 1 and 2, Study 3 provides support for deservingness and
ed
schadenfreude as key mechanisms underlying the relationship between participants’ concerns
about social justice and their participation in online shaming. Justice concerns facilitated
schadenfreude via participants’ belief that the offender was deserving of negative
iew
consequences. Schadenfreude, in turn, was associated with an increase in participants’ overall
engagement with the news article, as well as the severity of their comments. In this study,
there was evidence that the relationship between schadenfreude and the severity of
v
participants’ comments was significant when participants viewed stigmatizing shaming
re
comments – and not significant when they viewed non-stigmatizing comments – however the
overall test of moderation was marginal (p = .06) and should thus be interpreted with caution.
er
General Discussion
networking sites has led to a revival in public shaming. In the current studies we examined
two pathways through which online shaming could occur: (1) a justice motive – whereby
ot
deservingness, and moral outrage, and/or (2) a hedonic motive, whereby justice concerns
turn, associated with schadenfreude when witnessing the offender receiving punishment
rin
Overall, three studies provide support for a primarily hedonic motive facilitating
ep
online shaming behaviour. Participants’ concerns about justice indirectly influenced shaming
by increasing their perception that the offender in the news article was deserving of negative
Pr
Studies 1 and 3, once the effects of schadenfreude were accounted for, justice concerns and
34
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
moral outrage had no independent positive effects on either behavioural shaming outcome. In
ed
Study 2, justice concerns had a directly negative effect on participants’ overall engagement in
shaming (a suppressor effect). In Study 2 there was evidence that moral outrage was weakly
positively associated with participants’ engagement in shaming – but not the severity of their
iew
shaming. However, moral outrage did not consistently emerge as a mediator in any of the
other studies.
v
shaming is consistent with recent work implicating perceptions of deservingness and
re
schadenfreude as important emotional and cognitive reactions to information encountered on
social media (Cecconi et al., 2020; Wei & Lu, 2020). Further, our results confirm previous
er
research emphasizing that while justice concerns are important in driving initial reactions
following real (or perceived) wrongdoing, people are also motivated by emotions that derive
pe
satisfaction from seeing offenders receiving costly punishment (Bernsden & Tiggemann,
2020; Greenier, 2021; Jensen, 2012; Wang et al., 2019). That is, concerns about enforcing
morally acceptable behaviour may only tell one side of the story when it comes to online
ot
shaming – our results suggest that people may also need to experience a sense of malicious
deservingness, and moral outrage, moral outrage was only weakly positively associated with
ep
shaming in Study 2 – however it did not influence the severity of participant’s shaming. In
Studies 1 and 3, moral outrage had no direct effects on shaming – nor did it act as a mediator
35
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
Although research has pointed to moral outrage as an important antecedent of digital
ed
expressions of shaming and punishment (Brady et al., 2021; Crockett, 2017), our results seem
to align with work examining the paradox of moral outrage in the Internet age. Sawaoka and
Monin (2018; see also Sawaoka & Monin, 2020) found that while isolated expressions of
iew
moral outrage are seen as morally justifiable, outrage shared exponentially online is often
viewed by bystanders as disproportionate and unfair. Indeed, moral outrage is an emotion that
centres around people’s perceptions of morality and arises when one believes another has
v
violated their understanding of “right” and “wrong” (Jimenez-Leal & Cortissoz-Mora, 2021).
