“Free” Not to Engage: Neoliberal Ideology and Collective Action. The Case of the Yellow Vest Movement

Erupting in November 2018, the Yellow Vest (YV) movement has shaken up France. Demonstrations and strikes are not uncommon in the French landscape (Noiriel, 2018a). Indeed, the YV movement has been compared to the ‘sans culottes’ denunciation of representativeness, and of the split between ‘the people’ and the ‘privileged’ during the French Revolution (Grunberg, 2019). Nevertheless, it also has its own specificities. For instance, this grassroots movement has been largely decentralized, taking place in roundabouts and other local blockade points, alongside the demonstrations in larger cities. Moreover, social media websites, such as Facebook, have served as the mobilizing and coordinative tools of choice, replacing intermediaries such as Trades Unions (Boyer et al., 2019; Jetten, Mols & Selvanathan, 2020; Noiriel, 2018b). Concerning its demands, although the movement originally started because of a rise in fuel taxes, it has rapidly taken the form of a broader denunciation of the political agenda and of the institutions themselves (Boyer et al., 2019). In this context, we propose to look at the factors associated to the engagement in the YV movement. More specifically, in two correlational studies, we test the main hypothesis that the endorsement of neoliberal ideology is negatively associated to the engagement in the movement. Furthermore, we propose to look at the processes by which this relation may occur (via perceived personal control, system justification and identification with the YV).

tors associated to the engagement in the YV movement. More specifically, in two correlational studies, we test the main hypothesis that the endorsement of neoliberal ideology is negatively associated to the engagement in the movement. Furthermore, we propose to look at the processes by which this relation may occur (via perceived personal control, system justification and identification with the YV).

Engagement in the YV Movement as Collective Action
Collective action (CA) has been defined as individual behaviors carried out on behalf of a group, in the hope of achieving group-level changes (Wright, Taylor & Moghaddam, 1990). It may include both public action (e.g., participating in demonstrations) and private action (e.g., signing a petition; Radke, Hornsey & Barlow, 2016). Based on this definition, engagement in the YV movement should be predicted by the factors commonly considered as antecedents of CA (e.g., van Zomeren, Postmes & Spears, 2008). In their parsimonious Social Identity Model of Collective Action (SIMCA), van Zomeren et al. (2008) emphasize three comprehensive and direct predictors of engagement in CA: subjective injustice, social identity and collective efficacy.
Social identity, conceptualized as the shared knowledge of the social reality of the group directly predicts collective action (van Zomeren, Postmes & Spears, 2008) and all the more when it is politicized. A politicized identity can be described as a collective identity that underlies the awareness of a group's disadvantage, and the blaming of an external agent for the group's status differential (Simon & Klandermans, 2001;van Stekelenburg & Klandermans, 2013). It also means to be willing to and to actually engage in a power struggle with the perceived guilty party, in order to eradicate this differential (Simon & Klandermans, 2001;van Stekelenburg & Klandermans, 2013). Accordingly, politicized identity predicts CA to a greater extent than does non-politicized identity (e.g., Stürmer & Simon, 2004b;van Zomeren et al., 2008). The YV identity is considered here as a politicized identity because it answers to the aforementioned criteria (see also Jetten et al., 2020). 1 To better understand the role of social identity in the YV movement, it is important to distinguish between two types of disadvantages that may drive CA: structural and incidental (van Zomeren et al., 2008). The rise in fuel price at the origin of the YV movement can be considered typical incidental disadvantage. In such cases, the shared social identity develops around the specific issue as people recognize their common fate in this disadvantage (van Zomeren et al., 2008). However, in the case of the YV movement, the incidental disadvantage has soon been linked to broader social and institutional structures, turning the movement into a response to structural disadvantages at the expense of lower and middle social classes (e.g., unfair tax system; Noiriel, 2018b and see Study 1 and Appendix B), and more broadly of all of those who do not feel that the current French democratic functioning serves most people well (e.g., demand for a citizen referendum initiative; see Appendix B).
On a related note, as Jetten et al. (2020) point out, and despite the existence of social inequalities in France (Piketty, 2013), the French identity is historically grounded into a cultural value of equality. As such, Jetten et al. (2020) have proposed that besides acting on behalf of the 'have nots', the YV movement also aims to defend historical continuity in the content of the French identity, which is nourished by ideals of equality. Therefore, this movement lies at the intersection of a class struggle (Jetten et al., 2020;Noiriel, 2018b) and a willingness to defend French ideals of equality (Jetten et al., 2020) through the country's institutions and practices. Altogether, these arguments help understand what identities can underpin the YV movement and be relevant to investigate collective action to achieve group-level changes in this context.
The SIMCA stipulates that identity indirectly predicts CA, through its links with subjective injustice and collective efficacy. On the one hand, identity should facilitate inter-group comparisons and the perception of relative group-based deprivation, as well as the emotional experience of that deprivation. In fact, the model distinguishes between cognitive injustice (i.e., the perception of injustice) and affective injustice (i.e., the anger that results from this perception). It also specifies that affective injustice is a better predictor of CA than cognitive injustice, insofar as group-based emotions are more strongly linked to action tendencies (e.g., van Zomeren et al., 2008;van Zomeren et al., 2004). On the other hand, social identity is expected to predict efficacy because it offers a collective foundation for the disadvantaged to empower themselves, both individually and collectively (van Zomeren et al., 2008). In sum, the SIMCA predicts that identity, and to a greater extent politicized identity, efficacy and injustice should directly predict CA, and that identity should also indirectly predict CA via efficacy and injustice.

