Digitally Networked Participation and Lifestyle Politics as New Modes of Political Participation

Political participation has seen substantial changes in terms of both its structure and its scope. One of the most prominent venues of citizen engagement today is participation that relies on online means. Several approaches to online participation have attempted to understand its nature as a continuation of offline acts into the online realm, or as an independent form. In this article, we determine the place of online participation in the repertoire of political participation with greater precision. We ask whether, in particular, digitally networked participation (DNP) is an expansion of lifestyle politics, or whether there are empirical grounds to classify it as a new, independent mode of participation. We study a large variety of participatory activities using data from an online survey conducted among 2,114 politically active individuals in Belgium in 2017. We use an innovative measurement approach that combines closed ‐ with open ‐ ended questions, which allows us to explore new forms of partic - ipation that have previously not been considered or measured. Our results show that DNP is a core part of today ’ s activists ’ repertoire and a distinct mode of political participation that is clearly attractive to younger, critical citizens.


Introduction: Changing Activism
Since it was defined in the 1970s as "activities by private citizens that are more or less directly aimed at influencing the selection of governmental personnel and/or the actions they take" (Verba & Nie, 1972, p. 2), political participation has seen substantial changes in terms of both its structure and its scope (Norris, 2002;van Deth, 2001van Deth, , 2014. Stretching beyond early distinctions between conventional and unconventional, political participation now manifests in a wide variety of individualized, creative, expressive, and everyday forms of engagement with societal and political issues. Indeed, the very concept of political participation has seen substantial revisions in recent years to accommodate these changes, with van Deth's (2014) "conceptual map of participation" now offering a tool to identify any emerging activity as being within or outside one of four basic categories of participation. However, the question remains whether and how emerging forms of participation represent the emergence of more encompassing modes of participation. This study addresses the question whether the expansion of political participation represents the continuation of previous logics of participation in new forms and modes, or a more fundamental shift in the way citizens engage with politics.
One of the most prominent venues for the expansion of citizen engagement today relies on online means. Online participation, as it is often called, includes a wide variety of activities that range from the rather static posting of political content in government forums or signing a petition on a "petition warehouse website" (Earl & Kimport, 2011) like Change.org, to interactive activities such as responding to a local representative on Twitter, or posting a call for support about an issue of interest on Facebook (Dennis, 2019;Mercea, 2016). The hybridization of some online and offline forms of participation (Gibson & Cantijoch, 2013) and the increasing blurring of boundaries between these realms when it comes to political organization (Bennett & Segerberg, 2013;Bimber, Stohl, & Flanagin, 2009) have ensured that online and offline activism are nowadays strongly intermingled. Concurrently, several approaches to online participation find it to be a continuation of some offline acts into the online realm (Hirzalla & van Zoonen, 2011;Oser, Hooghe, & Marien, 2013).
Yet the use of social media to participate in politics might indicate more fundamental changes. The technological affordances of Web 2.0 technologies have altered the participation landscape, and various forms of participation may indeed offer an entirely new avenue for engaging in politics (Theocharis, 2015). In particular, authors have underlined the unique characteristics of what has been labeled as "Digitally Networked Participation" (DNP), defined as "a networked mediabased personalized action that is carried out by individual citizens with the intent to display their own mobilization and activate their social networks in order to raise awareness about, or exert social and political pressures for the solution of, a social or political problem" (Theocharis, 2015, p. 6).
Still, it might be premature to conclude that DNP constitutes its own mode of participation. A "mode" of participation refers to the notion that, resulting from some shared feature like a strategic logic, targets or arenas, specific political activities can be seen as expressions of an underlying style of participation (Theocharis & van Deth, 2018a, p. 17). Scholars generally find positive correlations between specific forms of participation, but being involved in one expression of a certain mode of participation makes it particularly likely that a person will be involved in other expressions of that mode of participation as well. For instance, a collection of activities related to political parties and the electoral arena-such as attending a party meeting or donating to an electoral campaign-are depicted as an institutionalized mode of participation. Likewise, actions that focus on the posting of political content in digital networks like social media could make up a distinct mode of political participation (Theocharis & van Deth, 2018a).
In this study, we argue that the jury is still out as to whether DNP is (i) a mode of participation in its own right or; (ii) one of multiple expressions of a broader, emerging mode of participation that includes many other "new" forms that share similar characteristics. Regarding (ii), which has so far remained empirically underexplored, some have suggested that DNP is an expression of a new mode of "personalized" or "individualized" participation (Bennett, 2012;Vromen, Loader, Xenos, & Bailo, 2016), and therefore not necessarily a mode of its own. Rather, it is clustered alongside other forms of "individualized collective action" (Micheletti, 2003, p. 34) such as, most notably, lifestyle politics or "the politicization of everyday life choices, including ethically, morally or politically inspired decisions about, for example, consumption, transportation or modes of living" (de Moor, 2017). As we will discuss below, both DNP and lifestyle politics can often be depicted as individualized, self-actualizing, and expressive participatory forms, and as taking place through everyday activities. In short, while some characteristics, like the focus on social media, clearly set DNP apart from other forms of participation, and while lifestyle politics is unique in its focus on material, do-it-yourself (DIY) activities like shopping, transportation, and energy use, there are also some clear overlaps with lifestyle politics that suggest that the two might be expressions of a shared, individualized mode of action.