re
In the collective action literature, moral outrage is consistently associated with action
designed to “right wrongs” – to reduce harm, injustice, and suffering (e.g., petitioning for the
er
rights of minority groups; Thomas et al., 2009; van Zomeren et al., 2004). Given this, an
individual who experiences outrage at breaches of moral standards may view the use of
pe
disproportionate punishment via viral shaming as a particularly unfair and immoral course of
action. In line with this possibility, recent work by Pundak et al. (2021) found that shaming
wrongdoers online was shaped by adherence to the nonmaleficence principle – a belief that
ot
one should avoid inflicting harm on others (Anderson et al., 2006). However, in some cases,
utilitarianism (i.e., when one believes shaming is key to preventing future harm and therefore
important for “the greater good”; Pundak et al., 2021). Accordingly, whether moral outrage is
rin
associated with online shaming may vary on the basis of people’s intuitive understanding of
public shaming as an effective means of achieving social justice – and lay conceptions of the
ep
36
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
in anonymity scores across condition, participants in the identifiable condition nevertheless
ed
reported feeling (in absolute terms) relatively anonymous. Indeed, it is likely that the online
nature of the current study contributed to participants feelings of anonymity. Thus, it would
be important for future work to provide a more detailed analysis of the role of anonymity on
iew
shaming behaviour in the context of studies that provide a more targeted manipulation of
anonymity. For example, manipulating “felt surveillance” by having participants believe their
comment will be viewed by a valued third-party (i.e., one’s employer) could test whether
v
anonymity is related to online shaming behaviour due to a perceived lack of consequences.
re
In Study 2 we found no evidence for the influence of social norms and Study 3 found
that norms had a marginal effect on only one path – the relationship between schadenfreude
er
and comment severity. Schadenfreude appeared to be a significant predictor of shaming
transgressor). Social norms had no influence on participants’ overall engagement with the
Despite these mixed effects, the marginal effect of social norms in Study 3 provides
us with some preliminary evidence regarding the role of social influence on shaming
tn
behaviour. In particular, it suggests that the responses of other users may be important in
shaping the nature of shaming participants engage in – specifically, the harshness of their
rin
response. This possibility is consistent with work outlining the power of social influence
online, and has important implications regarding our understanding of how more malicious
ep
forms of shaming can emerge on social media (Brady et al., 2020; Chia, 2020; Postmes &
Spears, 1998b; Postmes et al., 2000). Importantly, if we can understand the contextual factors
that aid in the promotion of stigmatizing shaming online, we can begin to develop possible
Pr
37
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
Limitations and Future Directions
ed
Our approach sought to maximise ecological validity by designing studies that
effectively manipulate) social norms as they operate in the real world – where there is the
iew
chance for norms to emerge and strengthen over time via people’s interaction with one
another. Indeed, in previous studies that have found support for the role of normative
influence in computer mediated communication (e.g., Spears, Lea & Lee, 1990), participants
v
were required to discuss issues via a synchronous chat program whereby norms were more
re
subtly delivered (and therefore able to emerge dynamically through participant interaction).
Thus, the strength of the evidence regarding social norms in the current studies may be due to
er
the relatively static way in which norms were manipulated – whereby participants were
shown a series of responses that they were led to believe were from “other users” and asked
pe
to respond. Accordingly, future work using paradigms more suited to capturing the dynamic
nature of social media could be used to assess how norms regarding the punishment of
Given the largely correlational nature of the present work we cannot make any claims
regarding the direction of effects or draw any conclusions regarding causation. One particular
tn
schadenfreude. Wang et al. (2019) propose that rather than deservingness as an antecedent to
rin
schadenfreude, people may experience schadenfreude first, before using claims about
experience of schadenfreude, and the experience of schadenfreude may, in turn, lead people
to endorse their belief that an offender is deserving of punishment more strongly. In the same
Pr
vein, the methods we employed here cannot disentangle whether schadenfreude precedes
38
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
shaming behaviour, or whether it is engagement in shaming itself that leads individuals to
ed
experience schadenfreude. Future research should therefore seek to explore the antecedents
and consequences of online shaming behaviour in a design that controls for participants’
engagement in shaming, therefore allowing for causal conclusions to be made regarding the
iew
order of effects.