System Justification and Neoliberal Ideology
However, while the SIMCA model helps to explain why people do engage in CA, it does not take into account the motivation people may have to justify the system and its inequalities (Jost et al., 2017), which can undermine engagement in system challenging CA Osborne et al., 2019). In fact, in Osborne et al.'s (2019) study, the more people justified the system, the less they engaged in system-challenging CA. According to system justification theory, people can be motivated to uphold and legitimize the systems (i.e., political, economic, social systems) in which they are embedded, even when these systems work to their disadvantage (see recent reviews by Friesen et al., 2019;Jost, 2019). In order to legitimize their surrounding systems, people can rely on system-justifying ideologies (e.g., meritocracy, gender stereotypes, benevolent sexism; e.g., Becker & Wright, 2011;Bonnot & Jost, 2014;Jost et al., 2012;Jost & Hunyady, 2005). In fact, Simon and Klandermans (2001) have suggested that legitimizing and hierarchyenhancing ideologies could indeed obstruct the politicizing process of identity (see also Cameron & Nickerson, 2009). In the present research, we are interested in one such system-justifying ideology, namely neoliberal ideology. Indeed, Azevedo, Jost, Rothmund, and Sterling (2019) have recently shown that neoliberal ideology is associated with general, economic, and gender system justification and therefore could well function as a system-legitimizing ideology (see also ).
An ideology can be defined as beliefs, opinions, and values that are shared among a certain group or society and that function as a means to interpret the world and envision the world as it should be (e.g., Jost, Federico & Napier, 2009). While defining neoliberalism is considered a difficult task (e.g. Venugopal, 2015), it generally refers to a set of economic and political inclinations in favor of the financialization of the economy, a deregulation of labor practices and capital circulation, and a privatization of national assets, while aiming to limit state intervention to its sovereign activities (e.g., enforcing the law) and to the planning of liberalization (Simon, 2016;Teo, 2018).
In such a neoliberal framework, people's position in society should be determined by their input (i.e., individual merit), and everyone should be able to compete against one another for resources. In fact, the government should guarantee a fair competition. Neoliberal ideology has also been described as a macro-cultural factor that manifests itself in people's psychology (e.g., Adams et al., 2019;Pulfrey & Butera, 2013;Ratner, 2019;Teo, 2018). For instance, individual freedom is framed as a sacred value (Harvey, 2007) and mainly underlies the notions of personal responsibility and self-determination (Teo, 2018).
Based on this specific perception of society characterizing neoliberalism, Bay-Cheng et al. (2015) have developed a measure of endorsement of neoliberal ideology (i.e., Neoliberal Beliefs Inventory) that includes the dimensions of perception of system inequality, competition, personal wherewithal and opposition to government interference. We will be using this measure in Study 1.
Moreover, neoliberalism is frequently defined as a form of governmentality (Foucault, 1978, cited in Laval, 2018Pyysiäinen, Halpin & Guilfoyle, 2017;Teo, 2018). As such, it dictates how people should govern themselves to be efficient neoliberal subjects (Scharff, 2016). Above all, neoliberal subjects have to be free (Adams et al., 2019;Zucker & Bay-Cheng, 2010): free from constraints (i.e., free will), which implies to be an autonomous decisionmaking agent (Adams et al., 2019;Pyysiäinen et al., 2017;Zucker & Bay-Cheng, 2010) and free to pursue personal aspirations. In this perspective, people are encouraged to 'be what they are and do what they want' (Adams et al., 2019;Teo, 2018;Türken et al., 2015). Positive feelings and happiness are valorized states that people are invited to achieve by making good choices (Adams et al., 2019;Teo, 2018). Therefore and going back to Bay-Cheng et al.'s (2015) scale dimension of personal wherewithal, people are said to be personally responsible for their own life and level of happiness (Adams et al., 2019). With those elements in mind, Study 2 will assess endorsement of neoliberal ideology by relying on people's belief in free will, tendency of singularity, and hedonism orientation.

Neoliberal Ideology: A Barrier to Engagement in CA
In sum, neoliberal ideology's framework and reach are broader than other ideologies that it encompasses (e.g., fair market ideology, meritocracy). Of notable interest for this research is how neoliberal ideology can be invoked to justify inequalities and decrease people's interest and perceived legitimacy of CA. Through its emphasis on personal responsibility and internal barriers that people have to overcome, it indeed minimizes the consideration of structural factors as sources of one's disadvantage and it presents self-regulation as the solution (Pyysiäinen et al., 2017). In doing so, it is likely to distance people from collective issues (Teo, 2018) and to discredit engaging in collective action to solve a disadvantage that goes beyond one's own condition while also delegitimizing it, because people should 'simply' put in more efforts to succeed (Brown, 2015;Teo, 2018).
For instance, in Codou et al.'s (2012) study, participants who were primed with advertisements reminding them of neoliberal values (e.g., 'behave good, behave bad, be you') eventually chose more statements referring to internal causal explanations (Dubois & Beauvois, 2005), thus minimizing the influence of external factors, compared to participants primed with neutral information. Thus, we expect that it is through the enhanced perception of personal control that neoliberal ideology leads to a greater perception that the system is just, because it entails that upward mobility and happiness are possible within the system, as long as one exercises one's control in putting the sufficient effort (Osborne et al., 2019). Because targeting and aiming for structural changes is perceived as illegitimate, and because one should not be preoccupied by group-level issues, we also hypothesize that neoliberal ideology should be negatively associated to CA, notably through its effect on politicized identity (Simon & Klandermans, 2001; and see Bonnot, Redersdorff, Girerd & Verniers, 2020;Fitz, Zucker & Bay-Cheng, 2012. This hypothesis will be tested in two correlational studies conducted in the context of the YV movement.
First, one of Study 1's objectives will be to analyze grievances expressed by the YVs, to determine the extent to which they convey a system-challenging view. Through a diversified sample, perceptual differences concerning the YVs' grievances, between YV identifiers, and nonidentifiers, will be investigated. Thus, in these preliminary analyses, identification with the YVs will be used as a categorical variable to distinguish between the grievances. However, the main analyses of both studies will focus on YV identification as an outcome of the influence of neoliberal ideology, while also predicting engagement in CA.
Consistent with the integrative model of Osborne et al. (2019), Study 1's purpose is indeed to test the SIMCA on engagement in the YV movement, while incorporating an ideological dimension, namely the endorsement of neoliberal ideology. In Study 1, its endorsement will be measured through the dimensions of perceived system inequality, competition, meritocracy (i.e., personal wherewithal) and government interference as proposed by Bay-Cheng et al. (2015) in the Neoliberal Beliefs Inventory. Those dimensions actually focus more specifically on people's perception of society. The objective of Study 2 is twofold. Its first objective is to replicate Study 1's predictive model by using a different measure of neoliberal ideology endorsement, focusing more on neoliberal governmentality and people's perception of themselves and others within, and in relation to, a neoliberal context (i.e., singularity, free will and hedonism), while including a measure of system justification and focusing on participants who are at least minimally supportive of the movement. Moreover, Study 2 will extend the scope of Study 1 in investigating the processes through which the endorsement of neoliberal ideology might be negatively associated to politicized identification with the YVs, and as a consequence, engagement in the movement.