Most surveys include only a limited number of very traditional participation measures without expanding to lifestyle or digital politics, and few studies are able to empirically assess these differences. Addressing this gap in the literature deepens our understanding of what sets apart emerging modes of political participation, which should, in turn, sharpen debates and future research about the origins, distinctiveness, and directions of found modes of participation. Our study includes a uniquely detailed measurement of forms of participation linked to individualized modes of participation. We study a large variety of participatory activities, ranging from traditional forms within the electoral arena and party politics to numerous expressions of protest, online participation, and lifestyle politics, using data from an online survey conducted among 2,114 politically active individuals in Belgium in 2017. An innovative measurement approach that combines closed-with open-ended questions is used. This approach allows us to both explore and validate whether new forms of participation that have previously not been considered or measured are empirically clustered within the modes they theoretically belong to.
Our results show that, even when measured alongside lifestyle politics, DNP proves to be an independent part of today's activists' participation repertoire, suggesting that technology-rather than individualization-is the main distinguishing feature at play. Furthermore, by comparing the core individual antecedents associated with validated modes of participation, we find that DNP is a mode of engagement that is uniquely attractive to younger individuals, those who are more dissatisfied with democracy, those having lower political trust, and those with a lower level of socioeconomic security.

DNP and Lifestyle Politics
DNP embeds characteristics that are inherent in Web 2.0 technologies (such as, e.g., social media), namely interactivity, collaboration, community-based input, interoperability, content curation, production and sharing (O'Reilly, 2012). In this way, DNP extends far beyond the more static-and in a way solitary-online political activities of the pre-Web 2.0 era (such as participating through e-government forums in an asynchronous with others fashion or clicking "donate" on a candidate's website). It is instead based on the idea that producing political content while activating one's networks via digital media with the aim of mobilizing others for social or political purposes is part of the repertoire of political participation (Theocharis, 2015).
Several studies employing the concept and type of activities underlined by DNP-including in the context of mobilization , participatory inequality (Bode, 2017), the participatory consequences of selective exposure (Feezell, 2016), feminist (Heger & Hoffmann, 2019) and youth activism (Ekström & Shehata, 2018;Pickard, 2019)-have shown that it has become a popular form of engaging in various areas of politics and civil society. Moreover, manifestations of DNP have been shown to be crucial for shaping political selfconcepts such as political self-efficacy and interest, strengthening people's political self-presentation on social media, and their public commitment to their image as politically active (Lane et al., 2019). As such, DNP has a variety of manifestations that may range from highly vocal, symbolic, and interactive (such as attaching #MeToo to a political post on Twitter, identifying with #ClimateChangeIsReal by changing one's profile on Facebook, or encouraging others to vote for a certain party by creating a YouTube video), to less intense acts (such as commenting through a tweet to an ongoing public discussion about a political issue) (Suk et al., 2019). Most important, DNP often is an expressive act usually not enabled in a topdown manner initiated by the government or some other organization interested in activating citizens. To the contrary, it relies on platforms developed for social networking, entertainment, self-expression, and socialization, and can thus be spontaneously political and targeted toward multiple actors and groups within and beyond the government in a bottom-up manner. This flexible, interactive, low-cost, and collaborative nature of DNP, along with the attention-grabbing potential it can have when (and if) it scales (Barberá et al., 2015;Margetts, John, Escher, & Reissfelder, 2011;Margetts, John, Hale, & Yasseri, 2015), has made it a popular way of engagement across the world (Theocharis & van Deth, 2018a;Anderson, Toor, Rainie, & Smith, 2018), and one with important effects on how citizens think about politics (Lane et al., 2019).
But while DNP can certainly be seen as a form of political participation with distinct features, it is not clear yet whether it makes up its own mode of participation. Understanding that is not only important in terms of being able to investigate the meaning and consequences of its adoption, but also for actually improving the way we measure online participation in the future. From a conceptual point of view, we consider it established that, following van Deth's "map of political participation," DNP should be classified as a form of political participation. 1 Thus, even though DNP typically does not make use of institutional participation channels (van Deth's "minimalist" definition), it does sometimes target the political decision making process or other societal problems in an extra-institutional fashion (van Deth's "targeted" definition), or is used to express political views and mobilize others (van Deth's "motivational" definition) (Theocharis, 2015). Notably, a similar argument has been made about lifestyle politics (de Moor, 2017).
