612217

research-article2015
GPI0010.1177/1368430215612217Group Processes & Intergroup RelationsThomas et al.

G
Group Processes & P
Intergroup Relations I
Article R

Group Processes & Intergroup Relations

Group interaction as the crucible of 2016, Vol. 19(2) 137­–151

© The Author(s) 2015

social identity formation: A glimpse at

Reprints and permissions:
sagepub.co.uk/journalsPermissions.nav
DOI: 10.1177/1368430215612217
the foundations of social identities for gpir.sagepub.com

collective action

Emma F. Thomas,1 Craig McGarty,2 and Ken Mavor3

Abstract
Many of the world’s biggest problems are being tackled through the formation of new groups yet
very little research has directly observed the processes by which new groups form to respond to social
problems. The current paper draws on seminal research by Lewin (1947) to advance a perspective as to
how such identities form through processes of small group interaction. Multilevel structural equation
modelling involving 58 small group discussions (with N = 234) demonstrates that focused group
discussion can boost the commitment to take collective action, beliefs in the efficacy of that action,
and members’ social identification with other supporters of the cause. The results are consistent with
the new commitment to action flowing from emergent social identities.

Keywords
collective action, small group interaction, social identity formation
Paper received 19 November 2013; revised version accepted 17 August 2015.

In order to effect large scale social change toward the future with social identities
we clearly need to do more than focus that have the capacity to engender and
individual minds on the future. Instead, sustain positive action.
we need large sections of the public to
embrace sustainable solutions collectively.
To promote collective action of this kind, (Postmes, Rabinovich, Morton, &
we need to fuse individual orientations van Zomeren, 2012, p. 194)

1Murdoch University, Australia

2The University of Western Sydney, Australia
3The University of St Andrews, UK

Corresponding author:
Emma F. Thomas, School of Psychology & Exercise
Science, Murdoch University, South Street, Perth, WA 6150,
Australia.
Email: emma.thomas@murdoch.edu.au
138 Group Processes & Intergroup Relations 19(2)

The field of social psychology is filled with The current research draws on the two prongs
research on attitude change, prejudice, of the Lewinian legacy by extending an analysis
self-concept formation, stereotyping, and of group interaction as it relates to social identity
emotions, most of which has ignored how formation and collective action. We suggest that
these processes operate in small groups. This an important part of social psychology’s ability
neglect means that many questions central to to contribute solutions to contemporary global
the human social experience have not been problems hinges on its ability to explain the ori-
investigated. gins or foundations of psychological group for-
mation. This is because many of the most
significant problems that confront us today are
(Wittenbaum & Moreland, 2008, p. 188) problems that require collective solutions (e.g.,
Jonas & Morton, 2012; Louis, 2009; Postmes
In 1947 Kurt Lewin, one of the founders of et al., 2010). Intriguingly, Lewin’s two legacies—
modern social psychology, published findings of group decision through social interaction and
research into what he termed “group decision.” action research—may, in combination, provide
Charged with the task of finding ways to change some significant insights as to how to generate
attitudes towards serving offal as a family meal as collective identities that promote sustainable
part of the World War II effort, Lewin compared action. The current analysis explores this possi-
the effects of listening to a lecture, with the bility by drawing on recent developments in the
effects of group discussion on samples of women social psychology of social identity formation
who made food purchasing decisions for their (Postmes, Haslam, & Swaab, 2005) and collective
families. The findings were striking. Of the action (Thomas, Mavor, & McGarty, 2012;
women who had listened to the lecture, only 3% Thomas, McGarty, & Mavor, 2009; van Zomeren,
went on to prepare offal. Conversely, of those Postmes, & Spears, 2008) to specify the role of
who had participated in a group discussion (about group interaction in promoting the development
difficulties facing women in serving offal), 32% of new social identities that are built for action.
subsequently served offal to their families. Lewin
(1958, p. 202) concluded that “discussion, if con- Group Interaction and Social
ducted correctly, is likely to lead to a much higher
degree of involvement.”
Identity Formation
The legacies of this study were twofold. Lewin Despite widespread recognition of the importance
was the first to demonstrate the powerful effects of Lewin’s results, research involving group inter-
of group interaction on changing attitudes and action has been relatively rare in social psychology
behaviour, findings that were mirrored some in the intervening decades. Several reviews of the
years later with the advent of the group polariza- trends in research in social psychology have docu-
tion phenomenon (Moscovici & Zavalloni, 1969; mented marked declines in studies involving social
Myers & Bishop, 1970). Lewin (1947, p. 150) was interaction (Haslam & McGarty, 2001; Moreland,
also the founder of action research, declaring Hogg, & Hains, 1994; Wittenbaum & Moreland,
“research that produces nothing but books will 2008). Haslam and McGarty (2001) account for
not suffice.” This point is mirrored in contempo- this trend as evidence of a focus on uncertainty
rary psychological debate where there is an reduction (and concomitant low tolerance for sta-
emerging consensus that (social) psychology can tistically and methodologically “messy” data) at the
and should do more to tackle issues of global cost of a truly interactionist social psychology
importance (e.g., human induced climate change, (Turner & Oakes, 1986; see also Wittenbaum &
global poverty; see, e.g., Bazerman & Malhotra, Moreland, 2008). It is also the case that research on
2006; Leidner, Tropp, & Lickel, 2013). intragroup processes and intergroup relations have
Thomas et al. 139