We tested our theoretical model in the same context in all three studies (an instance of
Islamophobia in the West), therefore it may be that the effects we have observed here apply
v
only in this context. Accordingly, future research should seek to establish whether these
re
effects can be generalized across different kinds of transgressions that differ in their nature
and severity, as well as in diverse cultural contexts – given the use and the effects of public
er
shaming and shame differ across collectivist and individualist countries (Wong & Tsai,
2007). Further, future studies could explore whether online shaming is influenced by certain
pe
aspects of the transgressor. Feather and Sherman (2002) found evidence that the perceived
experienced greater schadenfreude when they witnessed a high achiever suffer compared to
ot
1991). Ronson (2015) notes that in the case of Justine Sacco – a PR manager who gained
tn
infamy in 2013 when one of her tweets went viral – public anger may have been fuelled by
her status as a privileged white New Yorker, but also as a woman. Indeed, research has found
rin
that women are more likely to be the targets of online shaming (MacPherson & Kerr, 2020;
Muir et al., 2021). Testing the influence of such moderators (i.e., social status, gender, race,
ep
39
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
increasing severity in participants’ shaming according to how they communicated their
ed
disapproval (i.e., whether the offender was considered a good person who did a bad thing, or
above, shaming can encompass a range of behaviours ranging in severity. Thus, examining
iew
when people will choose to leave a comment condemning a transgressor for their behaviour
versus when they will work with other online users to uncover personal information about the
target (i.e., when online actions move into offline harassment) is worthy of future
v
investigation. Exploring the link between “low level” actions such as “liking” content and
re
more hostile actions has important implications for our understanding of how online shaming
can spiral out of control – and may provide important insights into contextual factors that can
er
be used to create safer online environments.
Conclusion
pe
The results from three studies point to perceptions of deservingness and
schadenfreude as important predictors of online shaming. Given the exploratory nature of the
current work and the paucity of existing research on online shaming, many avenues exist for
ot
future research. Social psychology is well placed to understand both individual and group
processes that may influence shaming behaviour – in particular, how certain features of the
tn
online environment and aspects of the transgressor may interact to influence the nature and
consume content and connect with others, we are hopeful that future research stimulates a
more comprehensive understanding of the dynamics of online shaming and its consequences.
ep
Pr
40
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
Funding
ed
This research received no specific grant from any funding agency in the public, commercial,
or not-for-profit sectors.
iew
Declaration of interest
v
Research and publication ethics
re
This research project has been approved by [anonymized for peer review].
Data availability
er
The data that support the findings of this study are available at
pe
https://osf.io/nr69m/?view_only=466a79bf3d4346598f873736bf502a37
ot
tn
rin
ep
Pr
41
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
References
ed
Akfirat, S., Uysal, M. S., Bayrak, F., Ergiyen, T., Üzümçeker, E., Yurtbakan, T., & Özkan, Ö.
S. (2020). Social identification and collective action participation in the internet age:
iew
A meta-analysis. Journal of Community & Applied Social Psychology, 1-15.
doi.org/10.31234/osf.io/uqb7a
Anderson, M., Anderson, S. L., & Armen, C. (2006). MedEthEx: A prototype medical ethics
v
advisor. In Proceedings of the national conference on artificial intelligence (Vol. 21,
2nd ed., p. 1759). Menlo Park, CA; Cambridge, MA; London: AAAI Press; MIT
re
Press; 1999.
Barlett, C. P., Gentile, D. A., & Chew, C. (2016). Predicting cyberbullying from anonymity.
er
Psychology of Popular Media Culture, 5, 171-180.
https://doi.org/10.1037/ppm0000055
pe
Bastian, B., Denson, T. F., & Haslam, N. (2013). The roles of dehumanization and moral
doi.org/10.1371/journal.pone.0061842
ot
Batson, C. D., Kennedy, C. L., Nord, L. A., Stocks, E. L., Fleming, D. Y. A., Marzette, C.
tn
M., ... & Zerger, T. (2007). Anger at unfairness: Is it moral outrage?. European
Bentler, P. M. (2007). On tests and indices for evaluating structural models. Personality and
rin
Berndsen, M., & Tiggemann, M. (2020). Multiple versus single immoral acts: an immoral
ep
person evokes more schadenfreude than an immoral action. Motivation and Emotion,
44, 738-754.