But whether DNP makes up a mode of participation cannot be established solely through this conceptual approach. For one, researchers have highlighted different dimensions of clustering, which result in sometimes competing expectations about how forms are clustered into modes of participation. Authors have organized forms into modes of participation based, for instance, on their "conventionality" (Barnes et al., 1979), the arena in which they're taking place (e.g., institutional vs. non-institutional; Norris, 2011), their target (e.g., state or non-state; van Deth, 2014), and associated norms of citizenship (Dalton, 2008a). Depending on which features are considered, the place of DNP in the wider repertoire of political participation has been imagined in various ways. In particular, two competing ideas have emerged.
Some scholars, referring to the characteristics and organizing impact of digital media and their particular affordances, straightforwardly expect DNP to make up its own mode of participation (Theocharis & van Deth, 2018a, 2018b. Individuals engaged in some form of DNP are expected to do so as a commitment to engage in politics through these venues, or because digital media gives them an entry point to becoming politically active (Ekström & Shehata, 2018). In both cases, DNP offers unique opportunities for individualized, self-actualizing, and expressive activities, which have been extensively discussed and documented in the literature (Bennett, 2012;Vromen et al., 2016).
Theocharis and van Deth's (2018a) empirical study confirms that DNP makes up its own mode of participation when studied alongside a range of common forms of participation. Yet DNP's specific relation to lifestyle politics-another quintessential area for the expansion of political participation (Stolle & Micheletti, 2013)-was not measured comprehensively in Theocharis and van Deth's study. Of course, there are several reasons to expect them to be distinct. First, even though the activities clustering under lifestyle politics sometimes take place in "lifestyle collectives" and may relate to the public promotion of lifestyle change, they typically focus on the domestic sphere (de Moor, 2017). By contrast, DNP activities are literally defined by their character as being embedded in more "open" social networks, such as Facebook and Twitter. Second, lifestyle politics has a clear material focus through its engagement with things like food, energy, transportation, or fashion, leading some to define it as a form of "sustainable materialism" (Schlosberg & Craven, 2019). DNP focuses on immaterial things like news and discussion. These distinctions should nevertheless not be overstated. For one, there is a clear overlap between DNP and lifestyle politics given the fact that DNP is often used to promote lifestyle politics (e.g., tweeting about the need to fly less; Stolle & Micheletti, 2013).
Other scholars, referring to logics of action and citizenship norms, suggest that DNP and lifestyle politics are both part of a broader emerging "individualized mode of action" used to express "self-actualizing" citizenship norms (Bennett, 2012;Loader, Vromen, & Xenos, 2014). According to Bennett, Wells, and Rank (2009, p. 106), "lifestyle politics entail greater personally expressive or self-actualizing affiliations that can be fluid and changing," and it is precisely because of this individualized, expressive, and fluid nature that DNP has been conceptually so closely associated with it (Gil de Zúñiga, Copeland, & Bimber, 2013). Bennett argues for instance that: Some of these [personalized] politics have specific consumerist styles, as in the many expressions of concern about the social or environmental realities beyond the brand image of popular products. Beyond consumer and lifestyle actions, large individualized collectivities have also emerged around broader political agendas with the help of various social and digital media (Bennett, 2012, p. 37).
Likewise, both DNP and lifestyle politics share key characteristics of what Micheletti (2003) has defined as "individualized collective action." They differ from more traditional forms of participation by taking place outside the realm of institutionalized politics and of formal organizations, and are often individual, direct, self-actualizing, and taking place through everyday activities. In turn, it has been argued, they are symptomatic of broad trends of the individualization of citizenship norms characteristic of late modernity. For instance, Beck (2006) and Giddens (1991) work on the risk society has argued that with globalization and the spread of, for example, environmental crises, it has become apparent that governments are unable to deal with society's main challenges. In response, citizens shift focus to engaging in "sub-politics," which replace engagement that is aimed at influencing government decisions with engagement targeting a much broader array of actors, and which are especially focused on DIY approaches to societal problems. Drawing on notions like postmaterialism (Inglehart & Welzel, 2005) and liquid modernity (Bauman, 2000), several authors (Bennett, 2012;Blühdorn, 2014) have argued that political participation is increasingly about self-actualization and the development of an identity through the performance of personalized political acts.
In sum, the notion that DNP and lifestyle politics cluster around a single individualized mode of participation presents a convincing alternative to the idea that DNP is a separate mode of participation distinguished by its reliance on social media. So far, these rival expectations have not been tested empirically.

Case Selection and Data Collection: The Belgian Activist Study
We use data collected through an online survey among politically interested individuals carried out in the Flemish part of Belgium in spring 2017. Belgium qualifies as a fairly typical European case for the study of political participation (de Moor, 2016;Deschouwer, 2018). According to Hooghe and Marien (2014, p. 7), "analyses of the European Social Survey have shown repeatedly that it [Belgium] is not exceptional with regard to participation patterns or political attitudes in the European context." According to Quaranta (2013), Belgium is comparable to most other European countries in terms of how different forms of participation relate to each other. Regarding lifestyle politics, Belgium does not score particularly highly, but like many other European countries, it does show an increasing trend (de Moor & Balsiger, 2018;de Moor, Marien, & Hooghe, 2017), while DNP has remained unexplored. The fact that Belgium is a fairly typical case for the study of political activism does not mean that our study presents generalized statements for other European countries. Rather, Belgium is sufficiently similar to other European countries so that our findings will not be biased due to specific circumstances or developments.