Study 1
We first hypothesize that the grievances expressed by the YVs will reflect a system-challenging stance. The comparisons between the grievances expressed by the YVs and the perceived grievances of people who do not identify with the YV, including on the issues of environmental protection, of constitutional change and of regulation of immigration, will be investigated but are more exploratory in nature.
We further hypothesize that the endorsement of neoliberal ideology will be negatively associated to the politicized YV identity and the two forms of injustice (van Zomeren et al., 2008), and will therefore be negatively associated to engagement in the YV movement (see Figure 1 for the full hypothesized model). 2 The links between general social identifications and politicized identification with the YVs will also be investigated in order to explore whether they form a viable basis for the YV identification (Simon & Klandermans, 2001;van Stekelenburg & Klandermans, 2013). We think of two non-politicized forms of identities that could be relevant in the context of this movement: national identification and identification with the lower and middle social classes. Indeed, in the YV context, national identification, envisioned here as critical attachment to the national group (more so than glorification, see Roccas, Klar & Liviatan, 2006), could be the common ground in what seems to be a very diversified movement (Bedock et al., 2018), and identification with the lower and middle classes could reflect a description of the movement in terms of a class warfare (Noiriel, 2018b). However, to be parsimonious and because our interest is more specifically focused on the politicized identification with the YVs, more so than on non-politicized identification, as in recent integrative models (Jost et al., 2017;Osborne et al., 2019), only identification with the YVs will be used as the predictor of the other SIMCA variables.
Finally, the way in which different politicized identities may correlate with one another, and with endorsement of neoliberal ideology will also be explored. Indeed, certain politicized identities are associated with system-challenging positions and may lead to engagement in systemchallenging collective action (e.g., feminist identification; Zucker, 2004), but others reflect system-supporting positions (e.g., identification with 'La manif pour tous'; a French movement that had demonstrated against the legalization of same sex marriage or identification with the Tea Party in the United States, Hennes et al., 2012). This will also allow us to investigate if endorsing neoliberal beliefs is associated to people disavow of any form of politicized identity, or only forms that are system challenging.

Participants
Four hundred eighty-nine participants (299 women) took part in the study. 3 While 12 participants completed the study during the fourth YV protest in Paris (December 8th), most of the sample completed the questionnaire online, between 8-20 December, using Qualtrics (distributed via social media websites, Facebook and Twitter). As suggested for SEM analyses (Kline, 2015), sample size was determined using the criterion of 20 participants per estimated parameter, leading, for 23 estimated parameters, to 460 participants. Participants ranged in age from 17 to 93 years of age (M age = 41.95, SD = 16.24). In total, 252 participants (51%) showed support for the movement: 173 participants from week 1 (November 17), 48 from week 2, 25 from week 3 and 6 from week 4. See Appendix A for more descriptive information about the sample.

Materials and Procedure
The measures were presented in the order in which they are described below. After the completion of the questionnaire, participants were debriefed and thanked for their participation.

Endorsement of neoliberal ideology
We used a shortened version of the Neoliberal Beliefs Inventory (NBI; Bay-Cheng et al., 2015). We took one item from three of the four dimensions of the scale (system ine- quality, competition, and government interference), and two for the personal wherewithal dimension, one clearly related to meritocracy, and another measuring people's sense of freedom in France. Thus, our final scale consisted of five items (e.g., 'Competition is a good way to discover and motivate the best people'; α = 0.77), assessed on a 7-point Likert scale ranging from 1 (strongly disagree) to 7 (strongly agree).

Social identifications
Participants' identification with French people was measured with one item, 'How close do you feel to French people?' on a Likert scale ranging from 1 (not close at all) to 7 (extremely close). To confirm the mode of identification this general national identification question taps into, participants' attachment to the in-group (r = 0.74, p < 0.001) as well as in-group glorification (r = 0.37, p < 0.001) were also measured on a 7-point scale ranging from 1 (strongly disagree) to 7 (strongly agree) using four items from Roccas, Klar, and Liviatan's (2006) scale. In order to assess participants' identification with their perceived social class, they were first asked to categorize themselves in one of four social classes: lower class, middle class, privileged or very privileged class. They were then asked 'How close do you feel to this social class?' on a scale ranging from 1 (not close at all) to 7 (extremely close).

Attitudes towards the Yellow Vest movement
Attitudes were assessed with one item 'How favorable are you to the YV movement?' on a 7-point scale ranging from 1 (not at all) to 7 (very much). If participants scored higher than the scale mid-point, they were then asked for how long they had been in favor of this movement, choosing between 'Week 1, 2, 3 or 4' with the dates in parenthesis to help them remember. We used this measure to describe the sample characteristics (see also the correlation Table 2), but it is not mobilized further.

Identification with the Yellow Vests
The item assessing social identification was adapted to measure identification with the YVs, (i.e., 'How close do you feel to the YVs?'), on the same 7-point scale. If participants scored higher than the scale mid-point, they were considered to share the politicized identification with the YVs. We refer to those as YVs, and to those who scored between 1 and 4 (included) as non-YVs.

Grievances of the Yellow Vests
Depending on participants' categorization as YVs or non-YVs, they were asked with an open question to either describe three demands of the movement (1) for them as YVs, (2) or as ascribed to the YVs by them. Participants were also asked to evaluate the importance of three given demands on a 7-point Likert scale, ranging from 1 (not important at all), to 7 (extremely important), (1) for them as YVs, (2) or as ascribed to the YVs by them. The three items concerned the regulation of immigration, environmental protection, and change in the French constitution. The regulation of immigration was covered because the representation of the YVs included allusions of racism and of a movement pertaining to the far right (Sebbah et al., 2018). Environmental protection was included because the movement was triggered by a reform on a fuel tax supposed to benefit environmental transition. Finally, the item on constitutional change aimed at quantitatively assessing YVs' desire for systemic, institutional change.