Pr
42
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
Billingham, P., & Parr, T. (2020). Enforcing social norms: The morality of public
ed
shaming. European Journal of Philosophy, 28, 997-1016. doi.org/10.1111/ejop.12543
Brady, W. J., Crockett, M. J., & Van Bavel, J. J. (2020). The MAD model of moral
contagion: The role of motivation, attention, and design in the spread of moralized
iew
content online. Perspectives on Psychological Science, 15, 978-1010.
doi.org/10.1177/1745691620917336
Brady, W. J., McLoughlin, K., Doan, T. N., & Crockett, M. J. (2021). How social learning
v
amplifies moral outrage expression in online social networks. Science Advances, 7, 1-
re
14. doi.org/10.1126/sciadv.abe5641
Braithwaite, J. (2000). Shame and criminal justice. Canadian Journal of Criminology, 42,
pe
281-298.
Brickman, P., Folger, R., Goode, E., & Schul, Y. (1981). Microjustice and macrojustice.
In The justice motive in social behavior (pp. 173-202). Springer, Boston, MA.
ot
doi.org/10.1007/978-1-4899-0429-4_9
doi.org/10.1016/j.jesp.2005.06.007
rin
Carlsmith, K. M., Darley, J. M., & Robinson, P. H. (2002). Why do we punish? Deterrence
and just deserts as motives for punishment. Journal of Personality and Social
ep
Cecconi, C., Poggi, I., & D’Errico, F. (2020). Schadenfreude: Malicious Joy in Social Media
43
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
Chia, S. C. (2019). Crowd-sourcing justice: tracking a decade’s news coverage of cyber
ed
vigilantism throughout the Greater China region. Information, Communication &
iew
Internet social interactions: “On the Internet, Nobody Knows You’re a Dog”.
Crockett, M. J. (2017). Moral outrage in the digital age. Nature Human Behavior, 1, 769-771.
v
doi.org/10.1038/s41562-017-0213-3
re
Eckert, S., & Metzger‐Riftkin, J. (2020). Doxxing. The International Encyclopedia of
Feather, N. T., & Sherman, R. (2002). Envy, resentment, schadenfreude, and sympathy:
doi.org/10.1177/014616720202800708
ep
Fox, J., Cruz, C., & Lee, J. Y. (2015). Perpetuating online sexism offline: Anonymity,
interactivity, and the effects of sexist hashtags on social media. Computers in Human
Gaskin, J., & Lim, J. (2018). Indirect effects. AMOS plugin. Gaskination's StatWiki.
44
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
Goldenberg, A., & Gross, J. J. (2020). Digital emotion contagion. Trends in Cognitive
ed
Sciences, 24, 316-328. doi.org/10.1016/j.tics.2020.01.009
iew
Haidt, J. (2003). The moral emotions. In RJ Davidson, KR Scherer, & HH Goldsmith (Eds.),
University Press.
v
Heider, F. (1958). The psychology of interpersonal relations. New York: John Wiley.
re
doi.org/10.1037/10628-000
Jagayat, A., Boparai, G., Pun, C., & Choma, B.L. (2021). Mock Social Media Website
er
Tool (1.0) [Computer software]. https://docs.studysocial.media
Jane, E. A. (2016). Online misogyny and feminist digilantism. Continuum, 30, 284-297.
pe
doi.org/10.1080/10304312.2016.1166560
Klonick, K. (2015). Re-shaming the debate: Social norms, shame, and regulation in an
Langlois, G., & Slane, A. (2017). Economies of reputation: The case of revenge porn.