The survey was conducted as part of the University of Antwerp Citizens Panel, which has run since 2004, and which was made available for the current study. 2 Participants in the panel were mainly recruited by inviting people in the street and through an online voting assistance application. The panel was not designed to be representative of the Belgian population. Instead, it presents a convenience sample, but one that, following the principles of quota sampling, offers sufficient diversity on a number of key characteristics to provide the statistical power needed to analyze relationships among a broad section of the population, including variations across age, sex, and education. Thus, even though members of the panel are disproportionately interested in politics, 3 more males (73 percent), older (sample 54, Belgian average 41), and better educated, 4 the panel has sufficient diversity on each of these indicators to be used for our analyses of DNP and lifestyle politics among the broader repertoire of participation.
This panel provides a strong advantage when compared with previous empirical studies investigating DNP using nationally representative data. General population surveys in Europe typically contain too few individuals who participate in a wide range of political activities. This makes the examination of DNP (or any other new form of participation) difficult to explore. While representativeness is important for generalizations, our sample's bias toward people with high levels of political interest gives us access to a part of the population that may exhibit sufficient diversity in political participation to explore new forms and their underlying relationships. In particular, the open-ended question approach (see below) was aimed at providing respondents with ample opportunities to mention different ways in which they participate. Confronted with the conflicting demands of representativeness and an encompassing repertoire of participation, we focus on the last; that is, we test empirically the plausibility of the rival expectations about DNP as a mode of participation.

Survey Design
To measure the repertoire of political participation we follow the procedure developed by Theocharis and van Deth (2018a; 2018b) combining a list of closed questions with an open-ended question. In our study, the 21 closed questions enquire about a mix of traditional and newer political activities that the respondent may have carried out over the last 12 months. Overall, this is one of the most extensive questionnaires on citizen participation fielded to date. Some of those activities we enquire about are often part of questionnaires employed in international comparative research (such as the European Social Survey), and some are new (cf. Baringhorst, 2015). We first ask whether the respondent has participated in a range of participation forms, like voting, contacting politicians, and protesting, as well as several forms of lifestyle politics like consuming less or changing one's mode of transportation for political or ethical reasons. Of course, we also ask whether they engaged in DNP acts, such as posting or sharing links on social media (Facebook, Twitter, YouTube, etc.) to political stories or articles for others to read. The full list of questions employed in our study can be found in Appendix I (DNP is expected to be measured by the last three items 18, 19, and 20).
Immediately after the closed-ended questions an additional open-ended question is presented, in which the respondents are asked to enumerate three activities they have engaged in and which are creative/self-expressive in nature. Following a short description with examples of relevant activities, the respondents were first invited to state (yes/no) whether they had been engaged in any such actions to express their political or social views or concerns during the last 12 months. All respondents confirming this last question were then asked: "what did you do?" and were consequently offered the opportunity to write down up to three activities in their own words. An open-ended question offers the opportunity to inductively discover emerging forms of participation, thus ensuring the validity of our measurement (see Appendix I for the exact question wording). This is especially important when it comes to identifying new, innovative, individualized acts that are currently less known to researchers.
The coding process of the responses to the open-ended question is detailed in Appendix II. A total number of 983 respondents (46.6 percent out of 2108; six missing) indicated that they had been engaged in actions aimed at expressing political or social views or concerns similar to the ones mentioned in the introductory text of the open-ended question. This is approximately 30 percentage points higher than the study with representative sample conducted by Theocharis and van Deth (2018a) (though the wording of our open question was mildly different), confirming our expectations that a more politically active sample would yield a greater volume of responses in the open-ended question. Of these respondents, 937 used the opportunity to specify their engagement in their own words: 54 mentioned only one particular action or cause, 115 respondents mentioned two, and 768 mentioned even three actions or causes. Consequently, a total of 2,588 actions or causes were registered for the openended question.
In order to make the actions mentioned accessible for quantitative analyses, a (re)coding procedure was developed including two steps. In the first step, on the basis of a screening of the responses provided (and the aim to cluster together and better specify expressive participation in politics), additional codes were created for the following categories: In a second step, all actions that were mentioned and were already unmistakably covered by one of the 21 items used in the closed-ended questions were deleted from the list of additional forms of participation and coded under the respective form already captured by the closed-ended item. This means that, for example, an answer specifying that one has "signed a petition" was coded as "signing a petition" and counted under the closed item for that form of participation. Similarly, posting content on social media "about lying politicians" was simply added to the item "posting political content on social media." In this way, the total numbers of people engaged in a specific activity mentioned in the closedended questions increased slightly after the corrections were made. 6 Many of the responses do not even come close to the definition of political participation as traditionally formulated by Verba and Nie (1972). Clearly, individualistic, self-expressive, and creative acts are an important part of the contemporary participation repertoire of politically active individuals. Yet, the most striking finding is the absence of responses to the open-ended question describing online forms of participation. While several respondents referred to activities that can be carried out via digital media, those activities were-in all cases but one-already tapped by the three items measuring DNP (items 18,19,and 20 in Appendix I). No digitally-enabled activities beyond those three items made up for an additional category in the coding procedure. While one might argue that this could be stimulated by only prompting the respondents with one example of an online activity (sharing photos in the turbulent 2014 Ferguson protest events), it is hard to believe that in a panel of more than 2,000 politically active people (who appear to be highly inventive in mentioning all kinds of actions) additional digitally-enabled activities would have been so overlooked.