Other politicized identifications
Similarly to the other measures of social identification, participants were asked 'How close do you feel to ecologists/to feminists?' 4

Other SIMCA variables
Based on van Zomeren et al. (2008), cognitive injustice was measured with one item ('I think the way the government treats the lower and middle classes is unfair'), affective injustice with one item ('I feel angry when I think about the way the lower and middle classes are treated by the government'), and collective efficacy with one item ('I think that the actions of the Yellow Vests can lead to real change in France'), on a 7-point Likert scales ranging from 1 (strongly disagree) to 7 (strongly agree).

Engagement in the Yellow Vest movement
We first asked participants if they already had participated in one or more of the movement's actions with a dichotomous yes/no question. We also assessed their behavioral intention to engage in the movement in the future, on a 7-point Likert scale ranging from 1 (not likely at all) to 7 (extremely likely). For both items, participants were given examples of possible forms of engagement (e.g., participating in marches, blockades, circulating information about the movement on the internet).

Socio-demographic measures
Lastly, participants were asked to complete a series of socio-demographic questions, including sex, age, and political orientation (see Appendix A).

Preliminary Analysis: Exploring the Yellow Vests' Grievances
In order to determine whether the grievances given by participants categorized as YVs (N = 197) were indeed system challenging, we first looked at their responses to the open question, where each participant could write down three grievances. The analysis was performed with the Atlas-ti software. In total, 568 responses were collected, containing 644 distinct elements. Each distinct element was coded into one or more of 223 elaborated codes, representing specific grievances and a miscellaneous category for ambiguous, hard to classify quotations.
The codes were grouped into 27 overarching categories, allowing for a clearer summary of the data. A description of the 10 categories collecting the most quotations can be found in Appendix B. Three major topics seemed to emerge from this data: (1) a desire to see the living conditions of the most disadvantaged improve, by increasing purchasing power or lowering taxes for that class (e.g., 'purchasing power = power to live in dignity in our country'); (2) the will to change political institutions and actors, following either their replacement or a change in the rules themselves, for instance by implementing a more direct and participative democracy (e.g., 'reform of the institutions for a more participative democracy'); (3) a recognition of class inequalities and a desire to remedy them, through a fairer tax system or a fairer distribution of resources, and also through the abolition of the ruling class and government officials' benefits (e.g., ' abolition of the government officials and ex-officials' privileges'). As expected, and consistent with Noiriel's (2018b) analysis, this does indicate that the YVs hold a system-challenging stance, directed towards current government representatives but also, and more broadly, towards the way society, its institutions and its distribution of resources are orchestrated.
We then looked at the grievances perceived by non-YV (N = 289) to compare them with those expressed by YVs. In total, 761 responses were collected, with 796 distinct elements listed. Each of these elements were coded into one or more of the 223 existing codes, and into 144 new codes. A description of the 10 categories collecting the most quotations can be found in Appendix C. What was quite noticeable in the comparison with the grievances expressed by the YVs, was that the categories 'Living conditions' and 'Prices' were overrepresented, which is mostly driven by the relatively larger number of quotations mentioning purchasing power, for the former category, and a decrease in fuel price, for the latter category. Similarly, the category 'Government' was also overrepresented, driven by quotations mentioning the President's resignation. On the other hand, the category "Democracy" is underrepresented (see Appendices B and C for percentages). Interestingly, two new categories emerged, one expressing a feeling that the YVs want to cause havoc in the country and attack the Republic and democracy (e.g., 'kill the Republic'). The other new category gathered codes expressing a certain illegitimacy of the movement and its grievances (e.g., 'they know nothing of what they want, poor morons').
We also looked at participants' responses on the three issues of immigration regulation, environmental protection and constitutional change (see Table 1). Adding to the qualitative data, results from the constitutional change item also indicated a desire of the YVs for systemic changes, to a greater extent than is perceived by non-YVs. On the question of immigration regulation, the results were quite similar for YVs and non-YVs, with a lack of consensus within each sample. Inspection of the data distribution for the YV sample indicated that the majority of participants either strongly agreed, or strongly disagreed with the item. The data distribution for non-YVs indicated a rather flat distribution, with responses almost evenly distributed across the scale. Finally, and interestingly, results for the environmental item showed the greatest gap between answers of the YVs and non-YVs. Indeed, while YVs perceived environmental protection as a very important issue, non-YVs perceived it as being rather unimportant to the YVs. Table 2 presents a summary of the variables' descriptions and intercorrelations. To test the hypothesized model (see Figure 1) according to which endorsement of neoliberal ideology would be negatively associated to engagement in the YV movement, through the SIMCA variables, we conducted SEM analyses on intention to engage in the movement (Model 1) and actual past engagement in the movement (Model 2, see Figure 1). Because of occasional missing data, 454 participants were used in the analyses. In order to test the indirect effects, we used bootstrapping with 1000 samples. The maximum likelihood estimator was used for Model 1, and mean-and variance-adjusted weighted least squares (WLSMV) was used for Model 2 where the outcome variable was dichotomous, except when testing the indirect effects with the bootstrap method where the diagonally weighted least squares (DWLS) estimator was used. We relied on the following standards of acceptable fit: root mean square error of approximation (RMSEA) and standardized root mean square residual (SRMR) ≤ 0.08 and comparative fit index (CFI) ≥ 0.90 (Browne & Cudeck, 1993;Hu & Bentler, 1999). 5 First, we looked at the correlations between general national identification and attachment to as well as glorification of the nation, in order to detect which of those two specific forms of national identification might underlie the more general measure. Consistent with our assumption and intent that our general measure of national identification reflects more of a patriotic but critical identification with the nation, general national identification was indeed positively correlated with attachment to the nation (r = 0.38; p < 0.001), but was not correlated with glorification of the nation (r = 0.05, p = 0.508).