tended to occupy separate conceptual and empiri- Elsewhere, Smith and Postmes (2011b) showed
cal trajectories (see Dovidio, 2013, for a review; that group discussion about a negative social ste-
Hogg, Abrams, Otten, & Hinkle, 2004). Whereas reotype (“women are bad at maths”) acted as a
the small group literature has focused on group buffer against stereotype threat.
performance, cohesion, and performance (e.g., In an important extension of this argument,
Levine & Moreland, 2006), the intergroup rela- McGarty, Bliuc, Thomas, and Bongiorno (2009)
tions literature has tended to focus on social iden- have proposed that small group interaction
tity, conflict, and competition (Dovidio, 2013). The provides a vehicle for the intensification of opin-
dearth of research into social interaction within ion-based group memberships (groups based on
the intergroup domain has arguably left some of shared opinions; Bliuc, McGarty, Reynolds, &
Lewin’s most important insights to wither on the Muntele, 2007). That is, targeted small group
vine. Indeed, one recent review identifies group discussion provides a site where people can
discussion as one of the most significant drivers of negotiate and validate a position about “how we
cooperative behaviour but notes a considerable want the world to be” (Gee, Khalaf, & McGarty,
lack of insight as to why this is the case (Meleady, 2007; Thomas & McGarty, 2009). Intensifying
Hopthrow, & Crisp, 2013; see also Hopthrow & opinion-based group memberships specifically
Abrams, 2010). Postmes, Spears, Lee, and Novak (rather than other broader social identities that
(2005) identify a similar lack of theoretical can be contested; Sani & Reicher, 2000) seems
resolution in relation to the group polarization particularly important since these identities can
phenomenon. be implicitly aspirational or future-oriented
A small number of recent papers have, how- (Rabinovich, Morton, & Postmes, 2010); they
ever, made progress in identifying the mechanisms are geared towards establishing (or preventing)
that underpin the effects of small group interac- a particular state of affairs (see Smith, Thomas,
tion as they relate to social identity formation, & McGarty, 2015). If our goal is to develop
social influence, and social action (Abrams, identities that engender action in the present
Wetherell, Cochrane, Hogg, & Turner, 1990). An but are sustainable into the future then opinion-
important contribution in this regard is Postmes based groups may be fruitful candidates (see
et al.’s (2005; also Postmes, Spears, et al., 2005) Thomas et al., 2009, for a detailed consideration
interactive model of identity formation, where it is of this point). Consistent with these arguments,
argued that inductive social identity formation occurs Thomas and McGarty (2009) found that group
through small group interaction. Inductive identi- interaction boosted social identification with an
ties form through communication, consensualiza- antipoverty opinion-based group and increased
tion, and negotiation about what it means to be a commitment to act in line with global poverty
group member and can stem from recognition of reduction efforts.
shared cognitions (thoughts and beliefs) amongst It also seems plausible that forming a novel
people (Swaab, Postmes, van Beest, & Spears, identity through small group interaction will
2007) or from a discussion of individual differ- affect some components of social identifica-
ences (Jans, Postmes, & van der Zee, 2011). tion more than others. Social identification is
Consistent with this account, social interaction has a multidimensional construct (see Ashmore,
been implicated in the expression of intergroup Deaux, & McLaughlin-Volpe, 2004, for a
hostility (Smith & Postmes, 2009, 2011a). Other review) where these identity dimensions may
research shows that group interaction is central to relate to connection, satisfaction, and self-defi-
the validation of social stereotypes (Haslam et al., nition (Cameron, 2004; Leach et al., 2008). In
1998; Stott & Drury, 2004), and can act to politi- the context of the current research, it might
cize and (under particular circumstances) radicalize take time (weeks, months, or years) for a group
sympathizers (Thomas, McGarty, & Louis, 2014). membership to become central to
140 Group Processes & Intergroup Relations 19(2)

self-perception. It might take the experience of efficacy that precede group formation.”
actual collective success or failure to influence Accordingly, EMSICA anticipates direct effects
group affect (e.g., Tausch & Becker, 2013). On of social identification on collective action but
the other hand, during the early stages of social identification also captures (and attenuates)
group formation, a bolstered sense of connect- the indirect effects of reactions to injustice (anger
edness to, or solidarity with, other group mem- or outrage) and efficacy. Put differently, the sub-
bers should arise from interaction (e.g., Jans jective feelings and beliefs associated with group
et al., 2011). Accordingly, the current paper membership become bound up in the social iden-
treats social identification as a multidimen- tity, rendering the direct links less important
sional construct and explores the role of the (Thomas et al., 2012; Thomas et al., 2015).
different identity components in the early Whereas SIMCA may be more relevant to identi-
stages of psychological group formation. ties that already exist (e.g., national, gender, or ethnic
categories), EMSICA seems particularly relevant to
understanding the interactive formation of emerg-
Collective Action for Social ing social identities that grow around the need to
Change take action. This is because social identity formation
Social psychology has also made recent progress is premised in knowledge about similar others’
in addressing another question close to Lewin’s views and judgements (e.g., Swaab et al., 2007). That
heart: how to motivate action for social change. is, it is difficult for people to form groups based on
The social identity model of collective action (SIMCA; opinions about how the world should be without
van Zomeren et al., 2008) suggests that people knowing what other people’s opinions are (Hardin
will engage in group-level efforts to bring about, & Higgins, 1996). If this is the case it seems likely
or subvert, social change when they believe that that discussions that allow people to express sup-
co-ordinated group efforts can be successful port for a cause and plan ways to achieve it should
(group efficacy; Bandura, 1997, 2000), experience contribute to the foundation for identity formation
motivating emotional reactions to the injustice rather than be supplementary to it (see also van
(anger or outrage; see Thomas et al., 2012; Zomeren, Spears, Fischer, & Leach, 2004). We
Thomas et al., 2015), and identify with relevant therefore expect that the encapsulated model of
groups that can mobilize action (social identity; collective action will be highly applicable to induc-
following Reicher, 1984, 1996). Indeed, van tively forming identities in interacting groups.
Zomeren et al. (2008) posit a particularly impor-
tant role for the social identity pathway, whereby
The Current Research
social identification has direct effects on collec-
tive action, as well as indirect effects on action The current research involves a structural analysis
through relevant reactions to injustice (e.g., anger of the outcomes of 58 small group discussions.
or outrage) and beliefs (group efficacy). Existing research identifies group interaction as a
More recently still, Thomas and colleagues vehicle for social identity formation (Postmes,
(Thomas et al., 2012; Thomas et al., 2009) pro- Haslam, et al., 2005) and opinion-based group iden-
posed the encapsulation model of social identity in collec- tification in particular (Thomas & McGarty, 2009).
tive action (EMSICA). EMSICA anticipates the The collective action literature, on the other hand,
same pathways as SIMCA (injustice, efficacy, and articulates a prominent role for social identification
identity) but proposes that new identities can in promoting collective action (van Zomeren et al.,
emerge out of shared reactions (injustice and 2008). This research integrates recent advances in
efficacy) to a state of affairs. Thomas et al., (2009, social identity formation and collective action to
p. 206) argued that: “in everyday social interaction describe the formation of action-oriented social
we might expect that in some situations it will be identities. In doing so, it answers recent calls to
the strong affective reactions and feelings of explicitly bridge research on intergroup relations
Thomas et al. 141