The results for the total of 34 forms of participation measured (voting +20 closed items +13 added forms) are summarized in Figure 1. The participation varies enormously and many lifestyle activities are highly popular. Aside from donating to, and volunteering for, social organizations, activities such as reducing energy, reusing, boycotting products, and buycotting are carried out by a significant part of our sample. DNP items, with percentages reaching as high as 43 percent for posting/sharing political links, and almost 40 percent for commenting on political discussions on social media, are also among the most popular activities. Institutionalized activities linked to party politics and campaigns are lower on the list with percentages ranging from 11 to 23 percent. Finally, a host of creative activities that have been classified using the openended question, populate the lower end of the figure. Obviously, the percentage of people involved in any of these actions is rather limited because (i) relatively unusual and highly individual-specific activities emerge here, and (ii) large numbers of responses to the open-ended question are included under the 20 closed items. Still, some of those activities are carried out by as much as almost five percent of the sample, which is quite remarkable when one considers that several electoral campaign-related activities hardly exceed 10 percent. To facilitate further analyses, the open-ended question answers are summarized in two variables: one for all lifestyle-related issues, that is, lifestyle (other), and one for all remaining responses, that is, participation (other).

Analytical Strategy and Findings
The Structure of Political Participation The main objective of this study is to validate whether DNP is (i) a mode of participation in its own right, or (ii) just one of the multiple expressions of a broader, emerging mode of participation that includes many other "new" forms that share similar characteristics. We, therefore, seek to empirically investigate and evaluate two theoretical dimensional structures: (i) one in which DNP is independent from lifestyle and every other mode, and (ii) one where DNP is part of an individualized mode of participation together with lifestyle politics. We expect that the rest of the forms of participation measured in our study will cluster in similar ways as in previous studies-that is, into institutionalized participation, protest, and volunteering (Bäck, Teorell, & Westholm, 2011;Barnes et al., 1979;Copeland & Feezell, 2017;Dalton, 2002;Gil de Zúñiga et al., 2013;Harris & Gillion, 2010;Parry, Moyser, & Day, 1992;Teorell, Torcal, & Montero, 2007;Theocharis & van Deth, 2018a;Verba & Nie, 1972;Zukin, Keeter, Andolina, Jenkins, & Delli Carpini, 2006-though, of course, not all of these studies have used these exact labels). Even though we have two theoretical ideas about the structure of political participation and the place of DNP and lifestyle politics in it, we first subject all items to exploratory factor analysis (EFA) to obtain a general overview. Instead of imposing a specific latent structure on the observed indicators, EFA allows the optimal number of factors to be determined based on different statistical criteria (Finch & French, 2015). This allows us to get an indication of how political participation is clustered. Subsequently, we subject the resulting solution to confirmatory factor analysis (CFA) in order to compare the model fit with that of theoretically different mode specifications. Specifically, we compare our model with the main rival specification whereby DNP and lifestyle politics are bundled together as one mode. We run both types of analyses in R, using the psych package (Revelle, 2018) for the EFA and the lavaan package for the CFA (Rosseel, 2012). 7 We base our initial selection of the number of factors to be included in the EFA model on eigenvalues greater than 1. The EFA is based on a tetrachoric correlation matrix using oblique rotation. The model achieves an acceptable fit based on conventional absolute fit indexes (Tucker-Lewis Index [TLI] = 0.856; root mean square error of approximation [RMSEA] = 0.072) and, most importantly, one that is consistent with our prior notions of what the structure would theoretically look like. 8 Table 1 shows the factor loadings for each mode, and includes the bivariate correlations between the different modes at the bottom. The results provide clear support for our first expectation, as DNP emerges as an independent mode of participation with three DNP items clustered together with very high loadings. The EFA also produces very interesting results with regards to lifestyle politics. While, unsurprisingly, it includes consuming less, recycling products, reducing energy, using alternative transportation means, producing own food/energy, and the relevant responses to the open-ended questions, it emerges as a distinct mode of participation from political consumerism. Even though both modes include buying less and correlate considerably (0.56), this finding is contrary to studies depicting them both as expressions of a single mode of participation (de Moor, 2017; Stolle & Micheletti, 2013). For our purposes, it is most important that DNP is only weakly correlated with both lifestyle politics (0.11) and political consumerism (0.21), which strengthens the idea that we are observing different participatory phenomena.