Model Testing
Secondly, among the subset of the data that self-categorized as lower or middle class (N = 349), the correlation between national identification and YV identification appeared larger (r = 0.43; p < 0.001), than the one between identification with the lower and middle classes and identification with the YV (r = 0.24; p < 0.001).  Table 3 for the results of Model 1). When tested on actual past engagement in the movement (Model 2), the model fitted the data very well, CFI = 1.00; RMSEA = 0.02 [0.00, 0.08]; SRMR = 0.04, and results were globally similar to those observed in Model 1 (see Table 4 for the results of Model 2). Since our main hypotheses were mostly supported in both models, and because the fit indices were better for Model 2, we will only describe the results on reported engagement (Model 2). We also tested two models where social class (Model 2a) and political orientation (Model 2b) were included as predictors of endorsement of neoliberal ideology, YV identification and engagement in the movement. Indeed, it was possible that those two variables appeared as confounding variables in our predicted model. Compared to people from the lower and middle classes, people from the upper class could indeed endorse neoliberal ideology to a greater extent insofar as it is more in line with their economic interests, and could also be less likely to engage in the YV movement that challenges the status quo. Finally, Azevedo et al. (2019) have also shown that, at least in their USA and UK samples, neoliberal ideology correlated positively with right wing ideologies, and conservatives are also less likely to engage in system-challenging CA (Hennes et al., 2012). However, since our main results remained unchanged, we only describe Model 2's results to be parsimonious and to maintain acceptable power (but see Supplementary Materials for the details of those analyses).
Results showed that, as expected, endorsing neoliberal ideology is negatively associated to politicized identification with the movement, and to the perception of affective as well as cognitive injustice (see Table 4). However, the indirect effect of neoliberal ideology on CA through affective injustice was non-significant (see Table 5). Most importantly, the indirect effect of neoliberal ideology on CA via politicized identification was significant (see Table 5), as was the total effect (Figure 2).

Exploratory analysis on politicized identities and neoliberal
ideology Finally, and interestingly, YV identification positively correlated with ecologist and feminist identification, two system-challenging identities (see Table 6), which again indicates that the YV identity reflects a system-challenging stance. Moreover, the more people endorsed neoliberal ideology, the less likely they were to identify as YVs, but also as ecologists and feminists (see also Fitz et al., 2012;

Discussion
The main purpose of Study 1 was to test an integrative model of CA including an ideological component (see Jost et al., 2017;Osborne et al., 2019), namely endorsement of neoliberal ideology, in order to predict engagement in the YV movement. A qualitative analysis of the grievances expressed by YVs showed that it could indeed be considered a system-challenging movement. In accordance with the SIMCA, collective efficacy as well as affective injustice were positively correlated with intention to engage in the movement (see Model 1). However, cognitive and affective injustice as well as perceived collective efficacy did not have a unique effect on past self-reported CA (see Model 2). Moreover and consistent with our hypotheses, endorsement of neoliberal ideology is negatively associated to engagement in the movement, notably through its association with politicized YV identification. As expected, it is also associated with a lower perception of injustice and to less anger in the face of injustice.  Intention to engage in CA; Attitudes = Attitudes toward the movement (higher scores mean more favorable attitudes); ENGMT = engagement in the YV movement (1 = No, 2 = Yes). Note that 65 participants did not answer the political orientation item. All scales went from 1 to 7 except for the political orientation measure that went from 1 (extreme-left) to 10 (extreme-right). *p < 0.05. **p < 0.01. ***p < 0.001.
Interestingly, while we did not have a specific prediction concerning the relationship between neoliberal ideology and perceived collective efficacy and did not include this link in the model, inspection of the correlations shows that they are indeed related concepts. In fact, endorsing neoliberal beliefs is significantly and negatively related with the perceived efficacy of the YV movement. Since neoliberal ideology is associated with a lower perception of inequality, it may well be that people perceive the YV movement as ineffective against inequalities that they do not acknowledge.
Finally, it appears that neoliberal ideology may undermine multiple, if not any form of politicized identification with system-challenging movements. The fact that all three politicized identities were positively correlated might suggest that developing a politicized identity on one issue increases the chances of developing politicized identities on other topics, presumably when such identities underlie the same view on the status quo, be it system supportive or challenging.
Several limitations can be highlighted in this study. Indeed, while our results suggest that neoliberal ideology is indeed a system-justifying ideology, the fact that we did not have a measure of system justification per se precludes a clearer conclusion. Study 2 aims at testing this hypothesis more directly by relying on a measure of system justification. Moreover, our measure of neoliberal ideology only included four dimensions as measured by the NBI scale, and very few items, and thus could not reflect the variety of values and beliefs (e.g., emphasized pursuit of happiness by hedonism, belief in free will) associated with this ideology. Thus, Study 2 will target other dimensions of neoliberal ideology with more extensive measures, to try to replicate the negative association between its endorsement and CA via YV identification. Also, several indicators used in the model tested in Study 1 relied on only one item, which has been done in the literature (e.g., Lantian et al., 2016;Osborne et al., 2019), but which can carry reliability concerns. Study 2 will rely on several-items scales.
Finally, while Study 1 contributes to the understanding of the way neoliberal ideology affects the various SIMCA predictors, it does not allow us to investigate the processes by which those effects appear. In investigating Note: *p < 0.05, **p < 0.01, ***p < 0.001. potential mediators of the effect of neoliberal ideology on politicized identification with the YVs, through perceived personal control and system justification, Study 2 will help fill this gap while also looking at a more delineated sample of self-defined YV.