(collective action) with intragroup processes (small Procedure

group interaction; see Dovidio, 2013). More specifi-
cally, the current research goes beyond the existing The procedure was similar across the three different
literature in three key ways. samples and the core factor is whether participants
First, we use multilevel structural equation engaged in interaction, or not (see also Thomas &
modelling to test the relative applicability of McGarty, 2009). All subjects read an information
EMSICA and SIMCA in the context of small sheet about the state of waterborne disease in devel-
groups engaged in social interaction. As antici- oping nations, the role of governments, and the
pated in the previous lines, we expect that EMSICA “Water for Life” movement (a United Nations
will better fit the data than SIMCA because it is movement which seeks to achieve clean water for
more readily applicable to new, interactively form- people in developing countries). At the end of this
ing groups. Second, we test social identification as information participants were asked to tick a box
a multidimensional construct in social identity indicating whether they supported (or did not sup-
formation. To date, the multidimensionality of port) the “Water for Life” movement. All partici-
identity has been underexplored in relation to the pants indicated their support.
effects of group interaction and collective action Participants were then given the information
more generally (see Cameron & Nickerson, 2009; that, the university was interested in the “Water for
Giguere & Lalonde, 2010, for exceptions). Life” campaign, and that their task was to come up
However, recent developments in the intergroup with strategies to help promote the “Water for
literature increasingly recognize that different Life” message.1 All participants were given the
aspects of identification can promote distinct written information that their strategies would be
(positive and negative) forms of intergroup written up and sent to various advocacy groups,
behaviour (e.g., Amiot & Aubin, 2012; Roccas, and posted on a website linked to the university.
Klar, & Liviatan, 2006). Accordingly, the current Participants were provided with a sheet of paper
research considers social identification at the com- upon which to write their recommendations.
ponent level to explore whether group interaction Participants in the group interaction conditions
exerts effects on all dimensions of identification. were formed into small groups of three to five
Finally, while our core aim is to test the applicabil- people and were left to engage in the group discus-
ity of SIMCA and EMSICA within interacting sion for half an hour. At the end of this period all
groups, we also seek to establish whether group participants completed a questionnaire. Participants
interaction affects identification, outrage, efficacy, assigned to a control condition completed the
and commitment to action compared to a relevant questionnaire without group interaction (i.e.,
(noninteraction) control. immediately after reading the information about
the “Water for Life” program; Study 1) or com-
pleted a comparable brainstorming task individu-
Method ally without group interaction (Studies 2 and 3).
Participants
Participants across three samples (N = 120;
Questionnaire
N = 103; N = 186 respectively) were primarily first Across all studies the questionnaire was titled
year psychology students at an Australian univer- “Attitudes towards support for people in develop-
sity who participated for course credit or people ing nations.” Measures were taken of social iden-
recruited on campus in exchange for A$10.00. tification, outrage, group efficacy, and collective
Altogether 409 people participated, of whom 256 action intention. Next, we detail the measurement
were female and 126 were male; there was no gen- strategy for each construct. All items were meas-
der recorded for 27 participants. Participants’ ured on an 11-point Likert-type scale anchored 1
median age was between 18 and 25; 90% of people (strongly disagree) to 11 (strongly agree) and were inter-
for whom age was recorded fell into this category. nally consistent.
142 Group Processes & Intergroup Relations 19(2)

Social identification. Identification with the pro- Results

“Water for Life” group was measured using Cam-
eron’s (2004) three-factor measure of social Intensification of Identity and Action
identification. Examples of these measures are: Through Interaction
“Being a supporter of programs such as ‘Water
Our main focus is on the processes underlying
for Life’ is an important reflection of who I am”
commitment to change but it is also important to
(centrality); “I have a lot in common with other
establish whether group interaction itself pro-
supporters of programs such as ‘Water for Life’”
duced increased identification, emotion, efficacy
(ingroup ties); “In general, I’m glad to be a sup-
beliefs, and commitment. The key data involve
porter of programs such as ‘Water for Life’”
observations that are nonindependent (due to
(ingroup affect). The three subcomponents were
group interaction) but we also had control condi-
internally consistent: centrality (α = .65–.79),
tion observations where people completed the
ingroup ties (α = .65–.85), and ingroup affect
same measures without group interaction. In
(α = .52–.76).
order to make the comparison between those who
had interacted versus those who had not we ana-
Outrage. Participants responded to four items
lysed the group data but used the control (col-
which read: “Thinking about the situation of peo-
lected at the same time from participants randomly
ple in developing countries, I feel angry/irritated/
allocated to the control condition) to establish a
outraged/livid.” Responses to these adjectives
baseline for each study. To prepare the group-level
were aggregated to form a scale (α = .84–.90).
data, we first subtracted the control group mean
for that sample from each interacting group mean.
Group efficacy. Group efficacy was measured with
The subtraction of the control group mean sim-
three items adapted from van Zomeren et al.
ply allows us to test the mean level in the interact-
(2004): “I feel that together the ‘Water for Life’
ing groups with the mean level in the control
program will be able to improve the water situa-
condition (analogous to using a t test to compare
tion in developing nations”; “The ‘Water for Life’
a mean to a scale midpoint). This strategy yields a
campaign will be a waste of time, effort and
deviation score expressing the strength of the
money” (reverse scored); and “The ‘Water for
effect in that group, relative to our best point esti-
Life’ campaign will be successful in its aims to
mate of the population level of the variable in the
provide safe and clean drinking water for those in
absence of group interaction (and taking account
developing countries.” These were combined to
of random variation across studies). There were
form a scale (α = .74–.89).
N = 58 groups based on 234 responses. As the
data were also clustered at the study level (i.e., they
Collective action intentions. Commitment to under- are comprised of three samples collected at differ-
take action to reduce the disadvantage of people ent points in time) all the following analyses also
in developing countries was measured by a series control for the effect of study.
of items designed to represent an escalating Table 1 shows the means and standard devia-
degree of commitment to the cause. An example tions for the group and control condition for
is: “I intend to support the ‘Water for Life’ move- each of the three samples, as well as the pooled
ment by attending a rally which calls for greater mean. Table 1 also displays a 95% confidence
government support for the initiative.” Other interval for the deviation from the control condi-
items were similarly worded and regarded signing tion is shown around the pooled mean (obtained
a petition, donating money, talking to others, by using the test of the intercept in the null
attending a rally, writing a letter to a member of model). In the absence of a direct multilevel
parliament, joining an advocacy group, and equivalent of between-groups ANOVA this
attending and organizing a “Water for Life” fun- allows comparisons of effects between the three
draising event (α = .87–.91). samples but without inflating the Type 1 error
Thomas et al. 143

rate by making multiple (post hoc) pairwise com-

0.06 [−0.67, 0.78]