The remaining factor loadings do not hold many surprises. Electoral and partyrelated activities cluster together in a second factor, as do volunteering and protest activities in a fourth and sixth factor, respectively.
To further strengthen our findings, we set up this specification as a CFA model and compare the fit statistics with those of the alternative specification. 9 Figure 2 plots the path diagram and visualizes the factor loadings. As expected, the CFA model confirms the six theorized modes of participation in our data set and, based on all conventional fit indexes, is a very good fit for the data (CFI = 0.952; TLI = 0.943; RMSEA = 0.046). The three DNP items cluster together with very high loadings 10 and DNP remains very weakly correlated with lifestyle politics and political consumerism. A comparison of that CFA model with a CFA model specification whereby lifestyle politics and DNP are merged together as a single mode of participation is a poor fit to the data according to all conventional fit criteria (CFI = 0.705; TLI = 0.769; RMSEA = 0.096). Evidently, DNP is not an extension or subcategory of an individualized mode of participation. Results from this highly politically active sample, then, validate those of previous studies showing DNP emerging as an independent mode of participation (Theocharis & van Deth, 2018a;2018b), even when measured alongside a more elaborate list of Notes: The first six columns represent factor loadings; loadings above the 0.30 cutoff point are shown in bold. The bottom part of the table shows correlations between different factors; h2 represents communalities; u2 represents uniquenesses; com represents the complexity of the factor loadings (see Revelle, 2018). N = 2,114; TLI = 0.856; RMSEA = 0.092. participation items, and in a different national context. Hence, our findings reinforce the idea that DNP is a genuinely new mode of engaging in politics.

Predictors of DNP and Other Modes of Political Participation
Given that DNP can clearly be distinguished from other modes of participation, we take a first exploratory look at how well-established socioeconomic and attitudinal variables that have been found to be key for engagement in different modes of political participation, predict it. Nevertheless, our focus here is on DNP, and not to engage in rigorous extensive analyses or profiling of activists. In the multivariate analysis seen in Table 2, we use a series of OLS regressions where each mode represents a dependent variable. 11 We control for a number of different sociodemographic and attitudinal variables such as political interest, political efficacy (internal and external), political trust, and satisfaction with democracy-all important predictors of participation based on past research (Brady, Verba, & Schlozman, 2013;Dalton, 2008b;Verba & Nie, 1972).
In line with previously reported findings, being younger is a highly statistically significant predictor for engagement through digital avenues-even in this relatively young sample of activists. 12 This is not the case with any other participatory mode, with volunteering being the only mode, in which being older makes it more likely to participate. Being male or female is irrelevant for engaging in DNP. Women are more likely to engage in lifestyle politics and political consumerism-which is also in line with previous research (Copeland, 2014;Stolle & Micheletti, 2013). While education is irrelevant for engaging in DNP, income is not. Those with lower levels of income are more likely to engage through this mode of participation. Part of the reason for that is probably the low-cost nature of DNP, though the coefficient is also negative (and significant) for both institutional participation and volunteering. As far as political attitudes are concerned, individuals that ideologically lean to the left, are more politically interested, have higher levels of internal efficacy, are less trustful of political institutions, and are more likely to engage in DNP. While this is, of course, a very particular sample we are looking at, it is interesting to see that those individuals more likely to engage in DNP fit well with descriptions of younger, "critical citizens" who are more distrustful of institutions of representative democracy, but at the same time by no means uninterested in politics, and eager and confident that they can make a change by engaging through alternative and potentially dissenting participatory avenues (Cammaerts, Bruter, Banaji, Harrison, & Anstead, 2014;Pickard, 2019).

Conclusions
Participation via the online realm is now an increasingly popular avenue for engaging in politics. For example, while a 2013 Pew report found that 39 percent of Americans had, at the time, taken part in some sort of political activity in the context of social media (Smith, 2013), in 2018 Pew reported that this percentage had increased to 53 percent (Anderson et al., 2018). Yet the measurement of online Note: Standard errors in parentheses. Dependent variables are additive indices of the latent variables identified by exploratory factor analysis (EFA). Levels of significance. *p < 0.05. **p < 0.01. ***p < 0.001. participation has been rather incomplete and inconsistent, leaving unanswered the question of whether it is a new and distinct mode of engaging in politics.
In this study, we explored the extensive participation repertoire of politically interested individuals in the Flemish region of Belgium. We used closed-ended as well as open-ended questions that allowed us to get an impression as to the various activities people undertake to engage in politics online and offline. Our analysis focused on understanding how (i) DNP fits into the repertoire of political participation and whether it is related to other recently emergent forms of participation with which it-theoretically-shares an individualistic component, and (ii) how individual-level characteristics that have been previously found to be important for political participation relate to DNP, as opposed to other modes of participation.