Study 2
The NBI scale used in our first study takes account of four dimensions of neoliberal ideology (i.e. structural inequalities, competition, meritocracy, state interventionism). To our knowledge, it is the only measure of neoliberal   ideology endorsement existing at this time. However, as emphasized in the introduction, the polysemy of the concept 'neoliberal ideology' warrants a more diverse measure of endorsement of neoliberal ideology. Consequently, in Study 2, we focus more closely on some other aspects of the ideology: a) the belief in free will b) the valorization of singularity and c) hedonism, which will be referred to thereafter as the 'Neoliberal Corpus' (NC). Apart from this measure which allows us to test the effects of other aspects of neoliberal ideology, we are seeking to achieve a better grasp of the mechanisms through which this ideology negatively predicts identification with the YVs by taking account of the psychological processes which may underlie it (system justification, perceived personal control; Azevedo et al., 2019;Osborne et al., 2019). The choice to include system justification as an explanatory factor of the effects of neoliberal ideology is based both on the theoretical elements Azevedo et al., 2019;Osborne et al.,2019) introduced above and on the results of Study 1. As argued in the main introduction, because of the close links between neoliberal ideology and perceived personal control, we now include this latter variable, expecting that it mediates the association between neoliberal ideology and system justification.
We issue the following hypotheses. A strong endorsement of neoliberal ideology (free will, hedonism and singularity) will be positively associated with perceived personal control (H1). Moreover, we expect that higher perceived personal control will be linked to more system justification (H2), that system justification will be negatively related to identification with the YV (H3) and that the higher the identification to the YV the more likely will be the participation in the movement (H4). Finally, we predict that endorsement of neoliberal ideology will be negatively correlated with participation in the YV movement via perceived personal control, system justification and politicized identity (H5). All our hypotheses are synthesized in Figure 3.

Method
Participants were contacted through social networks. In order to reach participants who were engaged in the YV movement, every Facebook group with the mention 'Yellow Vest' was approached between February and March 2019. We have also collected data with a paper version  of our questionnaire during a YV citizenship assembly (between February and March).

Participants
We received 210 answers but not all could be relied upon. We excluded participants who did not answer to 20% or more of the questionnaire. The final sample consisted of 201 persons, of whom 109 (54.2%) were women (M age = 37.61, SD = 14.7) (see Appendix D for more descriptive information about the sample). In this sample, 156 participants (77.6%) declared that they took part in the YV movement. We tried to collect a majority of people data who were actually engaged in the YV movement. Importantly, Study 2 takes place three months after the beginning of the movement (and after Study 1), thus the YV group stopped to be group around incidental disadvantages (van Zomeren et al., 2008) and became more established. This aim has complicated data collection and explains the smaller sample size compared to Study 1's sample. That said, statistical power remains satisfactory (see the results for justification of statistical power).

Materials and Procedure
The measures were presented in the order in which they are described below (see supplementary materials for the scales' items).

Free wil
We used the French version of the 'Free will And Determinism' scale (Carey & Paulhus, 2013) translated by Caspar, Verdin, Rigoni, Cleeremans and Klein (2017). Seven items measure the free will (e.g. 'people have a total free will'; α = 0.81), on a scale ranging from 1 (totally disagree) to 5 (completely agree).

Hedonism
To measure hedonism participants had to answer to the four items of the 'Life of Pleasure' subscale from the 'Orientation to Happiness Scale' (Peterson, Park & Seligman, 2005). The French version has been validated by Martin-Krumm et al. (2015). This scale is composed of four items, such as 'Life is too short to postpone the pleasures it can provide' (α = 0.77) to be answered on a Likert scale ranging from 1 ('not like me at all') to 5 ('very like me').

Tendency to singularity
This construct was measured using the Bouaziz-Chaara and Brée's six-item scale (2009). They answered questions such as 'being an original person is a part of my personality' (α = 0.85), on a 1 ('totally disagree') to 5 (' completely agree') Likert scale.

Index of neoliberalism
To test the measurement model of neoliberal ideology, we ran a confirmatory factor analysis (CFA) with the Lavaan R package (Rosseel, 2012). We posited that neoliberal ideology is a second-order factor composed of free will, hedonism, and singularity (see Figure 4). We used robust maximum likelihood (MLR) to compensate for violation of the multivariate normality assumption. To assess this model, we used the following criteria: CFI value above 0.90, and RMSEA and SRMR value below 0.08 (Browne & Cudeck, 1993;Hu & Bentler, 1999). The model robust indexes were satisfactory CFI = 0.95; RMSEA = 0.05, 90% CI = [0.03, 0.07]; SRMR = 0.06. The reliability of the second-order factor can be assessed by the coefficient ω h (McDonald, 1999). Results indicated that neoliberal ideology (second-order factor) had a good reliability (ω h = 0.81). Consequently, we used this global index composed of free will, hedonism, and singularity.

Perceived personal control
We used 10 items of the sphere of control scale (Paulhus & Van Selst, 1990) measuring perceived personal control ('I can learn almost anything if I set my mind to it'). Responses were given on a 7-point scale ranging from 1 ('totally disagree') to 7 (' completely agree'). This scale was subject to a transcultural translation following the methodology of Vallerand (1989). The internal consistency was not satisfactory (α = 0.63), consequently we ran an exploratory factorial analysis (Oblimin with the parallel method). We thus rejected complex items and kept the five remaining items to assess perceived personal control. The new Cronbach alpha was satisfactory (α = 0.79).

System justification
For this construct, participants completed the General System Justification Scale (Kay & Jost, 2003) translated by Vasilopoulos (2017, unpublished), on a Likert scale ranging from 1 (' completely disagree') to 7 (' completely agree'). Initially, this scale was composed of eight items but once again we met psychometric issues. We ran an exploratory factorial analysis (Oblimin with the parallel method). We kept the five items with the higher loadings ('In general, you find the society is fair') to reach a good internal consistency index (α = 0.87).

Identification with the YV movement
We adapted Sani, Madhok, Norbury, Dugard and Wakefield's scale (2014). Four items were used to appraise identification with the YV movement ('I feel I belong to the YV group'; α = 0.93). Responses were given on a 7-point scale ranging from 1 ('totally disagree') to 7 (' completely agree').

Effective engagement in the YV movement
Engagement was measured using the following dichotomous question 'Did you participate in the YV movement'.

Political tendency
A dichotomous variable appraised the political tendency ('Do you have a political tendency'). If they answered 'yes' we asked participants to position themselves on a continuum from extreme left-wing to extreme right-wing inspired by Urbanska and Guimond (2018).

