Online Discussion and the Moral Pathway to Identity Politicization and Collective Action

Research on the mobilizing potential of the Internet has produced some controversy between optimistic vs. skeptical perspectives. Although some attention has been paid to the effects of online discussions on collective participation, very little is known about how people’s experience of online interactions affects the key psychosocial predictors of collective action. The present research investigated whether use of the Internet as a channel for deliberation influenced the moral pathway to collective mobilization by shaping users’ politicized identity, thereby indirectly influencing collective action. Results showed that when people perceived online discussions as a constructive communication context, their politicized identity was imbued with the meaning of responding to a moral obligation, and willingness to participate in collective action was sustained. However, when participants perceived that online discussions were not constructive, their identification with the movement did not refer to moral obligation, and intention to participate in collective action was not sustained. Our discussion focuses on the need to deepen investigation of how people experience the particularities of interacting online, and on how this can affect psychosocial processes leading to collective action.

Politicized group identity is based upon group members' awareness that they share grievances and that their unsatisfying condition can be attributed to an opponent. As compared with non-politicized group identity, politicized group identity has a stronger normative content. Injunctive norms about 'the way the world should be', 'how it should change', and 'what we should do to make it change' lie at the core of its contents. As suggested above, these norms are socially validated and consensualized through intragroup discussion (Postmes, Haslam, & Swaab, 2005;Postmes, Spears, Lee, & Novak, 2005;Smith et al., 2015). As a result of the latter, group members converge on what defines them as a group: their collective interests and goals, the necessity to act collectively, and the specific actions they should undertake as a group. Indeed, group members develop a fully politicized group identity when they become aware they need to bring their fight into the political arena and make other parties (e.g., public opinion) take sides (Simon & Klandermans, 2001). Politicized group identity plays a key role in predicting collective action: the politicization process implies an encapsulation of other motivations to action and, feeding back, it further facilitates them (Thomas, Mavor, & McGarty, 2012; Turner-Zwinkels, Van Zomeren, & Postmes, 2015;van Zomeren, Postmes, & Spears, 2008).
Those motivations include both group-based variables and individual variables. With respect to group-based variables, attention has focused on group efficacy and group-based anger. Group efficacy includes subjective beliefs that members are capable of joining efforts and achieving collective goals, so as to change the situation in the desired direction (van Zomeren et al., 2008;van Zomeren, Spears, Fischer, & Leach, 2004). Groupbased anger stems from the appraisal that the status quo is unjust and that actions need to be taken against those responsible for the existing injustice (Leach, Iyer, & Pedersen, 2006;van Zomeren et al., 2004).
In terms of individual variables, recent psychosocial literature has focused on individual differences, such as sociopolitical beliefs or values, life experiences, and personal moral motivations (Duncan, 2012;Milesi & Alberici, 2018;Thomas et al., 2016;van Zomeren, 2013). Personality and life experiences are considered as distal antecedents of politicized group identity; on the contrary, moral motivations have been shown to be directly associated with it. Actually, the very process of politicization relies heavily on the moralization of the collective cause (Milesi & Alberici, 2018;Zaal et al., 2017). Moral motivations include moral convictions and a moral obligation to act. Moral convictions are metacognitions that one's own individual attitude towards the collective cause is rooted in one's own moral principles (Skitka & Morgan, 2014). Moral obligation to act corresponds with the feeling that action is what should be done, independent of personal interest, others' approval, or likelihood of success, because it is one's personal conscience that demands it (Vilas & Sabucedo, 2012). While both are associated with politicized group identity and, through it, they contribute to predict collective action (Alberici & Milesi, 2016;van Zomeren, Postmes, & Spears, 2012), moral obligation seems to be more closely involved in the politicization of group identity. This is probably because it also interacts with intragroup communication, and implies a stronger commitment to act in a public context (Alberici & Milesi, 2016;Milesi & Alberici, 2018;Vilas & Sabucedo, 2012). We start, then, from the idea that online discussions are a communication context where people, in specific ways, can ascribe a moral meaning to their sense of politicized identity.