7.11 (2.22) −0.23 (2.22) 7.20 (1.97) 6.89 (1.97) −0.31 (1.97) −0.33 [−0.66, 0.00]
0.64 [0.38, 0.80]
0.79 [0.46, 1.12]

0.29 [0.02, 0.56]

0.58 [0.27, 0.89]
parisons. As would be expected from repeated

Mean [95% CI]

sampling of effects of small to moderate size
with moderate sample sizes there is some varia-
Overall

bility in the significance levels of effects across

studies. It suggests that the strongest effects were
in Study 1 and weak or no effects in Study 3.
0.19 (1.69)
0.03 (2.11)
0.09 (1.13)

0.24 (1.43) 8.27 (1.30) 7.84 (1.50) −0.43 (1.50)

0.25 (1.74) 7.19 (1.95) 6.99 (1.98) −0.20 (1.98)
Analysis using MPlus (Muthén & Muthén,
2007) to control for nonindependence of observa-
Diff

tions revealed that the mean levels differed from

the control means for action intentions, efficacy,
0.68 (1.79) 6.06 (1.91) 6.25 (1.69)
0.52 (1.84) 5.00 (2.03) 5.03 (2.11)
8.39 (1.51) −0.36 (1.51) 8.51 (1.59) 8.60 (1.13)

and the ties and centrality components of identifi-

Discuss

cation, but not for the affective component of

identification or outrage (though the negative
70
18

effect was marginal). This demonstrates that group

interaction produced an intensified commitment
Control
Study 3

to action intentions and bolstered two of the

115

hypothesized drivers of action (efficacy, and two

of the three components of social identification).
Diff

EMSICA and SIMCA Models

Our next analyses utilized the deviation scores to
6.19 (1.79)
4.92 (1.84)

7.95 (1.43)
7.19 (1.74)

test the relationships amongst variables. Our goal

Discuss

was to compare the two theoretical models of

75
16

identity and collective action in predicting varia-

Table 1. Means (standard deviations) and cell sizes for measured variables.

tion in support for collective action amongst par-

0.84 (1.69) 5.51 (0.94)
1.53 (2.10) 4.40 (1.63)
0.37 (1.56) 8.75 (1.04)
6.76 (2.37) −0.42 (2.37) 7.35 (1.67)
0.85 (1.59) 7.71 (1.57)
1.39 (1.88) 6.94 (1.34)
Control
Study 2

ticipants in the interacting groups. By using the

deviation scores we can focus our attention on
28

the effects in the interacting groups only and con-

trol for sample differences by anchoring them
relative to the noninteraction controls in each
study. Correlations between variables across the
Diff

three samples are displayed in Table 2. To test our

core hypotheses we conducted path analysis using
6.18 (1.69)
4.95 (2.10)
8.51 (1.56)

7.97 (1.59)
7.35 (1.88)
Discuss

MPlus (Muthén & Muthén, 2007). The path

model was specified at the between-person level
98
24

of analysis but was “empty” at the between-

5.34 (1.49)
3.42 (1.33)
8.15 (1.40)
7.18 (1.97)
7.12 (1.67)
5.96 (1.71)

group and study levels of analysis. Since the

Control
Study 1

model considered only between-person relation-

ships the SRMR(within) is the most relevant
22

index to judge fit as other traditionally reported

Identity centrality

Action intentions

indices (CFI, RMSEA) are not reliable for multi-

Group efficacy
Identity affect

level models (Muthén & Muthén, 2007). A value

# of groups
Identity ties

of zero indicates perfect fit for SRMR(within)

Outrage

and a value of less than .08 is generally consid-

ered good fit (Hu & Bentler, 1999).
N
144 Group Processes & Intergroup Relations 19(2)

Table 2. Correlations between variables of interest.

Social Social identity Social identity Outrage Group Action

identity ties centrality affect efficacy intention
Identity ties 1 .48 .42 .32 .54 .62
Identity centrality 1 .29 .19 .44 .61
Identity affect 1 .42 .43 .50
Outrage 1 .36 .46
Group efficacy 1 .67
Action intentions 1

Note: As the results are from nonindependent data the conventional significance levels of the correlations have not been
provided in this table but see Figures 1 and 2 for the results of multilevel modelling.

Figure 1. Standardized weights obtained for the encapsulated model of social identity in collective action
(EMSICA).
Note. * p < .05. ***p < .001. Dotted pathways represent those that are nonsignificant. SRMR (within) = .06.

We first tested the EMSICA model whereby effect of identification through these variables).
social identification is a direct predictor of collec- This model also fitted the data well (SRMR within
tive action and fully mediates (encapsulates) the = .057); the standardized regression coefficients
effects of outrage and efficacy on action. This (γ) are displayed in Figure 2. It can be seen that
model fitted the data well (SRMR within = .06). there were large effects of social identification on
Figure 1 displays the standardized regression outrage, efficacy, and action; however, the indirect
coefficients (γ) for the multilevel model. It can be effects of identification on action through outrage
seen that all the paths were significant at p < .001. and efficacy were small and nonsignificant
Tests of the indirect effects of efficacy on action (ps > .17). Given the nonsignificance of the direct
through identification (IE = .57, SE = 0.04, paths between outrage–action and efficacy–action,
p < .001), and outrage on action through identifi- the model implies that correlations between effi-
cation (IE = .25, SE = 0.05, p < .001), were both cacy/outrage and action intentions are subsumed
significant. within their shared variance with social identifica-
We next tested SIMCA. The model included tion. This is consistent with the idea that social
social identification, outrage, and efficacy as direct identity encapsulates the effects of efficacy and
predictors of collective action; there were also outrage. Altogether, the results support the idea
direct links from identification to outrage and that social identification is playing an encapsula-
identification to efficacy to capture the facilitating tion role rather than a facilitating role in the inter-
effect of identification (and the potential mediated active formation of social identities.
Thomas et al. 145

Figure 2. Standardized weights obtained for the social identity model of collective action (SIMCA).
Note. *** p < .001. Dotted pathways represent those that are nonsignificant. SRMR (within) = .057.