Socio-demographic information
Participants completed the questionnaire indicating age, gender, level of education, and socio-economic situation (see Appendix D). Table 7 presents a summary of the variables' descriptions and intercorrelations. To test our hypothetical model (synthetized in the Figure 3), we performed a path analysis with the Lavaan R package (Rosseel, 2012). Specifically, we first used neoliberal ideology as a predictor of perceived personal control (H1). Second, we assessed the strength of the link between perceived personal control and system justification (H2). Third, we assessed the relationship between system justification and politicized identification with the YVs (H3). Fourth, we predicted engagement in the YV movement by identification (H4). Finally, we estimated the indirect effect of endorsement of neoliberal ideology on engagement through perceived personal control, system justification and identification with the YV (H5). According to Bandalos (2014), the robust version of DWLS performs better than the unadjusted WLS method with binary outcomes. This author suggests that samples sizes should not consist of fewer than 200 participants. Consequently, we used the WLSM method to run our analyses. In order to reduce the number of free parameters we did not include direct effects into the model equations if we did not have specific hypotheses about them. Thereby, we kept at around twenty the number of participants by free parameter. The model indexes were not satisfactory: Robust CFI = 0.95; Robust RMSEA = 0.10, 90% CI = [0.07, 0.13]; SRMR = 0.07. We analyzed the modification indices. Results suggested that perceived personal control should be included in order to predict participation in the YV movement. We tested this new model (see Figure 5). The model indexes were better than those of the previous model, Robust CFI = 0.98; Robust RMSEA = 0.07, 90% CI = [0.03, 0.10]; SRMR = 0.05. Path analysis results showed that the greater the endorsement of neoliberal ideology the higher the perceived personal control (a = 1.06, p < 0.001, R² = 0.26). This last factor was positively related to system justification (b = 0.39, p < 0.001, R² = 0.19) which was also negatively linked to YV identification (c = -0.66, p < 0.001, R² = 0.17). Finally, YV identification (d = 0.15, p < 0.001) and perceived personal control (b = -0.07, p < 0.001) predicted participation in the YV movement (R² = 0.52). The indirect effect of neoliberal ideology on engagement (abcd = -0.04, p = 0.002) was significant. All these results are consistent with our hypotheses (for more details see Tables 8 and 9). We also controlled for the effects of social class and political preference in the regression and path analyses. The effects remained similar to those of Model 4 (see supplementary materials). In order to respect the parsimony principle and to keep 20 participants per parameter, we decided not to include those control variables in the final model. The reader interested in those analyses can refer to the supplementary materials.

Discussion
The objective of Study 2 was twofold. Firstly, we wanted to replicate the negative association between endorsement of neoliberal ideology and engagement in the YV movement with another measure of neoliberal ideology. In accordance with our expectations and the results of Study Identification scale, PART = Participation. All scales which make up the neoliberal ideology index went from 1 to 5. All the other scales went from 1 to 7 except for the general system justification scale that went from 1 to 9. *p < 0.05, **p < 0.01, ***p < 0.001.
1, endorsement of neoliberal ideology was negatively associated with identification with and participation in YV CA. Next, we wanted to deepen our understanding of this effect. The results of Study 1 and of Azevedo et al. (2019) imply the justifying nature of neoliberal ideology. As expected, our results indicate that endorsement of neoliberal ideology is positively associated with the propensity of participants to consider the system as being just. In order to explain how this ideology may nourish the system justification process, we have assumed that perceived personal control (H1, H2) plays a mediating role, insofar as neoliberal ideology allows social hierarchy to be conceived of as the result of the exercise of freedom and personal control. Moreover, and in accordance with earlier work and the Note: *p < 0.05, **p < 0.01, ***p < 0.001, parameter coefficients are standardized.  results of Study 1, justifying the French system is negatively correlated with identification with the YV system-challenging movement (H3) : Osborne et al., 2019 and identification with the YV group predicts participation in CA (van Zomeren et al., 2008) (H4). Finally, our results support the supposed pathway (H5). In fact, it seems that the effect of neoliberal ideology on engagement in the YV movement is associated to an increased feeling of personal control, which in turn is positively associated to the propensity to consider the system as just. This phenomenon engenders a decrease in identification with the YV system-challenging movement and therefore decreases engagement in the movement. It is worth noting that an alternative model in which personal control and system justification are reversed fits the data more poorly (see supplementary materials for details).