Online Discussion and Collective Action
Discussing politics via the Internet has become a primary feature of contemporary political communication. A growing number of people use social media, online networks and blogs not only to stay informed, but also to Alberici & Milesi 145 exchange opinions, develop common understandings and, in some cases, learn how to mobilize for a collective cause. This is especially true for emerging social movement activists (Brunsting & Postmes, 2002).
As suggested above, according to many (e.g., Brunsting & Postmes, 2002;Shah et al., 2005), online interactions bring the potential to increase civic engagement and collective participation. In this regard, some recent studies found that online interactions may affect people's beliefs and emotions that, in turn, bring them to collective action. Moreover, it has been shown that discussions among group members can contribute to develop a sense of politicized identity, indirectly sustaining collective action (Alberici & Milesi, 2013, 2016. These studies highlighted the importance of taking into account not only how often people discuss online, but also how they perceive the quality of online interactions. Some preliminary results showed that those who perceive discussions as being of high quality charge their sense of politicized identity with the idea that collective action responds to an individual moral duty (Alberici & Milesi, 2016). The present research is aimed at deepening such preliminary results by: a) focusing on a larger range of specific characteristics and implications of discussing online, and b) analyzing how activists' perceptions of these characteristics affect the moral meaning of politicized identity. Indeed, the exchange of personal ideas about social reality can convert subjective individual perceptions into socially validated injunctive norms, upon which politicized identity can develop (McGarty et al., 2014;Smith et al., 2015). This, in turn, could indirectly affect the intention (or not) to mobilize collectively.
Compared to face-to-face discussions, online discussions present some specific features: they are mostly textbased, asynchronous, poor in visual and social cues, and often occur under conditions of anonymity (Valenzuela et al., 2012). Some have suggested that the text-based and asynchronous character of online interactions may stimulate more reflexive, rational, and argumentative conversations (Coleman & Gotze, 2001;Stromer-Galley & Wichowski, 2010). Some studies conducted on deliberative interactions showed that deliberation facilitates political participation (Katz, 1992;Warren, 1992). Deliberation corresponds to the process whereby a group of individuals think over a common or public problem and exchange rational and critical arguments about it with the aim of finding a solution that is likely to be agreed upon by those who are involved (Habermas, 1989). To be deliberative, discussions require a number of characteristics, such as rationality and reflexivity. For example, online deliberation is sustained when interactions rely on well-founded arguments (Halpern & Gibbs, 2013). Deliberation also is facilitated when individuals have the chance to learn new perspectives from others. In this regard, as online discussions often rely on networks characterized by weak ties, participants are likely to be provided with non-redundant and diverse information that may stimulate political learning (Valenzuela et al., 2012). Min (2007) compared the effects of face-to-face and online deliberation in experimental settings, and found that the latter was almost as effective as face-to-face deliberation in increasing participants' political engagement.
There also is evidence, however, that some features of online interactions may present an obstacle to deliberation. Text-based and asynchronous communication has been defined as a 'poor' form of communication, where the lack of non-verbal cues can easily produce misinterpretations and misunderstandings (Daft & Lengel, 1986;Walther, 1992). According to this view, text-based online interactions would not be efficient for empathizing with others or for exchanging emotions (Cheshin, Rafaeli, & Bos, 2011).
This drawback also is associated with the degree of conversational coherence of online discussions, that is, the extent to which messages relate to prior comments or the initial post (Halpern & Gibbs, 2013).

Online Discussion and Collective Action 146
Thus, on the one hand, some specific features of interacting online may lead to constructive and reflexive discussions where participants exchange and learn well-founded arguments; on the other, however, online discussions may be perceived as being non-constructive. This is due to a low conversational coherence which can foster a lack of understanding among users. Starting from these studies, we aimed at investigating how the way in which activists of a social movement experience online discussions affects: a) their sense of politicized identity and, particularly, b) the perception that collective action responds to an individual moral duty. Our core expectation was that when activists perceive online discussions as being reflexive and argumentative, the sense of moral obligation would be strongly associated with participants' politicized identity. This, in turn, would indirectly stimulate collective action. From a different standpoint, when activists participate in online discussions where conversational coherence is low and users frequently misunderstand each other, the sense of moral obligation might not be associated with activists' politicized identity. As a consequence, intention to participate in collective action would not be sustained.