Components of Social Identification encapsulated model of social identity in collective

action, the subjective feelings of being connected
There are two important results pertaining to the to, and belonging in, a group, acted as the con-
identity components: intensification of the com- ceptual and psychological link between group-
ponents due to interaction; and implications for based action-relevant emotions (outrage), beliefs
the strength of the components in the mediation (group efficacy beliefs), and actions (collective
relationship. Table 1 shows that social interaction action intentions). The causal ordering antici-
intensified commitment to ingroup ties (a psy- pated by the social identity model of collective
chological sense of connection to group mem- action also fit these data well, however the
bership) and centrality (centrality of the group absence of significant paths between outrage and
membership to self-definition), while ingroup action, and efficacy and action, suggests that
affect (feelings of satisfaction associated with identification is playing an encapsulation role in
group membership) was not bolstered in these these data. In other words, it seems that outrage
data. Given that affect may nevertheless play a and efficacy overlap with social identification and
predictive role (even if it was not boosted at the that identification is effectively capturing the vari-
mean level) we utilized a latent model with all ability associated with those reactions. Whereas
three components in the tests of SIMCA and SIMCA may be more relevant to identities that
EMSICA.2 Figures 1 and 2 show that a latent fac- already hold subjective meaning for participants
tor representation of identification fitted well, (e.g., social categories such as race, gender; mem-
though the nonsignificant loading suggests that bership of social movement organizations; see
the shared aspects of centrality and ties that pre- van Zomeren et al., 2008, for a meta-analysis), the
dict action are not shared with affect. encapsulating function of identity anticipated by
EMSICA may be more readily applicable to new,
Discussion interactively formed groups. In what follows we
The current research sought to develop an analy- consider the implications of current findings for
sis of the interactive origins of social identity in the literatures on group interaction and collective
collective action. Multigroup structural equation action more generally.
modelling involving 58 group discussions showed
that participating in a small group interaction bol- Social Interaction, Social Identity Formation,
sters two of the three dimensions of social iden-
and Sustainable Collective Action
tification (ties and centrality but not affect) and
also perceived efficacy and collective action inten- Lewin’s famous group decision study established
tions but not outrage. Moreover, in line with the the important effect of informal group
146 Group Processes & Intergroup Relations 19(2)

discussion. However, subsequent declines in the most relevant to social identity formation through
experimental study of social interaction and the interaction. We found that group interaction
separate trajectories of research on intragroup effectively bolstered two of the three dimensions
and intergroup processes ensured that the poten- of identification (effects on ingroup ties and
tial for group interaction to shed light on inter- centrality were greater than those for ingroup
group phenomena like collective action has affect) but that a single factor solution neverthe-
remained relatively untapped. Haslam and less fitted the data well. However, consistent
McGarty (2001; also Wittenbaum & Moreland, with arguments by Jans et al. (2011, p. 1133) who
2008) note that one of the most significant casu- suggested that a sense of connection to the
alties of the decline in research that involves group speaks to a more “organic process” of
group interaction is that it limits the study of inductive identity formation, it seems that devel-
consensus, a topic that should be of central con- oping a sense of psychological connection with
cern to social psychologists. This seems particu- group members (ingroup ties) may play a more
larly true of collective action, where group-based prominent role in capturing effects in the early
efforts to address inequality are central; knowing stages of group formation. This is not surprising
what relevant others think and intend to do seems as the ingroup ties component (in the Leach
fundamental to group formation (Klandermans, et al.’s [2008] typology named “solidarity”) may
1997; van Zomeren et al., 2004). be especially easy to bolster in small group inter-
Indeed, the little available research on small actions between strangers and this factor may
group interaction in intergroup contexts high- therefore be an especially good platform for
lights that people, tested in isolation (compared to action.
in interacting groups), will come to often mark- Finally, it is worth considering one alternative
edly different conclusions about social behaviour construal of these findings: that the results show
(Haslam et al., 1998; Smith & Postmes, 2009, nothing more than group polarization (i.e., the
2011a, 2011b; Smith et al., 2015; Stott & Drury, tendency for group discussion to polarize atti-
2004; Thomas & McGarty, 2009; Thomas et al., tudes in the direction in which they were already
2014). It is reassuring, then, that the current tending; Moscovici & Zavalloni, 1969). We agree
research supports the causal pattern of interrela- that the processes in operation here are closely
tions suggested by SIMCA and EMSICA (models related to what has been termed group polariza-
both derived in the absence of interaction; see tion. However, it is also the case that group polar-
Thomas et al., 2012, for a test of both models). ization is a more complex and sophisticated
We suggest that there is a distinction to be made phenomenon than it is commonly understood to
here between structural models, those models that be. Notwithstanding the conclusions of Isenberg
describe a set of predictors or correlates (in this (1986), there remains significant uncertainty
case, with collective action); and transformational about what produces the effects of group discus-
models, those that describe the processes and sion on attitude polarization (Postmes, Spears,
mechanisms associated with change (in this case, et al., 2005) and cooperation more generally
the processes through which identities are formed (Meleady et al., 2013). Indeed, the current data go
and transformed). This is akin to the distinction well beyond existing accounts of polarization by
between social identity as a fixed construct; and offering a substantiated account of process in
social identification as a process (Drury & Reicher, terms of the development of identity (Postmes,
2000, 2005). The current research sits between Haslam, et al., 2005; Thomas et al., 2009) and
these two extremes by identifying the dynamic demonstrate effects not only on the matters
and interactive ways in which people construe under discussion (as in standard group polariza-
their sense of self through social interaction. tion studies) but on a subset of other measures
Moreover, the current research explores the that were not discussed by the groups (including
specific dimensions of social identity that are one’s sense of self).
Thomas et al. 147

Limitations and Future Directions movement) is ill-defined for participants. If this