Pr
45
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
Communication and Critical/Cultural Studies, 14, 120-138.
ed
doi.org/10.1080/14791420.2016.1273534
Lapidot-Lefler, N., & Barak, A. (2012). Effects of anonymity, invisibility, and lack of eye-
iew
doi.org/10.1016/j.chb.2011.10.014
Lea, M., O'Shea, T., Fung, P., & Spears, R. (1992). 'Flaming' in computer-mediated
communication: Observations, explanations, implications. Harvester Wheatsheaf.
v
Leach, C. W., Spears, R., & Manstead, A. S. (2015). Parsing (malicious) pleasures:
re
Schadenfreude and gloating at others’ adversity. Frontiers in Psychology, 6, 1-13.
doi.org/10.3389/fpsyg.2015.00201
Lerner, M. J. (1981). The justice motive in human relations. In The justice motive in social
er
behavior (pp. 11-35). Springer, Boston, MA.
pe
doi.org/10.1080/14791420.2016.1273534
Loveluck, B. (2020). The many shades of digital vigilantism. A typology of online self-
MacPherson, E., & Kerr, G. (2020). Online public shaming of professional athletes: Gender
doi.org/10.1177/20563051211021378
Max, D. T. (2020, September 21). “The Public-Shaming Pandemic”. The New Yorker.
https://www.newyorker.com/magazine/2020/09/28/the-public-shaming-pandemic
ep
Montada, L., & Schneider, A. (1989). Justice and emotional reactions to the disadvantaged.
46
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
Moore, M. J., Nakano, T., Enomoto, A., & Suda, T. (2012). Anonymity and roles associated
ed
with aggressive posts in an online forum. Computers in Human Behavior, 28, 861-
867. doi.org/10.1016/j.chb.2011.12.005
Muir, S. R., Roberts, L. D., & Sheridan, L. P. (2021). The portrayal of online shaming in
iew
contemporary online news media: A media framing analysis. Computers in Human
O'Mara, E. M., Jackson, L. E., Batson, C. D., & Gaertner, L. (2011). Will moral outrage stand
v
up?: Distinguishing among emotional reactions to a moral violation. European
re
Journal of Social Psychology, 41, 173-179. doi.org/10.1002/ejsp.754
Owens, J. B. (1999). Have we no shame: Thoughts on shaming, white collar criminals, and
er
the federal sentencing guidelines. American University Law Review, 49, 1047-1058.
Postmes, T., & Spears, R. (1998a). Deindividuation and anti-normative behavior: A meta-
pe
analysis. Psychological Bulletin, 123, 238–259. doi:10.1037/0033-
2909.123.3.238
Postmes, T., Spears, R., & Lea, M. (1998b). Breaching or building social boundaries? SIDE-
ot
Postmes, T., Spears, R., & Lea, M. (2000). The formation of group norms in computer
doi.org/10.1111/j.1468-2958.2000.tb00761.x
Powell, C. A., & Smith, R. H. (2013). Schadenfreude caused by the exposure of hypocrisy in
ep
Preacher, K. J., & Hayes, A. F. (2008). Asymptotic and resampling strategies for assessing
47
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
Pundak, C., Steinhart, Y., & Goldenberg, J. (2021). Nonmaleficence in Shaming: The Ethical
ed
Dilemma Underlying Participation in Online Public Shaming. Journal of Consumer
Reicher, S. D., Spears, R., & Postmes, T. (1995). A social identity model of deindividuation
iew
phenomena. European Review of Social Psychology, 6, 161–198.
doi:10.1080/14792779443000049
Rodríguez-Gómez, P., Martín-Loeches, M., Colmenares, F., Ferreiro, M. V. R., & Moreno, E.
v
M. (2020). He had it Comin’: ERPs Reveal a Facilitation for the Processing of
re
Misfortunes to Antisocial Characters. Cognitive, Affective, & Behavioral
Sawaoka, T., & Monin, B. (2018). The paradox of viral outrage. Psychological Science,
pe
29, 1665-1678. doi.org/10.1177/0956797618780658
Sawaoka, T., & Monin, B. (2020). Outraged but Sympathetic: Ambivalent Emotions Limit
the Influence of Viral Outrage. Social Psychological and Personality Science, 11,
ot
499-512.