Research Context
Activists

Results
To test whether the four characteristics of online discussions were empirically distinct, we carried out principal axis factoring analyses with oblique rotation. Actually, the factor analysis extracted two factors that predicted 79.55% of the variance. Perceived soundness of argumentation and the perception that online discussions provided new information loaded on the first factor (47.48% of variance explained; both factor loadings were .92), while perceived presence of misunderstandings and perceived conversational coherence loaded on the second factor (32.06% of variance explained; factor loadings ranged from .81 to .86). Thus, while the first factor referred to the perception that online discussions were reflexive and relied on well-founded argumentation, the second factor referred to the perception that online discussions contained misunderstandings and a lack of conversational coherence. Based on this, we computed two separated indexes of these two different dimensions: a) reflexivity and b) lack of conversational coherence. These were then employed in the following analyses.   To test the predicted moderating effects of the activists' perceptions of online discussion, we performed separate hierarchical regressions with politicized identity as the criterion. Moral obligation and the other groupbased motivations (i.e., anger and collective efficacy) were entered in Step 1, followed by the frequency of online discussion and the perceived characteristics of online discussion in Step 2, and by each interaction term (moral obligation☓perceived reflexivity/lack of conversational coherence of online discussion) in Step 3. All variables were centred before entering the regressions. As shown in Table 2, Step 1 revealed significant main effects for moral obligation, as well as for collective efficacy and anger on politicized identity. Frequency of online discussion and the perceived characteristics of online discussion did not show main effects in Step 2.
Step 3 revealed significant effects for both the interaction terms. As predicted, moral obligation interacted

B (SE) B (SE) B (SE) B (SE) B (SE)
Step a Moral obligation and group-based motivations. b Step 1 + Online political discussion, perceived features of online discussion. c Step 2 + each of the two-way interactions between moral obligation and features of online discussion.
Simple slope analysis of moral obligation☓reflexivity of online discussion interaction revealed that the influence of moral obligation on politicized identity was significant for respondents who reported participating in online discussions featured by high reflexivity (B = .40, SE = .1; p < .01). It was not significant for participants who reported engaging in discussions featured by low reflexivity (B = .13, SE = .08; p = .11). Consistently, simple slope analysis of the moral obligation☓lack of conversational coherence interaction showed that the influence of moral obligation on politicized identity was not significant for those who reported participating in online discussions featured by low conversational coherence (B = .03, SE = .09; p = .74). It was significant for participants who reported engaging in online discussions featured by high conversational coherence (B = .42, SE = .13; p < .05).
Last, we investigated whether these interactions had an indirect effect on collective action intention through politicized identity. We tested mediated moderation effects for each of the independent variables investigated according to the specifications set out by Hayes's PROCESS macro for SPSS, using model 8 (Hayes, 2013) (see Figure 1). In each analysis, we controlled frequency of online discussions, collective efficacy, and anger.
Online Discussion and Collective Action 150 Consistent with the first mediated moderation, moral obligation maintained its (indirect) effect on willingness to collective action when activists perceived online discussion was characterised by high conversational coherence. However, when online discussions were perceived as having low conversational coherence, politicized identity did not mediate the effect of moral obligation on intention to collective action. Thus, collective action was not sustained through the investigated path when activists perceived they were participating in nonconstructive online discussions.