is the case, then it may be that some of the paths
One avenue for future research regards the (mar- observed here are more unstable at other time
ginal) negative effect of outrage in the group points (e.g., as participants become more familiar
interaction conditions: that is, while outrage was a with normative expectations for supporters). It is
strong qualitative predictor of identity, beliefs, also unclear whether the specific pattern of find-
and commitment to action (Table 2), group inter- ings observed here will generalize to other issues:
action may reduce felt outrage at the mean level Are opinion-based identities distinguishable from
(Table 1). Why might group discussion attenuate a broader ideological orientation (see, e.g., Bliuc
feelings of outrage? One possibility is that, hav- et al., 2015)? Finally, exploring the content of the
ing engaged in the interactive planning session, actual interactions is also a crucial avenue for
participants feel as though they have acted and future research: what is it specifically about the
this feeling of action assuages the emotional reac- social communication within interacting groups
tion (as in, e.g., Stürmer & Simon, 2009). Another that promotes identification and action?
possibility is that, as group members engage with More generally, Dovidio (2013) recently advo-
the problems confronting people in developing cated for a renewed focus on research that under-
countries, their emotions transition from “raw” stands the reciprocal relationship between
experiences of anger, to more situated “norma- intragroup processes and intergroup relations.
tive” expressions of emotion (see Thomas et al., Intragroup interaction is arguably the social psy-
2009); such a pattern is reportedly typical of chological crucible of these effects and should be
long-term activists (e.g., Groves, 1995). Future more routinely utilized in experimental social psy-
research could test the reliability of this effect chological research (see also Thomas, Smith,
and consider these possibilities. McGarty, & Postmes, 2010).
A further avenue for research concerns the
multidimensional nature of identification.
Sustainable Collective Action
Although we have modelled identification as a
latent factor, the finding that affect was not inten- The current research draws on seminal past
sified by social interaction (and nor did it share (Lewin, 1947, 1953) and recent research to con-
the strong predictive relationship of centrality sider the processes through which people form
and ties) suggests that the mediating (encapsulat- collective efforts to redress social inequality. We
ing) role of the components may operate some- argued at the beginning of this paper that social
what differently. Future research should consider change is likely to be predicated on the formation
the possibility that the different dimensions of of new groups that can promote action in the
identity may play a different role at distinct stages present but are sustainable into the future
of the identity formation process. Indeed, recent (Thomas et al., 2009; see also Jones & Morton,
research has demonstrated that social identity 2012; Postmes et al., 2012; Rabinovich et al.,
components have different effects on discrete 2010). We also suggested that opinion-based
parts of intergroup phenomena (cooperation vs. group memberships are useful candidate groups
competition; Amiot & Aubin, 2012). Accordingly, in this regard (McGarty et al., 2009; McGarty,
future research might consider the role of the Lala, & Thomas, 2012). Consistent with these
subcomponents in affecting collective actions, points, these results experimentally demonstrated
emotions, and beliefs, and at discrete stages of that opinion-based groups can emerge through
the “commitment trajectory” (i.e., in contexts social interaction and effectively capture action-
where people first become aware of a group, oriented reactions (outrage and efficacy) and
develop and sustain their commitment.) boost commitment to a pressing global problem
It is also possible that identification encapsu- (combating global poverty). Put differently,
lates efficacy and outrage simply because the through interaction group members generated
meaning of the group (and the broader social collective identities such that “what it means” to
148 Group Processes & Intergroup Relations 19(2)

be a supporter of an end to global poverty meant nature of norm formation, conformity and group
to experience those emotions (outrage), believe in polarization. British Journal of Social Psychology, 29,
the usefulness of action (efficacy), and to take 97–119. doi:10.1111/j.2044-8309.1990.tb00892.x
that action; all elements of a normative alignment Amiot, C. E., & Aubin, R. M. (2012). Why and how are
you attached to your social group? Investigating
(Thomas et al., 2009).
different forms of social identification. British Jour-
These processes have a powerful analogy in
nal of Social Psychology, 52, 563–586. doi:10.1111/
everyday social interaction where, as Klandermans bjso.12004
(1997, p. 211), argues “protest is staged by people Ashmore, K. D., Deaux, K., & McLaughlin-Volpe, T.
who come to share [emphasis added] a contentious (2004). An organizing framework for collective
identity.” It is precisely that process of becoming identity: Articulation and significance of multi-
that the current research has sought to specify dimensionality. Psychological Bulletin, 130, 80–113.
(Haslam, Reicher, & Platow, 2011; Reicher & doi:10.1037/0033-2909.130.1.80
Hopkins, 2001; Smith et al., 2015). The results Bandura, A. (1997). Self-efficacy: The exercise of control.
point to the practical importance of allowing New York, NY: W. H. Freeman & Co.
people to come to a new understanding of self Bandura, A. (2000). Exercise of human agency through
collective efficacy. Current Directions in Psychological
by harnessing the power of social interaction.
Science, 9, 75–78. doi:10.1111/1467-8721.00064
Funding Bazerman, M. H., & Malhotra, D. (2006). Econom-
ics wins, psychology loses, and society pays. In
The author(s) disclosed receipt of the following finan-
D. De Cremer, M. Zeelenberg & J. K. Murnighan
cial support for the research, authorship, and/or
(Eds.), Social psychology and economics (pp. 263–297).
publication of this article: This research was supported
Mahwah, NJ: Lawrence Erlbaum Associates.
by the Australian Research Council (DE120101029 to
Bliuc, A., McGarty, C., Reynolds, K. J., & Muntele,
the first author and DP0770731 to the second author).
D. (2007). Opinion-based group membership as
a predictor of commitment to political action.
Notes European Journal of Social Psychology, 37, 19–32.
1. Some of these instructions also contained experi- doi:10.1002/ejsp.334
mental manipulations designed to manipulate Bliuc, A.-M., McGarty, C., Thomas, E. F., Lala, G.,
emotional or belief content of the discussions Berndsen, M., & Misajon, R. (2015). Understand-
(as in Thomas & McGarty, 2009) and/or the task ing the climate change divide as an intergroup
focus of the group discussions. Some of these conflict between skeptics and believers. Nature
manipulations produced effects of small to mod- Climate Change, 5, 226–229. doi:10.1038/ncli-
erate size and together they represent another mate2507
source of variation. The effects reported here Cameron, J. (2004). A three-factor model of
are robust in that they hold over and above these social identity. Self and Identity, 3, 239–262.
effects (for example, as routinely applies in a doi:10.1080/13576500444000047
meta-analysis). Cameron, J., & Nickerson, S. L. (2009). Predictors of
2. We conducted exactly the same tests of EMSICA protest amongst anti-globalization protesters.
and SIMCA omitting the affect component from Journal of Applied Social Psychology, 39, 734–761.
the latent identification factor. Results were doi:10.1111/j.1559-1816.2009.00458.x
unchanged though the SRMR within index was Dovidio, J. (2013). Bridging intragroup processes
slightly worse (SRMR within = .064 for EMSICA, and intergroup relations: Needing the twain to
.060 for SIMCA), suggesting that it is worth meet. British Journal of Social Psychology, 52, 1–24.
retaining it in the models. doi:10.1111/bjso.12026
Drury, J., & Reicher, S. (2000). Collective action and
psychological change: The emergence of new
References social identities. British Journal of Social Psychology,
Abrams, D., Wetherell, M., Cochrane, S., Hogg, M. A., 39, 579–604. doi:10.1348/014466600164642
& Turner, J. C. (1990). Knowing what to think by Drury, J., & Reicher, S. (2005). Explaining enduring
knowing who you are: Self-categorization and the empowerment: A comparative study of collec-
Thomas et al. 149