Schrobsdorff, S. (2015, June 25). “When parents publicly shame their kids”. TIME.
tn
https://time.com/3935308/when-parents-publicly-shame-their-kids/
Smith, L. G., & Postmes, T. (2009). Intra‐group interaction and the development of norms
rin
which promote inter‐group hostility. European Journal of Social Psychology, 39, 130-
144. doi.org/10.1002/ejsp.464
ep
Smith, L. G., & Postmes, T. (2011). The power of talk: Developing discriminatory group
doi.org/10.1348/014466610X504805
Pr
48
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
Spears, R., Lea, M., & Lee, S. (1990). De‐individuation and group polarization in
ed
computer‐mediated communication. British Journal of Social Psychology, 29, 121-
Spring, V. L., Cameron, C. D., & Cikara, M. (2019). Asking different questions about
iew
outrage: A reply to Brady and Crockett. Trends in Cognitive Sciences, 23, 80-82.
doi.org/10.1016/j.tics.2018.11.006
Tait, A. (2020, April 4). “Pandemic shaming: is it helping us keep our distance?”. The
v
Guardian. https://www.theguardian.com/science/2020/apr/04/pandemic-shaming-is-it-
re
helping-us-keep-our-distance
Thomas, E. F., McGarty, C., & Mavor, K. I. (2009). Aligning identities, emotions, and beliefs
er
to create commitment to sustainable social and political action. Personality and Social
doi.org/10.1177/1368430215612217
Thomas, E. F., McGarty, C., Spears, R., Livingstone, A. G., Platow, M. J., Lala, G., &
tn
Mavor, K. (2020). ‘That's not funny!’ Standing up against disparaging humor. Journal
Thomas, E. F., Duncan, L., McGarty, C., Louis, W. R., & Smith, L. G. (2022). MOBILISE:
Thompson, J. D., & Cover, R. (2021). Digital hostility, internet pile-ons and shaming: A case
49
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
pleasure at the misfortune of others. Cambridge University Press.
ed
doi.org/10.1017/CBO9781139084246
Wang, S., Lilienfeld, S. O., & Rochat, P. (2019). Schadenfreude deconstructed and
iew
doi.org/10.1016/j.newideapsych.2018.09.002
Wei, L., & Liu, B. (2020). Reactions to others’ misfortune on social media: Effects of
v
deservingness. Computers in Human Behavior, 102, 1-13.
re
doi.org/10.1016/j.chb.2019.08.002
Wenzel, M., Okimoto, T. G., Feather, N. T., & Platow, M. J. (2008). Retributive and
er
restorative justice. Law and Human Behavior, 32, 375-389. doi.org/10.1007/s10979-
007-9116-6
pe
Wenzel, M., & Okimoto, T. G. (2010). How acts of forgiveness restore a sense of justice:
Wilton, C., & Campbell, M. (2011). An exploration of the reasons why adolescents engage in
traditional and cyber bullying. Journal of Educational Sciences and Psychology, 1(2),
tn
101-109.
Wong, Y., & Tsai, J. (2007). Cultural models of shame and guilt. In The self-conscious
rin
emotions: Theory and research (Eds.). Tracy, J. L., Robins, R. W., & Tangney, J. P.
Guilford Press.
ep
Yuan, K. H., & Bentler, P. M. (2001). A unified approach to multigroup structural equation
doi.org/10.4324/9781410601858-7
Pr
50
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608
Appendix A
Coding Guide for Qualitative Comments
ed
1 - No comment N/A
iew
2 – Benign (Non-shaming) - No condemnation
- No targeting of behaviour or person
e.g.
• omg
• this is sad
v
• how upsetting
re
3 – Shaming features - Features of shaming present
- Expresses disapproval
- Targets behaviour
- Condemns the behaviour
e.g.
er
• Sickening behaviour
• what a terrible thing to do
pe
4 – Stigmatizing shaming - Targets the person
- Condemns the person
- Labels the person as evil
- Undermines the person’s character
- Disrespectful of the person (insults, verbal
ot
abuse, name-calling).
e.g.
• what a disgusting person
tn
51
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4359608