Discussion
On the whole, results of our study support the claim that people's experience of online political discussions can shape the moral pathway to collective action. Findings showed that when activists perceived that the online discussions they joined were a constructive and reflexive communication context, where they exchanged wellfounded arguments and learned new information, their identification with the movement was imbued with the sense of responding to a moral duty. In contrast, when participants perceived that online discussions were not a constructive communication context, where there was low conversational coherence and people misunderstood each other, their identification with the movement did not refer to a moral duty. Thus, depending on users' Alberici & Milesi 151 experience of online discussion, the sense of moral obligation may or may not contribute to shaping the meaning of their politicized identity. Findings also showed that, through this process, perceptions of online discussion can indirectly affect willingness to participate in collective action. Indeed, when activists perceived online discussions as being reflexive, the moral pathway to collective action (through the mediation of politicized identity) was sustained. However, when online discussions were perceived as being poor in conversational coherence, then willingness to take part in collective actions was not stimulated by this path.
Thus, our findings enlarge research on the relation between computer-mediated communication and political participation, and they suggest some possible directions for future research. First, results suggest it is important to investigate not only how interacting online per se affects collective action, but also people's experience of online discussions. This latter factor, with its specific features, affects the psychosocial processes underlying the development of politicized identity that, in turn, gives rise to action. Our study confirmed that new technologies are not neutral for what concerns effects on collective action (see Spears & Postmes, 2015), in that group identities are influenced and transformed over time during online interactions (Alberici & Milesi, 2016;McGarty et al., 2014). Moreover, by reaching consensus online about the desired social reality, people define their group identities as aimed toward obtaining social change (Thomas, McGarty, & Louis, 2014).
The current study deepens the investigation of how online interactions moderate the moral path to collective action (Mazzoni & Cicognani, 2013;Vilas & Sabucedo, 2012). It seems that when activists interact online, and such interactions support deliberative practices, individual moral judgement about social reality is strengthened.
This may contribute to politicizing group identity by reinforcing the idea that there is a moral requirement for struggle in the public arena. The exchange of personal opinions about the desired social reality is likely to convert individual views of moral responsibility in a shared and socially validated norm, which forms the basis for developing a politicized identity. It is also possible that, when people participate in reflexive and well-argued discussions, they are likely to engage in post hoc moral reasoning; that is, they put forward arguments to support an a priori moral judgement (Haidt, 2001). This process would not be sustained, however, when online discussions are perceived to be non-constructive in terms of their conversational coherence.
A further point concerns the importance of more deeply investigating how people experience the specific features of interacting online. As suggested above, online discussions have some unique features, not experienced in face-to-face conversations. Depending on how people experience these aspects, online interactions may sustain collective participation on the one hand, but on the other, weaken the psychosocial processes underlying willingness to mobilize. This result could enrich the ongoing debate regarding the mobilizing potential of the Internet. It suggests attention should be focused not only on effects of the amount of online interactions, but also on the ways in which people use and experience new technologies (Kende et al., 2016). In this regard, people may show individual differences in the way they perceive the same online discussions. For example, our results showed that online discussions were perceived to be more or less constructive. Individual interest and knowledge of politics could be relevant here: more interested and expert participants could rely on different standards when they evaluate argumentative quality of online discussions, when contrasted with less interested or knowledgeable participants (Catellani & Alberici, 2012;Catellani, Milesi, & Alberici, 2014;Kahn & Kenney, 1999;Milesi, 2016). Research into online deliberation, and on its effects on political engagement, could certainly help in this direction (see Halpern & Gibbs, 2013).

Online Discussion and Collective Action 152
This study, to be certain, is in large part explorative. To corroborate our results, the investigation should be extended to other movements and contexts, and should involve larger samples. We also acknowledge that the current study considered only some of the specific features of online interactions. Further research could focus on other well-known characteristics of communicating through social media (e.g., Halpern & Gibbs, 2013), investigating their interaction with the psychosocial motives for collective action. For example, the perceived degree of civility of online discussions (i.e. the promoted respect and tolerance among users) could moderate the effect of action-relevant emotions (e.g. anger, contempt) on politicized identity and collective action.
Moreover, it must be acknowledged that there are different degrees of both participation in online interactions and engagement in online discussions. The questions we used in the present research may not have captured the full range of these factors. Not only can people just read the discussions as they unfold without intervening; they also can participate in the interactions in a variety of ways (e.g. asking or giving information, asking or making suggestions), and with different levels of intensity (e.g. Vaccari et al., 2015). Future research should investigate how these different ways and levels of engagement in online discussions interact with psychosocial predictors of collective action in influencing politicized identity and collective action.
Despite these limitations, our findings suggest it is worthwhile continuing to investigate those features of contemporary CMC that may be relevant to shaping the meaning of people's collective identities, thus sustaining or weakening their willingness to take part in collective actions.

Funding
The authors have no funding to report.