tive action and psychological outcomes. European Isenberg, D. J. (1986). Group polarization: A critical
Journal of Social Psychology, 35, 35–58. doi:10.1002/ review and meta-analysis. Journal of Personality and
ejsp.231 Social Psychology, 50, 1141–1151. doi:10.1037/0022-
Gee, A., Khalaf, A., & McGarty, C. (2007). Using group- 3514.50.6.1141
based interaction to change stereotypes about Jans, L., Postmes, T., & van der Zee, K. I. (2011). The
people with mental disorders. Australian Psycholo- induction of shared identity: The positive role
gist, 42, 98–105. doi:10.1080/00050060701280581 of individual distinctiveness for groups. Person-
Giguere, B., & Lalonde, R. (2010). Why do students ality and Social Psychology Bulletin, 37, 1130–1141.
strike? Direct and indirect determinants of col- doi:10.1177/0146167211407342
lective action participation. Political Psychology, 31, Jonas, K., & Morton, T. A. (Eds.). (2012). Restoring civil
227–247. doi:10.1111/j.1467-9221.2009.00750.x societies: The psychology of intervention and engagement fol-
Groves, J. M. (1995). Learning to feel: The neglected lowing crisis. Chichester, UK: Wiley-Blackwell.
sociology of social movements. The Sociologi- Klandermans, B. (1997). The social psychology of protest.
cal Review, 43, 435–461. doi:10.1111/j.1467- Oxford, UK: Blackwell.
954X.1995.tb00610.x Leach, C. W., van Zomeren, M., Zebel, S., Vliek, M. L.
Hardin, C. D., & Higgins, E. T. (1996). Shared real- W., Pennekamp, S., Doosje, B., … Spears, R. (2008).
ity: How social verification makes the subjective Group-level self definition and self-investment: A
objective. In R. M. Sorrentino & E. T. Higgins hierarchical (multicomponent) model of in-group
(Eds.), Handbook of motivation and cognition: The inter- identification. Journal of Personality and Social Psychol-
personal context (Vol. 3, pp. 28–84). Mahwah, NJ: ogy, 95, 144–165. doi:10.1037/0022-3514.95.1.144
Erlbaum. Leidner, B., Tropp, L., & Lickel, B. (2013). Bringing sci-
Haslam, S. A., & McGarty, C. (2001). A 100 years of ence to bear—on peace, not war. Elaborating on
certitude? Social psychology, the experimental psychology’s potential to promote peace. American
method and the management of scientific uncer- Psychologist, 68, 514–526. doi:10.1037/a0032846
tainty. British Journal of Social Psychology, 40, 1–21. Levine, J. M., & Moreland, R. L. (Eds.). (2006). Small
doi:10.1348/014466601164669 groups. New York, NY: Psychology Press.
Haslam, S. A., Reicher, S. D., & Platow, M. J. (2011). Lewin, K. (1947). Frontiers in group dynamics II.
The new psychology of leadership: Identity, influence and Channels of group life; social planning and
power. Hove, UK: Psychology Press. action research. Human Relations, 1, 143–153.
Haslam, S. A., Turner, J. C., Oakes, P. J., Reynolds, doi:10.1177/001872674700100201
K. J., Eggins, R. A., Nolan, M., & Tweedie, J. Lewin, K. (1953). Studies in group decision. In D.
(1998). When do stereotypes become really con- Cartwright & A. Zander (Eds.), Group dynamics:
sensual? Investigating the group-based dynamics Research and theory (2nd ed., pp. 287–301). Evan-
of the consensualization process. European Jour- ston, IL: Row Peterson.
nal of Social Psychology, 28, 755–776. doi:10.1002/ Lewin, K. (1958). Group decision and social change.
(SICI)1099-0992(199809/10)28:5<755::AID- In E. E. Maccoby, T. M. Newcomb & E. L. Hart-
EJSP891>3.0.CO;2-Z ley (Eds.), Readings in social psychology (3rd ed., pp.
Hogg, M. A., Abrams, D., Otten, S., & Hinkle, S. 197–211). New York, NY: Holt.
(2004). The social identity perspective: Inter- Louis, W. R. (2009). Collective action—and then what?
group relations, self-conception and small Journal of Social Issues, 65, 727–748. doi:10.1111/
groups. Small Group Research, 35, 246–276. j.1540-4560.2009.01623.x
doi:10.1177/1046496404263424 McGarty, C., Bliuc, A.-M., Thomas, E. F., & Bon-
Hopthrow, T., & Abrams, D. (2010). Group transfor- giorno, R. (2009). Collective action as the material
mation: How demonstrability promotes intra- expression of opinion-based group membership.
group cooperation in social dilemmas. Journal Journal of Social Issues, 65, 839–857. doi:10.1111/
of Experimental Social Psychology, 46, 799–803. j.1540-4560.2009.01627.x
doi:10.1016/j.jesp.2010.04.002 McGarty, C., Lala, G., & Thomas, E. F. (2012). Opinion-
Hu, L. T., & Bentler, P. M. (1999). Cutoff criteria for fit based groups and the restoration of civil society. In
indexes in covariance structure analysis: Conven- K. Jonas & T. Morton (Eds.), Restoring civil society
tional criteria versus new alternatives. Structural (pp. 250–264). Chichester, UK: Wiley-Blackwell.
Equation Modeling: A Multidisciplinary Journal, 6, Meleady, R., Hopthrow, T., & Crisp, R. J. (2013). The
1–55. doi:10.1080/10705519909540118 group discussion effect: Integrative processes and
150 Group Processes & Intergroup Relations 19(2)