Three major conclusions are reached. First, by measuring online participation through both closed-and open-ended questions, we see that the three items we used to capture DNP apparently tap a great deal of political activists' online participation. Our expectation that among highly politically active people the open-ended question would capture a number of new online acts did not materialize: the responses to the openended question did not show any such variation. This is despite the fact that our openended question was explicitly phrased in such a way as to prompt respondents to think of online activities through an example. Only a tiny proportion of the respondents mentioned online activities, and when these were mentioned, they almost always consisted of contextualizations of one of the already existing DNP items. If additional online (non-DNP) activities are not mentioned by thousands of politically active people, it is unlikely that we are missing an aspect of DNP that is critical for understanding online participation in the wider population. We do note, however, that this finding merits further investigation. It is possible, for example, that our participants simply did not recall as "online participation" some online activities which have, by now, converged with their offline counterparts. Signing petitions, contacting politicians, and donating money are cases in point, and we cannot rule out that the overall level of online participation was higher than what we report for DNP. Apart from the latter, however, we can conclude that online participation for political activists may be becoming concentrated to networked, digitally-enabled activities that have become available largely through social media platforms. That DNP might be concentrated in just a few key activities is an important contribution of this study, both in terms of conceptualization and in terms of measurement. Future research should further explore this finding, especially given the theoretically innumerable manifestations that DNP can take.
Our second conclusion is that DNP is clearly an independent mode of political participation. More important, DNP appears to be independent of lifestyle politicsanother mode of participation that is quintessentially associated with the expansion of political participation repertoires, and that has important characteristics in common with DNP. The structural analysis used in our study distinguished the three social media-based acts on one factor from a total set of >20 activities into a mode of digitally enabled, social media-based activities. This evidence corroborates and builds upon previous work on participation and DNP. Specifically, by adding uniquely detailed measures of lifestyle politics, we have been able to rule out the expectation that DNP and lifestyle politics may belong to a single mode of individualized action (Bennett, 2012). Importantly, we do not wish to argue that individualization is not an important driver of changes in political participation. It is, and both DNP and lifestyle politics are evidence of that. But individualization does not appear to be a defining feature uniting these two participatory phenomena. Rather, the use of social media is decisive in structuring the DNP-related area of the repertoire of political participation. Apparently, the emergence of DNP as a new mode of participation (i) is not a phenomenon isolated to one national context and (ii) is emerging in both nationally representative samples (Theocharis & van Deth, 2018a) and samples with politically active individuals. We note that this study also contributes to the debate about the nature of lifestyle politics. Specifically, our findings contribute to the discussion about whether political consumerism is part of lifestyle politics.
Finally, especially young people with lower economic security and trust in the political system-but with higher levels of political interest and internal efficacyare more likely to use DNP as an avenue for engaging in politics. These findings merit further investigation. If DNP is a clearly distinct mode of participation that provides an avenue for younger, less prosperous and more "critical" citizens, then perspectives that recognize that critical stances toward politics and democracy may be altering people's political behavior-both offline (Norris, 1999) and online (Shehata, Ekström, & Olsson, 2016;Vromen et al., 2016)-rather than marking their exit from politics-may be critical for explaining participation in the social media era. More important, DNP may be critical for a vibrant democracy in that area.

Notes
The authors are grateful to the coordinators of the University of Antwerp's Citizens' Panel who have kindly provided them with the opportunity to include their questions in the panel, as well as for collecting the data.
1. We apply van Deth's conceptual map because, unlike more traditional, static concepts, it is designed to accommodate the changing nature of political participation (for one exploration of its theoretical and empirical validity see Ohme, de Vreese, & Albaek, 2018). 2. See: https://www.ua-burgerpanel.be/. A total of 2,292 respondents participated in this survey (CASI; gross response rate 31 percent). Only those 2,114 who responded positively to at least one of our questions about political participation were kept in the data set. The questionnaire for the participation items was developed by the authors. 3. Almost all respondents in our study indicated that they were interested (N = 1,062) or very interested (N = 957) in politics. The average for the panel is an average interest of 8.6 when translated to a scale from 0 to 10. A representative Belgian election survey from 2014 (PARTIREP) shows an average interest of 5.12 on a scale from 0 to 10. 4. Our respondents left school at an average age of 22, suggesting a high and above-average level of education; panel = 64 percent higher education; Belgium = 35 percent "tertiary education" (OECD, 2014).

5.