General Discussion
The present research highlights interesting elements concerning the YV movement. Indeed, it does appear as a system-challenging movement, both on qualitative (Study 1) and quantitative data (i.e., positively associated with a wish for a constitutional change, Study 1; negatively associated with general system justification, Study 2). Engagement in the movement was also found to positively correlate with perceived injustice against lower and middle classes, anger in the face of this injustice, perceived efficacy of the movement to bring about change (Study 1; Table 2), and negatively with system justification (Study 2). Results were consistent across the two studies, despite relying on samples that vary in diversity (i.e., more heterogeneous in Study 1 than in Study 2).
In Study 1, those who identified as YV were also more likely to share other system-challenging politicized identities. Moreover, and consistent with the idea that the YV movement is also defending the French identity and its ideals (Jetten et al., 2020), the YV identity was found to correlate positively with national identification (more so than it did with social class). Importantly, national identification took the form of a critical attachment to the nation, and not its glorification.
The results of the two cross-sectional studies also show that endorsement of neoliberal ideology, assessed via several of its dimensions, predicts less engagement in the YV movement, a movement that appears to be a system-challenging one. Moreover, this research adds to the line of research on the psychological effects of neoliberal ideology (e.g., Adams et al., 2019;Azevedo et al., 2019;Bay-Cheng et al., 2015;Pulfrey & Butera, 2013;Teo, 2018) in showing that its endorsement is associated with a lower perception of inequalities, less anger in the face of such inequalities (Study 1) and a greater sense of personal control, which reinforces system justification (Study 2). Central to our hypotheses, endorsement of neoliberal ideology is also linked to a lower politicized identification with the YVs, which is likely to undermine CA, as revealed by the mediation models consistent across both studies. Therefore, by contradicting the tenets of a politicized identity, neoliberal ideology may impede the YV identity, as well as other politicized identities (see Study 1's results). At the same time, these results support the centrality of politicized identity as a major predictor of engagement in CA (van Zomeren et al., 2008).
This research also corroborates Jost et al.'s (2017) suggestion and Osborne et al.'s (2019) empirical findings that CA takes place in certain political, economic, social, and ideological contexts, that people may be motivated to defend and justify. To do so, people can rely on available ideologies, including neoliberal ideology (Azevedo et al., 2019), that impede the major predictors of CA (van Zomeren et al., 2008), and most notably politicized identity. It is worth noting that while the SIMCA find empirical support in Model 1 (Study 1), in Model 2 (Study 1), only politicized identity predicts past engagement. It may be due to the fact that accounting for past behavior involves the usual predictors of CA in different ways (see Becker & Tausch, 2015). Notably, actual engagement in CA can then reinforce politicized identification (Becker & Tausch, 2015;Stürmer & Simon, 2004a;van Zomeren et al., 2008), and this could explain why it is the only significant predictor of past engagement in the YV movement in Model 2 (Study 1).
Additionally, while anger predicts engagement in normative CA, it is not a good predictor of non-normative CA (Becker & Tausch, 2015;Jost et al., 2012). Indeed, contempt is the best emotion to predict engagement in violent non-normative CA (Becker & Tausch, 2015). Only anger was measured in Study 1, while our outcome variables could encompass many forms of CA, be they normative or non-normative. In fact, the YV movement has taken many forms, from petition signing to blockades, riots, and the destruction of property. Thus, the strength of the relationship between anger and CA (Study 1) may have been undermined by the variety of CA that participants could have in mind when they thought about engagement in the movement. While the focus of this research was not to test the specific predictors of engagement in these two types of protests, future research could investigate the specific effects of neoliberal ideology on normative and non-normative CA. In fact, it might well be that the influence of neoliberal ideology endorsement on a lower, or lack of engagement might be stronger for non-normative, compared to normative protest, insofar as non-normative protest underlies a dedicated attempt to achieve a largescale change and to affirm the unity of the group challenging the status-quo . Similarly, while efficacy seems to positively predict engagement in normative CA, it negatively predicts engagement in violent non-normative CA (Becker & Tausch, 2015). Thus, while testing the SIMCA was not the primary focus of this paper, another and more complex model could be useful to better understand engagement in the YV movement (e.g., Becker & Tausch 2015).
Some other limitations of this research can be outlined, one being its correlational nature. Indeed, while these results yield very promising and interesting results, an experimental study could directly test the prediction that exposure to neoliberal ideology undermines YV identification and engagement in the movement (see  in relation to feminist identification). Moreover, longitudinal studies could allow us to test their predictive power over time and across different forms of CA (Becker & Tausch, 2015;Jost et al., 2012).
Additionally, while it is assumed that the focus on personal control and the undermining of structural factors go hand in hand (e.g., Brown, 2015), we did not directly measure the latter to test this specific hypothesis. We did find that endorsement of neoliberal ideology was associated with a lower perception of systemic inequalities (Study 1), but it is not quite the same as demonstrating that neoliberal ideology leads people to minimize the perceived importance of structural factors on one's life. A single study measuring those two variables could shed light on this issue.
Finally, the diversity of dimensions used across those two studies to assess neoliberal ideology allows for a rich inspection of its involvement in the non-engagement in the YV movement, but it also raises the question of the psychological levels at which they each operate, from overarching values (e.g., hedonism) to more elaborated cognitive appraisals (e.g., government interference). While this was not the point of this paper, it might be necessary to more clearly define and operationalize this ideology in order to understand its influence on various behavioral outcomes (see , for similar interrogations).
In conclusion, neoliberal ideology, standing as a dominant ideology in Western countries (Navarro, 2007) is negatively associated to the engagement in system-challenging CA. Its specific system-justifying rationale seems to provide the basis for a depoliticization of social issues (Arfken, 2018;Beattie, Bettache & Chong, 2019;Brown, 2015), while people's disadvantage is described as resting on their own shoulders (Pyysiäinen et al., 2017). This depoliticization may thus undermine people's possibility to collectively challenge systemic injustices, ensuring a relative stability of the status quo and its inherent inequalities. However, even if neoliberalism is a 'hegemonic discourse' (Harvey, 2007), movements like the Yellow Vests and many others around the world are demonstrations that it is not omnipotent. People still draw from historical forms of solidarities while taking advantage of new forms of collaborations and organizing, rendered possible, for instance, by the social media. Thus, while CA might become more and more impeded in neoliberal contexts (Brown, 2015), there still exist ongoing dynamics of change that some people, all across the world, are trying to achieve.

Data Accessibility Statement
This manuscript is based on data collected as part of the first and second author's PhD.
The data can be accessed with the following link: https://osf.io/n7bz5/.

Notes
1 Note that the SIMCA has also been applied to opinionbased groups, which are groups that are defined by a shared opinion (Thomas et al., 2012). 2 It is worth noting that we do not expect endorsement of neoliberal beliefs to be associated with collective efficacy. Indeed, while we do believe that endorsing neoliberal ideology may lead individuals to perceive collective movements as illegitimate (Brown, 2015), all the more so when they fall outside a conventional framework, we do not believe it to be necessarily associated with the perception that they are inefficient.
Consequently, the relationship between neoliberal ideology endorsement and perceived collective efficacy is not specified in the model. 3 In total, 599 people accessed the study but 110 were excluded from the analyses because they did not complete the measures to the end and/or because the same IP addresses with the same socio-demographics completed the study more than once. 4 Measures of politicized identification with other movements (Unions; La Manif pour Tous and Génération Identitaire, a far-right movement) were included but could not be relied upon due to floor effects. 5 We did not rely on the Chi-square indicator because it is uninformative with samples larger than 400 (Kenny, 2015).

Additional Files
The additional files for this article can be found as follows: • Appendices. Additional socio-demographic infor-