suggestions for implementation. Personality and Social Psychology, 91, 698–711. doi:10.1037/0022-
Social Psychology Review, 17, 56–71. doi:10.1177/ 3514.91.4.698
1088868312456744 Sani, F., & Reicher, S. D. (2000). Contested identities
Moreland, R. L., Hogg, M. A., & Hains, S. C. (1994). and schisms in groups: Opposing the ordina-
Back to the future: Social psychological research tion of women as priests in the Church of Eng-
on small groups. Journal of Experimental Social Psy- land. British Journal of Social Psychology, 39, 95–112.
chology, 30, 527–555. doi:10.1006/jesp.1994.1025 doi:10.1348/014466600164354
Moscovici, S., & Zavalloni, M. (1969). The group as a Smith, L. G. E., & Postmes, T. (2009). Intra-group
polarizer of attitudes. Journal of Personality and Social interaction and the development of norms which
Psychology, 12, 125–135. doi:10.1037/h0027568 promote inter-group hostility. European Journal
Muthén, L. K., & Muthén, B. O. (2007). Mplus user’s of Social Psychology, 39, 130–144. doi:10.1002/
guide (7th ed.). Los Angeles, CA: Author. ejsp.464
Myers, D. G., & Bishop, G. D. (1970). Discussion Smith, L. G. E., & Postmes, T. (2011a). Shaping stereo-
effects on racial attitudes. Science, 169, 778–779. typical behaviour through the discussion of social
doi:10.1126/science.169.3947.778 stereotypes. British Journal of Social Psychology, 50,
Postmes, T., Haslam, S. A., & Swaab, R. I. (2005). Social 74–98. doi:10.1348/014466610x500340
influence in small groups: An interactive model of Smith, L. G. E., & Postmes, T. (2011b). The power
identity formation. European Review of Social Psychol- of talk: Developing discriminatory group norms
ogy, 16, 1–42. doi:10.1080/10463280440000062 through discussion. British Journal of Social Psychol-
Postmes, T., Rabinovich, A., Morton, T. A., & van ogy, 50, 193–215. doi:10.1348/014466610x504805
Zomeren, M. (2012). Toward sustainable social Smith, L. G. E., Thomas, E. F., & McGarty, C. (2015).
identities: Including our collective future into the “We must be the change we want to see in the
self-concept. In H. van Trijp (Ed.), Encouraging world”: Integrating norms and identities through
sustainable behavior (pp. 191–207). New York, NY: social interaction. Political Psychology, 36, 543–557.
Psychology Press. doi:10.1111/pops.12180
Postmes, T., Spears, R., Lee, A. T., & Novak, R. J. Stott, C., & Drury, J. (2004). The importance of social
(2005). Individuality and social influence in structure and social interaction in stereotype con-
groups: Inductive and deductive routes to group sensus and content: Is the whole greater than the
identity. Journal of Personality and Social Psychology, sum of its parts? European Journal of Social Psychol-
89, 747–763. doi:10.1037/0022-3514.89.5.747 ogy, 34, 11–23. doi:10.1002/ejsp.183
Rabinovich, A., Morton, T., & Postmes, T. (2010). Stürmer, S., & Simon, S. (2009). Pathways to collec-
Time perspective and attitude–behaviour consist- tive protest: Calculation, identification or emo-
ency in future-oriented behaviours. British Journal tion? A critical analysis of the role of group-based
of Social Psychology, 49, 69–89. doi:10.1348/014466 anger in social movement participation. Journal
608x401875 of Social Issues, 65, 681–726. doi:10.1111/j.1540-
Reicher, S. D. (1984). The St. Pauls’ riot: An expla- 4560.2009.01620.x
nation of the limits of crowd action in terms of Swaab, R., Postmes, T., van Beest, I., & Spears, R.
a social identity model. European Journal of Social (2007). Shared cognition as a product of, and
Psychology, 14, 1–21. doi:10.1002/ejsp.2420140102 precursor to, shared identity in negotiations. Per-
Reicher, S. D. (1996). “The Battle of Westminster”: sonality and Social Psychology Bulletin, 33, 187–199.
Developing the social identity model of crowd doi:10.1177/0146167206294788
behaviour in order to explain the initiation and Tausch, N., & Becker, J. C. (2013). Emotional reac-
development of collective conflict. European Jour- tions to success and failure of collective action as
nal of Social Psychology, 26, 115–134. doi:10.1002/ predictors of future action intentions: A longitu-
(SICI)1099-0992(199601)26:1<115::AID- dinal investigation in the context of student pro-
EJSP740>3.0.CO;2-Z tests in Germany. British Journal of Social Psychology,
Reicher, S. D., & Hopkins, N. (2001). Self and nation. 52, 525–542. doi:10.1111/j.2044-8309.2012.
London, UK: Sage. 02109.x
Roccas, S., Klar, Y., & Liviatan, I. (2006). The paradox Thomas, E. F., Mavor, K. I., & McGarty, C. (2012). Social
of group-based guilt: Modes of national identifica- identities can facilitate and encapsulate action rel-
tion, conflict vehemence, and reactions to the in- evant constructs. Group Processes and Intergroup Rela-
group’s moral violations. Journal of Personality and tions, 15, 75–88. doi:10.1177/1368430211413619
Thomas et al. 151

Thomas, E. F., & McGarty, C. A. (2009). The role of pro-social and hostile social movements. Interna-
efficacy and moral outrage norms in creating the tional Review of Social Psychology, 23, 17–55.
potential for international development activism Turner, J. C., & Oakes, P. J. (1986). The significance of
through group-based interaction. British Journal of the social identity concept for social psychology
Social Psychology, 48, 115–134. doi:10.1348/01446 with reference to individualism, interactionism and
6608X313774 social influence. British Journal of Social Psychology, 25,
Thomas, E. F., McGarty, C., & Louis, W. (2014). 237–252. doi:10.1111/j.2044-8309.1986.tb00732.x
Social interaction and psychological path- Van Zomeren, M., Postmes, T., & Spears, R. (2008).
ways to political engagement and extremism. Toward an integrative social identity model of
European Journal of Social Psychology, 44, 15–22. collective action: A quantitative research synthesis
doi:10.1002/ejsp.1988 of three socio-psychological perspectives. Psycho-
Thomas, E. F., McGarty, C., & Mavor, K. I. (2009). logical Bulletin, 134, 504–535. doi:10.1037/0033-
Aligning identities, emotions and beliefs to create 2909.134.4.504
commitment to sustainable social and political Van Zomeren, M., Spears, R., Fischer, A. H., &
action. Personality and Social Psychology Review, 13, Leach, C. W. (2004). Put your money where your
194–218. doi:10.1177/1088868309341563 mouth is! Explaining collective action tendencies
Thomas, E. F., McGarty, C., Stuart, A., Lala, G., Hall, through group-based anger and group efficacy.
L., & Goddard, A. (2015). Whatever happened Journal of Personality and Social Psychology, 87, 649–
to Kony2012? Studying a global online phenom- 664. doi:10.1037/0022-3514.87.5.649
enon as an emergent social identity. European Jour- Wittenbaum, G. M., & Moreland, R. L. (2008). Small-
nal of Social Psychology, 45, 356–367. doi:10.1002/ group research in social psychology: Top-
ejsp.2094 ics and trends over time. Social and Personality
Thomas, E. F., Smith, L. G. E., McGarty, C., & Post- Psychology Compass, 2, 187–203. doi:10.1111/j.1751-
mes, T. (2010). Nice and nasty: The formation of 9004.2007.00065.x