Although it is not always clear whether evaluating information can be depicted as an activity, these (very few) responses were not deleted from the data set because possible ambivalences might be due to the short-hand way the answers are summarized by the respondents. 6. The overlap between the responses to the two types of questions is clear from the number of responses: of the total of 2,588 actions mentioned, no less than 1,969 (or 76.1 percent) could be coded under one of the closed items. Moreover, whereas 937 respondents used the opportunity to mention additional new actions in their own words, only 464 (or 49.5 percent) mentioned at least one form not already covered by the 21 closed questions. 7. Psych's default extraction method aims to find the minimum residual solution through OLS. For dichotomous data, lavaan uses the Weighted Least Square Mean and Variance (WLSMV) estimator. This estimator uses diagonally weighted least squares to estimate the model parameters, but also the full-weight matrix to compute robust standard errors, and a mean-and variance-adjusted test statistic (Rosseel, 2019, p. 29). 8. We also attempt a five-factor solution, which produces a much poorer fit compared with the sixfactor model based on both absolute (TLI = 0.781; RMSEA = 0.113) and relative (BIC = 2.659 vs. 1.200 in the six-factor model) fit indexes. 9. The cutoff point to be used when selecting items is a subject of considerable debate, with some studies considering loadings as low as 0.3 and others as high as 0.5. To make use of as many theoretically relevant items as possible, in this study we begin by including items with loadings at 0.3 or higher, with the rationale to drop them later if the model is an unacceptable fit for the data. 10. This coherence among the three DNP items is even clearer when the parameter estimates are computed as weighted edges, with loadings of 1.00, 0.92, and 0.97 for sharing, commenting and mobilizing, respectively. 11. These are additive indexes constructed on the basis of each latent variable found in the model. 12. Perhaps with the exception of protest, our models do not yield high model fit, which should be explored further in future analyses.
In Berlin a group of people planted flowers in abandoned sites without permission, in order to make their neighborhood more beautiful; In Vienna hundreds helped creating a miniature model of a city that could have been built instead of bailing out banks. In Ferguson, thousands shared photos on social media to protest the shooting of a young black man by a white policeman. Some people grow their own food because they do not want to support large companies. Others repair defect items to reduce their waste. And some seek to reduce their impact on the environment by changing modes of transport or isolating their house. Many other examples of such expressive and everyday actions can be thought of and these are only some examples.
The respondent was invited to state whether she had been engaged in any such actions to express her political or social views or concerns during the last 12 months ("yes"/"no"; QID13).
All respondents confirming this last question were asked: "what did you do?" and consequently offered the opportunity to write down up to three examples in their own words (QID14_1, 14_2, and 14_3).

II. Coding Procedure for the Responses to the Open-Ended Questions on Political
Participation (Belgium Online Survey, Spring 2017) A total of 937 respondents mentioned at least one particular action or cause similar to the activities mentioned in the introductory text for the open-ended question. A total of 2,588 actions or causes were registered for the three opportunities following the open-ended question.
The number and variety of these responses reflect the expected variety of the actions and causes people nowadays use for political purposes. Yet many answers refer to forms of participation that had been clearly included in the list of 20 actions presented to them only a few seconds before: whereas 937 respondents used the opportunity to mention additional actions in their own words, only 464 (or 49.5 percent) mentioned at least one form not already covered by the 20 closed questions -all other responses to the open-ended question could (and should) have been taken care of by confirming items from the previously presented list. Apparently, a lot of respondents wanted to use the opportunity to specify their activities explicitly. This desire probably also explains the fact that the large majority of those respondents providing an example used all three options available (768 out of 937, or 82.0 percent).
In order to make the actions mentioned accessible for quantitative analyses, a coding and recoding procedure was developed including three steps: 1. Three new variables were constructed to specify additional forms of political participation mentioned after the closed questions were concluded. Because none of the actions mentioned was related to "voted in the last national election," this item was not used for coding. The first 20 entries for the new variables (codes 1 to 20), therefore, are identical to the 20 forms of participation enlisted for the closed question. On the basis of a screening of the responses provided (and the aim to specify lifestyle politics) additional codes were created (see main text). 2. All actions mentioned that were already unmistakably covered by one of the items used in the closed questions were deleted from the list of additional forms of participation-but, of course, coded under the correct form. For instance, the answer "refuse to buy Chinese products" was not coded as an additional form of participation, but recoded under the item "Boycotted products." In this way, the total numbers of people engaged in a specific activity increased after these corrections were carried out.
The overlap between the responses to the two types of questions is rather strong and visible in the number of responses: of the total of 2,588 actions mentioned, no <1,969 (or 76.1 percent) could be coded under one of the closed items. The increase is largest for the use of alternative forms of transportation, with 501 responses to the open-ended question resulted in 42 net increases. Yet even this increase of 1.9 percent is rather limited (the level of participation for this form increases from 57.1 to 59.0 percent after adding the results of the open-ended question).
Thirteen new variables were created for the remaining 619 unique responses to the three open-ended questions (see point 1 above) by transposing their respective codes (30-91) on three coded questions. In this way, each of these additional variables indicates whether an activity has been mentioned. For instance, variable Part50 shows whether the respondent mentioned any activity belonging to the category "Discus political and social issues; convince other people" in the very same way as variable Part6 (a combination of the closed question with added responses to the open-ended questions) shows whether a petition has been signed. Table A.1 shows the frequencies of the 13 unique codes as a percentage of the total numbers of responses to the three opportunities to mention specific actions or causes. Most additionally coded actions show very low frequencies-the largest category consists of "collecting garbage" found among almost four percent of the respondents with a total of 100 mentions.